94 
PHYSICS: D. L. WEBSTER 
the experimentally determined points by less than the limits of error, 
which are of course much larger at low potentials than at high. Since 
it is really a ratio of ionizations in an arbitrary amount of ethyl bromide, 
no importance can be attached to its absolute value, but its constancy 
seems very significant. 
Another way of testing this point is that of plotting the spectra at 
different potentials as in figure 3. For various reasons this method is 
even less accurate than the other, but the ratios of increase of the a 
and /3 lines are again approximately the same, and here it appears that 
even the y line increases similarly. 
These spectra are of interest in showing also the influence of a small 
impurity of ruthenium in the rhodium target in producing its own 
characteristic lines; but the most important information they contain 
is the location of the wave length whose quantum potential is the criti- 
cal value. Its position, 1.3% short of the y line, is independent of any 
errors entering uniformly in these potential measurements. Now this 
part of the spectrum is known to be marked by a sudden rise of absorp- 
tive power with decreasing wave length, and indeed the drop in the 
general radiation at this point is undoubtedly due to the influence of 
this absorption on the rays leaving the target, as they do, at a very small 
angle to its surface. This increased absorption, moreover, is known to 
be accompanied by a strong characteristic fluorescence, indicating that 
when a higher frequency oscillator has acquired whole quantum by 
absorption it undergoes a drop to the characteristic frequencies for 
emission. 
Now the results of this work show that to obtain any characteristic 
radiation by the impact of cathode rays, each of the latter must have 
have energy enough to satisfy one of these higher frequency oscillators. 
Hence it seems probable that in this case as well as in fluorescence the 
characteristic rays are produced by direct excitation of the higher fre- 
quency oscillators and their subsequent drop in frequency on emission. 
To sum up, these experiments show three points. First, to excite 
any characteristic radiation it is necessary to use a potential above a 
critical value which is the value required for general radiation of a wave 
length 1.3% shorter than that of the y Hne. Second, the lines all in- 
crease in the same ratio for any given increase of potential. Third, there 
is reason to beHeve that the characteristic rays are always a result of exci- 
tation of higher frequency oscillators, as in the case of fluorescence. 
lAmer. Phys. Soc, April, 1915; Physic, to., July, 1915. Recomputed on the basis ot 
Millikan's e their value is 6.50. 
^Heat Radiation, translation by Masius, p. 172, gives 6.415. 
3 Amer. Phys. Soc, Dec, 1915. These values are 6.57 and 6.59 respectively. 
