Abstract In order to constrain morphological theory one could hypothesise that all allomorphy must be conditioned locally, in the sense that morphemes must be adjacent (and perhaps within the same word) to be able to influence the selection of allomorphs. In principle the restrictive force of such a constraint would obviously be considerable, in particular with allomorphy where the determining factors are morphological rather than purely phonological and an adjacency requirement hence cannot be motivated on phonological grounds. It is also obvious, however, that the empirical import of such a constraint could vary depending on one's understanding of the crucial notions in terms of which it is stated, and of course also on one's analyses of the data where the constraint is invoked. The adjacency requirement may, for example, be conceived of as pertaining to linear contiguity of formatives or to operational succession, and with complex words resulting from prefixation operationally alternating with suffixation linear and operational sequences do not coincide. And evidently there are also differences of opinion concerning the notions of morpheme and allomorphy themselves, not to mention those concerning which general descriptive approach (such as Item and Arrangement, Item and Process, Word and Paradigm) to prefer. Nevertheless, irrespective of all possible disagreements about details as well as principles, some kind of constraint against distant conditioning of allomorphy would seem to command much intuitive appeal, and in one guise or another indeed has often been assumed, explicitly or tacitly, in morphological analyses and theories. Even if it should turn out that an adjacency constraint, however formulated, admits of exceptions, one would hope that these will be exceedingly rare and attributable to unusual circumstances of one kind or another. 1 See Plank (1982) for a more extensive discussion of this constraint, of its various formulations in the relevant literature, and especially of the empirical problems it encounters. There are indications, incidentally, that the constraint at issue pertains not only to word structure but also to phrase structure (cp. e.g. Plank 1984)