This dissertation is an investigation of a case of allomorphy between -au and -ā in duals and i-stem locatives in the Rigveda. Traditional accounts of this alternation present a generalized, purportedly original distribution of preconsonantal -a: and prevocalic a:v. Cases that do not fit into this generalization, such as preconsonantal -au, are attributed to the date of a hymn’s composition or phonological factors. This study represents an exhaustive compilation of each form in the Rigveda (over 3000 total), and it reveals a complex interaction among several factors. The date of composition and the phonological environment do have an effect on this distribution, but they cannot account for the whole picture. Another factor in the variation is lexical idiosyncrasies in the patterning of the allomorphic endings. In both i-stem locatives and duals, different stems are shown to have different patterns of distribution. Even after controlling for chronological and phonological effects and for lexical idiosyncrasy, one additional factor must be considered: poetic effects and the influence of surrounding forms. The traditional generalizations must be called into question, and indeed no simple generalization can be proposed that is able to account for the data in this case: it is only in the interaction of various factors on certain forms in particular contexts that the explanation can be found. The variation found in the Rigveda is reflective of variation and change taking place in the living language at the time of composition. In this dissertation, I explore the synchronic and diachronic aspects of these variations and the factors affecting them, and I investigate the distribution of the -a: and -au allomorphs in the duals and i-stem locatives, as well as related categories, with particular regard to explaining how the data of the Rigveda itself can be accounted for.