THE ILLUMrNATION OP THE ECLIPSKD MOON. 365 



ABC would be, as regai-ds solar light, absolutely dark. 

 The point P would be the furthest point at which rays 

 refracted through the lowest stratum of air would illuminate 

 an object in the shadow. 



And here I may remark that the rays which will reacli 

 any given point in the axis of the shadow after refraction 

 througli any given stratum of atmosphere, will be rays from 

 an infinitely narrow ring of the sun's disc and from no other 

 part. Confining our attention at present to the rays re- 

 fracted through the lowest stratum, we note that the rays 

 reaching the point B will be those only wliich have come 

 from the extreme margin or limb of the sun. Points be- 

 tween B and D will receive rays from successively smaller 

 and smaller circles of the sun's disc, until at D these circles 

 will have dwindled down to the centre itself, and rays of the 

 directly refracted class will have come to an end. But at 

 the same point D, where the directly refracted rays cease, 

 the cross-refracted rays begin to appear, and points between 

 D and F will receive rays from successively wider and wider 

 circles of the sun's disc, until at F we again meet with rays 

 derived from the extreme margin of the sun. To put the 

 case shortly : tlie earliest rays to converge will be rays from 

 the limh of the sun refracted through corresponding parts of 

 the eartli's atmospliero ; the latest to converge will be rays 

 from the limb of the sun refracted through opposite parts of 

 the earth's atmosphere ; while rays from the contre of the 

 sun will converge at an intermediate point. 



Wo have hitherto confined our attention exclusively to 

 the solar rays refracted thi-ough the lowest and densest 

 stratum of tlio earth's atmosphere. But what of the rays 

 which pass throiTgh the upper strata ? Precisely the same 

 series of events will happen to these, only each step will bo 

 lower down in the length of the shadow (or further from the 



