p. LECOMTE DU NOtJY 751 



rotation of the prism table), and the use of the dispersion formula, 

 given by Paschen (18): 



Then, the range of the spectrum embraced by a given sHt was checked 

 by moving a spectrum line across the slit, and reading the result 

 on the drum. The right edge of the D lines (sodium), for instance, 

 was brought in contact with the right edge of the sht (0.25 mm. 

 opening), and the reading made. Then it was moved toward the 

 left until the whole D line just disappeared, and another reading niade. 

 The result was 0.008 /x- This was done for the lines of copper 

 (0.4955 fi, 0.5292 /x), mercury (0.5461 m), sodium (0.5893 ju), and cad- 

 mium (0.6439 fj). For the infra-red, the data are pubHshed by 

 Hilger (19). 



It was decided to take the area covered at 2.66 ^ by a slit 0.025 

 mm. wide as unit, (0.026 fx) and to fix the slit in such a way, for every 

 wave length, as to cover the same range. A high sensitivity Leeds 

 and Northrup galvanometer was used in connection with the ther- 

 mopile, (galvanometer resistance = 12 ohms). 



Corrections of the Absorptions Due to the Spectrometer. — But, before 

 integrating the plotted curve, another very important correction 

 had to be introduced regarding the absorption by the golden mirrors, 

 because the energy distribution curve does not correspond to the 

 total amount received by the collimator slit, and because the ab- 

 sorption is much greater for short than for long wave lengths. Fig. 

 1 shows how the absorption varies for different wave lengths. 



If R is the coefficient of reflection, and 3 the number of mirrors, 

 the amount of energy reflected is expressed by 



I = loR' 

 It is easily seen that for 0.5 ju, for example, only — of the inci- 

 dent light is transmitted, and much less still for 0.4 ijl. 



The absorption by the prism amounts to very little. Theoret- 

 ically, from the formula, / = JoK", where c is the length of the path 

 of light in the rock salt and K a constant equal to 1 between and 

 9 {JL, it is equal to zero. We found that, practically, for the visible. 



