94 Recent Advances in Stellar Astronomy
der 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 then from a list
of naked-eye stars, and hence from one in which the
brighter stars have an egregious preference, and it has
already been seen that, in these stages, great brightness
means large mass.
A more searching test is found in the densities of stars
of the various sorts; for here we can make our comparison
quantitative instead of merely qualitative. The stars of
increasing temperature should have densities at which the
simple gas laws can be trusted to apply, at least approxi-
mately; the dwarfs should be so dense that we can be sure
that these laws fail of application; while the hottest stars
should have an intermediate density corresponding to the
region in which the gas laws are strikingly at work. From
a general knowledge of the properties of matter, we can
say with certainty that a density less than ten times that
of air falls in the first class, one greater than that of water
in the second, while the “twilight zone” between cor-
responds to densities in the neighborhood of one-tenth to
one-quarter that of water, and perhaps a little higher.
Now we have already seen that the redder giant stars are
less dense than air—the whiter ones being probably from
ten to fifty times denser; that the average density of the
А-stars is one-fifth that of the Sun, or one-third that of
water, while their individual densities range from about
fifty times that of air to that of water, and that the dwarf
stars have densities running from about that of water up
to four or five times as much. The agreement is perfect
throughout, and there can be no remaining doubt that the
proposed physical model represents what actually happens
in the stars.