1890.] 



Lord Rayleigh's Colour Box. 



141 



scale of my instrument ; so that I have taken our colour matches as 

 equal, aad referred all others to them, although, as will appear, we 

 do not quite agree with the majority of people. In order to under- 

 stand the numbers given in the following pages, it is necessary to 

 state that in the neighbourhood of normal vision a difference of one- 

 tenth of a division means a difference of about 2^ per cent, in the 

 ratio of intensities of red to green. I consider that the mean of five 

 readings of a practised observer should not differ by more than that 

 on different occasions. If two good observers place the Nicol half a 

 division different from each other, then each should be able in his 

 own mind to be certain that the other's match is wrong. The 

 difference of a whole division is generally very obvious after a little 

 practice with the instrument. 



It appears that differences of between a half and whole division 

 are very common, so that there cannot be any doubt of the real 

 existence of small differences, possibly following, as far as the number 

 of observations allow us to say, the ordinary law for deviations from a 

 mean. But the larger differences, such as Lord Bayleigh was the 

 first to observe, seem certainly to be more frequent than the distri- 

 bution of small differences would lead us to expect. As has already 

 been stated, it seemed better in the first instance to confine myself 

 to a careful examination of a limited number of cases than to extend 

 the enquiry too much before I could form an estimate of the accuracy 

 which is to be expected from a casual observer. I have examined 

 seventy-five : of these three proved colour blind ; four, of which three 

 belonged to the same family, showed the same peculiarity of vision as 

 Mr. Balfour and Professor J. J. Thomson, while one showed a large 

 difference in the opposite direction. 



In the following tables I shall call my own reading zero, and shall 

 take as the unit difference between myself and others one-tenth of a 

 division of the divided circle. The first table gives the ratio of red to 

 green used in the match, corresponding to the stated differences of 

 reading : 



Table I. 



