ON THE MECHANICAL EQUIVALENT OE HEAT. 
379 
Its correct length was, therefore, 
30 inches =- 0*0003 inch. 
As the barometer was only gi-aduatecl to 0*001 inch, no error was introduced in 
assuming the bar to be exactly 30 inches long. 
(Part. I., par. 48).—For the purpose of transferring this standard 30 inches to the 
bi’ass sleeve forming the scale of the barometer, a circular 
cast-iron surface plate was made. This plate had two 
pieces cut out of it, as shown in the sketch. The plate 
was fixed with its surface level, and then the brass sleeve 
was placed centrally upon it, standing upright on its 
lower bevelled edge. In this position the portion of the 
surface between the two grooves cut in the plate corres¬ 
ponded exactly to the surface of the mercury in the 
barometer between the two windows previously men¬ 
tioned. As it was probable that in actual use the lower 
bevelled edge would be slightly above the mercury sur¬ 
face, the sleeve was packed up by means of some very 
fine sheets of tissue paper till a line of light could be 
seen under it. Four sheets were necessary to effect this ; one of these was removed, 
and then the standard 30-inch bar was placed inside the brass tube, standing with 
one end on the surface plate. The upper bevelled edge was then adjusted till the 
line of light between it and the top of the steel standard was obscured, and the scale 
was made to read 30 inches in that position. 
Together with Mr. Foster I made this adjustment a number of times, but after 
once fixing the 30-inch mark, the reading of the length of the steel standard never 
varied by as much as 0*0003 inch from 30 inches. 
Unfortunately, the comparison was made at a temperature of 67° Fahr., while the 
standard temperature of the Whitworth gauges was 60° Fahr. A formula of reduc¬ 
tion of the readings of the barometer therefore became necessary at all temperatures. 
Taking for the coefficient of linear expansion of brass per ° Fahr. 0*000012 
,, ,, ,, ,, steel ,, 0*0000066 
,, „ ,, „ the mercury 
column of the barometer. 0*0001. 
Then at 67° Fahr. the true length of the brass barometer scale 
^ 1 -f 35 X 0-000Q066 
” 1 28 X 0*0000066 
= 30*000138 inches. 
3 c 2 
