Spectra of some of the Fixed Stars. 517 



contrasted with those of the stars just mentioned. It may be 

 that the length of the stellar atmosphere through which the 

 light passes is less, relatively to the intensity of radiation from 

 the photosphere, and so is insufficient to produce lines of the 

 same degree of blackness as would be produced if the atmosphere 

 were denser. The great intensity, however, of the light of Sirius 

 would rather lead to the conclusion that the atmosphere of va- 

 pours is itself highly incandescent. If so, might it not to some 

 extent replace with its own light the light which it has absorbed 

 from the photosphere behind it ? It matters little, however, for 

 the present purpose, whether or not either of these suppositions 

 be adopted. There is at all events a most striking difference be- 

 tween the effect on the colour of the star of the closely grouped 

 and very dark lines in the green and blue portions of the spectrum 

 of u Orionis and of the corresponding portion of the spectrum of 

 Sirius, in which the dark lines are faint and wholly unequal to 

 produce any noticeable subduing of the blue and green rays. 



We have not yet had an opportunity of testing by experiment 

 whether this hypothesis of the origin of the colours of the light 

 of the stars is also applicable to the remarkable exceptional class 

 of stars the light of which is of a decided green, blue, or violet 

 colour. Such stars are usually very small, and they are always 

 so closely approximated to other more brilliant stars, that it is 

 scarcely possible, with the apparatus which we employ, to obtain 

 separate images of the two spectra : and even were such separa- 

 tion easily practicable, the light of the strongly-coloured star is 

 usually so feeble that its satisfactory prismatic analysis would be 

 a matter of great difficulty. 



[One of the objects proposed in the construction of the spec- 

 trum apparatus with which the additional observations on Jupiter, 

 Saturn, and Mars were made, and which has been described 

 (p. 415) in connexion with those observations, was to make it 

 available for the prismatic observation of some double and mul- 

 tiple stars. 



Before commencing the observation of the spectra of the com- 

 ponents of a double star, it is necessary that the position-angle 

 of the stars should be approximative^ known. The spectrum ap- 

 paratus has then to be rotated upon the end of the telescope until 

 the direction of the slit becomes perpendicular to a line joining 

 the stars. When the instrument is in this position, the images 

 of the stars are elongated by the cylindrical lens into two short 

 lines of light parallel with the slit, and separated from each other 

 by a small interval. If the telescope be now moved in a direction 

 at right angles to that of the slit, either of the elongated stellar 

 images can, at pleasure, be made to fall upon the slit and form its 

 spectrum in the instrument. By adopting this method of ob- 



Phil. Mag. S. 4. No. 212. Suppl Vol. 31. 2 M 



