454 
DR. W. MANSEEGH VARLEY OX THE PHOTO-ELECTRIC 
Except for the fact that at the higher pressures no true saturation current exists, 
the sets of curves obtained can he simply explained by the ionic theory ol conduction. 
We regard the metal surface illuminated Ijy ultra-violet light in a similar manner to 
a hot surface, and suppose that negative corpuscles escape from the surface under the 
influence of light, and, in the absence of any force tending to drag them further, 
simply diffuse back into the metal. If now the illuminated surface is negatively 
charged, a field is set up and the corpuscles are enabled to escape, and a state of 
equilibrium will be established when the total number of corpuscles given out by the 
metal is equal to the sum of the number finding their way to the positive electrode 
and the numl)er diffusing back into the metal; and, as the field is increased, the latter 
number will become small compared to the former and we shall get a saturation 
curi'ent—neglecting, of course, any complications introduced by the presence of the 
surrounding gas. 
The curves of fig. 4 show that although we never get a saturation current at the 
higher pressures, yet at all pressures the current-E.M.F. curve is less steep as soon as 
a certain potential gradient has been reached. As the potential difference is further 
increased the current increases more slowly, lint uniformly, with the potential difference 
until another critical potential gradient has been reached, beyond which the current 
increases much more rapidly than the potential difference, and this rapid increase 
goes on until an actual s|)ark discharge takes place between the electrodes. From 
hg. 4 we see that in hydrogen at 20 milliins. pressure this latter stage begins when 
the potential gradient is about 80 cells per 3'5 milliins., or nearly 500 volts 
per centimetre. This second rise in the curve has been shown by Professoi' Townsend 
(‘Phil. Mag.,’ 0th series, vol. 1, p. 198, 1901) to be due to the ionisation of the 
molecules of the gas by the corpuscles themselves, and the potential gradient at 
which it liegins is inversely proportional to the ])ressure. 
The fact tliat the current-E.M.F. curves are not horizontal even v'hen we should 
exjiect saturation, and that they become steeper and steeper at their flattest parts as 
the pressure is decreased, seems to indicate that there is ionisation by collision to 
some extent at o-radients much below those at which the second rise in the curve 
leading up to the spark discharge commences. We have not as yet, however, 
succeeded in obtaining a thoroughly satisfactory explanation of this phenomenon. 
According to Paschen’s tables, quoted in Professor J. J. Thomson’s ‘ Ilecent 
Researches,’ the potential gradient necessary for a .spark to pass between jdates 
0'35 centim. apart in air is IIG electrostatic units per centimetre, or 34,890 volts per 
centimetre at 750 iiiillims. pressure, and in liydrogen 0'55 of this, or 1!),200 A'olts per 
centimetre. On tlie supposition, therefore, that ionisation by collision begins at 
potential gradients slightly lower than those necessary for the discharge to occur— 
which appeared to he the case at these moderately low ])ressures, the current-potential 
differeiice curve becoming ra|)idly steeper and steeper until the actual discharge^ 
passed—and that the sparking |)otentlal is inversely proportional to the pressure, 
