Abstract

This dissertation attempts to provide an answer to the following question: what must be the properties and the structure of physical space-time so that the inertial effects be satisfactorily explained? ;To answer that question in a precise manner, five meanings--logical, ontological, mathematical, empirical, physical--of the terms "absolute" and "relative" are distinguished according as they refer either to space-time or to motions. ;In the first part of the dissertation, this conceptual framework is applied to Newton's Principia, the Leibniz-Clarke controversy, Clarke's vindication by Euler and classical mechanics as set forth in contemporary textbooks of physics. ;In the second part, the classical electromagnetic theory and the special theory of relativity are examined in the light of these five distinct meanings. ;Finally, in the third part, the conceptual framework is applied to the general theory of relativity as it is found in the writings of Einstein, Weyl and Weinberg. A study of Mach's ideas and some cosmological considerations are included. ;This detailed analysis allows us to establish the following four points: So far, physics has not succeeded in completely eliminating absolute space-time; it is shown in a precise way in what sense and to what extent this elimination has not yet taken place. We can speak of a progress in the history of scientific theories from Newton to Einstein in the sense that absolute space-time, cause of the inertial effects, becomes more and more empirical, nonetheless without reaching Mach's ideal. Metascientific concerns, such as the demand to connect real effects with real causes, have played a significant role in the evolution of physics. The epistemological status of the space-time concept must be approached via the problematic of the origin of inertial forces, rather than via the problematic of empirically founding geometrical properties on the observable behaviour of rods and clocks. Writings by physicists--including Einstein--show that their main concern is to provide a satisfactory account of inertial forces; besides, the behaviour of metrical devices is not the same for all reference systems