3i8 REPORTS ON THE STATE OF SCIENCE, ETC. 



perceive various relations between these individual structures. Between 

 any pair selected at random there will in general be an obvious difference 

 both in size and shape. If we have a really large and miscellaneous selection 

 to examine, we shall see between various pairs practically every possible 

 kind of phenomenal relation that can exist between triangles. In general 

 there will be nothing unique in any of these relations, nothing to which we 

 can attach more importance in the case of one pair chosen at random than 

 in the case of any other pair. There is only one relation which will strike 

 us as unique : if we find that there are two or more triangles of the same 

 shape we can pick them out at once as a group characterised by the unique 

 relation of having a recognisable property common to all its members. 



Sameness of relation-shape is a symmetrical and transitive relation between 

 relations or relation structures, and it might seem that it is this property 

 which makes it uniquely important to perception. I don't think there are 

 grounds for this view. We do not in perceiving phenomenal relations 

 analyse them into their logical classes. The unique significance we attach 

 to a pair of triangles of the same shape when we pick them out from a medley 

 of pairs exhibiting other relations between their shapes is instinctive. It 

 does not depend on a conscious realisation that if (triangle) A is of the same 

 shape as B, B is also of the same shape as A, and further that if B is the 

 same shape as C then A is also of the same shape as C It might be sug- 

 gested that we do realise these facts almost simultaneously with the 

 perception of the triangles, but even if this were so only a mathematician 

 would also realise that the facts were of any unique importance as properties 

 of a relation. I can see no reason why any phenomenal relation should 

 have a unique significance for intuitive perception apart from association ; 

 and if we find, as we actually do from experience, that some particular kind 

 of relation has a unique significance it must be because it has become 

 indelibly associated in our minds with some unique type of experience of 

 outstanding frequency of occurrence. Broadly speaking, practically the 

 whole of our perceptual experience consists of the observation of permanent 

 objects and the ever-changing relations which .these permanent objects, 

 regarded as self-contained unchanging entities, may enter into with each 

 other. The changing relations between permanent objects are ephemeral ; 

 the relations between the moving motor-car and the mile posts are different 

 every time we look : but the relation structures constituting any given 

 aspect of the permanent objects, the motor-cars and mile posts themselves, 

 do not change, and are perceived every time we observe the same aspect 

 of the objects, whatever the circumstances of observation. Any particular 

 relation between different objects is therefore observed very rarely as com- 

 pared with the relations which characterise objects themselves, and which, 

 for a given object, are the same at all times. This is why ' sameness ' of 

 relation structure is perceived far more frequently than any other relation 

 between relations. It is the relation between the perceived relation structures 

 of any object on different occasions. 



It may be objected that when we identify the same object on different 

 occasions we are not really concerned with a relation between separate 

 relation structures observed on these occasions, but are merely perceiving 

 the one identical structure every time. This, however, is to confuse the 

 relation structure which constitutes any phenomenon with the abstract 

 relations involved in it. A phenomenal relation structure is not a structure 

 of abstract relations, but of instances of relations. For example, suppose 

 we arrange three billiard balls so that each ball is distant ten feet from each 

 of the others and that we also arrange three other billiard balls in the same 

 way. We have here two phenomenal relation structures in which all the 



