346 DEVELOPMENT AND PURPOSE CHAP. 



and given AB we have C, but we may have A or B without 

 C. We have, in fact, to assume a relation which we may 

 write (AB)f, and we may take the antecedents in time of 

 this collocation to be (ab}c. 



Now there are two possible interpretations of this rela- 

 tionship. The first turns on the Plurality of causes. The 

 antecedent (ati) c may be written as a single cause y. It is 

 the cause of A and B in this instance. But there are also 

 other causes S etc. of A (and perhaps of B as well). But S 

 yields AD, so that if we have AB here and AD there, it is 

 because in one case the cause is y, and in another <5. Before 

 considering this possibility, let us examine the alternative. 

 It may be that c is not a cause of A or B, but a condition 

 of the combination of their causes which are a and fr, while 

 it has its own direct effect C. Thus the existence of the 

 combined process leading from ab to AB is conditioned by 

 the process c-C. And unless we are to bring in a fourth 

 term, the necessity must be reciprocal, so that if we are to 

 find a whole in which variable collocations are reduced to 

 relations which hold of the parts as such, it must be one in 

 which a certain collocation of elements implies and is 

 implied by the existence of another element in the whole, 

 which element itself may or may not be a collocation of 

 more elementary parts. We have then a reciprocal rela- 

 tion, a mutual necessity as between two simultaneous things 

 or processes. 



In this, which constitutes the simplest case of conditioned 

 variation, we should have two processes necessary to one 

 another at successive stages. It would be indifferent 

 whether we said that AB and C were mutually necessary, 

 or ab and c. But this indifference depends on the assump- 

 tion that c has a single effect C. It is equally possible that 

 c may enter into other relations besides its relation to ab, 

 and in these combinations produce effects D, E, varying 

 from case to case. If that is so, we shall not find anything 

 co-existent with AB which implies and is implied by it. 

 We shall have to explain AB by reference, either to an 

 antecedent combination (ab]c^ or teleologically to some 

 subsequent combination (a/3) 7. All that we can say is that 

 AB must either belong to a whole, of which the parts 



