550 Mr. C. G. Barkla on Energy of 
variation in the absorption by aluminium is accompanied by a 
much greater proportional change in the ionizing power in air. 
As it was evident that there was a greater transformation 
of the more penetrating rays, further experiments were made 
to detect if possible a difference in the absorption of the 
secondary rays produced by a primary beam of intensely 
penetrating rays, and of the primary rays themselves, by plates 
of aluminium of equal thickness. As the average penetrating 
power of the radiations from different bulbs used did not 
differ sufficiently, one bulb was used, and the more absorbable 
rays were cut off by plates of aluminium. As the intensity 
of the primary and secondary beams was considerably reduced 
and the percentage possible error consequently increased, the 
accuracy of the former experiments was not attained. These 
experiments, however, seemed to indicate a slightly greater 
absorption of the secondary radiation than of the primary. 
Having proved the approximate equality in penetrative 
powers of the prumary and secondary radiation from air, it 
became a simple matter to investigate the relation between 
the energy lost in secondary radiation by a primary beam in 
its passage through air and the penetrating power of that 
radiation. ‘I'he apparatus was arranged as shown in the 
figure, but with plates C and D much nearer. The ratio of 
the rates of deflexion of the secondary and primary electro- 
scopes was determined for a given bulb in a fixed position 
when the current through it was varied, also when the bulb 
was soft and when hard. The ratio of secondar y to primary 
ionization was apparently slightly greater for the penetrating 
than for the more absorbable rays. To obtain a beam of very 
absorbable rays the pert radiation from copper was 
again used as the primary. 4 copper plate was placed as 
nearly as possible in the a seein previously occupied by the 
anticathode of the bulb, so that the boundaries of the beam 
were approximately the same as before. This was not obtain- 
able with perfect accuracy, because in the former case the 
radiation had proceeded trom practically a point source, 
while to obtain a beam of sufficient intensity from the copper, 
a plate several square centimetres in size was used. The 
difference in the two cases was, however, not considerable. 
The ratio of secondary to primary ionization was again deter- 
mined, and was found to be of the same order as when the 
direct beam from the oulb was used as the primary. 
The actual readings obtained are given below, but as the 
ionization produced in the secondary electroscope was only 
of the same order of magnitude as that occurring under 
normal conditions, the possible error was great. 
