158 



PROCEEDINGS OP THE AMERICAN ACADEMY 



The experiment is one containing so many errors that it could not 

 be expected that the experimental curves would be very accurate ; 

 but the large number of experiments made, on the other hand, give 

 the mean curve probably with considerable accuracy. 



An inspection of Plate II. shows us at once the important fact, 

 that, though the temperature varies very considerably, the geometrical 

 form remains very nearly, if not exactly, the same. 



Of course it is quite possible that further and more delicate experi- 

 ments may show slight variations in the form of this curve. 



In fact, it is well known from the experiments of a large number of 

 experimenters, and, indeed, from the change of color of the light as 

 the temperature of a source of radiation is raised, that the geometrical 

 form of the curve cannot be entirely independent of the temperature. 



But such phenomena might easily be caused by variations too 

 delicate to be detected by the thermopile. 



TABLE I. 



B line 11.15 



D line 12.45 



E line 14.35 



