NO. 8 WATER- VAPOR TRANSPARENCY— FOWLE 3 1 



since they constitute nearly all of the energy of a Nernst glower, and 

 still more its energy after water-vapor absorption, must furnish most 

 of the stray light, still it cannot be assumed that the stray light of 

 longer wave-lengths than 4 /a is negligible. Accordingly to the 

 ordinate i.iSy, there must be made another addition, which we may 

 call 2y, representing the stray light of all wave-lengths non-trans- 

 missible by quartz. By a special research the quantity 2y has been 

 determined for each place in the spectrum, as will be explained. 

 The corrected ordinate now becomes y^— i.iSyo — 2y. Finally each 

 ray is, on the other hand, depleted by sending stray light to all parts 

 of the spectrum. But as will be shown the depletion is so nearly the 

 same proportion of the intensity, namely, about 3 per cent, for all 

 the wave-lengths here considered that to correct for it is merely 

 to multiply all ordinates by the same multiplier, which may be 

 omitted. The increase in scattered energy, 2y, over 1. 18 times that 

 observed by the insertion of the quartz plate, ranges from 2 to ii 

 per cent according to the angular deviation. In addition its amount 

 varies over a range of about 2 per cent for the changes in the amount 

 of absorbing vapor occurring in this research. 



All the observations taken during 1916 and 1917 were corrected for 

 stray light by the process detailed in Appendix II. The entire 

 absence of appreciable strAy light in the part of the research made 

 with the 60° prism unfortunately led to a false feeling of security 

 in the first work with the 15° prism, and in much of the earlier work 

 with it no field-light experiments were made. Accordingly the 

 following indirect way was resorted to for its elimination : The 

 1916-17 work showed that with the carbon dioxide present in the 

 long tube no energy at wave-lengths longer than 1$ fi should reach 

 the bolometer. Accordingly all of the considerable deflection ob- 

 served there could be considered as stray light. The stray light foi 

 shorter wave-lengths was assumed to bear the same ratio to this 

 that the ordinates at the corresponding deviations of the stray light 

 curve of figure 18 do to each other. This involves building up an 

 ever increasing correction from a somewhat small observed value. 

 However, fortunately the larger in absolute amount it becomes, the 

 less its importance, for it becomes a smaller fraction of the total 

 observed deflection. The validity of this procedure is perhaps 

 shown by the depths thus obtained in the great band at 7 /i, where the 

 somewhat large deflection observed should be nearly all stray light. 



Before considering further troubles we will pass to a consideration 

 of the results as shown in figures 9 and 10. The form of the observa- 

 tions was already made familiar through figure 8. 

 3 



