282 THE SCIENCE OF LOGIC 



ing with ever increasing probability to the actual alternative ante 

 cedent, or combination of antecedents, from which those results 

 actually followed ; the greater, too, will be their value in re 

 vealing to us the real nature of the supposed alternatives, and 

 in thus &quot; eliminating chance &quot;. 



268. APPLICATION OF THE CALCULUS OF PROBABILITY TO 

 NATURAL AND SOCIAL PHENOMENA. The mathematical estima 

 tion of probability is best illustrated in games of chance, where it 

 is applied to specially prepared materials. Can we utilize it to 

 estimate probability in regard to the occurrence, or recurrence, of 

 complex natural and social phenomena, the causes of which we 

 know either not at all or only imperfectly? Attempts have been 

 made in various directions to apply it to such data with, how 

 ever, no remarkable degree of success. As a matter of fact, the 

 degree of rational expectation we may entertain about the future 

 recurrence of any natural phenomena which we have repeatedly 

 observed occurring in the past, but are unable as yet to refer to 

 its causes, such probability we never think of basing on the con 

 siderations that apply to an artificial game of chance, or of 

 measuring by means of the calculus outlined above. Rather, 

 in such cases, our first endeavour is to detect uniformities of 

 coexistence or sequence even among phenomena which appear 

 at first sight irregular and unconnected ; and we do so by com 

 piling statistics, and seeking for hitherto unobserved concomitant 

 variations (243, 249). Research in this direction often leads to 

 the discovery of causes. 



Suppose, however, that we are in the presence of a phenomenon 

 which sometimes happens and sometimes does not happen in 

 circumstances of the same general character : we may set ourselves 

 the task of discovering whether or not this set of circumstances 

 as a whole, so far as we know it, is &quot; indifferent &quot; to the occur 

 rence or non-occurrence of the event in question. We are face 

 to face with a result recurring irregularly in certain condi 

 tions. We want to find out, if possible, whether it is causally, 

 or only casually, connected with those conditions. For this pur 

 pose we see whether or how far we can &quot; eliminate chance&quot; by 

 applying the rule for calculating inverse probability (267). But 

 there are two considerations which reveal the difficulty of such a 

 procedure. One is that we may possibly know very little about 

 even the general set of antecedents to whose combination the 

 event, when it does occur, is really due. The other is that this 



