Effect of Magnetism on Time -pieces. 375 
from the magnet, and resumed its former position when- 
ever the magnet was withdrawn. 
No doubt now remaining as to the facts, and being in 
possession of the position of its poles, I proceeded to ex- 
amine the effects produced by this cause upon the watch’s 
rate of going. Having put on the pendulum spring, and 
replaced the balance in the watch, I laid the watch with 
the dial upwards, that is, with the plane of the balance 
horizontally, and in such a position that the balance when 
at its place of rest should have its marked side towards 
the north in this situation it gained 5' 35" in twenty- 
four hours. I then changed its position so that the mark- 
ed side of the balance when at rest should be towards 
the south, and, observing its rate of going for the next 
twenty -four hours, found it had lost 6' 48"— producing, 
by its change of position only, a difference of IS' S3" in 
its rate. It must be obvious to every person, that even 
this difference, great as it was, would be increased or di- 
minished as the wearer should happen to carry in his 
waistcoat pocket a key, a knife, or other article made of 
steel. This circumstance, taken along with the amount 
of the variation occasioned by the polarity of the balance, 
was fully sufficient to produce all the irregularity ob- 
served in the going of the watch. 
I then took away the steel balance, substituted one 
made of gold, and, having brought the watch to time, ob- 
served its rate of going, and found it as uniform as any 
watch of the like construction ; for, though it was a du- 
plex escapement, which is perhaps the best yet invented, 
at least for common purposes, it had no compensation for 
the expansion and contraction occasioned by heat and 
cold, and therefore a perfect performance was not ex- 
pected. 
Steel balances being commonly in use, and, on that 
account, easiest to be procured, and being on many ac- 
