548 
:ME. a. W. RUCKER AND DR. T. E. THORPE ON A :NrAGNETIC 
To this must now be added errors due to slight changes of position, to variability 
and uncertainty in the determinations of secular change, and to the use of different 
instruments by different observers. Taking all these into account, we do not think 
that it is too much to assume that the probable difference between the results of the 
two sets of observations at the repeat stations might be four times as great, or 
0•00224 metric unit. 
This estimate agrees very well with our experience in the 1886 Survey, during 
which on eight occasions we repeated the Horizontal Force observations at the same 
or at closely neighbouring stations, at intervals of less than two years. The places 
were Oban (Kerrera), Cambridge, Hai’penden, Manchester, Pteading, Dublin, Galway, 
and Enniskillen. The mean difference between the two measurements of Horizontal 
Force was 0'0017. 
If, then, we take the mean difference between two Horizontal Force measurements 
as 0’0020, since the tangent of the angle of Dip varies from about 2’5 to 3, the mean 
difference between the measurements of Vertical Force will vary from about 0'0050 in 
the south to O’OOGO in the north, on account of the error in the Horizontal Forces. 
The mean difference between the Dips obtained with the two needles used in each 
observation in our earlier survey was one minute of arc. The measurement of the 
Inclination was repeated at three of the above stations. The mean difference 
was F'4. 
As a change of F in the Dip alters the tangent of the angle of Dip by about one- 
thousandth part, the mean difference in the Vertical Force Disturbance would 
therefore be about 0‘0045, owing to the errors in the measurement of the Dip. 
On the whole, then, there seems every reason to expect an average difference 
of about 0'0070 metric unit between two independent measures of the Vertical 
Force made within a comparatively short interval of time. x4s the mean of the 
differences in Table XXIII. is only 0’0066, it does not appear that they can be 
affected to any appreciable extent by residual errors of any kind. In other words, 
the results of the two surveys as to Vertical Force Disturbances may be safely 
combined when both are reduced by the aid of the mean terrestrial isomagnetics. 
The range of the magnitude of the Vertical Force Disturbances at places where 
the surface is composed of sedimentary rocks, is about O'OGOO metric unit. On 
granite and gneiss the range is doubled, and in the neighbourhood of basalt (which 
we have avoided) it may be enormously increased. As, therefore, a single measure¬ 
ment does not on the average give a result correct to within d::0'0033 metric unit, 
we conclude that, although a doubt may sometimes arise as to the relative magnetic 
levels of two stations, the mean levels of groups of stations can be determined with 
adequate accuracy. This conclusion is borne out by a study of the maps of Aertical 
Force Disturbances given in our previous paper, and in the present Memoir (Map 12). 
No essential change, i.e., no change which would alter the positions of the principal 
ridge lines to any important extent would be produced if the Vertical Force Dis- 
