402 ACTUAL WORKING IN ORIGINAL RESEARCH. 



cient instances before we are logically justified in feeling 

 certain of it. 



Different pairs of facts or phenomena, whether they are 

 observed ones or not, differ in their degrees of essential 

 similarity from complete identity to entire difference ; and 

 in consequence of this, legitimate generalisation upon 

 them passes, by a series of insensible degrees, into the 

 feeblest analogy. And as the degree of essential similarity 

 between them differs in every different case, we can usually 

 judge correctly of the degree of real similarity only if 

 we have a full knowledge of the individual instances. 



These remarks show that to generalise with safety, and 

 to raise likely hypotheses by such means, requires much 

 self-discipline and an extensive knowledge of science, also 

 the still rarer qualification of being able to judge correctly 

 of the true extent of action of different scientific laws and 

 principles, and the essential circumstances necessary for 

 their operation ; only a wise man therefore can generalise 

 correctly. 



The mere act of generalisation is a comparatively easy 

 process ; and nearly all men are much too apt to employ 

 it, because it yields at little trouble and cost what appears 

 to be new knowledge ; they often use it without taking 

 care to base it upon proper or sufficient evidence ; for- 

 getting that it is much more safe to suspend the judg- 

 ment than to draw a general conclusion upon insufficient 

 evidence. The immoral practice of self-deception is to 

 many persons a great pleasure, so much so that it has been 

 said ' the pleasure surely is as great in being cheated as to 

 cheat.' 



In generalising upon facts, we always resolve them 

 into laws or principles ; and in generalising upon the 

 latter, we always resolve them into still more general 

 ones; and thus the final result of scientific research is 



