The meaning of Homeric εὔχομαι through its formulas

Chapter II. εὔχομαι in Sacral Contexts

Analytic Table of Contents

A. Formulas which conclude prayers (pp. 18–31)
          1. Introduction; formulas (A) and (B) (pp. 18–26)
          2. Formula (A) and the meaning of εὔχομαι (pp. 26–31)
B. Formulas which report prayers (pp. 31–66)
          1. Absolute usage of εὔχομαι (pp. 31–43)
                    a. Classified list of attestations (p. 31):
                              I: Ritual narrative formulas
                              II: κλύω + εὔχομαι formulas
                    b. Discussion of Section I attestations (pp. 31–37)
                    c. Discussion of Section II attestations (pp. 37–43)
          2. εὔχομαι + dative (pp. 43–53)
                    a. Classified list of attestations (pp. 43–45)
                              I: εὔχ
                              II: εὔχ  ̶  
                              III: εὔχ ⏑  ̶
                              IV: εὔχ  ̶  ̶
                              V: Anomalous attestations
                    b. Rationale of classification (pp. 45–52)
                    c. Meaning of εὔχομαι in this usage (pp. 52–53)
          3. εὔχομαι + infintive (pp. 53–66)
                    a. Classified list of attestations (pp. 53–54):
                              I: Future infinitive
                              II: Aorist infinitive
                    b. Discussion of Section I attestations: α 50–51, 59–60; Δ 100–103, 119–21 (pp. 54–57)
                    c. Discussion of Section I attestations (continued): Θ 526 (pp. 57–62)
                    d. Discussion of Section II attestations (pp. 62–66)
C. Conclusion (pp. 66–67) {17|18}

A. Formulas which conclude prayers

1. Introduction: formulas (A) and (B)

In Homer, εὔχομαι occurs in three easily distinguished contexts: (a) sacral, (b) legal, (c) secular. The method of contextual analysis makes the choice of a starting point between the three arbitrary, since determination of their historical interrelationship is secondary to a thorough investigation of each individually. We begin with the attestations in sacral contexts. More specifically, we begin with the systematically formular attestations of εὔχομαι in these contexts, on the assumption that these are traditional and therefore relatively archaic. [1] This is not to say that non-formular attestations are ipso facto later. Of necessity, the basic criterion for ‘formula’ status is recurrence, [2] but it is obvious that an obsolescent archaism may be as weakly attested as an innovation. Then there is the question of the completeness of our sample of epic diction. Both considerations make it advisable to decide the meaning and relative archaism of non-formular attestations on an individual basis and with the evidence from formular attestations already in hand.

Here is the most common εὔχομαι formula in sacral contexts:


ὣς ἔφατ᾿ …, κ. τ. λ.   ὣς ἔφατ᾿ εὐχόμενος
(without εὐχόμενος)   κ.τ.λ.
Concludes all direct speach
including prayers
vs. Concludes directly quoted
prayers only

{18|19} We can witness this contrast operating in epic diction by contrasting (A) with a specific case of # ὣς ἔφατ᾿…, κ. τ. λ. (without εὐχόμενος), viz, the formulaic line (B):

ὣς ἔφαθ᾿, οἱ δ᾿ ἄρα τοῦ μάλα μὲν κλύον ἠδὲ πίθοντο
                                                    (13 times)

(B) is a transformation of (A). This is immediately suggested by the accumulation of particles in (B) (δέ, ἄρα, μέν, μάλα). Furthermore, the cadence ἠδὲ πίθοντο #, which replaces the divine name + epithet/divine epithet cadence of (A), is probably nothing more than a gloss on the word κλύον which precedes it. Epic dialect attests other such line-final pairs, e. g. ἡγήτορες ἠδὲ μέδοντες # (Β 79, Κ 301, θ 97, ν 186, etc.), λωΐτερον καὶ ἄμεινον # (α 376, β 141), which consist of an archaic word (μέδοντες, λωΐτερον, κλύον and a living one to gloss it. Cp. English ‘time and tide (cognate with Ger. Zeit),’ ‘hue (cognate with Lat. heu) and cry’, ‘might and main (cognate with Lat. manus)’. As its parallelism with these phrases suggests, (B)’s κλύον in κλύον ἠδὲ πίθοντο # may well mean ‘hearkened, listened’ instead of the etymological (Frisk 1960–1970 s.v. κλέος) ‘hear’. [
4] This semantic specialization can be paralleled in the word which replaces κλύω in Greek, namenly ἀκούω ‘1. hear; 2. listen’ (LSJ9 s.v.). ‘List’ and ‘listen’ are themselves cognate with κλύω (C. T. Onions 1966 s.v. list).

The contextual constraint which necessitates the deletion of εὐχόμενος from (B) is apparent elsewhere. Another, though less well attested, formulaic line undergoes an exactly parallel transformation. Here are its two attestations:

(C) Α 453 ἠμὲν δή ποτ’ ἐμεῦ πάρος ἔκλυες εὐξαμένοιο,
               τίμησας μὲν ἐμέ, μέγα δ’ ἴψαο λαὸν Ἀχαιῶν·
               ἠδ’ ἔτι καὶ νῦν μοι τόδ’ ἐπικρήηνον ἐέλδωρ
(C1) Π 236 ἠμὲν δή ποτ᾿ ἐμὸν ἔπος ἔκλυες εὐξαμένοιο
                 τίμησας μὲν ἐμέ, μέγα δ᾿ ἴψαο λαὸν Ἀχαιῶν
                 ἠδ᾿ ἔτι καὶ νῦν μοι τόδ᾿ ἐπικρήηνον ἐέλδωρ

Both (C) and (C1) appear in sacral contexts. A man (Chryses, Achilles) is praying to a god (Apollo, Zeus) and referring to past occasions when he was heard. But in Book Ξ when Hera is trying to enlist the aid of the god Hypnos in her plan to seduce Zeus, she says to him:

Ξ 233 Ὕπνε ἄναξ πάντων τε θεῶν πάντων τ’ ἀνθρώπων,
         ἠμὲν δή ποτ’ ἐμὸν ἔπος ἔκλυες, ἠδ’ ἔτι καὶ νῦν
         πείθευ· ἐγὼ δέ κέ τοι ἰδέω χάριν ἤματα πάντα

Here again, εὐξαμένοιο has been deleted and ἔκλυες is glossed by πείθευ. [
6] Again, the change in context from sacral (man addressing god) to secular (god addressing god || man addressing man/men) motivates the deletion. The following diagram sums up the information thus far obtained on the contextual distribution of εὐχόμενος:


But the segments of (A) are not used in this way. Instead there is a series of variations on the line as a totality, as follows:

(A1) O 377 Ὣς ἔφατ’ εὐχόμενος, μέγα δ’ ἔκτυπε μητίετα Ζεύς
(A2) β 267 ὣς ἔφατ’ εὐχόμενος, σχεδόθεν δέ οἱ ἦλθεν Ἀθήνη
(A3) Ζ 311 Ὣς ἔφατ’ εὐχομένη, ἀνένευε δὲ Παλλὰς Ἀθήνη
(A4) Θ 198 Ὣς ἔφατ’ εὐχόμενος, νεμέσησε δὲ πότνια Ἥρη

{21|22} All these lines conclude prayers. They begin with # ὣς ἔφατ᾿ εὐχόμενος/εὐχομένη [
10] and conclude with a divine name. The only portion of the line which is subject to change is the verb characterizing the nature of the god’s response as markedly positive (μέγα δ᾿ ἔκτυπε, σχεδόθεν δέ οἱ ἣλθεν) or markedly negative (ἀνένευε δέ, νεμέσησε δέ). [11] All are examples of strong reactions of Athena or Zeus or Athena’s companion, Hera, to prayers by or against their special favorites. In (A1), Nestor has prayed to Zeus, and the thunderclap (μέγα δ᾿ ἔκτυπε) is an exceptionally favorable response proportional to the patriarch’s prestige and power elsewhere in the tradition (υ 101–3, Bacchylides Dithyramb 17.67–80 [Snell]). At β 267 (A2) Telemachus, who has just failed in his first attempt to play an adult role before the suitors and townspeople of Ithaca, is depressed and helpless. His prayer is substandard in form and Athena, his special protectress as she is also Odysseus’, rushes to help. [12] Conversely, (A2) and (A4) are negative responses of Athena and Hera to prayers by Hector for the breaking of Diomedes, a hero under their special protection as was his father, Tydeus. [13] Thus the variations (A1) – (A4) are poetic special effects, ringing changes on the rigid (A), which are contextually as well as formally restricted.

From this point of view, it is more difficult to dislodge εὐχόμενος from the text. It is possible that (A8) and (A7) preserve an archaic usage of the phrase # ὣς ἔφατ᾿ εὐχόμενος || in boasting contexts which predates the constitution of (A) as a unit and its fixation in sacral contexts. The poetics of battle-books, in which (A8) and (A7) occur, are such as to promote the preservation of such an archaism, or, alternatively, to stimulate the innovatory usage of the first half of (A) outside its conventional context. Consider, for example, the common battle-book formula:

δούπησεν δὲ πεσών, ἀράβησε δὲ τεύχεα ἐπ᾿ αὐτῷ

which occurs seven times (Δ 504, Ε 42, Ε 540, Ν 187, Ρ 50, Ρ 311) in the Iliad as a unit. Its pre-caesural segment is also attested with a different half-line following it nine other times, and each of the nine times the second {24|25} half of the line is different. Another example: in Ε 42–83 (forty-two lines) six heroes die, and each time a different line describes it:

Ε 42 δούπησεν δὲ πεσών, ἀράβησε δὲ τεύχεα ἐπ᾿ αὐτῷ
Ε 47 ἤριπε δ᾿ ἐξ ὀχέων, στυγερὸς δ᾿ ἄρα μιν σκότος εἶλε
Ε 58 ἤριπε δὲ πρήνης, ἀράβησε δὲ τεύχεα ἐπ᾿ αὐτῷ
Ε 68 γνὺξ δ᾿ ἔριπ᾿ οἰμώξας, θάνατος δέ μιν ἀμφεκάλυψε
Ε 75 ἤριπε δ᾿ ἐν κονίῃ, ψυχρὸν δ᾿ ἕλε χαλκὸν ὀδοῦσιν
Ε 83 ἔλλαβε πορφύρεος θάνατος καὶ μοῖρα κραταίη

The pressure of this variation aesthetic on the poet’s resources generates many new combinations and causes the preservation of old or creation of new phrases. [
19] Of the eleven different half-line segments used in the above examples, eight recur paired with other half-lines, three are unique, and only two pairs recur as a unit. So these six examples by no means exhaust the possible expressions which the poet has and uses elsewhere to describe the death of warriors. The heroic boast over their corpses is a recurrent feature of battle-books and is also likely to be introduced or concluded by unique phrases which can be archaisms or innovations.

Furthermore, the occasional breaking of contextual constraints is not uncommon in Homer and cannot be ruled out even outside battle-books. We will discuss an analogous case below (p. 133f.) in which a formula fixed (six times) in exactly this secular context of heroic boasts, viz. || εὐχόμενος ἔπος ηὔδα #, occurs once in a sacral context (Κ 461). There are no textual variants for Κ 461, nor is it necessary to emend: the line is either a compositional slip or an archaism predating the sacral-secular semantic split.

Yet it is worth remembering that (A6) and (A7) break both contextual constraints (crossing the boundary between sacral and secular) as well as dictional constraints against breaking (A) apart. The force of these constraints is evident from the contextual similarities of (A1)–(A5) and their rigorous formal parallelism to (A). While there is beguiling simplicity in the hypothesis that # ὣς ἔφατ᾿ εὐχόμενος is a scribal error, I have not yet found exact parallels to the shattering of precedents in (A6) and (A7).

For our immediate purposes, however, the difficulty in reading ὣς ἔφατ᾿ εὐχόμενος in (A6) and (A7) and the fact that # ὣς φάτ᾿ ἐπευχόμενος is actually attested in the Hymn to Apollo (A8) are more important than a clear-cut decision on the textual problem. Both considerations imply that (A) in whole or in part is felt by the manipulators of the epic language to be specifically constrained against usage in a secular context to which the word εὔχομαι per se is adaptable. This suggests that it has a different {25|26} meaning in sacral contexts than in secular ones. If there were free substitution of this formula or part of it in both sacral contexts and the context of heroic boasts, then it would be impossible to justify proposing two lemmata, one sacral, the other secular, in a definition of εὔχομαι. As it is, the simplest way to account for the constraint against free substitution is to postulate its realization in the epic poet’s mind as a semantic split. Again, the poet will not use an εὔχομαι formula designed for sacral contexts in secular contexts. It remains to be seen how consistent this is with the usage of other sacral εὔχομαι formulas, but the cancellation of εὐχόμενος and the substitution of ἐπευχόμενος in the textual tradition of (A8) and (A7) are consistent with this formulation.

In other words, we have provided evidence that the sacral-secular contextual split, which is the organizing principle of this and the next chapter, is a functioning reality to the epic poet. Still, this makes nothing explicit about what εὐχόμενος actually means in (A). Having selected one feature of the context in which it appears, namely the human and divine status of the respective speakers and hearers of speeches concluded by ἔφατ᾿ εὐχόμενος, we have shown that it is of crucial significance. A change in the status of the speaker or the hearer/hearers which alters the context from sacral to secular coincides with deletion of εὐχόμενος or textual problems regarding its presence. On this basis it is reasonable to hypothesize that there is a sacral lemma in the meaning of εὔχομαι. But the next step is to determine the content of that lemma. The simplest method is to examine not places where the word is deleted or textually doubtful but to establish its semantic field in a more positive way by studying words substituted for it in epic diction.

2. Formula (A) and the meaning of εὔχομαι

The following parallels to the opening of (A) occur in the last nine books [20] of the Iliad:

# Ὣς φάτ’ ἀπειλήσας || Φ 161, Ψ 184
# ὣς ἔφατο κλαίων || Χ 429
# Ὣς ἔφατο κλαίουσ’ || X 437, 515, κ.τ.λ.
# Ὣς φάτ’ ἐποτρύνων || Υ 373
# Ὣς φάτο λισσόμενος || Π 46

The common semantic factor in these participles after φημί is that they categorize in broad terms the speech which immediately precedes them as a threat, lament, exhortation, or supplication. Then εὐχόμενος in (A) should mean ‘praying’ broadly conceived, since to our way of thinking, ‘prayer’ is a category of speech on the level of ‘lament’, ‘supplication’, etc. As a dictionary definition, this may seem adequate, but in reality it raises more problems than it solves. What is the Homeric notion of prayer? How does it compare with ours? Does εὐχόμενος simply mean ‘praying’ as {26|27} opposed to ‘lamenting’, or does it have a more specific application to the speeches it concludes which allows it to function as a label on this level? If εὐχόμενος labels an act of speech by a man to a god, what kind of a speech is it? Here is a place to begin, with some examples of Homeric prayers:

a) κλῦθί μευ ἀργυρότοξ’, ὃς Χρύσην ἀμφιβέβηκας
Κίλλαν τε ζαθέην Τενέδοιό τε ἶφι ἀνάσσεις,
Σμινθεῦ εἴ ποτέ τοι χαρίεντ’ ἐπὶ νηὸν ἔρεψα,
ἢ εἰ δή ποτέ τοι κατὰ πίονα μηρί’ ἔκηα
ταύρων ἠδ’ αἰγῶν, τὸ δέ μοι κρήηνον ἐέλδωρ·
τίσειαν Δαναοὶ ἐμὰ δάκρυα σοῖσι βέλεσσιν.
Ὣς ἔφατ’ εὐχόμενος, τοῦ δ’ ἔκλυε Φοῖβος Ἀπόλλων

A 37-43

b) πότνι’ Ἀθηναίη ἐρυσίπτολι δῖα θεάων
ἆξον δὴ ἔγχος Διομήδεος, ἠδὲ καὶ αὐτὸν
πρηνέα δὸς πεσέειν Σκαιῶν προπάροιθε πυλάων,
ὄφρά τοι αὐτίκα νῦν δυοκαίδεκα βοῦς ἐνὶ νηῷ
ἤνις ἠκέστας ἱερεύσομεν, αἴ κ’ ἐλεήσῃς
ἄστύ τε καὶ Τρώων ἀλόχους καὶ νήπια τέκνα.
Ὣς ἔφατ’ εὐχομένη, ἀνένευε δὲ Παλλὰς Ἀθήνη.

Ζ 305-311

c) κλῦθί μευ αἰγιόχοιο Διὸς τέκος, ἥ τέ μοι αἰεὶ
ἐν πάντεσσι πόνοισι παρίστασαι, οὐδέ σε λήθω
κινύμενος· νῦν αὖτε μάλιστά με φῖλαι, Ἀθήνη,
δὸς δὲ πάλιν ἐπὶ νῆας ἐϋκλεῖας ἀφικέσθαι
ῥέξαντας μέγα ἔργον, ὅ κε Τρώεσσι μελήσῃ …
ὣς ἔφαν εὐχόμενοι, τῶν δ’ ἔκλυε Παλλὰς Ἀθήνη.

Κ 278-282, 295

d) Ζεῦ ἄνα Δωδωναῖε Πελασγικὲ τηλόθι ναίων
Δωδώνης μεδέων δυσχειμέρου, ἀμφὶ δὲ Σελλοὶ
σοὶ ναίουσ’ ὑποφῆται ἀνιπτόποδες χαμαιεῦναι,
ἠμὲν δή ποτ’ ἐμὸν ἔπος ἔκλυες εὐξαμένοιο,
τίμησας μὲν ἐμέ, μέγα δ’ ἴψαο λαὸν Ἀχαιῶν,
ἠδ’ ἔτι καὶ νῦν μοι τόδ’ ἐπικρήηνον ἐέλδωρ …
Ὣς ἔφατ’ εὐχόμενος, τοῦ δ’ ἔκλυε μητίετα Ζεύς.

Π 233-238, 249

These four passages, are discernible examples of composition by theme, [
21] of large units of language built on a common plan in a way which is appropriate to hieratic as well as oral poetic language. A Homeric prayer has the following structural elements: (I) Invocation of god or goddess with ornamental epithets, etc. (2) Claim that person praying is entitled to a favor on the basis of favors being granted, granted in the past, or to be granted, {27|28} or on the basis of a previous response which implies the existence of a contract between god and man based on past exchange of favors. [22] (3) Specific request for a favor in return, including an implied or explicit statement of the relevance of the favor to the particular god’s sphere. All of these elements can be amplified or minimized according to circumstances, as for example the grandiose and exotic invocation in Achilles’ great prayer to Zeus ( d) above) or the fierce conciseness of the request in Chryses’ prayer to Apollo (a)). And this five- or six-line pattern undergoes more radical transformations, as in:

e) Ψ 770 κλῦθι θεά, ἀγαθή μοι ἐπίρροθος ἐλθὲ ποδοῖιν

which is Odysseus’ complete prayer to Athena made ὅν κατὰ θυμόν when he is running the last lap of the foot race in Patroclus’ funeral games (claim to favor based on past granting of favor, etc. is omitted due to extenuating circumstances). Another expressive transformation:

f) “κλῦθί μοι, ὃ χθιζὸς θεὸς ἤλυθες ἡμέτερον δῶ
καί μ’ ἐν νηῒ κέλευσας ἐπ’ ἠεροειδέα πόντον,
νόστον πευσόμενον πατρὸς δὴν οἰχομένοιο,
ἔρχεσθαι· τὰ δὲ πάντα διατρίβουσιν Ἀχαιοί,
μνηστῆρες δὲ μάλιστα, κακῶς ὑπερηνορέοντες.”
ὣς ἔφατ’ εὐχόμενος, σχεδόθεν δέ οἱ ἦλθεν Ἀθήνη,

β 262-267

Telemachus doesn’t know the god’s name (it is Athena), and he is so demoralized by failing to impress the suitors he doesn’t know what to request (invocation substandard and conflated with grant of favor, request for favor omitted). [
23] When Priam solicits an omen from Zeus:

g) Ζεῦ πάτερ Ἴδηθεν μεδέων κύδιστε μέγιστε
δός μ’ ἐς Ἀχιλλῆος φίλον ἐλθεῖν ἠδ’ ἐλεεινόν,
πέμψον δ’ οἰωνὸν ταχὺν ἄγγελον, ὅς τε σοὶ αὐτῷ
φίλτατος οἰωνῶν, καί εὑ κράτος ἐστὶ μέγιστον,
δεξιόν, ὄφρά μιν αὐτὸς ἐν ὀφθαλμοῖσι νοήσας
τῷ πίσυνος ἐπὶ νῆας ἴω Δαναῶν ταχυπώλων.
Ὣς ἔφατ’ εὐχόμενος, τοῦ δ’ ἔκλυε μητίετα Ζεὺς

Ω 308-314

{28|29} he omits to claim the right to response, perhaps because he is pouring a libation as he prays (Ω 306) and a libation is a present favor or a symbol of past favors if not a promised one. [

For our purposes, it is important to conceptualize the prayer theme and these variants of it as a kind of deep structure with surface structure manifestations. [25] It would be misleading to conclude from the option of omitting the request feature and the claim to a favor on the basis of other favors granted that the only constant in prayers is the invocation and that εὐχόμενος accordingly refers to it only. The absence of a given element, like its elaboration or diminution, is a surface phenomenon, and εὐχόμενος is motivated by the globally conceived deep structure. In these terms one can understand the line which concludes this speech by Hector to his horses (!):

“Ξάνθε τε καὶ σὺ Πόδαργε καὶ Αἴθων Λάμπε τε δῖε
νῦν μοι τὴν κομιδὴν ἀποτίνετον, ἣν μάλα πολλὴν
Ἀνδρομάχη θυγάτηρ μεγαλήτορος Ἠετίωνος
ὑμῖν πὰρ προτέροισι μελίφρονα πυρὸν ἔθηκεν
οἶνόν τ’ ἐγκεράσασα πιεῖν, ὅτε θυμὸς ἀνώγοι,
ἢ ἐμοί, ὅς πέρ οἱ θαλερὸς πόσις εὔχομαι εἶναι.
ἀλλ’ ἐφομαρτεῖτον καὶ σπεύδετον ὄφρα λάβωμεν
ἀσπίδα Νεστορέην, τῆς νῦν κλέος οὐρανὸν ἵκει
πᾶσαν χρυσείην ἔμεναι, κανόνας τε καὶ αὐτήν,
αὐτὰρ ἀπ’ ὤμοιιν Διομήδεος ἱπποδάμοιο
δαιδάλεον θώρηκα, τὸν Ἥφαιστος κάμε τεύχων.
εἰ τούτω κε λάβοιμεν, ἐελποίμην κεν Ἀχαιοὺς
αὐτονυχὶ νηῶν ἐπιβησέμεν ὠκειάων.”
Ὣς ἔφατ’ εὐχόμενος, νεμέσησε δὲ πότνια Ἥρη,

Θ 185-198

B. Formulas which report prayers

That εὔχομαι in sacral contexts relates to the whole tripartite structure of Homeric prayers and not any one element of it emerges more clearly from a study of prayers reported by εὔχομαι rather than directly quoted. These fall into three groups which we shall survey individually, considering as well the evidence they provide with regard to the sacral/secular semantic split postulated above.

1. Absolute usage of εὔχομαι to report prayers

a. Classified List of Attestations

I. Ritual narrative formulas


II. κλύω + εὔχομαι formulas

γ 55 κλῦθι, Ποσείδαον γαιήοχε, μηδὲ μεγήρῃς
       ἡμῖν εὐχομένοισι τελευτῆσαι τάδε ἔργα
Α 453 ἠμὲν δή ποτ’ ἐμεῦ πάρος ἔκλυες εὐξαμένοιο
Π 236 ἠμὲν δή ποτ’ ἐμὸν ἔπος ἔκλυες εὐξαμένοιο
Ι 509 τὸν δὲ μέγ’ ὤνησαν καί τ’ ἔκλυον εὐχομένοιο
Π 531 ὅττι οἱ ὦκ’ ἤκουσε μέγας θεὸς εὐξαμένοιο
Α 381 εὐξαμένου ἤκουσεν, ἐπεὶ μάλα οἱ φίλος ἦεν

In these examples, the verb εὔχομαι is used without any grammatical complement whatsoever, just as it is in (A). This lack in itself suggests that εὔχομαι refers to the elements of Homeric prayer in their totality {31|32} rather than in part (e. g. ‘invoke’, ‘request’, etc.). Neither does internal analysis provide evidence for a specialized meaning.

b. Discussion of I: Ritual narrative formulas

I have divided the attestations in this group into two sections on the basis of formal and contextual distinctions, since they provide an objective point of departure for the semantic analysis of εὔχομαι. All of the passages in section I occur in narratives of ritual scenes. [32] The language of this sub-genre of Epic has its own formal peculiarities. [33] What characterizes it is an abundance of finite verbs in the indicative. μ 359 and ο 222 (cited above) are cases in point. Each contains three finite verbs, a feature of style rarely paralleled in other contexts of the Homeric poems, where the dictum ‘the line of poetry is a perfect sentence’ generally applies. A familiar example will remind the reader of the effect:

αὐτὰρ ἐπεί ῥ’ εὔξαντο καὶ οὐλοχύτας προβάλοντο,
αὐέρυσαν μὲν πρῶτα καὶ ἔσφαξαν καὶ ἔδειραν,
μηρούς τ’ ἐξέταμον κατά τε κνίσῃ ἐκάλυψαν
δίπτυχα ποιήσαντες, ἐπ’ αὐτῶν δ’ ὠμοθέτησαν·
καῖε δ’ ἐπὶ σχίζῃς ὁ γέρων, ἐπὶ δ’ αἴθοπα οἶνον
λεῖβε· νέοι δὲ παρ’ αὐτὸν ἔχον πεμπώβολα χερσίν.
αὐτὰρ ἐπεὶ κατὰ μῆρε κάη καὶ σπλάγχνα πάσαντο,
μίστυλλόν τ’ ἄρα τἆλλα καὶ ἀμφ’ ὀβελοῖσιν ἔπειραν,
ὤπτησάν τε περιφραδέως, ἐρύσαντό τε πάντα.
αὐτὰρ ἐπεὶ παύσαντο πόνου τετύκοντό τε δαῖτα
δαίνυντ’, οὐδέ τι θυμὸς ἐδεύετο δαιτὸς ἐΐσης.

A 458-468

The other three lines in this section are not, strictly speaking, formulas. Is the meaning εὔχομαι in these lines analogous to that in the formula # αὐτὰρ ἐπεί ῥ’ εὔξαντο? If they are analogous to it in form, there is reason to believe they are analogous in function. Here are the lines in question:

ο 222 ἦ τοι ὁ μὲν τὰ πονεῖτο καὶ εὔχετο, θῦε δ’ Ἀθήνῃ
ο 258 σπένδοντ’ εὐχόμενόν τε θοῇ παρὰ νηῒ μελαίνῃ
γ 45 αὐτὰρ ἐπὴν σπείσῃς τε καὶ εὔξεαι, ἣ θέμις ἐστί

Though the opening of o 222 is a formula (ἦ τοι ὁ μὲν. . ., passim) the combination of words in the rest of the line is unique in the Homeric corpus. However, the context, vocabulary, and style of this line, with its three finite verbs, belong to the ritual narrative sub-genre. It is probably an authentic {33|34} piece of traditional language that is simply unattested elsewhere. In any case, the lack of exact parallels to it is not sufficient evidence to invalidate an assumption that εὔχετο in ο 222 means the same as εὔξαντο in Α 458 κ. τ. λ.

This argument can be reinforced if the next two lines (ο 258 and γ 45) are also formally analogous to Α 458 κ. τ. λ., since ο 222 and ο 258 respectively begin and end a ring composition of 35 lines on the ancestry of Theoclymenus: the lines which frame such digressions in Homer are often verbally identical to say nothing of their functional equivalence. [38] But again ο 258 and γ 45 do not fit Parry’s strict definition of a formula. While both contain formulaic elements (θοῇ παρὰ νηῒ μελαίνῃ, cp. Α 300, HApoll 497; αὐτὰρ ἐπήν cp. α 293, Ο 147; ἥ θέμις ἐστί, cp. Β 73, Ψ 581) σπένδω and εὔχομαι are not metrically or morphologically fixed. [38a] However, it is poor method as well as counterintuitive to insist upon the strictness of Parry’s definition, especially when the collocation of εὔχομαι and σπένδω is attested elsewhere:

Π 253 ἤτοι ὃ μὲν σπείσας τε καὶ εὐξάμενος Διὶ πατρὶ
Ω 287 τῆ σπεῖσον Διὶ πατρί, καὶ εὔχεο οἴκαδ’ ἱκέσθαι

Clearly, then, the occurrence of these words together is traditional, as Benveniste has noticed. [
39] Now in these four lines neither of the verbs are in the aorist or imperfect indicative, but they are always in parallel construction, and all but one of the lines occur in a ritual narrative. [40] Given these considerations and also the lack of morphological and metrical fixity in the attestations, it seems likely that they are all transformations of an unattested ritual narrative formula containing εὔχομαι and σπένδω in the aorist or imperfect indicative. [41] If we can legitimately assume that transformations of this type do not destroy a formula’s identity, [42] then the collocation of εὔχομαι and σπένδω is another instance of a formula designed for sacral contexts which does not appear in secular contexts. Compared to # αὐτὰρ ἐπεί ῥ’ εὔξαντο ||, it has less probative value about the meaning of {34|35} εὔχομαι, since σπένδω occurs only in sacral contexts and in connection with it εὔχομαι need not carry any sacral connotations of its own. However, the two formulas share stylistic and contextual constraints, and accordingly it is difficult to allow that εὔχομαι has different meanings in each.

According to my hypothesis, however, εὔχομαι and σπένδω both refer to concrete actions, uttering words and pouring wine. This is suggested by the language of the ritual narratives in which they occur together and, for εὔχομαι, by the structure of speeches concluded by (A). Both are sacral actions with symbolic meaning and intent. A libation of wine is a token, wasted [49] gift of drink which symbolizes past, present, and future reciprocity between god and man, and which hopes for a return gift to confirm the existence of a reciprocal contract between the two. [50] Similarly, a prayer is a physical act of communication (in the etymological sense). It invokes a god, then it promises him a gift or reminds him of a past gift or calls attention to a present one; then, in return, it specifies a favor. In Greek, these words, εὔχομαι and σπένδω, are relatively pristine. They refer to physical actions endowed with meaning and formal design. Secondarily, some aspects and implications of these sacral acts become explicit and independent in the notions (a) ‘promise’, which is one possible part of the gift-time continuum in the ‘claim to response’ element of a Homeric prayer or an implication of the contract invoked in libation; and (b) ‘wish’, which is one part of the prayer structure, the request for a return favor, or one part of the libation rite, the wish for security: thus the semantic restrictions of the Latin cognates spondeō ‘pledge, promise’ (perhaps also *spons, spontis ‘wish’ [51] and voveō ‘pledge; wish.’ These Latin words are morphologically secondary to εὔχομαι and σπένδω as well as morphologically identical, [52] {36|37} so Benveniste, who allows that spondeō is semantically secondary [53] while supposing that voveō is primary, is formally and conceptually out of synchronisation. The present systematic study of εὔχομαι reporting prayers will bear out this hypothesis by demonstrating as plausible the notion that meanings such as ‘vow’ and ‘wish’ are too specific for the Homeric attestations. Other elements of proof will have to await the rest of the survey and the discussion of Indo-European evidence in an epilogue.

c. Discussion of II: κλύω + εὔχομαι formulas

The examples of absolute εὔχομαι in Section II are not fixed in an Epic sub-genre, but they do possess formal unity on another level. Three are attestations of a strict formula which is designed for and fixed in sacral contexts:

Α 453 ἠμὲν δή ποτ᾿ ἐμεῦ πάρος ἔκλυες εὐξαμένοιο
Π 236 ἠμὲν δή ποτ᾿ ἐμὸν ἔπος ἔκλυες εὐξαμένοιο
Ι 509 τὸν δὲ μέγ’ ὤνησαν καί τ’ ἔκλυον εὐχομένοιο

The contextual constraint upon this expression is comprehensible if εὐχομένοιο/εὐξαμένοιο has intrinsic sacral connotations. For our previous analysis of Α 453 (C) and Π 236 (C2) showed that ἔκλυες has no sacral connotations. It remains when the context becomes secular, while εὐχομένοιο/εὐξαμένοιο is deleted (see above, p. 19f.). The same point emerges more positively from the following expressions:

       δ 505 ἔκλυεν | αὐδήσαντος #
       γ 337 ἔκλυον | αὐδησάσης #
 HDem 299 ἔκλυον | αὐδήσαντος #
       Κ 47   ἔκλυον | αὐδήσαντος #
       Π 76   ἔκλυον | αὐδήσαντος #

αὐδήσαντος/αὐδησάσης is an exact metrical equivalent of εὐχομένοιο/εὐξαμένοιο, so the substitution of one for the other can only have been made on contextual grounds. These are not difficult to discover. While the antecedent of εὐχομένοιο/εὐξαμένοιο is always a man (Α 453: Chryses; Π 236: Achilles; Ι 509: generalized ‘men’) and his hearer a god (Α 453: Apollo; Π 236: Zeus; Ι 509: the Litai), the antecedents of αὐδήσαντος/αὐδησάσης are always men making exhortations or other non-sacral {37|38} speeches, and those who hear them are, with one exception, also men. [
54] Here is the exception:

δ 504 φῆ ῥ’ ἀέκητι θεῶν φυγέειν μέγα λαῖτμα θαλάσσης.
          τοῦ δὲ Ποσειδάων μεγάλ’ ἔκλυεν αὐδήσαντος·

As soon as Poseidon hears this boast of the lesser Ajax, he wreaks destruction upon him. [
55] So δ 505 is a line in which the formula ἔκλυες | εὐξαμένοιο # would be appropriate if εὐξαμένοιο could be used in εὔχομαι’s regular secular sense ‘boast’. But the poet will not use a formula designed for sacral contexts in secular ones. The freakish divine subject of ἔκλυεν in this secular context makes the use of αὐδήσαντος particularly felicitous to forestall confusion with the sacral formula. This example also shows clearly that the structure of the speech referred to by αὐδήσαντος/εὐξαμένοιο is a primary determinant of which word occurs. The divine or human status of its hearer is a secondary aspect of that structure. [56]

The substitution of αὐδήσαντος/αὐδησάσης for εὐχομένοιο/εὐξαμένοιο does more than verify the sacral-secular semantic split in εὔχομαι as against κλύω. It also implies that both αὐδάω and εὔχομαι are verbs of saying, one with sacral connotations, the other without. Here is another sign that εὔχομαι refers primarily to a concrete act, uttering words, just as σπένδω refers to making libation, σφάζω to slaughtering cattle, etc., and not to an element of the structure of these words, such as ‘wish’, ‘promise,’ ‘invoke,’ etc.

Three attestations in this section remain to be considered:

Π 531 ὅττι οἱ ὦκ᾿ ἤκουσεν μέγας θεὸς εὐξαμένοιο
Α 381 εὐξαμένου ἤκουσεν ἐπεὶ μάλα οἱ φίλος ἦεν
γ 55 κλῦθι, Ποσείδαον γαιήοχε, μηδὲ μεγήρῃς
       ἡμῖν εὐχομένοισι τελευτῆσαι τάδε ἔργα.

The first two expressions are not strictly formulas because of the switch in metrical position of εὐξαμένοιο/εὐξαμένου around the metrically fixed ἤκουσεν. But it is clear from other examples in the Homeric corpus that {38|39} ἤκουσεν is simply a lexical renewal of the petrified word ἔκλυεν in the formula ἔκλυεν εὐξαμένοιο#, and new flexibility for the participle is a correlate metrical renewal. [
57] The same phenomena recur together in the lines:

ι 497 εἰ δὲ φθεγξαμένου τευ ἢ αὐδήσαντος ἄκουσε
Π 76 οὐδέ πω Ἀτρεΐδεω ὀπὸς ἔκλυον αὐδήσαντος

(ι 497 is the only occurrence of αὐδήσαντος not in line-final position). The metrical and grammatical transposition εὐξαμένοιο # ~ # εὐξαμένου can be paralleled for many middle participles in Homer, which actually have three possible metrical positions:

ω 333 # οἰχόμενον~ ι 47 οἰχόμενοι || ~ οἰχομένοιο # α 281
η 40 # ἐρχόμενον~ λ 581 ἐρχομένην || ~ ἐρχομένοιο # α 408
Γ 307 # μαρνάμενον~ Ν 273 μαρνάμενος || ~ μαρναμένοιο # Ο 609

This situation suggests that the formula just discussed, viz.:

ἔκλυεν | εὐξαμένοιο #

is nothing more than an archaic transformation of (A) when it is stripped of elements attested elsewhere as independent formulas (ὣς ἔφατ᾿ and the adonic cadence, divine name + epithet/divine epithet). Compare the following:

(A) # – εὐχόμενος | τοῦδ᾿ ἔκλυε | – – ⏓ #
(C)                                     ἔκλυεν | εὐξαμένοιο #
      # – – ἤκουσεν || | εὐξαμένοιο #
      # εὐξαμένου ἤκουσεν ||

The semantic constraints on ἔκλυεν εὐξαμένοιο # do not in the least undermine a hypothesis that it is formally related to (A). Neither does the usage of # εὐξαμένου/εὐξαμένοιο # + ἤκουσεν, since in both attestations the hearer is a god (Α 381: Apollo; Π 531: Apollo) and the nouns to which # εὐξαμένου/εὐξαμένοιο # refer are men (Α 381: Chryses; Π 531: Glaukos). There is reason to believe, then, that the meaning of εὐξαμένου/εὐξαμένοιο # in this formula is identical to that in (A) and ἔκλυες εὐξαμένοιο #.

Perhaps it is legitimate to consider these three expressions one formula. When does a transformation destroy the functional identity of a formula? The relationship between ἔκλυες εὐξαμένοιο # and (A) is probably not synchronically perceptible, but consideration of the evidence which ἤκουσεν || + # εὐξαμένου/εὐξαμένοιο # provides for the sacral-secular semantic split clarifies this general problem. For the words recur together two other times:

Α 396 πολλάκι γάρ σεο πατρὸς ἐνὶ μεγάροισιν ἄκουσα
          εὐχομένης ὅτ’ ἔφησθα κελαινεφέϊ Κρονίωνι
          οἴη ἐν ἀθανάτοισιν ἀεικέα λοιγὸν ἀμῦναι … {39|40}
Φ 475 μή σευ νῦν ἔτι πατρὸς ἐνὶ μεγάροισιν ἀκούσω
          εὐχομένου, ὡς τὸ πρὶν ἐν ἀθανάτοισι θεοῖσιν,
          ἄντα Ποσειδάωνος ἐναντίβιον πολεμίζειν.

The subjects of ἀκούω and the nouns which εὐχομένου/εὐχομένης modify are all gods, and the speeches referred to are not prayers but boasts. Still these attestations do not violate the sacral-secular split. In both cases the ‘formula’ is enjambed, and the line-end is a boundary beyond which the formula might not extend. Compare the remarks of Parry on enjambed epithets:

and those of Hainsworth which are based on a slightly more comprehensive study of the phenomenon:

In Α 396 and Φ 475, therefore, the formula is

                                        || πατρὸς ἐνὶ μεγάροισιν ἄκουσα/ἀκούσω #

and εὐχομένου/εὐχομένης in the next line is an expansion or addition, not an integral part of it. Is this an example of the poet “drawing on his general resources of vocabulary,” to use Hainsworth’s phrase? On the contrary, it is possible, Hainsworth and Parry notwithstanding, for the poet to extend the formula beyond the line-end. Yet another case of this pairing of ἀκούω and εὔχομαι provides an example:

φ 210 οἴοισι δμώων· τῶν δ’ ἄλλων οὔ τευ ἄκουσα
          εὐξαμένου ἐμὲ αὖτις ὑπότροπον οἴκαδ’ ἱκέσθαι

{40|41} φ 211 is an instance of εὔχομαι + infinitive in sacral contexts which will be discussed in detail below, but what is of immediate concern is that here # εὐξαμένου retains the sacral sense it had when combined with ἀκούω within the line. (N. B. There are no examples of ἀκούω + secular εὔχομαι in the same line). I hypothesize that this is an extension of the formula, though a problematic one for the poet, and not an expansion, and that as far as he is concerned a formula usually does but may on occasion not lose its integrity when it crosses over the line-end. At such times an explanation is necessary, but it is surely no coincidence that the enjambed word in all three examples is also combined with εὔχομαι within the line twice. The poet’s mind is full of formulas, and the line-end is a somewhat fuzzy boundary between them.

This principle is confirmed in a striking way by the final attestation of absolute εὔχομαι, one which at first sight seems isolated lexically and contextually. It comes in the second line of a prayer:

γ 55 κλῦθι, Ποσείδαον γαιήοχε, μηδὲ μεγήρῃς
       ἡμῖν εὐχομένοισι τελευτῆσαι τάδε ἔργα.
       Νέστορι μὲν πρώτιστα καὶ υἱάσι κῦδος ὄπαζε,
       αὐτὰρ ἔπειτ’ ἄλλοισι δίδου χαρίεσσαν ἀμοιβὴν
       σύμπασιν Πυλίοισιν ἀγακλειτῆς ἑκατόμβης.
       δὸς δ’ ἔτι Τηλέμαχον καὶ ἐμὲ πρήξαντα νέεσθαι,
       οὕνεκα δεῦρ’ ἱκόμεσθα θοῇ σὺν νηῒ μελαίνῃ.

A clue to its provenance and meaning is preserved in an arcane source. Here is the first line of one of the Homeric ‘epigrams’ preserved in the Herodotean Vita Homeri 417 (Allen):

(D) κλῦθί μοι εὐχομένῳ κουρότροφε, δὸς δὲ γυναῖκα

The basic formal difference between this line and γ 55–6 is that the participle of εὔχομαι and its personal pronoun are dative singular in one place and plural in the other, a transformation which brings about enjambement of the latter. The close relationship of the epigram to γ 55–6 is evident, then, and perhaps we are dealing with another εὔχομαι formula designed for and fixed in sacral contexts. But if we carry the analysis a step further, the sacral meaning becomes more secure. For the first line of the epigram is only another transformation in the series we have been discussing, and γ 55–6 is a declension of it:

(A)                        εὐχόμενος || τοῦ δ᾿ ἔκλυε |
(C)                                                 ἔκλυεν | εὐξαμένοιο #
(D)           κλῦθί μοι εὐχομένῳ ||
γ 55-6          κλῦθι
                    ἡμῖν εὐχομένοισι ||

In this case, however, the participle of εὔχομαι is fixed and the κλύω form is mobile. For the switch

(A)                                        || τοῦ δ᾿ ἔκλυε |

                                                  vs. {41|42}

(D)                     # κλῦθί μοι …

compare the following:

1) α 125                              || ἡ δ᾿ ἕσπτετο


Β 484           # Ἔσπετε νῦν μοι …

2) Ε 655                              || ὁ δ’ ἀνέσχετο |


Ψ 587           # ἄνσχεο …

3) ζ117                                || ὁ δ’ ἔγρετο |


ο 46              # ἔγρεο …

It seems, then, that the enjambed formula in γ 55–6 does retain its identity. εὐχομένοισι || carries the same meaning as the other participles of εὔχομαι in (A), (C), and (D). It refers to the prayer in which it appears, a speech with the structure—invocation, grant of favor (the hecatomb mentioned at γ 59), request of favor—of speeches concluded by (A), and its meaning need be no more specific. It would be a mistake, for example, to say that since εὐχομένοισι appears in the invocation it means ‘invoking.’ Its formal relationship to (A) and a principle of economy applying to the evidence so far considered militate against such an interpretation.

Nevertheless, γ 55–6 requires explanation as an exception to the constraint against enjambed formulas. It is easy to justify this example as a product of the jeu des formules, in Parry’s phrase, yet the integrity of the formula might not be preserved beyond the end of a line. Someone, however, in the textual tradition may have been sensitive to this problem, for γ 56 is another place where the Mss. preserve a variant for a form of εὔχομαι (see also above, pp. 23–26). Allen’s d, i, j, and l families read:

γ 56 ἦμιν ἐπευχομένοισι || κ.τ.λ.

as opposed to the received text

# ἡμῖν εὐχομένοισι || κ.τ.λ.

The use of ἐπεύχομαι here resolves the enjambement problem, since it is not formulaic with κλῦθι, it can have the meaning ‘pray,’ and it is not subject to the rigorous contextual and phraseological constraints of εὔχομαι. In this situation, both readings are viable in terms of formulaic composition. ἐπευχομένοισι is eminently possible as a runover expansion of γ 55, and εὐχομένοισι is no less possible as a declension of # κλῦθί μοι εὐχομένῳ. The presence of both in the textual tradition is comprehensible if we assume rhapsodic transmission and the preservation of rhapsodic variants by sensitive or simply cautious Alexandrian scholars. On the other hand, it is not impossible that # ἡμῖν εὐχομένοισι is a scribal simplification of ἦμιν ἐπευχομένοισι, a mistake which happens to have remained because of the simple metrical switch ἦμιν/ἡμῖν (cp. ὣς ἔφατ᾿/ὣς φάτ᾿ in the previous example, p. 24). Yet it is a strikingly fortuitous error from the point of {42|43} view of formulaic composition, at a point where the technique crosses itself up exactly as in the previous cases (above, pp. 23–26). In any event, the provenance and meaning of εὐχομένοισι in γ 56, if it is the correct reading, is compatible with the other attestations of εὔχομαι in Section II.

2. εὔχομαι + dative




b. Rationale of classification

I have classified these twenty-seven attestations of εὔχομαι + dative on the basis of one criterion, the metrical shape of εὔχομαι in each. It emerges that classification on this basis coincides with classification by three other criteria, namely, line-placement, inflection, and the specific dative nouns which εὔχομαι governs. The result is a well-ordered system which is complex but nevertheless regular, and nineteen examples disposed into four classes of metrical shape display objective consistency with respect to all four criteria. Of the eight exceptions, five have an internal consistency of their own which can be accounted for and which also accounts for their exclusion from the system, and three are more complex special cases.

Our first purpose is to establish the efficacy of this classification scheme for the majority of the attestations. The five which belong in section I (γ 43, δ 752, ι 412, Ω 290, Υ 104) are consistent in inflection (2nd person singular present imperative), but the metrical position of εὔχομαι seems irregular and the collection of proper names in the dative heterogeneous, though they are metrically parallel in shape and placement. [60] However, comparison with the proper names in sections II and III reveals that a consistent group of three gods plus the collective θεοῖς αἰειγενέτῃσιν # are the only ones which occur. [61] As to the flottement in the position of εὔχεο around the beginning of the line, this is a common phenomenon for line-initial phraseology in Homer which finally has an explanation in the work of G. Nagy. In his words, {45|46}

Phraseological improvisation in this part of the line is the phenomenon which lies behind the fluctuation in meaning and vocabulary of the enjambed elements we have been discussing, and which Hainsworth describes as the poet “drawing on the general resources of his vocabulary.” Rather enjambement is improvisation, and improvisation does not preclude the use of fixed material. Instead it accounts for inconsistency in its use. Note that in our examples εὔχεο is always displaced by the same elements—ἀλλά, σύ, and γε—yet their order is not fixed. This mixture of fixed and free elements is a token of the phenomenon to which Nagy gives a name and which is again exemplified by the one instance of enjambement in this section. In Υ 104:

ἥρως ἀλλ’ ἄγε καὶ σὺ θεοῖς αἰειγενέτῃσιν

the x-sector phraseology # ἥρως ἀλλ᾿ ἄγε καὶ σὺ… has simply displaced εὔχεο to the beginning of the next line, after the fixed expression θεοῖς αἰειγενέτῃσιν # which it normally precedes.

When εὔχομαι + dative has the shape [[εὐχ –]] it occurs in three places in the line:

a)                         || ⏓ [[εὐχ –]]
b)                         || ⏑   [[εὐχ –]]
c) # [[εὐχ –]]

The relationship between a) and b) is a function of Severyns’ law, [
64] whereby there is interchange between formulas after the penthemimeral and trochaic caesuras, e.g.:

{46|47} The phraseological correspondence in one of our examples of P2 is less clear cut but nonetheless evident:


where Λ 736 includes in half a line the names of two divinities by deleting their epithets and putting one (Διί) in the slot between the trochaic caesura and [[εὐχ –]] and the other (καὶ Ἀθήνῃ) between [[εὐχ –]] and the end of the line. This sounds very complex, but an analogous process occurs in one attestation of # [[εὐχ –]] whereby the whole line is used to incorporate the same two divinities, this time with epithets:

ω 518 εὐξάμενος κούρῃ γλαυκώπιδι καὶ Διὶ πατρί

The other attestation of this placement, which is contextually related to this one, conflates the T2/P2 formula with a section I formula:

The origin of ω 521 in this process is significant, since it accounts for the metrical position of # [[εὐχ –]] as being a metrical and phraseological variant of # [[εὐχ ]] and thereby implies an interrelationship in the poet’s mind between two sections in the group εὔχομαι + dative. The formal similarities of sections I, II, and III suggest a horizontal relationship between them, so that it is perhaps legitimate to reduce all three to a system as follows:

To repeat, ω 518, by virtue of the metrical placement # [[εὐχ –]], and ω 521, by virtue of this and its dative name-epithet formula (|| Διὸς κούρῃ μεγάλοιο #) are tokens of a link in the poet’s mind between B) and A). Furthermore, the two placements # [[εὐχ ]] (inflectional fixation: εὔχεο) {47|48} and [[εὐχ– ]], || (inflectional fixation: εὔχοντο) are related as ritual prescription (imperative) and ritual description (imperfect indicative). The phraseological correspondence between these two ‘modes’ of the ritual narrative sub-genre has many parallels (see n. 40, n. 63 and p. 32f.) and, indeed, instances of it occur within sections I (γ 394: εὔχετ᾿, see below, p. 55) and II (Η 194: εὔχεσθε, cp. Η 200).

The last section comprises four enjambed attestations of # [[εὐχ– –]]. There is a mixture of inconsistency and consistency in # [[εὐχ– –]], whose metrical shape and position are fixed, but whose inflection is the same only when the dative nouns are the same:

IV.       γ 47                                          ἀθανάτοισιν
                    # εὔχεσθαι·
          Ψ 546                                         ἀθανάτοισιν#
                    # εὔχεσθαι·
          μ 333                                         θεοῖσιν#
                    # εὐξαίμην
          HΑpoll 237                                  ἄνακτι#
                    # εὔχονται

With so few attestations to analyze, this may be a coincidence, but the inflectional variety of this section does present a contrast in type and frequency to the relative fixity of the others:

I. εὔχεο throughout
II. εὔχοντο (3x) with one traditional (n. 63) variation, εὔχεσθε.
III. εὐξάμενος (4x) ~ εὐξάμενοι (1x) ~ εὐχόμενοι (1x).

Finally, the dative nouns paired with [[εὐχ ]] are lexically and in metrical shape distinct from those in the other sections. All this suggests that IV is not only improvisatory in itself but also vis-à-vis the system, which it overlaps neither in shape, phraseology, nor morphology. IV’s function is to compensate for the system’s limitations. (Coincidentally, this implies the functional reality of the system to the epic poet.) But nothing about section IV necessitates or suggests a different semantic function for εὔχομαι.

It is possible, then, to reduce these nineteen attestations to a coherent system with an improvised sub-system. The eight remaining attestations (p. 44f., V), which do not convincingly fit the system or any of its parts, are at first sight heterogeneous. But six of them are divergent in the same way. In ν 230 (σοί), Υ 451 = Α 364 and Α 86 (ᾧ) H 298 (μοι) and HApoll 495 (ἐμοί), the complements to εὔχομαι are not the names of divinities or a collective for them, but personal or relative pronouns.

The key to the ‘irregularity’ of these lines lies in the use of this pronoun. Its metrical placement is unfixed and the forms of εὔχομαι which govern it are irregular in inflection and position (ν 231: # εὔχομαι; Υ 451 = Λ 364: # εὔχεσθαι; A 87: # εὐχόμενος; Η 298: εὐχόμεναι ||), though at times one or the other coincides with a systematic usage. But it happens that the {48|49} antecedent of these pronouns is, in four (Υ 451 = Λ 364, Α 87, HApoll 495) of the six, the same god, Apollo. This coincides with another fact. Apollo is conspicuously absent from the divine names in the dative which occur in sections I, II, and III:


Yet as these pronouns and his presence in the adonic cadence of (A) attest, he is one of the gods to whom heroes normally pray. And furthermore he has two name-epithet formulas which have the same metrical shape as those in the preceding list:

Α 438, Π 513, Ψ 872                    || ἑκηβόλῳ Ἀπόλλωνι #
φ 267                                         || Ἀπόλλωνι κλυτοτόξῳ#

Unfortunately, both of these expressions begin with a vowel, while the name-epithet formulas which are used in the system begin with a consonant. [
65] Moreover, in the systematic usages the word which occurs before the caesura always ends with a vowel when it is followed by a formula of this type (I, II, III). By contrast, the words which occur before Apollo’s two dative formulas always end with a consonant (βῆσαν || Α 438, εἶπεν || Π 513, ἡπείλησεν || Ψ 872, θέντες || φ 267). Twice this consonant is a ν-movable before ἑκηβόλῳ, and there is independent evidence that this word began with a digamma, [66] but ostensibly the poet has no awareness of it. To use Apollo’s name systematically with εὔχομαι the poet would have to commit hiatus. Rather than do this, he unconsciously contrives to make Apollo the antecedent of a monosyllabic or disyllabic dative pronoun with which he can improvise: one of the examples is enjambed (Α 87), while the pronoun in the others (HApoll 495, Υ 451 =Α 364) occurs in the first foot. Thus the irregularity of these usages is a result of the narrowness of a system which they cannot enter, not of a breakdown within it. There is no reason to suppose that εὔχομαι has a different function here than it does elsewhere.

There are, however, two anomalous attestations in which the antecedents of the dative pronouns are not Apollo, and there is functional divergence in them (p. 45, V, B): {49|50}

Η 298 αἵ τέ μοι εὐχόμεναι θεῖον δύσονται ἀγῶνα
ν 230                                           σοὶ γὰρ ἐγώ γε
          εὔχομαι ὥς τε θεῷ καί σευ φίλα γούναθ’ ἱκάνω

The antecedent of μοι in Η 298 is Hector, which is extremely anomalous. This is the only place in all the Homeric corpus (including εὔχομαι in secular contexts) where a dative noun after εὔχομαι is not a god or collection of gods. I conclude that this does not demote εὔχομαι from sacral to secular significance, but that it promotes Hector from man to ‘god’. This is not surprising. Homeric heroes occupy an ambiguous place on the human-divine axis to begin with. [
67] In the passage where this line occurs, Hector is asserting his superiority to Ajax. “Let us stop fighting, since it’s dark, and part as friends,” he says. But the lines which follow belie his protestations of their equality:

H 294 ὡς σύ τ’ ἐϋφρήνῃς πάντας παρὰ νηυσὶν Ἀχαιούς,
          σούς τε μάλιστα ἔτας καὶ ἑταίρους, οἵ τοι ἔασιν·
          αὐτὰρ ἐγὼ κατὰ ἄστυ μέγα Πριάμοιο ἄνακτος
          Τρῶας ἐϋφρανέω καὶ Τρῳάδας ἑλκεσιπέπλους,
          αἵ τέ μοι εὐχόμεναι θεῖον δύσονται ἀγῶνα.

The real impact of this parallelism is that Hector is a ‘god’ while Ajax is merely a man. It is a powerful threat concealed as a compliment. So it is not surprising that, in a generous gesture of friendship, Hector proceeds to offer Ajax his sword as a gift. Ajax, beguiled, accepts. It is the sword upon which he throws himself in Sophocles’ play:

Ἐγὼ γάρ, ἐξ οὗ χειρὶ τοῦτ’ ἐδεξάμην
παρ’ Ἕκτορος δώρημα δυσμενεστάτου,
οὔπω τι κεδνὸν ἔσχον Ἀργείων πάρα·

(Soph. Ajax 661–663)

Furthermore, Hector’s ‘godhead’ is mentioned elsewhere in the Iliad. Achilles speaking:

Χ 393 ἠράμεθα μέγα κῦδος· ἐπέφνομεν Ἕκτορα δῖον,
         ᾧ Τρῶες κατὰ ἄστυ θεῷ ὣς εὐχετόωντο.

Here the explicitness of Η 298 is mollified by a simile. The same happens in two other speeches of Hector:

Θ 538 εἰ γὰρ ἐγὼν ὣς
          εἴην ἀθάνατος καὶ ἀγήρως ἤματα πάντα,
          τιοίμην δ’ ὡς τίετ’ Ἀθηναίη καὶ Ἀπόλλων,
          ὡς νῦν ἡμέρη ἧδε κακὸν φέρει Ἀργείοισιν.

{50|51} and speaking abusively to Ajax:

Ν 825 εἰ γὰρ ἐγὼν οὕτω γε Διὸς πάϊς αἰγιόχοιο
          εἴην ἤματα πάντα, τέκοι δέ με πότνια Ἥρη,
          τιοίμην δ’ ὡς τίετ’ Ἀθηναίη καὶ Ἀπόλλων,
          ὡς νῦν ἡμέρη ἧδε κακὸν φέρει Ἀργείοισι
          πᾶσι μάλ’, ἐν δὲ σὺ τοῖσι πεφήσεαι …

But the nakedness of his claim other men lay bare. Poseidon/Calchas speaking:

N 54 Ἕκτωρ ὃς Διὸς εὔχετ’ ἐρισθενέος πάϊς εἶναι.

As will be shown below, the formula εὔχομαι εἶναι is always used to state a man’s actual pedigree, not his genealogical pretensions. [

{51|52}Two attestations of εὔχομαι + dative which do not fit the system remain to be discussed. Both are combinations of εὔχομαι + dative formulas we have already discussed with altogether different formulas. The first combines the improvisational ‘formula’ of section III:

μ 333                                                   θεοῖσιν #
          # εὐξαίμην
γ 47                                                     ἀθανάτοισιν #
          # εὔχεσθαι
Ψ 546                                                   ἀθανάτοισιν #
          # εὔχεσθαι

with an expression analogous to those in the following series:

π 313 … φράζεσθαι                ἄνωγα # (5x)
π 316 … δεδάασθαι                ἄνωγα # (1x)
π 405 … παύσασθαι                ἄνωγα # (lx)
π 446 … τρομέεσθαι               ἄνωγα # (lx)
Ρ 357 … χάζεσθαι                   ἄνωγει # (lx)
Η 74 … μαχέσασθαι                ἄνωγει # (2x)
Λ 15 … ζώννυσθαι                  ἄνωγεν # (lx)
Π 8 … ἀνελέσθαι                     ἄνωγει # (lx)
Ο180 … ὑπεξαλέασθαι             ἄνωγε # (lx)
Α 313 … ἀπολυμαίνεσθαι         ἄνωγεν # (lx) produce the line:

*Ζ 240 καὶ πόσιας· ὁ δ᾿ ἔπειτα θεοῖς εὔχεσθαι ἀνώγει

The second combines the ritual narrative collocation of εὔχομαι and σπένδω (see above, p. 34f.) with a systematic formula of the type § I:

δ 752 εὔχε᾿ Ἀθηναίῃ κούρῃ Διὸς αἰγιόχοιο

to give:

*γ 394 εὔχετ᾿ ἀποσπένδων κούρῃ Διὸς αἰγιόχοιο

Another example of this combination which does not result in ‘irregularity’ with respect to the system is:

*Π 253 ἤτοι ὁ μὲν σπείσας τε καὶ εὐξάμενος Διὶ πατρί

(§ III) Here the jeu des formules results in a line which coincides with systematic usage of εὔχομαι , but in γ 394 (isolated inflection) and Ζ 240 (isolated metrical shape and position) it produces irregularities. Thus the formal anomalies of these two lines do not imply that the meaning of εὔχομαι in them is different from its meaning in Π 253 and the other systematic usages.

c. Meaning of εὔχομαι

All the examples in this group support the hypothesis of a sacral-secular split in the meaning of εὔχομαι, for none of the expressions used recur in secular contexts. In fact there are no examples at all of εὔχομαι in secular {52|53} contexts followed by a dative. The simple syntactic pattern εὔχομαι + dative is restricted to sacral contexts exclusively. Again, there is nothing intrinsically sacral about this syntax, but if there is something intrinsically sacral about εὔχομαι then the restriction and its inviolability make sense.

There is a difficulty, however, in this explanation. No actual prayers in the Iliad or Odyssey are addressed to ‘the gods’ collectively, while θεοῖς, ἀθανάτοισι, κ. τ. λ. are regular complements to εὔχομαι. But it is reasonable to assume that the contrast between particular and collective gods is a contrast between marked and unmarked categories. The examples of εὔχομαι + θεοῖς αἰειγενέτῃσιν# κ. τ. λ. are cases where the speaker or narrator chooses not to specify which god is prayed to (unmarked), whereas the rest do so specify (marked).

In conclusion, then, εὔχομαι in these attestations has the same meaning it does in (A) and when used absolutely. It signifies the utterance of a prayer with canonical tripartite structure.

3. εὔχομαι + infinitive

a. Classified List of Attestations

I. Future infinitive


{53|54} II. Aorist infinitive

Ω 287 τῆ σπεῖσον Διὶ πατρί. καὶ εὔχεο οἴκαδ᾿ ἱκέσθαι
φ 210                       τῶν δ᾿ ἄλλων οὔ τευ ἄκουσα
          εὐξαμένου ἐμὲ αὖτις ὑπότροπον οἴκαδ᾿ ἱκέσθαι
Ο 372 Ζεῦ πάτερ, εἴ πότε τίς τοι ἐν Ἄργεϊ περ πολυπύρῳ
          ἢ βοὸς ἢ οἰὸς κατὰ πίονα μηρία καίων
          εὔχετο νοστῆσαι, σὺ δ᾿ ὑπέσχεο καὶ κατένευσας
          τῶν μνῆσαι καὶ ἄμυνον, Ὀλύμπιε, νηλεὲς ἦμαρ …
Β 400 ἄλλος δ᾿ ἄλλῳ ἔρεζε θεῶν αἰειγενετάων
          εὐχόμενος θάνατόν τε φυγεῖν καὶ μῶλον Ἄρηος.
Ι 182 Τὼ δὲ βάτην παρὰ θῖνα πολυφλοίσβοιο θαλάσσης
          πολλὰ μάλ’ εὐχομένω γαιηόχῳ ἐννοσιγαίῳ
          ῥηϊδίως πεπιθεῖν μεγάλας φρένας Αἰακίδαο.
ο 353 Λαέρτης μὲν ἔτι ζώει, Διὶ δ’ εὔχεται αἰεὶ
          θυμὸν ἀπὸ μελέων φθίσθαι οἷσ’ ἐν μεγάροισιν·

b. Section I attestations

It is impossible to justify this classification as we have the two which precede it. Except in the doublet passages, which are special cases (see below), εὔχομαι occurs without convincing fixation to a specific infinitive. There is no formula or formula system composed of εὔχομαι + infinitive. In order to determine the meaning of εὔχομαι in these passages, our only recourse is to analyze individually their similarities and differences with respect to the classifiable usages. Then we can decide whether or not they demand modification of the hypothesis we have made about the meaning of εὔχομαι. This is the method we have employed above with regard to exceptional or anomalous usages.

The question then becomes, what similarities do these passages have to other usages of εὔχομαι? Notice that both doublets begin with the pattern εὔχομαι + dative which we identified earlier as an inviolable mark of sacral usage. This in itself calls into doubt the hypothesis of a new meaning of εὔχομαι. For once the syntactic pattern εὔχομαι + dative is articulated in the poetic line, the meaning of the word εὔχομαι is determined by all the other (twenty-nine) occurrences of εὔχομαι + dative in which the word seems to mean simply ‘pray’. [76] But there are still deeper similarities. The doublets in each case differ only in the inflection of εὔχομαι, which changes from imperative (# εὔχεο) to imperfect indicative (# εὔχετο). This is the traditional switch from ritual prescription to ritual description within the ritual narrative sub-genre attested in the εὔχομαι + dative system (see above, p. 48) and elsewhere. Finally, and most significantly, both doublets have the tripartite structure: invocation (πᾶσι θεοῖσι, Ἀπόλλωνι Λυκηγενέϊ κλυτοτόξῳ) followed by grant of favor (τεληέσσας ἑκατόμβας ῥέξειν, ἀρνῶν πρωτρωγόνων ῥέξειν κλειτὴν ἑκατόμβην) and request of favor (αἴ κέ ποθι Ζεὺς ἄντιτα ἔργα τελέσσῃ, οἴκαδε νοστήσας ἱερῆς εἰς ἄστυ {55|56} Ζελείης). [77] In view of this we are entitled to doubt altogether if the peculiarities (future infinitive with dependent request of favor) which are the basis of the accepted second meaning “vow” are peculiarities at all. For there are quoted prayers in the Homeric corpus in which the grant of favor is in the future tense, e. g.:

ἀλλά, ἄνασσ’, ἵληθι, δίδωθι δέ μοι κλέος ἐσθλόν,
αὐτῷ καὶ παίδεσσι καὶ αἰδοίῃ παρακοίτι·
σοὶ δ’ αὖ ἐγὼ ῥέξω βοῦν ἤνιν εὐρυμέτωπον,
ἀδμήτην, ἣν οὔ πω ὑπὸ ζυγὸν ἤγαγεν ἀνήρ·
τήν τοι ἐγὼ ῥέξω χρυσὸν κέρασιν περιχεύας.”
ὣς ἔφατ’ εὐχόμενος, τοῦ δ’ ἔκλυε Παλλὰς Ἀθήνη

γ 380-385

Here, it is true, the request of favor is not subordinate to the grant of favor, but in parataxis with it. Still, the parataxis casts doubt on the assumption implicit in postulating the meaning ‘vow’ that elsewhere the request of favor is conceptually basic to Homeric prayer. In reality there is as much reason to postulate ‘vow’ as the meaning of εὐχόμενος in γ 385—which no one has done—as there is to postulate it for εὔχεο/εὔχετο in the doublets. Furthermore, a prayer exists in which the request of favor feature is subordinate to a future (in time: grammatically, the subjunctive grant of favor [
78] —exactly as in the doublets:

πότνι’ Ἀθηναίη ῥυσίπτολι δῖα θεάων
ἆξον δὴ ἔγχος Διομήδεος, ἠδὲ καὶ αὐτὸν
πρηνέα δὸς πεσέειν Σκαιῶν προπάροιθε πυλάων,
ὄφρά τοι αὐτίκα νῦν δυοκαίδεκα βοῦς ἐνὶ νηῷ
ἤνις ἠκέστας ἱερεύσομεν, αἴ κ’ ἐλεήσῃς
ἄστυ τε καὶ Τρώων ἀλόχους καὶ νήπια τέκνα.
Ὣς ἔφατ’ εὐχομένη, ἀνένευε δὲ Παλλὰς Ἀθήνη.

Ζ 305-311

This prayer begins with a request of favor, has a grant of favor dependent upon it, and a further request dependent upon the grant of favor. One could hardly wish for a clearer sign that grant of favor can be dependent upon request of favor and vice versa.

It remains to account for the other undeniable peculiarities in these passages. Why the unique invocation? Why the unique parallelism in prayer structure? An answer suggests itself when we ponder another peculiar fact, that they are doublets in spite of there being no apparent obstacle to the use of any of these passages as singletons. The existence of such doublets makes sense in terms of poetic performance. They are virtuoso pieces in which the composer with formulas displays his ability not simply to form single lines from smaller units but to re-use with elegantly slight alteration (εὔχεο  εὔχετο) whole groups of lines. This taxes his memory and therefore makes them prized. The unique features are typical of the language of poetic sub-genres, each having its lexical, syntactic, and, in this case, compositional idiosyncrasies. And the unique features are shared because doublet language generates more doublet language, though it is based on—in fact it is a specialization of—other epic language.

c. Section I attestations (continued): Θ 526

If we accept Allen’s text, this passage is prima facie more anomalous than the ones just discussed. Like them, it has a unique invocation (Διί τ᾿ ἄλλοισιν τε θεοῖσιν #) and a dependent infinitive in the future, but unlike them and anything else in the corpus of Homer, the future infinitive is the verb of the request of favor clause. Conceptually, this presents no difficulties. In fact, we have just cited examples of request clauses in actual prayers in the subjunctive, a mood whose distinctness from the future tense cannot be pressed too far in this connection. Still it is worthy of notice that all other existing examples of εὔχομαι + infinitive of request (= all the passages in Section II above, p. 54) have an aorist and not a future infinitive. This seems especially significant since Θ 526 f. is structurally identical to the B passages and not to the doublets. Note the combination: dependent request infinitive, omission of grant of favor feature in Θ 526f., Ω 287, φ 211, Ι 183, and ο 353 and that there are no doublets for Θ 526f. and all the passages in B. The significance of all these items is difficult to assess. At least we can say that the peculiarities of Θ 526 f. are not easily explicable as sub-genre phenomena.

But the problems do not end here. For Allen’s text of Θ 526 contains one of only four examples of sacral εὔχομαι in the present indicative, as against 86 examples in past tenses, or tenseless imperatives, infinitives, moods and participles. The reason for this constraint against the present tense of sacral εὔχομαι is clear. The most common formulas with secular εὔχομαι are in the present tense (44 instances). This is an important and dramatic grammatical fact supporting the sacral-secular semantic split, and it is a constraint of the same type as the restriction of εὔχομαι + dative to the sacral sphere. There is no reason why εὔχομαι in the present tense cannot mean ‘pray’, just as there is no reason why εὔχομαι + dative cannot mean ‘boast’—unless the rule we have posited about the restriction of sacral formulas to sacral contexts has a converse, that secular formulas are restricted to secular contexts, and this restriction is implemented by or implementing a split in meaning.

It is true, however, that there are three other places where the present tense does occur in sacral contexts without textual variants. Not {58|59} unexpectedly, this restriction is so great as to be unstable. Even so, any sacral occurrences of present tense εὔχομαι must be textually suspect or due to obvious transformation. Is Θ 526 one or the other? To answer this question, we must detour for a moment and gain perspective from an analysis of the other attestations of sacral εὔχομαι in the present tense, which are as follows:

HApoll 237                                                   οἱ δὲ ἄνακτι #
          # εὔχονται
ν 230                                                           σοὶ γὰρ ἐγώ γε #
          # εὔχομαι ὣς τε θεῷ, …
ο 353 Λαέρτης μὲν ἔτι ζώει, Διὶ δ’ εὔχεται αἰεὶ
          θυμὸν ἀπὸ μελέων φθίσθαι οἷς ἐν μεγάροισιν·

HApoll 385                                                   ἔνθα δ᾿ ἄνακτι #
          # πάντες ἐπίκλησιν Τελφουσίῳ εὐχετάονται
ν 230                                                           σοὶ γὰρ ἐγώ γε #
          # εὔχομαι ὣς τε θεῷ, …
Χ 394 ᾦ Τρῶες κατὰ ἄστυ θεῷ ὣς εὐχετόωντο
θ 467 = o 181 τῷ κέν τοι καὶ κεῖθι θεῷ ὣς εὐχετοῴμην

Clearly the passages with enjambed, present tense, sacral εὔχομαι are transformations of those with εὐχετάομαι. The interchange of εὐχετάομαι with present tense εὔχομαι is a palpable sign of the epic tradition’s sensitivity to the constraint we have posited.{59|60}

Moreover, the same switch from Διὶ πατρί to Διί τ᾿ ἄλλοισίν τε θεοῖσιν is attested with a variation of this formula:


that uses ἐπεύχομαι instead of εὔχομαι. Compare:

Ζ 475 # – ἐπευξάμενος || Διί τ᾿ ἄλλοισόν τε θεοῖσιν
Γ 350, Ρ 46                        || ἐπευξάμενος Διὶ πατρί #


Θ 526 # εὔχομαι …             || Διὶ τ᾿ ἄλλοισίν τε θεοῖσιν #

So it is possible that the present tense of εὔχομαι in Θ 526 is genuine epic diction, [
85] and more so since the last example of sacral, present tense εὔχομαι, namely:

o 353 Λαέρτης μὲν ἔτι ζώει, Διὶ δ’ εὔχεται αἰεί

{60|61} may be a transformation of the same formula. We can suppose that the semantically unnecessary epithet πατρί # has been deleted and replaced by the word αἰεί, which is contextually à propos and metrically necessary to corrept the last syllable of εὔχεται. [
86] The position of Διί changes to accommodate the inflectional change of εὐξάμενος to εὔχεται. Compare:


Λ 736                                || Διί τ᾿ εὐχόμενοι καὶ Ἀθήνῃ #
o 353                                || ⏕ – Διὶ δ᾿ εὔχεται αἰεί #

where in Λ 736 a similar positional switch is accomplished to accommodate a connective and Athenas name.

In conclusion, then, defending Allen’s text at Θ 526 is difficult because of its lexical, grammatical, and structural peculiarities, but none of them are impossible to accept. On the other hand, the reading of Zenodotus is not at all problematic. His version of the line:

ἔλπομαι εὐχόμενος Διί τ᾿ ἄλλοισίν τε θεοῖσιν

removes the troublesome present-tense, sacral εὔχομαι and has a precisely paralleled transformation. Compare:

Γ 350, Ρ 46                        || ἑπευξάμενος Διὶ πατρί#
Ζ 475 … ἐπευξάμενος          || Διί τ᾿ ἄλλοισίν τε θεοῖσιν#
Zenodotus … εὐχόμενος || Διί τ᾿ ἄλλοισίν τε θεοῖσιν#

Allen’s line initial # εὔχομαι involves an additional metrical and inflectional switch vis-à-vis Γ 350 and Ζ 475 which is neither precisely paralleled nor impossible. [
87] The same is true of ἐλπόμενος || in Allen’s text, which occurs everywhere else in line-initial position (γ 228, Γ 112, Ξ 422, Π 281, Υ 180, Σ 260), while # ἔλπομαι, Zenodotus’ reading, occurs four times in exactly that position (HDion 28, Η 199, Η 353, Ρ 239). Finally, though the collocation of εὔχομαι and ἐξελάαν is lexically unparalleled and structurally {61|62} unique (see above, p. 58), the collocation of ἔλπομαι and ἐξελάαν (Zenodotus) has an almost exact parallel:

Ρ 495-6                                         || μάλα δέ σφισιν ἔλπετο θυμός
            αὐτώ τε κτενέειν ἐλάαν τ’ ἐριαύχενας ἵππους

So given these specific word patterns and the formular ethos of Homeric language, it is much more likely that the epic tradition generated Zenodotus’ reading than Allen’s. It is also simpler to believe that the traditional segment # ἔλπομαι εὐχόμενος was transposed to # εὔχομαι ἐλπόμενος by a scribe or, still more believably, a performer, [
88] —i.e. to suppose that Allen’s text is a banal slip—rather than that the tradition produced # εὔχομαι ἐλπόμενος from its own repertoire via the complex road we have reconstructed and instead of taking the shorter route to # ἔλπομαι εὐχόμενος. It is possible, one might add, that Homer himself made the banal slip, but for our purposes it does not matter who did. The line ceases to be a valid example of εὔχομαι + future infinitive. Instead it falls neatly into section III of the εὔχομαι + dative (above, p.43 f.) attestations without necessitating any modification in the word’s meaning.

d. Section II. attestations

As we have said, the lack of convincing formulas consisting of εὔχομαι + a specific infinitive (except, perhaps, in the doublet passages) dictates an individualized approach to the study of the word’s meaning in these attestations. But dissatisfaction with this methodological inconsistency tempts one to ask why there are no formulas in the first place. The doublets, we can say, do not share formulas with other attestations because they belong to a distinct poetic sub-genre. By the same token, the language in both of them is similar. But from the point of view of formulaic composition, the remaining attestations present a disconcerting picture. What has happened to the principle of dictional economy? Actual prayers provide a parallel but explicable dilemma. While their underlying structure is discernibly formalized, they vary considerably in language. On the one hand, this lack of verbal fixity is an aesthetic result of the variety of favors which men request and offer, and the elaborateness of the titles by which gods are invoked; on the other hand, it is a function of the small size of the compositional units which the epic poet has at his disposal. Large sized units are structurally fixed, and may use similar language, but the assembly of formulas into units longer than single lines which are repeated verbatim is a conspicuous, virtuosic act and not the norm. Prayers are examples of “composition by theme,” to use the term of Albert Lord, while the doublets are themes which have become gross formulas.

Let us demonstrate this in detail. The provenance of the first attestation, Ω 287, is evident from the following series:

γ 45 αὐτὰρ ἐπὴν σπείσῃς τε καὶ εὔξεαι, ἣ θέμις ἐστί
Π 253 ἤτοι ὁ μὲν σπείσας τε καὶ εὐξάμενος Διὶ πατρί
Ω 287 τῆ σπεῖσον Διὶ πατρί, καὶ εὔχεο οἴκαδ᾿ ἱκέσθαι

In γ 45 εὔξεαι occurs in absolute usage and in collocation with σπένδω (ritual narrative context). In Π 253, the collocation is conflated with the formula || ⏕ εὐξάμενος Διὶ πατρί (again, ritual narrative context). Ω 287 (ritual prescription) juxtaposes a metrical transposition of this conflation (the movement of Διὶ πατρί # to Διὶ πατρί || with the separately attested formula, || οἴκαδ᾿ ἱκέσθαι # (six times in other contexts).

Similarly, φ 210–211 is a member of the following series:

(A) # – – εὐχόμενος || τοῦ δ᾿ ἔκλυε | – – ⏓ #
(C)                                          ἔκλυες | εὐξαμένοιο #
(D) # κλῦθι …       εὐχομένῳ ||
Π 531                    ἤκουσεν ||                εὐξαμένοιο #
Α 381 # εὐξαμένου ἤκουσεν ||
φ 210                                                    ἄκουσα#
          # εὐξαμένου ἐμὲ αὖτις ὑπότροπον οἴκαδ᾿ ἱκέσθαι. #

In it an enjambed version of # εὐξαμένου/εὐξαμένοιο # + ἤκουσεν is juxtaposed with οἴκαδ᾿ ἱκέσθαι #.

Parallel to this usage of enjambed, line-initial εὔχομαι is Β 401, which belongs to a series: {63|64}


Here # εὐχόμενος is an x-sector improvisation combined with an infinitive phrase fixed in the rest of the line.

In two attestations, this situation is inverted: a line containing an infinitive is enjambed with a line containing εὔχομαι:

Ι 183 πολλὰ μάλ’ εὐχομένω γαιηόχῳ ἐννοσιγαίῳ
         ῥηϊδίως πεπιθεῖν μεγάλας φρένας Αἰακίδαο.
ο 353 Λαέρτης μὲν ἔτι ζώει, Διὶ δ’ εὔχεται αἰεὶ
         θυμὸν ἀπὸ μελέων φθίσθαι οἷς ἐν μεγάροισιν·

The lines containing εὔχομαι are both complex transformations of its formulas. For the first (Ι 183), compare the section I and III εὔχομαι + dative formulas:

(I) γ 43 εὔχεο νῦν, ὦ ξεῖνε, Ποσειδάωνι ἄνακτι
(III) ω 521 εὐξάμενος δ’ ἄρ’ ἔπειτα Διὸς κούρῃ μεγάλοιο

The name-epithet formula || γαιηόχῳ ἐννοσιγαίῳ # is the P2 variant of T2 || Ποσειδάωνι ἄνακτι #. (For this notation, see n.64.) But the position of εὐχομένω| | in Ι 183 is not elsewhere attested in the εὔχομαι + dative formulas to report prayers. The line has greater similarities to one used to introduce a prayer:

γ 54 αὐτίκα δ’ εὔχετο πολλὰ Ποσειδάωνι ἄνακτι

As for ο 353, we have already shown (p. 60f., above) that its post-caesural component is a possible but complex and precedent-breaking transformation of:


The infinitive in the lines enjambed to these (πεπιθεῖν || μεγάλας φρένας, I 184; || φθίσθαι οἷς ἐν μεγάροισιν #, ο 354) are again formulas with an independent existence, as their distance from εὔχομαι implies. Compare:

HAphr 7 τρισσὰς δ’ οὐ δύναται πεπιθεῖν φρένας οὐδ’ ἀπατῆσαι·
HApoll 275 ὣς εἰποῦσ’ Ἑκάτου πέπιθε φρένας, ὄφρα οἱ αὐτῇ
Ν 667 νούσῳ ὑπ’ ἀργαλέῃ φθίσθαι οἷς ἐν μεγάροισιν,

In sum, only one of these five aorist infinitives is not a hapax collocation with εὔχομαι, and all of them exist in other contexts; simultaneously, the analysis of εὔχομαι expressions in these lines amounts to a catalogue of the transformational and enjambement devices the poet has at his disposal. And the sixth example is not susceptible to formulaic analysis at all:

Ο 372 Ζεῦ πάτερ εἴ ποτέ τίς τοι ἐν Ἄργεΐ περ πολυπύρῳ
          ἢ βοὸς ἢ οἰὸς κατὰ πίονα μηρία καίων
          εὔχετο νοστῆσαι …

where the final phrase is without parallels.

{64|65}What account can we give of this situation? To begin with, it is true that the semantic reference of four of the infinitives (see above, p. 63) is not so diverse as to render their accession into a formula system impossible. But the other two (Ι 184 and ο 354) are not semantically compatible with them, so this view may be a distortion. It is important to give proper weight to the relative simplicity of formula fixation for the systematic usages, absolute εὔχομαι and εὔχομαι + dative. With absolute εὔχομαι, no potentially variable elements (such as favors granted or requested) need occur within the formula. In the case of εὔχομαι + dative of god, only four gods are prayed to, and, even so, the name-epithet system provides easy material for the construction of a system which includes only three. Development of a systematic language to incorporate all the elements of prayer structure—invocation, grant of favor, request of favor—is far more difficult. This makes it less surprising that no εὔχομαι + infinitive formula exists. Furthermore, it is an overstatement to say this. No strictly defined formula exists, but the combination εὔχομαι + οἴκαδ᾿ ἱκέσθαι # occurs twice, and # εὔχετο νοστῆσαι is at least a potential formula. It seems that we are witnessing in these attestations the still-birth of a formula system. To use theoretical terms, the complex thematic composition of actual prayers and compendious formulaic composition of reported prayers are at odds with one another. This is reflected by the formulaic analysis, which shows the poet and his medium extending themselves to report prayers in no more than two lines, [90] but without attaining verbal consistency except momentarily. The other solution to this struggle between formulaic and thematic composition is doublet composition, in which the whole theme attains the consistency of a single formula. The contextual relevance of such pairs is restricted, but we have just seen two examples of their use for reporting whole prayers. [91]

To admit this problem is not patronizing to the poet or his medium, since they are more than equal to it. If a formula system to report whole prayers is still-born, that is because the tradition can accomplish ad hoc compendious thematic composition of them. From this point of view, lines such as

Ω 287 τῆ σπεῖσον Διὶ πατρί, καὶ εὔχεο οἴκαδ’ ἱκέσθαι

where invocation (Διὶ πατρί), grant of favor (σπείσας), and request of favor (οἴκαδ’ ἱκέσθαι) all occur in a single hexameter, are testimonies to the resilience and versatility of poetic technique. The same structural completeness is achieved by other means in these passages:

Ο 372 Ζεῦ πάτερ εἴ ποτέ τίς τοι ἐν Ἄργεΐ περ πολυπύρῳ
          ἢ βοὸς ἢ οἰὸς κατὰ πίονα μηρία καίων
          εὔχετο νοστῆσαι …
                    Invocation: Ζεῦ πάτερ
                    Grant of favor: μήρια καίων (cp. δ 762–66)
                    Request of favor: νοστῆσαι
B 400 ἄλλος δ’ ἄλλῳ ἔρεζε θεῶν αἰειγενετάων
          εὐχόμενος θάνατόν τε φυγεῖν καὶ μῶλον Ἄρηος.
                    Invocation: ἄλλῳ … θεῶν αἰειγενετάων
                    Grant of favor: ἔρεζε
                    Request of favor: θάνατόν τε φυγεῖν κ. τ. λ.

In the other three attestations, there is yet another solution: the grant of favor element is ellipsed, and the reported prayer contains only invocation and request of favor. In my view, this is a surface phenomenon motivated by the compositional stress described. [
92] A grant of favor is understood but not specified by the poet in lines such as:

φ 210                        τῶν δ’ ἄλλων οὔ τευ ἄκουσα #
          εὐξαμένου ἐμὲ αὖτις ὑπότροπον οἴκαδ’ ἱκέσθαι.


o 353 Λαέρτης μὲν ἔτι ζώει, Διὶ δ’ εὔχεται αἰεὶ
          θυμὸν ἀπὸ μελέων φθίσθαι οἷς ἐν μεγάροισιν

(See also Ι 183–4). There is no need to assume that the combination εὔχομαι + request of favor infinitive is a sign that the word has a different meaning comparable to that of English ‘pray (request)’, any more than the doublet attestations of εὔχομαι + grant of favor infinitive are predicated on a meaning ‘vow’, which now we can view as the third in a series of strategies for reporting prayers.

C. Conclusion

This survey of εὔχομαι in usages to report and conclude prayers has arrived at and tested a hypothesis that the word has a sacral meaning ‘speak, say sacredly’ or ‘speak, say (a prayer)’. We began with a formulaic classification of its attestations in sacral contexts, and, by internal analysis of the classes, we provided formal, grammatical, and stylistic evidence for the existence of a sacral-secular functional split in its usage: sacral εὔχομαι formulas only overlap with secular εὔχομαι formulas when the text is doubtful and, concurrently, formulaic constraints are ambiguous or under stress. We also found that in sacral contexts εὔχομαι is constrained against usage in the present tense, a constraint which is not operative on secular {66|67} εὔχομαι, and that secular εὔχομαι is constrained against governing an indirect object, a constraint which is not operative on sacral εὔχομαι. In addition, we found a stylistic criterion which allowed us positively to identify as sacral phraseologically isolated attestations of εὔχομαι: their occurrence in single poetic lines containing more than one finite verb. Finally, on the basis of this stylistic criterion, analysis of formulaic transformations, and structural analysis of the speeches which sacral εὔχομαι introduces, we arrived at positive evidence that it designates an act of speech with a fixed tripartite structure whose essence is formal communication or exchange with divinity. The contrary assumptions that εὔχομαι sometimes means ‘vow’, sometimes ‘request’, are ad hoc, and the structural peculiarities of their supposed attestations are to be understood as surface structure transformations conditioned by the rigid yet adaptive constraints of epic compositional technique.


[ back ] 1. Relying on M. Parry’s (see Parry 1928 a = A. Parry 1971: pp. 1–190) analysis of the complexity and thrift of formula-systems and his common-sense conclusion that they could not be the work of an individual. Also, formulaic language is prima facie archaic or archaizing in any cultural setting.

[ back ] 2. In the strict sense of Parry’s first treatise in English on Homeric diction, a formula is “a group of words which is regularly employed under the same metrical conditions to express a given essential idea.” Parry 1930: p. 80 = A. Parry 1971: p. 272.

[ back ] 3. Examples of directly quoted prayers not concluded by (A): Β 419, Γ 324, Η 206, Η 181, Θ 245, ε 444, etc.

[ back ] 4. In (A)’s || τοῦ δ᾿ ἔκλυε, however, the sense ‘heard’ is probably maintained. See discussion below, n. 11.

[ back ] 5. Pluralization of the responders in (B) is a factor in its deletion of εὐχόμενος. A line which contains εὐχόμενος and has plural responders could easily have been made (e. g. *ὣς ἔφατ᾿ εὐχόμενος, τοῦ δ᾿ ἔκλυον––⏓# and enjambement if necessary). But in fact there are only two prayers actually addressed to more than one god: Γ 276ff. and Γ 298ff. (cf. Γ 296). The invocation at Γ 298ff. has a grammatical archaism of Indo-European date (Ζεῦ πάτερ… Ἠέλιός τε: see Schmitt 1967: p. 11, and n. 41), and Γ 298ff. (Ζεῦ κύδιστε μέγιστε, καὶ ἀθάνατοι θεοὶ ἄλλοι) is a part of the same ritual in which Γ 276ff. occurs. The responders to the first prayer are left unspecified, and Κρονίων (= the whole pantheon, see p. 60 below) is the single responder to the second. What we have in this passage is the only Homeric reflex of an archaic, Indo-European prayer type whose existence is otherwise attested in the report of (but never elsewhere their execution) prayers to πᾶσι θεοῖσι, Διι ἄλλοισίν τε θεοῖσιν, etc., in Myceanean offerings to pa-si-te-o-i (written as one word: KN FP 1, etc.), and in Vedic hymns to the Viśvedevā ‘all the gods’. The rare report of prayers to two gods (e. g. at Λ 736, Hesiod, Works and Days 465 and Theogony 441) is to be considered a secondary phenomenon, created within epic by analogy to defunct prayers πᾶσι θεοῖσι, and not necessarily reflecting current ritual practice. With regard to (B), synchronically speaking, its group of responders is a feature of social discourse, not of the sacral speech to which (A)’s εὐχόμενος is appropriate. A diagram in the text below takes account of this contrast in intimacy between sacral and secular speech.

[ back ] 6. # κλῦθι always (12x) occurs in this metrical slot, and its prosody and meaning seem appropriate. However, it has become restricted to sacral contexts (invocations in prayers), while, for example, its plural # κέκλυτε is restricted to secular ones (for a discussion of this, see n. 5). Apparently, Homeric words can evolve contextual restrictions which are originally independent of their meaning, as is the case with # κλῦθι, or dependent upon their inflection, as is the case with # κέκλυτε.

[ back ] 7. For an argument supporting the addition of ‘man’ in this diagram. see the next paragraph of the text.

[ back ] 8. Parry 1928a: pp. 50–1 = A. Parry 1971: p. 39.

[ back ] 9. See Parry 1937: pp. 59–63 = A. Parry 1971: pp. 414–18.

[ back ] 10. For the gender switch, compare:

          # καί μιν φωνήσας ἔπεα πτερόεντα προσηύδα Α 201, Β 7, Δ 284 κ.τ.λ.

with the variant used for the speech of a woman:

          # καί μιν φωνήσασ᾿ ἔπεα πτερόεντα προσηύδα O 35, O 89, α 269, κ.τ.λ.

[ back ] 11. The situation implies that (A)’s simple || τοῦ δ᾿ ἔκλυε is neutral in significance (= ‘heard’ only, not ‘hearkened’) although a god only once (Π 249–50) fails to fulfill the request of prayers concluded by (A). The ostensible neutrality of ἔκλυε and the almost perfect success of prayers can be reconciled if it is understood that prayers are based on the idea of reciprocity (as will be shown in detail below). A man who prays to a god is like a man giving a gift and hoping for (though in fact demanding) a return gift. “The form usually taken is that of the gift generously offered; but the accompanying behavior is formal pretense and social deception, while the transaction itself is based on obligation and self-interest” (Mauss 1967: p. 1). For a parallel in Greek, compare the polite optative in the γνώμη of Epicharmus: ἁ δὲ χεὶρ τὰν χεῖρα νίζει· δός τι καὶ λάβοις τί κα (Kaibel, frg. 273) ‘hand washes hand; give something and you might get something.’ Similarly the god only ‘hears’ a man’s prayer, when in fact he hearkens and accedes to its request. But the pretence of neutrality allows for exceptions such as Π 249–50. For Epicharmus’ parataxis of hand-washing and gift-reciprocity, cp. the custom of washing hands before praying (e. g. β 261) and before feeding a guest (e. g. α 146).

[ back ] 12. See below, p. 27 f. for a discussion of the structure of Homeric prayers. Telemachus’ lacks a specific request from the goddess.

[ back ] 13. For a discussion of the type of relationship between Odysseus, Tydeus, Telemachus, and Diomedes and their motherly protectresses Hera and Athena, see the classic article of Nilsson 1921.

[ back ] 13a. It, too, omits a request (see note 12).

[ back ] 14. The scene is also comparable to the earlier and unsolicited appearance of Athena to Achilles in Book Α (Α 194–222) in which she treats him as a mother would (pulls his hair from behind to restrain him) and tells him to turn his anger into words, not deeds (Α 211). Compare Thetis’ advice to Achilles in this passage:

          Α 363ἐξαύδα, μὴ κεῦθε νόῳ, ἵνα εἴδομεν ἄμφω

[ back ] 15. See T. W. Allen’s apparatus ad locos. His collation of papyri includes a single papyrus for each of the passages. P 16 (3rd Cent. A. D.) presumably reads # ὣς ἔφατ᾿ εὐχόμενος at Ε 106 since he does not report its reading, while at Υ 393 P 9 (actually a Syriac palimpsest – date unlisted) reads # ὣς φάτ᾿ ἐπευχόμενος. The manuscripts which read ἐπευχόμενος are h, Ο6, V3, Ve, and Eustathius at Ε 106, c, e, f, h, k, and l at Υ 393. h is Allen’s largest ‘family’, and it reads ἐπευχόμενος in both places.

[ back ] 16. As we shall show in greater detail below (see pp. 42, 59, n. 82) this is generally the case with both derivatives of εὔχομαι in Homer, viz. εὐχετάομαι and ἐπεύχομαι.

[ back ] 17. ὣς φάτ᾿ ἐπευχόμενος → ὣς ἔφατ᾿ ἐπευχόμενος (a diplology attested in Ludwich’s apparatus to Υ 393) → ὣς ἔφατ᾿ εὐχόμενος (by metrics, haplology, or memory of this opening elsewhere). Note that the diplology mentioned could simply be the result of contamination of two manuscripts each having one of the readings.

[ back ] 18. For another example which implies this for Zenodotus, see below, p. 58ff., and see Lehrs 1882: p. 258.

[ back ] 19. Note that it is hard to imagine a literate poet creating so many expressions for a single idea. Their superfluity is a corollary of Parry’s principle of dictional thrift rather than a refutation of it. Only a poet equipped with a traditional language has the means or desire to carry off such a display. For a poet who can write, it would be simply pretentious (see, for example, the condemnation of this practise in the article ‘Elegant Variation’ in Fowler4 1959).

[ back ] 20. This method of compiling evidence is necessary, since the concordances do not list ὡς, ἔφατ᾿, φάτο, κ. τ. λ.

[ back ] 21. For this term, see Lord 1960: pp. 68–98.

[ back ] 22. Note the preponderance of words with reciprocity implications (δός, χαρίεντ᾿, τίσειαν, κ. τ. λ.) and the dominant motif of reciprocating actions and reactions which permeates the discourse in all the examples (αὖτε, καὶ νῦν, ὄφρα, κ. τ. λ.). My analysis of this element in prayer structure is dependent on the work of M. Mauss (cited above, n. 11), who shows that in many societies the notion of reciprocity is a total social phenomenon with manifestations in political, social, economic, legal, and sacral (see esp. p. 12) spheres. For a demonstration of the presence of this phenomenon in Indo-European societies, see Benveniste 1966: pp. 315–26.

[ back ] 23. For a discussion of Athena’s response, see above, p. 22.

[ back ] 24. On the significance of libation in prayers, see below, p. 36f.

[ back ] 25. For these terms, see Chomsky 1965: Chapter 2, and for a fruitful application of them to Homeric studies, see Nagler 1967.

[ back ] 26. Ξάνθος is the name of one of Achilles’ divine horses, and Ποδάργη (cp. Hector’s Πόδαργος) is the name of the ἅρπυια who bore it (Π 149-154). Furthermore, Λάμπος is a name which belongs to one of the divine horses of dawn (ψ 246), the other being Φαέθων. The semantics of Λάμπος (‘bright’) and Φαέθων (‘shining’) are receptive to the name of Hector’s fourth horse, Αἴθων (‘burning’). There are, we postulate, three sets of divine horses in Homer, as follows:


So the names of Hector’s two pairs (the invocation at Θ 185) are each composed of the name of one divine horse plus a semantic/contextual associate of it. The derivational process at once unites them to and differentiates them from the other divine pairs without implying that they are so differentiated as to be mortal horses. Note that the name of Achilles’ explicitly mortal horse is Πήδασος (Π 153), which is semantically obscure and elsewhere attested as the name of a warrior slain in battle (Ζ 21-28), not a divine horse. For a structural analysis of the mythology and semantics of the horses of Achilles and the Dawn, see Nagy 1973: p. 164. For a parallel to the naming process above, cp. Ἕκτωρ—>Ἀστυάναξ (Ζ 403: Ἀστυάνακτ᾿· οἶος γὰρ ἐρύετο ᾿Ἴλιον Ἕκτωρ), and also Ὀδυσσεύς—>Τηλέμαχος, Ἑλένη —>Κλυταιμνήστρα (I owe these examples to G. Nagy.)

[ back ] 27. The (unattested) cadence *|| τοῦ δ᾿ ἔκλυον ὤκεες ἵπποι # was at least a possibility.

[ back ] 28. For another alteration in the surface structure of Θ 198, the alteration of || τοῦ δ᾿ ἔκλυε to || νεμέσησε δέ, see above, pp. 21–22.

[ back ] 29. For example, cp. the definition of prayer in the American Heritage Dictionary of the English Language (New York 1969): “a reverent petition made to a deity or other object of worship”. ‘Petition’ is defined as “a solemn supplication or request to a superior authority; an entreaty.”

[ back ] 30. Indeed, the meaning of || τοῦ δ᾿ ἔκλυε implies that no such suggestion can be explicit (see note 11).

[ back ] 31. Mauss 1968: p. 414, from a previously unpublished fragment entitled “Prière et les rites oraux.

[ back ] 32. Even without quoting the surrounding contexts, this is clear from the typical narrative particle combinations αὐτὰρ ἐπεὶ, αὐτὰρ ἐπήν, ἦ τοι ὁ μέν.

[ back ] 33. I use the term sub-genre after Householder and Nagy 1972. To conceive of the Epic genre as containing a series of sub-genres seems an effective way to account for the isolation of thematic and linguistic phenomena in specific contexts. An example of this has already been given in the discussion of battle-book formulas, see above, p. 24f.

[ back ] 34. Why? Perhaps because ritual itself is in essence stylized action. Then this style is its perfect verbal analogue. For a similar phenomenon in actual ritual language, see the concatenating, paired alliteration in the prayer to Mars preserved in Cato (De Agri Cultura, 141)—a feature of its style analyzed by C. Watkins (Lectures at Harvard University, 1968). Aesthetically, however, there is no comparison to the richness of detail and hypnotic verb rhythm of the Homeric passages.

[ back ] 35. Compare Mauss’ term: rite oral (above, n. 31).

[ back ] 36. Lévi-Strauss 1966: p. 10.

[ back ] 37. See the schematization of expressions with this opening in Parry 1930: p. 85 = A. Parry 1971: p. 276.

[ back ] 38. For example, see A 259 ἀλλὰ πίθεσθ᾿… A 274 ἀλλὰ πίθεσθ᾿…

[ back ] 38a. For an attempt to account for grammatical and metrical transformations of this kind in a definition of the formula, see Hainsworth 1968: pp. 36–39. See further below, n. 42.

[ back ] 39. Benveniste 1969: t. 2, p. 234.

[ back ] 40. The last, Ω 287, is a ritual prescription, i. e., a transformation of ritual description to the imperative. For other examples of formulaic interchange in ritual prescription and ritual description, see below, n. 63, p. 48, p. 55.

[ back ] 41. It is not necessary to suppose that such a line existed concretely. The notion of actual prototypes is spurious, see Nagler 1967: pp. 269ff. It is enough that parallels to such a prototype exist, e. g. αὐτὰρ ἐπεὶ σπεῖσάν τε πίον ὅσον ἤθελε θυμός (7 ´).

[ back ] 42. The transformation # σπένδοντ᾿ εὐχόμενον τε || => σπείσῃς τε || καὶ εὔξειαι involves metrical and grammatical changes not researched by Hainsworth (see n. 38a). But refusing to consider εὔχομαι + σπένδω as a traditional formula would defy common sense as well as suggestive linguistic evidence (see n. 52[????] below). Schematically, the metrical switch may be represented as follows:

          # – ⏕ – ⏕ –  || => – – || ⏕ – ⏕

This representation reduces the metrical transformation to a deletion of one syllable from the beginning of the formula and the addition of one syllable to the end, which is accompanied by a switch in placement from line-initial to triphthemimeral caesura. For a similar set of transformations, cp.

          Υ 143 # ἡμετέρῃς ὑπὸ χερσίν ||

          Κ 310, 397 # ἣ ἤδη χείρεσσιν || ὑφ᾿ ἡμετέρῃσι

          Θ 531, Σ 304 # νηυσὶν ἐπὶ γλαφυρῇσιν ||

          Ο 603 τὰ φρονέων νηέσσιν || ἐπὶ γλαφυρῇσιν

but more exact parallels are difficult to find given the stylistic peculiarities of the sub-genre. Cp. also:

          ε 197 # ἔσθειν καὶ πίνειν ||

          κ 272 # ἔσθων καὶ πίνων ||

          κ 427 # πίνοντας καὶ ἔδοντας ||

          β 305 # ἀλλά μοι ἐσθιέμεν || καὶ πίνεμεν…

          φ 69 # ἐχράετ’ ἐσθιέμεν || καὶ πίνεμεν…

For an example of a transformation which does destroy the identity of a formula, see below, pp. 39ff.

[ back ] 43. Benveniste 1969: t. 2, pp. 234–43.

[ back ] 44. I. e., sometimes it means ‘wish’ and sometimes it means ‘pledge’. This ambiguity makes sense in French, since voeu = ‘wish’ or ‘pledge’, as against English ‘vow’. Is Benveniste implying a typological parallel? If so, it is a petitio principii, since the split in French is inherited from Latin vōtum (the cognate of εὔχομαι and etymon of voeu) and he is trying to prove that the split in Latin is etymologically primary.

[ back ] 45. Benveniste 1969: t. 2, p. 235, p. 214.

[ back ] 46. One may compare, though Benveniste himself does not, the Latin cognate of Greek σπένδω, namely spondeō, which means ‘pledge, promise.´

[ back ] 47. Benveniste 1969: t. 2, pp. 236–7, p. 240. Homeric beasts are to be understood in the context of dēvōtiō, whereby a general pledged and sacrificed himself to the success of his army (Livy VIII, 10, 11, for the example of Decius Mus). When a Homeric hero εὔχεται about his γένος or his bravery, he is actually ‘pledging himself’ or ‘making a vow’. For a discussion of this argument, see c. III, below, n. 45 and p. 104.

[ back ] 48. Lewis and Short 1879: s.v. vōtum

[ back ] 49. For the idea of a wasted or conspicuously consumed gift, cp. the custom of potlatch among the Kwakiutl and Haida. See the discussion by Mauss 1967: pp. 31–45 and Benveniste 1966: pp. 323–326 for Indo-European parallels. See also Citron 1965: p. 4 on libation as gift.

[ back ] 50. The nature of this return may be implicit and ordained by custom or made explicit verbally. For an example of an inexplicit version of the rite, see the frequent libation made in Homer before going to sleep (γ 334, σ 419, Ι 712). In one place only does the narrator explain the addressee, η 137–138. It is Hermes, god of waking and sleeping, guide on dangerous journeys such as those to Death, and here, perhaps, to Death’s twin-brother (Π 672, 682, etc.), Sleep.

[ back ] 51. This is the folk etymology, see Ernout-Meillet 1959, s.v. *spons and Festus 440, 1: spondere Verrius putat dictum quod sponte sua, i. e. voluntate, promittatur, and Varro, L. L. 5, 69: spondere est dicere spondeo, a sponte: nam id valet et a voluntate. But the equation raises phonological problems which are unsolved. No other satisfactory cognate exists.

[ back ] 52. Both have o-grade of the root and –eye/eyo-suffix; see Meillet-Vendryes 1966: p. 190, § 292 for the type. Their formal identity in Latin is a suggestive fact with regard to the evidence for the traditionality of Homeric εὔχομαι + σπένδω. For an example of words paired together in Indo-European and so forced into parallel morphology, see Schulze 1933: pp. 257–261 on Tocharian tseke peke, Latin finxit pinxitque.

[ back ] 53. Benveniste 1969: t. 2, p. 215, in his discussion of spondeō: “En Latin une partie [l’ offrande] importante de la signification primitive a disparu”. See also Ernout-Meillet 1959: s.v. spondeō: “Le rite de la libation est indiqué par gr. σπένδω; l’obligation résulte du rite. Ceci indique, évidemment, le sens indoeuropéen, disparu en latin …” (p. 644).

[ back ] 54. γ 337 Mentes exhorting Telemachus and the Pylians; HDem 299 Keleos exhorting the Eleusinians; Κ 47 Hector speaking (not face to face) to Agamemnon, Π 76 Agamemnon (not face to face) to Achilles.

[ back ] 55. For φῆ introducing boasts, see below, p. 83. The adverb μεγάλ᾿ is never elsewhere attested with αὐδάω but several times with εὔχομαι in sacral contexts (see below, p. 113f.). The occurrence of the words τοῦ…μεγάλ᾿…ἔκλυεν in δ 505 (cp. (A) || τοῦ δ᾿ ἔκλυε…) suggests the presence of εὔχομαι formulas in the poet’s mind.

[ back ] 56. To my knowledge, moreover, this is the only example in Homer where anyone, divine or human, is said to hear a boast, which in Homer is constrained against having a specific addressee. Perhaps it would be more precise to speak of the addressee (i. e. the god invoked) rather than the hearer in specifying the constraints upon sacral εὔχομαι. See the example in the text below, the discussion of ἔκλυον ἠδὲ πίθοντο # above (p. 19) and cp. the following instance of a gloss by ἀκούω on the root noun κλέ(ϝ)ος of the verb κλύω:

          γ 83 πατρὸς ἐμοῦ κλέος εὐρὺ μετέρχομαι, ἤν που ἀκούσω

[ back ] 57. For φῆ introducing boasts, see below, p. 83. The adverb μεγάλ᾿ is never elsewhere attested with αυδάω but several times with εὔχομαι in sacral contexts (see below, p. 113f.). The occurrence of the words τοῦ…μεγάλ᾿…ἔκλυεν in δ 505 (cp. (A) || τοῦ δ᾿ ἔκλυε…) suggests the presence of εὔχομαι formulas m the poet’s mind.

          To my knowledge, moreover, this is the only example in Homer where anyone, divine or human, is said to hear a boast, which in Homer is constrained against having a specific addressee. Perhaps it would be more precise to speak of the addressee (i. e. the god invoked) rather than the hearer in specifying the constraints upon sacral εὔχομαι. See the example in the text below, the discussion of ἔκλυον ἠδὲ πίθοντο # above (p. 19) and cp. the following instance of a gloss by ἀκούω on the root noun κλέ(ϝ)ος of the verb κλύω:

          γ 83 πατρὸς ἐμοῦ κλέος εὐρὺ μετέρχομαι, ἤν που ἀκούσω

[ back ] 58. Parry 1928a: p. 207 = A. Parry 1971: pp. 164–165.

[ back ] 59. Hainsworth 1968: p. 108. While Parry’s remarks concern name-epithet formulas, Hainsworth’s are based on a study of enjambed noun-epithet formulas.

[ back ] 60. All have the form || – ⏔ – – ⏓ # except || κούρη Διὸς αἰγιόχοιο #. This is simply a P2 variant of Διὸς κούρη μεγάλοιο #, see p.46 below and n. 64.

[ back ] 61. These are also the gods which fill the adonic cadence of (A), except Apollo is omitted. On the reasons for and consequences of his omission, see below, p. 48f.

[ back ] 62. Nagy 1974: p. 99.

[ back ] 63. Again, there is one example of ritual prescription, Η 194, which is the only instance in Section II of deviation from the inflection εὔχοντο. See Η 200, which fills the prescription, n. 40 above.

[ back ] 64. Severyns 1944–48: t. 2, pp. 96–97, p. 54.

[ back ] 65. Apollo does have an alternative which begins with a consonant, Λυκηγενέϊ κλυτοτόξῳ, which is in fact twice attested with εὔχομαι, but it is frozen in a group of three lines (Δ 101 ff. = Δ 119 ff.). This severely restricts its availability to the poet.

[ back ] 66. ἑκηβόλος < ἑκάς (?)< ἕ < ϝε < *swe + –kas (as in Skt. śata-śaḥ ‘hundred by hundred’). For evidence of the digamma, see Boeot. ϝεκαβόλος, Frisk 1960 s.v. ἑκήβολος and the hiatus in Ε 54 οὐδὲ ἑκηβολίαι, ᾗσιν τὸ πρίν γ᾿ ἐκέκαστο.

[ back ] 67. For an excellent demonstration of this and its relation to cult practice, see Clader 1973. This line is one of the few direct mentions of cult in the lliad and Odyssey, but its importance underlies the whole corpus.

[ back ] 68. The only other figures with this genealogy are Herakles (λ 604), Hermes (HHerm 214, 230), and the Muse in these lines:

          θ 487 Δημόδοκ’, ἔξοχα δή σε βροτῶν αἰνίζομ’ ἁπάντων·

                    ἢ σέ γε Μοῦσ’ ἐδίδαξε, Διὸς πάϊς, ἢ σέ γ’ Ἀπόλλων·

[ back ] 69. The description of her at n 221 ff. is reminiscent of Hermes, god of thieves, trickery, etc. rather than Athena. Note Athena’s remarks when she reveals herself:

          ν 298 ἐγὼ δ’ ἐν πᾶσι θεοῖσι # μήτι τε κλέομαι καὶ κέρδεσιν·

Another instance of overlap between Hermes and Athena: Athena’s wearing of Hermes’ sandals in α 96–98, cp. ε 44–46, Ω 340–342.

[ back ] 70. At 299–300, she explicitly denies that Odysseus had recognized her. But she may have been fooled.

[ back ] 71. See above, n. 67.

[ back ] 72. For discussion of still another significant anomaly in ν 230–231, see below, p. 58f.

[ back ] 73. Or of hero pretending to divine status (Η 298) or of god pretending to lack it (ν 230–231). These are the only other possibilities.

[ back ] 74. On εὔχομαι…#ἐξελάαν (Θ 526-7 see below, pp. 57ff.

[ back ] 75. See Frisk 1968: s.v. εὔχομαι; Citron 1965: pp. 85–86. These two doublets are the only textual basis for the meaning ‘vow’ in Epic.

[ back ] 76. Of course, the possibility exists that it means ‘vow’ in all those cases as it does here, though none of the above mentioned authorities would agree. This is beside the point. All we are arguing here is that given this pattern, the meanings should be consistent.

[ back ] 77. That this is the request of favor is not immediately apparent, but comparison with examples of εὔχομαι + aorist infinitive make it clear. See φ 211, Ο 372-4, Ω 287. In the context Pandaros is praying for his life (= return home thematically speaking; see Frame 1971: passim; compare also another εὔχομαι + aorist infinitive attestation: Β 401 #εὐχόμενος θάνατόν τε φυγεῖν, κ.τ.λ.).

[ back ] 78. In spite of αὔτικα νῦν at Ζ 308, the difference between the future infinitives in the doublets and the subjunctive ὄφρα… ἱερεύσομεν cannot be closely pressed.

[ back ] 79. Perhaps the real force of these transformatioins can be brought home to an English speaker by the following sentences:

          Joe, you scratch my back, I’ll scratch yours.

          Joe, if you scratch my back, I’ll scratch yours.

          Joe, if I scratch your back, you’ll scratch mine.

          Are any of these statements petitions? promises? wishes?

[ back ] 80. Mauss 1968: 383.

[ back ] 81. See Allen’s apparatus ad loc. The Mss. are Allen’s h family, M13, U10, V11, the testimonium is Plutarch’s Vita Homeri 118.

[ back ] 82. For the relative contextual freedom of the two derivatives of εὔχομαι, compare the earlier discussions of εὔχομαι/ἐπεύχομαι variants (pp. 23ff., pp. 41ff.). For another example of the freedom of εὐχετάομαι as against the restrictions on εὔχομαι, cf.

          Λ 761 πάντες δ᾿ εὐχετόωντο θεῶν Διὶ Νέστορί τ᾿ ἀνδρῶν

where εὐχετάομαι is used with a ‘man’ in the dative (cp. our discussion of the only example of εὔχομαι + mortal in the dative, pp. 50 ff., and on Λ 761, see the evidence on Nestor’s mortality/immortality gathered by Frame 1971: pp. 108-128. Note also the close association of Zeus and Nestor in this line; see above, p. 22f.).

[ back ] 83. Stanford 1962: p. 346.

[ back ] 84. For an example of this principle in the cultural context of oral epic, see Lord 1960: p. 286, n. 2, where a Yugoslav singer is cited who refuses to commit himself to a title for his song other than “Two pashas spent the winter…”, which is its first line.

[ back ] 85. Note that this explanation resolves problems about the unique invocation in Θ 526 since it can be explained by terms other than genre (see above, p. 58).

[ back ] 86. Note the poignant contrast between ἔτι and αἰεί in the complete passage:

          o 353 Λαέρτης μὲν ἔτι ζώει, Διὶ δ’ εὔχεται αἰεί

                    θυμὸν ἀπὸ μελέων φθίσθαι οἷς ἐν μεγάροισιν.

If our analysis is correct, the switch of πατρί to αἰεί is a brilliant stroke.

[ back ] 87. # εὐχ is well attested (εὔχομαι + dative, Section I) though it is inflectionally restricted to the imperative, while # εὔχομαι in sacral contexts occurs enjambed in the line ν 231 (which is in other respects anomalous, see pp. 50ff., n. 72). Neither of these are exact parallels, then, but they offer some support to Allen’s text.

[ back ] 88. Scribal transpositions are usually ΑB => BA (for examples, see Adam 1965: v. II, p. 523), while this transposition is metrical and morphological, and of a type that occurs in spoken language, e. g. “he came running” “he ran coming”, etc.

[ back ] 89. Seeabove, n. 77, on the relationship between coming home and escaping death. For οἴκαδ᾿ ἱκέσθαι vs. οἴκαδε νοστῆσαι, cp. Δ 103 = 121 # οἴκαδε νοστήσας, Σ 60, 441, οἴκαδε νοστήσαντα, and cf.

          Α 60 ἂψ ἀπονοστήσειν, εἴ κεν θάνατόν γε φυγοῖμεν.

[ back ] 90. The average length of actual prayers is about five lines.

[ back ] 91. Note that one of the εὔχομαι + pronoun formulas for Apollo in the dative (Υ 451 = Λ 364) is the only repeated whole line with εὔχομαι aside from (A). The compositional stress there is analogous.

[ back ] 92. For a parallel in actual prayers, cp. the one which Odysseus makes while running the foot-race in Patroclus’ funeral games:

          Ψ 769 κλῦθι θεά, ἀγαθή μοι ἐπίρροθος ἐλθὲ ποδοῖιν.

                    ὣς ἔφατ’ εὐχόμενος· τοῦ δ’ ἔκλυε Παλλὰς Ἀθήνη …