20 INFRA-RED ABSORPTION SPECTRA. 



From the equations, 



k =(90°—/)) + (90°— 0= i8o°— r — - and r = ;n +* 

 we obtain the desired relation, p = (f> -\- m — i, where (f> is the angle of 

 the prism, m is the angle between the mirror and opposite face of the 

 prism, i is the computed angle of incidence of sodium light, for mini- 

 mum deviation, and p is the required rotation of the prism table. All 

 we have to do, then, is to turn the prism-mirror table so that the beam 

 from the slit returns on its own path and forms an image of the slit upon 

 the slit. Then rotate the prism table through the angle p, and the beam 

 will be reflected from the mirror and enter the prism at the proper angle 

 of incidence for minimum deviation. 



A second method of minimum deviation adjustment used by Stewart^ 

 was also tried, and the results obtained agreed with the preceding. 



Using these three methods, viz, the non-minimum deviation (con- 

 stant emergence, as with the small spectrometer) method, and Wads- 

 worth's mirror-prism device for constant mimimum deviation, which 

 was adjusted by the two methods just mentioned, the results obtained 

 were in excellent agreement for the large spectrometer. But when 

 these results were compared with the work done with the small spec- 

 trometer it was found that all the maxima, of a certain number of com- 

 pounds, were shifted by a constant slight amount toward the longer 

 wave-lengths, which would have been unimportant had it not been for 

 the question of harmonics. Now, the work with the small spectrometer 

 had also been found consistent. This was established by the repeated 

 examination of certain sharp emission and absorption bands which 

 were equivalent to the comparison spectrum in the optical region. 

 After weeks of intercomparison only one explanation of this discrep- 

 ancy could be found, viz, that on account of the slight curvature of the 

 image of the slit found the small spectrometer, in projecting the sodium 

 lines upon the radiometer slit, which served as a " zero reading " in 

 testing the constancy of the adjustments and as a starting point in cali- 

 brating the apparatus, the whole spectrum was thrown forward about 

 30". Consequently every spectrometer circle reading represented a wave- 

 length situated 30" farther toward the infra-red than that indicated on 

 the calibration curve. Applying this correction to about a dozen com- 

 pounds, the location of the maxima, as found with the small apparatus, 

 coincided exactly with that found v/ith the large spectrometer. 



In the discussion of the curvature of the slit, in Kayser's Spectroscopy 

 it is shown that such a shifting toward the long wave-lengths, as noticed 

 above, is likely to occur. 



^Stewart : Phys. Rev., xin, p. 257, 1901. (The details of his method of adjust- 

 ment were given the author in a private letter.) 



