1 8 INFRA-RED ABSORPTION SPECTRA. 



tion can be derived. It has been noticed that as the colHmator arm is 

 rotated the different wave-lengths must emerge from the prism at a con- 

 stant angle in order to be reflected into the radiometer slit. Hence we 

 need to compute this quantity, sin i, but once, viz, for the sodium lines. 

 Reverse the rays. Then from the relations, sin i{=n sin r^, 1'^=^ — r^, 

 and sin i. = n sin ra, we obtain the value of ig very readily, while the 

 deviation is 8 = j'2 — c, where c = <j) — i^. 



The indices of refraction of rock salt used in these computations are 

 due to Rubens and Trowbridge,^ to Paschen,- and to Langley.^ The 

 indices of refraction of quartz are due to Rubens.* 



ADJUSTMENT OF LARGER SPECTROMETER. 



The question of harmonic relations among the absorption bands 

 depends so much upon the accuracy of the adjustment of the apparatus 

 that it will be necessary to go into details, even at the risk of being 

 prolix. After struggling with the question of harmonics for a year 

 and a half, in which the prism has been remounted and adjusted a great 

 many times, it was found that the use of the prism by the non-minimum 

 deviation (" constant emergence ") method is precarious, since a slight 

 error in the adjustment of the setting for minimum deviation of the 

 sodium line may cause a great error in the infra-red. However, this 

 is not so difficult as obtaining the " zero setting," in which it is neces- 

 sary to project the image of the sodium lines within the radiometer slit. 



Accordingly, the large spectrometer was set up. It was used in two 

 ways, viz, the constant emergence method already described, and also 

 the mirror-prism method of Wads worth (loc. cit.), in which the prism 

 table is rotated and the spectrometer arms remain fixed. In the latter 

 all the different rays pass through at minimum deviation, after setting 

 for minimum deviation of the sodium lines. 



The long spectrometer arms, i meter, make the apparatus unwieldy, 

 and since the prism and its accompanying mirror (fig. 2) were also 

 large, all the adjustments were made after mounting the latter upon 

 its table. 



After adjusting the faces of the prism about a vertical axis by reflect- 

 ing the beam from the collimating slit back on its own path and forming 

 an image of the slit upon the slit, from the two faces, the prism and 

 mirror faces were made parallel in the same manner by tilting the mirror 

 in such a manner, by means of adjusting screws, that the beam reflected 



^Rubens & Trowbridge ; Ann. der Physik (3), 61, p. 224, 1897 '■ also Rubens : Ann. 

 der Physik (3), 53, p. 267, 1894. 



^Paschen : Ann. der Physik (3), 53, p. 337, 1894. 



^Langley : Annals Astrophys. Obs., vol. i, p. 258. 



*Rubens : Ann. der Physik (3), 45, p. 260, 1892 ; 46, p. 529. 1892. 



