516 THE PEINCIPLES OF SCIENCE. [chap, 



yet it is uo more than the observed phenomena of light 

 and heat force us to accept. We cannot deny even the 

 strange suggestion of Young, that there may be independent 

 worlds, some possibly existing in different parts of space, 

 but others perhaps pervading each other unseen and 

 unknown in the same space.^ For if we are bound to 

 admit the conception of this adamantine firmament, it is 

 equally easy to admit a plurality of such. We see, then, 

 that mere difficulties of conception must not disci'edit a 

 theory which otherwise agrees with facts, and we must 

 only reject hypotheses which are inconceivable in the 

 sense of breaking distinctly the primary laws of thought 

 and nature. 



Conformity with Facts. 



Before we accept a new hypothesis it must be sliown 

 to agree not only with the previously known laws of na- 

 ture, but also with the particular facts which it is framed 

 to explain. Assuming that these facts are properly 

 established, it must agree with all of them. A single 

 absolute conflict between fact and hypothesis, is fatal to 

 the liypothesis ; falsa m uno, falsa in omnihus. 



Seldom, indeed, shall we have a theory free from 

 difficulties and apparent inconsistency with facts. Though 

 one real inconsistency would overturn the most plausible 

 theory, yet there is usually some probability that the fact 

 may be misinterpreted, or that some supposed law of 

 nature, on which we are relying, may not be true. It may 

 be expected, moreover, that a good hypothesis, besides 

 agreeing with facts already noticed, will furnish us with 

 distinct credentials by enabling us to anticipate deductively 

 series of facts which are not already connected and 

 accounted for by any equally probable hypotliesis. We 

 cannot lay down any precise rule as to the number of 

 accordances which can establish the truth of an hypothesis, 

 because the accordances will vary much in value. While, 

 on the one hand, no finite number of accordances will 

 give entire certainty, the probability of the hypothesis 

 will increase very rapidly with the number of accordances. 



' Voung's Works, vol. 1. p. 417. 



