^3 J ROYAL SOCIETY OF CANADA 



of decreasing wave-length of the spectral lines.^ Au explanation of some 

 of the cases which constitute this excess over the probable number may 

 be found in the idea of an induced action. 'That is, let the decrease in 

 wave-length be attributed to increase in the velocity of the atmospheric 

 convection currents, the minimum wave-length corresponding to the 

 maximum velocity of these currents at the period of maximum activity 

 in the atmosphere and the induced eifect following this condition and 

 tending to prolong it as already explained (§4 and §5). There will thus 

 occur a slow falling away to minimum velocity of the convection cur- 

 rents and with it a corresponding slow increase in wave-length of the 

 spectral lines. Thus if in any of these cases the shifting of the lines 

 is due to atmospheric changes instead of being due, as ordinarily sup- 

 posed, to orbital motion of the source of light, the idea of an induced 

 action would explain why it takes longer to pass from the time of mini- 

 mum to the time of maximum wave-length than it does to pass from 

 maximum to minimum wave-length, that is, why the star is apparently 

 moving longer towards than away from the observer. 



Tliere are a considerable number of " spectroscopic binaries," whose 

 periodic variations in light — if such exist — are not great and it is nat- 

 ural to expect changes in luminosity to accompany considerable line- 

 displacements. Possibly the presence of a very deep absorbing atmos- 



m. M 



phere may account for this ; or — since the " tidal effect," — ^ — , involves 



the mass, m, of the revolving star as a direct factor, while the relative 



velocity,-! /(,7i -f j^)( ^ )' contains it only as a term in a factor, and 



further sinoe in the former expression the dimension of r is — 3 while in 

 the latter it is — y^, the assumption of relatively greater brightness of the 

 revolving star combined with relatively small mass — m much smaller 

 than M — or with considerable separation of the two bodies, may explain 

 why the velocity is appreciable while the variations in brightness are not. 



§8, There are also a number of particular cases such as the light- 

 variations and line-displacements of Novae which may be explained by 

 this convection hypothesis. 



I wish to express here my gratitude to the Director, Dr. King, for 

 his ki*ndness awd suggestions. 



Dominion Observatory, Ottawa, May, 1909. 



1 Noted by J. Millpi- Barr who gives plausible explanations of the pheno- 

 menon in Jour. Roy. Asfron. Soc. Can. 2, 70-81. 1908 : 3, .50-53. 1909. Since the 

 publication of these papers the proportion has been reduced by the study of more 

 cases. The convection hypothesis offers an alternative explanation. 



