1205 
of this kind and we further know from the existence of a limit 
of the Röntgen-radiation on the side of the small wavelengths, that 
they cannot take place in accordance with the theory. According 
to RutHerrorp’s experiments the relative number of the emitted 
recoil-rays is strongly dependent on the rapidity or range A of the 
primary a-particles, as shown by the following table: 
ie '720 5,3 4,5 3,7 3,0 ete. 
Vi == 100 77 51 25 0. 
This might be interpreted as indicating that the radiation of 
rotational impulse decreases rapidly with the speed v, so that with 
falling v there are less and less particles which are able to reach 
the stationary orbit. If this view is correct, we would have in 
RurnerForD’s table a new way along which to penetrate into the 
riddle of the quantum theory. 
Side by side with particles which have completed the transition 
into the stationary orbit, others are to be expected, even with the 
highest velocities, which owing to a higher initial impulse have not 
succeeded in doing so. The directed radiation must therefore be 
surrounded with a scattered radiation. According to a kind personal 
communication of sir Ernest Rutarrrorp’s something of that kind 
is found experimentally: a new experimental method has shown 
that the recoil-rays are in reality less homogeneous than appeared 
originally and that side by side with the rapid H-particles observed 
at first, there are others of smaller speed’). 
As suggested above it is probable that the large deviations from 
Maxwerr’'s theory, as required for a sufficiently strong radiation, 
are limited to the range of very high accelerations. This makes it 
doubtful, whether a similar approach to the stationary orbit n = 1 
is to be expected as to the orbit n = 0; for the range near it 
correspondends to a much greater distance from the nuclens. 
On a different occasion we hope to discuss the question, bow 
the stationary orbits are distributed for nuclei other than of hydrogen. 
We shall only mention here, that for heavy atoms the equations 
(17) and (18) owing to the high value of the nucleus charge # 
make the steps of the discrete angular distribution so small, that 
the result cannot differ appreciably from a distribution in accordance 
with classical statistics. 
1) Cf. E. RUTHERFORD, Phil. Mag. 41 (6), p. 307, 1921. 
