MARYLAND GEOLOGICAL SURVEY 443

with scale erect, for example, would be taken, then the needle would
be turned around 180° in the stirrup—this by reason of the octagonal
form of the magnet and stirrup was a very simple matter—and the read-
ing of the new position obtained, about 8-10 readings on the average
being taken in all. The inversion of the magnet in the stirrup gave the
means of correcting for non-coincidence of the magnetic axis and of the
geometric axis of the magnet. Generally, additional readings for de-
clination would be taken at the close of the other observations, usually
at the end of the oscillation observations for intensity. A factor of
prime importance, however, is the elimination of the torsion in the
silk fibre by which the magnet is suspended. This torsion must either
be removed or allowed for in some manner. The general method is
to remove it before making the declination observations by suspend-
ing instead of the magnet a copper bar of the same weight. Every
one who has made observations in the field knows that generally the
removal of the torsion is a tedious matter, especially if the observer is
obliged to wait for it and is prevented from doing something else in
the meanwhile. In times of windy weather the removal of the torsion
by this method is especially likely to put the observer in a bad humor.
It was my endeavor, therefore, to reduce the torsion factor to such
an extent that the error due to the probable amount of torsion in the
fibre at any time would not cause a greater error in the declination
than the reading error of the horizontal circle, which was divided by
two opposite verniers to whole minutes and allowed 1/2 minute to be
estimated. The number of silk fibres supporting the magnet was
reduced from two to one and the fibres invariably soaked in glycerine
before using. For this purpose a small bottle of glycerine containing
silk fibres was always carried along. If the amount of torsion in the
single silk fibre was such that the position of rest assumed by the
copper weight was at right angles to the magnetic meridian, then this
amount of torsion was equivalent, on the average, to an angular devia-
tion of the magnet from the position it would have assumed had there
been no torsion of somewhat over 1'. For 30° of torsion the angular
deviation would be about 0/4. The torsion, with the exception of a
few special instances, was rarely removed in the field, but instead was