2) Q. V. L. CHARLIER, 
It is not my intention here to discuss the function f,(T)'. It 
suffices to observe that the temperature may be supposed to have a 
close correlation to the spectral type. 
The correlation between the spectral type and the diameter of 
a star is, probably, less pronounced. The small stars, when contrac- 
ting, have not the same power as the big ones to reach the high 
temperature characterizing a star of spectral type B or even A or PF. 
On the other hand a big star, having once reached the maximum tem- 
perature, cools gradually and thus runs through a series of tempera- 
tures according to the spectral-scheme: 
MA UREA RGP NAS BA BENE CET 
without materially altering its diameter. 
This view of the development of the stars was long ago pro- 
posed by LockvER and has since been worked out, on other grounds, by 
HERTZSPRUNG, RUSSELL, SHAPLEY a. o. In the Meddel. N:o 68 I have 
shown, through a discussion of the proper motions of all stars down 
to the 6" magnitude, that the mean absolute magnitudes and the dis- 
persion in the magnitudes of the stars of different spectral types are 
connected by a linear relation and that the fluctuations in the lumino- 
sities of the stars of type B (and even of type A) must be very small, 
The absolute magnitude is defined in these investigations as the 
apparent magnitude of a star at a distance of one Siriometer, Deno- 
ting the absolute magnitude by M we hence get from (1) 
|| = R 1002 m 
or 
(2) M = —5 log Rk. 
From this relation it follows that from the constaney of M for 
the stars of type B also the constancy of R for these stars may be 
concluded. 
3. The determination of R is based on the proper motions and 
the radial velocities of stars brighter than the 5" magnitude. "There 
are 299 B-stars in the heavens brighter than this magnitude. Of these 
! Compare Meddel. N:o 67. 
