316 Transactions of the Royal Society of South Africa. 
employed in the reflection, falls off inversely as the square of the sine of 
the glancing angle of the incident radiation. 
There is one serious difficulty, however, which is also noticed by 
Millikan * as an outcome of his work on the photoelectric determination of 
Planck's " A." The primary radiation used is homogeneous ; the oscillators 
in the substance, whether solid or gaseous, possess definite frequencies, yet 
any radiation will ionize any gas, or cause the emission of corpuscles from 
any solid. Millikan supposes that there must exist in the substance a few 
oscillators whose period equals that of the ultra-violet light he employs ; by 
analogy are we to assume that there exist in any substance a few oscillators, 
whose period exactly equals that of the primary radiation ? Millikan claims 
that photoelectrons are bound electrons. Then from our experiments on 
the ionization of gases we must suppose that this accidental number varies, 
as the fourth power of the atomic number of the elements composing them. 
I do not think this is feasible. Also if one absorbing substance be used and 
radiations of various frequencies are absorbed in it, then the accidental 
number of oscillators tuned to the frequency of the incident radiation varies 
as according to the law of absorption of various wave-lengths. 
Another way of stating that the absorption of energy varies as and 
the number of corpuscles produced varies as is that the absorption of 
energy per corpuscle is a constant and independent of N. This is true if 
the kinetic energy of the corpuscle is a measure of the energy absorbed 
in its production. The speed of the ejected corpuscle depends only upon 
the frequency of the incident radiation. 
In a letter to Nature (March 4th, 1915, p. 7) Barkla described an 
experiment pointing to the conclusion that for each quantum of energy 
hn^ absorbed from the primary radiation, one high-speed electron was 
emitted whose energy was characteristic of the incident energy, together 
with one quantum each of the K and L radiations. In other words the 
absorption of energy from the primary radiation per election 
= Imvj^^ + hnj^ + hn^ + • . . 
Whatever the relative values of the terms may be the first term is 
eventually transformed into terms similar to the second and third. There 
is evidence enough to show that the secondary radiation only emerges after 
a ring of electrons has been depleted of one of its members. The criterion 
is that the energy absorbed from the corpuscle at each collision must be 
equal to or /m^ according to the ring depleted. 
The number of high-speed electrons produced varies as N^, this is so for 
both gases and solids and holds for radiation of a definite wave length. 
The true absorption can therefore be put into the form 
*' Phys. Eev., vol. vii, p. 387. 
