392 
ME. WAEEEN DE LA ELT] ON THE 
Column 10 gives the ditFerences betvreen the mean distances in column 8, and those 
in column 9, or De La Kue — Farley. The mean of the differences will be seen to he 
+4"T; that is, the distances of the centres of the sun and moon come out greater than 
the computed distances by 4"T. This tends to show that the semidiameters of the sun 
and moon jointly, are less in reality by 4" than their tabular values. It is not intended 
to urge this as an absolute proof, but merely as supporting that view, which is further 
corroborated by the fact that the first contact occurred later, and the last contact sooner, 
than the predicted times. The distances of the sun and moon’s centres, obtained by 
calculation from the angular opening of the cusps, will be presently employed to furnish 
data respecting the commencement and end of the eclipse. See. ; and it will be seen that 
the times thus obtained differ from those derived from the peripheral distances, and that 
they approach more nearly to the predicted times. The optical distortion of the sun’s 
image would occur in the direction of a radius, and would not affect the numbers derived 
from measurements of the angular opening of the cusps, provided the picture were con- 
centric with the optical axis of the instrument ; while it would affect the numbers based 
on the measures of the distances of the sun and moon’s peripheries, so that the quantity 
4" is probably, from that cause, in excess of the true correction. The measurements of the 
angular opemngs of the cusps, and the measurements of the distances of the peripheries, 
both present peculiar ditficulties. The difficulty of determining the precise termination 
of the cusp, especially when blunted by a lunar mountain, leads one to make the 
angular opening greater than it ought to be, and, consequently, the cosines and the 
distance of the centres less than they really are. On the other hand, the optical distortion 
of the image, combined with the irregularities of the peripheries of the sun and moon, 
tends to make the measurements of the distances of the peripheries, and consequently 
the distance of the centres, greater than they are in reality. These liabilities to error 
have, therefore, in the two cases, an opposite effect on the final results ; hence a mean 
of the numbers obtained by the two methods will probably approach very nearly to 
the correction to be applied to the semidiameters of the sun and moon taken conjointly. 
