and Polarization of Heat. 157 



non-luminous heat, and with incandescent platinum it was ex- 

 tremely faint. My subsequent experiments gave for the propor- 

 tion of the depolarizing effect to the whole heat which reached 

 the pile when the plates E and F were crossed, 



X on-luminous Heat. Incandescent Platinum. Argand. 



.00 .016 .03 



But upon performing this experiment with a thicker plate, namely 

 that before alluded to in (53) and (54), I found that where it was 

 interposed between the crossed polarizing and analyzing plates, the 

 quantity of heat which reached the pile was increased by that inter- 

 position by about 0°.5. Hence we have the singular spectacle of 

 the transmission of heat being greater when a thick obstacle is 

 interposed, whilst the direct effect is actually diminished by the 

 interposition of a thin one. This effect was of the most marked 

 character with heat from incandescent platinum ; with dark heat 

 the result was quite analogous, but within narrower limits. With 

 unpolarized dark heat, I found that the thin plate stopped 30 

 out of 100 rays, whilst the thick one stopped 65, or more than 

 twice as much. 



57. The depolarizing effect of mica was tried under every 

 variety of circumstance, and with the most conspicuous and co- 

 incident results. The quantity of light accompanying the heat, 

 appeared by no means to regulate the quantity of heat depola- 

 rized. The heat emitted from platinum, of a full red, (and 

 therefore not vividly incandescent), was one of the most favour- 

 able. Heat from an Argand lamp, with glass chimney, was also 

 employed, and absolutely non-luminous heat from brass about 

 700°. I also employed mercury in an iron vessel, at about 500°, 

 and found the results admirably marked. Pursuing the experi- 

 ment as the temperature of the mercury descended, I found the 

 effect still very sensible at 220°, and then thought of trying hot 

 water, which I had not done since I devised the telescopic 

 method of observing the galvanometer, (6). The result was, 



