CHAP, ii SCIENTIFIC RECONSTRUCTION 251 



the present, and that judgment is liable to err through 

 defect or confusion of memory. The picture that he 

 has even of his own past is not a simple and straight- 

 forward reproduction of that which he has actually lived 

 through. Memory is not a cinematograph. It brings 

 together mutually relevant data, it selects and rejects. It 

 analyses and constructs. Pure experience then, in the 

 sense of the sum of the contents present from time to 

 time to consciousness, is only a material on which the mind 

 works, and it is for any one of us a slender material rela- 

 tively to the wide range of our thought. 



What we have to enquire then is by what methods 

 thought treats this material and whether these methods are 

 valid? The broad answer to the first question is that 

 thought acts on its material, (i) by decomposing or analy- 

 sing it into elements, (2) by bringing different elements 

 together, without being necessarily confined in so doing 

 to the empirical order, (3) by taking the relations which it 

 so finds under certain conditions as true of reality in 

 general, and (4) by comparing its results and correcting 

 them one by another. The broad answer to the second 

 question is that this process of correlation and correction 

 can be so adequately performed as to yield results which, 

 in their general application, will hold true. 



2. The central difficulty here turns on the conditions of 

 valid generalisation. We have no a priori guide on the 

 point, for, in fact, simple and uncritical generalisation goes 

 far beyond the limits of certainty. We do not learn to 

 generalise as some have thought. fc We learn not to gener- 

 alise as often as we wish. What conditions of generalisa- 

 tion then may be held valid, and why? The difficulty of 

 finding any satisfactory reply to this question has been the 

 persistent stumbling-block in the way of any theory of 

 experiential reconstruction. In particular, it has led both 

 in the theory and in the practice of science to a view which 

 would confine valid reasoning to deduction and allow to 

 experience only the secondary function of corroboration. 

 Reasoning being thought of as essentially deductive in 

 character must be based accordingly on first principles 



