SUITABLE CLASSIFICATION OF IDEAS. 53 



the classification of ideas in accordance with the mental 

 powers, such as into immediate and mediate, is not a 

 primary one, because the human mind is fashioned by 

 nature, much more so than nature by the human mind. 

 A more fundamental classification would be one in accord- 

 ance with the actual and possible existences of nature, and 

 the truths of the various sciences, which are themselves 

 determined by the separate forces and substances of 

 nature, and are therefore formed upon a mechanical and 

 mathematical basis. A less suitable division of ideas for 

 the purposes of scientific discovery is into observations, 

 comparisons, general ideas, inferences, and hypotheses. 

 In the logical classification of scientific ideas, also, every 

 class should differ from every other class by some distinct 

 mark or marks. None of the classes should overlap each 

 other ; and in order to make the classification complete, 

 no class of ideas should be omitted ; l we should also not 

 divide into primary classes instances of different degrees 

 of action of the same power, such, for instance, as acts of 

 the mind into conscious and unconscious ones. 



Scientific ideas are related to each other in a multi- 

 tude of ways ; and the relationships (except in the case of 

 ideas of imaginary existences) are determined by those of 

 the real and known existences to which the ideas corre- 

 spond, and which they represent. Notwithstanding the 

 almost infinite number of ideas, and their complexity 

 of relation, Mr. Alfred Smee, F.E.S., even proposed a 

 relational machine, or a kind of mechanical dictionary, 

 designed to show by mechanical means all the relations of 

 any one term or idea to all others. It was not, however, 

 really constructed, for he himself admitted that it would 

 cover a space as large as London. 



1 See Chapter on Classification. 



