CONTRIBUTIONS TO SCIENCE. 471 



emits very little light indeed, with the white disc, so that the 

 amount of light received from the central circle is approxi- 

 mately proportional to the angle of the white sector. On 

 spinning the top, the central circle appears of a neutral gray, 

 which is the same as a dull white, and the darkness of 

 the gray increases with the amount of the black sector intro- 

 duced. If, now, the outer circle appears yellow compared 

 with the central gray when the top is spinning, it is found 

 that, by increasing the amount of blue and diminishing the 

 red and green, the circles may be matched in hue. Similarly, 

 if it appears too green we must diminish the green and in- 

 crease the red and blue, while if it appears purple we must 

 increase the amount of green. After a little experience it 

 is easy to recognise the character of the change required. 



Now, suppose that we have three and only three colour 

 sensations, and that, for the sake of argument, these 

 correspond respectively to red, green, and blue. Then if 

 two colours contain the same amount of red, the same 

 amount of green, and the same amount of blue, they must 

 be in every respect identical. There are, therefore, three 

 conditions to satisfy in order that two colours may match. 

 But in the case of the top, two other conditions must be 

 also fulfilled, for the portions of the smaller discs employed 

 must exactly fill up the circle, and the same must be true 

 of the larger discs. Hence there are in all five conditions 

 to be fulfilled in order to make a colour match, and in 

 general five discs will have to be employed and matched, 

 two against three or three against two. If there had 

 been four primary colour sensations, we should have re- 

 quired in general to employ six discs to ensure a match, 

 and so on for any other number. Maxwell showed that 

 with a normal eye a match could always be obtained with 

 five discs, but that with a colour-blind person a match 

 could always be obtained with four discs, thus demonstrating 

 that the normal eye possesses three independent colour sensa- 

 tions, while the eye of the colour-blind possesses only two. 

 This conclusion will appear more evident from the experiments 

 with the colour box, which will be described presently. 



