548 
Proceedings of the Royal Society of Edinburgh. [Sess. 
beam of light. The one beam under comparison must, of course, go through 
the cell with the solution, and its intensity is diminished ; we must have a 
means of diminishing the intensity of the other beam to the same value, and 
at the same time of letting us know in what ratio it is diminished. Nicol’s 
prisms are of no use for this purpose, for Iceland spar and Canada balsam 
absorb all wave-lengths below ’300 ja. Neither can a micrometer slit be 
used as in Vierordt’s spectrophotometer. The brightness of a continuous 
spectrum can, it is true, be diminished by diminishing the width of a slit, 
but there are no continuous spectra below *300/*. In this region the only 
possible source of light is the iron arc or the electric spark, and these sources 
give line spectra. And if we have two line spectra, one above the other, 
and gradually close the one slit, we only diminish the width of the lines, 
not their brightness. A rotating sector cannot be used, because it does 
not diminish the intensity of the beam, but only the time during which it 
acts. The effect is the same as far as the eye is concerned, but it is different 
with a photographic plate. . Twice the intensity for half the time does not 
produce the same density. 
We were therefore compelled to fall back upon the inverse square law, 
and in order to study its possibilities a preliminary experiment was made 
with a spectrograph with glass prism and lenses. A ground glass plate was 
fixed permanently 14 cm. in front of the slit. This ground glass plate was 
large enough for the rays from it to fill the full aperture of the collimator. 
A small inverted incandescent mantle was mounted so that it could be 
moved back and forward in the line of the collimator, but always at such a 
distance in front of the ground glass that the intensity of illumination on 
the latter could be regarded as inversely proportional to the square of their 
relative distance. Draw-slides were placed across the upper and lower 
halves of the slit. A glass cell with the solution to be investigated was 
placed before the draw -slides, between the latter and the ground glass. 
The incandescent mantle was placed at a standard position near the ground 
glass, one-half of the slit was opened, and an exposure made. The cell 
was then replaced by a similar cell filled with water, the relative distance of 
incandescent mantle and ground glass increased — say doubled — and another 
exposure made for exactly the same time, the other half of the slit being 
used. If we assume, now, that the intensity of the incandescent mantle 
remains unaltered, the intensity of illumination in the second case is 
only one-quarter what it was in the first case. If, therefore, we seek out 
those wave-lengths for which the density on the adjacent spectra is equal, 
we obtain the wave-lengths for which one-quarter of the incident light 
is transmitted through the solution, i.e. for which lCT Acc * — Hence Acd. 
