from Mr. 346, given under above, where the speaker is a Chandâla or man of the lowest caste, who may be held to speak a low form of Apabhrança. It is perhaps another of those Aryan roots which Sanskrit has rejected. The classical language uses instead masj, Latin mergere. H. भेट, भेंट 66 a, to meet " (to join any one), a, "to close, shut," P. भेड, S. भेट and भीड, G. भेट, M. भिड, both a and n, B. भेट, भेड, 0. ; and H. "to stand close to, to be crowded," P. id., S. G. id., M. ЯT, B. FHE"to approach near to," O. f "to be tight." The

There is also a sub

From the meaning I

was led to suppose “near,” which,

general idea is that of closeness or a crowded state. stantive “a crowd." 66 (Vol. I. p. 176) a derivation from a Sanskrit p.p.p. however, has been disputed. The question must for the present be left undecided.

H. "to efface," P. S. id., and more common H. fn, "to be effaced, to fail, wear out" (as a writing or engraving), and so in all. Of this stem, all that can be said is, that it is probably connected with मृष्ट "rubbed,” p.p.p. of √, though one would expect a Pr. fg or yz, and H. 3. There are two other stems ending in Z, which present nearly the same difficulty, viz.:

H. पिट “to be beaten," ऐसा करोगे, तो पिटोगे “If you act thus, you will get a beating," P. fuz, M. faz, both a and n, B. and 0. fuz a, and H. qz a, "to beat," not in the others. In Prakrit there is fug "to beat," fufue ad fuscafe “Having beaten this slave, turn him out" (Mr. 354, again in the mouth of a Chandâla), and fufzugŝu fau yulfa fg “I must roll about again like a beaten jackass" (Mr. 107). Here, unless this is a nonSanskritic old Aryan root, we can only refer to fug "ground, broken," p.p.p. of √ fqq “to grind," but this is hardly satisfactory, as this root has a descendant, H. "to grind," and fun, "to be ground."

H. n, “to lie,” "to be in a recumbent posture," and fen, “to

wallow,” P. लेट, लिट, S. लेट, G. M.id.

Probably connected with

; but there does not appear to be any Prakrit root to which it can be traced. The nearest Sanskrit root is "to lie down;" loṭṭai= svapiti.-Pischel, Hem. iv. 146.

There is next to be noted a small group of stems ending in, concerning which also there has been some controversy.

Skr. √"buy," ix. turfa and Huta, Pa. fauifa, Pr. fanu, H. कीन, S. गिन्ह (is not the हृ here due to some confusion with Pr. गेण्ह = ग्रह किन. ug “take”?), B. O. f. This is a single verb, the complications occur in the following compound with fa, fant "sell," ix. famuifa, Pa. fafauifa, Pr. fafauz, s. fafau a, “to sell,” 0. fa, Gipsy biknáva. But in H. faa is n, "to be sold, to be exposed for sale,” as चावल आज सस्ता बिकता “ rice is selling cheap to-day.” In M. बिक is both a and n, as बिकेल तसें पिकेल " when it is ripe it will sell." So also P. S. fan, "to be sold."

For the active H. has a sometimes pronounced "to sell," as a "he is आज चावल बेचता selling rice to-day. " P. बेच, G. वेच, B. वेच.

When we remember that

all verbs are prone to take the forms of the Bhû type, it is intelligible that fa should mean both "to sell" a, and "to be sold" n, for the Parasmai of the Bhû form would be विक्रयति, and the Atmane विक्रयते, and the final syllable being rejected as in √ ✈ and √ ✯ mentioned above, the stem resulting in both cases would be fa. S. and Gipsy have retained the या of the Pr. विक्किमद्. But whence comes the च in बेच ? "to pull, drag,” is a similar word.

H., commonly pronounced P. खिच्च and खिंज, G. M. खेंच, B. खेच and खेच, O. id. Also H. खिच n, "to be dragged,” B. fat, få"to be dragged or distorted (the face), to grin, make faces, writhe," M. f. From the meaning we are led to think of Sanskrit √¶“to drag," and although this root has been shown to have given rise to another pair of verbs kaṛh and káṛh, and in composition to ni-kal and ni-kál, yet it is not impossible that, used

in a different sense, it may have originated another set of words like khench and its congeners."

H. “to arrive" n, written in various ways as 4, 7, पहुछ, P. पहुंच, S. पहुच, G. पोहंच, पोंच, Mid., B. पहुंच or पहुंच, 0.457. In the dialects are some curious forms, as Marwari पूग and ч, which also occurs in Chand, and in Nepali. Chand uses also a form पहू, as दिन दोय मंझ नीके पहंत ॥ “ In two days one easily arrives (there).” Pr. R. i. 175. In Old-Gujarati also there is a verb usta, e.g. तळ कहे नारदने ने वखाण भाव न पहोतो " Says Nala to Narada,

this story does not arrive at mind" (i.e. is not probable).—Premanand Bhat, in K. D. ii. 74. S. पहुंचणु has p.p.p. पडतो, which latter looks as if it were from +, but this will not account for the . Hoernle (Ind. Ant. i. 358) derives this word from the old Hindi adverb “near," and कर "make," assuming a change of into; but though this change occurs in the ancient languages, there are only very few and doubtful traces of its existence in mediæval or modern times, and I do not think we can safely base any argument upon so rare a process. Hoernle goes so far as to consider H.

"to call," as the causal of ч, which

he says was (or must have been) anciently y. There is another possible

प्राघूर्ण

derivation from Skr. "a guest," which becomes in H., but

this fails to explain the final .

Some light may perhaps be thrown on the subject by some stems in the moderns ending in, for as

so arises from + (Vol. I. p. 326).

arises from

+,

Thus :

"cleave,"

a "he is

H. a, “send,” P. id. Here we have Skr. √f

"separate." Causal fa, which would make a passive

made to separate," i.e. "he is sent away." If we take the active causal as the origin of this word, we must admit an elision of the vowel between d

1 See note to Krsh in § 20. The cognate verb ainchna is also in use in the moderns.

VOL. III.

5

and y; or, taking the simple passive fa, we may assume that there "to be sent," from which the active “to send,” has

was a neuter

been formed; bhij, however, is not found.

H. बज ”, “ to sound,” P. id., S. वग and वज़, G. वज, M. वाज, B. id. Also H., “to play (music)," and n, "to sound." Probably from Skr. √ वद् “speak,” causal वादयति, Pa. वदति and वदेति, the passive of the causal is Skr. वाद्यते, Pa. वज्जति "to be beaten," i.e. "to be caused to speak,” as vajjanti bheriyo “drums are beaten." Hence the modern baj. The short form baj is apparently due to analogy.

§ 23. It is the business of the lexicographer, rather than of the grammarian, to work out the derivations of all the verbs in these languages, and even he would probably find the task one of insuperable difficulty in the present elementary state of our knowledge. It is hoped that the examples and illustrations given above will have enabled the reader to gain some insight into the general principles which have governed the modern languages in the process of forming their verbal stems. To conclude this part of the subject, I will now point out what seem to me to be the laws deducible from the examples above given, and from many others which, to avoid prolixity, I have not cited.

Single neuter stems are derived (i) from the Prakrit present tense of Sanskrit neuter verbs, or (ii) from the Prakrit passive past participle, or (iii) Prakrit has assumed one form for all parts of the verb, which form has been handed down to the modern languages almost, if not entirely, unchanged. Types of these three processes respectively are ho, baith, and uth.

Single active stems are formed from the Prakrit present of active verbs, and in cases where the verb in Sanskrit is not conjugated on the Bhû type, Prakrit usually, and the moderns always, adopt the Bhû type. Here, also, Prakrit has occasionally taken one form of root and used it throughout, and

the moderns have followed the Prakrit. Types of these classes are parh, kar, and ghen.

In the double verbs two leading processes are observable. Where the root is conjugated actively, or is active in meaning in the ancient languages, the modern active is derived from it, and in that case the modern neuter is derived from the Prakrit form of the Sanskrit passive, as in labhanu, lahanu, or as chhor, chhut. Where the ancient root is neuter, the modern neuter is derived from it, and in this case the active is derived from the ancient causal, as in tut, tor, or mar, mâr.

These rules, if further research should eventually confirm them, do not provide for every modern verbal stem, as there are many whose origin is obscure and doubtful. It is highly probable that as we come to know more about these languages, we shall find out other processes which will throw light upon the method of formation of many now obscure stems.

It should here also be noted that even where the same stem occurs in the same, or nearly the same, form in all the languages, it is not used in the same phase in all. Marathi and Sindhi have different sets of terminations for neuter and active, so that the fact of the neuter and active stem being the same creates no difficulty, the distinction of meaning being shown by the terminations. Thus in M. i, if treated as a neuter, would be conjugated thus: Present gânthato, Past gânthalâ Future gânthel, etc.; but if as an active, thus: Present gânthito, Past gânthilen, Future gânthil. In this language, therefore, we often find a verb used either as active or neuter; while in Hindi, which has one set of terminations for all stems, the difference between active and neuter can only be marked by the stem. In several rustic dialects of Hindi, however, and in the medieval poets, we often find the neuter verb with a long vowel, but confusion is avoided by giving to the active verb the terminations of the causal, thus "to grow big," "increase," makes its active "to make big," and rustic and