^{page 419 note 1} The paper is continued from the Publications of the Modern Language Association of America, xxvi, 2, 1911, pp. 379-418.

^{page 424 note 1} I cannot attempt, in the present paper, to trace the development of the various powers of the Romance conditional from its original one of past futurity, but must content myself with merely mentioning some of them, here and there,

^{page 427 note 1} There is a very rare use of the subjunctive in Greek to express the same idea in a question, as in Il., 1, 365. I know of three sure cases only. It is probable, then, that the subjunctive did not make any considerable contribution to the “subjunctive” construction of this kind in the languages in which subjunctive and optative are merged in one mood.

^{page 429 note 1} Rarely, but much more satisfactorily, even if the source is not mentioned, a category which one has laid down is promoted to a new office in the work of another, or appears without change, but under a new name. In a review of the Hale-Buck Latin Grammar in the Classical Review, Sonnenschein held that I had proposed too many categories. But he has since that time taken one of the new ones, that of obligation, and, in The Unity of the Latin Subjunctive: A Quest, 1910, has made it the foundation of his scheme for the entire mass of subjunctive uses. In the New Latin Grammar, 1912, he finds the second part also of my phrase to be helpful in the form “obligatory or proper.” He also, in the latter work, practically employs my category of natural likelihood, in the phrase, “what …. is likely to happen.” Bennett, in his recent Syntax of Early Latin: The Verb, found that the category “potential” did not suffice, and was forced, in phrases that seem difficult, to distinguish the “should-would potential” (my “subjunctive of ideal certainty”), from the “can-could potential” (the “potential” in my terminology). He was also obliged to add a “subjunctive of duty or fitness” (my “subjunctive of obligation, propriety, or reasonableness”). He further, through his translating, pointed out that the subjunctive sometimes expresses an act as “natural,” as “likely” (my “subjunctive of natural likelihood”). Having previously adopted my “volitive subjunctive” and “subjunctive of an act anticipated” in his grammar, he thus at the present moment has, under one name or another, every one of my new categories.

In Germany, my views have fared better. Thus Delbrück in his Comparative Syntax, speaking of modern work upon the Latin subjunctive, mentions mine (“vor allen von W. G. Hale”), and, in his treatment of Greek, says that he “now classifies the Greek subjunctive uses with Hale,” referring to my Anticipatory Subjunctive in Greek and Latin, which was in good part a criticism of his own previous positions. Similarly Maurenbrecher, in his recent work upon Greek Syntax, expressly says that his treatment of the subjunctive is based upon my Anticipatory Subjunctive. And even one American grammar of Greek seems to have been influenced in part by my system. Babbitt, § 563, writes, “A statement of a future possibility, propriety, or likelihood, as an opinion of the speaker, stands in the optative with (Potential Optative) …. Note.—Observe that the potential optative may express all shades of opinion, from mere suggestion of possibility to ideal certainty, and the English rendering should be made to suit the context.” I mention these saving facts lest English-speaking workers should infer, from my diatribe about the indifference of grammarians in general, that the views which I offer for consideration need not detain them.

^{page 435 note 1} The subject matter of the earliest literature of these languages is such as to offer extremely little opportunity for expressions of an obligation in an individual case. There is thus no positive evidence against the existence of such a use. It is most unlikely too, that, even if the optative was first used to say “one ought to do so-and-so,” it could have been prevented from being used to say “you ought to do so-and-so,” etc.

^{page 435 note 2} In the above, I use the phrase “my theory of the free use of the optative to express obligation.” It is commonly referred to as Elmer's theory (the reference being to Latin), as e. g., by Morris, Bennett, and Sonnenschein. Elmer, while deriving the power differently, namely, from what I have named the subjunctive (for Latin) of ideal certainty, did use the idea and the name in print before I did. As he himself explained in the Latin Leaflet, he heard both idea and name mentioned so freely in my class-room at Cornell University, where he began his teaching as my assistant, that it did not occur to him that it was not common doctrine. The theory was, in point of fact, first put forward by me, and forms one of the principal foundation-stones for my whole mood-system of the Indo-European verb.

This system, though fully formed in 1886, and often discussed with my colleagues and students, was not put forth until 1901, when I published an abstract of it (intending a fuller treatment elsewhere) under the title Leading Mood-Forces in the Indo-European Parent Speech, in the Proceedings of the American Philological Association, xxxii. It is also briefly given in the Hale-Buck Latin Grammar.

^{page 445 note 1} Brugmann had put forth a similar theory, before me, to explain the origin of the historical present, and the present indicative in the sense of a future in independent statements. But his theory did not touch upon the use in dependent clauses, nor upon the historical concurrence of this present indicative with the other moods. My conception is that the set of forms from which that which we now call the present indicative came down (or, as I should like to call it, the primitive verb), served, with the help of the situation, of gesture, facial expression, etc., to convey all modal ideas, as well as all temporal ideas.

I have discussed the question more fully in my Anticipatory Subjunctive, and in Controlling Conceptions in Syntactical Study, School Review, June, 1902.

^{page 448 note 1} Inheritance from the subjunctive of the parent speech.

^{page 448 note 2} Inheritance from the optative of the parent speech.

^{page 448 note 3} Special Latin development.

^{page 452 note 1} In an address before the Michigan Schoolmasters' Association, in March, 1911, I treated the subject of these two articles, more briefly, but with the inclusion of some things here omitted. A number of specialists followed me. The address and the discussion were published in the School Review, and in the University of Michigan Linguistic Studies, ii, 1911.

^{page 453 note 1} In the familiar style, commands are often introduced by ut, which is probably not due to ellipsis, but came into the sentence as the formal opposite of ne (see “formal ut,” Hale-Buck Latin Grammar, p. 261, ftn. 2). Romance que, che, que, as used in qu'il vienne, is probably descended from this.

^{page 453 note 2} In Latin, Romance, and Germanic the subjunctive of request, of consent (“concession”), and of proviso is probably of both volitive and optative origin. But these forces are here placed, for convenience, under the head of the volitive.

^{page 454 note 1} The determinative clauses—probably the most frequent of all relative clauses, though until lately without a name—are clauses which tell what person, what thing, etc., is meant. They are mostly clauses of fact, and so in the indicative. Thus in “cursed be the hand that fired the shot.” They fill out an incomplete determinative pronoun—whence the name.

^{page 455 note 1} English may in the common expression of purpose is historically not potential, but a volitive subjunctive (like German möge), expressing ability as aimed at, precisely as in Latin ut vivere possis.

^{page 455 note 2} The Greek subjunctive condition is of volitive and anticipatory origin, the Greek optative condition of optative and potential origin, and the Latin subjunctive condition probably of all four origins (see the author's Subjunctive and Optative Conditions in Greek and Latin, in Harvard Studies in Class. Phil., xii, 1901).

^{page 456 note 1} The quod sciam phrase (which is succeeded in Romance by que je sache, etc.) probably owes its mood to the fact that it nearly always follows a negative or a word meaning first, last, or only—types with which, in Latin, the subjunctive is especially associated. See the author's Cum-Constructions, German translation, Teubner, 1891 (problem not solved at date of American edition. The current explanation had been that the clause was “potential”).

^{page 456 note 2} The Latin clause of rejected reason, with non quod, quia, or quo (whence Romance pas que, etc.), is only an outgrowth of the clause of reason in “indirect discourse.”