any class other than the Bhû, it still keeps the personal endings of Sanskrit for that conjugation; thus from √ “go,” we have

Pa. S. I. यामि, 2. यासि, 3 याति P. 1 याम, 2. याथ, 3. यांति.

which differs from Sanskrit only in omitting the visarga in P. 1.

The imperative follows the type of the present, and may be thus compared with Sanskrit Parasmaipada,

Skr. S. 1. पचानि, 2. पच, 3. पचतु; P. 1. पचाम, 2. पचत, 3. पचंतु. Pa. S. 1. पचामि, 2. पचहि, 3. पचतु; P. 1. पंचाम, 2. पचथ, 3. पचंति. and with the Âtmanepada, thus

Skr. S. 1. पचै, 2. पचस्व, 3. पचतां; P. 1. पचाम है, 2. पचध्वं 3. पचंतां. Pa. S. 1. पचे, 2. पचस्सु, 3. पचतं; P. 1. पचामसे, 2. पचहो, 3. पचंतं. Here the S. 1 Parasmai seems to have arisen from some confusion with the present, as also P. 2. Noteworthy is S. 2, with its ending f, which, though only found in classical Sanskrit in the second, third, seventh, and ninth conjugations, has crept into all in Pali, and has continued on into the medieval period, thus Chand

तिन सु गल्हू अच्छी कहहि ॥

Say thou a good word of them."-Pr. R. i. 9.

where कहहि = Skr. कथय (हि). In Vedic Skr. हि appears in all the conjugations. Of the Âtmane forms P. 1 seems to be derived from an older form, masai. P. 2 should perhaps be read hvo, not who, in which case it is a regular resultant from Sanskrit dhv.

The potential is the Sanskrit optative (lin), thus—

PARASMAI.

Skr. S. 1. पचेयं, 2. पचेस्, 3. पचेत्; P. 1. पचेम, 2. पचेत, 3. पचेयुस्. Pa. S. 1. पचेय्यामि, 2. य्यासि, 3. व्य; P. 1. °य्याम, 2. ° व्याथ, 3. ०य्यं.

ÂTMANE.

Sk. S. 1. पचेय, 2. पचेथास्, 3. पचेत ; P. 1. पचेमहि, 2. पचेष्वं, 3. पचेरन् . Pa. S. 1. पचेय्यं, 2. पचेथो, 3. पचेथ; P. 1. पचेय्या म्हे, 2. ° व्यहो, 3. पचेरं .

In this tense the point specially to be noticed is the tendency to simplify not only the root-syllable, but the range of terminations also. Having got the syllables eyya as the type of the tense, Pali seeks to avoid all further distinctions, and to use as much as possible the personal endings of the present tense. It sometimes conjugates the potential according to the types of other classes, and in this respect follows the lead of the present less faithfully in this tense than in the imperative. Thus, though in the present and imperative of kar, it follows the Sanskrit, and has karoti, karotu, yet in the potential it treats kar as if it belonged to the Bhû class, and has kareyyâmi as though from a Sanskrit kareyam instead of the actual kuryâm. There are other peculiarities about this tense which are not here noticed, as having no bearing upon the subject of the modern languages.

The imperfect has been, to some extent, mixed up with the aorist (lun), and both, together with the perfect, lead us into considerations which are of interest only for Pali itself, not having survived or had any influence on modern developments. They may therefore be passed over as immaterial to our present inquiry.

The future, on the contrary, offers many interesting peculiarities, especially, as will be seen hereafter, in reference to Gujarati and some of the rustic dialects of Hindi. The future is a different tense in the modern languages, and every scrap of information which can help to elucidate it deserves special notice. It runs thus in Pali ("")

Skr. S. 1. IfAQIFA, 2. °ufa, 3. °ufa; P. 1. •QTHH, 2. qy, 3. vifa. Pa. S. 1. गमिस्सामि, 2. स्ससि, 3. स्सति ; P. 1. स्साम, 2. स्सथ, 3. स्संति.

Here the only noteworthy feature is the change of into. The Âtmanepada follows the same rule throughout. Although the tendency to keep that form of the root which exists in the present leads to divergences from the Sanskrit future type, yet instances occur in which the Sanskrit type is preserved. These occur in reference to that very troublesome feature in the Sanskrit verb, the intermediate, which is sometimes inserted between the root and the termination, and sometimes not. When it is not inserted, the euphonic laws of Sanskrit require that the final consonant of the root be changed to enable it to combine with the initial consonant of the termination. Thus "cook," when it has to take the future termination ष्यति, becomes पक् and पक + ष्यति = पक्ष्यति. Here Pali sticks to the form q, because it is used in the present and makes its future fafa as though there had been (as there probably was in colloquial usage) a Sanskrit future ufafa with the intermediate inserted.

In a certain number of verbs, however, it has two forms, one as above retaining the root-form of the present, and the other a phonetic equivalent of the Sanskrit. Kuhn1 gives the following examples, to which I add the Sanskrit for comparison.

Skr. √ लभ् “get,” future लप्स्यते. Pali लच्छति but also लभिस्तति. "speak," " वक्ष्यते.

The consonantal changes are in accordance with the treatment of the nexus as explained in Vol. I. p. 304. The striving after uniformity is seen, however, in the retention of the alternative forms having the same type as the present, and it is, moreover, worth observing that the forms which reproduce the type of the Sanskrit without the intermediate seem by degrees to have been misunderstood. The illiterate masses, and even those better instructed, seem to have missed the issati which so generally indicated to their minds the future tense, and regarded those forms which had not this familiar sound as present tenses. So they made double futures by adding the issa to them. Thus from "to see," future fa, Pali

made a form dakkhati, but the people by degrees took this for a present, and made what to them seemed a more correct future dakkhissati. I mention this here as I shall have occasion hereafter to discuss the much-debated question of the origin of the familiar modern stem dekh "see" (see § 17). Another instance is

Skr. √ शक् “be able,” future शक्ष्यति. Pa. सक्विति, whence vulgo afgufa.

In one case Pali has a future which points back to a Vedic form:

Skr. "weep."

Vedic future रोत्स्यति. Pa. रोच्छति.
Classic ditto रोदिष्यति.,, रोदिस्सति.

Occasionally the स्स is softened to ह as in काहति, काहिति from aftwfa, Skr. affa. This is noteworthy with reference to Bhojpuri and the eastern Hindi dialects generally.

§ 5. It used to be held that Pali was a descendant of the Mâgadhi dialect of Prakrit, but this opinion is now, I believe, exploded. Though the question is not yet set at rest, it would seem to have been fairly established that Mahendra was a

native of Ujjayin, and that the language which he carried to Ceylon was the ordinary vernacular of his own province.1 This dialect was not very different from that of Magadha, and Mahendra may have slightly altered the Mâgadhi sayings of the great master, by his Ujjayini pronunciation, while retaining the name Mâgadhi out of deference to the sacred associations which clustered round the birthplace of Buddha.

Be this as it may, the nearest Indian dialect to Pali seems undoubtedly to be the Prakrit of the Bhagavatî, a sacred book of the semi-Buddhist sect of Jainas. If Hemachandra, himself a Jain and author of several works on Prakrit, were available for reference, our task would be easier; as yet, however, none of Hemachandra's writings have been printed or edited. Weber's articles on the Bhagavatî are at present our only source of information.2

In the Jaina Prakrit the ten conjugations of the Sanskrit verb are, with few exceptions, reduced to the Bhû type. In this respect it goes further than Pali, treating as verbs of the first conjugation many which in Pali retain the type of other conjugations. The fifth, seventh, and ninth conjugations, which in Sanskrit insert with certain variations, are all reduced to one head by regarding the as part of the root, as is also the case with the of the fourth class. The a inserted between the root and termination of the Bhû class is used throughout, though occasionally weakened to i, or changed to e from some confusion between this and the e = aya, which is the type of the tenth class. The following examples will illustrate the above remarks.

1 Kuhn, Beiträge, p. 7.

2 Pischel's admirable edition of Hemachandra's Grammar (Orphanage Press, Halle, 1877) has reached me just as this work is going to press, and too late to be of use for this edition, except for a few hasty notes here and there. Mueller's Beiträge zur Grammatik des Jainaprakṛit came into my hands about the same time. I find it enables me to add a few illustrations to this section, which, however, was written in the latter part of 1875.

VOL. III.

2