ASTRONOMY: H. N. RUSSELL 
399 
red end of the spectrum? There are certainly very heavy absorption 
bands in the red in these spectra; and further evidence in favor of this 
hypothesis is found in Coblentz's measures of Alpha Herculis,^^ which 
show this star, of Class Mc, radiates far more heat in proportion to its 
light than do stars of Class Ma, and also in Hertzsprung's^^ observation 
that the very faint dwarf stars of Class Mb are not nearly as red as their 
small luminosity, and probable low surface brightness, would lead one to 
suppose. A careful study of the color indices, and, if possible, of the 
spectral energy curves, of the stars of Classes Ma, Mb, and Mc is much 
to be desired. The extraordinarily red stars S Cephei^^ and +43°53,29 
which have color indices exceeding five magnitudes, should be included 
in such a study. 
3. One other stellar characteristic which may be investigated without 
knowledge of distance is variability of brightness. If we really under- 
stood the causes of stellar variabiHty, we should probably have ad- 
vanced a long way towards the solution of the whole problem of stellar 
evolution, if not have solved it completely. But, in spite of the great 
number of variable stars, the variety of the phenomena which they 
represent, and the accuracy with which they can now be observed, the 
humiliating admission must be made that no even tolerably satisfactory 
theory of the causes of the variation exists, except for the eclipsing 
variables, and in this case it is based on the proposition that, except 
for the accident of eclipse, the components are not variable at all! 
Successful attack upon the problem of intrinsic stellar variation will 
probably demand the correlation of all the data that can be brought 
together from every accessible source. In the case of regular variables, 
precise light curves are of importance, and many stars still await in- 
vestigation, — some of them visible to the naked eye, and long known to 
be var'able. The new photometric methods of precision — especially 
the photoelectric cell — have opened a wide field in the study of bright 
stars with small variation, in which important results have already 
been obtained, — notably by Stebbins^^ and Guthnick,^^ — and more may 
be anticipated. 
(a) Former suspicions of changes in form of the light curves appear 
to have been unfounded in the case of eclipsing variables; but similar 
changes are beheved with good reason to exist among Cepheid vari- 
ables.^2 Xo prove their reality — still more to discover their laws — 
demands very precise observations, preferably by two or more observers 
at different places and the same time. 
(b) Changes in color, as well as in brightness, appear to be the general 
— ^perhaps the invariable, rule among eclipsing variables, and especially 
