44 



Now we derive from 2 determinations (the interval being only 2 

 magnitudes) : 



rr=:0".001 ± 0"0007. 



According to Shapley (I. c. p. 10) the diflference Faü.-Mt. W. is 

 constant = -|- 0"'.24 and as Harv. = Mt. Wilson photovis., we have 

 also: Fag.-Harv. = -[- 0"'.24. We have taken Fag.-Harv. = -]- 0'".2, 

 so that the magnitudes used should be correct. Upon closer inquirj', 

 however, the difTerence Fag. -Shapley appears not to be constant, 

 but to vary with the magnitude. We have determined the errors of 

 Fagerholm's scale by comparing the magnitudes of 156 stars, and 

 afterwai'ds we have calculated the numbers A,a for the corrected 

 magnitudes. Now we derive for the parallax from only one deter- 

 mination that can be used : 



By telling off Shaplky's catalogue we find for the parallax the 

 values .-T == 0".001 and Jt = 0".002. Summing up, we may assume 

 for the parallax of this cluster: 



jr = 0".002. 



For this cluster Shapi,ey determined the colour indices of all the 

 stars, perceptible on the plate within a circle with a radius of 12'. 

 But here, too, no great valne can be attached to a comparison of 

 the distribution of colours, found by Shapley for every J/, with 

 Schwarzschild's table. For it is not certain that all stars up to 

 13'". are visible on the plate, and just here the separation of cluster 

 stars and back-ground stars offers great difïiculties. According to 

 Shapley the distribution of colours, expressed in percentages of the 

 numbers of stars of determined absolute magnitude, is as follows: 



