413 
imi arising from its rotation. 
needle became stationary during the rotation, the direction 
of its north end was observed. The point o on the plate was 
again made to coincide as quickly after the rapid rotation as 
possible, and the direction of the needle observed, in order to 
see if that rotation had produced any permanent change in 
the iron ; the same was done when the point 180 again coin- 
cided. Observations precisely similar to these were made 
when the upper edge of the plate revolved from south to 
north. 
Although the centre of the plate was stationary, and the 
needle was placed in certain positions with respect to it, I 
consider, as before, the situation of the centre of the plate 
with reference to the plane passing through the centre of the 
needle perpendicular to the dip ; and its angular distance 
from this plane, the equator, was measured on a circle of 
9 inches radius parallel to the meridian, passing through the 
centre of the needle, and at the distance 1.45 inches from it, 
so that the centre of the needle was always at tliis distance 
from the edge of the plate, east or west. As the needle was 
only two inches in length, and the rim of the compass 
divided into degrees, the direction of the needle could not be 
observed nearer than to 5', and indeed scarcely to that de- 
gree of accuracy. The mode which I was under the neces- 
sity of adopting in adjusting the compass to the several posi- 
tions did not admit of extreme accuracy, so that these 
positions may be considered as liable to errors amounting to 
1®, or perhaps rather more, in angular distance from the 
equator ; but as my principal object was the comparison of 
the deviation due to the slow and rapid rotation of the plate, 
when its centre was in precisely the same position with re- 
spect to that of the needle, this was not very material : it 
