IMPARTED TO A VACUUM BY HOT CONDUCTORS. 
521 
saturating potential might become much greater, as is shown by the following curve, 
taken, also with apparatus 7, at a pressure of '02 millim. In this case the current is 
not saturated till a potential of about 280 volts is reached. Duiliig this experiment 
the temperature of the filament, as indicated 
by the defiection produced by the thermo¬ 
couple, was kept constant. 
The bend in the curve at about 20 volts 
seems to indicate that a sort of saturation 
occurs here. The subsequent increase of 
current would then be exjfiained by the ions 
produced by collisions as the electromotive 
force was increased. On this supposition, 
when we again reach the flat part of the 
curve at 280 volts we must sup230se that 
every collision jDossible at this pressure 23 ro- 
duces ions. Similar considerations exjfiain 
Fig. 9. 
120 
the gradual slojDe of the curve in fig. 8 
aftei saturation. O’wing to the peculiar shape of the electrodes it was not jDossible 
to calculate the magnitude of the effects. 
With this curve it is interesting to comiDare one obtained at a slightly lower 
pressure ('013 millim.) with the other form of aj^paratus. In this case the current 
used to heat the filament was kept constant while its resistance decreased in the ratio 
of 1‘025 to 1 during the observations. This does 
not imply that the temperature altered, since 
heating a carbon filament steadily decreases its 
resistance when cooled and measured again at the 
original temj^erature. The absolute value of the 
current is also some twenty times as great as that 
in the preceding curve. It will be seen that in 
this case saturation was reached with about 160 
volts. 
It is evident that in this case the bend at 20 
volts does not a 2 )pear. This may be due to the 
gi’eater magnitude of the current and smaller 
pi-essure, which makes the Townsend effect less 
frig_ 10. comparison. With still higher joressures the 
current for low voltages is small and increases 
moie lapidly as the voltage is raised than it would if it were proportional to the 
potential difference. The curve then joasses through a singular ] 3 oiut, the current 
increasing less and less raj^idly with the voltage until the saturation value is reached. 
The current afterwaids remains stationary for some time until it begins to increase 
3 X 
VOL. CCI.—A. 
