stellar clusters 



35 



19. The colour-index has been used by Shapley to determine the distances 

 of some globular clusters, namely Messier 3, 5, 13 and 15. He has obtained for a 

 number of stars in these clusters a negative colour-index (compare the Appendix to 

 this memoir) and concludes l:o that these clusters contain stars of the spectral 

 type jB, 2:o that they consequently must be situated at a very large distance (1000 

 Siriometers or more). (Proceeding of the Mat. Acad, of Sc. Vol. 3, p. 267 ff.). He 

 bases his computations upon certain adopted values for the magnitude of the B-stars. 



Against the conclusions of Shapley I find it appropriate to raise the following 

 objections. First it is not selfevident that to a negative colour-index shall correspond 

 the spectral type B. 1 refer regarding this question to the Appendix of this memoir. 

 From the measures of Parkhurst we find to a colour-index equal to — 0.2 and even 

 to — 0.4 a mean spectral type AQ. At any rate we may assume that the spectral 

 types found in the globular clusters scarcely are bluer than, say, jB 8 or 5 9 and 

 already for these types we matj expect — taking into account the exceptional circum- 

 stances ruling in the globulars — to find dwarf stars of so low luminousity as 

 M =1 *. I admit that the negative colour-index of some stars in the globular clusters 

 offers some difficulty against accepting their galactic nature. But I find not the 

 argument to be decisive, especially taking into account the general objections 

 against the extragalactic nature of tliese objects advanced in § 17. 



Another method, also leading to large distances, for determining the distances 

 of the clusters has been proposed by Zbipel and applied by him to M. 3 (Arkiv 

 för Mat., Astr. och Fys. 1916). It is based on a relation found by Miss Leavitt 

 between the period of a variable star and its magnitude, a relation previously used 

 by Hebtzsprung to determine the distance of the small Magellanic cloud. The 

 applicability of the interesting relation of Miss Leavitt to the cluster variables are 

 however very doubtful and, according to my opinion, erroneous. I have the intention 

 to argue this point more fully, when a discussion of the variables now in preparation 

 has been completed, the chief argument being that the variables having a period 

 shorter than 24 hours have an essentially lower luminousity than those of longer 

 period. The latter have, I think rightly, by Hektzspbung been determined as giant 

 stars, whereas the former probably are dwarfs, thus confirming the conclusion re- 

 garding their luminosity drawn in the present investigation. 



20. The globular clusters and the great star cloud in Sagittarius. 



We have found that the globular clusters are most abundant in the square GD^^ 

 and the surrounding ones, i. e. in the constellation Sagittarius and its neighbours. 

 It is well known that in this part of the heaven are to be found some of the most 

 brillant parts of the Milky Way. Which is the relation of the globulars to these 

 galactic clouds? 



* In A. S. P. 26, 198 (1914) Adams announces the discovery of a star (O, Eridani) of the 

 spectral type AO, which possesses an absolute magnitude equal to 8.7. 



