NO. 8 



WATER- VAPOR TRANSPARENCY FOWLE 



35 



with different slit widths occurred, it would serve to assure the 

 absence of errors from diffraction in the tube observations. 



Such a cold day was finally obtained in February, 191 7, with only 

 0.028 cm. ppt. HoO in the optical path. An excellent set of observa- 

 tions was obtained and the mean result of the day, so far as it 

 diverged from the curve of the spectroscope alone, is plotted with 

 simple crosses under the extreme right curve (1916-17 observations). 

 It shows : First, that an increase in water vapor from 0.003 to 



Fig. 10. — Energy curves showing the absorptions produced by water vapor 

 in the 15° rock-salt spectrum of the Nernst lamp. Ordinates are intensities 

 of energy. Abscissae are proportional to prismatic deviations from 1.8 m. 

 Unless otherwise stated, dates are as indicated in Figure 9. 



0.03 cm. (lo-fold) produces no appreciable change in transparency 

 of the region from 9 to 13.4 jx. Second, that beyond from 13.4 jx the 

 additional carbon dioxide contained in the tube becomes very effec- 

 tive, and produces practically complete absorption at about 14.6 p.. 

 The cross-hatched portion shows the proportion of energy cut off by 

 absorption. This may be assumed to be the added effect due to the 

 carbon dioxide in the tube over that present in the spectroscope. 



It has already been noted in the section on " Diffraction Effects " 

 that the coincidence (up to the region of the carbon dioxide band) of 



