176 SCHLESINGER— RADIAL VELOCITIES. [April 20, 



apertu'"es from twelve to fifteen inches have proven nearly as efficient 

 for the determination of radial velocities as those attached to tele- 

 scopes of two or three times their aperture. 



If, therefore, we are to seek any considerable extension of these 

 observations to fainter stars, we are forced to turn to the objective 

 prism, the great advantage of this form of spectroscope being its 

 economy of light. Few slit spectrographs now in use utilize more 

 than one per cent, of the light that falls upon the objective of the 

 telescope to which they are attached. With the objective prism, 

 as much as twenty-five per cent, may easily be realized with a proper 

 choice of materials in the objective and in the prism, and providing 

 also that the thickness of the prism is not too great. Furthermore, 

 with the slit spectrograph we secure the spectrum of only one star 

 at a time, whereas with the objective prism all the stars in a consid- 

 erable area of the sky impress their spectra upon the same plate. 

 On the other hand, the difficulty of securing points of reference from 

 which the shifts in the lines may be measured has thus far proved 

 an insurmountable obstacle in the way of obtaining even fairly accu- 

 rate velocities by this means. 



Spectra produced by difl^raction gratings are nearly normal ; that 

 is, the distance between any two lines is approximately proportionate 

 to their difference in wave-length. In such spectra the shift due to 

 velocity is greater for lines of greater wave-length. Consequently, 

 the spectrum is lengthened if the star is receding from us and is 

 shortened if the star is approaching. On the other hand, spectra 

 produced by prisms are far from being normal, lines in the region of 

 greater wave-length being crowded together, while those at the violet 

 end are more widely separated. In this case, shifts due to velocity 

 become reversed in their relative amounts, being now greater for 

 lines of shorter wave-lengths. Prismatic spectra are accordingly 

 shortened for receding stars and lengthened for those that are ap- 

 proaching us. We see, then, that the lengths of both prismatic and 

 diffraction spectra are changed by radial velocity, and accordingly 

 the latter can, theoretically at least, be found by measuring the dis- 

 tance between two lines widely separated in the spectrum. 



The possibility of determining velocities in this way seems first 



