138 THE EVOLUTION OF 



accuracy. It is actually represented in fig. 3 l . The 

 period of revolution is fifty-two years. When first dis- 

 covered the companion was approximately in the position 

 a in the figure. For many years it remained visible 

 through telescopes of high quality as it traced the part of 

 its orbit remote from Sirius, but in the year 1890 as it 

 passed b it disappeared to the finest instruments in the 

 world as its light was lost in the brilliant rays of the star. 

 From these it has now emerged and is again visible at c. 



By repeatedly measuring the position of the companion 

 with respect to Sirius we have obtained its relative orbit. 

 If, however, we record the positions of both star and 

 companion with reference to fixed stars in the same tele- 

 scopic field, we may obtain the actual orbit of each orbit. 

 This has been done, with the result shown in fig. 4. It 

 appears from this that the distance of Sirius from, the 

 common focus O of the ellipses is at every instant one half 

 that of the companion, from which we arrive at the con- 

 clusion, assumed in our previous study of the figure, 

 that the mass of Sirius is double that of the com- 

 panion. The result is remarkable in view of the fact that 

 the star gives more than 30,000 times the light of the 

 companion. 



The period of mutual revolution of two attracting masses 

 increases with their distance apart, or, more accurately, 



1 The apparent orbit as actually traced from observation is not that 

 of fig. 3. In the case of a body describing an elliptical orbit round 

 a centre of attraction as focus, the actual form of the orbit as seen 

 from a distance will only be preserved if it is presented squarely to the 

 line of vision. If oblique, the orbit is seen in perspective. It still, 

 however, appears elliptical, but the degree of ellipticity is changed. 

 Happily, however, the attracting star no longer appears in the focus, 

 and this fact enables us to tell whether the orbit is tilted, while the 

 displacement of the star from the focus of the perspective ellipse enables 

 us to determine the extent of tilt and to allow for it. The figures 

 illustrating the lecture are those of the actual orbits, the distortion due to 

 tilt having been corrected. 



