Research on software evolution focuses on one hand on empirical investigations studying changes in long-living software systems, on the other hand on methods and tools, how such evolutionary behaviour of software can be controlled or supported. This paper departs from the observation that the empirical work and the tool-or methods-builder's work are quite often only obliquely related. Too often, the two camps depart from a tokensemantics of the word evolution. However, by ignoring the precise meanings of the words software and system, chances for misunderstandings are high. Therefore, an attempt to sharpen semantics of these three terms is made by stratifying their scope. On the basis of these sharpened definitions, some exemplary attempts for dealing with Software Evolution are mentioned..