Metallic Vapours in an Exhausted Space. 103 
walls of the tube and from the anode (this being supposed to 
be made of a solid conductor such as graphite, iron, silicon, &c.). 
The ionized vapour emitted by the cathode (the are still 
running in AB) begins to propagate slowly along the tube; 
and when it reaches the anode and the vacuum is improved by 
the continued action of the pump, the are will start up, go out 
again, and repeat this many times, until the greater part of 
the gases has been liberated from the anode. ‘The arc is 
then allowed to run for some time (it is best to open the side- 
branch are when the main arc has become stable), the tube 
being constantly connected to the pump; and when the 
mercury in the pump, by its hammering sound, indicates that 
only traces of gas are given off, the tube is ready for the 
experiment. It is difficult, if not impossible, to get rid of all 
the gases occluded in the anode and on the walls of the vessel. 
With a little experience it is, however, easy to recognize the 
moment when it is safe to seal the lamp off the pump, and 
have a lamp which will run for many hundred hours without 
a serious impairment of vacuum. 
Another factor must be taken into consideration while 
carrying out the experiments. The presence of inert, non- 
conductive mercury vapour acts in much the same way as 
the presence of foreign gases, so that the starting is easier 
when the tube is cold than when it is hot. If, after the arc 
had been running for some time, the current is interrupted 
and the starting attempted again, three intervals can be 
distinguished. In the first one, which lasts a few minutes, 
the startingis easy. In thesecond one, which may last from a 
few minutes up to a few hours, the starting is not instantaneous. 
In the third one, when the lamp is more or less cooled down, the 
arc starts again instantaneously. The length of the first and 
second intervals depends upon the length of the tube, the 
magnitude of the impressed voltage, and the care with which 
exhaustion has been carried out before sealing off the lamp: 
the longer the lamp, the smaller the impressed voltage, and 
the lower the vacuum the shorter is the first and the longer the 
second interval. With very long tubes of 50-60 ins., and an 
applied voltage of only 120 volts, it is often necessary to 
let the tube cool down for two or tkree hours before the 
starting becomes instantaneous. It is possible that this 
behaviour of the tube is not so much due to the presence of 
mercury vapour as to the presence of traces of foreign gases 
driven out by the heat of the are and reabsorbed by the walls 
of the tube after they had become cooled down. Still, the 
phenomenon is to be observed with long tubes even after very 
careful exhaustion and while the tube is connected with the 
