1910-11.] Absorption of Light by Inorganic Salts. No. I. 523 
and scale at one metre my thermopile and galvanometer gave a throw 
of 06 cm. per sq. mm. of sensitive surface. When the area of the surface 
that can be utilised is taken into consideration, this is six times as sensitive 
as Boys’ radio-micrometer. 
The spectroscope, thermopile, and galvanometer were the same as I 
used for determining the efficiency of metallic filament lamps,* only the 
spectroscope was fitted with a new and better prism. It was calibrated 
in the same way as formerly, by bringing the thermopile into coincidence 
with various spectral lines in the visible spectrum and by using the 
absorption maxima of water in the infra-red at *966/*, T500yu, and 1'956/x, 
and the minimum at 1'708/x as determined by Aschkinass.-j* The spectroscope 
had glass lenses and a glass prism, but as only aqueous solutions were used 
and as water absorbs before glass does, glass was quite satisfactory. 
A straight Nernst filament was used as source of light, and at first 
it was focussed directly on the slit of the spectroscope by a lens. Two 
similar glass cells were taken, and one was filled with water and the 
other with the solution to be investigated. The cell filled with water 
was placed in front of the slit, a wooden screen placed before the cell, and 
the zero noted. The screen was then removed, the light allowed to fall 
on the slit, and the reading noted. The screen was then replaced and the 
zero again noted. The zero was always drifting, and the deflection was 
calculated from the mean of the readings before and after the deflection. 
The cell with the water was then replaced by the cell with the solution 
and the deflection obtained for the latter. The second reading, divided 
by the first, then gave 10 -Ac L 
This method was, however, unsatisfactory on account of stray heat. 
Radiant heat, of course, is to a certain extent reflected from both lenses 
and scattered from the surfaces of the prism, and consequently the spectrum 
is not quite pure. As the deflections at one part of the spectrum used 
are fully twenty times the deflections at another part, diffuse heat from 
the first part may appreciably alter the reading at the second part. 
Attempts were made to eliminate the effects of this stray heat with water 
filters, ferrous ammonium sulphate filters, and six different kinds of coloured 
glass. Also screens were tried at different places inside the instrument. 
But all these attempts were unsuccessful. 
I eventually removed this cause of error in a satisfactory way by 
resolving the light spectrally before it reached the slit, and this method 
* “ The Efficiency of Metallic Filament Lamps,” Proc. Roy. Soc. Edin., xxx. p. 555 
(1910). 
t E. Aschkinass, Wied. Ann., lv. p. 401 (1895). 
