THIRD SERIES.— DEPOLARIZATION OF HEAT. 191 



to discriminate the different lengths of waves of these kinds of heat, as I had for- 

 merly supposed, and shews that the variation of x must be very small, or else 

 (what is improbable) that it is constantly proportional to the variation of the re- 

 tardation — e. 



43. A^Uthe three figures give as nearly as possible a value of 1.4 for —^ at a 



thickness of depolarizing mica, equal to .020 inch, or .07 for a thickness of .001 

 inch. Let us compare this with the case of light. The sum of the retardations 

 for the various mica plates, as given in art. 33, amounts to .000199 inch ; the 

 sum of the thicknesses in the next article is .0361 inch, consequently the mean 

 value of the retardation or o — e is .0000055 for a thickness of mica of one-thou- 

 sandth of an inch. But the length of x for extreme red is .0000266, for extreme 

 violet, .0000167 inch. Hence for a plate of mica .001 inch thick the values of 



o — e 



are 



55 



For extreme Red light, . . . -^gg- = -'^o? 

 For extreme Violet light, . . . -^ = .329 



° 167 



For Heat, . . • .... = .07 



44. If we assume the retardation, or o — e, to be the same for all lengths of 

 waves, and for heat as for light, we immediately deduce the value of x, or the 



length of a wave of heat. For since for a plate .001 inch thick, — ^ = .07, as above, 



o — e = .0000055, we have 



.07 



about three times as long as a wave of red light, and four and a half times that 

 of violet. But it is always to be remembered, that this proceeds on the supposi- 

 tion of the retardation being invariable. 



45. I have taken the trouble to calculate and project in a similar manner 

 my original observations on Depolarization given in the First Series of these re- 

 searches, art. 74, in order that, though probably less accurate, they might form 

 a check upon the results just given. The plates then employed, and marked 

 No. 1 and No. 2, (which are not to be confounded with those so designated in 

 this paper) had thicknesses (deduced from the retardations) of .0072 and .0036 

 inch. I have the gratification to find that the computed results agree almost pre- 

 cisely with those just obtained, although from the accidental thicknesses of the 

 two plates employed the observations with these alone do not enable us to select 



the appropriate value of , there being at least two values which still remain 



ambiguous, but when taken in conjunction with the observations of art. 41, the 

 ambiguity is at once removed, and the numerical value of x comes out almost ex- 



