402 Prof. S. P. Langley on Invisible Heat- spectra 



cases respectively 185° C, 88° C., 46° C, and -18° C. In 

 the last instance, the cube was colder than the bolometer 

 strips, and the deflections obtained were negative ; though 

 small, they were distinctively measurable, the greatest being 

 —12 divisions of the galvanometer-scale. The position of 

 the maximum in the ice-curve is indicated, but the curve 

 itself is, on this scale of ordinates, sensibly coincident with 

 the straight line. (See PL IV. fig. 1.) We have in fig. 3 the 

 curves representing the radiation from these sources, in which 

 the abscissae are proportional to the indices of refraction in 

 the rock-salt prism, but the ordinates only approximately so 

 to the deflections of the galvanometer due to the heat at the 

 corresponding points ; since we are not here principally 

 concerned with the relation of the amounts of heat emitted to 

 the temperatures of the sources of emission, but chiefly with 

 the secondary effect of the progressive movement of the 

 maximum which is clearly shown. 



Date of observa- 

 tion. 



Approximate 



temperature 



of source. 



Approximate 



temperature 



of excess.. 



Deviation of 



maximum ordinate 



in heat curve. 



1885. 

 October 7 



815° C. 



525 



330 



330 



300 



178 



179 



119 



100 



99 



40 

 -02 



803° C. 

 505 



318 

 310 



275 

 185 

 152 

 126 



88 



79 



46 

 -18 



39° 08' 

 39 03 

 39 01 

 39 00 

 38 42 

 38 35 

 38 35 

 38 25 

 38 22 

 38 27 

 38 00 

 37 40* 



.. 3 



7 



3 



September 26 ... 

 March 20 



August 19 



March 20 



October 7 



3 



March 21 



„ 24 





It is to be observed of each of the curves in fig. 3 that 

 though nearly all the area is seen, yet, owing to the exten- 

 sion of the heat-curve toward the right, the length shown 

 is limited here by the size of the plate, whereas the extremity 

 measured in each curve (except of course the solar one) does, 

 in fact, correspond to an index of less than 1*45. We give 

 above a table showing the dates of observation, the approxi- 

 mate temperatures of excess, and the approximate deviations 

 of the ordinate corresponding to the point of maximum heat 

 in the (rock-salt) prismatic spectrum. We should observe 

 that the higher temperatures are here only determined with 

 an approximation sufficient to make it certain that there is a 



* The position of the maximum in this last case depends upon a 

 single observation of some delicacy and is liable to subsequent correction. 



