DIFFUSE MATTER IN SPACE 
377 
where x 0 — hv 0 /RT. If E is the average initial energy of an electron after 
expulsion and T 0 the corresponding initial temperature, we have 
%RT 0 ~ E = hv — hv 0 . 
Hence by (257*4) we find 
For large values of x 0 the initial temperature approximates to § T, the 
next approximation being 
Afterwards the mean temperature tends to rise above this initial value, 
because the slowest electrons are weeded out most quickly by capture. 
If this cause operated alone the average temperature would become equal 
to T. But we have seen that some fraction of the initial energy (| to is 
gradually lost by the process ( d ). The conclusion is that the temperature 
of interstellar matter will be between §T and T. 
In considering a suitable average value to adopt for T it must be 
remembered that we are only concerned with radiation of short wave 
length capable of ionising the atoms, and therefore the hottest stars must 
be given most weight. I therefore suggest a temperature about 10,000° 
for interstellar matter. The interesting point is that when ionisation alone 
is operating the enfeeblement of the radiation makes no difference to the 
temperature assumed by diffuse matter; T 0 depends on the relative in 
tensities for different frequencies and not on the absolute intensity. For 
example, diffuse matter round about the orbit of Neptune should be cooler 
than the average; the sun, by liberating large numbers of slow-moving 
electrons cools the interstellar material in its neighbourhood. 
The weak point of the investigation is that we have assumed the 
radiation of the stars to follow the black body law at very short wave 
lengths—an assumption which has little theoretical or observational 
justification (§ 229). Numerical results must therefore be uncertain. I do 
not think there is any thermodynamical principle that forbids interstellar 
material attaining a temperature higher than the effective temperatures 
of the stars if there happened to be absorption bands in stellar spectra so 
placed as to cut down the number of expulsions with small energy. 
258. In certain spectroscopic binaries the absorption lines H and K 
of calcium are found not to partake of the orbital motion shown by the 
other spectral lines. Evidently these lines do not arise in the atmosphere 
of either component; they are formed either in an envelope surrounding 
the whole system or during the passage of the light through interstellar 
Fixed Calcium Lines.