Now let us look to French. On June 3, 1679, the French Academy decreed that the participles present should no longer be declined.23

What was the meaning of this decree? Simply what may now be found in every French grammar, namely, that commençant, finissant, are indeclinable when they have the meaning of the participle present, active or neuter; but that they take the terminations of the masculine and feminine, in the singular and plural, if they are used as adjectives.24 But what is the reason of this rule? Simply this, that chantant, if used as a participle, is not the Latin participle present cantans, but the so-called gerund; that is to say, the oblique case of a verbal noun, the Latin

If verbal adjectives in ing existed in Anglo-Saxon, Bopp's theory would certainly remove all difficulties. We should then have to admit two forms, substantives in ung and adjectives in ing, converging into the modern English participle in ing. But no such adjectives exist in Anglo-Saxon, and I do not see how to explain their sudden appearance except by adopting the theory of the late Mr. Garnett.

By means of such phrases as 'he was flogging,' instead of 'he was a flogging,' forms in ing without a preposition entered into a new grammatical category. They were felt as participles, and were allowed to enjoy all the privileges of the forms ending in ende, which they gradually supplanted. The same thing has happened in French. To a Frenchman aimant is much the same as what amans was to a Roman ; and it is only by analysing such constructions as 'une femme aimant ses amants' that we discover aimant as the representative of the Latin Gerund amando, and not of the Latin participle amans.

One more word about Henneberg! In the dialect of Henneberg the substantive termination ung is pronounced ing. We find Übing, Verwässeling, Verwonnering, instead of Übung, Verwechselung, Verwunderung. This is the only light which the Thuringian dialect throws on the change of Anglo-Saxon ung into English ing, though, as Grimm remarked, the suffix ing extends far beyond Thuringia.

"Cf. Egger, Notions élémentaires de Grammaire comparée: Paris, 1856, p. 197. La règle est faite. On ne déclinera plus les participes présents.-B. Jullien, Cours supérieur, i. p. 186.

24 Diez, Vergleichende Grammatik der Romanischen Sprachen, ii. p.114.

cantando corresponding to the English a-singing, while the real Latin participle present, cantans, is used in the Romance languages as an adjective, and takes the feminine termination-for instance, 'une femme souffrante,' &c.

Here, then, we see again that in analytical languages the idea conveyed by the participle present can be expressed by the oblique case of a verbal noun.

Let us now proceed to a more distant, yet to a cognate language, the Bengali. We there 25 find that the so-called infinitive is formed by te, which te is at the same time the termination of the locative singular. Hence the present, Kariteki, I am doing, and the imperfect, Karitekilâm, I was doing, are mere compounds of âki, I am, âkilâm, I was, with what may be called a participle present, but what is in reality a verbal noun in the locative. Kariteki, I do, means 'I am on doing,' or 'I am a-doing."

Now the question arises, Does this perfectly intelligible method of forming the participle from the oblique case of a verbal noun, and of forming the present indicative by compounding this verbal noun with the auxiliary verb to be,' supply us with a test that may be safely applied to the analysis of languages which decidedly belong to a different family of speech? Let us take the Bask, which is certainly neither Aryan nor Semitic, and which has thrown out a greater abundance of verbal forms than almost any known language. Here the present is formed

26

23 M. M.'s Essay on the Relation of the Bengali to the Aryan and Aboriginal Languages of India. Report of the British Association for the Advancement of Science, 1847, pp. 344-45. Cf. Garnett, l. c. p. 29.

*See Inchauspe's Le Verbe basque, published by Prince Louis-Lucien Bonaparte. Bayonne, 1858.

by what is called a participle, followed by an auxiliary verb. This participle, however, is formed by the suffix an, and the same suffix is used to form the locative case of nouns. For instance, mendia, the mountain; mendiaz, from the mountain; mendian, in the mountain; mendico, for the sake of the mountain. In like manner, etchean, in the house: ohean, in the bed. If, then, we examine the verb,

we see again in erorten a locative, or, as it is called, a positive case of the verbal substantive erorta, the root of which would be eror, falling; 27 so that the indicative present of the Bask verb does not mean either I fall, or I am falling, but was intended originally for 'I (am) in the act of falling,' or, to return to the point from whence we started, I am a-falling. The a in a-falling stands for an original on. Thus asleep is on sleep, aright is onrihte, away is onweg, aback is onbæc, again is ongén (Ger. entgegen), among is ongemang, &c.

This must suffice as an illustration of the principles on which the Science of Language rests, viz. that what is real in modern formations must be admitted as probable, or at least as possible, in more ancient formations, and that what has been found to be true on a small scale may be true on a larger scale.

But the same illustration may also serve as a warning. There is much in the science of language to tempt us to overstep the legitimate limits of inductive reasoning. We may infer from the known to the

"Cf. Dissertation critique et apologétique sur la Langue basque (par l'abbé Darrigol). Bayonne, p. 102.

GENERALISATION.

unknown in language tentatively, but not positively. It does not follow, even within so small a sphere as the Aryan family of speech, that what is possible in French is possible in Latin, that what explains Bengali will explain Sanskrit; nay, the similarity be tween some of the Aryan languages and the Bask in the formation of their participles should be considered as an entirely exceptional case. however, after establishing the principle that the Mr. Garnett, participle present may be expressed by the locative of a verbal noun, endeavours in his excellent paper to show that the original Indo-European participle, the Latin amans, the Greek typton, the Sanskrit bodhat, were formed on the same principle:-that they are all inflected cases of a verbal noun. In this, I believe, he has failed, 28 as many have failed before and after him, by imagining that what has been found to be true in one portion of the vast kingdom of speech must be equally true in all. This is not so, and cannot be so. Language, though its growth is governed by intelligible principles throughout, was not so uniform in its progress as to repeat exactly the same phenomena at every stage of its life. As the geologist looks for different characteristics when he has to deal with London clay, with Oxford clay, or with old red sandstone, the student of language, too, must be prepared for different formations, even though he confines himself to one stage in the history of language, the inflectional. And if he steps beyond this, the most modern stage,

"He takes the Sanskrit dravat as a possible ablative, likewise sas-at, and tan-vat (sic). It would be impossible to form ablatives in åt (as) from verbal bases raised by the vikaranas of the special tenses, nor would the ablative be so appropriate a case as the locative, for taking the place of a verbal adjective.

DIFFERENT TREATMENT OF DIFFERENT LANGUAGES. 25

then to apply indiscriminately to the lower stages of human speech, to the agglutinative and radical, the same tests which have proved successful in the inflectional, would be like ignoring the difference between aqueous, igneous, and metamorphic rocks. There are scholars who, as it would seem, are incapable of appreciating more than one kind of evidence. No doubt the evidence on which the relationship of French and Italian, of Greek and Latin, of Lithuanian and Sanskrit, of Hebrew and Arabic, has been established, is the most satisfactory; but such evidence is possible only in inflectional languages that have passed their period of growth, and have entered into the stage of phonetic decay. To call for the same evidence in support of the homogeneousness of the Turanian languages, is to call for evidence which, from the nature of the case, it is impossible to supply. As well might the geologist look for fossils in granite ! The Turanian languages allow of no grammatical petrifactions like those on which the relationship of the Aryan and Semitic families is chiefly founded. If they did, they would cease to be what they are; they would be inflectional, not agglutinative.

If languages were all of one and the same texture, they might be unravelled, no doubt, with the same tools. But as they are not-and this is admitted by all-it is surely mere waste of valuable time to test the relationship of Tungusic, Mongolic, Turkic, Samoyedic, and Finnic dialects by the same criteria on which the common descent of Greek and Latin is established; or to try to discover Sanskrit in the Malay dialects, or Greek in the idioms of the Caucasian mountaineers. The whole crust of the earth is not made of lias, swarming with Ammonites and