EMISSION VELOCITIES OF PHOTO-ELECTRONS. 
223 
with an amount of energy e . hi and that e . Vf, represents the loss of energy by the 
time it has emerged from the surface. We may regard this e. V(, as being made up 
of two parts, . e the loss of energy in passing out of the molecule and Vg. e the loss 
of energy in passing through the retarding surface layer. It is very probable that Vg 
is zero for surfaces prepared by distillation of metals in vacuo. The effect of admitting 
oxygen for a short time to the apparatus, when the disc was covered with a new 
surface of cadmium, was to increase Vg from 0 to about '3 volt. A previous research* 
showed that oxygen was effective in reducing the emission velocities from Cd and Zn 
and other metals while hydrogen was ineffective. This is in agreement with a view 
advanced by Sir Oliver Lodge and others, that oxygen is mainly responsible for the 
contact potential. On this theory, however, the oxygen layer is positive with respect 
to the metal, while to account for the reduction in the velocity the layer should be 
negatively charged. 
VjC may be identified with the work required to take an electron away from a 
molecule. Unless the quantity hi exceeds V^, light is incapable of producing a 
separation of the electron from the molecule. If the Vq given by these experiments 
is equal to Vj, then the smallest frequency capable of exciting the photo-electric 
effect should be identical with the smallest frequency capable of producing ionisation 
in the vapour of the metal. An assumption is made here that ionisation in gases by 
light is essentially the same thing as the photo-electric effect. 
An inspection of the values of Vq in Table V. shows that the work of withdrawing 
an electron from a molecule is least for calcium, the most electropositive element in 
the table, and greatest for the more electronegative elements. 
(vii.) We may apply these principles to determine the work required to ionise a 
molecule of oxygen. From my experimentsf on the ionisation of the air by ultra¬ 
violet light, I concluded that air (or more probably the oxygen in it) is only ionised to 
an appreciable extent when light of wave-length shorter than about X 1350 is absorbed 
in it. An examination of the results given in Palmer’s paper| justifies this estimate. 
What value shall we give to k for oxygen ? Fortunately the values of k do not alter 
much, so a mean may be taken without much error. The mean of all the Fs (omitting 
k for the electropositive calcium) is 3’62 x 10“A The mean k for Bi and Sb is 3'66 x 10“^b 
and these elements resemble oxygen somewhat more than the other elements. The 
value of Vj (= kn) works out to be 8'0 volts. Dember§ finds that the work required 
to ionise a molecule of air is 1'26 x 10“^^^erg. Dividing this by e = 4'65 x 10“^“ we get 
the ionising potential to be 8T volts. 
(viii.) When zinc is in combination with chlorine, which has a strong attraction 
for the electron, experiment shows that light of wave-length X 1849 is incapable of 
* Hughes, ‘Proc. Camb. Phil. Soc.,’ XVL, p. 167, 1911. 
t Hughes, ‘Proc. Garni). Phil. Soc.,’ XV., p. 482, 1910. 
I Palmer, ‘Phys. Rev.,’ XXXIL, p. 1, 1911. 
I Franck and Hertz, ‘Verb, d. D. Phys. Ges.,’ p. 967, 1911. 
