MEMOIR XXVIIL] DISTRIBUTION OF HEAT IN THE SPECTRUM. 49 1 



TABLE IV. Distribution of Heat by Quartz. 



Series I. Series II. 



(1) Heat of the whole visible spectrum 100 100 



(2) " more refrangible region 49 53 



(5) " less " " , 51 47 



Series I. represents twenty -seven experiments, Series 

 II. twelve. In the former two quartz prisms were used 

 to increase the dispersion ; in the latter only one was 

 employed. 



Perhaps it may not be unnecessary for me to say that 

 I have repeated these experiments many hundred times 

 during a period of several months, including the winter 

 and the summer, varying the conditions as to the hour of 

 the day, arrangement of the apparatus, etc., as much as I 

 could, and present the foregoing tables as fair examples 

 of the results. Apprehending that the heliostat mirror, 

 which was of speculum metal, might exert some disturb- 

 ing influence on account of its faint reddish tinge, I re- 

 placed it with one of glass silvered on the front face, but 

 could not detect any substantial diiference in the results. 



The important fact clearly brought into view by these 

 experiments is, that if the visible spectrum be divided 

 into two equal portions, the ray having a wave-length of 

 5768 being considered as the optical centre of such a 

 spectrum, these portions will present heating powers so 

 nearly equal that we may impute the differences to errors 

 of experimentation. Assuming this as true, it necessarily 

 follows that in the spectrum any two series of undula- 

 tions will have the same heating power, no matter what 

 their wave-lengths may be. 



But this conclusion leads unavoidably to a most im- 

 portant modification of the views now universally held 

 as regards the constitution of the spectrum. When a 

 ray falls on an extinguishing surface, heat is produced ; 

 but that heat did not pre-exist in the ray. It arose from 

 the stoppage of ether waves, and is a pure instance of 



Co 



