792 
in the image plane is determined, the theory can be tested. As it 
appeared in preliminary observations that the temperature had 
influence on the position and the distribution of intensity *), care 
was taken that the surroundings of the echelon remained at constant 
temperature during the measurements. The measurements have been 
carried out with regard to three components of the green mercury 
line (A = 546,1 uu). 
2. The determination of the intensity. The intensity was determined 
by a photographic method. The method used, the description of 
which follows here, is analogous to that which Miss R. Riw1n *) 
applied for the determination of the absorption. 
In order to prevent complications in consequence of difference in 
time of exposure, kind of plate, development etc., all the spectra 
belonging together were photographed on one plate with the same 
time of exposure’). The blackening of the plate then depends 
exclusively on the intensity of the incident light. When the functional 
relation between blackening and intensity (curve of blackening) is 
known, the second quantity can be found from the first. As the 
components of the system of lines examined by us have only a very 
small difference in wave-length, a curve of blackening need be 
constructed only for one wave-length. 
To find this curve the following course was taken. The spectrum 
of the green mercury line was photographed with a definite position 
of the echelon. Then different light-reducers *), which weaken the 
light of the green mercury line in known ratio, were successively 
placed before the slit of the collimator, and with the same position 
of the echelon the spectrum was repeatedly photographed. 
To gauge the light-reducer, the light of the mercury lamp (Wes- 
TINGHOUSE CoopErR-Hrwitt, 220 V., 3.5 A) is concentrated by con- 
densers on a surface thermopile of Morr, which was connected with 
a galvanometer of Morr. Filters ensured that only the light of the 
wave-length 546,1 uu fell on the thermopile. The reducers were 
placed immediately before the thermopile. The ratio in which the 
light is weakened is found by division of the deviation of the gal- 
1) Phys. Zeitschr., 21, 16, 1920. 
2) These Proc. Vol. 23, p. 807. 
3) Paget, Orthochromatic, Extra Special Rapid plates were used. The development 
took place for about 10 minutes with a glycin developer. 
4) For reducing the light solutions of chromealum were used in different concen- 
trations in air tight vessels. To prevent turbidity a litlle sublimate was added. The 
vessels were gauged anew a few times with an interval of some weeks. No change 
could be perceived in the absorption. 
