ROTATION OF THE GALAXY — EDDINQTON 245 



If the absorption occurs uniformly in the cloud, the mean distance 

 of the stretch traversed by the star's light should correspond to 

 halfway. The distance of the veiling cloud should, therefore, always 

 come out to be half the distance of the corresponding stars. A glance 

 at the table will show how closely this is fulfilled. To speak frankly, 

 I should have been better pleased to see more discordance, since 

 the closeness must to some extent be put down to rather outrageous 

 luck — as the authors indeed recognize. 



The transverse proper motions can be examined for differential 

 rotation in like manner, but I am skeptical as to whether they add 

 very much to the evidence. In treating the radial velocities we have 

 gained greatly by using the most distant stars and, granting that 

 sufficiently luminous stars can be found, there is no more difficulty 

 in determining radial velocities at 2,000 than at 20 parsecs distance. 

 But proper motions depend on measurements of arc, and what we 

 gain through the magnification of the effect by distance we lose in 

 the reduction from linear to angular displacement. The effect on 

 the apparent angular motion is thus the same at all distances and 

 remains always on the verge of what is detectable observationally. 

 However, so far as the evidence goes it is favorable to the theory. 

 An extensive investigation by Sir Frank Dyson gave correctly the 

 center in longitude 330°, but the magnitude of the differential rota- 

 tion was somewhat smaller than that deduced from the radial 

 velocities. He pointed out that the analysis also indicated certain 

 larger terms not explicable by rotation — a fact which seems to spoil 

 the significance of the result. 



When I decided to lecture on this subject, I thought I was going 

 to describe the newest of the various methods employed in our search 

 for information about the stellar system. I was mistaken ; it is the 

 oldest. Vixerunt fortes ante Agamemnona multi. Precisely this 

 method was employed by Gylden in 1871.^ Using the proper motions 

 then available he discovered the double-period term sm2{l — lo) 

 and explained it just as we have done. To make the conclusion 

 more convincing he made a test of the validity and efficiency of the 

 method by applying it to the apparent motions of the asteroids, 

 using them as a model for illustrating differential rotation as I have 

 used Saturn's ring. From the motions of the asteroids he deduced 

 the direction of the center of the system, viz, the sun; his error was 

 about 6°. That would be one way of finding the sun again if ever 

 it ceases to be visible ! Keturning to the stellar system Gylden re- 

 marked that a definitive determination of the center was not at 

 present practicable because the common rotational motion was pre- 



» Indications of Laws Governing Stellar Motions (in Swedish) Of vers. K. Vetensk. 

 FSrhandl., vol. 28, p. 947. I am Indebted to Professor Llndblad for the reference and 

 information. 



