and magnetic rotations. 521 
and has ascended. Such an hypothesis might explain Expe- 
riments 18 and 19, if, on measuring its influence, the cause 
were found sufficiently powerful ; but it appears from Expe- 
riment 21, that when the needles are not visibly bent in the 
form of vanes, this cause is not sufficient to produce the 
effect, although it acts in the same direction. The heated 
air above a revolving plate will revolve in the same direc- 
tion ; and if it pass through the interstices of the muslin, it 
ought to impress motion in the same direction as the plate 
from which it rises. The heated air above the zinc screen 
had not this rotatory motion ; and we have seen by experi- 
ment 21, that its vertical action was not sufficient to produce 
the effect. 
In many of the experiments no heat was employed ; and 
yet the same motions resulted, as is apparent by Experiments 
5 and 6 for direct motions, and by Experiment 10 for the 
retrograde. 
5th. Vibrations excited in the apparatus by the motion of 
the jack. This cause was decisively refuted by Experi- 
ment 17, in which they were measured, and found to be 
quite insensible. Also in Experiments 18 and 19 the appa- 
ratus was not in motion, yet the needles did move. 
6th. The torsion of the silver wire to which the needles 
were attached. Until I had measured this torsion by revers- 
ing the direction of the moving plate, as in Experiment 15, I 
had some suspicion that the heat of the lamp might have 
altered the elasticity of the wire ; and although I did not 
then see how the reverse motion could be well accounted 
for, yet I was desirous, if possible, of removing altogether 
this disturbing action. Some of the previous experiments in 
