Third Series. — Befrangihility of Heat. 1 89 



The preceding results give values of ju, all too high ; that for 

 light is known by other methods to be only 1*53 or rs*. 

 This arises from the increasing intensity of the partially re- 

 flected light with the obliquity of the incident ray which makes 

 the apparent transition from partial to total reflection too ra- 

 pid, and consequently gives the index too great. The true 

 indices may be approximately found by diminishing these 

 numbers by '07 ; but the relative results are the most im- 

 portant. 



The results which we have obtained apply, it must be re-^ 

 collected, only to the 'predominant kind of heat in any source, 

 and that we have as yet got no information respecting the 

 composition of a ray and the amount of dispersion. 



It is very easy to see that were the mathematical conditions 

 of the experiment (p. 183) fulfilled, we should be led to an 

 exact analysis of heat, more perfect far than we have any 

 prospect of obtaining in the case of light, considering the dif- 

 ficulty of applying the photometer to coloured light. Were 

 the curve in Plate IV. fig. 2, solely representative of the 

 progress of reflection due to the heterogeneity of the rays, the 

 increment of intensity between any diagonal E and another 

 C, or DjT would denote the proportion of the entire heat in- 

 cident, which lies between the limits of refrangibility assigned 

 by the diagonal. Thus an entire ray would be decomposed 

 into parcels of known proportions, between given intervals of 

 refrangibility. The case is considerably different. Though 

 the points of contrary flexure agree remarkably well, as we 

 have seen, the curves are in some cases much more flattened 

 than in others, where the source of heat is the same ; owing 

 probably to the greater parallelism of the rays at one time 

 than at another, depending on the distance of the source of 

 heat from the lens. 



We can, therefore, in this way form but an imperfect idea 

 of the comparative homogeneity of the different kinds of heat. 

 Such comparisons can only be made advantageously by com- 

 paring the results obtained in immediate succession from one 

 and the same source with interposed screens of different qua- 

 lities, as in the comparison which we instituted between heat 

 direct from Locatelli's lamp, and that transmitted by glass, 

 (p. 186). 



The facts respecting refrangibility, which may now be con- 

 sidered as ascertained, serve to render our ideas much more 

 precise in several respects. For instance, (1.) the range 

 of mean refractive indices for heat is small, all the modifica- 

 tions which we have considered lying within a range of "O^, 

 or between \'5\ and 1*55 nearly, which is little more than 



