the same direction, we can imagine two ways in which the action
proceeds: either the supply of gaseous molecules at the surface of the
negative pole must run short and the phenomena come to an end, or the
molecules must find some means of getting back. I will show you an
experiment which reveals the molecules in the very act of returning.
Here is a tube (Fig. 14) exhausted to a pressure of 0.001 millimeter
or 1.3 M. In the middle of the tube is a thin glass diaphragm, C,
pierced with two holes, D and E. At one part of the tube a concave
pole, A', is focused on the upper hole, D, in the diaphragm. Behind
the upper hole and in front of the lower one are movable vanes, F and
G, capable of rotation by the slightest current of gas through the
holes.

[Illustration: FIG. 14--PRESSURE = 0.001 MM. = 1.3 M.]

On passing the current with the concave pole negative, the small veins
rotate in such a manner as to prove that at this high exhaustion a
stream of molecules issues from the lower hole in the diaphragm, while
at the same time a stream of freshly charged molecules is forced by
the negative pole through the upper hole. The experiment speaks for
itself, showing as forcibly as an experiment can show that so far the
theory is right.

This view of the ultra-gaseous state of matter is advanced merely as a
working hypothesis, which, in the present state of our knowledge, may
be regarded as a necessary help to be retained only so long as it
proves useful. In experimental research early hypotheses have
necessarily to be modified, or adjusted, or perhaps entirely
abandoned, in deference to more accurate observations. Dumas said,
truly, that hypotheses were like crutches, which we throw away when we