________________
No. 8.]
SANJAN PLATES OF BUDDHAVARASA.
145
or short. At all events the same must be compared with the Eastern Chalukya Kokkili, the Telugu-Choda Kokili,' the Chola Kokkilli, eto. The correct form seems to be a cerebral l, single or doubled. The name is certainly Dravidian. The same is probably the case with - Kalvivana, 1. 29. Palakėsi is likewise in its first part a Dravidian name; compare Kanarese puli, tiger, where it is of interest that the l of this word is dental and not cerebral. Also the remaining names are most likely of a similar kind. At all events, it seems quite certain that in ordinary words the two forms of la are correctly distinguished as mentioned above,
Now it is a curions fact that we find exactly the same state of affairs in some other inscrip. tions, viz. the spurious plates of Dharasēna II of Valabhi, dated Saka 400, and the spurious Gurjara grants of Saka 400, 415, and 417. Here also there are some few exceptions. In the Bombay Museum Plates of Dharasēna IJS we find the sign which usually denotes an initial or a doubled l written in the name Giriviligrāmah, l. 20, while the intervocalic form is used in lagna in trinägralagna-, 1. 27. In the Umēta grant of Dadda we find jivalokah, 1. 12, and trināgralagna., PI. II, 1. 9, with the intervocalic, and the name Gilaka-, PL. II, 1. 15, with the initiall. In the Bagumră grant7 we find the same words jivalokah, l. ll; triņāgralagna-, 1. 25, with the intervocalic, and the name Ushilathana, 1. 18, with the initial form, The Ilio plates likewise have jivalokah, 1. 10, and trinägralagna., 1. 21, with the intervocalio, and the name Akulēšvara., 1. 14, with the initial form of la.
It will be seen that the exceptions are everywhere of the same kind. In the first place, we find some standing terms in which the l has been treated as intervocalic, though it is the initial of the second part of a compound word. There is nothing extraordinary in this. We know from the history of the Prakrits in India that compound words were often treated as one word and suffered such phonetic changes as resulted from the word-Sandhi. It is also of interest to note that the exceptional use of intervocalio l in these inscriptions always recurs in the same words, an additional proof that all these forged records proceeded from the same source. The remaining exceptions to the role, where the initial form is used when a single l occurs between vowels, are all found in proper names, just as was the case in the Sanjän plates.
The necessary inference from this state of things is that at the time when these plates were engraved the sound of 1 differed according to its use. When it was initial or formed a compound consonant, it had a sound other than when it was used as a single consonant between vowels.
Now it is a well-known fact that a corresponding state of affairs is found in a number of Indo-Aryan vernaculars of the present day, such as Marāțhi, Rajasthani and Gujarati. As I have shown elsewhere, every old initial I and sach l's as are derived from old compound consonants in Marathi, and, I may add, in other languages of the same kind, remain dental, while on the other hand every old single unoompound l between vowels becomes cerebralised. I have also drawn attention to the fact that this is an ancient feature of certain Indo-Aryan vernaculars. In the so-called Paisachi dialects, which belong to the oldest Prakrits, it was already fully developed. 10 According to the oldest Indian tradition the old Paisāchi was spoken somewhere in the neighbourhood of the Vindhyas, 11 and we have no information that the same
Ep. Ind., Vol. V, p. 128 ; Vol. IX, pp. 51, 188.
Ep. Ind., Vol. V, p. 123 n. Ep. Ind., Vol. IX, p. 89 n. 1, and so forth. • Kielhorn, Inscriptions of Northern India, Nos. 346-349. Ind. Ant., Vol. X, pp. 277 .
Ind. Ant., Vol. VII, pp. 61 f. Ind. Ant., Vol. XVII, pp. 199 f.
. Ind. Ant., Vol. XIII, pp. 115#. • JR48., 1902, pp. 417 ft. 10 Of. Pinebel, Grammatik der Prakritsprachen, para. 260.
11 Of. Konow, ZDMG., Vol. LXIV, pp. 95 #. I cannot find that my arguments have been weakened by Grierson, ibidem, Vol. LXVI, pp. 49 f.
