From the time of Vararuci on, most scholars, with the exception of some Gujarat Jain writers, have tended to view the Prakrit dialects in terms of Sanskrit and the divergencies from regular Sanskrit derivation have dominated discussions. T. Burrow's studies on the language of the Kharoshti documents from Chinese Turkestan (1931) brought a new perspective to Middle Indo-Aryan, and showed how much Prakrit was of intrinsic interest in its grammatical structure. One of the most striking characteristics of Prakrit is an internal and not a Sanskrit-dominated feature, namely the bewildering profusion of declensional endings and the apparently haphazard manner in which they can be used. Thus in the case of the nominative-accusative plural of the common neuter nouns in -a the grammars state that-ā, -āïm, -āï and -āni are all used in Prakrit (Pischel 1900: 255). It is well known that some of these endings are dialectal variants as will be discussed below, but this does not account for the entire situation. Such a profusion of alternants, and free variation of the kind postulated by Prakrit grammars would be unthinkable in any natural language. It would seem obvious that syntactic and stylistic variation must be involved. An attempt has been made in a recent paper (Schwarzschild (1977)) to show this in connexion with the locative endings and a similar situation is discussed here in relation to the neuter plural endings.