396 DISTRIBUTION OF HEAT IN THE SPECTRUM. [MEMOIR XXVIIF. 



of the instrument should be set to this. The other, 

 which may be termed the working zero, is formed by 

 coupling the pile and the multiplier together, and intro- 

 ducing the screen n between the intromitting slit and 

 the prism. On doing this it will probably be found 

 that the index will deviate a few divisions. Its po- 

 sition should be accurately marked at the beginning 

 and close of each set of measures, and the proper correc- 

 tion for them made. The disturbing influences of the 

 mass of the prism, of the mirror, and of the pile itself, 

 are thus eliminated. As respects the last, it should not 

 be forgotten that it may be affected by changes in the 

 position of the person of the experimenter himself. 

 With the intention of diminishing these errors, I have 



O ' 



usually covered the upper and lower portions of the con- 

 cave mirror, dd, with pieces of black paper, so arranged as 

 to leave a band of sufficient width to receive and reflect 

 the entire spectrum, a a, Fig. 89, is the upper paper, b b 

 the lower, c c the uncovered reflecting 

 band, receiving the spectrum, r v. Had 

 the spaces thus covered been permitted to 

 reflect, they would have rendered more in- 

 tense the image of the prism with its ex- 

 Fl 'g- 89 - traneous heat. 



As regards the multiplier, care must be taken to avoid 

 disturbance from aerial currents. I have one of these 

 instruments of French construction, which could not be 

 used in these delicate researches until proper arrange- 

 ments were applied. It was covered with a glass shade. 

 The slightest cause occasioned currents in its included 



O 



air, which perpetually drifted and disturbed the needles. 

 For this reason, and also for more accurate reading, it is 

 best to view the position of the index through a small 

 telescope. 



The combination of needles being nearly astatic, at- 



