Metallic Vapours in an Exhausted Spuce. 97 
One of the methods consists in bringing the two mercury 
surfaces, which have to serveas electrodes, into contact, then 
separating them and lengthening slowly the arc by moving 
one of the mercury electrodes away. 
The other method is founded on the principle of breaking 
down the initial resistance (dielectric strength) of an evacuated 
space to passage of current by a high-voltage discharge. 
This method was further developed with a view to commercial 
application of the are as a source of light by Peter Cooper 
Hewitt (Electrical World and Engineer, p. 641, April 20, 
1901). 
From a practical point of view the two methods are in- 
convenient. The first one requires a large mass of mercury, 
and the starting, especially of long ares, is slow and uncertain. 
The second one needs rather complicated arrangements for 
the production of the high-voltage shock, and has other 
disadvantages which need not be discussed here. 
The new way of starting the passage of a current of low 
voltage through metallic vapours that is to be described in 
this paragraph solves, from a practical standpoint, the problem 
of starting long arcs ina convenient way and instantaneously, 
without the help of any higher voltage than that necessary 
for maintaining the arc; while from a theoretical point of 
view it throws some light on the mechanism of the con- 
ductivity of the arc in general. The results obtained can be 
conveniently expressed by using the modern terminology 
of the theory of ionization in gases, and the principle dis- 
covered can be considered as the first application, to my 
knowledge, of that theory to a practical problem. 
The following series of experiments will serve to lead up by 
degrees to the recognition of that principle. 
EHaperiment 1.—Fig. 1 (PI. III.) represents a glass tube 
ABCD, which has four cups A, B, C, D, filled with mercury. 
The tube is exhausted by means of a mercury-pump of the 
Sprengel type to as higha degree of vacuum as possible. 
The dimensions of the tube are not very essential for the 
success of the experiment, so long as the distance between C 
and A does not greatly exceed 10 inches. By means which will 
be described later, an arc is established between the electrodes 
B and D, the current being derived from a direct-current 
source of moderate voltage g (for the sake of simplicity one 
can assume that are BD is being started by means of a high- 
voltage discharge). A source of direct current, g’, of mode- 
rate voltage (100-200 volts), independent of the first one, 
is connected to the two cups A and ©. vr and7, are two 
variable resistances. Under ordinary conditions, in absence 
Phil. Mag. 8. 6. Vol. 7. No. 38. Feb. 1904. H 
