The need for incorporating a valid account of phonological universals into a general model, or metatheory, of phonological descriptions may be taken as self-evident. In the first place, such a treatment of universals significantly simplifies descriptions of the phonologies of particular languages, since the universals, once incorporated into the metatheory, need not be explicitly mentioned in each description. Even more important, the incorporation of universals into the metatheory constitutes-as Chomsky and other generative phonologists have often pointed out-a kind of 'explanation' of otherwise apparently idiosyncratic features of particular phonological systems. That is, those features of particular systems that reflect phonological universals are, by definition, predictable, and this predictability is automatically accounted for if such features are assigned not to the particular systems themselves but, rather, to a metatheoretic model that specifies the features of phonological systems in general.1 The present paper is concerned with a class of phonologlcal universals that has thus far apparently not been scrutinized by generative phonologists: universals having to do with assimilation. The paper will attempt to show that, as in other cases, the ineorporation of these universals into the general phonological model can simplify the description of particular phonological systems and can explain otherwise apparently idiosyncratic features of these systems. Evidence in support of these claims will be drawn primarily from the phonological system of the Akan language of Ghana.2