This paper will present new evidence which bears crucially on an empirical question not only directly relevant to the future of generative phonology but also clearly pertinent to the field of linguistics as a whole, as I have shown elsewhere (E. Maxwell 1978 a, b, 1979 a, b, c, forthcoming a, b, c, d). It will be argued that all previous putative analyses of this problem were either a) completely without empirical motivation or b) void of content, and were therefore theoretically insignificant and empirically inconsequential. The facts upon which the present analysis are based necessitate a principled and empirically crucial interaction of the various rules and components, for which none of the previous methodologically naive hypotheses were able to account. The independently motivated approach delineated here furthermore excludes on principled grounds a set of logically possible sequences from occurring in any natural language. As no counterexamples were found in the investigation of our sample (all known human languages), it is therefore claimed that these findings constitute a linguistic universal. It will finally be argued that any further investigation of the problem outlined above, must of necessity, for the reasons just given, incorporate this paper’s working hypothesis and theoretically substantiated methodological framework.