Abstract

The problem of relation between statistical mechanics (SM) and classical mechanics (CM), especially the question whether SM can be founded on CM, has been a subject of controversies since the rise of classical statistical mechanics (CSM) at the end of 19th century. The first views rejecting explicitly the possibility of laying the foundations of CSM in CM were triggered by the "Wiederkehr-" and "Umkehreinwand" arguments. These arguments played an important role in the debate about Boltzmann's original H-theorem and led to the so called statistical H-theorem proposed by Boltzmann himself. (For the history of these early debates we refer to Brush's monograph (Brush 1976).) After CSM had been brought to "canonical form" by the Ehrenfests, (Ehrenfest and Ehrenfest 1959) the physicists turned away from the foundational problem leaving it to mathematicians to worry about in the form of what has become called the ergodic theory. In retrospect, the physicists' general mood seems to have been the hope that ergodic theory establishes rigorously what is needed to found CSM on CM and which had been expressed essentially by Boltzmann already (Wightman 1985). However, very few physicists followed closely the developments in the mathematical theory of dynamic systems. One of those who did was the Russian physicist N.S. Krylov. (For a brief description of Krylov's personal life we refer to the papers in (Krylov 1979).).