THE CRITICAL PHILOSOPHY 73 



of the inadequacy of analytic propositions to provide a basis for 

 science. But it is equally important to recognize that there 

 could have been no such problem, had not Kant accepted the 

 dogmatic doctrine, that analysis must yield final elements. In- 

 deed, the very division of propositions into analytic and synthetic 

 rests on this assumption. For no proposition could be deter- 

 mined as synthetic, unless a complete definition of its terms had 

 exhibited their ultimate disparateness. Moreover, the recogni- 

 tion of the necessity for synthetic connection means that the 

 terms of thought are ultimately simple elements, possessing no 

 inherent relationships. It is only because a synthetic proposition 

 connects B to an A, which in itself, as A, has no relation to -B, 

 that the connection is conceived to demand an explanation. The 

 validity of synthetic relationship can never be grounded in the 

 nature of the terms themselves. 



This, then, is the problem of criticism: How is the validity of 

 these indispensable relationships to be explained? The solution 

 criticism finds in the assumption, that pure thought supplies to 

 experience certain universal modes of relationship, to which every 

 particular experience must be subject, as a condition of its be- 

 longing to experience at all. This assumption has its justification 

 in the fact that without it the validity of thought must remain 

 unexplained. It is the only possible means of accounting for 

 the element of necessity in experience, of accounting for the fact 

 that experience is a unity and not a chaos ; and hence it must be 

 accepted. What makes the assumption necessary, however, it is 

 important for us to note, is the fact that relations are conceived as 

 external to the terms related. It is only relations between terms 

 already determined as A's and 5's, that a priori forms of thought 

 are needed, or are competent, to explain. If experience does not 

 yield terms given as essentially discrete, the a priori forms must 

 remain inoperative. For if the A's and B's are not discrete, 

 then their relationship to each other must constitute in part their 

 determination as A and B. Consequently, not only is there no 

 need of assuming universal forms of thought to account for their 



