THE MOON S FACE. 
265 
indicates a predominant approximation to vertically, and 
it therefore accords better with the phenomena than does 
the law of incidence angle derived from the theory of cosmic 
meteors. The introduction of the hypothesis of a Saturnian 
ring thus accomplishes much toward the reconciliation of the 
impact theory with the circular outline of the lunar craters. 
Whether it secures complete adjustment is not immediately 
apparent, and as the question of concordance or discordance 
is important to the impact theory, the discussion has been 
carried somewhat further. 
The inquiry has followed three lines: First, an investiga¬ 
tion of the ellipticity of lunar craters; second, an experi¬ 
mental investigation of the relation between incidence angle 
and ellipticity of impact craters; third,a more refined inves¬ 
tigation of the orbital relations affecting the incidence 
angles of moonlets. 
In the investigation of the ellipticity of the lunar craters 
I made use of a series of photographic negatives made at 
the Lick Observatory and deposited by the director of that 
observatory with the Smithsonian Institution. On these 
negatives the moon’s disk has a diameter of from 5 to 
5J inches, so that measurements of some refinement can be 
made. It was found practicable to determine the conjugate 
diameters and through them the ellipticities of 120 craters. 
In three-fourths of these the ellipticity is less than .1; in 
eleven-twelfths it is less than .2; in twenty-nine-thirtieths 
it is less than .3.* 
* Measurement was limited to the larger craters because the photo¬ 
graphs did not indicate the outlines of the smaller with sufficient precis¬ 
ion. Craters near the limb were- ignored because inequalities in the 
heights of their rims might vitiate the results, and also because the per¬ 
spective foreshortening makes it practically impossible to determine the 
directions of the longest and shortest diameters. Probably some of the 
measures made are affected by the latter difficulty and the ellipticities 
consequently underestimated, but the uncertainty thus introduced is 
thought to be less than would be added by throwing out all the craters 
obliquely seen and thus reducing the number of instances on which gen¬ 
eralization is based. The number of measurements could be largely 
extended by direct telescopic observation. 
