languages. That these three branches of the I. E. stock should have, late and (apparently) independently, worked out this process so similarly, creates a strong feeling that there may be a fundamental underlying cause that holds good for them all. Hence a theory that seems to explain the facts of Latin very well, but breaks down utterly when applied to Greek, needs very strong proof of its validity before it can be received with confidence.

(3). The theory advanced must be in accord with the laws of simple unconscious growth in language. Upon this rock has split many a brilliant hypothesis. The validity of any theory is in doubt if it cannot bear the scrutiny of the question: Is the process here assumed conceivably a description of what might actually take place in language growth, and are there any simple wellestablished parallels in which such a process has actually taken place? I feel that we sometimes expect the true explanation to be so abstruse and far to seek that there is danger of overlooking a simple (and perhaps true) one lying close at hand; in actual fact, the secret of some changes may be found in a thing so simple as phonetic decay or a leveling by analogy. A striking case of this last appears in the subsequent history of the unreal speechform, when the indicative in Old French takes the place of the Latin subjunctive in the si-clause. The entrance of the indicative into the protasis of the unreal conditional sentence seems due to the analogy of the many si-clauses that use that mode, for in cases where si is omitted or repeated by que the original subjunctive is still retained in the unreal condition.1

Keeping in mind the above necessary conditions of a valid theory, it now remains to find some way of explaining the change by which the imperfect subjunctive in Latin came to leave its early function and finally became the accepted expression for the present unreal conditional sentence. If an explanation is to be sought which may apply to Greek and the Germanic languages as well as to Latin, it must be sought in something common to all these languages. The obviously common factor is the unreal thought-form. If the key to an explanation does lie in the unreal thought-form, it is to the past rather than to the present unreal

1O. M. Johnston, Modern Language Notes, xiv, p. 270 ff. Compare what Sweet has to say of the history of the speech-form in English; New English Grammar, $2280-2.

that attention should be first directed.' I hope to show that there is a peculiarity of this past thought-form which may serve to explain how its original speech-form (imperfect subjunctive) first became associated with the present idea, and, later, identified with it. The past unreal conditional sentence, strictly speaking, has no tense-force of its own, but it is opposed to, or reflects, the time of various past realities. In the nature of things, the time of past events is not all of one kind; sometimes it concerns only one point (aorist), at others it is continuous (imperfect); sometimes it implies a present result or continuance (perfect definite). The past unreal conditional sentence can reflect any of these time aspects. For instance, one man might say to another, 'You are very hard on that foster child,' and the other reply, 'If he had not deceived me on the day he entered my home, I should have loved him as a son.' In this sentence the protasis is opposed to a reality of the aoristic variety, but the apodosis is opposed to a reality that extends all the way from that point of time up to, and including, the present. In fact, the same sentence might perfectly well be used in reply, if the first speaker's remark be taken

'At this point I part company with other investigators. Partly as the result of the unclearness arising from a failure to differentiate speech-form and thought-form, many have tried to account for the use of a past tense-form as the expression of the present unreal thought-form either by stretching the present unreal thought-form a little or by finding in it some implication which might be brought out by the use of a past tense-form. For the first of these see Blase, Geschichte des Irrealis, p. 16; Dittmar, Studien zur Latein. Moduslehre (Leipzig, 1897), $300, takes almost the same view, Wer einen Irrealis ausspricht, versetzt sich nämlich jedesmal in die Vergangenheit, wenn auch in eine, die nur um ein paar Sekunden zurückliegt.' On the other hand, it is sometimes claimed that there is an implication of impossibility in the unreal thought, and that this justifies the use of a past tense-form, since a past tense implies impossibility of fulfillment. (See Blase, l. c. p. 14). I trust that the method of attacking this question which I am about to outline above, may appeal to the reader as more in accord with the working of the laws of language growth, and hence, more likely to be correct.

...

2 Examples of the perfect definite variety are of special interest for this discussion, Cicero, Phil. II. 36. 90; Qui tu vir, di immortales, et quantus fuisses, si illius diei mentem servare potuisses! Phil. X. 4. 9; Si enim C. Antonius, quod animo intenderat, perficere potuisset, ... Macedoniam perdidissemus. p. Mur. 13. 29; In qua (defensione oratoris) si satis profecissem, parcius de eius laude dicerem. In the first two cases the thought so clearly includes the present that the author resumes with the imperfect subjunctive in the clause that follows. Cf. Livy XXI. 40. I (supersedissem), Tac. Agr. 34 (constitisset); A. J. P. XXI. p. 268 ff.

as referring strictly to some special instance of harshness in the present (rather than to the attitude in general); in that case the reply is a defence of the present position primarily, and its present force is very clear.

This is the open door through which the imperfect subjunctive in Latin may have first become associated, and then identified, with the present unreal thought-form. Even as late as Plautus, past tense usage in general is not very sharply differentiated, and it is very likely that, at the time the imperfect and the pluperfect subjunctive were the accepted expression for the past unreal condition, these tenses were used more or less interchangeably. In certain cases the imperfect or the pluperfect chanced to be opposed to a past reality of such a nature that there was nothing to keep the hearer from thinking of the present as well as the past. The next step would be to use these past tense-forms when consciously including both past and present. In the stages following, as the past speech-form came to be used as the expression of an opposition to realities whose past aspect was less prominent than the present, and, finally, to those whose thought was purely present, the imperfect gained upon the pluperfect and became the chosen expression for the present unreal conditional sentence. That the imperfect rather than the pluperfect should make good its claim on the present meaning is not to be wondered at in view of the fact that the imperfect and the pluperfect naturally form a pair-the pluperfect is, 'so to speak, the perfect of the imperfect." In Greek, the choice of the imperfect indicative rather than the aorist may have to do with the fact that the imperfect is more closely bound to the present, being made on the same stem.

That this development of meaning outlined for past tense-forms is quite possible and in accord with the laws of language growth is shown by the quite parallel and well-established process by which a perfect form like novi takes on present meaning. The perfect definite meaning 'I have become acquainted with', implies the present result 'I know'; this associates that past form with a present meaning, thus opening the door to the use of that form when only the present result is thought of, i. e. novi comes to be used freely like a present, with a loss of feeling that it is properly

1 Gildersleeve-Lodge Latin Grammar, §241.

a past. To the Latin student, it would be very interesting to examine the earliest cases of the imperfect subjunctive which show a distinctly present meaning, but this is denied us. In Plautus the process is so far advanced that few cases can be found where the old proper past meaning surely occurs-the imperfect subjunctive is pretty well established in its new function. Homer, however, seems to be just at the critical point of change in the function of the imperfect indicative, and the theory above proposed may be examined in the light of his usage.

There seems a general agreement among Greek scholars that the imperfect indicative in this idiom refers to continued or repeated past action. It is possible that it does more than this, as may appear from the following examples:

Od. iv. 178-9;

καί κε θάμ' ἐνθάδ ̓ ἐόντες ἐμισγόμεθ' · οὐδέ κεν ἡμέας
ἄλλο διέκρινεν φιλέοντέ τε τερπομένω τε.

On this passage Goodwin3 quotes Monro as saying 'the imperfect poyóueda takes in the present time, we should (from that time till now) have been meeting. In criticism of this he adds, 'It seems to me that, according to the Homeric usage, we can find no more

The fact that the perfect definite in general allows a primary sequence shows how inherent is present force in this tense-use.

2 I can hardly refrain from noting that it seems to me a mistake to insist, as Ameis does, that, in so early an author as Homer, this tense is always sharply differentiated from the aorist, that is, is always restricted to continued or repeated past action. Certainly such a claim is unreasonable in the case of v, for there is no aorist form to use. See also Od. iv. 732 ff.;

εἰ γὰρ ἐγὼ πυθόμην ταύτην ὁδὸν ὁρμαίνοντα·

τῷ κε μάλ' ἡ κεν ἔμεινε καὶ ἐσσύμενός περ ὁδοῖο,
ἡ κέ με τεθνηκυῖαν ἐνὶ μεγάροισιν ἔλειπεν.

Ameis explains the imperfect λɛɛv (he would have left me dead') as denoting continued action; but what of the aorist uɛive ('he would have remained')? I think that a person with no prepossession for either meaning would have settled on tueive rather than heεv as an expression denoting continuance. I am aware that this is not an altogether simple case, for the verb meaning of μέvo implies continuance, and in the following line the unreality lies not in ěheɩev but in tε0vηкviaν. But even so, I am not satisfied by Ameis' explanation of thεεv. If early Greek usage is anything like that found in Latin, I should not be surprised to find aorist and imperfect in this idiom not so far differentiated but that Homer could use as suited his verse either uɛve or ἔμενε, ἔλιπε οἱ ἔλειπε.

3 Greek Moods and Tenses, $435, foot-note.

in Oáμa éμoyóμedá ke than we should have had frequent meetings, and the rest comes from the context.' Goodwin denies (1. c.) that the imperfect in Homer is ever used in the present unreal conditional sentence, but even he admits that, in this case, there is nothing to prevent the hearer from thinking of the present as well as of the past (and the rest comes from the context'). This is the first step in the development outlined above for a past unreal form to take on present meaning. Monro seems to have felt (and I am not at all sure but that the feeling was right) that Homer has taken a second step-that Menelaus is here represented as consciously expressing opposition to a reality of the perfect definite type, and intends to include the present as well as the past.❜

Od. v. 311;

τῷ κ ̓ ἔλαχον κτερέων, καί μευ κλέος ἦγον ̓Αχαιοί·

Here Odysseus, in fear of perishing in the sea, has just expressed the wish that he had fallen in the battle over Achilles' body; line 311 tells what would have happened in that case (r). The first clause of the line is apparently opposed to a reality of the aoristic variety 'I should have enjoyed funeral honors,' but the case of hyov seems different. Odysseus may well have been thinking of the present as well as the past. Perrin, in his school edition, feels the present force so strongly here that he renders in his note 'would be carrying, wherever they went, i. e., spreading or cherishing.' In any case, this too is a situation in which the hearer would be justified in feeling a present force, even granting that the original speaker was not thinking of this especially himself.

Od. xiv. 61-2;

ἢ γὰρ τοῦ γε θεοὶ κατὰ νόστον ἔδησαν,

ὅς κεν ἔμ ̓ ἐνδυκέως ἐφίλει καὶ κτῆσιν ὅπασσεν.

In this passage the swine-herd Eumaeus is talking to (the unrecognized) Odysseus, expressing the opinion that the gods.

1 In the second edition of his Homeric Grammar (§324) Monro adopts Goodwin's statement of the usage in this idiom.

2 At the end of his foot-note Goodwin adds a nearer approach to the later usage perhaps appears in Il. xxiv. 220; εἰ μὲν γάρ τίς μ' ἄλλος ἐκέλευεν, ἐξ any other (had?) commanded me.' In this passage Priam means to emphasize the fact that the command is from Zeus and must be obeyed. The unreality lies not in ééλevev (for he has been ordered), but in åλλos; a fair rendering might be if it were some other that had bidden me.' In such a case I should not care to insist on present force for the imperfect.