THIRD SERIES.— DEPOLARIZATION OF HEAT. 189 



2 + «, or 3 — a, &c. In the case of the two examples given above, we have for 

 the Argand 



|i=^:!L=.629; /E^=±.793 

 F2 8,38 V F- 



And ^^=^= .29 or .71 or 1.29 or 1.71, &c. 



For the dark heat, ^= |g- = -915; J^ = ± -957 



And ^=1= .41 or .59 or 1.41 or 1.59, &c. 



X 



The true value must be such, that, when a number of plates are employed, °~* 

 must mcrease uniformly with the thickness of the plates. 



39. Clearly to mark this, and at the same time to combine the results by gra- 

 phical interpolation, I projected the numbers obtained as above in the way shewn 

 in Plate XL Figs. 1, 2, and 3. On a horizontal line spaces representing the thick- 

 ness of the plates (art. 34) were set off as abscissae, and a few of the ambiguous 



values of ^— as ordinates, which are marked by dots. It was then easy to se- 

 lect those points thus set off, through which a straight line could most nearly be 

 drawn, representing the linear relation between the thickness of the plate and 



the quantity ~^, (both vanishing when the thickness = 0), and inspection of the 



figures will shew that no doubt can attach to the choice of the ambiguous num- 

 bers, and also that the straight hne represents in general remarkably closely the 

 course of those points. 



40. There is one exception to this statement, and it is an important one. It 

 wUl be observed that in all the three figm-es the interpolating line, instead of pass- 

 mg through any of the dots set oft' for the mica plate No. 3, bisects exactly twcf 

 dots, which are nearest to one another in the case of dark heat, — wider apart 

 with incandescent platinum, and Avidest of all in the case of the Argand-lamp. 

 The explanation is complete and satisfactory. The interpolating line in all these 



cases gives a value of -^ = \, which gives a value of ^ = 1 ; in other words, in- 

 fers a total polarization of the heat in the horizontal plane (or in the case of light 

 total darkness, when the polaiizing and analyzing plates are parallel) which we 

 know can only occur when the heat is absolutely homogeneous. The want of ma- 

 thematical coincidence in this case infers the admitted physical condition of want 

 of homogeneity in the incident rays. Hence, we infer that dark heat is most ho- 

 mogeneous ; next, that from incandescent platinum ; and, least of all, that from 

 the Argand. 



