258 Proceedings of the Royal Society of Edinburgh. [Sess. 
sufficient to give good illumination in the violet. The Nernst filament is, 
of course, unmeltable and seems to be almost ideal for the purpose. I have 
found it best to use those employed for lantern projection work, which 
require to be heated with a Bunsen to make them start. 
II. The first type of flicker spectrophotometer I constructed is shown 
diagrammatically in fig. 1, and is a simple modification of the instrument 
already referred to.* In that instrument a double-image prism was 
employed to bring the two beams accurately edge to edge ; in this case a 
prism of greater power is used, which has the effect of completely super- 
posing them. The two beams X and Y, whether proceeding from two 
different light-sources as in “ comparison spectrophotometry,” or from a 
single light-source as in “ absorption spectrophotometry,” are focussed by 
the lens L x on the collimator slit. After passing in turn through the 
collimator lens L 2 , the dispersing prism P 1 , and the telescope lens L 3 , they 
are brought to a focus at the double-image prism P 2 , and thence pass on as 
one through the nicol N into the small telescope T, and thence to the 
observer’s eye. In the plane of the spectrum immediately in front of P 2 
there is placed, as usual, a shutter with a slit which permits only the 
light of the desired part of the spectrum to pass. On looking into the 
telescope, which it is convenient to focus on a screen pierced with two 
exactly similar apertures for the two beams and placed between L 2 and P 1 , 
the observer sees these two apertures in monochromatic light and super- 
posed on each other. To make a measurement, the opaque disc shown in 
fig. 2 is placed anywhere between L 2 and L 3 , and by its rotation alternately 
cuts off the light from each beam. In these circumstances, unless the two 
beams are equally bright, a strongly marked flickering sensation is produced, 
and this furnishes a delicate criterion of their equality of brightness. The 
relative intensity of the two beams can be altered as required by rotating 
the nicol N, the reading on the circular dial of which, when the flickering 
has ceased, gives the required measurement after the usual “reduction.” I 
find it most convenient to drive the disc by means of a little water turbine, 
as such a motor can be made perfectly steady-running and practically 
noiseless. The water-tap, or a brake on the axle, furnishes a ready 
means of adjusting the speed, and should be arranged to be near the 
observer’s hand, as the best speed depends on such variable factors as the 
colour of the light, the intensity, etc. 
The apparatus thus arranged suffers, like most polarising spectrophoto- 
meters, from the error caused by the partial polarisation in the vertical 
direction imparted to the light by the prism P r This error can be done 
* Loc. cit. 
