eg. 34' from a western azimuth, and 29 deg. 30' from amplitude;
but on the following afternoon, where the variation should have been
nearly the same, azimuths gave 24 deg. 12' and an amplitude 23 deg. 43' west; the
mean 5 deg. 35' less than off the Start. The same compass was always used,
and the ship's head was at west (magnetic), or within one point of it, in
all the cases; but in the first observations the compass was placed on
the binnacle, and in the last, was upon the booms. In order to ascertain
clearly what effect this change of place did really produce, I took
observations a few days afterward [MONDAY 27] with every compass on
board, and Mr. Thistle did the same upon the booms, ten or twelve feet
before the main mast, where the compasses were as far removed from any
quantity of iron, as they could be placed in any part of the ship. The
head was south-west by the steering compass, our latitude was 38 deg. 1'
north, longitude 14 deg. 18' west, and the results were as under.

Binnacle. Booms.
Variation from an azimuth compass by Walker, deg. ' deg. '
marked No 1: mean of both sides of the vane. 25 47 22 17 W.
From a ditto marked No. 2, 25 35 19 15
ditto marked No. 3, 24 41 21 27
Walker's meridional compass, 25 46 - -
Ditto used as a common azimuth, 25 51 20 35
Compass made by Adams, 25 44 21 9
--------------
Means, 25 34 20 57 W.
--------------

Thus a change of place from the binnacle to a little before the centre of
the ship, produced an alteration of 4 deg. 37' in the mean variation, the
same way as, but a less quantity than Mr. Thistle had found it off the
Start, when the ship's head was west. The true variation I judge to have
been 23 deg., and that the observations on the booms showed 2 deg. too little,
and those on the binnacle 21/2 deg. too much. The error in excess, upon the
binnacle, appeared to continue so long as the ship was in the northern
hemisphere and the head to the westward; but it diminished gradually as
we approached the equator, and the observations on the binnacle and booms
then nearly coincided. This example is sufficient to show the