ON THE NATURE OF LIFE 203 



may commit suicide (also indepeadently of all the others) on the 

 same day. Now an insurance company taking risks of houses 

 being burned down by accident, or of people committing suicide, 

 calculates its possible liability from the application of the 

 theorems of probability, and it would safely ignore such risks as 

 those we have just mentioned. 



The latter are incredibly small. The chance that a decilitre 

 of gas will separate into two portions of different temperature is 

 also incredibly small, but it is not zero. In the ordinary affairs 

 of life we neglect small risks, and say they are " practically zero," 

 or " infinitesimal," but when we apply such chances in specula- 

 tions concerning the origin and fate of the universe we must not 

 dismiss them unless we are sure that they are really insignificant 

 in the conditions. Now we are not going to extend Boltzmann's 

 results obtained from a study of the kinetic theory of gases to 

 the whole universe that is, we must not suppose that a reversal 

 of the second law depends on the collision end on of every mole- 

 cule. All that we suggest is that it is possible in some way 

 other that entropy may decrease in our universe instead of 

 increasing. This is a logical possibility, and, given certain 

 arbitrary conditions in a very limited system, it is a probability, 

 the numerical value of which can be estimated. Further, we are 

 compelled to postulate that somewhere or other, or some time 

 or other, the second law of thermo-dynamics must reverse itself 

 that is, some time or somewhere entropy must decrease, or have 

 decreased, in our universe, otherwise we shall be compelled (as 

 Sir William Thomson was) to postulate a beginning, or creation. 



Disentropic Phases in the Universe. Neglecting, in the mean- 

 time, the numerical value of the probability that entropy may 

 decrease (or that unavailable universal energy may become 

 available), we. may now proceed to consider the possible history 

 of our universe, taking as our " conceptual model " the changes 

 that occur in a small volume of gas left to itself. In the latter, 

 then, incredibly great periods of time may pass, and during these 

 the molecules of our gas are moving and colliding in a haphazard 

 fashion. Nothing happens in the system considered as a whole; 

 it does no work, although the total quantity of energy contained 

 in it is conserved. But some time or other, if we wait long 

 enough, the system changes, and its original, segregated condition 

 becomes restored. This condition in which the gas system 

 contains available energy and can do work lasts for an in- 



