repetition of M. Arago's experiments on mag7ietism, 493 
situation of the compass, the dip, and the axis of rotation, by 
laws not very easy to assign. This will explain some ex- 
pressions quoted by Mr. Barlow from his correspondence 
with one of the authors of this paper, which may appear 
otherwise to militate against the general view here taken. 
35. This diminution of the total effect by a more general 
distribution of the magnetism, was imitated by sticking a great 
number of needles vertically through a light cork circle, all 
being strongly magnetised, and having their north poles 
downwards, so as to form a circle, or, as it were, a coronet 
of magnets. This apparatus suspended centrally over a re^- 
volving copper disc, was not sensibly set in rotation. In this 
case, when at rest, the south polarity induced in the plate 
would be disposed in spots accumulated under each needle ; 
but these spots, elongated and blended by the effect of rota- 
tion, must produce a nearly uniform circle of south polarity, 
whose equal and contrary actions on all the needles would 
keep up the equilibrium, and prevent the coronet from ac- 
quiring a tendency either way. 
36. One consequence of this reasoning, which deserves 
trial, is this — that if the axis of rotation of an iron shell be 
situated in the direction of the dip, the spots occupied by its 
poles will not change their places by rotation, and conse- 
quently no deviation of the compass ought to take place from 
that cause. The experiment however is very delicate ; and 
care must be taken to remove any magnetised bodies whose 
influence might induce subordinate poles in the shell, whose 
places would shift by rotation. The compass therefore in 
this case cannot be neutralized by a magnet ;* but we must 
* In Mr. Barlow’s experiments, the large and powerful bar magnets used to 
