166 THE ROYAL SOCIETY OF CANADA 
charged it not only ionizes the ether molecule but itself combines with 
the negative ion, thus becoming neutral. If, however, the rest-atom 
is uncharged when it meets the ether molecule, it ionizes the ether, 
but goes on its way unchanged. We thus see that the charged 
condition of the rest-atom at the end of recoil depends on the nature 
of the last atom which it succeeds in ionizing. 
This does not seem surprising when one stops to consider that 
ionization implies high relative velocity between the ionizing and 
ionized bodies. In general the relative velocity is due to the motion 
of the smaller and more stable body, but this is not necessarily the 
case as is evident in the collision of the recoiling rest-atom with oxygen 
molecules. Under these circumstances the moving body may quite 
possibly be ionized as well as the molecule which it strikes. If on 
the other hand a recoiling rest-atom meets a large complex molecule 
(e.g. an ether molecule) it may ionize the latter without being itself 
ionized. Of course the experimental facts could be equally well 
explained by assuming that the rest-atom was only affected by the 
last collision. However, this assumption seems less natural. 
In the above discussion nothing has been said regarding the charge 
carried by the rest-atom at the instant it is formed from the emanation, 
because it can not possibly affect the problem, unless the rest-atom 
is not onPy negative at birth, but on recoil only collides with ether 
molecules. Under these circumstances it might still be negative 
at the end of the recoil and would therefore be collected on the anode. 
It would, however, be ionized and become neutral or positive on collision 
with a single air molecule. Removal of the air completely enough to 
avoid such a collision is practically impossible in these experiments. 
The question might possibly arise whether a gas might not be 
found in which the rest-atoms would all be initially negatively charged. 
This seems very unlikely as it would mean that the rest-atom would 
in reality have to become a large negative ion in its recoil through 
the gas, and we expect to find such large negative ions only in the case 
of aggregates. 
CONCLUSIONS. 
1. The conditions which should hold in an ideal testing vessel 
for this work have been discussed. It has been shown that the cylin- 
drical type of vessel used by many investigators does not satisfactorily 
fulfill those conditions. A parallel plate type of vessel which does 
fulfill those conditions has been described. 
2. All the available data indicates that the rest-atoms of radium, 
thorium and actinium are affected similarly by the surrounding gas 
and behave similarly in electric fields. The slight differences, which 
do exist, are differences of degree and not of kind. 
