252 DEVELOPMENT AND PURPOSE CHAP. 



which cannot be proved. But it is admitted, that while 

 there are some first principles which are true axioms, 

 needing no proof, there are others which at the outset are 

 mere assumptions, taken up for the purpose of seeing what 

 flows from them. These conclusions can be tested by experi- 

 ence, and if there is agreement, the assumption on which 

 they depend stands uncontradicted. It may be true. If 

 further results are elicited and the agreement with experi- 

 ence continues, it becomes difficult to believe that an 

 assumption which works so well can be false. When it has 

 stood very wide and complicated tests, we need not trouble 

 ourselves to question it further. We may take it as true. 

 This is the only way by which experience can establish a 

 generalisation. Any such generalisation is at first a hypo- 

 thesis, and in proportion as its consequences are found to 

 conform to fact it becomes a recognised theory. 



But though this account is a fair description of what 

 is often the course of discovery, it is in no sense a theory 

 of proof, since it involves the fallacy inherent in the 

 c inverse ' method. If the hypothesis is true, certain 

 observable facts will follow. They do follow, therefore 

 the hypothesis is true. This is inherently bad logic, and 

 the theory that there is no proof obtainable from experience 

 but this is the parent also of much bad science. That 

 discovery should follow this course, that scientific explana- 

 tion should take this form and that scientific men should 

 shut their eyes to its defects as logical demonstration, are 

 all equally natural results of the position of our experience. 

 We are conscious that it does not, as it stands, yield us the 

 fundamentals of reality, but is an effect or appearance of 

 a more deeply set real order. What, under these circum- 

 stances, is more natural than to go outside experience, to 

 make a bold conjectural attempt to seize on some of the 

 fundamentals of the real order, to take up this position as a 

 point of view from which experience will become intelli- 

 gible, to reason out as one only can reason from the centre 

 what effects must follow, and if they coincide with that 

 which we actually find, to rest assured that it is no mere 

 coincidence, but the hand of truth? All this is, we say, 

 natural, but that does not make it less fallacious, it does not 



