FAILURE OF ALL OUR CLASSIFICATIONS. 207 



practice of classification also reacts upon and strengthens 

 the powers of attention, abstraction, and continuous 

 thought, because it leads to the formation of general con- 

 ceptions, in which ideas are connected together by their 

 natural relations and similarities. 



New classifications generally require new knowledge, 

 therefore in original research we are nearly always obliged 

 to employ the classifications which already exist. We 

 employ existing classifications for the purpose of suggest- 

 ing new hypotheses, and for testing those we have 

 imagined ; we also use them to aid us in detecting simi- 

 larities and differences, and in finding causes and other 

 relations of phenomena. On much less frequent occasions 

 we classify existing knowledge in various ways to raise 

 new questions, or we classify it in special ways to ascertain 

 if it supports a particular hypothesis. If our object is 

 to obtain the most philosophical view of science, and that 

 most conducive to discovery, we should classify scientific 

 truths, , not necessarily according to their apparent or 

 most obvious similarities, but according to their most 

 intrinsic and essential ones, so far as these are known. 



Classification of knowledge leads to discovery, because 

 it sometimes puts unrecognised general truths before us 

 in such a form that we can perceive them; for instance, 

 by arranging the elementary substances in the order of 

 the relative number of atoms contained in a given volume 

 of each, it was found that those containing the most 

 closely-packed atoms were, in nearly all cases, the most 

 paramagnetic, and those in which the atoms are farthest 

 apart were the most diamagnetic. A new discovery, 

 therefore, is sometimes disclosed and represented by 

 means of classification. 



At other times we classify new results, for the purpose 

 of discovering their cause or relations. As some orders of 



