SURVEY OF THE BRITISH ISLES FOR THE EPOCH JANUARY 1, 1886. 
75 
Values of m. 
Observed. 
Mean. 
Corrected. 
0'0091945 
0-0091963 
0-0091954 
0-0091965 
Values o/H. 
Observed. 
Mean. 
Corrected. 
#7. 
Reduced. 
1-8363 
1-8368 
1-8365 
1-8360 
- 0-0051 
1-8309 
At places where the deflection experiment was not made, the moment of the magnet 
was calculated from the other moments observed during the same tour. It was 
assumed that the strength of the magnet diminished regularly, and a linear expression 
was determined by equations of condition which gave the rate of decrease. A careful 
analysis of the Scotch observations, which were the first reduced, proved that this 
method was satisfactory. 
The greatest difference between observation and calculation occurred at Kirkwall, 
and amounted to 0'234 per cent. It is certain that in this case the difference is not 
due to errors of experiment, but to a real change in the moment, as the observations 
obtained at the next station (Lerwick) confirmed it. We are also able to assign a 
probable cause for the alteration, as the “Coventina” was caught in a “roost” oft’ 
Stromness, and experienced rather rough treatment, during which the magnet may 
have been jarred. 
The mean difference (Irrespective of sign) at all the stations was 0’057 per cent. ; 
and if we put aside about one-fifth of the whole number of stations at which, owing to 
bad meteorological or magnetic weather, or some similar cause, the conditions under 
which the observations were taken were not very favourable, the mean difference 
at the remainder was 0'028 per cent. 
It must, of course, be remembered that this quantity is not a measure of the error 
of experiment, but of the algebraical sum of that error, and of the deviation of the 
rate of decrease of the moment of the magnet from perfect uniformity. The conclusion 
we came to from the discussion of the Scotch observations was that if the observed 
moment differed from that calculated by the linear formula by OT per cent., either a 
real change had taken place in the magnet or the obse)’vation had been affected by 
some disturbing cause, which it was in general easy to specify. In the later tours the 
moments of the magnets changed very slowly, and it was sufficient to take the mean 
L 2 
