primitive a had become e in the European parent speech. (Ueber die Spaltung des a-Lautes etc., in the Berichte der Königl. sächs. Ges. der Wiss., 1864, pages 9-42.) This view, which long prevailed, sustained its first shock from the discovery that the Armenian shares this e. Ought this language to be reckoned among the European ones, or must we drop the pre-supposed originality of the a, and assume that an e or å existed in the parent speech, and became a in those few languages which do not retain it, which a is not distinguished (at least in the written language) from other varieties of a, so that the lack of originality would fall, not on the European, but on the Indo-Iranian side?

The truth of this assumption appears more than probable when we consider the relation of the Indian (and Iranian) palatals to the following vowels. We observe that in numberless instances a palatal springs from a guttural through the influence of a following i (so, for example, ojiyān belongs to the positive ugrás). We often find that a has the same influence as i, but only the a which corresponds to e in the other languages, as for instance the a of the reduplication-syllable in cakára, from kar "make", which corresponds to e in Greek, German etc., or the a of ca, which answers to a Greek and Latin e, and so on. It follows from these facts, which could be multiplied indefinitely, that the above-mentioned a of the Sanskrit must have borne some resemblance to i, and hence that it must have been e or å, which proves the existence of an e in all Indo-European languages. Now if it can be proved (and it can) that this palatalization which we find in Sanskrit must have extended back into the parent speech, the originality of this e (a) is as surely demonstrated as it is possible to demonstrate any assertions of this nature. Without entering here into further details, I will refer the reader to an essay by JOHANNES SCHMIDT (Kuhn's Zeitschrift, 25, page 1 seq.), comprehending the previous labors of AscOLI, BRUGMAN, FICK, COLLITZ and others, on the whole question of the two guttural series and the European e, in which he discusses the tenability of the new theory with the strictest attention to detail.

Beside the discovery of the Indo-European e-vowel we

may range the very probable supposition (originating with OSTHOFF) that a "sonant" (i. e. syllabic) r, such as we are acquainted with in Sanskrit, and which is represented in Greek by ap (pa), already existed in the primitive speech. (Zur Frage des Ursprungs der germanischen n-Declination, in Paul & Braune's Beiträge, 3, page 1 seq.) According to this view, the a of the Greek aorist tрañоv can no longer be regarded as an isolated remnant of a former condition of the language, whose preservation is due to a certain confusion of the linguistic instinct (CURTIUS, Grundzüge, 5th edition, page 52), but pa is the regular representative of the Indo-European r-vowel, which is appropriate to the aorist. No less important is BRUGMAN'S assumption of a syllabic nasal (Nasalis sonans in der indogermanischen Grundsprache, in CURTIUS' Studien, 9, page 287 seq.), and similar researches.

These discoveries, showing, as they did, in a most striking manner that even trifling differences in pronunciation (e. g. the difference between å and a) are propagated with wonderful fidelity throughout the ages, contributed in great degree toward increasing the respect felt for the regularity of all phonetic change.

BOPP had granted to sounds the privilege of changing, now in this way, now in that, within the same linguistic period, and this freedom had been considerably restricted by POTT, SCHLEICHER and CURTIUS; when, now, still other important limitations were added, the idea naturally arose that phonetic laws admit of absolutely no exceptions. We shall show later in detail that this new theory necessarily caused greater stress to be laid upon the principle of analogy.1)

1) The first person who, to my knowledge, clearly expressed the view that the phonetic laws admit of no exceptions, is LESKIEN (Die Declination im Slawisch-Litauischen und Germanischen, Preisschriften der Jablonowski'schen Gesellschaft in Leipzig, Leipzig 1876, pages XXVIII and 1). He says:

"In my investigations I have started with the principle that the form of a certain case, as we meet with it, can never result from an exception to phonetic laws which are observed elsewhere. To prevent misunderstanding, I will add : if by ‘exception' be understood those cases where the expected phonetic change has not taken place from definite ascertainable causes, such as the absence of Lautverschiebung in German in phonetic

We must mention in conclusion that those scholars who advocate the infallibility of phonetic laws have often emphasized the fact that the natural constitution of language is not manifested in the cultivated tongues [Kunstsprachen], but in the dialects of the people. The guiding principles for linguistic research should accordingly be deduced, not from the obsolete written languages of antiquity, but chiefly from the living popular dialects of the present day.

These are the principal views, endeavors and hypotheses which usher in a new treatment of the problems of linguistic science. The chief of these problems will now be discussed in the following chapters.

CHAPTER V.

THE AGGLUTINATION THEORY.

We have shown in the previous pages how the so-called "agglutination theory" originated with FRANZ BOpp, and it has been at least intimated what part this hypothesis has

groups like st etc., where one rule to a certain extent interferes with another, then of course there is nothing to be said against the statement that phonetic laws are not infallible. For the law is not nullified in such circumstances, and works as we should expect it would do wherever these or other disturbances, i. e. the influence of other laws, are not present. But if we admit arbitrary, accidental deviations, such as are incapable of classification, we virtually confess that language, which forms the object of our research, is inaccessible to scientific investigation."

The remarks of OSTHOFF and BRUGMAN are in the same spirit (Morphologische Untersuchungen, 1, page XIII):

"All phonetic change, in so far as it occurs mechanically, is accomplished in accordance with exceptionless laws, i. e. the direction of the phonetic movement is always the same among all members of a linguistic community, except where a division into dialects takes place; and all words in which the sound subject to this phonetic movement appears under like circumstances, are without exception affected by the change.”

Beside this we find also the extreme view that all phonetic laws work blindly, with a blind natural necessity, and the like.

played in the further development of linguistic research. It will now be my task to ascertain what degree of probability can be adjudged to it.

Every analysis of the Indo-European inflectional forms must start with the fact that certain inflectional endings of the verb show a great resemblance to certain pronominal stems. The termination of the first person, -mi, calls to mind at once me, mi-hi and the rest of the series, and in the same way the -ti of the third person reminds us of the pronominal stem ta, which appears in tóv etc. The endings of the second person also exhibit certain analogies to the corresponding pronoun, although these are not so unmistakable as in the case of the other two persons. Now BOPP explained this similarity by assuming that the pronouns were affixed to the verb, which accordingly possessed no endings before the affixion; and the idea of agglutination expressed in this hypothesis became the prevailing one in his entire explanation of inflection. But it is evident that beside Bopp's assumption it is possible to form others, with the same fact for a starting point. Thus far two such hypotheses have appeared; the first with the assumption that the endings existed first, and the pronouns were formed from these by freeing them from the stem, the evolution theory; and the second, according to which the pronouns and the endings arose independently of each other, and were afterwards brought into relation, the adaptation theory.

I will first discuss these two hypotheses.

The evolution theory is older than the agglutination theory, since FRIEDRICH V. SCHLEGEL was its first advocate; yet there exists no authentic demonstration of it, for neither AUGUST WILHELM V. SCHLEGEL, nor LASSEN, nor any other scholar of this school has opposed anything except negation to BOPP's arguments. Under these circumstances we must have recourse to the works of three men, neither of whom can be regarded as a recognized exponent of SCHLEGEL's doctrine; I refer to CARL FERDINAND BECKER, MORITZ RAPP and RUDOLPH WESTPHAL. What C. F. BECKER, the once renowned author of the Organism, can bring forward in support of the originality of the personal suffixes, is essentially reduced to the following observation :

"Since the word was originally a member of a sentence, the grammatical relation was given in the beginning with the notion of the word, and its inflection was given with the word itself. The word as expression of the notion, and the inflection as expression of the grammatical relation, are equally old and original."

But this reasoning would only be valid if we were obliged to assume that everything which is thought finds expression in language. Now it is evident that this is by no means the case, and therefore nothing prevents us from assuming that the idea of relation had existed long before it was expressed by linguistic means. Accordingly, from this deductive method of observation we can draw no conclusion respecting the age of the expression of grammatical relation.

As regards the second of the above-mentioned men, MoRIZ RAPP of Tübingen, I will refer the reader to a notice of his Comparative Grammar by STEINTHAL (Kuhn's Zeitschrift, 2, page 276 seq.), where precisely the point in question is discussed. On the other hand, the views of RUDOLPH WESTPHAL, as set forth in his Philosophisch-historische Grammatik der deutschen Sprache (Jena, 1865), and his Methodische Grammatik der griechischen Sprache (Jena, 1870), demand a more detailed consideration.

WESTPHAL'S system is, in brief, as follows. In the development of language three periods can be distinguished, according to the formation of the roots. In the first, things are characterized independently, in the second, in relation to human thought, in the third, in relation to each other. (Phil.hist. Gr., page 98.) In the first period the nominal stems originated, in the second verb-inflection, in the third noun-inflection. By means of the root a name was given to being, as that in which a definite motion or activity is manifested. Now this root is sometimes employed to designate independent being, but usually it is phonetically changed for this purpose, being amplified by the addition of an ă, i or u. WESTPHAL expresses himself as follows, concerning the meaning of this amplification :

"In contrast to the monosyllabic verbal root, the concrete noun thus obtains a dissyllabic form, whose terminal vowel is