184 
Proceedings of the Royal Society 
we find, from the position of the surface CD, a sin D as the dis- 
tance between successive surfaces of similar phase parallel to CD, 
that is to say, as the length of the wave of the light propagated 
in the direction normal to CD. Similarly, by drawing perpen- 
A \ 
diculars upon the successive envelope surfaces through C from 
the first opening to the right, we get for the same wave length 
^ a sin D 2 from the second image, ~ a sin D 3 from the third, and so 
LI O 
on. In the case of white light, the separation into its component 
colours exhibited in each lateral image enables us, by observing 
the deviation of each colour of the spectrum, to measure the wave 
length of light of that colour. 
The lateral images are thus easily accounted for in the imaginary 
case, in which the transparent intervals are of infinitely small 
breadth. Gratings have been constructed by ruling sensibly dark 
lines upon glass so closely that the breadth of the transparent 
interval is only a small fraction of the length of wave. The 
explanation of the images seen through these is the same as that 
just given for the imaginary case. 
Suppose, however, the width of the 
spaces to be so much greater than the 
length of wave, that the small inclined 
surface AC which covers the opening, 
as seen in the direction AP normal 
to AC, stretches obliquely across the 
exact length of a wave of the inci- 
dent light, the surface AC, which 
would be the locus of the same, or at least concordant phases 
of vibration if light were propagated in the direction AP, 
