Second Series. — Polarization by Refraction. 355 



out altering our views as to the probable identity of the cause of 

 both those physical agents. With heat from incandescent pla- 

 tinum the effects are extremely well marked. Thus with tour- 

 malines E and F (see First Series, 22) I obtained for the ratio of 

 the quantities of heat transmitted, with axes of crystals parallel, 

 and axes crossed, 100 : 76 ; or 24 per cent, of the heat was 

 polarized (Jan. 19. 1836). When one of these tourmalines was 

 combined with a mica plate, marked G, polarizing by transmis- 

 sion (see below art. 20), the proportion was 100 : 62, or 32 per 

 cent, polarized of heat from incandescent platinum. 



16. With dark heat incomparably greater difficulty was ex- 

 perienced. Excessively little heat could be obtained through 

 the combined mass of tourmaline and the glass to which it is ce- 

 mented, and of that little it appeared that but a minute portion 

 was polarized, or at least absorbed by the action of the former. 

 At one time I seriously doubted whether any perfectly dark heat 

 came out of tourmaline polarized in one plane only. I have rea- 

 son to believe that, in my first experiments, there was a source 

 of error, arising from the form of the plates, which was not ad- 

 verted to formerly. I have, however, satisfied myself that even 

 dark heat is capable of being acted upon by tourmalines in the 

 same manner as light. In my experiments the quantity appa- 

 rently polarized did not exceed one-seventh or one-eighth of the 

 small quantity transmitted. This was in combination with a 

 polarizing mica plate (marked I). 



§ 3. On the Laws of the Polarization of Heat by Refraction or 



Transmission. 



17. In my last paper on this subject, I stated the fact of the 

 polarization of all the kinds of heat which I tried by transmis- 

 sion through thin bundles of mica placed obliquely. I stated 

 the difficulties which I experienced, and the quantitative errors 

 to which the results were liable. I shewed at the same time that 

 these errors were of a kind calculated to mask the effects of po- 



