the assumption that the Pāli, and so the Prakrit, are derived from the Sanskrit, deserves the preference over the converse view to which D'Alwis appears here and there to be not indisposed, viz., that the Pāli, as being the most ancient Prākṛit which has been handed down to us, stands higher in point of originality and independence than the Sanskrit. For it is clear that the Sanskrit, both in its phonetic system and flexions, stands much closer to the common mother of it and the Pāli than the latter does, 108 and has consequently a far superior right than it to be regarded as the representative of that parent language. A perplexing circumstance connected with this question, and one which leads to many sorts of mistakes, is that we have unfortunately no proper name for that stage of the language which lies at the foundation of both the 'sister dialects,' the Pāli (and Prākṛit) and the Sanskrit, i.e. for the Vedic vulgar speech; for the names bhashā and vyāvahārikī are not sufficiently pregnant; and one is consequently thing-that which is pre-eminent-that which is the natural or quiescent state of anything-not made.' Hence it is clear that the correct and primary sense of the word Prākṛita,—indeed that which was originally assigned to it, despite the so-called 'common acceptation,'-was 'original,' 'root,' 'natural.' By the Prakrit was therefore at first meant the original Indian language, as distinguished from the apabhransa, 'the ungrammatical,' and the Sanskrit, signifying [from sam altogether', or ‘together,' and kṛita 'done' =‘altogether,' or 'completely made, done, or formed'] that which has been composed or formed by art, adorned, embellished, purified, highly cultivated or polished,' and regularly inflected as a language." See in opposition to this view of the sense of the word prākṛita, Hemachandra's interpretation of it given above in p. 44, and Lassen's explanation, quoted in p. 51. In a review of Prof. Weber's Ind. Literaturgeschichte, in the Journal of the German Oriental Society for 1853, p. 605, Prof. Roth thus expresses himself on this question: "Prākṛit (according to the derivation which seems to me the correct one) signifies that which has its foundation in another thing, 'the derived,' or the 'to be derived.' expression is one formed by grammarians, and has a grammatical sense. grammarians say e.g. samhitā pada-prākṛitis; the Sanhită texts have for their foundation the words, i.e. that form of speaking and writing the texts in which the end and beginning of the words which follow one another in a sentence are brought into harmony with the general phonetic laws of Sanskrit has for its foundation the single words conceived in their original form. The Sanhità text is thus prākṛitā in relation to the word-text, the pada-pāṭha: it is a derived text made for a scientific purpose. I would understand the word prākṛita in the same sense, when it is applied to the dialects." In Böhtlingk and Roth's Sanskrit Lexicon the sense "customary," "common," is assigned to the word prākṛita when generally used, while of the dialect so called it is said: "The 'common' speech is that spoken by the people, which the grammarians derive from Sanskrit."

The

The

108 This, however, can afford us no reason to deny that the Pāli has actually preserved older forms than the Sanskrit. [Note of Prof. Weber.]

at a loss how to designate it. Benfey's excellent remarks at p. 245 of his article Indien (which unfortunately has not yet been re-written), regarding the dying out of the Sanskrit as a vernacular language in the sixth century B.C. labour under this disadvantage, that they apply the name Sanskrit for a period for which it is in no way applicable.' The views of Burnouf and Lassen on the relation of the Pali to Sanskrit are thus stated in their Essai sur le Pali, pp. 138, ff:

"The Pali is derived from the Sanskrit, according to certain rules, for the most part euphonic, which do not allow the derivative language to admit certain sounds and combinations of consonants, common in the parent tongue. These modifications apply equally to the substantive portions of the words and to their terminations and inflections. It hence results that there is no grammatical form to be found in Pali of which the origin may not be discovered in Sanskrit; and that there is no occasion to call in the influence of any foreign idiom to explain the modifications to which the Pali has subjected the Sanskrit.

"When the Pali, as a derivative from Sanskrit, is compared with other dialects having the same origin, it is found to approach far more closely than any of those others to that common source. It stands, so to speak, on the first step of the ladder of departure from Sanskrit, and is the first of the series of dialects which break up that rich and fertile language. But it appears that the Pali, which contained in itself the germs of alteration already greatly developed, was arrested in its progress all at once, and fixed in the condition in which we now find it, i.e., in a state of almost immediate connexion with the language from which it proceeded. In fact the greater part of the words which form the basis of the one, are found without modification in the other; those which are modified can all be traced to their Sanskrit root; in short, no words of foreign origin are to be found in Pali."

"We shall not enter into new details regarding the manner in which the Pali has been derived from the Sanskrit. The laws which have guided the formation of that language are the same which we find at work in other idioms in different ages and countries; these laws are general, because they are necessary. Whether we compare the lan

guages which are derived from Latin with the Latin itself, or the later Teutonic dialects with the ancient languages of the same stock, or the modern with the ancient Greek, or the numerous popular dialects of India with the Sanskrit, we shall see the same principles developed, the same laws applied. The organic inflections of the parent languages are seen to exist in part, but in a state of evident alteration. More commonly they will be found to have disappeared, and to have been replaced, the case-terminations by particles, and the tenses by auxiliary verbs. The processes vary in different languages, but the principle is the same; it is always analytic, whether the reason of this be that a synthetic language happens all at once to become the speech of barbarians who do not understand its structure, and therefore suppress its inflections, and replace them by other signs; or whether it be that when abandoned to its natural course, and as a necessity of its cultivation, it tends to decompose and to subdivide the representative signs of ideas and relations, just as it unceasingly decomposes and subdivides the ideas and the relations themselves. Pali appears to have undergone this last sort of alteration; it is Sanskrit, not such as it would be spoken by a strange population, to whom it would be new; but pure Sanskrit, becoming altered and modified in proportion as it becomes popular. In this manner it still preserves its declension, instead of replacing it by particles, as the modern dialects of India do. One form only, the ablative in to might pass for the commencement of the analytic declension; but it is already found in the parent language. A great number of Pali forms might be cited to prove that the modifications, which it has made in the Sanskrit, are of the same kind as those which the Italian, among other tongues, has made in the Latin. Thus the assimilation of consonants, which in Italian makes letto from lectus, and scritto for scriptus, is one of the principles of Pali."

The

The Pali, in the precise form in which we find it in the Ceylonese books, could scarcely have been a vernacular language. At least, it exhibits a variety of refinements which could hardly have been employed in common speech; but seem likely to have been confined to the language of composition, or introduced after the Pali had ceased to be the spoken tongue of the followers of Buddha, and had become consecrated to the service of religion and literature: just as the gram

mar of the Sanskrit itself became regulated by more fixed and rigid rules, after it had been removed from the deteriorating influences of vernacular use. Such a peculiarity is the use of interpolated, or the retention of otherwise disused, consonants to obviate the inharmonious sounds which would arise from the collision of vowels. No less than nine letters, y, v, m, d, n, t, r, 1, and g, are employed for this purpose, as is shown in the following examples, viz.:

This peculiarity of attention to euphony is common to the Pāli with the Sanskrit; and though the means they use are for the most part

109 Clough's Pāli Grammar, p. 11. On this subject I translate the following remarks made by Dr. Kuhn in a review of the first edition of this volume, in his Beiträge zur Vergleichenden Sprachforschung u.s.w. vol. iii. p. 241, f.; "As regards the interpolation of euphonic letters treated of in p. 82, I cannot entirely agree with the author when he claims them for the written language alone: the greater part of them show that they are by no means what are called interpolations, but the old auslaut" (i.e. concluding consonant) "which the preceding words had in an earlier stage of the language. I have briefly treated of them in the first volume of these Beiträge, p. 126, and here only repeat that I now regard only the y and the v in na-y-imassa, and ti-v-angikam as real euphonic interpolations; and that I look upon the latter as having proceeded from y." I subjoin a translation of the remarks referred to by the writer as having been previously made by him in the 1st vol. of the Beiträge, p. 126:"A comparison with the Sanskrit shows that only a few of those apparently interpolated consonants are due to an actual interpolation, as the others are remains of an earlier condition of the language. The Pali has almost entirely rejected the final mutes, and the few cases in which such are found are to be regarded as exceptions. When for instance etad eva is found in place of the ordinary etam eva, this is an archaism which Lassen rightly explains by the close juxtaposition of the following eva to the preceding etad. In the same way we are to explain tasmāt iha from the Sanskrit tasmād iha, sabbhir eva from sadbhir eva, chhalabhiññā from shalabhijñās, which as a technical designation preserved the old form (see Turnour, Mahavanso, p. 31, 1, and elsewhere), puthageva from prithag eva, prageva from prāg Of the remaining instances nayimassa, tivangikam, and itonayati are indeed to be regarded as cases of consonantal interpolation, whilst lahum essati and attadattham may still remain doubtful. Clough further states, in p. 14, that m is sometimes introduced as an augment before both vowels and consonants; that thus chakkhum anichcham (Sanskrit chakshur anityam) stands for chakkhu anichcham, avamsiro for ava siro; but in the first case, as in that of lahum essati, perhaps another

eva.

different, yet in neither case could the refinements employed in writing have been practised in the language of ordinary life. The Pāli has other characteristics (borrowed from the Sanskrit) which could scarcely have been very common in the vernacular dialects of Northern India, supposed to have been contemporary with it; such as the use of desiderative, and nominal verbs; like jighachchhati, he wishes to eat; pabbatāyati, he resembles a mountain; puttiyati, he treats like a son.110 Fausböll observes in his introduction to the Dhammapada (p. vi.) that the antiquity of that work is proved by the character of its language, which approaches closely to the Sanskrit, even in some of its oldest forms, and differs widely from the diction of the prose Sūtras, explanation is possible, whilst in the second case, the assumption of an interpolation appears decidedly wrong, as the final mute of the Sanskrit was probably nasalized. Nevertheless, Clough's rule appears to be correct, since at least Turnour's text shows some other examples of this interpolation. Thus in p. 50, line 14, ewan te-m attano nāmañ katwā janapadañ bahuñ, unless perhaps te-m is here mutilated (verstümmelt) from the Sanskrit te ime; and in p. 52, line 4, yatra-m-ichchasi tam aññatra yakkhehi wijite mama, where, however, certainly the metre declares itself (spricht) not only against the interpolation of m, but also in favour of the elision of the final a of yatra." It is to be observed that the same interpolation (if I am right in so calling it) of more than one letter (as in yatha-r-iva for yathā-iva), is to be found in the language of the Gāthās in the Lalita vistara, which will be treated of further op. This shows that the process did not begin in Ceylon.

The following are instances collected from the Lalita vistara of the euphonic insertion of consonants between vowels which may be compared with the cases of a similar character which have been adduced in the text as occurring in Pāli.

110 Some desiderative verbs and nouns must, however, have been in ordinary use in the Prakrits; as we find in the modern vernaculars some words which have their origin in desideratives. Thus the Hindi bhukh, hunger, must come from bubhukkhā, a Prakrit corruption of bubhukshā. The Hindi piyās, thirst, too, is probably derived from pipāsā, though it may also have been compounded of pī +āsā, a desire to drink.