278 Mr. Barlow on the secondary deflections produced 
riments, but without being able at that time to reduce them 
to uniform laws. From the experiments we did make, it 
appeared, however, that if the ball was raised just above the 
plane of the table, and the compass carried round the table, 
proceeding, for example, from the north towards the east 
or west, the deteriorated branch of the needle receded from 
the ball; and this happened also beyond the east and west 
points to a certain azimuth, after which the deteriorated 
branch approached the ball. Precisely the same occurred 
when the ball was placed just below the table, beginning 
however now at the south instead of the north. The points 
of change being in this case between the north and east or 
west point, and in the former between the south and east 
or west point. If the ball was placed exactly with its 
horizontal section in the plane of the table, the law ap¬ 
peared still more anomalous; but in carrying the compass 
round the ball in the magnetic equator, or plane of no 
deviation, the deteriorated branch in all cases approached 
the ball. 
In page 55 of the first edition of my essay, I have stated 
that some of the discrepancies I found between the observed 
and computed deviations, were probably due to an unequal 
distribution of magnetism in the two branches of the needle ; 
and there could be no doubt that this was actually the case 
in the present instance; but I had no idea of the great 
amount of error to which my first observations might have 
been subject, had this inequality of magnetism been greater 
than it was. In the experiments above referred to, the error 
amounted to 2°, 3°, and even 5® and upwards; but as the 
actual amount depended principally upon the extent of dete- 
