1. “Swift Horses” from Proto-Indo-European to Greek

Horses are nearly ubiquitous in the early recorded poetries of the Indo-European world, and one particular facet of this presents a perfect starting point for our discussion: a reconstructible Proto-Indo-European poetic expression describing horses. For the earliest poetries of Greece, India, and Iran not only treat horses in ways that are strikingly similar but even utilize some of the same poetic vocabulary for describing them. In particular, horses in Homer are described with the phrase ὠκέες ἵπποι, “swift horses,” while the Avestan Gathas use the corresponding āsauuō aspåŋhō, and the Sanskrit Vedas use āśavas aśvās. [1] All three of these descend directly from the same PIE phrase, *h1ōk̑éu̯-es h1ék̑u̯-ōs. [2] This shared history indicates that the phrase *h1ōk̑éu̯-es h1ék̑u̯-ōs is likely to have been in use in the poetry of the PIE speaking community and that the Homeric ὠκέες ἵπποι reflects a direct continuation of this prehistoric usage. [3]
This is not to say, of course, that PIE poetry deployed this phrase in precisely this form and precisely this order on all occasions, although it may be assumed that it did at times. As Edwards remarks in his criticism of this conventional understanding of formulas, a formula is not a thing “handed down over the ages, like a mummified cadaver, fixed in memory.” [4] Instead, what are generally perceived as formulas, and will conventionally be called so here, are the results of oral compositional techniques which constantly and innovatively exploit thematic, phonetic, and metrical associations to facilitate composition in performance. These techniques, which may more truly than formulas be considered the inheritance of one generation of oral singers from another, render certain patterns as naturally, not prescriptively, frequent products of such composition. That is to say that the sorts of scenes and activities in which horses are likely to be mentioned will restrict the vocabulary items with which they occur and that the poetic devices employed within the singer’s cultural tradition, along with the metrical restrictions of the poetic genre, will further facilitate certain associations of vocabulary which appear as formulas in the fixed and popular sense. [5] In some limited cases, such as Parry’s famous noun-epithet formulas, specific coincidences of vocabulary are so frequent that they do genuinely function as bound units, but these are exceptional. [6]
*h1ōk̑éu̯-es h1ék̑u̯-ōs should not be imagined as the only PIE poetic expression for horses, but simply as the one that yielded ὠκέες ἵπποι, the prolific Greek equine formula. Yet this phrasing must have been used frequently in PIE poetry and the cultural and poetic pressures that caused this frequency must have been sufficiently continuous and long-lived for the constituent elements of this phrase to maintain a strong association into the Greek poetic period. Thus, the particular ways in which this phrase functioned in Proto-Indo-European and in Greek should be defined, as far as is possible, so that a comparison of these formulas may reveal how this maintenance of practice may be accounted for in light of the linguistic and poetic developments of the Greek epic, since these developments may be assumed to have eradicated many other PIE poetic features in the descendant traditions. Such an understanding will prove instrumental in our appreciation of early Greek poetic treatment of horses.
The importance of the horse to PIE as well as Greek culture indicates a continuum of horses’ significance in the intervening phases between these two cultures, and the importance of speed to the horse’s value is easily accepted. [7] There would, however, certainly have been numerous ways to say “a fast horse” in Proto-Indo-European. Therefore, to establish that this particular phrasing is likely to have occurred frequently in PIE poetics, and that its appearance in Greek, Vedic, and Avestan verse is not simply coincidence, the phrase must be shown to exhibit poetic characteristics, as it does quite readily.
In fact, the phrase *h₁ōk̑éu̯-es h₁ék̑u̯-ōs demonstrates several features of PIE poetic technique. [8] The consonantal sequence h₁-k̑-u̯-s occurs identically in both words, forming an alliteratively bound pair. [9] The phrase seems, in fact, to form a jingle, and would presumably have been quite catchy to the ears of a PIE speaker. [10] Yet this phrase would, of course, have appeared in other cases, just as it does in the later traditions, and the phonetic correspondences manifest in the nominative would not be completely identical to those found in other forms. The accusative plural and genitive plural formulas, to draw examples from those occurring most frequently in Homer, may be reconstructed, respectively, as something like *h1ōk̑u-ms h1ék̑u̯o-ms and *h1ōk̑u̯-ōm h1ék̑u̯-ōm. The basic schematic pattern of consonantal alliteration is maintained throughout all the cases, although the final consonant or consonant cluster in each is different; likewise the vocalic features retain poetic qualities in each case, but less neatly than in the nominative. Phonetics alone, then, support the possibility of this phrase having been poetically useful and, therefore, likely to have been deployed by the PIE poet. This evidence, taken together with the phrase’s appearance in our three later IE poetic sources, supports the hypothesis that the later IE formulas descend from a genuinely PIE poetic phrasing.
There is, however, another, less obvious, poetic figure represented here, which makes this conclusion even more compelling. For the two words involved in *h1ōk̑éu̯-es h1ék̑u̯-ōs derive from the Proto-Indo-European *h1ék̑u̯o-, ‘horse’, and *h1ōk̑u-, ‘swift’, the similarities of which are immediately striking. As mentioned previously, the consonantal makeup of both words is identical. Additionally, the difference between the semi-vowel in the former and the fully vocalic u of the latter is simply the expected allophonic variation of the phoneme u in pre-vocalic and final position, which is to say they are essentially the same sound. The final o of *h1ék̑u̯o- is simply the nominal suffix (o/e) used to form o-stem nouns, as in both the Greek and the Latin second declension. The principal difference evident in these two words, then, is in the central vowels, yet when one recalls that ablaut of e and o was a fundamental element of PIE morphology, even this distinction appears to be of superficial, derivational, significance. Therefore, the first of this pair of words is, probably, a suffixed noun of the same root of which the second word is a lengthened o-grade adjective. They seem to be a noun and adjective pair deriving from a common root *h1ek̑-, and sharing the same base meaning, ‘swift’. [11] These words would, therefore, form a figura etymologica, translatable as something like ‘swift swifties,’ akin to Greek, ἔπος ἐπεῖν, Vedic ávocāma vácaḥ, and Avestan uxðā vacå, all descending from the PIE *u̯eku̯os u̯eku̯, “to speak a speech.” [12] Unlike those figurae etymologicae, the “swift horses” formula does not maintain its transparently etymological quality in the later traditions, but in the PIE phase of the language the phrase would have exhibited this pronouncedly enough to have been readily recognized by its hearers. [13] The distinctiveness of this highly specialized poetic feature along with the demonstrated artfulness of phonetic arrangement makes it highly likely that this phrase is the ancestor of Homeric ὠκέες ἵπποι.

“Swift Horses” in Greek

The fact that this PIE phrase is so unlike ὠκέες ἵπποι is paradoxically reaffirming. For ὠκέες ἵπποι is relatively devoid of obviously poetic qualities, especially in comparison to its cognates, Avestan āsauuō aspåŋhō and Vedic āśavas aśvās, despite the fact that it is more frequently utilized. One would, in fact, expect a reconstruction of its linguistically earlier forms to restore to the phrase some earlier poetic features. This sort of reconstruction is routinely utilized on a smaller scale regarding Homeric phrases with original digamma, such as ὀλίγον γόνυ γουνός, reconstructed as ὀλίγον γόνυ γoνϝός so as to reveal an earlier and much tidier euphonic sequence (Iliad XI 547). Similarly we find ὠκέες ἵπποι shortly after its loss of the digamma in ὠκέες (< ōkéu̯es), and the conversion of the labiovelar, the qu-sound, in ἵπποι (< híku̯oi), which would have euphonically patterned with both the velar k-sound of ὠκέες and the labial semi-vowel sound of the lost digamma, and sounded something like ὠκέϝες ἵκϝοι. [14] Phonically then there is good reason for this pair of words to have maintained its close poetic relationship until a period only shortly before our Homeric verse was recorded, since the comparatively unpoetic appearance of ὠκέες ἵπποι would appear to be a very recent development.
However, even though the phrase had only recently changed at the period of the recording of our Homeric texts, it had changed nevertheless. [15] The fact that this phrasing was not finally lost at a point after the occurrence of these linguistic shifts and just before the period that was to be preserved in our textual tradition requires explanation. For just like historical and dialectical word forms, oral formulas should be employed only so long as they continue to be useful, which may indeed be a very long time depending on how each is individually affected by diachronic change. Poetic formulas then should not be expected to have been preserved simply for the sake of tradition, as oral poets are neither rigid conservators of poetic forms nor daring innovators, but rather practical craftsmen. They continue to use the words, formulas, and techniques which they inherit so long as a use can be found for these elements within the confines of their contemporary poetic systems, even if this is a new use. Therefore the poetic artfulness responsible for the early development of the poetic treatment of this phrase is only a partial explanation of its prevalence and persistence in Homer. It explains why the Greek oral poets would have inherited the phrase initially, but not why they would have preserved it and continued to use it after the occurrence of the linguistic shifts that permanently altered its poetic qualities. To understand this, one must examine the applications of the phrase in early epic itself.
The phrase occurs declined into nearly every case. The frequency of each case is charted below in Table 1.1. Although Iliadic uses dominate, and will receive most consideration here, I list all occurrences in Homer and Hesiod simply to provide a good impression of the phrase’s proliferation throughout early Greek poetry. The symbol / indicates a line end, and ellipses mark wherever other words intervene between the two under consideration.
Table 1.1. Frequency of ὠκέες ἵπποι formulas
Phrase (basic) No. of Occurrences in Iliad No. of Occurrences in Odyssey No. of Occurrences in the Homeric Hymns No. of Occurrences in Hesiod
ὠκέες ἵπποι / 10 1 1
ὠκέες…ἵπποι / 1 1
ὠκέας ἵππους / 18 2 2
ἵππους / ὠκέας 2
ὠκειάων /
2 1
ὠκειάων /
This formula system then seems to exhibit very regular behavior, as may not be surprising for one of such common use. The declensional distribution is in the normal pattern, as outlined by Hainsworth, in that the accusative is the most common, followed by the nominative, and then the genitive. [16] The dative and vocative forms are unknown, and the nominative and accusative formulas are the only ones that appear with any frequency. These two most common formulas also happen to be of the shape ˉ ˘ ˘ ˉ ˉ and occur always after the diaeresis and run to the verse end, filling up the so-called adonean clausula. These forms then do make up something akin to formulas in the very rigid sense, much like Parry’s noun-epithet formulas, and thus belong to a rather small network of such heavily regularized expressions. [17] Within this network, however, this phrase is not at all unusual, since most such phrases occupy this exact position. There is a strong incentive for the creation of rigid formulas at the end of a poetic verse, especially in dactylic hexameter. Unlike the rest of the verse, in which permissible resolution in the arsis allows the occurrence of either a long syllable or two short syllables, the sixth foot of a line of dactylic hexameter is formally bound to be spondaic. The fifth foot is slightly freer, in that resolution is still technically possible, yet so bound in practice that fifth-foot spondees are rare. Thus phrases of this metrical type, ˉ ˘ ˘ ˉ ˉ , were extremely prone to this sort of regularizing, so long as their sense was useful. The phrase ὠκέες ἵπποι, given its especially desirable metrical characteristics and semantic suitability to heroic verse, was particularly fit for this sort of regularizing, as unusual as this development may be within the more general practices of oral poetics. [18]
Thus, although the date of the beginning of dactylic hexameter verse, a Greek innovation, is debated, [19] it should nevertheless predate the phonetic evolutions that permanently altered the appearance of this phrase. There is good reason to think that this phrase had attained prominent formulaic status in Greek heroic verse before the alteration of its originally inherited phonetic and poetic character. By the time of the transformation of the characteristics that had once preserved this expression, other uniquely Greek uses adhered to the phrase. I assume that during most of its life it behaved as most poetic expressions do, as words loosely associated by phonetic, metrical, and thematic pressures so as to appear together frequently, but not in any one consistent position or orientation. This was probably true at the point at which dactylic hexameter verse developed among Greek-speakers, after which the idiosyncrasies of this new meter occasioned a tendency for this phrase to appear at line end and in one particular order, thus elevating it to the realm of truly regular formulas. The prominent position to which this expression had risen may have been sufficiently powerful to contribute to its preservation in the Greek oral poet’s vocabulary despite changes in its phonetic nature. It is interesting to note that this phrase’s Sanskrit and Avestan counterparts, although still alliterative, are not as regularized in position. The deployment of the phrase in Homer is anomalous, but it may be this very anomaly that best testifies to the utility of the phrase and best explains its proliferation and longevity there.
Beyond a phrase’s linguistic and metrical assets, however, may lie more literary qualities which also contribute to its proliferation, as is especially true here. For even a very metrically useful formula should not be expected to have been preserved solely for the sake of occupying one position, but may have developed uses dependent on cultural and artistic resonance, and one of these may actually be particular to highly recognizable formulaic units such as this. Once even limited regularity of deployment develops, this may itself propel the formula into genuine prominence, necessarily imbuing the phrase with new literary potential, since these rigidly bound formulas acquire allusive capabilities. Such development is clearly present here, as an overview of the deployment of the nominative case alone illustrates. The nominative formula occurs ten times in the Iliad, and two of these appear in identical lines in book XVI, at line 383 and 866: ἵετο γὰρ βαλέειν· τὸν δ’ ἔκφερον ὠκέες ἵπποι, “for he longed to strike him, but his swift horses bore him away.” [20] Although these lines are not in extreme proximity to each other, they are close enough to have featured in the same performance and so might seem like a purely mechanical result of oral composition—in other words, this line was fresh enough in the poet’s mind that he simply reused it to provide himself time to plan the next lines. Yet the second occurrence does more than simply fill space, representing as it does the deliberate deployment of a poetic device. The first of the two lines occurs at the beginning of Patroclus’ aristeia at the moment when, carried away by his seemingly superhuman, nearly Achillean, prowess, he desires to kill Hector and storm Troy himself. The second appearance occurs just after Patroclus has died, as Hector, finished boasting over the corpse, tries to catch Automedon, who has taken the horses of Achilles from Patroclus. The line then is the same, but the referents have changed: 383—“for he [Patroclus] longed to strike him [Hector], but his [Hector’s] swift horses carried him [Hector] away”; 866—“for he [Hector] longed to strike him [Automedon], but his [Automedon’s] swift horses carried him [Automedon] away.” Hector has transformed from pursued to pursuer, has taken the place of Patroclus as doomed hero, now tragically chasing his own death.
It is also important that the first use of this formula refers to Hector’s horses, but recalls the verses about Achilles’ horses that precede it:
…ὠκέες ἵπποι
ἄμβροτοι, οὓς Πηλῆϊ θεοὶ δόσαν ἀγλαὰ δῶρα.
…the swift horses, the immortal ones which the gods gave to Peleus as glorious gifts.
Iliad XVI 380–381
The verses that follow this are repeated after the final occurrence of the formula as well, and in fact end the book:
ἵετο γὰρ βαλέειν· τὸν δ’ ἔκφερον ὠκέες ἵπποι
ἄμβροτοι, οὓς Πηλῆϊ θεοὶ δόσαν ἀγλαὰ δῶρα
for he longed to strike him, but his swift horses carried him away, the immortal ones which the gods gave to Peleus as glorious gifts
Iliad XVI 866–867
Although the formula itself, ὠκέες ἵπποι, remains the same, the greater semantic units that it anchors have been rearranged to highlight the dramatic shift that has occurred in the narrative. The irony of the fact that the immortal horses of Achilles could not catch the horses of Hector has given way to the sudden impotence of Hector and the ineluctability of his impending fate, for with the death of Patroclus the dramatic action of the Iliad switches to Hector’s own death.
This formula, furthermore, has a particular association with Hector. Of its ten occurrences in the Iliad, four have to do with him, and the only repeated extended formula based on the nominative ὠκέες ἵπποι is Ἕκτορος ὠκέες ἵπ- ποι (VIII 88, XVI 833). In addition, a full half of the occurrences of the ὠκέες ἵπποι formula appear in book XVI. Book XVI develops this association between Hector and the formula ὠκέες ἵπποι, and manipulates its contextual significance in service of the work’s greater narrative structure.
One final point to be considered is that the second occurrence of this extended formula immediately follows the lines αὐτίκα δὲ ξὺν δουρὶ μετ’ Αὐτομέδοντα βεβήκει / ἀντίθεον θεράποντα ποδώκεος Αἰακίδαο, “then with his spear he immediately pursued Automedon, the godly servant of the swift-footed [Achilles] son of Aeacus” (XVI 864–865). Achilles, the figure whose impending return has informed this entire formulaic discourse, is finally linked to this formula explicitly. For although Achilles himself is not yet present, his formulaic epithet is joined to the ὠκέες ἵπποι formula by their common element, the adjective ὠκύς, “swift,” which appears in both of the epithets most common for Achilles: as the second element of the compound ποδώκης, “swift-footed,” and as an independent adjective in πόδας ὠκύς, the uncompounded form of the same phrase. The valor of horses connotes the valor of the horses’ heroic owner, [21] and it is no coincidence that the salient feature of the formula designating valorous horses finds its second most frequent expression in formulas attached to Achilles. The changing role of Hector has been dramatized by the changing semantics of the ὠκέες ἵπποι formula, and now, as this formula’s newest application foreshadows Hector’s impending demise, it echoes the Achillean epithet so as to foreground the epic’s ultimate hero, whose preeminence will occupy the remainder of the work. Book XVI consistently reinforces the association of Hector with this “swift horses” formula but ultimately reminds the audience that Hector’s heroism will soon give way to that of Achilles. The frequency and prominent position of this formula render it more exploitable in the intersections of formulaic and narrative technique, thus providing a new mechanism by which the phrase maintained and even expanded its position in Greek epic verse.
The extent of this application, however, is not limited to strict reiterations of the formula, but also encompasses modified formulas, the significance of which is influenced by the frequent and maintained use of the base formula. Modification of the nominative, ὠκέες ἵπποι, occurs once in Homer and once in Hesiod, and does not initially appear to be a modification of this formula at all, but of the secondary formula ἵπποι…ὠκύποδες, “swift-footed horses.” The Hesiodic use of this formula may indeed be treated as a modification of this derivative formula system, [22] but the Homeric occurrence is best understood within the ὠκέες ἵπποι system and may indeed serve as a useful link between these two. The Homeric formula occurs in the chariot race of book XXIII: φαίνετ’, ἄφαρ δ’ ἵπποισι τάθη δρόμος· ὦκα δ’ ἔπειτα / αἳ Φηρητιάδαο ποδώκεες ἔκφερον ἵπποι, “suddenly the pace of the horses was stretched to the utmost and then quickly the swift-footed mares of the son of Pheres bore him off” (XXIII 375–376). Although we have the compound ποδώκεες rather than the familiar ὠκέες, when the line is considered from only the fourth foot onward, it will be noted that -ώκεες ἔκφερον ἵπποι is simply a reversal of the first two words of the extended formula already mentioned: ἔκφερον ὠκέες ἵπποι. It will also be noticed that there occurs, in the line immediately preceding this, the word ὦκα, “swiftly,” in precisely the same metrical position that would normally be occupied by ὠκέες. Thus, a sufficient number of the expected elements of the ὠκέες ἵπποι formula is present for these lines to be evaluated as a reorganized expression of this basic formula.
An explanation for this unusual structuring may, at least initially, be sought within the mechanics of oral verse composition. This line’s first four feet, which belong strictly to the previous sentence, involve the word ἵπποισι, “horses” in the dative plural. It would have been irregular to fill the last two feet of the verse with the normal nominative formula because that would have caused unusual repetition of the noun. Yet the formula, being in the mind of the poet, may still have provided a model for the last two feet of the verse, as indicated by the preservation of the word ὠκύς in adverbial form and perhaps by the phonetic resemblance in the replication of the unvoiced labial stop, the pi, of ἵπποι in ἔπειτα. Thus, one of the vocabulary elements of the base formula is preserved in addition to the formula’s signature phonetic sequence, and with this formula so phonetically present in the last two feet of line 375, the poet may have been disinclined to use it in the same position so soon afterward in line 376. Instead, the poet created a seemingly unique formulation based on the models presented by these two other derivative formulas, that of the Hector and Patroclus episode and the ποδώκεες extensiἱons (discussed below) to convey the essential idea. [23] Thus, this formulaic deviation may paradoxically be conditioned by the regularity and commonality of the basic formula. [24]
Yet this modification also highlights a shift in the formula’s semantic import, as becomes clear when considered in the light of the following line: αἱ Φηρητιάδαο ποδώκεες ἔκφερον ἵπποι / τὰς δὲ μετ’ ἐξέφερον Διομήδεος ἄρσενες ἵπποι, “[suddenly the pace of the horses was stretched to the utmost and then quickly] the swift-footed mares of the son of Pheres bore him off, but the stallions of Diomedes bore him on after them” (XXIII 376–377). [25] The poet seems at this moment to be very concerned with highlighting the distinction in gender between the two groups of horses. In all of Homer, the definite article is used with the “swift horse” formulas only this one time. It is important to note that Homer regularly uses masculine forms of ὠκύς in these expressions, even when female horses are involved. Genitive plural forms are an exception and will be discussed shortly. There is also no specific word for “mare” in Greek (the word ἵππος is simply accompanied by a feminine form of the definite article, or the phrase θῆλυς ἵππος, “female horse,” may be used [26] ). It is therefore difficult for the poet to specify that the horses under discussion are either male or female, and it is apparently rarely of concern to do so. This unique definite article, however, does distinguish the horses as female, and its prominent positioning does so quite dramatically. These particular horses, in fact, are gendered not only here but in their first mention in the catalogue of ships, when they are modified by the feminine adjective ἄρισται (II 763). These horses are apparently noteworthy in their gender, for reasons that I will discuss below, and the poet is being careful to highlight this fact. The formula ὠκέες ἵπποι did not need to be divided in order to include a definite article, but the marked innovation in such a standard formula draws special attention and therefore serves to spotlight the poet’s effort. [27] Finally, it is worth noting that this gendering of the horses may even be employed to occasion something of a sexual pun in the next two lines that would be dependent on a double entendre for the verb ἐπιβαίνω: Τρώϊοι, οὐδέ τι πολλὸν ἄνευθ’ ἔσαν, ἀλλὰ μάλ’ ἐγγύς· / αἰεὶ γὰρ δίφρου ἐπιβησομένοισιν ἐΐκτην, “those [male horses], the ones of Trojan birth, were not far behind, but very close, for they always seemed about to mount the chariot” (XXIII 378–379). [28] The potential for nonformulaic horse expressions to derive meaning from formulaic expressions testifies to the fact that ὠκέες ἵπποι, despite its change in phonetic character, attained a literary status which developed from, and reciprocally maintained, its formulaic prominence.
I would like to move now from the usage of the nominative forms of this phrase to an analysis of the usage of the genitive and accusative. I do this in part for the sake of thoroughness, to offer the reader a full survey of the Homeric artfulness represented herein. I also hope that in the course of this, however, I will build an impression of a marked tendency for rather extreme phonetic patterning in and around these expressions, which will be relevant in analysis of other, related equine formulas at the end of this chapter.
In moving to the genitive formula, ἵππων ὠκειάων, we see another example of explicit equine gender, and although this version is relatively rare, it is still quite fascinating in its formulaic behavior. It occurs only twice in unmodified form, but presents a unique deviation from the other formulas in that it is reversed in word order, and fills three feet, an entire half of a line. One possible explanation for the reversal in word order may be that in this particular version the phrase takes up more space than it does in its other cases: it extends beyond the diaeresis that generally defines these formulas, and which may be the principal force determining the shape of most verse-end formulas. Although both the adjective-noun possibility and noun-adjective would scan identically, and would occupy the same space in the line, the adjective-noun formula would have obliterated this diaeresis, while the noun-adjective leaves it intact.
This conclusion seems to be confirmed by another formula entirely. The only other occurrence of the word ὠκειάων in Homer is in the formula νηῶν…ὠκειάων, “of the swift ships,” which appears twice: νηῶν ἐπιβησέμεν ὠκειάων (Iliad VIII 197); νηῶν ἐπιβαινέμεν ὠκειάων (Odyssey ix 101). These formulas, νηῶν ὠκειάων and ἵππων ὠκειάων, would have scanned identically, and in both examples the adjective, ὠκειάων, occurs at verse end. The conventions of diaeresis may have rendered the formulas ὠκειάων ἵππων and ὠκειάων νηῶν equally undesirable and given rise to the word order in both cases. It is also noteworthy that these, again, are explicitly female horses. Yet, unlike the previously mentioned female horses, the gender of these particular horses does not seem to have special significance. In this case, the feminine form of the adjective seems to have been preferred because the usual masculine form would not have scanned in dactylic hexameter.
Finally, just like the nominative, the genitive formulas also exhibit a modification by separation, although only in one instance: Δεξιάδην ἵππων ἐπιάλμενον ὠκειάων “[he threw his spear at] the son of Dexias as he jumped upon his swift mares” (VII 15). The purpose for this deviation from formula does not seem to be the highlighting of any narratological oddity, nor does any metrical consideration necessitate the separation of the adjective and noun, as Δεξιάδην ἐπιάλμενον ἵππων ὠκειάων would scan perfectly well. Perhaps this modification was made instead to reflect the dramatic action of the line by presenting words that seemed to be jumping from their place just as the character involved jumps upon his horses. Thus end the rare genitive forms of the “swift horses” phrase.
The most interesting phenomena, however, belong to the accusative formula, ὠκέας ἵππους, and these will lead us to an analysis of these formulas’ poetics in general. The accusative phrase generally occupies the same position in its verse as the nominative, but appears twice with modification by verse-end separation: [29]
ἐξῆγεν πολέμοιο δυσηχέος, ὄφρ’ ἵκεθ’ ἵππους
ὠκέας, οἵ οἱ ὄπισθε μάχης ἠδὲ πτολέμοιο
ἕστασαν ἡνίοχόν τε καὶ ἅρματα ποικίλ’ ἔχοντες·
[Polites] led him away from the grueling war, until he came to the swift horses which awaited him away from the battle and war, those which had a charioteer and chariot beautifully wrought.
Iliad XIII 535–537
χερσὶν ἀείραντες φέρον ἐκ πόνου, ὄφρ’ ἵκεθ’ ἵππους
ὠκέας, οἵ οἱ ὄπισθε μάχης ἠδὲ πτολέμοιο
ἕστασαν ἡνίοχόν τε καὶ ἅρματα ποικίλ’ ἔχοντες·
[his comrades] lifting him in their arms, bore him away from the struggle, until he came to the swift horses which awaited him away from the battle and war, those which had a charioteer and chariot beautifully wrought.
Iliad XIV 429–431
The two appearances of this modification introduce the same repeated lines. Similar examples of runover epithets have been studied by Bassett, [30] who showed that they are used to form a bond between one idea and another, preceding one. I suggest that this particular example demonstrates a further purpose. These lines describe two separate, but very similar scenarios: both depict wounded heroes being helped to their horses by a comrade in order to flee from a fight which they cannot win. The fact that the person aided to his horse in the second example is Hector (the first is Deiphobus) also demonstrates a continuity of application in that it parallels the allusive manipulation of the nominative phrase, used specifically to dramatize Hector’s dramatic arc. It seems that the expected formula shape may have been avoided in order to draw attention to the unusual ignominy of the act depicted and to provide a device by which these two scenes could be linked. The deviation from the expected formula shape draws attention to the unusual event being described, and this modified formula is then exploited for its own allusive potential, which is nevertheless still dependent on the unique status of the base formula.
The preservation of this formula despite fundamental alterations to its phonetic structure is understandable on literary as well as metrical grounds. Yet the loss of certain sorts of phonetic artistry ought, in common formulas, to give rise to new varieties, which reinforce the formula’s newfound prominence. Further investigation into the deployment of the accusative case of the ὠκέες ἵπποι expression demonstrates that such new phonetic associations did indeed develop. The accusative case, ὠκέας ἵππους, is even more heavily regularized than the nominative, so is a particularly apt subject for this sort of investigation. It is, in fact, so regularized that even three of its extensions attain formulaic status themselves: ἔχον ὠκέας ἵππους; ὑφ’ ἅρμασιν ὠκέας ἵππους; ἐλαύνομεν ὠκέας ἵππους. A truly unusual regularization of the phonetic patterns around the phrase, however, shows the persistence of the dental nasal, the nu, immediately preceding the formula in nearly every occurrence:
Iliad III 263: τὼ δὲ διὰ Σκαιῶν πεδίον δ’ ἔχον ὠκέας ἵππους
Iliad V 240: ἐμμεμαῶτ’ ἐπὶ Τυδεΐδῃ ἔχον ὠκέας ἵππους
Iliad V 261: ἀμφοτέρω κτεῖναι, σὺ δὲ τούσδε μὲν ὠκέας ἵππους
Iliad V 275: τὼ δὲ τάχ’ ἐγγύθεν ἦλθον ἐλαύνοντ’ ὠκέας ἵππους
Iliad VIII 254: εὔξατο Τυδεΐδαο πάρος σχέμεν ὠκέας ἵππους
Iliad VIII 402: γυιώσω μέν σφωϊν ὑφ’ ἅρμασιν ὠκέας ἵππους
Iliad VIII 416: γυιώσειν μὲν σφῶϊν ὑφ’ ἅρμασιν ὠκέας ἵππους
Iliad X 527: ἔνθ’ Ὀδυσεὺς μὲν ἔρυξε Διῒ φίλος ὠκέας ἵππους
Iliad XI 127: εἰν ἑνὶ δίφρῳ ἐόντας, ὁμοῦ δ’ ἔχον ὠκέας ἵππους
Iliad XI 760: ἂψ ἀπὸ Βουπρασίοιο Πύλονδ’ ἔχον ὠκέας ἵππους
Iliad XII 62: ἀφραδέως διὰ τάφρον ἐλαύνομεν ὠκέας ἵππους
Iliad XV 259: νηυσὶν ἔπι γλαφυρῇσιν ἐλαυνέμεν ὠκέας ἵππους
Iliad XVI 148: τῷ δὲ καὶ Αὐτομέδων ὕπαγε ζυγὸν ὠκέας ἵππους
Iliad XVII 465: ἔγχει ἐφορμᾶσθαι καὶ ἐπίσχειν ὠκέας ἵππους
Iliad XVIII 244: χωρήσαντες ἔλυσαν ὑφ’ ἅρμασιν ὠκέας ἵππους
Iliad XXIII 294: διογενής, ὑπὸ δὲ ζυγὸν ἤγαγεν ὠκέας ἵππους
Iliad XXIII 516: ἀλλὰ καὶ ὧς Μενέλαος ἔχ’ ἐγγύθεν ὠκέας ἵππους
Iliad XXIV 14: ἀλλ’ ὅ γ’ ἐπεὶ ζεύξειεν ὑφ’ ἅρμασιν ὠκέας ἵππους
Odyssey iii 478: καρπαλίμως δ’ ἔζευξαν ὑφ’ ἅρμασιν ὠκέας ἵππους
Odyssey iv 28: ἀλλ’ εἴπ’, ἤ σφωϊν καταλύσομεν ὠκέας ἵππους
The only two exceptions occur at V 275 and X 527, and line V 275 can readily be accounted for as a modification of ἐλαύνομεν / ἐλαυνέμεν type of formula seen in XII 62 and XV 259. The anomaly of line X 527 occurs within the much-disputed Doloneia, which is notorious for its unexpected diction and syntax, so does not detract strongly from the appearance of a genuine poetic habit. [31]
Thus we have seen the formula expand to encompass not only the fundamental semantic unit, the basic accusative formula, but also to influence elements that are phonically regular, but semantically unrestricted. Furthermore, the regular extended formula, ἔχον ὠκέας ἵππους, involves additional phonetic attraction in that the phase is always preceded by a voiced dental, a delta: δ’ ἔχον ὠκέας ἵππους (III 263; XI 127); Πύλονδ’ ἔχον ὠκέας ἵππους (XI 760). A slightly altered version of this also appears in Τυδεΐδῃ ἔχον ὠκέας ἵππους (V 240). Although extra- formulaic features may be of obvious utility in helping a poet structure a line which he knows will end with a common formula, it is unclear at the moment why these particular examples were so regularized. The delta and the nu are homo-organic in articulation, so there may have been a quasi-alliterative quality to this composition, and the nu, coming as it does before an initial vowel, may have its origin in a moveable nu that was extended to wide distribution through the mnemonic demands of the craft and performers’ own poetic sensibilities.
Table 1.2. Frequency of related dual formulas
Formula Iliad Odyssey Homeric Hymns Hesiod
ταχέ’ ἵππω 2
χαλκόποδ’ ἵππω 2
In addition to the extended phonetic artistry of the formula there is extra-formulaic metrical consistency. The foot preceding ὠκέας ἵππους is dactylic in all but two instances: the already irregular V 275, in which the use of the dual may help explain the exception, and XVII 465. There is a nonessential but broadly consistent extra-formulaic metrical component to the deployment of this formula. Finally, although the Homeric and Hesiodic treatments of horse formulas often differ slightly, it is interesting—if admittedly potentially coincidental—that these two practices are also preserved in Hesiod: ἡνίοχοι βεβαῶτες ἐφίεσαν ὠκέας ἵππους and Κύκνε πέπον, τί νυ νῶιν ἐπίσχετον ὠκέας ἵππους (Shield 307; 350). Although these phonetic patterns may appear trivial, they seem to reflect an unusually strong tendency toward phonetic patterning in and around these formulas, perhaps influenced by the sheer frequency and regularity of their usage.
To see the true extent and significance of this patterning, however, we must now turn to dual formulas. The phrase ὠκέες ἵπποι does not actually occur outside of the plural. The absence of the singular is interesting but not completely surprising, since horses most frequently appear in teams and only rarely is an individual horse called to the audience’s attention. The absence of duals is, however, more surprising since chariots could indeed be pulled by a pair of horses, and pairs of horses are mentioned as normal in Homer. [32] Isolated dual formulas, semantically related to the ὠκέες ἵπποι formulas, do occur (see Table 1.2).
Table 1.3. Frequency of related plural formulas
Formula Iliad Odyssey Homeric Hymns Hesiod
μώνυχες ἵπποι / 8
μώνυχας ἵππους / 25 1 1
καλλίτριχες ἵπποι / 3
καλλίτριχας ἵππους / 8 3 1 1
καλλίτριχε…ἵππω / 1
χρυσάμπυκας ἵππους / 1
χρυσάμπυκας…ἵππους / 3
The expression ταχέ’ ἵππω, “fast horses,” is essentially synonymous with ὠκέες ἵπποι, and χαλκόποδ’ ἵππω, “bronze-footed horses,” although not synonymous, does describe the horses’ feet, calling to mind the formula πόδας ὠκὺς Ἀχιλλεύς, “swift-footed Achilles,” as well as κύνες πόδας ἀργοὶ ἕποντο, “swift-footed dogs followed,” so both may be considered related to ὠκέες ἵπποι. [33] The most interesting element of these substitutions is their phonic correspondence with the principal expression, in their preservation of the unvoiced velar, the kappa, of ὠκύς, once with the corresponding aspirated counterpart, a khi, and once as the identical unaspirated velar of ὠκύς in echo of the initial aspirate. Although ὠκύς itself does not appear in the dual, maybe a simple fault of our sources, these formulas clearly reflect it phonetically.
The assertion that these dual formulas, ταχέ’ ἵππω, and χαλκόποδ’ ἵππω, are meaningful phonic responses to ὠκέες ἵπποι may seem untenable, based as it is on only one corresponding consonant. There are several other formulas that do not as fully accord with ὠκέες ἵπποι semantically, but are nevertheless extremely useful for comparison (see Table 1.3). [34]
These formulas, combined with ὠκέες ἵπποι and its inflected counterparts, comprise a list of the most common formulas involving horses in Homer, and every one of them displays an unvoiced velar, of either the kappa or the khi variety, at the beginning of the syllable preceding ἵπποι, just as the dual formula ταχέ’ ἵππω does, and indeed they often precede that velar with another at the beginning of the word, just as the dual formula χαλκόποδ’ ἵππω does. Admittedly, the formula χαλκόποδ’ ἵππω does not exhibit the velar in precisely the same position as the other formulas, but, in light of the other evidence, it may be considered to belong to the same nexus of phonic and semantic formulation. It seems then that the network of Homeric horse-formulas is influenced by the association between the word ἵππος and the unvoiced velar, for which the most obvious explanation is the prominence of the word ὠκύς in the formulaic system. We then have a phrase that is not simply a common element in epic diction independently, but that is part of an entire network of phonetically and semantically related phrases.
The fact that these unvoiced velars seem to serve as the obligatory counterpart to ἵπποι may descend from the consonantal phonetics of the PIE expression itself, *h1ōk̑éu̯-es h1ék̑u̯-ōs. Although the phrase ὠκέες ἵπποι is sufficiently widespread to justify the assumption of a purely synchronically determined phonetic pattern, the fact that, prior to the recent loss of the labiovelar, the phrase had itself contained two closely positioned velar sounds raises a tantalizing prospect: that the strength of the association of velar sounds with this expression was conditioned by the expression’s earlier alliterative character, which manifested a continuity of influence. If this is true, the phonetic quality of the phrase is not as far removed from its earlier linguistic phases as it initially appears. For although the expression itself does not reflect its earlier character well, the broader phonetic network in which it operates does. This would represent a continuity of phonetic treatment dating to the period when this phrase took part in the freer and more dynamic mechanics of oral composition, existing alongside the novel phonetic attractions arising after its later prominence in a consistent position. Early Greek verse does not, of course, exhibit euphonic patterning as prominently as some other IE poetries, and certainly not as prominently as some early Germanic verse or even Vedic. It is, however, reasonable to expect to see some examples of it nevertheless, perhaps especially in and around its oldest expressions, as I believe occurs here. A recognition of this may ultimately be key to seeing the relationship of these expressions to several other, less obviously related formulas.

The Formulaic Significance of Hades κλυτόπωλος

An application of the foregoing formulaic analysis to an unusual epithet of Hades will showcase just how analytically far-reaching an appreciation of euphonetic patterning can be. The adjective κλυτόπωλος (“of famous foals”) appears five times in early Greek poetry, thrice in the Iliad as an epithet of Hades, once in the fragmentary Hesiodic Catalogue in reference to Ion, and later in a Pindaric fragment, where it refers to Poseidon. [35] The application of the term to Poseidon, the god of horses, is no surprise, and Ion, although not a uniquely famous horseman, is a Greek hero for whom excellence in horsemanship in expected. It is the application of the term to Hades that has confused scholars for quite some time, but I think that the emphasis on its mythological connection to Hades in the Iliad has precluded research into its function in oral verse. The origin of κλυτόπωλος is linked to most of the formulas outlined above, formulas that all convey the idea of “good horses.” Recognizing the position of κλυτόπωλος within this formulaic network helps us to chart the diachronic evolution of this network as a whole.
The formulas described above contain an unvoiced velar in the word pre-ceding ἵπποι, and thus contribute to an overall sense of phonetic similarity among these expressions, at the level of a basic kp consonantal sequence. The semantic similarity is even easier to spot, as these particular expressions function quasi-synonymously, each having the core semantic value of ‘good horses’. [36] The particularities of κλυτόπωλος will be discussed momentarily, but within this broad group most describe the horses’ quality by focusing on their speed, either directly, such as ὠκέες ἵπποι and ταχέ’ ἵππω, or through metonymy, such as χαλκόποδ’ ἵππω and μώνυχες ἵπποι. καλλίτριχες ἵπποι and χρυσάμπυκες ἵπποι describe the horses’ quality through their beauty. Rarely are these differences particularly narratologically significant. Despite the particular honorific attributes highlighted by any one expression, they all fundamentally indicate ‘good horses.’
The consistency of the semantic, phonetic, and metrical quality exhibited by these expressions is striking and should not be dismissed as sheer coincidence. I suggest, in fact, that these formulas constitute an especially rich example of a genetically related family of formulas, similar to those studied by Nagler. [37] In such a family new expressions or formulas can be generated as something like varied allomorphic realizations of a stable underlying nexus of metrical, semantic, and phonetic characteristics. In the process of oral performance phonetically similar expressions may cluster around a unique theme, especially in common metrical positions. In this case that would mean that as a poet reached line end, the position where each of these formulas is most common, if the poet planned to express the idea of ‘good horses,’ a variety of different formulas may have been employed, either by generation or recollection, with the aid of a persistent underlying phonetic structure. [38]
κλυτόπωλος, differs from these expressions because it is not an adjective-noun sequence but instead a singular bahuvrīhi-type compound adjective, i.e. it identifies a possessor of good horses rather than the horses themselves. It nevertheless exists in the same basic semantic sphere as an expression that conveys the semantic notion of ‘good horses’. [39] It also resembles these formulas at the phonetic level. In this case πῶλος (foal) appears rather than ἵππος, and κλυτός (famous) appears rather than one of the various adjectives already described, but a similar structure, anchored by corresponding k...p sounds is still evident. Finally, in hexametric verse this word occurs only at line end, the most frequent position for the other ‘good horses’ expressions. κλυτόπωλος then resembles the other members of this family at the semantic level, at the phonetic level, and at the metrical level, so satisfies the criteria for inclusion in this network. If, in the course of a performance, a poet reached line end and wished to express the core semantic idea of ‘good horses’ in a way that describes an individual who has good horses, rather than the good horses themselves, he could have generated or employed this expression by drawing on the same phonetic structure on which his stock of ‘good horses’ formulas was built.
It should be noted that the word’s appearance in the Hesiodic Catalogue perfectly reflects its treatment in Homer, occurring in the same position as in the Homeric texts:
ἥ οι Ἀ]χαιὸν ἐγ ἐγ̣[είνατ’ Ἰάονά τε κλυ]τ̣ό̣π̣ω̣λ[ο]ν
μιχθ]εῖσ’ ἐν̣ [φιλότητι καὶ εὐε]ι̣δέα Διομήδην
who bore to him Akhaios, and Ion of famous foals, and glorious Diomedes, having mingled in love
Solmsen frag. 10a23–24
The only other early usage occurs in Pindar, and is the only one that deviates from this pattern:
Ζηνὸς υἱοὶ καὶ κλυτοπώλου Ποσειδάωνος
the sons of Zeus and Poseidon of famous foals
Maehler frag. 243.2
This is, however, a metrical outlier, occurring in non-hexametric verse, and Pindaric poetry was presumably composed with the aid of writing, so this particular example does not need to have an origin in oral verse mechanics. It is not uncommon, however, for Pindar’s compositions to employ ancient phraseologies or to display vestiges of older technique. [40] In any case the Pindaric usage does nothing to obscure the character of this term or related terms in earlier verse. [41]
To gain an understanding of the full significance of this term’s relationship to the horse formulas discussed above, a look at other uses of the word πῶλος, “foal,” in the epics is instructive. The word is used relatively seldom, at least in comparison to ἵπποι, “horses,” and it usually occurs in compounds. The very few examples of πῶλος occurring in uncompounded form appear in the immediate vicinity of the word ἵππος and may be employed there to avoid repetition. In its more common compound forms it usually occurs at verse end, often in the formula Δαναῶν ταχυπώλων, “of the Danaans of swift foals.” This appears ten times in the epics, and it too exhibits the same phonetic character as the broader family under discussion. πῶλος also appears twice in the name Ἐχέπωλος, “the possessor of foals,” which also occurs at line end and also demonstrates the same phonetic character.
Πρῶτος δ’ Ἀντίλοχος Τρώων ἕλεν ἄνδρα κορυστὴν
ἐσθλὸν ἐνὶ προμάχοισι Θαλυσιάδην Ἐχέπωλον·
Antilokhos first slew one of the helmeted Trojans among the forefighters, noble Ekhepolos, son of Thalusios.
Iliad IV 457–458
Ἴλιον εἰς εὔπωλον, “toward Ilium of the good foals,” is the only common usage not accounted for by these phonetic and metrical constraints.
The recognition of πῶλος within this network is especially exciting because it helps us to chart the development of the network with unusual precision by allowing us to identify an early and a late phase. *h1ōk̑éu̯-es h1ék̑u̯-ōs surely proliferated in Greek before the period documented in our texts and, most importantly, before the significant phonetic developments discussed above, specifically, the conversion of Greek labiovelars, the ku̯ sounds, to p in many contexts, including when they were followed by an o as happened here. [42] Again, in the period just before that which our texts document, this formula was certainly in use, but did not sound like ὠκέες ἵπποι, but instead like ὠκέϝες ἵκϝοι, with two unvoiced velar sounds. The organizing phonetic structure then was not, in fact, the k...p sequence that the Homeric formula family documents. The original underlying phonetic structure seems instead to have been an alliterative k...k.
This presents a problem, however, for the two halves of this formula family, the expressions anchored to a final ἵππος unit and those anchored to a final πῶλος unit, like κλυτόπωλος. The etymology of πῶλος is still the subject of speculation, but it must descend from something like *polH-. Thieme has suggested that this comes from the verbal root *ku̯el-, [43] meaning “to roam,” but its perfect cognition with such words as German Fohlen and Gothic fula, and even English “foal” makes that impossible. The unvoiced labiovelar ku̯ sound could indeed become a labial p in Greek but would not have become a fricative f in the Germanic languages. A p, as in *polH-, would yield precisely what we see in the Greek and the Germanic cognates. [44] This means that πῶλος always began with a p sound and that the underlying phonetic structure for the πῶλος formulas was always k...p, rather than the k...k of the ἵππος formulas.
The best explanation for the relationship of these two groups must lie in a diachronic evolution of the phonetic schema of which all these formulas are a realization. This is an evolution that would have occurred in tandem with the changing phonetic character of the Greek language. The ἵππος formulas, or more historically, the ἵκϝος formulas, must represent an earlier phase in the generation of “good horses” expressions. In early oral composition “good horses” expressions must have employed an alliterative k...k pattern, perhaps rooted in the figura etymologica of the “swift swifties” phenomenon. As the phonetic evolution of the Greek language altered the ἵκϝος formulas to ἵππος formulas, the schema upon which all of these expressions were founded altered as well, developing from a k...k structure to a k...p. After phonetic change resulted in this k…p sequence this new scheme became generative itself and thenceforth formulas could be added to the network with a base in ἵππος or πῶλος. This does not mean, of course, that each unique ἵππος formula must antedate each unique πῶλος formula, but instead that the general practice of generating and employing ἵππος formulas must have a start date anterior to the start date of the incorporation of πῶλος formulas.
The Iliadic treatment of κλυτόπωλος then provides excellent evidence for understanding the development and deployment of horse formulas generally in Greek oral poetry. But what of its application to Hades? Although few scholars have looked into this term very deeply, those who have have focused entirely on explaining why Hades would be known as a possessor of famous foals in the first place. Horses do not after all feature very frequently in his mythology. Although my own argument here has not yet dealt with this issue, it has been implicit throughout my reasoning that there was indeed some special significance in the application of the term to Hades. My argument in fact assumes that the Greek oral poets deployed this term for one who possesses good horses on the model of “good horses” formulas precisely because there was an immediate utility to such a term in their performances, and this argument must not conclude, I think, without attempting to identify what that was.
The first step in this process should be an examination of the occurrences of the term themselves:
Sarpedon speaking to Tlepolemus:
σοὶ δ’ ἐγὼ ἐνθάδε φημὶ φόνον καὶ κῆρα μέλαιναν
ἐξ ἐμέθεν τεύξεσθαι, ἐμῷ δ’ ὑπὸ δουρὶ δαμέντα
εὖχος ἐμοὶ δώσειν, ψυχὴν δ’ Ἄϊδι κλυτοπώλῳ.
I declare that slaughter and dark death will be fashioned for you, by my hands and that you, conquered by my spear, will give glory to me and your soul to Hades of famous foals.
Iliad V 652–654
Odysseus speaking to Socus:
σοὶ δ’ ἐγὼ ἐνθάδε φημὶ φόνον καὶ κῆρα μέλαιναν
ἤματι τῷδ’ ἔσσεσθαι, ἐμῷ δ’ ὑπὸ δουρὶ δαμέντα
εὖχος ἐμοὶ δώσειν, ψυχὴν δ’ Ἄϊδι κλυτοπώλῳ.
I declare that there will be slaughter and dark death for you today, and that you, conquered by my spear, will give glory to me and your soul to Hades of famous foals.
Iliad XI 443–445
Meriones speaking to Aeneas:
εἰ καὶ ἐγώ σε βάλοιμι τυχὼν μέσον ὀξέϊ χαλκῷ,
αἶψά κε καὶ κρατερός περ ἐὼν καὶ χερσὶ πεποιθὼς
εὖχος ἐμοὶ δοίης, ψυχὴν δ’ Ἄϊδι κλυτοπώλῳ.
If I should hurl [my spear] at you, and strike your middle with my sharp bronze, although you are strong and trust your hands, you would quickly give glory to me and your soul to Hades of famous foals.
Iliad XVI 623–625
The first thing that one notices is that Ἄϊδι κλυτοπώλῳ is not the only common element among these passages, but that these lines echo each other generally, as if a traditional threat has been adapted three times. This looks then like an element of a type-scene, a narratological unit, larger than any single formula, in which elements tend to cluster in a given type of scene. One may still ask, however, why it is κλυτόπωλος that we find embedded in these important lines rather than one of Hades’ other epithets. There are five such epithets in the Homeric poems: ἴφθιμος (“mighty”), ἀδάμαστος (“unconquerable”), πελώριος (“monstrous” or “huge”), ἀμείλιχος (“implacable”), and πυλάρτης (“gate fastener”). It is important that these are not all strictly interchangeable, however. Each one is metrically unique and served its own function in the process of composition in performance. [45] These six epithets in fact form their own formulaic network, one that overlaps with the network under examination here, and in which κλυτόπωλος plays an important role.
The term κλυτόπωλος is unlikely to have been chosen from this network for metrical reasons alone, and we should still look to these lines for nonmetrical motivations. Since it is not this epithet alone that repeats but the threat’s language generally, the question that we must ask is not just why Hades would have famous foals but why his possession of them would be significant in this particular type of scene, in which this language was apparently resonant enough to become traditional. The answer here must be related to the passages’ striking similarity of context: each occurrence is part of a threat that sudden death will soon befall the target of the speaker’s aggression. What, then, does the epithet κλυτόπωλος have to do with such sudden death? It is with this question in mind that we should turn our attention to the two scholars who have studied this word in depth, Paul Thieme, mentioned earlier, and Martin P. Nilsson. [46] Although neither approached the word in quite this light, each offers very useful suggestions about the application of κλυτόπωλος to Hades.
Thieme’s explanation has roots in the IE underworld and the etymology of the word “Hades” itself, which is complicated. He suggests that the word is comprised of the prefix sṃ- ‘with’ and the verbal root *u̯ei̯d- ‘to see’. Thieme’s etymology would roughly mean ‘seeing together’, and would have a perfect cognate in the particularly loaded Sanskrit term samvedanam, referring to the act of reuniting with one’s forefathers in the afterlife. [47] He suggests that the limited information that we have about the early IE afterlife suggests that this realm was imagined, at least in part, as a meadowland, and therefore may have contained horses. [48] If Thieme is correct, then these lines and the word κλυτόπωλος within them would serve to call to mind the place to which the threatened man may soon be going. This works very well here, because the epithet would not be incidental but specially suited to this particularly narratological circumstance, making sense of its unique occurrence in minatory exchange. Although this could be a sufficient explanation for the expression, it does require that the term be a fossilization, because the belief that there were horses in the underworld is not attested in archaic Greece. This expression would need to have been preserved by poetic habit after its original significance had been forgotten. Although this is possible, I think it is useful to explore the possibility that this phrase had synchronic significance as well, to ask if it meant something special to the archaic poet and audience, admitting that such a meaning does not need to be the meaning of its origin. This is, I think, an especially important line of inquiry given scholarly debate about Thieme’s etymology. [49]
For this I turn to Nilsson, whose explanation is followed by most scholars. He suggests that the term adheres because of Hades’ use of horses in the abduction of Persephone. [50] This theory holds that Hades did not always await the arrival of new souls to the underworld, but instead came on occasion to the realm of the living to collect them. This particular element of Hades’ behavior would be typified in the abduction of Persephone. Hades’ horses would feature in this epithet because they are the means by which he hunts down his victims; they function metonymically, like the Grim Reaper’s sickle. Contrary to Nilsson’s suggestion, however, I do not believe that there is good evidence showing that Hades was often imagined carrying souls to the underworld via his chariot, [51] but instead Persephone seems to be the only figure about whom this is said to have happened. I suggest that this is the best starting point for this analysis. For although horses are not a frequent feature of Hades’ mythology, his abduction of Persephone by chariot was an exceedingly important and prolific story. Reference to Hades’ horses then would not remind the archaic audience of the underworld generally but of the sudden and violent abduction of Persephone specifically. The import of this minatory usage may, in fact, be that the victim is about to go to the underworld suddenly, just as Persephone did. The epithet need be neither incidental nor vestigial here but may have worked specifically to enhance the resonance of the language of the associated threat. Such a reading is, of course, impossible to verify conclusively but seems to me to provide the most promising hypothesis since it allows room for diachronic evolution of the phrase while still pointing the way to a synchronic utility that aided in the epithet’s survival.
Although the epithet κλυτόπωλος appears only three times in the Iliad, its occurrence offers us special insights concerning the prehistory of Homeric horse formulas. It also highlights the profound role that the phrase ὠκέες ἵπποι has in the history of Greek epic, as well as the place that it should have in our analysis of Homeric equine poetics. For not only do the metrical deployment and phonetic structure of κλυτόπωλος reflect common metrical and phonetic characteristics among the formulas related to ὠκέες ἵπποι, but the linguistic history of the epithet’s final element, πῶλος, allows us a glimpse of diachronic evolution within Homeric horse formulas generally. Its p sound has always been a p sound, while the p sound of formulas anchored by ἵππος was originally a k sound, and this allows us to identify two groups within this family of formulas, those that were always built on a k…p structure, like κλυτόπωλος, and those whose historic k…p structure reflects a prehistoric k…k structure, like ὠκέες ἵπποι. Given the ancient poetic qualities of the k…k formulas and the extreme age of their most prominent representative, ὠκέες ἵπποι, it must be the case that the production of ἵππος formulas predates the production of πῶλος formulas. κλυτόπωλος then proves a precious window into the deep history of these oral formulas as well as the practices of the poets who employed them. Although linguistic changes in a language must sometimes render poetic expressions obsolete, κλυτόπωλος provides beautiful evidence of how oral formulas survive those changes, and how traditional poetic expressions sometimes persist and even multiply, yielding new forms and new meanings over the millennia.


[ back ] 1. For the sake of enhancing the transparency of cognition, when Sanskrit phrases are to be compared to cognates I will provide the Sanskrit without sandhi (the phonetic changes that take place at word boundaries) and without phonetic change to consonants in absolute final position, whereby consonants are pronounced differently when they end a phrase). I will, of course, preserve all relevant phonetic changes when actual poetic lines are quoted. The dating of these literatures is of central importance to this sort of argumentation but, as with all oral poetry, is extremely problematic. The process of composition involved in all cases was very long in duration, but the linguistic eras represented principally by our surviving texts may both be safely assumed to predate that of the Homeric corpus (which will be considered the 8th century BCE). The compositional date of the Old Avestan corpus will be considered to be very roughly 1000–500 BCE; Beekes 1988:xi. The age of the Rigveda will be taken to be roughly 1500 BCE; Witzel and Jamison 2003:65.
[ back ] 2. For further evidence of this see Schmitt 1967:238.
[ back ] 3. For a similar conclusion from the point of view of our Sanskrit sources see Durante 1971:93.
[ back ] 4. Edwards 1988:29.
[ back ] 5. Compare Watkins on formulas for “immortal fame,” 1995:173–178; also Lord 2000:13–29.
[ back ] 6. Lord 2000:30–67.
[ back ] 7. For artistic documentation of the preeminence of horses in ancient Greece, see Markman 1943. For a survey of the economic and political significance of horses in Greece, see Camp 1998.
[ back ] 8. Some of the ensuing analysis also appears in Platte 2014.
[ back ] 9. The predilection of PIE verse for alliteration may be inferred from the frequency of this stylistic device in later traditions, but is also frequently revealed in PIE poetic reconstructions; cf. Watkins 1995:28–49. Watkins’ *égu̯hent ógu̯him represents a similar alliteration of labiovelar elements, although voiced in this case, as well as a similar alternation in root vowel. It is also interesting to note that even in Latin poetics, an area in which innovations so readily obscure evidence of PIE poetic inheritance, alliteration in very early verse is quite common, as early Saturnian verse makes clear.
[ back ] 10. My use of the word “jingle” is influenced by Watkins’ usage. See, for example, Watkins 1995:328. For an overview of PIE poetic technique in general see Schmitt 1967, especially 221–284. For an overview of poetic technique with an eye toward Greek inheritance specifically see Durante 1971, especially the first volume.
[ back ] 11. Watkins 1995:23.
[ back ] 12. The fact that these words formed a figura etymologica has been pointed out elsewhere: Katz 2010:361.
[ back ] 13. These examples of later phrases of the ἔπος ἐπεῖν type do not reflect precisely the same sort of derivational relationship in their constituent members, but should not be expected to, as vowel ablaut was no longer active in the linguistic phases represented here. Nevertheless, a continuity of underlying practice seems evident. English “sing-song,” although not historically traceable to PIE, demonstrates something slightly more akin to the PIE ablaut pattern represented here.
[ back ] 14. I must make clear here that the word ἵππος may never have contained an actual labiovelar, but instead a velar followed by a labial, with a syllabic break occurring between the two sounds, that is ἵκ-ϝος rather than ἵ-κϝος. I am assuming here that the conversion of this consonantal group occurred at roughly the same time as that of the labiovelars. This should be a safe assumption since the two developments proceeded in much the same way. That the loss of labiovelar was a late development in Greek is guaranteed by its presence in Mycenaean Greek.
[ back ] 15. I do not consider the Homeric text to be a perfect preservation of any one moment of poetic-linguistic development, but rather the result of a more fluid and evolving oral and textual tradition that only became fixed as the current text late in its history. I will, however, treat our text practically as representing a broad period of poetic-linguistic development.
[ back ] 16. Hainsworth 1968:48.
[ back ] 17. Due to their particularly regular deployment, formulas of this type were central to much of Parry’s initial work, and have long exerted great influence on approaches to formulaic behavior as a whole.
[ back ] 18. This formula is largely restricted to the Iliad, where horses appear much more frequently than in the Odyssey.
[ back ] 19. See Nagy 1974.
[ back ] 20. All textual translations are mine unless otherwise noted.
[ back ] 21. Cf. the βραδέες ἵπποι, “slow horses,” of Nestor, who has grown too old for combat, to be discussed in Chapter 2; Iliad VIII 104.
[ back ] 22. I say this only because, in Hesiod, formulas of the “swift-footed horses” variety are preferred to the simple “swift horses” formulas favored in Homer, so within the confines of Hesiodic verse one may prefer to analyze this formula within its own system.
[ back ] 23. Given the paucity of our sources, we cannot be certain that this formulation was not more common than our one preserved occurrence would lead us to believe. However, even if other examples did occur, as may be likely, they may still be understood within the model offered here.
[ back ] 24. These formulaic modifications do not have to be seen as derivatives exclusively, but may draw on the same sorts of phonetic, metrical, and thematic associations that occasioned the creation of the early freer formulaic networks out of which the later rigidly bound use developed. Yet even if this is the case, given the regularity of the final formula, this modification must be synchronically informed by the ὠκέες ἵπποι formula even if both descend, to some extent, from common practices diachronically.
[ back ] 25. Cf. Hainsworth 1968:94.
[ back ] 26. Cf. Iliad V 269.
[ back ] 27. Once the decision had been made to begin the line with the article, there may also have been a tendency to move one part of the upcoming formula into greater proximity to it.
[ back ] 28. On the sexual usage of ἐπιβαίνω see Liddell 1996:624, sub voce AIII3. Aristotle uses it of quadrupeds generally: ἐπιβαίνοντος ἐπὶ τὸ θῆλυ τοῦ ἄρρενος, “with the male mounting the female” (History of Animals 539b26). Compare the double meaning of ἐπεμβάτης in Anacreon’s famous lyric about the Thracian girl, to be discussed in Chapter 3, pages 74–76.
[ back ] 29. For a full discussion of the mechanics of verse-end separation, see Hainsworth 1968:105–109.
[ back ] 30. Bassett 1926.
[ back ] 31. A wide-ranging study of the idiosyncrasies and possible origins of the Doloneia can be found in Danek 1988.
[ back ] 32. See Delebeque 1951:143.
[ back ] 33. For the mechanics of synonymic substitution in Homeric formulas, see Paraskevaides 1984. It is very interesting that in the two occurrences of the phrase, χαλκόποδ’ ἵππω, it is immediately followed by the word ὠκυπέα at the start of the next line: VIII 41, XIII 23. It seems that even without a dual form of ὠκέες ἵπποι to deploy, the phrase itself is still somewhere in the mind of the poet, conditioning a bond between the two words that runs across the line boundary: χαλκόποδ’ ἵππω / ὠκυπέα.
[ back ] 34. Μώνυχες and ὠκέες are the two most common epithets of horses in Homer (Delebecque 1951:149).
[ back ] 35. There are also examples of the term from later literature, but I have omitted them from this discussion because their late date removes them from the world of oral poetry. Triphiodorus uses it to describe the Achaeans, Ἅλωσις Ἴλιου, 92; Maximus Astrologus uses it to describe Selene, Περὶ Καταρχῶν, 5.75, 6.151, 6.261. It also appears once in the Papyri Magicae Graecae to describe Helios, Preisendanz 2.89.
[ back ] 36. On the commonality of speed as a basic approbative value of horses in IE cultures see Matasovic 1996:73–4.
[ back ] 37. Nagler 1967:269–291.
[ back ] 38. Such associations may also explain the unusual consistency of the nu that so often precedes the accusative formulas discussed above.
[ back ] 39. cf. Apion’s definition: ‘ἱππους ἀγαθοὺς <ἔχων>’, Γλῶσσαι (Sammlung griechischer und lateinischer Grammatiker) 3, fr. 51.1.
[ back ] 40. Pindar may, in fact, showcase other members of this family not attested in Homer. λεύκιπποι, and λευκόπωλοι, the bright horses, for example, are both absent from Homer but display the same underlying phonetic structure as the Homeric ‘good horses’ formulas.
[ back ] 41. One more term, πλήξιππος (“lasher of horses”) may also be added to the list of related expressions in Homer, since it features the same phonetic structure and in three of its four appearance in the Iliad it appears at verse end (II 104, IV 327, XI 93). Admittedly, it does not agree with the others semantically in quite as neat a fashion, but it does indicate one’s status as horseman and so could function in the same broad nexus of expressions. Moreover, the word is slightly unusual because compounds that feature ἵππος as their final element are fairly rare, as is generally the case of compounds in which a verbal element precedes a nominal, i.e. the “pickpocket” compounds. The system proposed here may provide some justification for this unusual form.
[ back ] 42. I must repeat here that the word ἵππος may never have contained an actual labio-velar, but instead a velar followed by a labial, but that this difference should be insignificant for this analysis. See note 14.
[ back ] 43. Thieme 1968:143–148.
[ back ] 44. There are other cognates that support this as well. See Beekes 2013:1266 and Frisk 1970:2.634.
[ back ] 45. ἴφθιμος is particularly interesting because it is only used of Hades himself in the Odyssey, while in the Iliad it occurs in proximity to Hades but modifies the souls and the heads of the men who die and are sent to him (I 3; XI 55).
[ back ] 46. I omit here the argument of Verrall (1898) that the term had nothing to do with horses at all but rather with “ranging” and “haunting,” as if connected to πωλέομαι. Pindar’s application of the word to Poseidon, the horse god, makes his idea difficult to accept.
[ back ] 47. Puhvel also points out (1987:109) that the Vedic Yama is the saṃgámanam jánānām, “ingatherer of the people.” The more traditional etymology traces the word to *ṇu̯id- ‘unseen’, which not only corresponds with the basic notion that death is unforeseeable, but is particularly apt given Hades’ possession of a cap that imbued its wearer with invisibility (cf. Apollodorus Bibliotheca I 2.1 and The Shield 226–227; Pherekydes 3 F 11). See Frisk 1970:1.33.
[ back ] 48. Although beyond the scope of this argument, there are some other pieces of Indic evidence that could be added to bolster the connection between horses and the ruler of the dead. The name of the Indo-Iranian underworld god, Yama (Sanskrit)/Yima (Avestan) does appear to have something to do with reining, and the Sanskrit noun yama when it does not appear as a name can indicate a rein. Yama is also said to have had particularly good horses, but that is true of the subjects of too many Vedic hymns to be useful: e.g. hiraṇyakaṣyānsudhurān hiraṇākṣānayaśśaphān aśvānanaśyato, “horses with golden girdles, good under the yoke, golden eyed, and iron hoofed, immortal” (Taittirīya Āraṇyaka–6). For other similarities between the Greek and Indic afterworld see Puhvel 1987:139. On the Yama analogues in Greek mythology more generally see Ehni 1890:196–209.
[ back ] 49. On objections to Thieme’s etymology see Beekes 2013:34.
[ back ] 50. Nilsson 1941:I 425.
[ back ] 51. The only evidence that could support this, that I know of, comes from epitaphs that occasionally describe Hades snatching (ἁρπάζω) the dead: e.g. Inscriptiones Graecae IIÏ 12629, Supplementum Epigraphicum Graecum 25.298. Even then, however, this seems usually to indicate the death of a young woman and may then still be a reference to the Persephone myth.