SURVEr OF THE BRITISH ISLES FOR THE EPOCH JANUARY 1, 1891. 
549 
turbanoes were arbitrarily increased or diminished by 33 (t.c., 0’0033 metric unit) 
even if the alterations were designed to produce the maximum change in the form of 
the lines. 
The mean differences shown in Table XXIII, when taken irrespective of sign, thus 
appear to be of the magnitude which the error of experiment would lead us to expect, 
and, since the algebraical means are in all cases much less, there is no reason to believe 
that residual errors tend to accumulate in any particular district. 
It is evident that some of the remarks which have been made with regard to the 
Vertical Force do not apply to the Declination and Horizontal Force. As we have 
laid no stress on the relative magnitudes of the Horizontal Disturbing Forces it is 
unnecessary to discuss them, but their direction is an absolute quantity, and anything 
which tended to alter the mean direction all over a particular district would make the 
results untrustworthy. 
The change in the calculated direction of the Horizontal Disturbing Force produced 
by a given error in the Horizontal Force or Declination, depends upon the magnitude 
of the Disturbing Force itself, and it is evident that no valid result can be attained, 
unless the magnitude of the Disturbing Force exceeds the error of experiment. 
We have already seen that the error of experiment in the measurement of H is on 
the average about dzO'OOlO. 
Hence there will be no reason for surprise if, when the Horizontal Disturbing Force 
is only 0'0020 or 0'0030 metric unit, two different measurements reduced by means 
of different isomaw’netics lead to discordant results. 
O 
Putting aside, on this account, the results at St. Leonards, Tunbridge Wells, and 
Charleville, we have tabulated the angles which the Disturbing Forces make with the 
geographical meridian at the repeat stations according to the two Surveys. 
In the fourth column is the difference between the results, and in fhe fifth 
the difference when the two are reduced to the mean isomagnetics. 
