3 * WHAT IS SCIENCE? 



To make this important matter clear,- it may be well 

 to suggest a procedure which might be adopted. We 

 might make both the normal and the abnormal people 

 look at the objects through a red glass. Through the 

 glass, of course, everything will look the same colour to 

 both normal and abnormal, but different objects will 

 appear different shades of the same colour. The pink 

 objects A, B, C, . . . will all appear the same shade, 

 and so will the greenish objects X, Y, Z . . . ; but the 

 former will appear a lighter shade than the latter ; and 

 they will appear a lighter shade, not only to the normal 

 people, who see the difference of colour when the red 

 glass is not interposed, but also to the abnormal,who do 

 not see this difference. Here then universal agreement 

 has been attained ; every one agrees that through the 

 red glass the objects look different. Accordingly we 

 regard the appearance through the red glass as a better 

 basis for science than the appearance without the red 

 glass ; we say that scientifically the objects are different 

 in colour if they appear different through the red glass, 

 and we call one set of people normal and the other 

 abnormal because one set agree and the other do not 

 with the distinction based on this truly scientific criterion. 



It is a very remarkable fact that, wherever we find 

 such abnormal people under permanent hallucinations 

 (and we find them in regard to all the senses), we can 

 always find another test which, in the manner just 

 described, enables us to restore universal agreement. It 

 is this fact, which could not be anticipated, which makes 

 science possible, and gives its great importance to the 

 test of universal agreement. 



But perhaps the reader may doubt whether there are 

 really judgments about which every one agrees, if we are 

 allowed to include people with the most extremely 

 abnormal sensations, such as the totally blind or the 

 totally deaf. The doubt can only be removed by quoting 



