74 Journal of the Asiatic Society of Bengal. [N.S., XII, 



Before proceeding, I must correct a mistake into which I 

 have incurred in my " Notes on the Grammar of the Old Western 

 Rajasthani, with special reference to Apabhramsa and to Guja- 

 ratI and Maravarl. l " I am hardly responsible for it, as when 

 I wrote the ' fc Notes," I had never been in India and for all 

 information concerning pronunciation in GujaratI and Mara- 

 varl, had completely to rely on the accounts given by others, 

 which I afterwards found to be incorrect. In the first chapter 

 of the aforesaid " Notes," I had stated that the ai and ail of 

 Old Western Rajasthani become e, 6 in Modern GujaratI and 

 ai, au in Modern Maravarl. This is inaccurate. In both 

 Modern GujaratI and Maravarl, the ai, aii of Old Western Raja- 

 sthani become e and o. 



What I mean by e and 6 is a wide sound of the e and o 

 vowels, approximately, not exactly, corresponding to the wide 

 sound of a in the English word " hat," and o in " odd." The 

 difference is mainly in the quantity, the Maravarl vowels e and d 

 being more prolonged in pronunciation than the corresponding 

 vowels in the two English words quoted above. It is therefore, 

 originally, a long wide sound. In contradistinction to it, Mara- 

 varl possesses also a narrow sound of the same vowels e and o, 

 which I will mark by an acute accent, thus e 9 6. This sound 

 corresponds to the sound of a in c< care " and o in " old," and 

 can be quantitatively both long and short — the latter case is 

 very rare, — whereas the wide sound can be only long. Now, in 

 Maravari — and so in GujaratI — the distinction between the 

 wide and narrow sound of e and o is of primary importance. 

 There are many words, which are identical in form and 

 differ only in that one contains a wide e or o and the other 

 a narrow one. It is strange that no GujaratI grammarian has 

 ever realized that the real difference between the two sounds of 

 e and o in Gujarat! is not one of quantity, but one of quality. 

 The case here is very analogous to that of French and Italian, 

 where we also find two sounds of e and o, one narrow and the 

 other wide. 



In Modern GujaratI there is nothing to distinguish the two 

 different sounds of e and o in the writing. Both are repre- 

 sented by a single matra, thus : % stays for both he and ke, and 

 *irr for both led and ko. In Old Maravarl manuscripts, though the 

 distinction is by no means generally observed, there is a ten- 

 dency to represent the wide sound by two mdtrds and the narrow 



by a single matra. Thus : ££ = «, ke = % kd = ^, Jtd = ^t. This 

 tendency is evidently based on an orthographical tradition, and 

 the history of the language shows that the Maravarl spelling is 

 the correct one. 



From an etymological study of all words which contain 



1 Indian Antiquary, Vol. XLIII-XLIV (1914-15). 



