54G 
MR. 0. W. RICHARDSON ON THE ELECTRICAL CONDUCTIVITY 
was let into the apparatus and the current again measured under atmospheric 
pressure at the same temperature. It was found that there was no detectable leak 
till a temperature of 60° higher was reached, when one division of the galvanometer 
scale (3 X 10~® ampere) was obtained with the wire at a potential of — 200 volts. 
The small value of the currents at atmospheric pressure is probably due to the 
difficulty of saturating them. 
It was thought conceivable that the ionisation at low pressures might he due to 
the gas molecules hitting the hot wire and becoming ionised thereby. If we assume 
that the maximum current would correspond to each molecule producing one ion, we 
can calculate its value in any given case. If we take the number of molecules in a 
cubic centimetre of gas at 0° C. and 760 millims. to be 2 X 10^^, then the number 
which hit unit area of the wire per second is X 2 X 10^^ approximately, where u 
(the scpiare root of the mean velocity square) may be taken as 5 X 10^' centims. per 
second for air. The number which strike unit area of the wire per second at 
1 millim. pressure is therefore 2’2 X 10'*^, which gives a saturation current of 
14'3 X 10^° electrostatic units, or 47‘3 ampk’es per square centimetre. At a pressure 
•0016 this current would become ’OS ampere per square centimetre. As a matter of 
fact, during the experiments, a current of 2'0 amperes per square centimetre was 
obtained at •0016 millim. pressure. This is twenty-five times the maximum value 
oljtained by supposing each molecule to produce one ion; so that it is highly 
improbable that any considerable part of the conductivity investigated is due to ions 
produced in tliis way. 
Another way of considering this question is to calculate the number of times each 
molecule of air inside the cylindrical electrode must collide with the filament per 
second to produce the ol:)served current, assuming that each collision sets free one 
cor])uscle. In the experiments in question the cylinder had a volume of about 
I cul). centim., so that each molecule present would have to pass backwards and 
forwards between the filament and the cylinder some 10^ times each second. This 
seems to he an impossible feat for an uncharged molecule. 
Both these points of view lead to the conclusion tliat the corpuscles are not 
produced by a dynamical action between the molecules of the surrounding gas and 
the surface of the metal. In fact, all the experimental results seem to point to the 
view that the corpuscles are produced from the metal l)y a process similar to 
evaporation. The effect of the surrounding gas, of impurities in the wire, and of its 
previous history are to Ije regarded as due to alterations in the property of the metal 
wliich corresponds to latent heat in the theoiy of evaporation. 
§ 4. The Edison Effect. 
It will readily be seen that the results which have l)een obtained furnish a 
complete explanation of the phenomenon known as the Edison effect. Edtsox first 
discovered this eftect 1)y connecting an insulated electrode, which was symmetrically 
