Electricity from a Nernst Filament. 257 
strongest fields at my disposal (900 lines per sq. cm.), the 
filament being charged to +200 volts. 
Thus the carriers of the positive electricity discharged from 
a Nernst filament at low pressures are particles much heavier 
than corpuscles. Inall probability they are bodies of molecular 
dimensions. 
In connexion with the magnetic deflexion of the negative 
leak from the filament, it remains to add that I have found it of 
the utmost importance to surround the filament and electrodes 
by an earth-connected metal tube. If this be not done, the 
walls of the glass tube get charged up, and, when dealing 
with large currents, this may prevent the magnetic field from 
producing any diminution in the leak. 
The author hopes at an early date to make some experiments 
on the discharge from a Nernst filament in different gases. 
{t will be interesting to see if an increase in the leak takes 
place when hydrogen is substituted for air, corresponding to 
what happens in the case of platinum. 
Summary of the principal results. 
(1) From a glowing Nernst filament there is at all pres- 
sures a discharge both of positive and negative electricity. 
(2) The negative leak behaves in a perfectly regular way, 
diminishing only slowly with the time, and is not liable to 
those sudden variatious which occur in the case of the negative 
leak from platinum. During the whole course of the expe- 
riments the negative leak did not suddenly change without 
any apparent cause more than three or four times. 
(3) The negative leak at certain pressures shows the effect 
of ionization by collision. 
(4) The negative leak when no ionisation by collision occurs 
increases as the pressure is diminished until a pressure of a 
few millimetres is reached. It then remains constant down 
to a low vacuum. 
(5) The negative leak increases rapidly with the temperature 
of the filament. 
(6) At low pressures for a considerable range the negative 
leak under a high potential is proportional to the pressure. 
(7) From a filament which has not been heated before, there 
is a large initial positive leak which rapidly dies away. After 
long continued heating of the filament the positive leak 
becomes fairly steady, but still continues to diminish with the 
time. 
(8) The positive leak increases with the temperature of the 
filament, but not so rapidly as the negative leak. 
(9) The carriers of the negative leak in a vacuum are 
Phil. Mag. 8. 6. Vol. 8. No. 44. Aug. 1904. S 
