NO. 8 



WATER-VAPOR TRANSPARENCY — FOWLE 



53 



in the region from 9 to 15 /a and yet the additional amounts of vapor 

 produce no further absorption there? 2d, In treating of the earth's 

 radiation outward, complete absorption for atmospheric amounts of 

 vapor has been assumed for wave-lengths greater than 20 fi. Is this 

 certain? The first point could be settled only by observing the 

 radiation through a path containing no water vapor. Dr. Coblentz 

 (Proc. Nat. Acad. Sc. 3, p. 504, 1917) has just published observations 

 ■on a black body radiating at 800° C. Comparing the radiation after 

 passing through a tube containing 0.00045 cm. ppt. H,0 with that 

 when the tube was exhausted he found the absorption due to this 

 amount of vapor to be about 0.9 per cent. If all the absorption had 

 been produced in the region between 9 and 15 /x,, the mean absorption 

 in this region must have been 12 per cent. 12 per cent, then, is the 

 maximum absorption permissible here with this amount of vapor. 

 Assuming the absorption in the known water-vapor bands is propor- 

 tional to the amount of vapor, within the range from no vapor to the 

 smallest amount used in the present research, the following table is 

 probably representative of the absorptions from wave-length to 

 wave-lens:th in Dr. Coblentz' case. 



Total energy same scale, 6825. 



The absorption computed from the above data would be 1.3 per 

 cent, which is probably within experimental error the same as that 

 found by Dr. Coblentz. This confirms the conclusion of great trans- 

 parency in the re'gion between 9 and 15 ju,. 



As to the second point, recent nocturnal experiments by Mr. L. B. 

 Aldrich on Mount Wilson with the pyranometer, employing salt and 

 sylvine screens tend to confirm the assumption that the atmosphere 

 transmits no appreciable quantity of radiation from the earth of 

 wave-lengths greater than 20 /x. 



APPENDIX I 



CORRECTION TO OBSERVED ENERGY CURVE FOR WIDTH OF SLIT AND 



BOLOMETER 



The true monochromatic intensity in a spectrum would occur only 

 with the use of an infinitely narrow slit and then the intensity would 

 be infinitely small. As both the slit and the observing device must 



