SURVEY OP THE BRITISH ISLES FOR THE EPOCH JANUARY L 1891. 
31 
1886 survey, and have, as was to be expected, found small inaccuracies in tho 
latitudes and longitudes of several of them, as given in our earlier paper. The 
correct values are entered in the Table of Errata at the end of this Memoir. In no 
case have the errors been of any importance except in so far as they might make it 
more difficult to recognize the precise spots at which we chserved. 
It is, however, very probable that the latitudes and longitudes of others of tlie 
1886 survey stations are affected with similar small errors, but, if these cause 
any doubt, the written descriptions and bearings of conspicuous objects which v/e 
recorded will in general be sufficient to resolve it. 
Methods of 2'aking, Correcting, and Tabulating the Observations. 
On all these points the methods adopted were, with one or two exceptions, 
practically identical with those used in our previous survey. The modifications of 
the telescope of the Kew Magnetometer introduced by us (‘Phil. Mag.,’ Aug. 1888, 
p. 122), were applied to all the instruments, and were found to facilitate the sun 
observations. 
Some new forms have been used in the tabulation of the observations- In 
particular a convenient double page form has been used for the calculation of the 
sun’s azimuth, in which a place is provided for every detail of the calculations. 
In order to remind the observers of the various points to be attended to in 
the choice of a station and during the observations, a form was provided, in which 
questions were asked which they answered at the time of observation and while still 
at the station. 
As the number of stations to be dealt with was so large, it was necessary to make 
the vmrk of computing as light as possible. Except, therefore, in the case of the two 
supplementary surveys made by ourselves in 1889 before the large extension of the 
survey was determined on, we omitted the more rigorous method of making the two 
values of H and m (the moment of the magnet), given by each experiment, abso¬ 
lutely independent (“Mem.,” 90, pp. 72-73). The values of the quantities v and A, 
defined in our last paper (pp. 70-76), were also used with the signs changed, as it 
was found that fewer errors were made in the computing if all such quantities were 
entered as corrections instead of as errors. 
The following Table gives the mean values of P (in metric units) obtained for each 
year during the survey. For the purpose of comparison the part of the Table 
published in our former paper is added. 
