826 
values of A,r in the second computation are considerably larger 
than the corresponding values of 4,» in the first computation. Also 
the values of A,/ are larger than those of 4,4. We are thus led 
to the same conclusion as before, viz: the reality of the non-periodie 
part of the perturbation is not assured. and the only thing that can 
be asserted with certainty is that the non-periodic part cannot have 
any considerable irregularities and that no other periods are possible 
than the Saros of 18.03 years. 
The following tables contain the principal quantities occurring in 
the computations. Table V gives for each eclipse the values of 7, 
1. 1,’ and those of dn and dn’ computed by the formula (10). The 
first column of the table contains the time ¢ counted in synodic 
months from the beginning of the Saros. The time ¢= 223 of any 
Saros is, of course, identical to the time ¢=O of the next Saros. 
The arrangement of the eclipses in groups of six is very clearly 
shown. The several groups begin at 
aoe |) 41, 88, 129 and 176 
and end at 
bai Poh tale ij boj 165 and 212. 
Table VI contains the purely periodie part of the perturbation 2, 
and 2,’ according to the two computations. The similarity between 
the different Saros-periods is very striking. In the mean motion this 
similarity is even more apparent than in the longitude. The mean 
motion is not contained in the table. but can easily be derived from 
the longitudes, as it is the difference of two successive values of 2, 
(or 2. We see from this table that in the first computation the 
amplitude of the periodic part is fairly constant for the first eight 
periods and begins to increase after the eighth Saros. The difference 
between the -extreme values of 4, oscillates between 700 and 830 
in the periods I to VIII, and then gradually increases up to about 
1200 for the Saros XII. In the second computation the difference 
between the extreme values of 4’; is more constant and varies between 
about 950 and 1100. 
The remarkable agreement between the results of the two compu- 
tations justifies the expectation that the general character of the per- 
turbations in longitude produced by an absorption of gravitation will 
be sensibly the same for any assumed distribution of density within 
the body of the earth, which is at all within the limits of probability. 
The conclusions arrived at in Part I are thus not restricted to the 
particular hypothesis which was there introduced, but have a much 
wider bearing. 
