SURVEY OF THE BRITISH ISLES FOR THE EPOCH JANUARY I, 1891. 537 
attains a certain value lies somewhere between the places at which it is less and greater 
than that amount respectively ; (2) that in some few instances it is impossible to decide 
between alternative paths, and that in these cases the lines show what, in our judg¬ 
ment, is the most probable representation of the facts ; (3) that, in some cases, purely 
local disturbances have possibly led us to depict the true lines as departing more from 
the terrestial lines than they really do. We are, of course, fully aware of this risk, and 
we can only say that in the few cases in which the evidence for such deviations 
depends on one station only, they are not more remarkable than those whose real 
existence is conclusively proved. 
Reduction of the Terrestrial Lsomagnetics Obtained from the 1886 Survey 
TO THE Epoch of the New Survey, and Determination of the Mean 
Calculated Terrestrial Isomagnetics. 
We published in our “ 1890 Memoir” tables of the calculated values of the elements 
at the intersections of whole degrees of latitude and longitude for the epoch January 1, 
1886. 
By adding (algebraically) to these the secular changes for the five years 1886-91 
given in Table XIV., p. 530, we obtain values which, if the terrestrial isomagnetics 
for 1886 and 1891 were absolutely correct, and if the secular change had been per¬ 
fectly determined, should agree with the numbers given in Tables III., VI., and IX. 
As a matter of fact there is a very good agreement throughout the centre of the 
Kingdom, but near the borders (where the errors of the isomagnetics are likely to be 
relatively large) the discrejiancies are considerable. 
This being the first occasion on which a comparison has been possible between two 
sets of isomagnetics drawn from observations made at large numbers of stations, with 
a short intervening period of time, it is desirable to give the details of the accord or 
disagreement between them. 
In the following tables, therefore, three numbers are given as corresponding to each 
point defined by whole degrees of latitude and longitude. 
The first of these is obtained from the 1886 survey and the secular change. 
Thus, according to Table VIII., “ 1890 Memoir,” p. 322, the declination at lati- 
tuue 55, longitude 4, was 20° 57’‘4. From Table XIV. of this paper the secular 
change at the same place has been 36'‘3. From these data the declination on 
January 1, 1891, would be 20° 57''4 —36''3 = 20° 21'T, which is entered in Table XVII. 
According to Table III., p. 508, however, the value found by the present survey is 
20° 20''1. This number is also entered in Table XVII., and the difference, 1'‘0 
(taken positive when the 1886 survey gives the larger result), is the algebraical sum 
of the errors introduced in the operations of determining the positions of the two sets 
of isomagnetics and of reducing them to the same epoch. 
mdcccxcyi.—A. 3 5 ; 
