232 , ROYAL SOCIETY OF CANADA 



of the satellite is comparable in magnitude to the diameter of the .tar 

 the two disturbances on the opposite sides of the star will be unequal 

 and consequently two unequal maxima in the light of the star may be 

 observed, the greater when the satellite is acting towards the observer 

 and the smaller when it is acting away from him. This may account 

 for the secondary maximum in the light-curves of some variables. 



§ 7. 'The general results of the investigations of variable stars may 

 be explained by this hypothesis as follows : 



( 1 ) Stars which show symmetrical variation in luminosity are near- 

 ly all precise short-period variables of the hydrogen and helium types. 

 In this case it is assumed that on account of the low density and mobility 

 of hydrogen and helium and of the probably very great depth and rarity 

 of the atmosphere in this type of star the atmospheric changes respond 

 readily to the changing attraction of the satellite, the induction effect 

 being' practically negligible, that is, JS" = P + 7 = F, Of course there 

 are a great number of short-period symmetrical variables which do not 

 vary continuously and which are perhaps best explained by the eclipse 

 theory. 



(8) Stars which have asymmetrical or irregular variation are solar- 

 type or red stars, that is, stare which contain in their atmosphere the 

 vapors of metals and even of chemical compounds as well as the lighter 

 gases. Nearly all the variables of this class spend the greater part of 

 the period decreasing in luminosity, a fact which is explained by as- 

 suming that the induction effect is considerable on account of the pres- 

 ence of the vapors of metals and possibly of matter in different phases. 

 There are three distinct divisions in this class: (a) Variables of long 

 period, or "month-variables," that is those whose periods are reckoned 

 in months; (b) Cepheid variables or " day- variables," whose periods are 

 several days; (c) Cluster- variables or "hour-variables," whose periods 

 are a few hours. These divisions are represented by the light-curves^ 

 shown in Fig. 2. The variables of division (a) exhibit irregularities in 

 their periods and in the extent of their variations, while those of divi- 

 sions (b) and (c) are very precise in the lengths of their periods and 

 in their light-changes. This is explained by supposing that in the case 

 of the " month-variables " there may be several satellites acting in dif- 

 ferent periods, or that the changes in attraction are very slow and small 

 so that the fitful disturbances in the atmospheres become more apparent, 

 or that as in the sun, the spotted or disturbed areas are distributed 

 irregularly in longitude and as the star rotates the light emitted is of 



1 Light-curve (a) is from Miss Agues Clerlie's "The System of the Stars; 

 (b) Pickering, Annals, Harvard College Observatory, 46, 1903; (c) Bailey, 

 Ibid., 38, 1902. 



