522 
MR. 0. W. RICHARDSON ON THE ELECTRICAL CONDUCTIVITY 
Fig. 11. 
again as the voltage is raised. This increase is doubtless due to new ions produced 
by the collisions of the negative ions, since it is in all respects similar to the effects 
described by Professor Townsend.* All these characteristics are shown l)y the upper 
curve in fig. 11, which was obtained with apj)a- 
ratus 6. In this case the pressure increased from 
•65 to '85 niillim. during the experiment; the 
heating; current was maintained constant and 
<D 
equal to 1'07 amperes. 
When the jDressure of the gas is raised further, 
the current at low voltages becomes still smaller 
than before, the current E.M.F. curve being 
always concave to the axis of current. In this 
case the current never becomes saturated, owing 
to the collision effect coming in before the 
saturating potential is reached. Under these 
circumstances the current increases more and 
more rapidly with the potential as the latter is 
raised. These characteristics are very well shown by the lower curve in fig. 11, 
which was taken with apparatus 6 at a pressure of 2 millims. The filament was 
heated by a constant current. 
It will be noticed that several of these curves are very similar to those obtained 
with hot platinum wires by Professor McClelland.! 
In all these cases it was found that in retracing the observations backwards, the 
curves never quite coincided with those obtained first. These effects, which were of 
the nature of hysteresis, were partly attributable to change in the conditions while 
the observations were being made. Such changes were, for example, increase of 
pressure due to gas given off fi om the walls, change in the temperature of the carbon 
heated-by a constant current owing to the permanent alteration of the resistance of 
a cai’hon filament produced by heating, &c. Even when such disturbances were 
eliminated as far as possible, the curve could never be made to return on its outAvard 
path. The form of the curve was always the same, but the value of the current on 
the return curve was invariably smaller than on the outward one ; in a particular 
case, when the pressure increased from 2‘4 to 2‘6 millims., the current with about 200 
volts was reduced to one-third its value on the return journey. 
After having investigated in some detail the connection between the current and 
the electromotive force when the filament was maintained at a constant temperature, 
the connection between the saturation current and the other conditions was next 
examined. The experiments to be described are therefore concerned Avith the 
relation betAAmen the saturation current and 
* Townsend, ‘Phil. Mag.,’ Feb., 1902. 
t ‘ Camb. Phil. Proc.,’ vol. 11, p. 296. 
