162 



PEOCEEDINGS OF THE AMERICAN ACADEMY 



In comparing the spectra of the different bodies with each other, we 

 wish to study the distribution of heat, and not the relative- total 

 emission. Accordingly, the curves of Plate IV. are all reduced to 

 approximately the same size. An examination of this plate indicates 

 a relation which further study may show to be of considerable impor- 

 tance, though the experiments thus far give hardly more than a hint 

 of what the relation ia. 



If we regard each curve as an irregular triangle, and, starting from 

 each of the augles, draw lines of symmetry toward the opposite sides, 

 the point of intersection of these lines, or, if they do not intersect in a 

 single point, the centre of the small triangle of intersection, may be 

 regarded as a centre of area. 



The wave-lengths of these several centres of area are found to be 

 related to the molecular weights of the radiating substances. 



They are not in exactly the same proportion, but, at least, they are 

 in the same order, so that the greater molecular weights have the 

 greater wave-length, and large or small intervals in one quantity are 

 represented by large or small intervals in the other. 



These results appear in Table V., in which the first column gives 

 the substances, the second the molecular weights, and the third the 

 wave-lengths of the centres of area. In the fourth column are given 

 the abscissas of the point in the arbitrary scale adopted in the plates. 



TABLE V. 



It would be useless to attempt the formation of an empirical equa- 

 tion connecting these quantities, as the experiments so far made are 

 not sufficiently accurate to warrant it. In fact, in single observations 

 the centres of area near together are sometimes reversed in order. 

 Still, the mean curves point out a relation, which could hardly be 

 accidental for all of the six substances. 



The curves are all of so nearly the same form that the distances 

 between even the extreme geometrical centres is not very great, and 



