OF AUT5 AND SCIENCES. 371 



other hand, a photometric measure has a great advantage on the score 

 of independence, as it may be repeated many times in succession in a 

 single evening. An observer cannot repeat a naked-eye conifjarison 

 many times without being prejudiced in the kiter measures by those 

 whicli have preceded them. 



The photometer used in the following measurements is essentially the 

 same as that described in volume xi. of the Annals of the Harvard Col- 

 lege Observatory, p. 4, figs. 1 and 2. A double-image prism is placed 

 between the object-glass and eyepiece of a telescope, and in front of the 

 eyepiece a Nicol prism is inserted. A circle divided into degrees 

 serves to show the angle through which the Nicul is turned. Two ad- 

 jacent stars may be compared by this instrument with great accuracy. 

 Two images of each will be formed by the double-image prism, and 

 their relative brightness varied at will by turning the Nicol. Each 

 image in turn will disappear when the Nicol is revolved 90°. There 

 will therefore always be four positions in which the brighter image 

 of the fainter star will be precisely equal to the fainter image of the 

 brighter star. 



/3 Persei. 



The measurements of (3 Persei were made by comparing it with oj 

 Persei, a fifth magnitude star 90' distant. The two images were 

 formed by two Rochon prisms, which produced a separation of about 

 100'. It was therefore necessary that they should be placed very 

 near the object-glass of the telescope, in order that the images of the 

 two stars should be near together. The focal length of the telescope 

 is about seventeen inches, and its available aperture is limited by the 

 size of the prisms to about an inch. A larger aperture would be 

 preferable, but is not very important on account of the brightness of 

 the stars. The telescope is placed horizontally with a right-angled re- 

 flecting prism in front of the object-glass. The line of sight is there- 

 fore horizontal, even when objects in the zenith are observed, and 

 the stars are followed by rotating the telescope around its axis, and 

 also by turning the stand around a vertical axis. The four images 

 are placed in line, and the two central ones, which are compared, are 

 reversed by moving the prisms to and from the object-glass by a handle 

 attached to the tube carrying them. This reversal was essential, 

 and was always made after the second setting in each set. Errors 

 due to the position of the images are thus com{»letely eliminated. As 

 the two images are seen on the same background under precisely the 

 same conditions, many sources of error are eliminated. The couven- 



