TRANSACTIONS OF THK SECTIONS. 



553 



eclipse her sui-face was all red, but towards its close a kind of 

 twilight spread over that part towards the place of emersion, 

 which became at length so bright as to make the moment of its 

 occurrence inappreciable. As soon as a portion of illuminated 

 surface was decidedly seen, the above zone could again be traced, 

 and was seen till about one third of the disk had reappeared. 

 This zone the author supposes to be the effect of those rays 

 that pass through the atmosphere, their illumination being 

 greatest near the edge of the shadow and gradually decreasing 

 towards its axis, wliich, as he presently shows, it is improbable 

 that any of them ever reach except at a distance far beyond 

 the moon. The reddish light must proceed from some other 

 cause, which the author does not attempt to conjecture ; it is 

 commonly ascribed to the absorption of the more refrangible 

 rays in their transmission, and illustrated by reference to the 

 setting sun and the clouds of evening. But if this absorption 

 be effected by the vapours diffused in the atmosphere rather 

 than by the air itself, we are forced to conclude that no light 

 can pass through the I'egion where they occur bi'ight enough 

 to be sensible when reflected from the moon. 



The deviation of a ray of light passing through our atmo- 

 sphere will be twice the horizontal refraction of that stratum 

 of air which is at the vertex of its path. Considering this 

 stratum as an elementary zone of a refracting sphere, it will form 



an image of the sun at the distance —. — pr-n when r =s zone's 

 * sm 2 H 



radius and H the horizontal refraction. In this image all the 



rays transmitted by the zone must be found ; and the inverse 



proportion of their areas would give the ratio of direct sunshine 



to that of the refracted light were air perfectly transparent. 



This ratio can easily be transferred from the image to the lunar 



section ; and summing the effect of any number of these zones, 



we obtain the lunar illumination derived from the source. 



Let, for instance, T I and T G be refracted rays coming from 

 the lower and upper points of the sun's disc, and also R I and 

 R G. I G will be the image ; after which the rays diverge and 



