144 ANNUAL REPORT SMITHSONIAN INSTITUTION, 19 3 2 



contrary experience is impossible; the second occurs because the 

 contrary experience is too improbable. 



This distinction is entirely theoretical; there is nothing in the 

 observations themselves to suggest which type a particular regularity 

 belongs to. We recognize that " impossible " and " too improbable " 

 can both give adequate explanation of any observed uniformity of 

 experience, and the older theory rather haphazardly explained some 

 uniformities one way and other uniformities the other way. In the 

 new physics we make no such discrimination; the union obviously 

 must be on the basis of (b) not (a). It can scarcely be supposed that 

 there is a law of nature which makes the holding of 13 trumps 

 in a properly dealt hand impossible ; but it can be supposed that our 

 failure to find equilateral triangles with unequal angles is only be- 

 cause such triangles are too improbable. Of course, my remark 

 does not refer to the theorem of pure geometry; I am speaking of 

 regularities of our experience and refer therefore to the experience 

 which is supposed to confirm this property of an equilateral triangle 

 as being true of actual measurement. Our measurements regularly 

 confirm it to within the highest accuracy attainable and no doubt 

 will always do so; but according to modern theory that is because 

 a failure could only occur as the result of an exceedingly improbable 

 coincidence in the behavior of the vast number of particles concerned 

 in any experimental measurement. 



We must, however, first consider the older view which distin- 

 guished type (a) as a special class of regularity. Accordingly there 

 were two types of natural law. The earth keeps revolving round the 

 sun because it is impossible it should run away. Heat flows from a 

 hot body to a cold because it is too improbable that it should flow the 

 other way. I call the first type primary law, and the second type 

 secondary law. The recognition of secondary law was the thin end 

 of the wedge that ultimately cleft the deterministic scheme. 



For practical purposes primary and secondary law exert equally 

 strict control. The improbability referred to in secondary law is so 

 enormous that failure even in an isolated case is not to be seriously 

 contemplated. You would be utterly astounded if heat flowed from 

 you to the fire so that you got chilled by standing in front of it, 

 although such an occurrence is judged by physical theory to be not 

 impossible but improbable. Now it is axiomatic that in a deter- 

 ministic scheme nothing is left to chance ; a law which has the ghost 

 of a chance of failure cannot form part of the scheme. So long as 

 the aim of physics is to bring to light a deterministic scheme, the 

 pursuit of secondary law is a blind alley since it leads only to proba- 

 bilities. The determinist is not content with a law which prescribes 

 that, given reasonable luck, the fire will warm me; he admits that 



