320 ANNUAL OF SCIENTIFIC DISCOVERY. 



moon's cdo^e. Since, then, there is an increase both of the actinic 

 power and of tiie heat, it would seem that these eflfects are real, 

 since the methods of ol)servin<2: thom arc so totally ditVerent that 

 no error in one could be introduced into the other. The only ex- 

 phmation of the phenomenon that seems possible is to assume the 

 presence of a lunar atmosphere. The corona would then be 

 caused l)v refraction, light ri'achini2: the observer from parts of the 

 sun already eclipsed. Althou<;h, for various reasons, this hypoth- 

 esis is unsatisfactory, yet it is strengthened by other observa- 

 tions. The protuberances have often seemed to indent the moon's 

 edge, — an appearance usually ascribed to irradiation. Several of 

 the photographs, however, show this same et^ect ; and in some of 

 them the exposure was so short, and the edges of the protuber- 

 ances are so well defined, that it cannot be caused by the intensity 

 of their light, but must have its oris^in outside of the eye of the 

 oljserver. It is noticeable on all sides of the moon, sometimes in 

 half-a-dozen protuberances in a single photograph. An atmos- 

 phere of rapidly increasing density might produce this effect by 

 reflection, and of course would not influence the corona, if it was 

 caused by refraction. On this supposition, reliance could not be 

 placed on measurements of the moon's diameter by occultations, 

 or by contacts during eclipses, and would account for the uncer- 

 tainty of this constant. 



Tlie principal reason for supposing the corona a portion of the 

 sun is, that, during totalit}', it does not appear to move with the 

 moon, but remains concentric with the sun, or, more properly, is 

 brightest where the sun's edge is nearest. Many of the photo- 

 graphs show this very well, the difference on the two opposite 

 sides of the moon being verv marked. Now, this eff'ect would 

 be explained equally well by supposing the corona caused by 

 refraction. For the centres of the sun and moon never difter 

 during totality by more than half a digit, while the breadth of the 

 corona is sometimes several times as much ; so that meridy cover- 

 ing a small portion of it wouhinot produce a greater diminution 

 of light than would be caused by a slight change in the direction 

 of the sun's rays shining through a lunar atmosphere. On the 

 other hand, it is difficult to conceive of an atmosphere dense 

 enough to produce these eff*ccts, and yet so trans})arent that the 

 edges ot" the full moon are perfectly distinct, and that the light of 

 the sun, during an eclipse, should be increased rather than ilimin- 

 ished. Again, we should expect that such variations would be 

 produced Ijy changes of temperature that they could scarcely fail 

 to be detected. 



We, then, conclude that the polariscope gives only negative 

 results, and cannot be regarded as proving that the light is re- 

 flected. The evidence of the spectroscope needs confirmation, 

 since the dark lines may have been invisible, owing to the feeble 

 light of the corona. But if the observations with it are correct, 

 the self-luminous character of the corona is established. The 

 thermometric and actinic experiments point towards a lunar atmos- 

 phere as the cause of the corona. 



In the above, I have endeavored to give the evidence in favor 



