McCie.tianp anp Hackurr—Secondary Radiation from Compounds. 31 
fal 
Snyw, A 
and therefore Ke 
Ky 
} 
1 
nw, My 
a, 
7 
Ma 
ay 
Sm, ua Ky 
(a7 
or finally ee —, 
7) 
TEL 
where m, is the mass of any element per unit volume of the compound, and 
fu, &, and x, are the true coefficient of absorption, density, and transformation- 
constant of that element. 
Before we can use this expression to enable us to calculate « for a compound, 
we must know the true coefficient of absorption (1) for the constituents of the 
compound. Ina previous paper,* by one of us, the relation between p and the 
usually observed coefficient was worked out; but it was also shown that the 
methods hitherto employed in measuring the coefticient of absorption involved 
considerable inaccuracies owing to the fact that the effects of secondary radiation 
had always been neglected. We have not, therefore, the data necessary to supply 
accurate values of » for the elements. 
An approximation to accuracy will be to take mw to be proportional to the 
density d of the elementary substance. With this assumption the expression 
giving « for a compound takes the simple form— 
Lm ky 
Sm 
From some work going on in this laboratory at present, it is probable that the 
ratio »/d will be found to increase with the atomic weight of the substance. If 
so, since « for all elements increases as the atomic weight increases, calculations 
made from the above simple expression should give too small a value for x for a 
compound in which an element of high atomic weight is combined with one of 
low atomic weight. It should give an accurate result for compounds composed of 
elements of nearly equal atomic weights. The error introduced should, in any 
lig 
d 
x. This is practically what has been found to occur, as will appear from the 
results shown in the following table. ‘The differences between the calculated 
and the observed values are always small, and in the cases in which the difference 
is appreciable the compound is made up of elements of very different atomic 
weights, and the calculated value is less than the observed value. 
case, always be small, because of the form in which — occurs in the expression for 
* Trans. Roy. Dublin Soc., vol. ix., part 1., 1906. 
