234 ROYAL SOCIETY OF CANADA 



localities of electric discharges. Quite often, too, the emission and ab- 

 sorption lines of the same element show different displacements, and 

 frequently the lines due to various elements are shifted differently: 

 phenomena which may be caused by convections of different character 

 in different localities in the star's atmosphere, as illustrated in the case 

 of the sun. 'There will, of course, be shifts in the spectral lines due to 

 the velocity of the star as a whole and in general the " velocity-curves " 

 will be complex, containing the two components due to atmospheric dis- 

 turbances and to motion of the star. In cases where there are shifts in 

 the lines, all of which agree closely with the assumption of orbital mo- 

 tion of the source of light, without accompanying changes in the character 

 of the lines or in the intensity of the light of the star, it is safe to sup- 

 pose that the shift is entirely due to orbital motion of the star. On the 

 other hand, where there are accompanying variations in the character 

 and intensity of the light it seems reasonable to assume that the line-shifts 

 are due, entirely or partially, to atmospheric disturbances particularly 

 when these shifts do not agree closely with the assumption of motion 

 of the star in an orbit. 



(4) About a dozen solar-type stars have been investigated which 

 have maximum luminosity very nearly when the spectral lines are of 

 minimum, wave-length, while minimum luminosity and greatest wave- 

 length of the spectral lines occur nearly together, the curves of changing 

 light and changing wave-length being almost identical in appearance^- 

 This class is illustrated by the curves in Fig 3, which shows the velocity- 

 curve and light-curves of IF Sagittarii^ plotted with the maximum 

 brightness and maximum negative velocity (minimum wave-light) up- 

 ward showing the slow falling away to minimum brightness and maxi- 

 mum wave-length which occur siniultaneously. An additional curve 

 gives the light-intensities in terms of the light of a tenth-magnitude 

 star. As shown before (§5) this phenomenon can be explained by the 

 changing action of a satellite which affects the velocity of the convection 

 currents thus affecting simultaneously the light of the star and the posi- 

 tion of the spectral lines : the lines having maximum frequency of vi- 

 bration or minimum wave-length at or near the time of greatest lumin- 

 osity, the wave-length and luminosity corresponding throughout the 

 period since they are both dependent on the same variable. JN early all 

 of these stars show the slow falling away from maximum brightness 

 whicli is presumably due to the inductioin-effect. 



(5) A large proportion of the stars whose periodic spectral-line 

 displacements have been studied show a greater time of increasing than 



1 Albrecht, L.O.B., ^, No. 138, 1907; Astrophysical Journal, 25, 330-348, 

 1907. Duncan, L.O.B.. 5. No. 151, 1009. 



2R. H. Curtiss Astrophysical Journal, 20, 149-181, 1904; 22, 274-275, 1905. 



