UNDEK THE INFLUENCE OF CHEMICAL ACTION. 
41 
set by the photo-electric currents and shown at A. The right-hand limit could not be 
determined at this stage. 
Another estimate of the value of T for chlorine can be got by assuming that for COCl 2 
and comparing the widths of corresponding parts of curves such as I. and II. in fig. 17. 
Keeping to the rapidly falling parts of these curves, which are comparatively free from 
errors due to the uncertain zero and to the disturbing causes in its neighbourhood, I 
find that the width from the 66§ per cent, point to the 4 per cent, point, is for chlorine, 
1 *387 volts, and for C0C1 2 , 0-840 volt. Taking the value of T for COCl 2 to be 3300, 
this would give the value for chlorine as 3300 X -= 5450. This is considerably 
higher than the value got from a study of fig. 21, but it will be remembered that the 
corresponding operation on fig. 20 also gave a low value for COCl 2 . Under the circum¬ 
stances probably the best estimate that we can make for chlorine is the mean of the 
two values, viz., 4900°K. 
In each case the position of the true zero given by assuming the Maxwell distribution 
falls within the limits set by the direct photo-electric measurements. If we assume 
that the position of this zero is correctly fixed in this way, we can get another check 
on the relative values of T. The displacement in volts from the zero to the place where 
the current is a fixed small fraction of the saturation value, small enough to avoid the 
disturbances in the neighbourhood of zero volts, should be in the proportion of the 
respective values of T. The three sets of data give the following residts, using 0 -05 
as the value of the small fraction :— 
COCl 2 Cylindrical anode —- 
zero . . . . = 1 -84 
5 per cent. . . = 0-52J 
COOL Spherical anode — 
zero . . . 
5 per cent. . 
Cl 2 Spherical anode — 
zero . . . . = 3 -11 
5 per cent. . . = 1 -36 J 
>Dk 
Displacement =1-32 volts. 
= 0-40| 
^ ^ >Displacement = 1 -20 volts. 
1 
Displacement =1-75 volts. 
The numbers 1 -32, 1 -20 and 1-75 are in the proportion 3300, 3000 and 4400, and 
are therefore in agreement with the values already obtained. 
§ 10.— Thermo-chemical Considerations. 
An upper limit to the value of T can be obtained from thermo-chemical data. From 
£ Recueil de Constantes Physiques,’ pp. 333 and 339, it appears that the heats of certain 
relevant chemical reactions are as follows :— 
VOL. CCXXII.—-A. 
G 
