t 4S9 ] 
index fg, exactly like the former, except that it is 
kept in an horizontal pofition, by the fcreve k bearing 
cigainft the end of the iron rod. When the experiment 
is to be made, the pendulum is fir ft put into a box, 
and gradually heated by a large lire, to a confiderable 
degree, being often turned, that every part may be 
equally expofed to the fire. And having continued 
flint up in the box for l'oine time after it is removed 
from the fire, that the two rods may be heated as 
uniformly to the fame degree as pofiible, the pendu- 
lum is hung upon the wire, and the two indexes 
made to ftand nearly in an horizontal pofition. The 
two graduated plates h i are then flid upon a wire, 
till the divifions in each mark’d s are pointed to by 
the indexes. As the pendulum cools, the lower in- 
dex will be feen gradually to defcend but if the 
ends of the two fcrews, in the ball of the pendulum, 
aft upon proper parts of the levers, the upper index will 
continue in the fame place. If the ends of the fcrews 
are either too far off, or too near the centres of the 
levers, the index will either rife or defcend ; and, by 
comparing the number of divifions it has varied, with 
thofe which the lower index has varied, a near efti- 
mate may be made, how much the fcrews require 
to be alter’d ; and, in a very few trials, they may 
eafily be adjufted to a very great exadnefs. In 
order to make an actual trial, how far this contri- 
vance of the pendulum will anfwer the end propofed, 
it is neceftary, that the clock, to which the pendulum 
is fitted, be made with great exadnefs, and intirely 
to be depended on : For otherwife the experiments 
will be very uncertain, as I found in the clock I firft: 
made ufe of. 
0-31 
I have 
