( 297 ) 



This does not agree with what we have found here, in the supposition 

 that Helmiioltz's '•'yellow-wiiite" is also yellow-white in our scale, i. e. 

 is also represented by a positive number in Schmidt's scale. We 

 also tind here with the whitest stars that when they become fainter 

 the colour becomes more yellow to just the same degree as with 

 the yellower stars. Xow the expression "yellow-white" is vague, but 

 if we consider that what is called white in the scale of Schmidt is 

 whiter, that is to say bluer tlian the light of Sirius, and that the 

 solar light, the standard for white for ordinary optical considerations, 

 if weakened to the brightness of a star, in the scale of Schmidt 

 would be called 3 a 4 (Capella 3, 4), then the principal colour, 

 if Helmholtz's theory is true, instead of being yellow-w-hite would 

 still lie on the blue side of the Sirius light. 



§ 3. After the colours had thus been reduced to the brightness 

 3,0, they had still to be freed from the ijifluGiice of the atmosphere, 

 which makes them redder. This cannot be done with the desired 

 accuracy, because neither time nor altitude are given along with the 

 observations. The influence at high and mean altitudes is probably 

 very small, and the obsei-ver is sure to have taken care that most 

 of the stars were observed at a proper altitude (for in.stance between 

 SO"* and 60*^). Therefore this correction is only practically important 

 for the few southern stars which always remain near the horizon ; 

 in these cases it will be possible to represent the variation of colour 

 by a correction depending on the declination. Instead of the declina- 

 tion of the star we have taken the declination of the B. D.-zone 

 which Osthoff has added to his catalogue. 



For each spectral class we have determined mean colour- values 

 for all stars north of the equator, and for the stars south of the 

 equator we have formed the deviations from these class-means which 

 then were arranged according to their declination and combined to 

 mean values for groups of stars. We have excluded, however, those 

 classes in which too few northern stars occurred, namely I, II and III. 



The means found are : 



20 



Proceediugs Royal Acad. Amsterdam. Vol. IX. 



