455 
DISCHARGE FROM METALLIC SURFACES IN DIFFERENT GASES. 
ionisation by collision should commence in hydrogen at 20 millims. pressure at a 
potential gradient somewhat less than 520 volts per centimetre, which is in close 
agreement with the value taken from the corresponding curve in fig. 4, and given 
above, viz., 500 volts per centimetre. 
When the pressure is decreased below a few tenths of a millimetre, the character 
of the curves alters completely, and we now begin to get perfect saturation currents. 
This critical pressure corresponds to that at which the mean free path of the ions 
becomes comparable with the distance separating the electrodes. As soon as this is 
the case, the number of molecules which can be struck by the corpuscles sliot ofi‘ from 
the illuminated electrode becomes very limited and a small potential gradient will he 
sufficient to cause saturation. Further decreasing the pressure will decrease this 
saturation current, until at very low pressures the whole current will he practically 
due to the corpuscles shot ofi: from the cathode, as only a very small fraction of these 
wdl strike a molecule in the course of their journey from one electrode to the other. 
The curves in fig. 5 show most clearly the variation of the currents at these low 
pressures. We see that there is a sudden change in the curves between pressures of 
0-2 and 0-05 millim., and we know from other considerations that the mean free path 
of the corpuscle in air is of the same order as the distance between the electrodes 
(0-35 centim.) at about these pressures.^ 
The curves at the very lowest pressures agree with Lenard’s result (‘ Ann. der 
Physik/ vol. 2, p. 359, 1900) for cathode rays produced by ultra-violet light, that in 
a vacuum a very large change in the potential difference produces no change in the 
current. 
The curves of fig. 0, obtained ^¥ith a platinum cathode in air, present no peculiar 
features, being similar to those obtained with the zinc cathode. The fact that the 
electrodes in this case were only 2 millims. apart introduces slight modifications, the 
sudden change in the curves which occurs when the mean free path of the ion and 
the distance apart of the electrode become comparable, appearing at a liigher pressure 
than before when the electrodes- were 3'5 centims. apart, as we should naturally 
expect. 
The curves given in figs. 7 to 10, showing the relations between tlie photo-electric 
currents in the three gases air, hydrogen, and carbon dioxide, all show that at the 
higher pressures the current is greater in carbon dioxide than in air, and greater 
again in air than in hydrogen for all potential gradients up to those at which the 
genesis of ions by collision begins. They also show that this stage begins in 
hydrogen at a much lower potential gradient than in air or carbon dioxide, the 
potential gradient required to cause ionisation by collision in these two gases being 
approximately the same. This is what we should expect, on the supposition 
* Townsend, in ‘Phil. Mag.,’ February, 1901, p. 224, gives the mean free path of an ion in air as 
4'3 X 10 ceiitim. at 760 millims. jmessure, corresjronding to a mean free path of O'34 centim. at 
O'10 millim. pressure. 
