with (with the final sound unvoiced as it appears in the Scottish legal preposition outwith) and as off, the final sound here also being unvoiced. If he migrates to England or to Australia he will probably in course of time adopt the pronunciation with a voiced final sound. In the course of years habit will become second nature, and in this respect the speaker's pronunciation will become identical with that of his neighbours. It is clear, however, that changes of this nature cannot take place on a large scale. If a large number of persons migrate in a body and continue to live in close intercourse with one another and but little in contact with the outside world, changes such as take place in the pronunciation of the individual emigrant do not occur. There can be no imitation of alien sounds, for there are none; no greater effort to be intelligible is required, for the audience has not changed. Hence it has been often remarked that a population which history shows to have remained undisturbed for very long periods in the same geographical situation manifests but little change in its language. Thus in Arabia and Lithuania the population has remained practically unmixed in the same habitat for thousands of years, with the result that the languages spoken there remain at the present day the most archaic members of the linguistic families to which they respectively belong.

hydrogen and one of oxygen will make water, and they will make | if he remains in his native place, as a rule pronounces them as nothing else at any time or at any place the world over. Phonetic laws, however, do not hold true universally. They are often curiously limited in the area to which they apply. In ancient Greek, for example, the sound -s- between two vowels, which had been handed down from the original language whence Greek and the sister languages are derived, regularly disappears; in Latin, on the other hand, it changes into -r-; thus an original genitive of a neuter substantive we find represented in Greek by yéve-os, a form which comparison with other languages shows to be traceable to an earlier *genes-os, preceding the separation of the languages, while the same original stem with a different vowel in the ending appears in Latin as gener-is. Similarly an early *euso appears in Greek as evw, in Latin as uro. This disappearance of original intervocalic s pervades all Greek dialects the apparent exceptions come under the heading of analogical change; with a very few exceptions similarly explicable Latin intervocalic s has become r. But Latin was originally limited to a very small part even of Italy, and the next neighbours of the Latins on the east and south-the Sabines, Campanians and Samnites-retained this intervocalics without changing it into r. On the other hand, the neighbours to the north-east-the Umbrians in and beyond the Apennines shared in this rhotacism. Yet the Celts, who bordered on the Umbrians along the Po, and who spoke a language in many respects very closely akin to the dialects of Italy, in this regard agree rather with Greek than the Italic languages. In Latin, again, the period of action of the law which changed intervocalic s into r did not in all probability exceed the century from 450 B.C. to 350 B.C. So unlike, indeed, are phonetic laws to the laws of natural science in universality that an opponent of the dogma which declares that phonetic laws have no exceptions has compared them with the laws of fashion. The comparison is not so outrageous as it may seem at first sight. For in language there are two kinds of sound change, that which is unconscious, universal at a given time and within a given area, and, on the other hand, that which belongs only to a particular class or clique, deviates consciously from the pronunciation of the majority, is therefore not universal, and exercises no permanent influence on the language. The second kind of sound change corresponds exactly to the laws of fashion; it is in fact one of them. Such sound changes are the pronunciation of the English ending -ing as -in', which was fashionable in the middle of the 19th century. This had, though probably without the knowledge of those who used it, an historical justification in the earlier forms from which most of the English words now ending in -ing are descended, and which survive in numerous local dialects. A similar conventional mispronunciation was the lisp affected by some would-be artistic persons at a somewhat later period. Belonging to an entirely different social stratum, and now equally obsolete, was the London pronunciation of the first half of the 19th century typified in Tony and Sam Weller's treatment of v and w in the Pickwick Papers. This, however, made a much nearer approach to being a genuine dialect peculiarity. It undoubtedly pervaded the pronunciation of the lower classes in London at one time; had it survived it might conceivably have spread over a wider and wider area until it embraced the whole population of England. A later change, that of the diphthong ai into ei (so that day, daily are pronounced dy, dyly), has spread from Essex and the East End of London over a large part of London and of the adjacent counties, and is still widening its range both geographically and socially. The history of these sound changes has not yet been investigated in detail with the thoroughness which it deserves.

From what has been said it will be obvious that a phonetic law is only an observed uniformity in the treatment of a sound or a combination of sounds within a linguistic area at a given time. In the definition the term linguistic area is a very variable quantity. Thus it is a phonetic law that a sound of the original Indo-European language, the precise pronunciation of which cannot be determined, but which was at any rate a palatal sound (k), appears in the Indo-European group (Sanskrit, Zend, Old Persian, with their descendants), in Armenian, in Balto-Slavonic and Albanian, in the form of a sibilant, while in Greek, the Italic dialects, Germanic and Celtic, it appears as a k-sound (see INDO-EUROPEAN LANGUAGES). Here the linguistic area is extremely wide, and it is clear that the difference between the two groups of languages must be dated back to a very early period. Again, it is a phonetic law of Greek that the original combination st- at the beginning of words is retained in Greek. How then are we to explain the existence side by side of σTÉYOS and réyos? The former apparently complies with the law, the latter does not. The former has by its side the verb oreyw, while réyos is supported only by the rare réyn. Yet the forms of the verb and substantive found in the Germanic languages leave no doubt that the forms without s- represent an extremely old form, for the English thatch could not have changed its original t- into th- if it had been preceded by s-, the law being as strict for English as for Greek that initial st- remains unchanged. On the other hand, a phonetic law may be limited to a very small area. Thus in the dialect of Eretria, and nowhere else within the area of the Ionic dialect of ancient Greek, do we find the change of the sound which appears elsewhere in Greek as -between vowels into -p-: σίτηριν for σίτησιν (acc. sing.), Tapaẞaivwpiv for Tapaẞaivcov (3rd pl. subjunctive). Why παραβαίνωριν παραβαίνωσιν this change should take place here and nowhere else we do not know, although it may be conjectured that the cause was a mixture with immigrants speaking a different dialect, a mixture which ancient tradition supported. Undoubtedly such mixtures are the chief conditions of phonetic change, the effect of which is universal. The manner in which the change takes place is that the basis of articulation, the method in which the sound is produced, becomes changed. Thus along the "Highland line in Scotland, where the English and Gaelic-speaking populations had their linguistic frontier for centuries, the wh- of English, the There is, then, a part of sound change which is a matter of Anglo-Saxon hw-, becomes universally f-, wha? becoming fa? fashion and which is conscious. This sound change appears white, fite, &c., f being the sound which it was most easy to substi frequently in the pronunciation of individuals who have migrated tute for the difficult hw-. The history of Spanish in the different from one part of a country to another. In many parts of communities of South America excellently illustrates this point. Scotland, for example, the prepositions with and of appear in After the discovery of America there was a large influx of dialect only in the forms wi' and o', which were originally the Spaniards into Chile, who ultimately, and chiefly by intermarunaccented forms. In the conscious attempts to pronounce riage, incorporated amongst them a considerable element from them as they appear in literary English, the educated Scotsman, amongst the native Araucanian Indians. The result has been

fore if in Modern English the plural of child had been childs. Bui in spite of the common tendency to make the plural of all nounstems in -s, child has gone in the opposite direction and has not only maintained its -r, but has added to it the -en of stems like oxen and eyen. In Wiclif we find a similar plural to calf, calveren, but here calves has long replaced in the literary language both the earlier forms.

66

that the language of Chile is Spanish, pronounced not with the genuine sounds of Spanish, but with the sounds of the Araucanian language substituted for them. Elsewhere in Spanish America the language of the conquerors remained comparatively pure, because the Spaniards were much fewer in number, and had therefore to maintain themselves as a caste apart. For the same reason Latin has split up into the numerous branches which we know as the Romance languages. The particular line of (b) Let us now take another instance from the English verb. development which, e.g. French followed as compared with In Old English the different persons of the preterite indicative Spanish or with the language of the Rhaetian Alps was condi- in the so-called strong (irregular) verbs were generally distintioned by the nature of the sounds in the language which pre-guished by different root-vowels; rīdan, “to ride," and bindan, ceded it in the same area, and which was spoken by the ancient Gauls who adopted Latin. The difficulty found in all of these cases is precisely of the same kind as that which an adult at the present day speaking one language finds in attempting to learn the pronunciation of another language. On the one hand, it is only with the greatest difficulty that muscles for many years accustomed to perform one set of movements can be forced into performing another set which are very similar but yet not identical; on the other hand, to an untrained ear the difference between the two sounds may remain unappreciated. The result is that the new language is pronounced with the sounds of the speaker's original language. If the new language is adopted by a whole people to whom it was originally foreign, the children naturally learn it from their parents with the sounds of the old language which has now become obsolete. Thus the basis of articulation is changed, and if, as was the case with Latin, this process be frequently repeated among peoples speaking languages with articulation widely differing one from another, it is clear | that a series of different dialects of the adopted language has been created. This kind of change is immediate and universal throughout the whole area where linguistic change has taken place.

to bind," for instance, form their preterities thus; ic răd, dū ride, hē rād, wē, gë, hïë ridon, and ic band, dû bunde, hẽ band, wē, gë, hie bundon. In modern English this difference in the rootvowels has been abandoned, and rode, bound now stand for all persons, rode being the modern phonetic equivalent of the 1st and 3rd sing. rãd, while bound represents the u- form of bindan. When one form or set of forms ousts other varying forms from the same paradigm, the change is described variously as material or logical analogy. Inasmuch as a similar process of levelling to that seen in rode has been carried through in all preterites of Modern English, regularity prevails even here, though a few traces of the old conflict are still visible in such poetic forms as sung for the preterite side by side with sang. But when we look to its results in the individual verbs we soon find that the choice amongst the different forms which might have served as starting-points has been entirely arbitrary. It is indeed impossible to say why the old singular form should have been chosen as a model in one case, as in rode, and the old plural form in another, as in bound. From these and numerous similar instances we must draw the conclusion that it is beyond our power to ascertain whence analogical changes start, and to what extent they may be carried through when once begun. Analogical change, on the other hand, does not affect the All we can do is to classify carefully the single cases that come pronunciation of a language as a whole in the way that phonetic under our observation, and in this way to investigate where change does, but is confined to the formation, inflexion, syntax such changes are especially apt to take place and what is their and meaning of single words or groups of words, and therefore is general direction. As to the latter points, it has been observed very apt to bear an entirely arbitrary and irregular character. before that levelling of existing differences is one of the chief A few instances will be sufficient to illustrate this and also to features in analogical change (as in the case of rode and bound). show how the apparently irregular phenomena of analogy may As to the former, it must be borne in mind that, before any anabe classified. (a) In Old English a certain number of substantives logical change can take place, some mental connexion must exist formed their plurals by mutation of the root vowels, as fōt, fēt, between the words or forms serving as models and those which or bōc, bēc. In Modern English this system of inflexion has been are remodelled after the types suggested to the minds of the preserved in some cases, as in foot, feet, and altered in others, as speakers through the former. Of such natural mental combinabook, books. Now, while foot, feet and book are the regular tions two classes deserve special notice: the mutual relationship modern phonetic equivalents of the old fōt, fet, bōc, the plural in which the different, say inflexional, forms of the same word books can in no way be phonetically traced back to the old bēc, stand to each other, and the more abstract analogies between the phonetical equivalent of which in Modern English would be the inflexional system of word-groups bearing a similar character, *beech. The only possible explanation of a form like books is as, for instance, the different declensions of nouns and pronouns, that the older bec was at some date given up and replaced by an or the different conjugations of verbs. The instance of rode, entirely new formation, shaped after the analogy of the numerous bound may serve to illustrate the former category, that of books words with a plural in -s without modification of the root-vowel. the latter. In the first case a levelling has taken place between Such changes, which are very numerous, exemplify the first kind the different forms of the root-vowels once exhibited in the of analogy, which is generally termed formal analogy. Other different preterite forms of ridan or bindan, which clearly examples are the almost entire disappearance from the language constitute a natural group or mental unity in consequence of of the forms in er and en, which were earlier used as plurals in their meaning. The form of rode as a plural has simply been English. That they were originally stem and not case suffixes taken from the old singular räd, the long a of which has become does not affect the point. In Middle English, as in Modern in Modern English ō, that of bound as a singular from the old English, oxen was spelt as a plural; oxen survives, but eyen, except plural bundon, the u- sound of which has in Modern English come in such dialect forms as the Scotch e'en, has been replaced by the to be pronounced as a diphthong. In the case of book, books for form in -s: eyes. Similarly in Middle English the suffix -er bōc, bēc, this explanation would fall short. Although we might existed in many words which had been originally of the neuter say that the vowel of the singular here was carried into the plural, gender. Thus the plural of child was childer, of calf was calver, yet this would not explain the plural -s. So it becomes evident traces of which, besides the survival in dialect of childer and of that the old declension of bōc, bēc was remodelled after the calver (become by the 16th century in northern Scotch car- declension of words like arm, arms, which had always formed pronounced as cahr-which is still in common use), are to be their plurals in -s. The changes indicated may generally be found in the place, and hence personal, names Childer-ley and shown by a proportion, the new analogical formation being the Calver-ley. The old plural of brother was brether, where the unknown quantity to be ascertained. Thus in the case cited suffix, however, contained an original -r, not -s changed into -r, above, arm: arms = book: x; and clearly the form to be as did childer and calver. In Old English, alongside the form for ascertained is books. Isolated words or forms which are no child making a plural childer, there had been a masculine form part of natural groups or systems, inflexional, formative or synmaking its plural in -s. It would not have been surprising there-tactical must be regarded as commonly safe from alterations

through analogy, and are therefore of especial value with regard | standard, it is only in very recent times that syntax has to establishing rules of purely phonetic development.

Historical

Syntax.

received methodical treatment from the comparative point of (c) In syntactical analogy the mental connexion between the view. It may indeed be said that almost the two series of constructions between which the change takes whole fabric of the comparative syntax of the place is generally still more conspicuous. The connexion may Indo-European languages as it exists to-day has be one of similar or of contrasted meaning. In Latin, adjectives been reared by one man-Professor Berthold Delbrück of Jena. of fullness, like other adjectives, no doubt originally were followed In a series of brilliant studies beginning with a pamphlet on the by the genitive case; participles, on the other hand, were followed Locative, Ablative, and Instrumental, published in 1867, and by the instrumental ablative. Thus Plautus in the Aulularia continued in his Syntactical Researches (Syntaktische Forsch813 and elsewhere could say aulam auri plenam, a pot full of ungen) in five volumes, comprising a treatment of the gold," or 802 aulam onustam auro, a pot laden with gold." | conjunctive and optative moods in Sanskrit and Greek (1871), From these the transition was easy to the construction aulam the theory of the Sanskrit tenses (1877), the order of words in onustam auri, as if in English one should say (as was possible early Sanskrit prose (Catapatha Brahmana; 1878), the foundain Earlier English), a pot laden of gold." In English, con- tions of Greek syntax (1879), and the syntax of the oldest Santrasted words often tend to assimilate their syntactical construc- skrit (Altindische Syntax), dealing exclusively with the literature tions. Thus, the adjectives like and similar are followed by of the Vedas and Brahmanas (1888), Professor Delbrück laid the preposition to (though in Modern English like need have no the foundations for his treatment of comparative syntax in preposition), and upon the analogy of such words, different and three volumes (1893, 1897, 1900), which has formed the averse, with which correct speakers and writers couple from, are completion of Brugmann's Grundriss der vergleichenden by no means rarely followed by to. Nor is it uncommon to Grammatik der indogermanischen Sprachen. The only work hear or to see differ with instead of differ from, upon the analogy by another hand (on a large department of the subject) of agree with. Curiously enough, Latin, from which differ is which deserves to be mentioned by the side of Delbrück's descended, is found to follow the same analogy even in good studies is the small treatise by Hübschmann on the theory writers. Thus Cicero (Academica Pr. ii. 143 ) combines dissidere of the cases (Zur Casuslehre, 1875). For the comparative with cum, as later does Seneca (Epistulae, 18. 1). neglect of this field of investigation there are several reasons. The earlier philologists had so much to do in determining the languages which should be included within the Indo-European group, and in organizing the field of research as a whole, that it is not to be wondered at if they were unable to devote much attention to syntax. In the 'seventies, when attention began to be more directed towards comparative syntax, the remarkable discoveries made by Verner with regard to accentuation, and by Brugmann, Collitz and others with regard to the phonology of the Indo-European languages, again distracted attention from the subject. Moreover, the research in itself is infinitely more difficult than that into sounds and forms; for the latter may be carried on by the help of grammars and dictionaries with a comparatively small knowledge of the literature of any individual language, while on the other hand the study of syntax is impossible without a thorough and intimate knowledge of the literature and modes of expression in each separate language. It is not, therefore, matter for wonder that Delbrück has confined himself in the investigation of syntax to a part only of the languages whose sounds and forms are discussed by Brugmann in the earlier volumes of the Grundriss. To cover the whole ground is beyond the powers of a single man, and there is a great lack of preliminary studies on the syntax of many of the languages.

(d) In the development of analogy in meaning, similarity of sound is often the effective cause. Thus impertinent is properly irrelevant, not to the point, and is still so used in legal language; its more common signification of " saucy arises from its accidental resemblance in sound to pert, a word which curiously enough has reversed its meaning, being now used in the sense of mal-apert, while the Old French apert, aspert (a confusion of Lat. apertus, "open," with expertus, skilled"), meant both 66 open and skilful." Thus from very early times the verbs fly and flee have been confused, though they are of entirely different origins. When Middle English began to lose its verb endings in -en, it was very easy for the verb leren," teach," and lernen, "learn," to be confused. Hence frequently in Elizabethan English learn stands side by side with teache in the same signification. Cf. Tottell's Miscellany, p. 129 (Arber):

"I would not have it thought hereby

The dolphin swimme I meane to teache:
Nor yet to learn the Fawcon flie:

I rowe not so farre past my reache."

It is true that the distinction between phonetic and analogical change has always been acknowledged in comparative philology. At the same time it cannot be denied that analogical changes were for a long time treated with a certain disdain and contempt, as deviations from the only course of development then allowed to be truly "organic" and natural, namely, that of gradual phonetic change (hence the epithet "false" so constantly attached to analogy in former times). Amongst those who have recently contributed most towards a more correct evaluation of analogy as a motive power in language, Professor Whitney must be mentioned in the first place. In Germany Professor Scherer (Zur Geschichte der deutschen Sprache, 1868) was the first to apply analogy as a principle of explanation on a larger scale, but in a wilful and unsystematic way. Hence he failed to produce an immediate and lasting impression, and the merit of having introduced into the practice of modern comparative philology a strictly systematic consideration of both phonetic and analogic changes as co-ordinate factors in the development of language rests with Professor Leskien of Leipzig, and a number of younger scholars who had more or less experienced his personal influence. Amongst these Brugmann, Osthoff and Paul rank foremost as the most vigorous and successful defenders of the new method, the correctness of which has since been practically acknowledged by most of the leading philologists of all shades of opinion.

The New

School.

While the syntax of individual languages was one of the first features which attracted the grammarians' attention, at any rate in so far as particular authors differed from a given

One of the most difficult problems connected with syntax, but primarily, as it appears, a question of morphology, is the origin of grammatical gender. It cannot be said to be an advantage to the languages which possess it, while languages which, like English, have dropped it except for an occasional metaphor, suffer no loss. Nor is the problem confined to the history of gender in the substantive. Even more perplexing is the introduction of gender into the adjective. The pronouns of the first and second persons, which are certainly very old, show no trace of gender; the pronouns of the third person, which are more of the nature of deictic adjectives, generally possess it. To the question how grammatical gender arose in the substantive, the answer was till comparatively recently supposed to be that primitive man was given greatly to personification, endowing inanimate things with life and attributing to them influences benign or the reverse upon his own existence. The answer is not quite sufficient, for though this tendency to personification, which philologists have perhaps unduly decried or altogether denied, might account for life being attributed to inanimate objects, it hardly explains why some should be treated as mascu line and others as feminine. Nor is it true, as has also been suggested, that in the case of the lower animals the generic name for the larger and stronger animals is masculine and that for the smaller or weaker feminine. In both Greek and Latin the wolf

sex.

is masculine and the fox feminine, but the lamb or the chicken | the number of the original cases had once been determined, the which the fox robs from the fold or the henroost is rarely feminine, sifting of the pro-ethnic usages attaching to each case was generally masculine. Nor does this explanation account for tolerably easy, for besides Sanskrit and (to a less extent) Latin, the mouse in those languages being of the masculine gender, Lithuanian and Slavonic have kept the pro-ethnic case system while the ferret or cat which caught them is feminine (yaλn, almost complete. The ideas also which had to be expressed feles). An explanation which completes the theory of personifi- by the cases were on the whole of a very concrete character, so cation, if it does not altogether drive it from the field, has been that here the problem was much simplified. On the other hand, put forward by Brugmann.1 In its briefest form this explana- the ideas expressed by the forms of the verb are of a much more tion is that gender was attached to certain suffixes because they subtle nature, while the verb system in all languages except chanced to occur frequently in words which markedly implied Greek and Sanskrit has broken down earlier and more completely In the Indo-European languages the commonest suffix than the noun. It is clear that the verb of the original Indoindicating feminine gender is a. According to this theory it European language possessed two voices, and forms correspondhad originally nothing to do with gender, but as some early words ing to what we call the Indicative, Subjunctive, and Optative for woman or wife ended with this sound it came to be identified moods, and to the Present, Imperfect, Future, Aorist, and Perfect with feminine gender. Similarly the ending os in o-stems tenses. The imperative mood seems primitively to have been occurred often in names connected with males and so became confined to the second person singular, just as the vocative, identified with the masculine gender. But many stems indicate which, like the imperative is a stem form without suffix, was either gender indifferently, and even the very old sex words confined to the singular. The infinitive, as is well known, is father and mother have the same ending. But when masculine in all languages of this system not originally a verbal but a and feminine endings have been attached to certain suffixes in substantival form. The pluperfect, where it has developed, this way, how comes it that in one series of stems the neuter should seems to be a mixed form arising from the application of aorist be marked not by an absence of all suffix but by a separate endings to a perfect stem. Thus far the history of the verb suffix in -m? These are the o-stems, other forms of which have system is tolerably clear. But when we attempt to define the been markedly identified with the masculine gender. As this original meaning of the moods and of the tenses we pass into a characteristic, like the others mentioned, goes back apparently region where, in spite of assiduous investigation in many quarters to a time before the separation of the Indo-European languages, during recent years, the scanty amount of light thrown on the explanation can hardly pass beyond speculation. It is, however, problem has only served to make the darkness visible. As to be noted that the neuter form of the nominative is phonetic- regards the tenses, at least, it has been shown that without doubt ally identical with the accusative form of the masculine, and there is no difference in formation between present, future and it has been ingeniously argued that such forms were used aorist stems, while the earliest meaning of the perfect was that originally in the accusative, such neuters not forming the subject of a special kind of present expressing either repeated or intensive to a verb. To the same writer the most plausible explanation action or a state. It has also been proved that the original of the presence of gender in the adjective is due, viz. that gender meaning of the aorist is not past in time, and that in fact the began with the deictic pronoun *so "that man, *sā that only element whereby these languages could express remoteness woman," and that hence it passed to the adjective with which in time was the augment. The augment seems to have been the pronoun was so frequently accompanied. If this explanation originally a pronominal deictic particle. Thus, as there was no be right, analogy has brought into the Indo-European languages original pluperfect, as neither perfect nor aorist originally the useless multiplication of gender marks in such sentences as referred to past time, and as the future, except in Lithuanian the Latin hae pulcrae feminae caesae sunt, where the feminine (with slight traces in Slavonic) and the Indo-Iranian group, gender is indicated no less than four times without any obvious cannot be clearly distinguished from the aorist, the system as gain over the English These fair women were slain, where a method of expressing time absolutely breaks down. The grammatical gender is no longer obviously indicated at all. tenses in fact did not originally express the times when the action took place, but the type of action which took place. Thus the present system in the main expressed continued or durative action, the aorist only the fact that the action had taken place. The action indicated by the aorist might have been of considerable duration, or it might have been begun and ended in a moment; its characteristics in this respect are not in any way indicated by the aorist form, which intimates only that the action is viewed as a completed whole and not as a continuous process. The present system, however, is built up in a great variety of ways (thirty-two according to Brugmann's enumeration). It is a priori unlikely that such a multiplicity of formations had not originally some reason for its existence, and Delbrück thinks that he has discovered a difference in syntactical value between various forms. The reduplicated present forms of the type seen in Sanskrit jigāti, Greek didwu, &c., he regards as expressing originally an action which consisted of repeated acts of the same nature (iterative), though this iterative meaning frequently passed into an intensive meaning. Presents of the type seen in Sanskrit tr' ṣyati, is thirsty," and Greek xalpw, am glad " (for*xapw), where the (y) of the suffix has modified the first syllable and disappeared, he regards as cursive―i.e. they express continuous action without reference to its beginning or end. Verbs which have regard to the beginning or end of the action he calls terminative, and finds them represented (a) in verbs with -n- suffixes, Sanskrit ṛnoti, öpvvoi, “sets in motion," ayvμ, “break to pieces "; (b) in verbs with the suffix -sko-, Sanskrit gáchati, goes (to a definite destination), Greek Baokw, &c. The roots he classifies as momentary (punktuell) or non-momentary, according as they do or do not express an action which is begun and ended at once.

Closely related to this question is that of the history of the neuter plural, which was first fully worked out by Professor Johannes Schmidt of Berlin. The curious construction, most common in ancient Greek, whereby a neuter plural is combined with a singular verb, is now demonstrated to be an archaic survival from the time when the neuter plural was a collective singular. Thus a word like the Latin iugum was a single yoke, the plural iuga, however, which was earlier iugā, was a collection of yokes, with the same final a as is found generally in feminine substantives. The declension ought therefore to have been originally: nominative iugā, genitive iugās, &c., like mensa, &c., of the first declension. But as iuguum was used in the neuter singular for both nominative and accusative, iuga when it was felt as the corresponding plural was used for the accusative as well as the nominative, while the other cases of the plural were taken over from the masculine o-stems, with which the singular neuter in -o-m was so closely connected. That collective words should be used for the plural is not surprising; the English youth, first an abstract, next a collective, and finally an individual, is a case in point.

For the early history of the syntax of the verb Greek and Sanskrit are important above all other languages, because in them the original forms and the original usages are better preserved than they are elsewhere. And it is in the verb that the great difficulties of comparative syntax present themselves. The noun system is so well preserved in several languages that, when 1 Techmer's Internationale Zeitschrift für Sprachwissenschaft, iv. 2 B. I. Wheeler, Journal of Germanic Philology, ii. 528 sqq. Pluralbildungen der indogermanischen Neutra (1889).

100.