380 
The Galactic System of Stars [ch. xiy 
352. Against this must be noticed the apparent difficulty that the stars 
which constitute the galactic system seem to be of very different ages; at 
first sight it hardly seems probable that the .M-type dwarfs can have been 
born out of the same nebula as the huge and vigorous 0 and 5-type stars. 
We have however seen that the atoms at the centres of the spiral nebulae 
(§ 325) and of the white dwarfs must be completely ionised, and that this 
renders them immune from annihilation (§ 120). No limit can be assigned 
to the ages of the atoms which form the giant M or the giant 0 and B type 
stars if only we can suppose that they have been shielded from decay in the 
hot central regions either of a spiral nebula or of a white dwarf throughout 
the main part of their lives. 
In a spiral nebula, as in a star, the atoms of highest atomic weight must 
sink to the centre, so that it is these, broadly speaking, wffiich are preserved 
from decay. As the development of the nebula proceeds, layer after layer is 
shed by the shrinking main mass and condenses, first into clusters and then 
into stars. The stars which are born first, coming from the outermost layers 
of the nebula, will have the lowest atomic weights, and as they contain the 
highest proportion of atoms of the “permanent” elements, will be least 
luminous per unit mass. Those which are born last will contain the atoms 
of highest atomic weight and so will have the greatest luminosity per 
unit mass. 
At any instant the ages of the stars in existence, as measured from the 
time when their atoms first condensed into stars, will vary greatly, but the 
ages of the atoms of these various stars will all be the same, all dating back 
to the creation of the original parent nebula. We shall find the lowest atomic 
weights and the lowest luminosity in the stars which appear to be oldest, and 
high atomic weights and high luminosity in stars which appear to have been 
recently born. This is in general agreement with what is observed in the 
galactic system, but in addition certain specific facts of observation seem to 
have some bearing on the question. 
Shapley * has found that in various globular clusters, the brightest stars 
are red, and the faintest are blue. If the stars of a globular cluster are 
arranged in a temperature-luminosity diagram (§ 57) the upper half of the 
main sequence is missing. In the open clusters precisely opposite conditions 
prevail. The Pleiadesf, Perseus, Ursa Major and other clustersJ do not 
possess a single red giant; all the giant stars are main-sequence stars, and 
the giant branch which leads up to the giant M type stars is absent. In 
clusters such as MW, which are of intermediate type, both kinds of giant 
stars appear and the temperature-luminosity diagram is of the usual reversed- 
* Mount Wilson Contributions, Nos. 115-117 (1915), 129 (1917), 133 (1917), 151-157 (1918) 
160, 161 (1919). 
f Trurapler, Lick Obs. Bulletin, No. 333 (1921), p. 114. 
X Lundmark, Lick Obs. Bulletin, No. 338 (1922), p. 151.