350 THE WORLD MACHINE 



travel at nearly a thousand times the velocity of sound, but 

 its vibrations are crosswise to the line in which it travels. The 

 motion of the source of light does not occasion a change of 

 pitch that is to say, of the light colour but this movement 

 is disclosed by a slight displacement in the position of the lines 

 of the spectrum. 



The placing of the lines is determined by the wave-lengths 

 of the different rays. If the source of light be approaching, 

 the waves will be crowded a little closer together, with the result 

 that the lines will be pushed up a little farther towards the 

 violet end of the spectrum. Conversely, if the source of light 

 is flying away from us, with sufficient speed, the wave-lengths 

 will be drawn out a little, the spectral lines will be displaced 

 slightly towards the red end. By a marvellous refinement of 

 measurement, this displacement can be utilised to reckon the 

 speeds of stars moving directly in the line of sight that is to say, 

 which have a motion radial to that of the earth and the sun. 



This is accomplished by means of photography. If the 

 spectrum of a moving star be registered upon a photographic 

 plate, side by side with that of some source of light within the 

 laboratory, whose lines exactly correspond with those of the 

 star, the minute change of position due to the star's motion 

 may be observed. 



By this means it has been possible, to some extent, to check 

 the calculations as to the speed of the stars reckoned from the 

 measures of parallax and of proper motion across the line of 

 sight, or at least to render these calculations probable. If, for 

 example, the speed of a number of stars of about equal magni- 

 tude, and therefore probably of about the same distance from 

 the earth, be computed first by the earlier method, then as 

 many more by the spectral method, we may compare the 

 averages. If they roughly agree, it is evident that in neither 

 the one instance nor the other are we dealing with optical 

 illusions. The chances against this concurrence being the result 

 of a mere hazard has enormously increased as the number of 

 motions thus reckoned increases, and with this increase the 

 concurrence grows closer. Were it possible only to reckon the 

 speed of the stars by means of their parallax, it might still be 

 open to supposition that these computations were grounded 

 upon some periodic disturbance like that of stellar aberration 

 and nutation. Were the estimates grounded only upon the 



