ix EXPERIENTIAL RECONSTRUCTION 159 



scious they are stated in abstract terms, and figure as 

 axioms. In this analysis Kant gave the first critical 

 account of the nature of axioms. For the axiom rests not 

 on apparent self-evidence, the psychological feeling of 

 certitude, but on the correlating function which it formu- 

 lates in general terms. We shall return to this point later. 

 Here we may be content to remark that in the modern 

 philosophical movement from Kant onwards we have criti- 

 cism attacking the systematisation effected by thought at 

 both ends. We have it applied alike to the primary data 

 and the supreme correlating principles. Kant himself was 

 clear that these principles have no validity and no real 

 meaning, except in relation to that which is given in 

 experience, and his criticism as distinguished from his 

 quite inconsistent reconstruction has so far the same ten- 

 dency as that of the British empiricists. Thought is that 

 which has the function of correlating experience. What 

 is true is in the last resort judgment based on a duly cor- 

 related experience, and thought is the function of corre- 

 lating experience. There was needed, accordingly, a logic 

 of experience, or a scientific induction, and to elaborate 

 such a logic is as much the problem of modern as the 

 formation of a deductive logic was the task of ancient 

 thought. 



The evolution of the modern theory of method, how- 

 ever, has not been determined by philosophical impulses 

 alone. The revival of experimental science preceded the 

 modern movement in metaphysics, and the development 

 of mathematics engendered new methods of handling 

 experimental results. In particular the discovery by 

 Newton and Leibniz of methods of calculating from rates 

 of change to results, or vice versa, gave an enormous exten- 

 sion to applied mathematics. It became possible in 

 physics to assume the action of very simple forces, to 

 calculate the result of their operation in complex or remote 

 cases and to compare the result with accurately measured 

 observation. This is the method, not devoid of liabilities 

 to error, which has on the whole determined the advance 

 of physics. In fields where simple and measurable forces 

 could not be so assumed and where observation at first 



