80 Proceedings of Royal Society of Edinburgh. [sess. 
wrong, but the error of this quantity is certainly small, and cannot 
sensibly affect the observed shifts. 
After the observations had thus been properly reduced, I pro- 
ceeded in the following way. My object being first to obtain the 
pure effect of the annual shift, I corrected for the diurnal and lunar 
displacement, using the above formulae, and then arranged the 
observations into groups of 25 to 30 single measurements accord- 
ing to the longitudes of the sun. This first collection of data is 
shown in Table I. 
Table I. 
© 
Distance 
in t.m. 
Weight. 
© 
Distance 
in t.m. 
Weight. 
1901 
1903 
0 
150 
0-3904 
1 
179 
0-3801 
2 
3 
165 
•3913 
1 
204 
•3813 
§ 
219 
•3901 
l 
3 
222 
•3807 
2 
3 
1902 
1901^ 
o 
341 
•3764 
1 
338 
•3609 
1 
14 
•3693 
1 
17 
•3579 
1 
93 
•3794 
1 
68 
•3655 
1 
161 
•3859 
1 
100 
•3729 
1 
208 
•3853 
1 
176 
•3809 
1 
189 
•3809 
1 
1903 
o 
1905 
348 
•3621 
1 
3 
o 
32 
•3629 
2 
3 
71 
•3649 
1 
64 
•3631 
1 
82 
•3655 
1 
75 
•3644 
1 
93 
•3671 
1 
147 
•3738 
o 
3 
115 
•3725 
f 
201 
•3747 
o 
3 
Unfortunately, the low position of the sun, combined with the 
very unfavourable atmospheric conditions, rendered observations 
between November and February quite impossible. But in spite 
of these unavoidable gaps the effect of the annual displacement 
is very clearly shown in the following graph (fig. 2), which 
demonstrates the sinoidal character of the annual curves quite 
plainly. In accordance with the theoretical formulae, we find a 
