PHOTOGRAPHY OF THE LUNAR SURFACE. 113 



ency or a colony of the earth. To be more precise, and considering the 

 question from the standpoint of Laplace, the satellites ought to be con- 

 sidered as fragments of the planet they accompany, these fragments 

 having been derived from its equatorial and superficial layers. The 

 materials which predominate in the original body ought to be repre- 

 sented in the satellite, and especially the fluids and other light sub- 

 stances which gravity tends ceaselessly to conduct to the surface. This 

 supposition is confirmed by the low mean density of the moon, hardly 

 greater than one-half of that of the earth. It might generally be sup- 

 posed that the division of the original atmosphere would have taken 

 place in proportion to the masses of the two bodies, and one ought not 

 to expect to find the air on the surface of our satellite forming a layer 

 as dense as that on the terrestrial globe. The ratio of the surface to 

 the volume of the moon is four times as great as that of the earth, and 

 gravity, reduced to a sixth part of its value, would more or less effi- 

 ciently counterbalance the expansive force of gases and vapors. The 

 lunar atmosphere would therefore be expected to extend to a much 

 greater height than ours, producing a corresponding decrease in its 

 density. For these two reasons it may be predicted that the air at the 

 surface of the moon can not possess a refracting power as great as one- 

 fiftieth of that on the earth's surface. Observation, however, shows 

 that it must be very much less than even this. 



VIII. — INDICATIONS THUS FAR OBTAINED BY DIRECT OBSERVATIONS. 



If our satellite possessed a layer of air of appreciable density its 

 presence would surely be revealed in several different ways : (1) The 

 -diameter of the moon would be increased, its apparent contour being- 

 no longer determined by straight tangents drawn from the observer to 

 the lunar globe, but by rays which are tangent to its surface, after hav- 

 ing been curved in its atmosphere. Assuming that refraction is still 

 sensible at a height of five kilometers above the lunar surface, this would 

 result in an increase of about five seconds in its apparent diameter, 

 and in a band of about two seconds width, there would be a superposi- 

 tion of images, which would detract very much from the sharpness of 

 the moon's outline. Bright stars would, before disappearing, seeinfor a 

 few seconds to travel on the luuar disk. (2) The apparent diameter, 

 calculated from the duration of the occultation of stars, would, on the 

 contrary, be smaller than the apparent diameter defined by geometric 

 tangents. A phenomenon similar to that observed at the rising and 

 setting of the stars would result. Refraction transforms the luminous 

 rays into curvilinear trajectories, with their concavity always turned 

 toward the center of the earth. Therefore, for example, the time of 

 sunset is retarded and that of sunrise is hastened by the same amount. 

 In the case of an occultation the effect would be twice as great seen from 

 the earth as it would be for an observer on the moon at the point of 

 tangency of the luminous ray emitted by the star. The ray, to reach us, 

 S M 98 -8 



