552 Proceedings of the Royal Society of Edinburgh. [Sess. 
longer exposures X ==- 223//. was reached. Spark lines were obtained down 
to 202//, (Zn 29). The length of the iron spectrum on the plate from 
'233//, to about '530//, was about 4 cm. Eleven exposures could be 
obtained on one plate. 
The method of working was as follows : The iron-arc spectrum was 
first of all calibrated by means of the cadmium-zinc spark and a table 
drawn up by which the wave-length of any portion of it could be read. 
The ratio of the light transmitted by the mesh to the light transmitted by 
the quartz cell filled with water and placed before the lower face of the 
prism was next determined from the positions of the quartz plates for 
equal density on the photographic plate. This ratio varied slightly 
throughout the spectrum, but the latter could be conveniently divided into 
two regions, for each of which the ratio could be taken as constant. Then 
certain standard positions of the quartz plates and B 2 were determined 
which gave equally spaced values of A cd and therefore of A, A cd being of 
course defined by the equation I — I 0 10~ Acd . The quartz cell was next filled 
with the solution, the absorption of which was to be measured, placed in 
position, and exposures were made for the different standard positions of the 
quartz plates. The plate was then developed and examined under a micro- 
scope of low power. The points at which the two comparison spectra 
matched were thus determined. The table then gave the wave-lengths, 
and the positions of the quartz plates gave the corresponding values of A cd. 
Owing to the sharp edge P of the quartz prism PQRT (fig. 2) the compari- 
son spectra touched one another very sharply, and it was thus possible to 
determine very accurately the positions at which the intensities matched. 
One exposure sometimes gave A for as many as five points, thus indicating 
at least two maxima and two minima in the curve representing A as a 
function of X. 
Only one quartz cell was used, its inside thickness being 1 cm., and 
for most of the salts more than one strength of solution had to be used 
in order to bring; A within the range of the instrument for all wave- 
lengths. The results obtained are tabulated below. In each case the 
first column gives the concentration of the salt used in gramme-molecules 
per litre, the second gives the wave-lengths, and the third gives the corre- 
sponding values of A. 
