148 ANNUAL BEPOET SMITHSONIAN INSTITUTION, 1923 



character of the phenomena. We may at least say with confidence 

 that the surface temperature, as well as that in the interior, will 

 reach a maximum and then diminish, until at last the mass will 

 shrink nearly to the greatest density which it can possibly attain, 

 and end tty cooling off almost like a solid body. During the early 

 stages, while the temperature is rising, the body will be of large 

 diameter. As it contracts its surface will diminish, but its surface 

 brightness will increase, so that the amount of light which it gives 

 out will not change much. It will, however, grow whiter as it 

 gets hotter, until it reaches its maximum attainable temperature. 

 By this time it will be much smaller in diameter than at the start, 

 but only a little fainter. But after it begins to fall in temperature, 

 while still contracting, the situation is different. There are now 

 fewer square miles in its surface, and less light given out per square 

 mile, so that its light will fall off rapidly, and it will grow fainter 

 and redder until at last it disappears. 



During its history, therefore, it will pass through any surface 

 temperature lower than the maximum twice — once when of large 

 diameter, low density, great luminosity, and rising temperature, and 

 again when its diameter is small, density high, luminosity low, and 

 temperature falling. It is obvious that these contrasted groups of 

 characteristics are exactly those which differentiate the giant and 

 dwarf stars. The theoretical and observed pictures, indeed, agree 

 not merely in their general outlines, but in every detail. For ex- 

 ample, the lower the temperature selected for study, the greater will 

 be the theoretical difference between the groups of stars of rising 

 and falling temperature, and the greater is the actual difference be- 

 tween the giant and dwarf stars. The approximate equality in 

 brightness among the giant stars of the various spectral classes, and 

 the great differences among the dwarfs, find also a complete expla- 

 nation. 



Stars of large mass, as can easily be shown, should attain a greater 

 maximum temperature than those whose mass is smaller, and should 

 be more luminous than the latter, for the same surface temperature, 

 especially in the giant stages. The great masses and luminosities of 

 the B stars are thus accounted for. They are not massive because 

 they are hot, but hot because they are massive. Lesser masses never 

 attain the B stage of temperature, but stop at A; and still smaller 

 ones may not get beyond class F, or even G. As we go down the 

 spectral series, therefore, we are continually adding to our list stars 

 of mass too small to get into any of our earlier groups at all — so it 

 is no wonder that the average mass decreases for the redder stars. 

 The fact that the masses of the giants average high, whatever their 

 spectral type, is probably an effect of observational selection. We 

 have picked them from a list of naked-eye stars, and hence from one 



