ON THE SENSITIVE STATE OF VACUUM DISCHARGES. 
617 
exhausted tubes each of the two positive columns of the double unipolar discharge 
assumes a thick tongue-shaped form, the tips pointing towards each other, so in 
higher exhausts we find two long pencil-like columns each tapering away to a point, 
and each accompanied by its attendant phosphorescence. Either of these can be used 
for the purpose of examining the properties of this phosphorescence, and the results 
will hold good in all cases, however the pencil-like column be obtained. There is 
no such pencil-like column when a negative unipolar is observed. 
The first thing to do in the analysis of this attendant phosphorescence is of course 
to ascertain the direction and, if possible, the source of the streams of molecules which 
produce it. This was done by means of a tube into which had been introduced a 
number of crooked pieces of wire and small bits of glass. When the attendant 
phosphorescence was brought up to these it was found that (subject to the qualifica¬ 
tions to be given later on) the shadow of each object lay in the same normal section of 
the tube as the object itself, and was, roughly speaking, in the position in which it 
would have been had the seat of the discharge been the opposite side of the tube. 
This conclusively showed that the streams producing this attendant phosphorescence 
did not come from the negative terminal, but were due to local action in the tube, and 
that the direction of these streams was normal to the axis of the tube. 
Further and more minute investigation has since shown that these results require 
some modification when the part of the attendant phosphorescence which is being 
examined is situated at some bend or corner of the zigzag course of the positive 
luminosity, and that they can only be taken as accurate where the luminosity is not 
very greatly curved. This result, so far from affording any argument against the 
theory that this phosphorescence is due to local action, strongly supports it, as showing 
that peculiarities in the direction of the molecular streams are only found when we 
should, from the circumstances of the case, expect to find peculiarities of local action. 
No doubt in such cases the direction of the streams is a resultant of conflicting 
tendencies. We shall, therefore, neglect such cases for the present, and confine our¬ 
selves to the case of a fairly straight positive column where, as we have already said, 
the streams at each point come to the glass along paths lying in the normal section of 
the tube at that point. 
We see at once that we have here a phenomenon of almost unexampled importance 
in the analysis of the modus operancli of the positive discharge. We find that its 
passage is accompanied by discharges of negative all along the tube, the molecular 
streams accompanying which move in directions normal to its axis. Taking as a 
provisional hypothesis* that these start from the sides of the tube, we see that we 
have a process of relief continually going on. But the most remarkable point seems 
to be that these streams either start in the direction of or are subsequently directed to 
the thin column of positive luminosity. This leaves scarcely any room to doubt that 
* We shall presently see that there are experimental grounds for believing this hypothesis to be a 
correct one. 
