316. Passing over for the present the question whether the quasi-imperative or the quasi-future use is to be regarded in each case as representing the original meaning of the mood, we proceed to consider the uses in subordinate clauses. Here the main distinction is that between "final" and "conditional" ;if these terms are used with some latitude: especially if we rank with the final clauses not only those which distinctly express the end or purpose of an action, but also all clauses which are referred to the time of the governing verb. It is true that this distinction does not always apply: e. g. to the subjunctive in
Δαναῶν ὀλοφυρόμεθʼ αἰχμητάων,
οἵ κεν δὴ κακὸν οἶτον ἀναπλήσαντες ὄλωνται·
or to the optative in
ἀλλὰ πολὺ μεῖζον . . .
μνηστῆρες φράζονται, ὃ μὴ τελέσειε Κρονίων
For there the relative clause is in sense a parenthesis, and is construed accordingly as an independent sentence. Again, in
ἔσσεται ἦμαρ ὅτʼ ἄν ποτʼ ὀλώλῃ κτλ.
φρασσόμεθʼ ἠὲ νεώμεθʼ ἐφʼ ἡμέτερʼ ἦε μένωμεν
δείδιε γὰρ μὴ λαιμὸν ἀποτμήσειε κτλ.
and generally in object clauses, the subordinate clause does not express end; but the time from which it is regarded as spoken is fixed by the governing verb, in the same way that the time of a true final clause is fixed by the action of which it gives the end. For the present purpose, accordingly, there are two kinds of clause to be considered, (1) final and object clauses, and (2) conditional clauses.
Regarding the meaning of the subjunctive and optative in final clauses there can be little doubt. The subjunctive in most instances follows either a 1st Person (present or future), or an imperative; that is to say, it expresses the immediate purpose with which the speaker announces his own action, or commands the action of others. Hence, by a natural transference, it comes to express the purpose of another person (viz. the subject of the principal clause). Similarly the optative, whether as the mοοd of wish or of expectation, comes to express a wish or expectation not now felt, but spoken of. Again, by virtue of its character as a softened or less confident future, it naturally expresses a purpose that does not lie within the speakerʼs own sphere of action or direct influence.
It shoud be noticed, too, that the relation which we imply by the term "final clause" may exist without grammatical subordination, i.e. without a particle such as ἵνα or ὡς to introduce the clause. Thus in
Il. 6.340 ἀλλʼ ἄγε νῦν ἐπίμεινον ἀρήία τεύχεα δύω
the meaning would not be altered by saying ἐπίμεινον ἴνα δύω. So in
Il. 18.121-125 νῦν δὲ κλέος ἀροίμην καὶ . . . στοναχῆσαι ἐφείην, γνοῖεν δʼ ὡς δὴ δηρὸν ἐγὼ πολέμοιο πέπαυμαι
the last wish is evidently also the result hoped for from the fulfilment of the preceding wishes (so that γνοῖεν δέππ ὡς γνοῖεν).
In conditional clauses, on the other hand, the condition or supposition is not subordinated to the time of the governing verb, but is made from the present point of view of the speaker. The question arises: What is the original force of the subjunctive and optative in this use?
In the case of the subjunctive we naturally look to the quasi-imperative use. It is common to use the imperative as a way of stating a supposition, as when we say "let it be so," meaning "if it is sο" (cp. Latin cras petitο, dabitur). This view is confirmed by the fact that negative conditional clauses take μὴ, not οὐ; that is to say, they are felt to be akin to prohibition rather than denial. Thus ὃς μὴ ἔλθῃ literally means not "who will not come" (ὃς οὐκ ἂν ἔλθῃ), but "who shall not come," i. e. whom we are not to suppose coming.
Similarly we may understand the optative in these clauses as the mood of concession—"admitting this to be so"—and so in a negative sentence, ὃς μὴ ἔλθοι "whom I agree to suppose not coming." For the choice of the mood does not depend on the greater or less probability of the supposition being true, but on the tone in which it is made—on the degree of vividness, as Mr. Goodwin says, with which it is expressed (Moods and tenses, § 455).
It may be objected that on this view we ought to have εἰ οὐ, not εἰ μή, whenever the verb is in the indicative. But there is no difficulty in supposing that μή was extended to the indicative on the analogγ of the clauses with the subjunctive and optative; just as μὴ ὤφελον is an extension from the common use of μὴ in wishes. And this is strongly supported by the circumstance that in fact εἰ οὐ with the indicative occurs several times in Homer.
Il. 15.162 εἰ δέ μοι οὐκ ἐπέεσσʼ ἐπιπείσεται κτλ. (so 178)
Il. 20.129 εἰ δʼ Ἀχιλεὺς οὐ ταῦτα θεῶν ἐκ πεύσεται ὀμφῆς
Il. 24.296 εἰ δέ τοι οὐ δώσει ἑὸν ἄγγελον κτλ.
Od. 2.274 εἰ δʼ οὐ κείνου γʼ ἐσσὶ γόνος κτλ.
See also ll. 4.160; Od. 12.382, 13.143. On the other hand, in the very few examples of εἰ οὐ with a subjunctive, the οὐ goes closely with the verb, viz. Il. 3.289 (οὐκ ἐθέλωσιν), 20.139 (οὐκ εἰῶσι). On the whole, therefore, it is probable that the subjunctive in conditional clauses represents the tone of requirement in which the speaker asks us to suppose the condition to be true, and that the optative. implies concession, or willingness to make the supposition involved.
317. Original meaning. Whether the use of the subjunctive as an emphatic future was derived from its use to express will, or vice versa, and whether the optative originally expressed wish or supposition, are questions which take us back to a very early period in the history of Indo-European speech. The two moods are found in the same uses (generally speaking) in Homer and in the Veda; the formation of these uses therefore belongs in the main to the period before the separation of the different languages—to the period, indeed, when the original parent language was itself in course of formation. The problem therefore is one on which comparison of the earliest forms of the known Indo-European languages can hardly throw any light. It is as though we were asked to divine whether the use of shall in commands (thou shalt not kill) or in predictions (you shall see me) is the older, without recourse to earlier English, or to other Germanic languages. Some considerations of a general character may however be suggested.
a. The subjunctive is strongly differentiated from the imperative by its personal endings, and especially by the existence of a 1st person.
b. In most languages it will be found that the imperative meaning is expressed in more than one way. Thus in Sanskrit we find the imperative proper, the injunctive, the subjunctive, and the optative; in Greek the imperative, the subjunctive and certain uses of the future. The reason of this is evident. Variety in the expression of will and wish is one of the first needs of human society. The form which has been appropriated to express command is unsuitable to courteous request, still more unsuitable to humble entreaty. Accordingly other forms are used, precisely because they are not imperatives. In time these acquire a quasi-imperative character, and fresh forms are resorted to as the same want of a non-imperative mode of expression is again perceived.
c. The use of the secondary endings in the optative points to the conclusion that in its origin it was a mood of past time. The tendency to use a past tense in wishes, and in some kinds of supposition, may be amply illustrated from English and other modern languages.
d. The uses with οὐ go far to show that the quasi-future sense of the subjunctive and optative is at least as primitive as the quasi-imperative sense. If the strong negation οὐ γένηται is derived by gradual change of meaning from a prohibition, the appearance of οὐ is difficult to explain.
e. The use of the Subj as an imperative may be compared to the Attic use of the future in a "jussive" sense, and in final clauses to express purpose (Goodwin, p. 373). The change from an expression of will to one of expectation is one to which it wοuld be much more difficult to find a parallel.
318. Conditional Protasis with εἰ. The derivations that have been proposed for the particle εἰ or αἴ are too uncertain to furnish ground for any theory as to the manner in which the conditional protasis may have been formed. The question arises for us on the passages in which εἰ with the optative is used to express a wish, Thus in εἴ τι καλέσειε I pray sοme one to call we may take the clause as conditional, with a suppressed apodosis (καλῶς ἂν ἔχοι or the like). Or we may follow L. Lange in holding that the Clause is not subordinate at all, the particle εἰ being originally a kind of affirmative interjection, used to introduce expressions of wish and supposition. We can thus explain the ordinary complex conditional sentence as made up of two originally independent clauses
(1) a wish or supposition, introduced by εἰ, and
(2) an assertion of the consequence to be expected from its being realised.
On this theory the clause of wish introduced by εἰ is not an incomplete sentence, derived from a complex sentence by omission of the apodosis, but is one of the elements from which the complex sentence was itself developed.
The latter of these views has a priori the advantage of deriving the complex from the simple, and it has some apparent support in Homeric usage. We find in Homer
- Wish, standing alone.
ὡς ἀπόλοιτο καὶ ἄλλος ὅτις τοιαῦτά γε ῤέζοι
- Wish followed by an independent clause expressing expectation of a consequence.
Od. 15.180 οὕτω νῦν Σεῦς θείη, ἐρίγδουπος πόσις Ἥρης·
τῷ κέν τοι καὶ κεῖθι θεῷ ὡς εὐχετῴμην
Il. 13.55 σφῶϊν δʼ ὧδε θεῶν τις ἐνὶ φρεσὶ ποιήσειεν,
αὐτ θʼ ἑστάμεναι κρατερόῶς καἰ ἀνωγέμεν ἄλλους·
τῷ κε καὶ ἐσσύμενόν περ ἐρωήσαιτʼ ἀπὸ νηῶν
- Wish, with εἰ, εἰ γάρ, εἴθε, etc., but without apodosis.
Il. 4.189 αἲ γὰρ δὴ οὕτως εἴη, φίλος ὦ Μενέλαε.
Il. 11.670 εἴθ’ ὣς ἡβώοιμι, βίη δέ μοι ἔμπεδος εἴη, κτλ.
- Wish, with εἰ, εἰ γάρ, εἴθε, etc., followed by a clause of consequence.
Il. 7.157 εἴθʼ ὣς ἡβώοιμι, βίη δέ μοι ἔμπεδος εἴη·
τῷ κέ τάχʼ ἀντήσειε κτλ.
Od. 15.536 αἲ γὰρ τοῦτο, ξεῖνε, ἔπος τελέσειε Κρονίων·
γνοίης χʼ οἵη ἐμὴ δύναμις καὶ χεῖρες ἕπονται
- Supposition, with εἰ, followed by a clause of expectation.
Il. 7.129 τοὺς νῦν εἰ πτῶσσοντας ὑφ’ Ἕκτορι πάντας ἀκούσαι,
πολλά κεν ἀθανάτοισι φίλας ἀνὰ χεῖρας ἀείραι
The similarity in these examples is manifest. The type in the first four sets consists of a clause of wish, either alone (1 and 3) or followed by a clause of consequence (2 and 4). Again, (5) only differs from (4) in punctuation, so to speak: the two Clauses are taken together, and the the εἱ clause is no longer an independent supposition, but is one made with a view to the expressed in the clause with κεν. And this, it is contended, was the result of a gradual process, such as we find whenever parataxis passes into hypotaxis.
319. Final Clauses with εἰ. An argument for Langeʼs view of the original force of εἰ is found in the use in final clauses, such as εἴμι εἴ κε πίθηται. The meaning here is essentially different from that of the conditional sentence I go if he listens; and on the ordinary hypothesis, that εἰ originally expressed a condition, it is difficult to account for the two uses. But if εἰ is a mere interjection, introducing wish or supposition, it is intelligible that the clause should be conditional or final, as the context may determine.
320. The formula εἰ δʼ ἄγε, with the varieties εἰ δʼ ἄγετʼ (Il. 22.381) and εἰ δὲ (Il. 9.46 & 262), is often used in Homer to introduce an imperative or subjunctive (§ 275). It has generally been supposed to be elliptical, standing for εἰ δʼ ἐθέλεις ἄγε, or the like. And εἰ δʼ ἐθέλεις is actually found with an imperative in a few places.
ll. 19.142 εἰ δʼ ἐθέλεις ἐπίμεινον
Od. 16.82, 17.277 (cp. 3. 324). It has been pointed out, however, by Lange, in his dissertation on this question1
that εἰ δʼ ἐθέλεις is only found where it introduces a distinct second alternative. Thus in Od. 16.82 the context is: "I will send the stranger wherever he desires; or if you choose (εἰ δʼ ἐθέλεις) take him into your house." So Od. 3.323-4 ἀλλʼ ἴθι νῦν σὺν νηΐ . . . εἰ δʼ ἐθέλεις πεζός κτλ. But with εἷ δʼ ἄγε this is not the case. We find it at the beginning of a speech; as
Il. 6.376 εἰ δʼ ἄγε μοι, δμωαί, νημερτέα μυθήσασθε
Od. 2.178 ὧ γέρον, εἰ δʼ ἄγε νῦν μαντεύεο κτλ.
so Il. 16.697, 17.685; Od. 12.112, 22.391, 23.35.
Or in the apodosis of a conditional sentence, as
Od. 4.831 εἰ μὲν δὴ θεός ἐσσι, θεοῖό τε ἔκλυες αὐδῆς,
εἷ δʼ ἄγε μοι κτλ.
so Il. 22.379-381.
Or to express an appeal which is consequent upon something just said.
Il. 1.301 τῶν οὐκ ἄν τι φέροις ἀνελῶν ἀέκοντος ἐμεῖο
εἰ δʼ ἄγε μὴν πείρησαι
(ay, come now and try)
Cp. ll. 8. 18.
Il. 1.523 ἐμοὶ δέ κε ταῦτα μελήσεται ὄφρα τελέσσαω·
εἰ δʼ ἄγε τοι κεφαλῇ κατανεύσομαι
(so come, I will nod my head).
Il. 23.579-85 εἰ δʼ ἄγ ἔγὼν αὐτὸς δικάσω, καί μʼ οὔ τινά φημι
ἄλλον ἐπιπλήξειν Δαναῶν· ἰθεῖα γὰρ ἔσται·
Ἀντίλοχʼ, εἰ δʼ ἄγε δεῦρο . . .
come I will be judge myself . . . so come, Antilochus, take this οath
See also Od. 1.271, 9.37, 21.217, 24.336.
321. Conclusion. Notwithstanding these arguments, the common explanation of the εἱ clause of wish (as primarily a clause of supposition) seems to be the more probable one.2
- The uses of εἰ present a marked correspondence with those of the relative and its derivatives. Note especially the use of ὅτε μή as almost exactly = εἰ μή.
- The analogy εἶτα : εἰ :: ἔπειτα : ἐπεί makes it likely that εἰ was originally temporal. The fact that εἶτα is not Homeric takes something from the force of this argument.
- The use of alternative forms of wish, and the use of some form of apposition to express wish, are phenomena which can be exemplified from many languages; cp. the Latin ο si, German wenn, wenn nur, etc. And ellipse of the apodosis occurs with εἰ clauses of other kinds; see § 324.3
- The εἰ clause, whether of supposition or of wish, is specifically Greek, whereas the chief meanings of the optative—wish, concession, supposition—are much older, being common to Greek and Sanskrit. Hence the εἰ clause was formed at a time when the Optative of Wish had long been established in use. The presumption surely is that the εἰ clause, when it came to be used as a form of wish, was a new way of expressing wish. It would probably be adopted at first as a less direct form, suited for wishes couched in a different tone (as εἴθε is confined to hopeless wish).
- The only use of εἰ not obviously expressive of supposition is that which is seen in the isolated phrase εἰ δʼ ἄγε, of which Lange has given an exceedingly probable analysis. Possibly bhowever the εἰ of εἰ δʼ ἄγε is not the same wοrd as εἰ if, but an interjection, like εἶεν and Latin eia. We may gο further, and point out that the δέ of εἰ δʼ ἄγε has been shown by Lunge himself to be out of place, hence the true form may be εἷ’ ἄγε, like Latin eiα age.
It may be observed, in conclusion, that the question of the εἱ clause is quite distinct from the question of the original meaning of the optative. It is possible to combine Langeʼs theory of εἰ vwith Delbrückʼs earlier view of the optative as originally the mood of wish,4
but Lange himself does not do so. He regards the εἰ clause of supposition (Fallsetzung) as developed independently of the εἰ clause of wish. His main thesis is that εἰ does not imply a correlative particle, or an apodosis (καλῶς ἂν ἔχοι or the like), so that the two meanings of εἰ γένοιτο—suppose it happened and would that it happened—belong to originally distinct meanings of the optative γένοιτο. That is to say, the development of εἰ if with various moods—optative, subjunctive, indicative—was parallel to an entirely distinct development of interjectional εἰ with the Optative of Wish.
322. Homeric and Attic uses. The main difference between Homer and later writers in regard to the moods may be said to be that the later uses are much more restricted. Thus the subjunctive is used by Homer in principal clauses of every kind—affirmative and negative, as well as prohibitive, interrogative, etc. In Attic it is confined to the prohibitive use with μή, and the idiomatic hortatory and deliberative uses.
Again, in subordinate clauses the important Homeric distinction between the "pure" subjunctive and the subjunctive with ἄν or κεν is almost wholly lost in Attic. In clauses of conditional meaning, whether relatival, tempοral, or introduced by εἰ, the subjunctive with ἄν has become the only generally allowable construction, the pure subjunctive being confined to a few instances in poetry. With the optative, on the other hand, an equal uniformity has been attained by the loss of the use with ἄν or κεν. In short, of the four distinct Homeric constructions
1) ὅς ἔλθῃ (ὅτε ἔλθῃ, εἰ ἔλθῃ, etc.)
2) ὅς ἂν (or ὅς κεν) ἔλθῃ (ὅτʼ ἂν ἔλθῃ, ἑὰν ἔλθῃ, etc.)
3) ὅς ἔλθοι (ὅτε ἔλθοι, εἰ ἔλθοι, etc.)
4) ὅς ἂν (or ὅς κεν) ἔλθοι (ὅτʼ ἂν ἔλθοι, ἑὰν ἔλθοι, etc.)
the language dropped the first and last, with the result that as ἄν alwayσ accompanied the subjunctive and was absent from the optative, it ceased to convey a distinct meaning, independent of the meaning given by the mood. In other words, the use became a mere idiom. The change, though apparently slight, is very significant as an evidence of linguistic progress.
In regard to final clauses the most noticeable point is the use of the relative with a subjunctive. In this respect Homeric Greek agrees with Latin, while in later Greek the subjunctive was replaced, generally speaking, by the future indicative. It is also worth observing here that in Homer, as has been said (§ 316), the final clause in the great majority of instances expresses the speakerʼs own purpose, not a purpose which he attributes to a person spoken of; see §§ 280, 281, 285, 286. In other words, the subordination of the clause to the governing verb does not often go so far as to put the 3rd person for the 1st (e. g. φράσσεται ὥς κε νέηται = he will consider—"how am I to return"). The further license by which a past purpose is thought of as if still present—sο that the subjunctive is used instead of the optative—is not Homeric (§ 298).
- 1De fοrmula Homerica εἰ δʼ ἄγε cοmmentatiο, Lipsiae 1873.
- 2This is also the conclusion maintained by Mr. Goodwin, who discuses the question very fully in the new edition of his Moods and Tenses (pp. 376 ff.).
- 3This is also the conclusion maintained by Mr. Goodwin, who discuses the question very fully in the new edition of his Moods and Tenses (pp. 376 ff.).
- 4This view was proposed in Delbrückʼs Syntaktische Fοrschungen (vol. i. p. 13), but is withdrawn in his recent work (Altindische Syntax, § 172).