298 
THE SOURCE OF STELLAR ENERGY 
elements accumulate. But according to § 195 diffusion is so slow that 
there is no appreciable flow of the heavy elements to the centre. In fact 
in the parent star of Capella the heavy elements would probably diffuse 
upwards owing to radiation pressure. It is true that the tendency may 
be for heavy elements to be evolved at the centre; but against this we 
have the stirring of the star by the circulatory currents caused by its 
rotation. Presumably the rotation of the parent star must have been 
rapid since otherwise it would not have divided. This objection might 
possibly be surmounted by the idea of stratified circulation (§ 199). 
We may turn, however, to wider pairs and groups such as the 
Taurus cluster where the primitive material from which the stars 
separated can have had no very definite central condensation; pairs of 
stars showing the same anomaly as the two components of Capella can 
be picked out. Wherever in a coeval group of stars we find the more 
massive stars with the lower effective temperature the paradox arises; 
they liberate more energy per gram at lower temperature and density 
and since they have been doing this through the past their store cannot 
be less exhausted. 
It may be suggested that in the close binaries the stars have not yet 
reached a steady state so that we are not justified in inferring the amount 
of liberation of subatomic energy from the radiation of these stars. A 
star must reach the state of balance in a period of the order of the Kelvin 
time-scale, that is to say, about 100,000 years for giant stars. If there is 
anything in this suggestion the anomaly should be conspicuous in the most 
recently formed binaries; these are presumably the eclipsing variables 
with separation not much greater than the dimensions of the stars. It 
appears to be the general rule that in ordinary giant pairs the more 
massive component has the lower effective temperature (as in Capella); 
but almost without exception this is reversed in the eclipsing variables, 
the fainter and less massive component having the lower surface brightness. 
It is just those stars in which the anomaly would be pardonable which 
fail to show it at all. 208 
208. The foregoing difficulties arise in a comparison of giant stars 
with one another and with a star of the main series. We might perhaps 
hope that an explanation confined to stars of the main series would be 
a simpler problem to start with. There is the great advantage that effects 
of temperature in stimulating the liberation of subatomic energy are 
eliminated since the central temperature is approximately constant along 
the main series. The first thing that strikes us is the enormous exhaustion 
effect. The liberation per gram by Krueger 60 is of the liberation 
per gram by V Puppis. Moreover Krueger 60 has a much higher density; 
it is not unnatural to suppose that the rate of liberation is proportional