Third Series. — Variable Polarizahility of Heat. 105 



little, and in no sort of proportion to the characteristic action 

 of mica even in moderate thicknesses. This will be more 

 evident, if we compare the ratios of the heat from different 

 sources transmitted in the two cases, taking the heat from 

 the lamp sifted by glass as the standard for each column. 



Plates I and K. Mica -016 inch. 



Locatelli with glass . . 100 100 



Locatelli 116 79 



Incandescent platinum .108 70 



Brass at 700° .... 96 21 



Heat of 212° .... 62 11 



I need hardly add, that so remarkable a result as that the 

 heat sifted by glass should be less readily transmitted by the 

 thin mica laminae, than the direct heat from a lamp, was care- 

 fully verified. 



Since, then, the first four kinds of heat are transmitted 

 without any great difference of proportion, by the piles I and 

 K, and since, especially, the heat from a lamp sifted by glass 

 and that from dark brass possess almost exactly similar cha- 

 racters in this respect, it is very clear that we have a new 

 ground for rejecting as untenable M. Melloni's supposition, 

 that the apparent differences of polarization in my experi- 

 ments, arose from the unequal proportions of heat absorbed 

 by the mica piles when the source varied. 



Admitting, then, the fact of the variable index of polariza- 

 tion exhibited by heat of different qualities similarly treated, 

 we are tempted to inquire what explanation can be offered 

 of it. This question, inferring for its answer a knowledge 

 of the nature of heat, we are not prepared to answer with 

 confidence. My former suspicion of its being due solely to 

 the difference of the refractive index of mica for heat of dif- 

 ferent kinds, I am disposed to retract as inadequate, or at least 

 to suspend my judgement respecting it. I at one time thought, 

 that, supposing the mica bundles unequally permeable to heat 

 from different sources, a difference of ratio in the total heat 

 reaching the pile with the plates I and K^ parallel and crossed 

 might be accounted for. But a careful analysis of the cir- 

 cumstances convinced me, that the absorptive action, if as- 

 sumed the same for common and polarized heat, could pro- 

 duce no such effect. One of the most plausible suppositions 

 which occurred to me was this, — that, supposing the reflec- 

 tion of luminous heat to take place more copiously at the 

 mica surface than that of dark heat, and supposing the angle 

 of incidence to be that of total polarization, since the refracted 

 ray contains as much heat (if heat be like light) polarized 

 perpendicular to the plane of incidence, as is reflected and 



