Mr Kaye, The Selective Absorption of Rontgen Rays. 241 
The insulation was of sulphur throughout. The electroscope 
was out of the path of, and well shielded by lead from stray 
Rontgen radiation. The connections of the ionization chamber to 
the electroscope were, as is usual, well shielded by earthed metal 
tubes. 
Measurements. 
In a comparison of the Rontgen rays, from two metals of the 
anticathode, electroscope readings were taken alternately a great 
many times — first using one metal and then the other. It is 
essential that the incident stream of cathode rays should remain 
steady and constant for a proper comparison. An induction-coil, 
such as was used, working with a hammer-break interrupter, 
cannot always be relied upon to run perfectly steadily and well for 
any length of time. It needs very careful adjustment of the 
current in the primary, of the hammer-break and its tension screw, 
to get good results for the particular pressure inside the tube. 
The surfaces of contact of the platinum break require attention 
from time to time — they should be kept clean and smooth and 
parallel. The amount of sparking at the hammer-break inter- 
rupter must be as small as it is possible to get it. When a coil is 
working well it emits a steady purr, unbroken by kicks and 
bumps at the break. 
For steadiness of running, the current in the primary must be 
much smaller than that which would make the coil give its 
longest spark. In other words, the coil must run on a light load, 
and it is therefore better to use as big a coil as is convenient. 
In spite of all precautions, only a small fraction of the readings 
actually taken were good enough to use. By comparing different 
pairs of metals, and thus getting a number of cross-checking 
results, an accuracy of 1 or 2 per cent, could probably be ensured 
for the final readings. 
From time to time the pressure in the tube was read by the 
McLeod gauge, and a reading was taken of the length of the 
alternative spark gap. By taking a number of successive readings 
with the electroscope for the same anticathode, one could tell 
whether the coil were working well or not. 
The usual length of a reading was from 20 to 50 seconds, and 
the coil was given two minutes rest between successive readings. 
This, of all the methods tried, was found to be the most satis- 
factory for yielding results which were concordant. 
Results. 
For the results given below a constant pressure of '005 mm. of 
mercury was maintained in the tube, and the length of the 
alternative spark gap was - 8 cm., which is equivalent to a mean 
