DISCHARGE OF ELECTRICITY FROM GLOWING CARBON. 213 
suggested by one of us,’ that the arc discharge is fully 
established. 
In our experiments, the currents when arcing commences 
have been estimated, but we have been unable to deduce 
the magnitude of the potential gradient at the anode 
surface. The value of this latter factor will include a part 
due to the projection of electrons from the surface of the 
anode, which will be greater the higher the anode tem- 
perature. 
In the above description it is considered that the develop- 
ment of the arc from the non-luminous discharge takes 
place in two stages; the first one commences with the 
* somewhat copious emission of positive ions from the anode 
surface, when the energy of its bombardment by negative 
ions reaches a critical value; the second one begins when 
the velocity, at collision with gaseous molecules, of elec- 
trons projected from the cathode, commences to increase 
on account of the presence of positive ions near the cathode 
surface; it is completed when this velocity is such as to 
start a mode of conduction through the vapour column, 
characteristic of the fully developed arc, in which, perhaps, 
electrons are handed on from atom to atom through the 
column, as suggested in the paper just mentioned. 
The view we have taken of the origin of the change from 
the non-luminous to the are discharge receives some sup- 
port, we think, from the observations of the differences in 
the voltages required for arcing given in table II. When 
the hot carbon is vertically below the cool one, the flow of 
molecular ions is helped by the convection current of hot 
gas, whereas in the reverse position the flow is opposed by 
the current; the value of IE’ may thus, for the same 
potential difference, be different in the two cases. In 
1 Pollock, Proc. Elect. Assoc. N.S.W., 1908-9. 
