430 Sir James Dewar [Jan. 28$ 



above the horizon. This was three-quarters of an hour after noon 

 on May 9, 1920, following a night of drizzling rain with steadily 

 improving conditions during the morning ; air temperature was 

 16° C. in the shade (22° C. in the sun), and Z/E was <»-68, corre- 

 sponding to T z of — 8° C. With one of the small white clouds 

 included in the zone of exposure, the ratio increased to n-78. Low 

 daylight values were also observed on several occasions in March and 

 April, 1921, generally with short intervals of clear blue sky between 

 showery periods associated with blue sky and white broken clouds. 

 The ratios then were frequently of the order of 0'7, and sometimes 

 a little below. 



Typical determinations of sky radiation, such as those made at 

 Hump Mountain, Xorth Carolina, and also at Calama, Chile, by 

 Moore and Abbott with their electrical pyranometers, showed that 

 the radiation from the whole sky when very clear may be as low as 

 one-fourteenth that of the sun, rising with cloudy skies to as much 

 as 60 per cent, of that normally due to the sun alone. * The same 

 observers also found lower values from the upper than from the 

 lower zones of the sky, but on no occasion was there actual loss of 

 radiation to a daylight sky, however clear. 



There are of course several absorption bands in the visible trans- 

 mission spectrum of liquid oxygen "f which suggests the possibility 

 of appreciable absorption from the daylight sky ; but careful 

 measurements would be necessary to define its amount, and on the 

 other hand, the charcoal thermoscope responds strongly to the 

 radiation from sunlit clouds, even when a cooled glass screen is 

 interposed. The apparent loss of earth radiation to clear daylight 

 zenith, as indicated by the charcoal thermoscope, may therefore be 

 regarded as sufficiently definite to provoke further enquiry. 



Absorption by Atmospheric Constituents. 



The increase observed in the Z/E ratio just after sunset and 

 sunrise would be partly accounted for by the temporary increase of 

 water vapour content in the surface air by condensation in the first 

 case and evaporation in the second. This absorption by water 

 vapour at various wave lengths has been extensively studied by 

 Fowle and Aldrich. J A sensitive bolometer measured the energy in 

 different parts of the heat spectrum as distributed by a rock-salt 

 prism, after passage through a chamber over 100 metres in length. 

 By long-focus mirrors the path could be greatly extended. The 



* Annals A sir. Phys. Laby., vol. iv. p. 259, et seq., Measures of Bright- 

 ness of the Sky, etc. etc. 



t Proc. Roy. Inst., xv. p. 556. 



J Ann. Smithsonian Astrophysical Observatory, iv. 27 et seq. 



