392 STUDIES IN SPECIAL SENSE PHYSIOLOGY 



terms of three independent variables ; that is to say, from the 

 point of view of stimulations, normal colour vision is trichromatic. 

 It is important to notice that a choice of variables is, from the 

 theoretical standpoint, arbitrary ; indeed, if a table be constructed 

 in terms of three primaries, A', B', C', a second can be deduced 

 in terms of another three, A", B", C", because, in view of what 

 has been said, we can evidently define any one of the new 

 variables, e.g. A", in terms of the old ones by a linear relation of 

 the form 



' + c.C'. 



The process, in fact, merely involves a change of the co-ordinate 

 axes. Of course, if we choose the three primaries so close together 

 that we cannot experimentally reproduce all the spectral stimulus 

 values, our table will involve negative directions, but this is of no 

 theoretical importance. 



Summarising the points dealt with in the last few pages, we 

 find that 



(1) Fixing our attention on stimuli alone, mixing results are 

 functions of two variables and graphically expressible by means of 

 a plane figure. 



(2) Colour differences can be expressed in terms of three chosen 

 stimuli, the choice being theoretically unrestricted, but in practice 

 certain real stimuli are selected for convenience. 



I must reiterate here the fact that all these statements purport 

 to be merely descriptive and to resume experimental observations 

 as briefly as possible. Colour diagrams and the assertion that 

 colour vision is trichromatic are only short ways of expressing 

 experimental results ; they contain no hidden theoretical assump- 

 " tions, and their truth, or falsehood, is purely a matter of observa- 

 tion and necessary inference from such observations. It is essential 

 to separate the theory I propose to discuss from the data which 

 have given birth to it. 



SECTION III. PARTIAL COLOUR-BLINDNESS 



"We have seen that, from the purely experimental point of 

 view, the characteristics of normal colour vision admit of relatively 

 simple arrangement, that in fact all the results of stimulation can 

 be expressed in terms of three different stimuli and no more ; we 



