diurnal variation of the magnetic needle^ ^c. 341 
earth and the atmosphere, should be symmetrical. Another 
circumstance in which the experiments differ from the hy¬ 
pothesis, is this : that in these the needle is not only of finite 
dimensions compared with the distance of the revolving poles 
from the centre, but its length is even considerable in this 
respect, being a fifth of that distance; and this will modify 
all the results. 
I have before mentioned that the intensity of the magnet 
whose north pole was upwards in these experiments, and 
which, for the sake of distinction, I will now call N, was 
considerably greater than that of the magnet whose south 
pole was upwards, and which I call S. Their intensities were 
such, that when they were severally placed to the west of the 
needle, so that their axes were in a horizontal line drawn 
through its centre at right angles to the magnetic meridian, 
and their nearest ends 4-2 inches from that centre, the devia¬ 
tions of the north end of the needle were : for the magnet N, 
60°40' W., and 59°4 o'E., according as its north, or its south 
pole was that nearest to the needle; and for the magnet S, 
26° OQ> E, when the south pole was nearest to the centre, and 
31° 00' W., when its north pole was nearest. I afterwards 
increased the intensity of the weaker magnet by passing the 
poles of the other alternately over it from centre to ends. 
After this the deviations observed as before were: for the 
magnet N, 60“ 20' W., 58° 20' E.; for the magnet S, 39° 40' E., 
41° 30' W. In this state the magnets were again adjusted to 
the instrument as before, excepting that the distance of their 
poles from the centre was 10 inches instead of 10 25 inches, 
as, at the latter distance, the stem of the support to the 
compass was liable, in some cases, to interfere with their 
