ASTRONOMY: ADAMS AND SHAPLEY 
139 
A method of applying the Harvard system of classification to the de- 
termination of spectral type through the aid of numerical relationships 
between the intensities of the hydrogen lines and certain other selected 
lines will be described by one of us in another Communication. The use 
of this method gives in the case of b Cephei: 
At maximum F4, At minimum G2 
The variation in type, accordingly, amounts to 8 divisions of the Har- 
vard scale. 
An independent determination of the spectrum and its change can 
be made from the comparison of the mean visual and photographic light 
curves. Nearly twenty visual curves have been made during the last 
hundred years, but only the one by Stebbins is based on measures with 
a photometer. The only photographic curve is due to Wirtz. It is 
possible to reduce both curves to the international magnitude scale; the 
difference between them for any phase gives the color index, which may 
be transformed into spectral type by known relations. Hence we de- 
rive in stellar magnitudes: 
Visual Range - 4.25 - 3.49 = 0.76, 
Photographic Range = 5.15 - 3.90 = 1.25; 
and for the spectral class at light maximum F2, at minimum G4. The 
time of plate 7 4571 coincides with velocity minimum but succeeds the 
minima of the light and color curves by several hours. Allowing for 
this we derive from the color curve, for comparison with the direct 
classification above, the satisfactory result: 
At maximum F2, At minimum GO 
An important conclusion is that all genuine color variations observed 
in Cepheids may be directly interpreted as normal changes in spectral 
class. 
Associated with the variation in spectral type in 5 Cephei is a varia- 
tion in the intensity of certain spectrum lines which have been found to 
fluctuate with the intrinsic luminosities of the stars in which they occur. 
The use of the intensities of these lines in calculations of absolute stellar 
magnitudes will be described elsewhere. The application of the method 
to the case of 5 Cephei is of considerable interest because of the accurately 
known range of variation in apparent, and hence, of course, in absolute 
magnitude. Unfortunately the spectrum of the star at maximum (F4) 
is of a type not well suited for the use of the method and the results are 
necessarily approximate. The variation in absolute magnitude found 
in this way, however, is entitled to considerably more weight than are 
