( 292 ) 



good deal more than those obtained formerly when the undivided 

 figures were simply multiplied. But on the other hand there are 

 other cases, particularly those in which a metasubstituted substance 

 is nitrated, where this calculation does not agree with the experiment 

 by a long way. If we take into account the figures of proportion for the 

 single positions we obtain as a rule a much better approach to 

 the figures observed by means of the products than with the sums, 

 even in the case where the two substitnents present are unequal, when 

 Huisinga's method of calculation cannot be applied. The proof there- 

 of is laid down in the subjoined table which gives the figures of 

 pi'oportion in which the isomeric nitroderiva tives are formed from 

 the substances at the top of the columns, with the figures obtained 

 from both the sums and the products. 



found 16.0:84.0 8.7:91.3 19.7:80.3 11.4:88.6 



product 17.7:82.3 17.7:82.3 23.3:76.7 23.3:76.7 



A fuller account of this investigation will appear in the Recueil. 

 Amsterdam, org. lab. Univ. 1906. 



Astronomy. — ''The relation hetioeen the spectra and the colours 

 of the stars." By Dr. A. Pannekoek. (Communicated by 

 Prof. H. G. VAN DE Sande Bakhuyzen). 



(Communicated in the meeting of September 29, 1906). 



The close relation between a star's colour and its spectrum has 

 long been known. The stars of the l^S 2^ and 3^ types are usually 

 called the white, the yellow and the red stars, although accurately 

 spoken the colour of the so-called yellow stars is a very whitish 

 unsaturated yellow colour and that of the so-called red stars is deep 

 yellow mixed with very little red. In a paper read at Dusseldorf ^) in 



') Die Farben der Gestirne. Mittheilungen der V. A. P. Jahrg. 10. S. 117. 



