In dialectology, there are two competing conceptualizations of linguistic space: one based upon dialect areas/regions and the other on the idea of a dialect continuum. To date these two have been neither theoretically nor methodologically unified. In this article an attempt is made to furnish a theoretical basis from which both concepts can be derived, and a compatible method for the evaluation and interpretation of empirical data is introduced. Gaetano Berruto’s suggestion that varieties be seen as concentrations of co-occurrent variants serves as a starting point; when applied to linguistic geography this implies that diatopic varieties can be conceived of as regions of concentration within a geolinguistic continuum. As such, they are fuzzy and can overlap one another. A prototype-theoretical dimension is added to the model, allowing varieties to be understood as abstract dialect types that are manifested to varying degrees in concrete individual dialects. To operationalize this concept, factor analysis, a statistical technique for dimensionality reduction is employed as a category-building instrument in linguistic geography; its mode of operation is eminently suited to identifying and quantifying regions of concentration, i. e., plausible structures within a geolinguistic corpus. At the same time, linguistic co-occurrences are employed to identify systematic relations between variants. Applying the method to lexical data from the Sprachatlas von Bayerisch-Schwaben reveals the usefulness of the procedure, not just for the categorization of linguistic space, but also in the search for previously unidentified spatial structures.