DESCRIPTION OF APPARATUS AND METHODS. 1/ 



employed. Its refracting angle was 59° 57' 43" for the center of the 

 faces. The prism was securely and accurately mounted upon the spec- 

 trometer table, and a dish containing phosphorus pentoxide was placed 

 over it. When not in use it was covered with a glass bell- jar. as a 

 further protection against moisture. 



The whole apparatus was inclosed in a tin box to exclude air cur- 

 rents, moisture, and stray radiation from the radiometer and prism 



(fig. I a). 



With the small spectrometer the prism was set at minimum deviation 

 of the sodium lines by means of a Gauss eye-piece, which was placed in 

 the focal plane of the second mirror. 



ADJUSTMENT AND CALIBRATION OF APPARATUS. 



Since the radiometer must remain stationary, the successive portions 

 of the spectrum must be projected upon its vane by one of two methods, 

 viz, by rotating the prism, or by rotating the collimating arm of the 

 spectrometer. This involves the problem of keeping the prism at mini- 

 mum deviation. With the small spectrometer it was not possible to 

 have the different rays pass through the prism at minimum deviation 

 by rotating the prism table as in the mirror-prism device described by 

 Wads worth. ^ An automatic attachment, like that used by Paschen and 

 others, continued to give trouble, so it was discarded and the colli- 

 mating arm was moved, while the prism remained stationary after set- 

 ting it for minimum deviation of the sodium lines. As a consequence, 

 when the collimating arm is revolved about the prism, the different 

 wave-lengths emerge from the prism at a variable angle, which is, of 

 course, no longer the minimum deviation angle, except for the sodium 

 lines. This makes the computation of the dispersion curve more com- 

 plex, since we can use the minimum deviation formula 



i=sin(^±i) 

 2 \ 2 ^ 



n sm 



2 V 2 



only for the sodium lines. The appropriate formula as used by Asch- 

 kinass^ is 



i\ ^arc sin (sin cf> \/n- — sin^ i^ — cos <f> sin (2) 

 w^here z\ is the angle of emergence, <f> is the angle of the prism, n is the 

 index of refraction, and /, is the angle of incidence for the new wave- 

 length. The deviation, 8, from the sodium line is then, 8=^0 — ii, where 

 the sodium line is at minimum deviation. This formula is very un- 

 wieldy, and I preferred computing the calibration (dispersion) curve 

 from the following simpler relations, from which the aforesaid equa- 



iWadsvvorth : Phil. Mag. (5), 38, p. 346, 1894. 

 ^Aschkinass : Ann. der Physik (3), 55, p. 406, 1895. 



