262 DEVELOPMENT AND PURPOSE CHAP. 



say, ABC becomes DEF, there may be an element F com- 

 mon to FGH, a phase in quite a different process LKM. 

 In this case F is said to have two distinct causes, but (a) the 

 causes as wholes have effects which differ as wholes, and 

 (b) each cause is still a phase in a transition which leads 

 inter alia to F. (3) These assumptions explain scientific 

 induction both in its successes and its difficulties and 

 failures. They show that the problem of thought is that of 

 disentangling the relevant from the irrelevant. In experi- 

 ence as it comes to us the two are involved in what at first 

 appears as a hopeless tangle. Continuity yields the first 

 clue. What persists amid change is taken provisionally as 

 self-determined, i.e. as substantial, and the process that 

 goes on steadily and is repeated when circumstances differ 

 is treated as self-determining, i.e. as causal. The principle 

 thus roughly carried out by common sense may be simply 

 formulated thus. Let an element A be introduced into an 

 environment BC, and be the starting-point of a process A-a. 

 This process is not due to BC as such, because it did riot 

 arise till A was introduced. But BC may contribute to it. 

 Then let A also be introduced into the environment DE 

 having nothing in common with BC, and let the same result 

 follow. The process A-a is then not conditioned by any 

 part of the environment, that is, it is self-determining. 



This statement of the method of scientific induction is 

 open to criticism along two lines. One attacks its form or 

 principle, and founds itself on the Plurality of Causes. B 

 or C, it argues, may be the cause or part cause of a in the 

 one case, D or E in the other. But we have seen that the 

 plurality of causes is a doctrine of limited application. BC 

 on the one hand, EF on the other are either permanent, and 

 if so do not yield any element of a, or they are phases in a 

 transition, or processes. If either of these processes might 

 be regarded as leading to one of the conditions of a, the 

 same cannot be true of the other, for they are ex hypothesi 

 alike in no respect. They neither are nor are becoming 

 alike. If the hypothesis then be granted the universal 

 relation A-a is established. 



But at this point more substantial difficulties are opened 

 up. How can so much be known about the concomitants ? 



