The author aims: (1) to show that generative phonology uses essentially the method of internal reconstruction which has previously been employed only in diachronic studies in setting up synchronic underlying phonological representations; (2) to show why synchronic analysis should add the comparative method to its arsenal, together with whatever adjustments in theory are required to make this accommodation; and (3) to present some empirical results of a first approximation to a pandialectal phonology which has been worked out by the author. This paper represents a revision of a paper read at the Conference on Historical Linguistics in the Perspective of Transformational Theory held at the University of California, February 1, 1969. A shorter and more revised version will appear in the Conference papers to be published by the Indiana University Press. (AMM)