NO. 8 



WATER-VAPOR TRANSPARENCY — FOVVLE 



67 



and reflections at the rock-salt prism and plate by table i. Intensity taken 

 as 100,000 at 1.8 [x. 



Line 8 corrected intensities in the Nernst lamp spectrum approximating the 

 conditions of line 7 but differing at the greater wave-lengths probably because 

 of the decreasing absorption of energy by the lamp-blackened surface of the 

 bolometer strip. (See discussion relative to comparison of black-body spec- 

 trum in section so headed in main body of paper). 



It was found that the total energy scattered from any region is 

 only about 3 per cent of that belonging to that region. The angle 

 of incidence of the radiation on the image forming mirror was 

 slightly less than 3°. The intensity of scattered energy falls off very 

 rapidly to each side from the central image (see line 5). If the 

 intensity is 100,000 in the central image, at an angle 10' it amounts 

 to only I of I per cent, at 20', i/io of i per cent, then falling ofif 

 more slowly, it amounts to i/ioo of i per cent at 100'. The total 

 field energy (line 6) at 10' amounts to 5 per cent of that of the lamp 

 spectrum upon which it is superposed, at 60' it amounts to nearly 

 50 per cent and at 100' to over 500 per cent. As already stated, in a 

 solar spectrum formed with the same apparatus, the stray light at 

 10 /x was more than 100 times as intense as that which belonged there. 



The above data were obtained with silver-on-glass mirrors, the 

 surfaces of which were in excellent condition, freshly polished to 

 a hard, compact surface. The following table shows a comparison 

 with results obtained several months later after the mirrors were 

 so badly tarnished as to be unfit for work in the visible spectrum. 

 It will be noted that the scattering had increased relatively more for 

 the greater deviations from the central image. 



Table 18— Increase of Scattering with Tarnishing 



Deviation (') . . . 

 Wave-length (/u) 

 Increase ratio. . . 



100 

 17-5 

 1.70 



It would be of interest to know how far this work is applicable to 

 determine stray light in other cases. The above results were a side 

 issue of the more extensive investigation described in this report, and 

 no more time could be spared to investigate this matter. In the work 

 on the solar spectrum, involving resilvering of all the mirrors, a 

 different distribution of energy in the spectrum, in which the upper 

 infra-red much more greatly predominated over the longer wave 

 portions than with the Nernst glower, produced curves of stray light 

 which were the same within the limits of observational errors. Part 



