UNDER THE INFLUENCE OF CHEMICAL ACTION. 
33 
On referring to fig. 12 it may be noticed that the values taken the following day and 
shown thus © on that diagram, are not in good agreement with the crosses on this 
part of the curve. However, they are nothing like so reliable. The reading at + 0 T2 
volt is only 15 per cent, of the electrometer drift; that at — 0 T2, 37 per cent.; that at 
— 0-42, 80 per cent. ; and it is not until the comparatively large relative current at 
— 1-0 volt is reached that the accuracy becomes better than that of the worst point 
shown in fig. 16. 
§ 8 .—The Effect of the Different Gases on the Photo-electric Threshold Frequency. 
In the one case for which this frequency has been closely estimated (COCb) we have 
seen that it was very near to X 4900 (6-02 X 10 14 < v 0 <6-25 X 10 14 ). This Va is 
not to be regarded as a reliable constant, but one which varies with apparently trifling 
changes in the conditions. Thus in one C0C1 2 experiment no photo-electric current 
could be got with the full light of the mercury lamp through the blue filter (mostly 
X 4355). The chemical effect was working quite well and there was nothing otherwise 
to differentiate this experiment from others which gave quite large photo-electric currents 
with X 4355. With this exception, so far as I can recollect or tell from the record, 
measurable photo-electric currents were always obtainable with X 4355 both in COCl 2 , 
chlorine and water vapour. With COCl 2 it is probable that the threshold frequency was 
always higher than that of the green line X 5460 (frequency 5 *49 X 10 14 ). At any rate, 
tests made from time to time with the green filter never gave any deflections which 
could be relied on to mean anything. As regards chlorine, all that can be said is that 
the threshold frequency was less than that of X 4355 (6-89 X 10 14 ) in all the experiments 
here referred to. In some of the tests with water vapour present it was found that 
not only X 4355, but also the green line X 5460 and the yellow line X 5769 — X 5790 
gave photo-electric currents. The same was true of the light from a photographic 
dark-room lamp, and photo-electric currents were also obtained with the radiation from 
an incandescent lamp filtered through a solution of iodine in carbon disulphide of such 
strength as to make the lamp quite invisible through it. Evidently when water vapour 
is present the threshold frequency can lie in the infra-red part of the spectrum. 
§ 9. —Analysis of the Curves. 
The obvious way of attacking the problem of the distribution of kinetic energy among 
the emitted electrons is to take the curves, such as figs. 4, 7 and 12, and find the incre¬ 
ments in current corresponding to equally spaced intervals dY along the voltage axis. 
These will be proportional to the number of electrons whose energies lie between eV 
and e (V + dY). This method will certainly give correctly, to the degree of accuracy 
within which it can be operated and to the accuracy within which V is known, the 
distribution of velocity among the electrons as they in fact reach the receiving electrode. 
VOL. ccxxn.—A. F 
