and Polarization of Heat. 163 
in Newrown’s rings, and which is equal to twice the distance be- 
tween the plates in that experiment. For example, with the 
thin mica film mentioned in (56), which polarized light circularly, 
the tint produced (between crossed polarizing and analyzing 
plates) corresponded (by Nrwron’s table) to an interval of about 
five-millionths of an inch between the surfaces of glass, or to a 7e- 
tardation, (o —e), of .00001 inch. The film, marked No. 2, which 
gave plum-red of the first order (65), gives a retardation of 
00002. The film No. 1 (65), gives .00004 inch. From these 
data, then, having the value of E* (68), it is clear that we may 
calculate the value of a, or the length of an undulation of heat.* 
70. In our present case we have always made 7 = 45° ; whence 
E’ =F’ sin*s (°=*); and of course 0? = F*— E*. But in an 
experiment we must not use the direct indication of the multi- 
plier, when the polarizing and analyzing planes are parallel, for 
the total quantity or F’; for a large proportion of the heat is not 
completely polarized, and in order to compare the values of E? 
and. F’, we must determine the value of each directly, that is, not 
only how much is depolarized, but how much is polarized by the 
mica plates. This I did by ascertaining alternately with the 
quantities of depolarization, the total intensity of the polarized 
part of the heat, which reached the pile. This was effected by 
rendering the polarizing and analyzing plates parallel and per- 
pendicular to one another ; whilst the principal section of the 
interposed mica remained parallel to one or other, so as to exer- 
cise no depolarizing influence. 
71. To illustrate this mode of investigation, I shall give as an 
example the very last series of experiments made on this subject, 
* Of course this is only true on the supposition that rays of heat and light are 
equally retarded. ‘This is not demonstrated, but it is probable that they are nearly 
so, since that part of the heat which accompanies the spectrum is so, and the disper- 
sion in the case of double refraction is inconsiderable. 
x Q2 
