SURVEY OF THE BRITISH ISLES FOR THE EPOCH JANUARY I, 1891. 551 
It is satisfactory to be able to state that, for the most part, stations at whicli the 
Horizontal Disturbing Force falls below this limit occur in groups, showing that the 
apparent smallness of the forces is due to a real absence of disturbance. Again : 
though at all, or very nearly all, the stations where a difficulty of interpreting the 
direction of the Horizontal Disturbance arises, its magnitude falls below the limit, yet 
in most cases the direction of the small Disturbing Forces agree well with each other, 
and with the results obtained at neighbouring stations where the disturbances are 
greater. The concordance is, in fact, so good as to prove that if we have made any 
mistake in the above discussion, it is in over-estimating the average magnitude of 
♦ 
the error of experiment. 
As regards the discrepancies at other stations, to whatever cause they are due (for 
we do not think that they are always errors of experiment in the ordinary sense), and 
large as they may at first sight seem, they are for the most part unimportant if the 
exact nature of our inquiry be borne in view. Our conclusions are based upon two 
independent groups of data : (1) the relative values of vertical disturbance at neigh¬ 
bouring stations; (2) the general drift of the direction of the Horizontal Forces. 
Apart from the fact that these lead to the same conclusions, our stations are so 
numerous that considerable errors in the directions of the Horizontal Disturbing 
Force at a few places could not seriously mislead us. 
The means (taken irrespective of sign) of all the numbers in the last two columns 
of the table on p. 550 are about 19° and 13° respectively. Hence, on the average, our 
results are correct to 9°’5 if referred to independent terrestrial isomagnetics, and to 
6°'5 if referred to the mean lines. This accuracy is quite sufficient for any use we have 
made of the observations. As to the occasional larger errors, we represented them 
graphically in our previous paper (“ 1890 Memoir,” p. 269), and the two sets of arrows 
are shown for the repeat stations in Map 13 of the present Memoir. At no less than 
eight stations the directions are so nearly the same that we cannot depict them sepa¬ 
rately on the scale on which the map is drawn. 
An inspection of the remaining fourteen is sufficient to show that our conclusions 
would not be affected if we were arbitrarily to reject either of the directions at a ny 
station. 
The substantial agreement of the two surveys is further illustrated by the following 
tables. They are arranged exactly like tlie tables on pp. 510. 515, and 521, to show a 
number of points through which the terrestrial isomagnetics pass. 
The upper figure in each square refers to the isomagnetics deduced from the 1891 
survey alone; the lower to the mean isomagnetics for January 1, 1891, deduced 
from the two surveys. 
