SURVEY OF THE BRITISH ISLES FOR THE EPOCH JANUARY I, 1891. 
33 
[ 
1 Observer. 
Declination. 
H (metric). 
Dip. 
Year. 
No. of 
observations. 
Probable 
error. 
No. of 
observations. 
Probable 
error. 
No. of 
observations. 
Probable 
error. 
' Geat .... 
1890 
.55 
±0'6 
53 
±0-0007 
77 
±0-5 
Beiscoe . . . 
1890 
47 
±0-6 
44 
±0.0007 
56 
±0-4 
Watson . . . 
1891 
67 
±0-5 
72 
±0-0006 
95 
±0-3 
These figures are very satisfactory, and are a sufficient proof of the high standard of 
the work of these gentlemen. 
o 
Secular Corrections for the Reduction of the 1891 Survey to Epoch. 
It has been already explained that in the year 1890-91 we covered the country 
with a new network of stations which were to serve as a means of calculating the 
position of the terrestrial iso-magnetics on January 1, 1891. The corrections for 
secular change had, therefore, only to be determined for the short periods from April, 
1890, and from October, 1891 to the epoch. 
For this purpose we used the table given at the end of our last Memoir (“ Mem.,” 90, 
p. 325). As the corrections had to be applied to the observations made in 1892 for a 
longer interval of time—about 18 months—and as the period of maximum sunspots 
was approaching, it was necessary to inquire whether the same method of calculating 
the secular change could be applied to the work of that year without appreciable error. 
This could be tested for the south of England by comparing the actual change in 
the Declination at Kew between January 1, 1891, and July 1, 1892, with the tabular 
values of the secular variations for the same period. 
To assure ourselves that the observed results were free from any temporary 
disturbances, we calculated them in several different ways as follows :— 
Mean Declination at Kew for the whole of 1890 and 1891 = 17° 46''2 
,, for December, 1890 .= 17°46''8 
,, ,, January, 1891.= 17°45''7 
Mean of the two months . . 17° 46'’2 
Hence we get the same result for January 1, 1891, whether we take the mean of 
the two months or of the two years of which that day is the central date. 
We could not apply the same test to July 1, 1892, as the reductions were made at 
the end of that year before the Declinations in the later months were known. We 
therefore took the lour months from May to August, J 892, and compared the mean 
with that of the two months, June and July. 
MDCCCXCVI.—A. f 
