VELOCITY OE WHITE AKD OE COLOTJHED LIGHT. 
275 
intensity is increasing with increase of speed the star should have a red tinge. 
Observation confirms these statements. Hence the observations can he explained on 
the assumption that blue light travels quicker than red light. 
The analytical expression of this result is quite simple. §§ 2 and 3 of the mathe¬ 
matical theory give the following values for the intensities of red and violet light 
(indicated by the suffixes R and V): 
ri.=!{2(i-K)-jp+j5-} 
1. Light increasing with increase of speed k RJ 
But if the velocity for violet is greater than for red light, N v is greater than N K , 
and hence I B is greater than I v in the even phases, and the return light will appear 
tinged with red. 
r 
2. Light diminishing with increase of speed < 
E 
n 
i 
i 
hi— 2 12(1 K )~\~2 ) 1 — 
E=2{ 2 ( 1 ~ k )+P-1-^} 
Here, on the other hand, if N v be greater than N 1{ , we have I v greater than I B , and 
the return light will be tinged with blue, in the odd phases. 
First measurements of the difference in velocity of red and 'white light. 
While we were quite prepared to examine every possible source of error in these 
new and unexpected conclusions, we considered it to be of first importance to attempt 
to get, even in a rough manner, some actual measurements of the difference in velocity 
of led and blue light, on the assumption that such a difference is the explanation of 
oui results. From the above figure it will be seen that the speed of rotation necessary 
to give equality of lights must always be greater for blue than for red light. It 
is also clear that the difference in speed of rotation for red and for blue light bears 
the same ielation to the absolute speed of rotation for either of those colours as the 
difference m velocity between rays of red and blue light bears to the absolute velocity 
2 N 2 
