2 9 4 D O C U M E N T S 3 7 3 , 3 7 4 S E P T E M B E R 1 9 2 2 A closed mechanical system originating from any given initial state describes a path of the following character along its energy hyperspace. Given a segment of the hypersurface, there is a temporal limit of the ratio: lingering period/total time, which can be written in the form [ ][4] where [ ][4] is a continuous function on the surface independent of the special initial state. A priori it would be possible that such a continuous function did not exist at all if it could be proved that it does exist, its independence from the chosen initial state would probably also be provable. But this first statement—the validity of which I personally do not doubt for the systems coming statistically into consideration—is, to my knowl- edge, still unproven. I cannot see that you have proved it anywhere, either. As long as it is not proven, an important term is missing in the chain of proofs for statistical mechanics. (That some of the consequences of the law partly are in part not valid in nature, in my opinion can only prove the non-validity of the mechanical equa- tions.) Regards from your colleague. 373. To Michele Besso [Berlin,] 26 September 1922 Dear Michele, Somewhere around the 3rd or 4th of October I’ll be visiting you in Bern in transit to Japan. Then we can talk about everything that I, the wretch, did not write you in response.[1] Warm regards, yours, Albert P.S. Zangger is here. I met with him today. 374. To Eberhard Zschimmer Berlin, 27 September 1922 Esteemed Doctor, The observations in your essay appear right to me, at least from the aspect of physics, as the only one I can judge with certainty.[1] However, in my view the important question for a comparison between relativity theory and Kantian philosophy does not stand out prominently enough: Are the spatio-temporal forms, etc., that [also] underpin relativity theory “a priori” just suitable means of