NO. 8 WATER-VAPOR TRANSPARENCY FOWLE 59 



LIMITATIONS 



It is evidently impossible by any such process as that just developed 

 and described to obtain, in general, from an impure spectrum, a pure 

 spectrum, except in the case of a continuous energy curve such, for 

 instance, as would be expected from an incandescent black body. A 

 line cannot be made to appear which has been completely obHterated 

 by the impurity of the spectrum although one which has been ren- 

 dered shallow may be deepened, and a maximum which has been 

 lowered may be made to approximate its proper height. 



APPENDIX II 



STRAY LIGHT AND ITS DETERMINATION 



The most intense region in the rock-salt prismatic spectrum of 

 the Nernst glower lies at wave-lengths less than 4 fi. From this 

 region, which is transmissible by quartz, most of the field light comes. 

 At 4 |U, the intensity of the spectrum had already decreased nearly to 

 i/io. Interest in the present research centers in wave-lengths greater 

 than 4 fx and not transmissible by quartz. Energy proper to any place 

 in the spectrum of wave-length greater than 4 /* has been dis- 

 tinguished from that scattered into it from the intense short-wave- 

 length spectrum region by observing what portion of the deflection 

 remains when a plate of quartz is inserted between the source of 

 energy and the spectroscope slit. Since the quartz is opaque to 

 radiation of wave-lengths greater than 4 fx, the deflection observed 

 through quartz must be all false and due to energy of wave-length 

 less than 4 fi. Without the quartz, the deflection represents this false 

 energy, plus the true longer-wave radiation, plus certain other cor- 

 rections to be presently considered.' 



Turning again to figure 8, the significance of the central curve 

 a'h'c'd'e' as a means of determining this field light will be considered. 

 This curve, which is essentially an energy curve of the source ob- 

 served through a plane parallel quartz plate 1/2 cm. thick, consists of 

 two parts of quite different significance, namely, region a and region 

 b'c'd'e'. 



Region a'. — Curve a' indicates approximately the intensity of 

 energy in the lamp spectrum for wave-lengths between 0.6 and 4.0 fi 

 for which quartz is nearly transparent. The dift'erence between the 

 areas a and a' (after making certain correction for the absorption in 

 the quartz of the longer waves between 3 and 4 fx) measures the 

 amount of energy of this region reflected from the quartz surfaces. 



