114 PHOTOGRAPHY OF THE LUNAR SURFACE. 



would, in fact, have to traverse the lunar atmosphere twice, being 

 twice deflected, but always in the same direction. To be concise, the 

 duration of an occultation would be shortened and the diameter of the 

 disk, calculated from the time of entry and exit, would be less than the 

 true diameter, and, a fortiori, less than the diameter, given by micro- 

 metric measurements or by transit observations. In the case of an 

 oblique passage of a star behind the disk the trajectory of the star 

 would be sensibly inflected. (3) On days preceding or following new 

 moon the narrow horns, which terminate the illuminated portion, 

 ought to be prolonged beyond their geometric limits — a twilight effect. 

 (4) In an eclipse of the sun that portion of the lunar disk not projected 

 on the sun's surface should appear, partly, at least, surrounded by an 

 aureole similar to that which has often been observed during the 

 transits of Venus. The light from the solar corona would be reinforced 

 in these parts, the refracted rays adding their effects to those which 

 are directly transmitted. (5) The spectral rays of atmospheric origin, 

 compared with those of solar origin, should be relatively more pro- 

 nounced in the light reflected by the moon than in that received 

 directly by the sun. (6) Finally, if one assumes that the atmosphere 

 of the moon contains a considerable quantity of water vapor, as does 

 our own, it would appear improbable that this vapor, submitted to 

 extreme temperature changes due to days and nights fifteen times 

 as long as our own, would not condense to form clouds or deposits of 

 snow readily visible. 



Most of these indications have repeatedly been observed, according 

 to the statements of astronomers worthy of belief. They can, however, 

 hardly be considered as established by sufficiently precise and con- 

 cordant testimony, depending, as they do, on such data as the prolonga- 

 tion of the horns by night and the theoretical duration of occultations. 

 To cite one more fact, the only one on which a numerical evaluation can 

 be based, and which has only been brought to light quite recently,- it is 

 to-day admitted that the semidiameter of the moon, deduced from 

 meridian observations should be diminished two to two and a half sec- 

 onds, to conform to observations of occultations and eclipses. At the 

 beginning of this century, on account of the lack of sufficiently numerous 

 and exact measurements, the two values of the diameter were regarded 

 as identical. Based on this assumption, the illustrious Bessel was of 

 the opinion that he could affirm that the atmospheric density at the sur- 

 face of the moon could not be greater than ¥ J ¥ of its value on the sur- 

 face of the earth. If the same calculation be repeated on the basis of 

 the recent discussion of accumulated observational data, it would lead 

 to the assumption that the moon does possess an atmosphere and that 

 its density is slightly greater than that assigned by Bessel as a maxi- 

 mum limit. 



It would be wrong, however, to accept this conclusion as absolute. 

 We do not know the diameter of the moon, as determined by direct 



