D O C U M E N T S 3 7 1 , 3 7 2 S E P T E M B E R 1 9 2 2 2 9 3 it does not follow, however, that one is not allowed to facilitate the description by a suitably chosen frame of reference without violating the relativity postulate. If, for ex., I approximate the real world by a “cylinder world” with evenly distributed matter and choose the time axis to be parallel to the one generated by the “cylinder,” this does not mean the introduction of an “absolute time.” Either way, there is no such system of coordinates in the universe in which the formulation of the natural laws would be preferred. As regards the real world, an exact definition of such a system of coordinates would be impossible anyway, even if the real world could be roughly approximated by that cylinder world. The principle of relativity does not contend that the world is describable in an equally simple or even in the same way as all systems of coordinates, only that the general laws of nature are the same with respect to all systems (more precisely: that the hypothetically possible natural laws should only be weighed against each other, as regards their simplicity, in their gen- erally covariant formulations). 371. To Franz Selety Berlin, 25 September 1922 Esteemed Doctor, Thank you very much for sending me your paper[1] and for your letter.[2] I sent a comment on the paper to the Annalen and hope that the editors will send you a correction proof when it has been typeset.[3] I have nothing substantial to object to in the first part of your paper [4] however, I see that you have not completely com- prehended the arguments in the general theory of relativity about the closed space connected with Mach’s thought.[5] They are presented in a booklet that recently appeared in Braunschweig by Vieweg: Four Lectures on the Theory of Relativity [6] I will ask the publishing house to have a copy sent to you. With kind regards. 372. To Edgar Zilsel[1] Berlin, 25 September 1922 Esteemed Colleague, I have not been able to put as much time into the study of your paper[2] as would have been desirable.[3] But as far as I can judge, the main problem remains unsolved, which seems to me to consist in the following: statistical mechanics assumes the validity of the following statement without proof.