In order to say anything useful about the “mixed aorist” imperative ὄρσεο, we must see it in relation to the other forms of the verb ὄρνυμι as they are used in the epic. The basic usage is in the middle and perfect, with intransitive sense; compare the middle forms of the Latin cognate orior. Nearly all the forms of ὄρνυμι attested in Homer are third person singular, so these will be cited. The perfect ὄρωρε evidently belongs to a very old level of the epic language. It (as also the pluperfect ὄρώρει) is virtually restricted to the final position of the verse (exceptions A 658, M 177, e 294 = i 69 = μ 315) and particularly to formulaic phrases like the following: … νεῖκος ὄρωρε (ὀρώρει) H 374 etc. … ὀρυμαγδὸς ὄρωρε (-ει) B 810 etc. … βοὴ δ’ἄσβεστος ὄρωρεν (-ει) L 530 etc. The augmented pluperfect occurs once: Σ 497-8 … ἔνθα δὲ νεῖκος//ὠρώρει … where the usage is obviously secondary to the regular verse-end νεῖκος ὄρωρε. The antiquity of (ὄρ)ωρε is also supported by the Vedic Sanskrit perfect ára (RV 3.30.10 ví ara “opened”, verse-final) and Hittite ari “comes.” [1] Furthermore, we can be certain that the athematic aorist middle ὧρτο is ancient. Such formations are isolated and unproductive even in Homeric Greek (cf. ἇλτο, γέντο, ἔδεκτο, λέκτο); [2] this is sufficient proof of their archaism. In addition, there are parallel forms in other ancient Indo-European languages: Vedic aorist middle, indicative árta “came,” injunctive arta ; Hittite present arta(ri ) “stands” and middle preterit artati. The development from the original Indo-European perfect or middle was parallel in all three languages. In Homer, ὧρτο is used more freely than ὄρωρε, but still tends to follow certain patterns, such as these: ῾ὼς ἔφατ’· ὦρτο δ’ ἔπειτα … Ψ 708 etc. ὦρτο πολὺ πρῶτος μὲν ἄναξ ἀνδρῶν … H 162, Ψ 288 … ὦρτο δ’ἐπ’αὐτοὺς … E 590 etc … ὦρτο δ’ἀὕτή. M 377 etc.

For the present tense, the formation of ὄρνυσι is paralleled by Vedic rnóti “sets in motion” and Hittite arnuz(z)i “brings.” Like the Greek active, the Hittite form has a causative sense. The active of the present system in Homer is limited to the imperatives ὄρνυθι (cf. Hittite arnut ), ὄρνυτε. The present middle ὄρνυται and imperfect ὦρνυτο are common. Since both the imperfect and the aorist are used as narrative tenses (though perhaps with different nuances), ὦρνυτο can serve as a metrical variant of ὦρτο in narration. Compare: C 754 ῾ὼς ἔφατ’, ὄρνυτο δ’αὐτίκ’ ‘Οϊλῆος ταχὺς Αἴας, C 708 ὣς ἔφατ’, ὦρτο δ’ἔπειτα μέγας Τελαμώνιος Αἴας. The dactylic shape of the imperfect is especially useful in the fourth or fifth foot; so in these positions it is often used rather than the aorist. When, however, the passage is in the present tense, and reference is made to a prior time, the imperfect cannot be used interchangeably with the aorist. In three such passages, we find a dactylic form of the aorist middle, namely, ὤρετο: [3] M 279f, … ὅτε τ’ ὤρετο μητίετα Ζεὺς νίφεμεν… “when Zeus has begun to snow” (in a simile); Ξ 397 … ὅτε τ’ ὤρετο καιέμεν ὕλην “when it (fire) has begun to burn the forest” (in a simile); X 102 … ὅτε τ’ ὤρετο δῖος ‘Αχιλλεύς, “when godlike Achilles arose” (Hector speaking). Thematization of an archaic athematic form is a predictable innovation; in fact the same thing happened in Vedic Sanskrit, where árata is a renewal of árta. [4] There was another possible dactylic form of the aorist in Homer, the reduplicated ὤρορε. It occurs only five times, with uncertainty whether it replaces the perfect ὄρωρε (as in N 78 and θ 539) or the causative aorist ὦρσε (as in B 146, τ 201, ψ 222). The sigmatic form is the regular causative aorist in Homer, occurring frequently and in various metrical positions; that is, it is a productive formation. The future tense, on the other hand, is very rare. We find only two possible examples. At Y 140, the expected form for a liquid verb occurs, middle and intransitive: αὐτίκ’ ἔπειτα καὶ ἄμμι παρ’ αὐτόθι νεῖκος ὀρεῖται. A causative future participle ὄρσουσα occurs at Φ 335, where Zenodotus read ὄρσασα and Pseudo-Plutarch Vit. Hom. 2.108 quoted ὀρέουσα (both presumably regularizations of an abnormal form). [5] The first person plural ὄρσομεν, though clearly subjunctive where it occurs: Δ 15-16 ἤ ῥ’αὖτις πόλεμόν τε κακὸν καὶ φύλοπιν αἰνὴν ὄρσομεν, ἦ φιλότητα μετ’ ἀμφοτέροισι βάλωμεν: may at some time have been taken as a future indicative and thus have provided a base for the participle ὄρσουσα. Homeric usage clearly shows a link between this participle and the sigmatic aorist; compare M 253: ὧρσεν ἀπ’ ‘Ιδαίων ὀρέων ἀνέμοιο θύελλαν, and Φ 335 εἴσομαι ἐξ ἁλόθεν χαλεπὴν ὄρσουσα θύελλαν. Therefore, there is insufficient reason to cite ὄρσω as a future of ὄρνυμι. It is simply the subjunctive of ὧρσα. As will be seen, this is quite different from the relation of λέξομαι and ἔλεξα.

A complication arises when we consider the forms ὄροντο (γ 471) and ὀρώρει (Ψ 112). The latter is identical with the pluperfect of ὄρνυμι; the former could be the third person plural of ὧρτο (or ὤρετο). But when these two verses are compared with ξ 104, another conclusion is required: ξ 104 ἐσχατιῆ βόσκοντ’, ἐπὶ δἀνέρες ἐσθλοὶ ὄρονται γ 471 δαίνυνθ’ ἑζόμενοι· ἐπὶ δἀνέρες ἐσθλοὶ ὄροντο Ψ 112 πάντοθεν ἐκ κλισιῶν· ἐπὶ δ᾿ ἀνὴρ ἐσθλος ὄρώρει. These forms have to be assigned to the verb ὄρομαι, which is otherwise attested only in Mycenaean, e.g. PY Ae 134 o–pi ta–ra–ma-<ta>-o ge–to–ro–po–pi o–ro–me–no “guarding the cattle of Thalamatas.” [6] Notice the construction with opi, as in Homer with ἐπί. It has usually been assumed that ὄρομαι is cognate with ὁράω and thus originally had an initial w. [7] But if this were so, we would expect wo- in Mycenaean. [8] There is no evidence of a digamma in the Homeric passages either. It seems better, therefore, to associate ὄρομαι (and οὖρος, ἐπίουρος, ἔφορος) with Avestan haraiti (and harvaiti) “guards”, and Latin seruare “to protect”; that is, to posit a root * sor- instead of * wor-. [9] Both the noun ἔφορος and the active verbal forms cited by Hesychius: ὅρει· φυλάσσει, ὥρειν: φυλάττειν show the stage in which the initial s has been reduced to h. The further development of psilosis in Aeolic and East Ionic dialects caused the forms of ὄρομαι and ὄρνυμι to be confused. For the sense, ἐφοράω “oversee” approached ἐπὶ … ὄρομαι. In consequence, ὄρομαι lost its separate existence.

It remains to consider the apparently anomalous forms ὀρέοντο and ὄρσεο. The former is explained by Magnien as imperfect of the future ὀρεῖται. [10] Meillet and Vendryes accept this interpretation. [11] Bechtel, on the other hand, calls it an iterative in *-eyo- of ἔρετο, cited by Hesychius in the meaning ὡρμήθη. [12] He is followed by Pokorny and Chantraine. [13] There is, however, some internal evidence that ὀρέοντο is an innovation in the epic. It occurs only twice: B 398 ἀνστάντες δ’ὀρέοντο κεδασθέντες κατὰ νῆας, Ψ 212 ἡ μὲν ἄρ’ ῾ὼς εἰποῦσ’ ἀπεβήσετο, τοὶ δ’ὀρέοντο ἠχῆ θεσπεσίη … With the former, we may compare N 738f: οἱ μὲν ἀφεστᾶσιν σὺν τεύχεσιν, οἱ δὲ μάχονται παυρότεροι πλεόνεσσι, κεδασθέντες κατὰ νῆας. Here the phrase κεδασθέντες κατὰ νῆας is used correctly: “scattered among the ships” as an existing state. In B 398, on the other hand, the phrase makes less good sense. It should mean that the men were already scattered among the ships; but, since they are gathered in an assembly, it has to mean “they went scattering” to various places. This sense would seem to require a present rather than a stative participle (though the verb is not used in the present system in Homer). The incongruity of the tense in B 398 suggests that this versed is secondary to N 739. In Ψ 212 also, the usage can be shown to be secondary. We have already seen that phrases with ἀπεβήσετο are often replacements of those with ἀπέβη (above, chapter three), e.g. E 133: ἡ μὲν ἄρ’ ὣς εἰποῦσ’ ἀπέβη γλαυκῶπις ‘Αθήνη. For the second half of Ψ 212, compare N 833f: ὣς ἄρα φωνήσας ἡγήσατο: τοὶ δ’ ἅμ’ ἕποντο ἠχῆ θεσπεσίη … It seems that Ψ 212 is a conflation of two formula-types. If, therefore, ὀρέοντο appears in only two verses, each demonstrably secondary, it is more probable that it is an analogical formation of the epic language than that it is a survival of some old derivational category.

How did the innovation come about? The comparison of ὦρτο and ὤρετο suggested that a thematic vowel could optionally be added before the desinence. As was stated above, ὄροντο, though attested only from ὄρομαι, would be the plural of ὦρτο. For an archaic third plural from the athematic stem with full grade in the desinence, [13a] compare the Vedic cognate ranta “moved themselves” (RV 7.39.3). The following analogical diagram can be set up:

The analysis of ὄρσεο is a little more complicated. As middle and intransitive, it clearly cannot be connected with the active causative aorist ὦρσε or with the causative future participle ὄρσουσα. If, as Wackernagel said, it is an analogical development from ὄρσο, the athematic imperative corresponding to ὦρτο, [14] why is it not *ὄρεο? Of course, a tribrach would have no metrical advantage, whereas a dactyl is desirable, and especially at the beginning of the verse, where imperatives are usually found. [15] But metrical advantage is not sufficient for the creation of an otherwise unmotivated form. The following series of events is proposed as a fuller explanation. When intervocalic s was lost in Greek, the middle imperative desinence came to be perceived as -o instead of -so. Consider the hesitation shown in restoring –so after a vowel. [16] Then ὄρσε was resegmented as ὄρσ-ο instead of ὄρ-σο. The s was now taken as an extension of the stem. Then the thematic vowel e was inserted before the desinence –o:

The developments from ὧρτο, ὄροντο, ὄρσο, to ὤρετο, ὀρέοντο and ὄρσεο are in accord with the principle that a compound morpheme tends to replace a simple morpheme which has the same function. [17] A listing of the instances of ὄρσεο follows. In the two verses where the contracted form ὄρσευ is read, an earlier version with ὄρσο can be reconstructed; that is, instead of ἀλλ’ ὄρσευ πόλεμόνδε, one might have had ἀλλ’ ὄρσο πτόλεμόνδε. [18]

The imperative λέξεο looks similar in formation to ὄρσεο, but in order to make the relation precise we must look into the situation more closely. The verb formed from the root λεχ-, seen in the noun λέχος, occurs in extant Greek texts only in aorist and future (though Hesychius attests the present and the perfect participle: λέχεται: κοιμᾶται λελοχυῖα· λεχὼ γενομένη). According to F. Specht, the root * legh- in Indo-European had punctual aspect: “to lay oneself down,” not “to be lying down.” [19] Thus the athematic [20] ἔλεκτο, though it is not formally distinguished from an imperfect, is regularly aoristic in Homer. The verb also has a future tense (originally a desiderative) λέξομαι (-εαι, -εται). In the epic these are the basic forms. But ἔλέκτο was not entirely satisfactory as an aorist. Not being marked as an aorist, and being in any case a synchronically unmotivated form, it could come to be used as an imperfect through misinterpretation of formulae. For example, a verse such as I 664: τῷ δ’ἄρα παρκατέλεκτο γυνή, τὴν Λεσβόθεν ἦγε, “a woman laid herself down beside him, etc.” could be understood as “a woman was lying beside him.” The durative sense is clear in I 565: τῆ ὅ γε παρκατέλεκτο χόλον θυμαλγέα πέσσων, “beside her he (Meleager) lay nursing his heart -sore anger.” Compare also [Hesiod] Scutum 46: παννύχιος δ’ἄρ’ ἔλεκτο σὺν αἰδοίη παρακοίτι “all night long he lay with his modest wife.” The aorist function of ἔλεκτο could not be renewed by thematization, as ὦρτο was by ὤρετο, because there was no synchronic awareness of a verbal stem λεχ-. Before s and t the velar stops fell together; we can see the resulting confusion between λεχ- and λεγ- in δ 451-3 (Proteus and the seals): … πάσας δ’ἄρ’ ἐπῴχετο, λέκτο δ’ἀριθμόν· ἐν δ’ ἡμέας πρώτους λέγε κήτεσιν, οὐδέ τι θυμῷ ὠίσθη δόλον εἶναι· ἔπειτα δὲ λέκτο καὶ αὐτός. The first λέκτο must mean” counted”; the second, “lay down.” Consider also the ambiguity of the active aorist ἔλεξα. Furthermore, *ἐλέχετο would look even more like an imperfect than ἔλεκτο. The opposition with the imperfect ὤρνυτο establishes ὤρετο as an aorist, but there is no such opposition in the verb from λεχ-. The creation of a new aorist began with the interpretation of λέξομαι as aorist subjunctive. [21] This may be seen in Δ l30f: ἡ δὲ τόσον μὲν ἔεργεν ἀπὸ χροός, ὡς ὄτε μήτηρ παιδὸς ἐέργη μυῖαν, ὄθ’ ἡδέϊ λέξεται ὕπνῳ, and in Ψ 171f: ἀλλ’ἄγε μοι, μαῖα, στόρεσον λέχος, ὄφρα καὶ αὐτὸς λέξομαι· ἦ γὰρ τῆ γε σιδήρεον ἐν φρεσὶν ἦτορ. When λέξεται was an aorist subjunctive, then ἐλέξατο was created as a corresponding aorist indicative. This form has not only the benefit of a distinct aorist marking (- sa -) but also the metrical advantage of a dactylic ending. One might ask, why was not ὦρτο renewed in the same way, giving *ὤρσατο? Clearly, the answer is that ὄρνυμι already had a sigmatic aorist ὦρσα, well established in the causative function. A middle form *ὤρσατο would also be causative (transitive), like ἐστησάμην (b 431, Z 528) which contrasts with the intransitive ἔστην. The causative ἔλεξα (J 252, V 635), on the other hand, is evidently later than ἐλέξατο.

As for the athematic imperative λέξο (Ω 650, κ 320), it was inevitable that its stem should appear to be λεξ-, when the only motivated forms of the verb had that stem. Why do we get λέξεο rather than λέξαι (as perhaps we do at τ 44, before a vowel, with λέξεο elided as a variant reading)? The form of λέξεο is evidently not based on the sigmatic aorist ἐλέξατο. Is it based on the future λέξομαι? There are difficulties in the way of positing a future imperative, as has been seen in the case of οἴσετε etc. Leumann had to imagine a missing link (Vermittlungsglied ) λέξεσθε, which could be a future indicative used imperativally and for which he compares ὄψεσθ [22] (see above, chapter two). But λέξεσθε does not in fact occur, either as future or as imperative. The future occurs only in the singular and always verse-initial (except Δ131, quoted above, and of course in compounds). If λέξεο were based on the future, one would expect it to occur in the same metrical position as the future, especially since imperatives tend to be verse-initial of their own accord. But λέξεο occurs only in the fourth and fifth feet (once each): I 617 οὗτοι δ’ ἀγγελέουσι, σὺ δ’αὐτόθι λέξεο μίμνων τ 598 ἔνθα κε λεξαίμην· σὺ δὲ λέξεο τῷδ’ ἐνὶ οἴκῳ. Leumann suggests that λέξεο was the model for ὄρσεο, the two verbs forming an opposed pair (“lie down” and “get up”). [23] But ὄρσο and ὄρσεο are more common than λέξο and λέξεο (ὄρσο occurring five times in Homer, ὄρσεο seven, λέξο and λέξεο twice each). Hence it is more probable that the influence went the other way, i.e. that λέξεο was modeled on ὄρσεο.

In contrast to ὄρσεο and λέξεο, another “mixed aorist” imperative may actually be from the future. The form ἀείσεο occurs once or twice in the Homeric Hymns: H. 17.1 Κἀστορα καὶ Πολυδεύκε’ ἀείσεο, Μοῦσα λίγεια, Η. 20.1 Ἥφαιστον κλυτόμητιν ἀείσεο Μοῦσα λίγεια. Though the manuscripts have ἀείδεο in the latter verse, the middle ending makes the future stem probable, as the present is normally active. [24] The two examples are both in introductory lines of hymns. Some hymns begin “I shall sing,” like H. 23.1: Ζῆνα θεῶν τὸν ἄριστον ἀείσομαι ἠδὲ μέγιστον. Others begin with an appeal to the Muse, “Sing, O Muse,” like H. 19.1: ἀμφί μοι ‘Ερμείαο φίλον γόνον ἔννεπε Μοῦσα. Compare A 1 and a 1. The “future imperative” ἀείσεο seems to be a conflation of the first-person future type and the imperative type, preserving the metrical shape and position of the future indicative ἀείσομαι. This last criterion, it will be remembered, λέξεο did not satisfy, although its form could have been connected with the future; ὄρσεο differed even in the form of its stem from the future. So these imperatives had to be explained via the aorist rather than the future.