48 



C. V. L. Charlier 



The resulting relation between the spectral index and the corresponding mean 

 color-index is 



(4) . c = + 0,436 S 



or the same relation as from the Harvard observations. 



The agreement with the former reduction of Parkhuest's observations (for- 

 mula (2)) as well as with the reduction of the Harvard observations is almost com- 

 plete. We conclude that the spectral-types as determined by Pakkhurst do not 

 give rise to any systematic deviations in the color indices. Moreover we find a 

 very satisfactory agreement with the Harvard determinations of the photographic 

 magnitudes. The systematic deviations between Parkhurst's and King's color- 

 indices are therefore probably dependent on the color equation in the Harvard 

 visual magnitudes found by Messrs Müller and Kempf. 



26. The near agreement between the results from Harvard and Yerkes, when 

 both series are compared with the Ptr-magnitudes, makes it possible to combine 

 both series and thus obtain a fuller material for determining the correlation between 

 color-index and spectral type. 



We thus get a collection of 405 stars, for which the visual magnitudes all 

 are known from the Potsdam General Catalogue, whereas the photometric magnitudes 

 are determined in Yerkes or Harvard. 



The mean spectral index of these stars is 1.48 [= Fö), the mean color-index 

 is -j- 0.65. The dispersion in the spectral indices is 1.39 and in the color-indices 

 0.66. The coefficient of correlation between spectral- and color-indices is 



r = -\- 0.9317 ± 0.0066, 



confirming the high correlation found from Parkhurst's measurements. 

 The lines of regression have the form 



c = + 0.440 s + 0.004, 



(5) -r -r , 

 ^ ' S = -\- 1.975 c -4- 0.186. 



The color-indices used should therefore be diminished by 0.004, a correction 

 of no importance as the color-indices are given only to two decimalplaces. 



27. Besides the american determinations of photographic magnitudes we 

 possess the Göttingen catalogue of Schwaezschild embracing 3522 stars between 

 0^ and -f- 20" declination. These observations have been discussed by Schwarz- 

 schild in his introduction to the »Actinometrie der Sterne der B. D. bis zur Grösse 

 7.5 etc.» (1912). Like the american astronomers he only computes one regression 

 line, namely that giving the mean color-index corresponding to different spectral- 

 types *. A complete discussion of the Göttingen observations were without doubt 



* In a note (p. 18) he calls, however, attention to the desirability of making a fuller research 

 of the questions of correlation. 



