394 DISTRIBUTION OF HEAT IN THE SPECTRUM. [MEMOIR XXVIII. 



be obstructed, a remark that applies especially to the in- 

 visible rays of less refrangibility than the red. For this 

 reason the mirror, d d, must be placed at such an obliquity 

 to its incident rays as to throw the focal images sufficient- 

 ly on one side. Yet this obliquity must not be greater 

 than is actually necessary for that purpose, or the purity 

 of the second spectrum, with its Fraunhofer lines, will be 

 interfered with. At the place of the third focus, arising 



from the reunion of the 

 dispersed rays, is the ther- 

 mopile, <7, connected by its 

 wires, k k, with the multi- 

 plier, m. 



Whenever any of the 

 visible rays of the Fraun- 

 hofer spectrum are inter- 



cepted by advancing either 

 of the screens h, i, the im- 

 age on the face of the pile 

 ceases to be white. It be- 

 comes of a superb tint, an- 

 swering to the combination of the non-intercepted rays. 

 A slip of white paper placed for a moment in front of the 

 pile will satisfy the experimenter how r magnificent these 

 colors are. It is evident, therefore, that by this arrange- 

 ment the pile will enable us to measure the heat of any 

 particular ray, or of any selected combination of rays. 

 The screens can be arranged so as to reach any desig- 

 nated Fraunhofer line. 



The pile I have used is of the common square form ; 

 a linear pile would not answer. The focal image on the 

 pile is of very much greater width than the slit , on ac- 

 count of the obliquity of the front face of the prism. 



By removing the screen 7^, and placing the screen i so 

 that its edge coincides with the line A of the Fraunhofer 



