224 
MESSRS. W. SPOTTISWOODE AND J. FLETCHER MOULTON 
terminals, and the other occupies the vastly longer period that elapses between two 
consecutive sparks.* 
The probability that both kinds of discharge are really pulsatory is increased by the 
consideration that the stake are formed by the discharge. This increases the difficulty 
of supposing that a strictly continuous current could imitate effects which we have 
seen to be caused by discharges known to be instantaneous and disconnected. If the 
evidence given in the paper as to the form of the discharge from one stria to another 
(see page 205) be considered sufficient, these remarks have still greater weight, for it is 
scarcely conceivable that a strictly continuous current should take so strange a course. 
The passage from stria to stria must then be taken as disruptive and discontinuous, 
and if this be granted, then, as striae are only particular cases of terminals, it follows 
almost as a matter of course that all discharges in rarefied air are equally so. And 
this consideration becomes of greater weight when we consider the intimate relation of 
striae and their component parts to what we have termed the physical unit of a striated 
discharge. Such a discharge is, as we have seen, made up of repetitions of a unit 
consisting of the bright surface of a stria, the dark space between it and the next stria 
and the hazy interior of the latter. Such a structure then must either be necessary 
for the passage of the discharge or its natural consequence. Take the hollow cone 
discharge produced by the positive non-relief-effect. Here we compel a positive 
discharge to leave the interior surface of the glass under such circumstances that an 
equal amount of negative electricity is left behind to satisfy the original positive 
discharge. This it effects by causing the positive column to stop suddenly at a 
considerable interval from the place of the discharge, and to terminate in a sharply- 
defined bright head, which is in all respects identical with the outer surface of a stria, 
while on its part the negative discharge forms the hazy interior of the hollow cone. 
Thus the discharge requires and compels the formation of a unit of the structure above 
described. It seems in the highest degree improbable that structures formed and 
maintained to satisfy the needs of disruptive discharges (which have to leap from the 
one part and alight on the other of the differently constituted terminal portions of the 
structure) should also be formed and maintained by discharges which, in virtue of 
their continuous character, must be supposed to flow with the same evenness as in 
a metallic current, and to remain obedient to the law that the same amount of 
electricity flows across every section of the circuit in the same infinitesimal portion of 
time. 
But if we admit that the ordinary discharge is discontinuous, the rapidity of the 
intermittence must be very great. For if this were not the case the individual 
pulsations would be of considerable magnitude, and would cause sufficient rise and fall 
of electric tension outside the tube to give to us the phenomena of the sensitive state. 
But if they are thus rapid there is no difficulty in the non-sensitiveness of the 
* This is selected as the most striking instance. But most of the phenomena of the relief and non¬ 
relief-effects would probably be found on examination to alford instances of it. 
