NATURE OF GENERALISATION. 401 



this constitutes an important part of actual discovery, 

 because when we recognise a similarity or identity in those 

 facts or phenomena, we extract from those facts a new 

 general truth. In generalising, however, even upon com- 

 plete proof, we do not create new truth but only evolve it, 

 because the statement of the identities in detail and that 

 of the principle or general proposition which embodies 

 them, are but equivalent to each other ; and in stating the 

 general truth we merely affirm in fewer words that which 

 has already been admitted in a greater number. 



In the second case, the process of generalising consists 

 in extending the general idea by means of the faculties of 

 imagination and inference to facts or phenomena which we 

 have not actually perceived, and thus propounding new 

 hypotheses to be verified, or questions to be answered. When 

 we infer that certain objects we have never perceived (and 

 in some cases may never be able to perceive) are similar in 

 some respects to those we have perceived, our generalisa- 

 tion is to a greater or less extent uncertain, and, strictly 

 speaking, we do not make a discovery. A general truth 

 is not completely established until all the facts which 

 support it are found and observed, but in proportion as a 

 principle is more extensive and liable to fewer exceptions, 

 so may we, by generalising, justifiably assume its existence 

 in unseen cases ; for instance, we may with a high degree of 

 certainty assume that the law of gravity operates univer- 

 sally, without being able to verify it in every particular case ; 

 in fact, not a principle of nature exists which has been 

 verified in all its instances. In this way, inference, based 

 upon suitable knowledge, sometimes enables us to assert 

 with safety that which we cannot prove. We cannot, with 

 certainty, arrive at once at a general truth ; we may, how- 

 ever, guess it, and then prove it, or we may be led to it by 

 experience ; but in every case we must build it upon sum*-* 



D D 



