64 
PHYSICS: E. DERSHEM 
But 
Whence 
AA>^i-V + 2/cosfl). 
nr 
Denning, as usual, the resolving power to be X/AX, we have by dividing X 
by each side of the inequality 
X ^ nr\ 
AX d cos 6 (s + It cos 6) 
From this it is apparent that the resolving power may be increased by in- 
creasing the order of the spectrum and the distance between the crystal and 
the plate and also by decreasing the width of the source and the thickness 
of the crystal. To increase the resolving power by any of these means re- 
sults in a loss of intensity which must be compensated for by an increased 
time of exposure. To secure the best results in any given case requires a 
selection by experience of the Dest relative values of these quantities which 
will depend upon the kind of crystal used and the hardness of the X-rays. 
It is also apparent by an inspection of figure 1 that the true position of the 
line on the photographic plate is to be obtained by measuring to the outer or 
most deviated side of the image and then subtracting one half of the width of 
the source. This does not in general coincide with the position of the most 
intense part of the image and since the point of greatest intensity is the one 
obtained by an ionization chamber method the latter can never give results 
of the greatest accuracy. 
In the experimental work the endeavor was made to obtain as high re- 
solving power and as accurate measurements as possible. A Coolidge tube 
with a tungsten target was used with a rock salt crystal to obtain the results 
given in the table. These results are certainly accurate to within 0.1% in the 
case of the L radiations and 0.8% in the case of the K radiations. 
By the use of a crystal of rock salt which was first waxed to glass and then 
ground to a thickness of 0.019 cm. the widening of the K lines due to penetra- 
tion into the crystal was reduced to such an extent as to cause the doublets 
to be clearly separated in the spectrum of the first order and this is not pos- 
sible if the thickness of the crystal is not limited. 
In the case of the L group of lines the resolving power as defined by the 
above formula was less than 170 but nevertheless 19 separate and distinct lines 
were obtained and this very naturally suggests that if it were possible to ob- 
tain such resolving powers in X-ray spectroscopy as have been obtained in 
A0 
2d cos 0 
nrA\ 
d cos 6 
> s + it cos e, 
