100 Dr. E. Weintraub on the Arc in 
a solenoid. On closing the switch P the current goes through 
the solenoid and the mass of mercury filling the cups AB. 
The energized solenoid pulls the iron rod out of the mercury,, 
thus separating the mercury surfaces A and B, and a spark 
is formed which starts up the are in the main tube BD. By 
providing a magnetic cut-out energized by the current flowing 
in OD, the current in the side branch AB can be cut out 
automatically as soon as the arc BD starts up. The current 
in the side branch need not be large, and a small part of an 
ampere is sufficient to start the ionization process on the 
cathode B and the flow of ions in the tube BD. The length of 
this latter tube is regulated only by the voltage impressed on 
it, and with 120 volts tubes 55 ins. long can be readily started.. 
The instantaneous starting up of such a long arc is one of the 
most beautiful and effective experiments for the purposes of 
demonstration. 
The arrangements arrived at in figs. 2 and 3 represent a 
convenient way of starting a mercury arc, or an are discharge 
through any other metallic vapour, instantaneously, by means. 
of a source of the same voltage as that on which the are runs. 
stable. 
The initial spark can also be produced by means of a Holtz 
machine or induction-coil, the two poles being connected to. 
the electrodes A, B of the starting branch in the same way as. 
was the source of moderate voltage. This spark is, however, 
somewhat less efficient in consequence of the very small flow. 
of current produced by it. 
The experiments 4 and 5, illustrated by figs. 2, 38, do not 
succeed if B is made the anode and A and D the two cathodes. 
of the arc. The arc AB starts, of course, in the same way 
as before. The main arc BD does not start, however, even: 
after the ionized vapour propagating from the are AB has. 
filled the space BD, whether D is a mercury or graphite- 
anode. 
The difference in the behaviour of the two electrodes in 
respect to the starting of the arc is shown by all these expe-. 
riments. In the ordinary way of starting an arc between two. 
electrodes, such as two carbons, or two metals, by bringing 
them into contact and separating them, the process is entirely 
symmetrical, and no conclusion as to the difference between 
the two electrodes can be drawn. 
By the experiments described above, especially experi- 
ment 2, this process has, so to speak, been divided into two, 
and the predominating réle of the cathode shown. This does 
not prove that no ionization at all takes place at the anode, 
but only that the primary one takes place at the cathode, and 
