43^ The Account of a 
face of the bar was brought into an horizontal plane by means of 
screws and a spirit level. The brass points on the upper sur- 
face of the bar were brought into a right line, by stretching a 
silver wire along the top, and pressing the bar laterally with 
wedges, till all the points fell under the wire. Part of the 
chain was then placed on rollers, which rested on narrow slips 
of wood fixed on the side of the plank, about five inches below, 
and exactly parallel to the bar ; and whilst it was fastened to 
an adjusting-screw near one end of the plank, it was kept 
straight on the rollers by a weight of fifty-six pounds. 
From the extremities of the 20 feet on the edge of the bar, 
two fine wires with plummets were suspended, which were im- 
mersed in vessels of water, the wires hanging so as nearly to 
touch the chain. One end of the chain being then brought 
under its wire, by means of the adjusting- screw, a fine point 
was made on the chain coinciding with the other wire. This 
part of the chain was then shifted, and another 20 feet mea- 
sured in the same manner ; and the operation continued till 
the length of each chain was thus obtained at five successive 
measurements. The result was, that in the temperature of 
51-i 0 , in which the operation was performed, the chain A was 
found to exceed 100 feet by 0,114 inches, and the chain B, by 
0,058 inches. Now, according to the table of expansions in 
Vol. LXXV. Phil. Trans, the expansion due to T Fahrenheit 
on 100 feet of cast iron is 0,0074 inches, and that of the chain 
being 0,0075, their difference is 0,0001, and therefore for 2|° 
it will be 0,00025 9 consequently, as the points were put on the 
bar in the temperature of 54®, and the chains measured in 514° 
or 2 1® less, their lengths in the temperature of 54*, agreeing 
with the points on the bar, will be 
