Abstract
Within the standard structuralist approach, the theoretical description of a system by means of an empirical theory T is regarded as an extension process in which partial models are extended into models of T by supplementing suitable T-theoretical functions. Thereby, it is taken for granted that the base sets, on which these functions or relations are defined, can be assumed as given independently of the theory in question. My aim in this paper is to show that, in many cases, this assumption is not justified. I will argue that this fact—which may be seen as a mere technical complication on first sight—turns out to have profound consequences for core concepts of the structuralist framework: First, the concept of intended applications has to be reconsidered. Second, the extension process has to be described in a more differentiated and complex way. Third, the theory-dependent determination of base sets has major implications concerning the defense mechanisms of empirical theories against conflicting observational data