ii SCIENTIFIC RECONSTRUCTION 269 



and explanation, the two aspects of the work of reason in 

 thought, appear to take us outside the content of what is to 

 be proved or explained. But if this process be generalised, 

 it inevitably leads us to something which is neither proved 

 nor explained. In proof of proposition A I adduce pro- 

 position B, and in proof of B I adduce C, and so forth as 

 long as may be. But wherever I may have arrived in this 

 process, my first proposition is always unproved. Similarly 

 I would explain an event, and I do so by referring it to B, 

 and B I refer to C and so on. But wherever I may have 

 arrived in the process, the event from which I start is always 

 unexplained. This way of conceiving reason, then, leaves 

 its work necessarily incomplete ; there must be something 

 unreasoned. It also leaves it dependent, for what is 

 reasoned out follows from what is not reasoned out. 



This leads us to ask whether there is not another way 

 of regarding the work of reason which is not thus self- 

 mutilated. Let us, as before, suppose that we have formed 

 a certain judgment A, which we hold with a certain measure 

 of conviction. We ask, however, for proof of it ; we find 

 another judgment B which is formed independently, again 

 with a certain measure of conviction, and which on com- 

 parison is found to necessitate A as a consequence. Now 

 if we take B as certain, as a definitely established truth, A is 

 established along with it, and the question of truth or 

 falsity is closed. But even if B is not certain, if it is only 

 probable, still its probability will affect our belief in- A. 

 Provided that that belief has anything short of the maxi- 

 mum of intensity, it will be strengthened by the corro- 

 boration of an independently probable argument, just as 

 it would be pro tanto shaken by conflict with some inde- 

 pendently probable argument. The degree of validity 

 which we reasonably attribute to A is, in short, dependent 

 on two factors, 1 on the force of conviction with which the 



1 1 assume for the sake of simplicity that the judgment A is immediate 

 in the sense that it rests on processes which cannot be analysed out. If 

 we suppose this analysis to be effected and all the empirical data and 

 psychological laws contributing to the judgment A to be set out in the 

 form of explicit judgments, then the validity of A would simply depend 

 on all the independent judgments leading up to or corroborating it, each 

 one of these having a definite felt force of its own. 



