Vol. 8, 1922 
PHYSICS: R. C. WILLIAMSON 
259 
of ionization by collision for the higher accelerating potentials. Hence 
we may conclude that the positive ionization is due to radiation between 
3100 A and about 1850 A (probably the shortest wave-lengths present in 
the arc spectrum), and also that under the conditions of the experiment 
ionization by collision was of a lower order of magnitude than that due to 
radiation. 
If the vapor jet is cut off by means of the shutter under conditions as 
shown in figure 1, approximately the same effect is produced on the current 
as by inserting the screen in the path of the hght. This confirms the 
view that the negative residual current is due to photoelectric leak, and, 
furthermore, indicates that the scattered light which produces this leak 
comes largely from the diaphragming system and is not due to any marked 
scattering or resonance by the vapor. 
In preliminary tests of the ionizing power of the mercury arc spectrum, 
as a function of wave-length, the retarding and accelerating potentials were 
set at one and ten volts respectively. Absorbing screens were successively 
inserted in the beam of radiation, and the observed currents were plotted 
against the lower transmission limits of the screens, figure 2. Curve abc 
represents the effect of the screens on the photoelectric current from 
surfaces due to stray light, being taken with no vapor present; a'h'c^ 
and a"h"c" were taken with vapor issuing from . the nozzle. No screens 
for the region below 2900 A were available at the time abc and a'b'c' were 
taken. These curves show a positive current superposed upon a negative 
photoelectric surface current. This positive current appears at about 
2900 A and increases quite rapidly as the wave-lengths decrease. The 
ionization does not seem to be associated with resonance phenomena or the 
absorption of a single wave-length in the vicinity of 2900 A. Rather it 
appears to be a continuous function of the wave-length, starting near the 
convergence wave-length of the principal series and increasing quite rapidly 
as the wave-length decreases, giving a continuous absorption of energy 
below this limit. It is of interest to note, in this connection, the region of 
continuous absorption described by Wood^ in the case of sodium vapor 
which extends from the convergence frequency of the principal doublet 
series on down in the ultra-violet, thus corresponding with the region of 
ionization and consequent absorption of radiation observed for potassium. 
The above results can be summarized as follows: 
It is believed that definite evidence has been obtained of the ionization 
of metallic vapors by light. 
The results indicate that there is a long-wave limit in the case of this 
ionization which can be calculated from the relation Ve = hv by means 
of the ionizing potential. 
The ionizing power of the radiation is a continuous function of the 
wave-length below this limit, increasing as the wave-length decreases. 
