PROBABILITY AND ENTROPY 143 



tuations have occurred, there is nothing to show 

 that the earth has gro^vn colder on the average. 



In any case this unidirectional flow of energy 

 from the sun without apparent source concerns 

 the law of conservation of energy rather than 

 the law of dissipation. If we have to state the 

 latter in conformity mth scientific usage, we 

 must not talk of the universe of which we know 

 so little, but of some isolated system which can 

 be studied at will. Thus the law of Clausius 

 must read, "Any system left to itself ap- 

 proaches in a single direction a definite state of 

 equihbrium, or, in other words, its entropy in- 

 creases steadily toward a maximum." It is this 

 statement that must be challenged. 



However, before proceeding with this analy- 

 sis, I must point out the great influence of the 

 unidirectional process that we are now consider- 

 ing upon the concept of time. In the pure ge- 

 ometry of space-time there is no dissymmetry 

 between up and down, or between forward and 

 backward in time; nor have w^e found in the 

 laws of mechanics, to which that geometry is so 

 admirably fitted, any evidence of such dissym- 

 metry. Even our new picture of the phenomenon 

 of radiation gives no intimation of unidirec- 

 tional time. On the other hand, all of the inci- 



