of Electrons, and the Spectrum of Canal Rays. 687 



where the first maximum is for ions moving with small 

 velocities in all directions on the average, and the second is 

 for canal-ray particles. If the curve (J) gives the intensity 

 of the radiation of an ion as a function of its velocity, the 

 curve (E) will be that for the resulting radiation. It has the 

 required form, the first maximum giving the undisplaced and 

 the second the displaced line. Whether the present (n) 

 curve, or that assumed by Stark, be the correct one, can only 

 be decided by experiment ; the present one has the advantage 

 of leading to no contradiction with the electron theory. 



§ 28. In §§ 15, 16 Stark shows that, as the average 

 velocity of the canal-ray particles increases, the intensity of 

 the displacement line increases; and further that the light 

 emitted [normally] to the direction of motion is polarized,, 

 the electric vibration parallel to the motion predominating 

 very slightly. 



Equations (35), (37), (39) show that both these effects are 

 to be expected from theory, but both are due to inequalities 

 of distribution of axes ; since the polarization is small these 

 inequalities are small, and therefore the change in intensity 

 to be expected on this account is only small. 



Stark finds that the change is rapid, and all the more rapid 

 the higher the frequency. 



These changes are most probably due to the action of 

 collisions. Grreat velocity implies collisions of great violence 

 and of short duration, and therefore great disturbance of the 

 ion and preponderance of high harmonics. In the present 

 state of our knowledge of the details of collisions detailed 

 explanations of Stark's results can hardly be expected. 

 Without entering further into the matter I think there is. 

 sufficient prospect of accounting for them on the basis of. 

 collisions, and that it is premature at present to invoke a new 

 hypothesis as to the action of radiation pressure, which, 

 moreover is difficult to reconcile with the electron theory. 



JVote I. — Since this paper was written Stark (Phys. Zeitschr.. 

 8. p. 79) has himself admitted the possibility that collisions may 

 play a part in canal-ray emission. 



Note II.— It is incorrect to call (u r , v' ', w') the velocity in 

 the svstem (X'); the expression for this velocitv has been 

 given by Einstein {Ann. Phys. 322. p. 916). The results 

 of this paper are, however, not appreciably affected thereby. 



