'9>2.] SCHLESINGER— RADIAL VELOCITIES. 181 



application is a matter involving serious difficulties. A number of 

 modifications have been proposed with a view to obviating one or 

 another of these difficulties. Thus Stewart" suggested that, instead 

 of making exposures upon two dififerent plates, two objectives and 

 two oppositely placed prisms be employed for making two simul- 

 taneous exposures side by side upon the same plate. The plate 

 would then have to be inclined equally to the two incident beams, 

 and, under ordinary circumstances, it would be possible to secure 

 only one pair of spectra upon each plate. Another modification is 

 that due to Comstock," who proposed that the two halves of the 

 objective be covered by two prisms having their refracting edges 

 turned toward each other. These prisms are to be compound and 

 of the " direct-vision " variety, so as to yield for each star a pair of 

 spectra in close juxtaposition. Here, however, we encounter the 

 difficulty of securing prisms of sufficient size, for prisms of this 

 description would have to be very thick in order to give spectra of 

 sufficient dispersion. The present writer has suggested still another 

 modification,® in which the two plates are taken simultaneously by 

 means of two independent (but similar) cameras and prisms, both 

 being enclosed in a constant temperature case provided with two 

 suitable windows of plane parallel glass. 



The advantages of securing the two photographs at the same 

 time are two- fold; first, this obviates any necessity for considering 

 refraction, a very bothersome matter when large fields are in ques- 

 tion. Second, the so-called guiding error is eliminated. Whether 

 the telescope is driven entirely by clockwork, or whether the observer 

 attempts to secure more perfect guiding by introducing slight cor- 

 rections by hand, the spectra will still wander a little from their 

 mean positions and the place at which an observer will bisect a 

 line in the spectrum will depend somewhat upon the nature of the 

 guiding. If, however, the direct and the reversed spectra are se- 

 cured simultaneously, the guiding error will be the same for both 

 and will have no efifect upon the derived velocity. 



In this method of determining velocities there remains a difficulty 



'^ Astrophysical Journal. 23, 396, 1906. 

 ^ Astrophysical Journal, 23, 148, 1906. 

 * Science, 30, 729, 1909. 



