length of the pendulum vibrating seconds . 47 
at a certain temperature, and altitude of the barometer, in an 
infinitely small arc, the great weight being above. 
The frame of the support was now elevated, the pendulum 
was inverted, placed in the Ys, with the great weight below , 
and the knife edges being gently let down as before on the 
agate plane, the same process with respect to the observa- 
tions was followed, which has just been described. And if 
the mean temperature differed from that in the former posi- 
tion of the pendulum, the mean number of vibrations was 
corrected for such difference of temperature, the expansion 
of the pendulum being known by experiments hereafter to 
be detailed, and consequently the gain or loss in 24 hours by 
a given change of temperature. 
The mean number of vibrations thus found, differing from 
that given in the former position of the pendulum, the second 
weight was moved, the number of vibrations again deter- 
mined, and the pendulum being inverted, the process was 
repeated until the vibrations in 24 hours, in either position 
of the pendulum were brought as near to an equality as could 
readily be effected by means of this weight ; it was then 
firmly secured in its place. 
Whatever alteration may be made in the arrangement of the 
weights, the effect on the vibrations ( except in one particular in- 
stance ) will be the same in both positions of the pendulum , always 
increasing or diminishing their number in both cases, though in 
different degrees ; and, the vibrations will be least affected by such 
change when the great weight is below, and will consequently be 
nearest to the truth in this position. No doubt, therefore, can 
arise, as to the kind of correction required. The number of 
