MI.M..IK XXVIII.] DISTRIBUTION OF HEAT IN THE SPECTHUM. 395 



spectrum, all the invisible heat radiations of less refran- 

 gihility than the red are cut off, except the contaminat- 

 ing ones arising from the general diffusion of light by 

 the substance of the prism. Under these circumstances 

 the image on the pile will be white, and the multiplier 

 will give a deflection representing the heat of the visible 

 and the extra-violet regions. If then the screen be ad- 

 vanced still farther, until it has intercepted all the less 

 refrangible regions up to the sodium line D or a little 

 beyond that is, to the optical centre of the spectrum 

 the tint on the face of the pile will be greenish - blue, 

 and the multiplier will give a measure of the heat of 

 the more refrangible half of the visible spectrum, to- 

 gether with that of the ultra-violet rays : the latter por- 

 tion may, however, be eliminated by properly using the 

 other screen, k. 



Besides the error arising from stray heat diffused 

 through the spectrum in consequence of the optical im- 

 perfection of the prism, there is another which may be 

 recognized on recollecting the relative positions of the 

 prism, the concave mirror, and the face of the pile. It is 

 evident that the prism, considered as a warm or a cool 

 mass, is a source of disturbance, for the mirror reflects its 

 image that is, the image of the prism itself to the pile. 

 After the intromitted sunbeam has passed through the 

 prism for a short time, the temperature of that mass has 

 risen, and the heat from this source has become inter- 

 mingled with the proper spectrum heat. But this error 

 is very easily eliminated. It is only necessary to put a 

 screen, n, in the path of the incoming ray, between the 

 slit and the prism, and note the deflection of the multi- 

 plier. Used as we are here supposing, the multiplier 

 has two zeros. The first, which may be termed the 

 magnetic, is the position in which the needles will stand 

 when no current is passing through the coil. The scale 



