( 299 ) 



With 9 stars (out of 355) the deviation exceeds a unit of colour ; 

 the reduced colours are here : 



/? Can. maj. Ill 1,2 8 Hydrae XIII 5,2 7j Persei XV^ 6,8 

 o^ Cygni 1X1,4 ft Persei XIV 5,5 llUrs.min. XV5 6,6 

 d Delphini IX 3,8 o, Cygni XfV 6,5 5 Ononis XVII 7,9 



In this investigation we have, as it was said before, excluded the 

 c- and «-«c-stars, the L (bright lines), the P (peculiar spectra) and 

 the C (composed spectra). It is important to examine the c and the 

 «c-stars among them more cl.osely in order to see whether thej show 

 a distinct difference in colour from the «-stars of tlie same class- 

 number. In the mean 11 «c-stars give a deviation of -j- 0,1 (from 

 -f- 0,5 to — 0,3), and 12 c-stars -|-0,7; so these last ones are a little 

 redder than the dt-stars. Here, however, the great individual deviations 

 are very striking ; the extreme vahies are : 



Q Cassiop XIII + 2,5; -/, Orionis III + 1,8 ; 47?Camelop F/+ 2,0; 

 3FCamelop F7^+ 1,5; n I^eonis VII — 0,3 ; /? Orionis VI — 1,2, 



The differences are very great, but no regularity can be detected. 



§ 5. The results found solve a problem which in my former paper 

 remained unsolved, namely where in the continuous series of spectral 

 classes shall we have to look for the maximum of radiating power. 

 The colour-numbers show very distinctly a fall in the first classes, a 

 minimum between the 4^'^ and the 5"^^^ class and then a continual 

 rise. The stars which in order of evolution directly follow on 

 y Orionis {n Aurigae, n Hydrae, u Herculis) have the whitest colours ; 

 both the earlier and the later stages of evohition are yellower; 

 classes I and II agree in colour best with class VIII. Therefore, in 

 so far as we are entitled to derive the entire radiation from the 

 colour, the maximum of radiating power lies between the 4"' and 

 the 5*'' class. 



The mean colour-numbers for each of the groups formed before are: 



20* 



