32 
MR. A. W. RUCKER AND DR. T. E. THORPE ON A MAGNETIC 
Magnetometer. 
Year. 
No. 60. 
No. 61. 
No. 70. 
No. 73. 
1884 
0-000817 
1885 
0-000866 
1886 
0-000828 
0-000753 
1887 
0-000809 
0-000693 
1888 
0-000800 
0-000706 
1889 
0-000840 
0-000800 
J890 
0-000786 
0-000707 
0-000934 
1891 
1891 
0-000799 
0-000811 
J 0-000753 
0-000925 
1892 
0-000814 
0-000718 
0-000907 
0-000734 
In 1891 the final reduction of the observations made with No. 60 was begun 
before the conclusion of the field work for the year. Hence the results were divided 
into two batches, corresponding to dates before and after July 31, and P was deter¬ 
mined separately for each. The same course was adopted for No. 70, but the two 
values of P were, in this case, found to be identical. 
It must be borne in mind that the number of observations on which the above 
results depend are very different in different years. Thus, in 1889, only 9 determi¬ 
nations of P were made with No. 61, and 21 with No. 60. 
The values for No. 61, in 1891 and 1892, both depend on 15, and that of No. 73, 
in 1892, on 16 observations. On the other hand, the results obtained with the 
instruments which were more frequently used, depend on between 50 and 90 
independent determinations. 
The coloured glasses used to shield the eye in the sun observations are generally 
placed in the eye-piece of the telescope. When the brightness of the sun is variable, 
owing to the rapid alternation of cloud and clear sky, it is necessary to change the 
glasses frequently and quickly. This can hardly be done without turning the 
telescope about its axis, which is inconvenient and involves error if all the readings 
are not made exactly at the intersection of the cross-wires. We have, therefore, 
had the glass screens placed in front of the object-glass, so that they can be used 
without touching the telescope. If they are wedge-shaped, this causes a slight 
displacement of the image, which can, however, be determined once for all, and 
allowed for. All the glasses used were very nearly jjlane. The corrections rarely 
exceed 20", but they were applied in all cases. 
Errors of Experiments. 
The probable errors of the observations of the Assistant Observers were worked out 
for the first year in which each of them was employed. The results were as follows :— 
