104 
Gaseous Stars 
[ch. m 
follows that the calculated value of N*/A varies approximately as /a 8 ' 5 . For 
stars of large mass the variation is different but of comparable amount. Thus 
a small error in the value assumed for /a, will result in a large error in the 
values calculated for N 2 /A. 
With this in mind let us examine the significance of the values for N*/A 
shewn in Table XIII. 
94. Atoms of moderate weight have an atomic weight A which is approxi 
mately twice their atomic number N, so that A is approximately four times 
N*/A. In more massive atoms the factor is larger; for uranium 92 and 
A — 238, so that N*/A = 35-5 and A = 6‘7 (N 2 /A). Thus all the entries in the 
last column represent atomic weights far higher than that of uranium, and in 
most cases atomic weights of thousands at least. 
These values of N*/A have been calculated relative only to a very special 
model of stellar structure. If we had complete confidence that this model gave 
a true picture of the structure of a star, and if we further had complete 
confidence in Kramers’ theory of the absorption of X-radiation and of the use 
we have made of it, we could only conclude that stellar matter shewed a 
stopping power for radiation far beyond that of any known terrestrial atoms, 
and hence that it consisted of atoms far heavier than uranium. 
There is no á priori objection to supposing that stellar matter consists of 
atoms heavier than uranium ; indeed we shall see later that such a possibility 
must be taken very seriously into consideration and that there is much to be 
said for it. The only atoms with which the chemist is acquainted are the 
atoms which occur in the outer surfaces of the stars and near the surface of 
the earth, which in turn has probably at one time been formed out of the 
surface layers of the sun. Thus the atoms of chemistry are “ surface-atoms ” 
and there is no valid reason for expecting them to form a fair sample of the 
atoms of the universe as a whole. It is a priori far more probable that their 
very lightness has caused them to float to the surfaces of astronomical masses, 
and that concealed in the far depths of these bodies are atoms of far higher 
atomic weight, whose extreme weight has caused them to gravitate to the 
centres of the stars. If so terrestrial chemistry may properly be described as 
“ surface-chemistry,” and cosmical chemistry, of which terrestrial chemistry is 
a branch, deals with a wider range of elements. 
But the acceptance of this hypothesis does not make it possible to accept 
the values of N*/A shewn in the last column of Table XIII. These figures 
have been calculated for an assumed molecular weight p — 2'5. Now even 
completely broken up uranium has an effective molecular weight 2’56 which 
is greater than this; atoms of atomic weight far above uranium would in all 
probability have still higher values of ¡i even if they were completely broken 
up, and our calculated central temperatures are inadequate to break up even 
atoms of uranium completely.