Celestial Chemistry. 397 



vapours forming the stellar atmosphere, and which correspond 

 to tbe bright lines they would form in an incandescent state, 

 be strongest and most numerous in the more refrangible portions 

 of the spectrum/ then the star would have a red or orange tint, 

 because this part of the spectrum would suffer least absorption; 

 while, on the contrary, should the red and yellow portions 

 have most lines, the blue and green rays would predominate in 

 the colour of the star. The frequency of the lines in the 

 orange, green, and blue of Aldebaran's spectrum, previously 

 pointed out, is strongly in favour of this theory, as the red is 

 left little affected, and the other tints are subdued by the dark 

 lines, a Orionis has the red, green, and blue rays much dimi- 

 nished, which produces the orange colour of this star; and 

 ft Pegasi, which much resembles a Orionis in its spectrum, has 

 a deep yellow hue. In Sirius, which is of a brilliant white, 

 there are no lines sufficiently intense in any particular part of 

 the spectrum to interfere with our receiving the light in about 

 the same proportion, as to the quantity of the different coloured 

 rays, to that in which it starts from the incandescent light- 

 giving surface. 



It became a matter of great interest to test this theory 

 with respect to the double stars, which present the most 

 marked difference in colour; but the faintness of the blue or 

 green companions rendered the observations very difficult. 

 Still, with the second pair of prisms, which because of their 

 less dispersive power give a more brilliant spectrum, observa- 

 tions of ft Cygni and a Herculis were obtained, which accord 

 remarkably with the theory just propounded. Thus, in the 

 bright star of ft Cygni, which is light orange in colour, the 

 blue and green rays are full of lines, while the few in the red 

 and yellow are far apart; in the small star, on the other hand, 

 which is deep blue, the red and orange part of the spectrum is 

 full of groups of fine lines that interfere with those rays, and 

 leave the blue end, which has lines few and far between, domi- 

 nant in the light of the star. The colours of the components 

 of a Herculis, which are severally reddish and green, appear 

 to be produced in strict conformity to the same hypothesis ; 

 and although the theory cannot be considered as established by 

 these instances, it must be regarded as exhibiting a high 

 degree of probability, while the singular variability in the 

 colours of stars at different times, which cannot fail to occur 

 to our readers as not at present explained, may yet prove 

 to be capable of some elucidation by further investigations 

 in the same direction. Of course change in the chemical 

 constitution of the investing atmospheres would be an ob- 

 vious cause of change of colour; but these latter changes 

 are too frequent, and recur with such unvarying regularity 



