328 
PROF. C. G. BARKLA ON X-RAYS AND THE THEORY OF RADIATION. 
from this assumption occurs when primary and fluorescent radiations differ consider¬ 
ably in wave-length ; when the primary is of just shorter wave-length than the 
characteristic radiation, the error becomes vanishingly small, consequently cannot 
affect the accuracy of the determination of the maximum transformation coefficients. 
As will be seen later, there is also strong evidence that even when the primary and 
fluorescent radiations are widely different in wave-length, the error is certainly not 
large, if it is of appreciable magnitude. 
The substances Cr, Fe, Ni, Co, Cu, Zn in which the transformation coefficients have 
been determined are not suitable for accurate experimental determination of the 
energy of the corpuscular radiation—nor if they were, would they be the most 
Fig. 4a. 
suitable for the purpose of the present investigation. There is, however, little or no 
disadvantage in obtaining the transformation coefficients in the substance bromine 
by extrapolation from the above results, for the results experimentally obtained 
with elements of lower atomic weight are so regular as to leave little room for error 
in the results so determined. 
The advantages in the choice of bromine as the experimental substance are (l) it 
can be obtained in the gaseous state alone or in combination with light elements, so 
that ionizations may readily be obtained and the energy of corpuscular radiation accu¬ 
rately measured ; (2) its K spectral lines are of about the most suitable wave-length for 
