The Light of the Stars 57
these molecules : they break apart into their constituent
atoms, and are dissociated. The bands disappear, while
the lines of titanium are reinforced, the oxygen, mean-
while, going as it were into hiding and requiring a much
more violent stimulus to rout it out. But why should the
other red stars of the R and N types show a different
set of bands, and why should these and the bands char-
acteristic of the М-stars never be found together? Cur-
tiss has given a beautiful explanation. Both sets of bands
are due to compounds—one to a metallic oxide—the other
to some compound of carbon. It is very likely indeed
that these two are chemically incompatible, and incapable
of existing together. When the carbon is in excess it
takes away the oxygen from the titanium—as it does from
iron-ore in the blast furnace. When the oxygen is in
excess it combines with all the carbon—getting it into
some form in which it does not produce the characteristic
bands—and what is left over combines with titanium. So
here, at a temperature low enough for the chemical com-
pounds to form, we have a real chemistry of the stars—
not merely physics—and our first distinction seems to be
the old familiar one between an oxidizing and a reduc-
ing atmosphere.
We might now reasonably expect to find that the hot-
test stars were the brightest, and the cooler stars fainter,
but the facts are against us. Among the stars of enor-
mous luminosity of which we have already spoken, those
in Orion’s belt are of Class BO, at the top of the scale;
but Alpha Cygni is of Class A2, Canopus of Class F, and
Antares actually of Class M, at the other end of the
series. A general study of the relations between the spec-
tral types of the stars and their absolute magnitudes is
therefore of importance. The material required for the