102 Dr. E. Weintraub on the Arc in 
closing the switch s a few minutes pass before an are is 
established between A and D. It seems, therefore, that the 
presence of ionized vapour in itself is not sufficient to start 
the ionization at the cathode, but that it facilitates the action 
of the other causes. 
§ 2. 
Role of Vacuum. Function of the Carbon Filament. 
The phenomena described in the first section, especially 
that of the instantaneous starting of a long are, take place 
only when the vacuum in the lamp is as hi oh as can be 
reached by careful exhaustion with a good mercury pump. 
Measurements by means of a Mcleod gauge have shown 
that when the pressure of foreign gases (air, hy drogen, 
&c.) exceeds about °01 mm., the starting of the main arc in 
‘the arrangement shown by fig. 2 is not instantaneous, but 
takes some time; and the ionized vapour, coming from the 
cathode of the auxiliary arc, is seen to creep up slowly i in the 
tube until it reaches the anode, and then eventually causes 
the starting of the arc. If the vacuum becomes lower than 
the limit given above, the arc starts with difficulty, or not 
at all. As the vacuum becomes poorer and poorer the pheno- 
mena become more complicated, and the regularities observed 
in a high vacuum become obscured. In order, therefore, 
that the experiments described in Section 1 should succeed, 
the vacuum must be high and the gases from the walls and 
from the electrodes must have been driven off. In repeating 
the experiments of Section 1 it would therefore be advisable 
to begin with short tubes (5-6 ins.), as the reproducing of the 
phenomena i in long tubes requires some experience and skill, 
which can be eained only by actual work on the subject. 
The experiment relating to the starting of an arc by means 
of a side-branch may therefore be described in detail. After 
the tube has been exhausted on a good Sprengel pump, the 
mercury that is to serve as electrode is poured in, or better 
distilled in, from an attached glass vessel connected to the 
main tube by means of a narrow glass tube that allows the 
vessel to be sealed off subsequently. When enough mercury 
is collected in the cups A and B (letters refer to fig. 2), 
so that by slight shaking the two mercury surfaces can be 
easily brought into contact, the connexions to the current 
source are made and the are inthe side-branch is started. At 
this stage it is best to have in the side-branch about 4 amperes. 
If the vacuum is high enough, the main arc in the tube BD 
will probably flash up, but will immediately go out in con- 
sequence of the large amount of gases given off from the 
