74) J. A. POLLOCK AND A. B. B. RANCLAUD. 
in the experiments under discussion, the arc forms long 
before the current is saturated, while there is yet a potential 
rise at the cathoe and therefore no possibility of ionisation 
by collision near the hot surface, one must look to the 
anode for the origin of the change from the non-luminous 
to the arc discharge. Asa confirmation of this view, we 
have noticed that the change to the luminous régime of 
current flow is always heralded by the appearance of a 
small white hot spot somewhere on the face of the positive 
earbon. After the advent of the spot, the development to 
the full are takes place too rapidly for its phases to be 
followed by the eye, although, if the circuit is immediately 
opened, the formation of the luminous discharge may be 
prevented. 
If I is the current of negative ions, EH the potential 
gradient at the anode surface and 4 the length of the last 
free run of the ions, at the end of which they collide with 
the anode, the energy reaching the anode surface per second 
is I HA; it is here suggested that for the arc to form, the 
potential difference between the carbons must reach a value 
necessary to make the magnitude of IEA sufficient to 
raise a portion of the anode surface to such a temperature 
that positive ions are somewhat freely emitted. These 
ions, in travelling to the cathode, first annul the hot carbon 
rise of potential, and then by accumulating near the 
cathode surface create a cathode fall of potential. With 
a cathode fall the positive ions bombard the cathode sur- 
face with considerable energy thereby raising its temper- 
ature; as aresult, the electrons are projected in greatly 
augmented numbers, and with enhanced velocity, their 
speed being still further increased during the free flight in 
the now reversed field. When the velocity of the electrons, 
at the end of their average range, reaches the value of 
2°6 x 10° centimetres per second, it may be considered, as 
