ON THE SENSITIVE STATE OP VACUUM DISCHARGES. 
631 
leaving a negative terminal. Now it is very difficult to imagine that these molecular 
streams can commence to leave the terminal before the negative discharge begins, or 
that they can continue after it has ceased ; so that in the absence of anything pointing 
to an opposite conclusion, we are justified in assuming that the period of emission of 
the molecular streams is included within that of the negative discharge. There is not 
the same justification for assuming that these molecular streams continue to be 
emitted during the whole of the period during which the negative discharge is leaving 
the terminal, but we think it probable that such is the case, though it may be that 
they are not of the same intensity throughout so as to be equally capable of causing 
phosphorescence. Seeing, then, that it is probable that the two periods of emission 
are identical, and that this supposition will not in any way affect the validity of the 
argument, inasmuch as the period of the emission of the negative discharge is un¬ 
doubtedly as long as that of the molecular streams, we shall treat these two periods as 
the same. 
In order to compare the periods of time occupied in the actual emission of the two 
discharges respectively, we shall commence by showing that the negative discharge 
occupies a 'period greater than that required by the particles composing the molecular 
streams to go the length of the tube but comparable with it. 
This proposition is placed beyond doubt by the various phenomena depending upon 
the interference of the molecular streams and esjiecially virtual shadows. It is quite 
clear that if we give relief to a portion of the tube near the positive terminal during 
the progress of a discharge with positive intermittence, the streams of molecules from 
the sides of the tube must start before those from the negative terminal do so. It is 
true that the difference may be infinitesimal, but at all events it exists. Now we find 
that the streams of molecules that come from the negative terminal are interfered with 
and diverted by streams from the side of the tube. Hence these latter streams must 
have continued to flow at least as long as it has taken the molecules from the negative 
terminal to arrive at the point of the tube where relief is given. It may be that the 
main portion of these relief streams has passed across the tube before the arrival of 
the streams from the negative terminal, but the relief streams must be still continuing, 
or they could not interfere with the others. This last remark may serve to explain 
the fact that relief phosphorescence is usually diametrically opposite to the place where 
the relief is given, so that the streams that form it are not perceptibly swept down¬ 
wards along the tube towards the positive terminal by the streams from the negative 
terminal.* The main body of the relief streams has had time to cross the tube and 
impinge on the glass before the streams from the negative terminal can reach the spot. 
But when we attempt to produce relief effects in the immediate neighbourhood of the 
negative terminal the case is far different. The two streams are then on equal terms, 
* It would not be wise to attach too much weight to this. The interference of two molecular streams 
at right angles is not very great when it does not take place in the immediate neighbourhood of the source 
of one of them. 
4 M 2 
