An account of physical explanation derived from the instance view of scientific explanation is outlined, and it is shown that this account does not cover explanations by theories which contain theoretical functions. An alternative account, also derived from the instance view, is proposed on the basis of Sneed's account of theories. It is shown that this account does cover theoretical explanations. Finally, it is shown that this account can accommodate explananda that record errors of measurement.

The philosophical debate whether the epistemological and conceptual structure of science is better characterized as hierarchical or as holistic cannot be decideda priori. A case study on general relativity should help to clarify our representation of this section of physics. For this purpose Sneed's model-theoretic approach is used to reconstruct the structure of relativity. The proposed axiomatization of general relativity takes into account approximations and utilizes local models for a realistic view on the functioning of the theory. A central objective (...) of the paper is to give an explication of the approximative empirical claim of general relativity, which is designed to identify the empirical content of the theory in a weak hierarchical form. (shrink)