VELOCITY OE WHITE AND OE COLOURED LIGHT. 
239 
The smallness of the correction due to the term-in our experiments can be seen 
9 + P 
by examining the details of the observations. 
In our experiments c/=yf almost exactly. Most of our measurements were made 
with the 12th equality. Some how'ever were made with the 13th equality. We will 
now examine the formula for this case. 
9. Let us consider the case of the r + 2] th equality, r having the same value as before. 
Here B is increasing, A is decreasing. We have y> A =r+3, p =r-\ j-2. 
I"a=I"b Or ^{ 2 ( 1 - K) +(r+3)- 1 -g}=^{2(l- K ')-(r+2 ) + U} 
Since E" B still =pE" A , this becomes 
rn" f 
2(l- K )+r+2--=p^2(l- K ')-(r+2)+ 
n 
w±. 
But at the r— l| th equality 
tyi 
2(!— ac) —(r+2)+— =p —a c)+r— 
n 
9$ A 
subtracting we have 
2 (r+2 ) + 2 p (r +1) = 1 + ^ K+ n") + n") 
r+2 
If now D a and D B be so related that g=~ —p this equation becomes 
2(r+l)(g+p )=^(«'+«") 
n + n" , TT 2 m(n'-\-n")D B 
R =~—— and \ =— -—^— 
2(r+ 1) 
If however the true value of N B is 
N„=W (j * ■ 
2(r +1) \ g + p (g + pY 
To approximate to the value of g-\-p on the assumption that k=k=^ we notice 
that the above equations reduce to 
