Abstract

Scientific realism postulates that science aims for truth in both the domains of the observable and the unobservable, and is capable of achieving this aim, at least approximately. From the realist perspective our current scientific theories are on the right path to their aim, encapsulating a significant degree of theoretical truth. A key argument supporting this viewpoint is the continuity observed between successive scientific theories, interpreted as the preservation of truth. However, we contend that this continuity argument is problematic in significant cases. Features of older theoretical schemes frequently do not persist in subsequent theories but _emerge_ as limiting cases with restricted applicability domains and fine-grained structures that differ from what the older theories deemed possible. The alterations accompanying theory replacement are often more drastic than what is assumed by realist intuition, justifying skepticism about continuity serving as an indicator of theoretical truth preservation. We propose an argument wherein continuity is instead viewed as a consequence of empirical success preservation.