189 
of Edinburgh, Session 1869-70. 
Different measurements pay be got by the same observer at dif- 
ferent times from his uncertainty as to the points in the spectrum 
at which he should consider each colour to begin and end. This 
uncertainty is usually considered to be obviated by using solar 
light, and measuring the deviations of the dark lines in the spec- 
trum ; but as these lines are the parts of the spectrum from which 
no light comes, the process can scarcely be called the determina- 
tion of the wave length of light. 
Since the above measurements were made, an improvement was 
made in the apparatus by which the gratings were constructed, and 
finer gratings were made, which gave more brilliant spectra, by 
reason of the greater number of apertures from which similar 
phases of vibration came to the eye. With one of these, consisting 
of transparent spaces drawn at the rate of 3000 to the inch, a new 
set of measurements was taken in the following manner 
EF represents a rectangular piece of 
wood upon which is pinned a piece of 
paste-board ABCD, whose edge AFB ^ 
is an arc of radius 20 inches and 
centre at E. The chord AB is divided 
into tenths of an inch by perpendicu- 
lars to it meeting the arc. Touching 
the arc are placed, but not fixed, two 
pieces of tin bent as represented at Gr, 
each having a narrow slit so situated 
that the bottom of the one slit is on a 
level with the top of the other, and 
carrying a small piece of candle im- 
mediately behind the slit. The grat- 
ing is held at E, and the pieces of tin 
are moved along the arc until the 
colour observed in each spectrum is in 
the same vertical line at F. The dis- 
tance between the two slits is then 
read upon the graduated chord, and 
the half of that distance divided by 
20 inches is the sine of the deviation. In this case the second 
spectrum from each light was observed, and the observed dis- 
2 b 
VOL. VII. 
