ON THE MECHANICAL EQUIVALENT OF HEAT. 
363 
switch worked in a funnel forming the top of a pipe leading to waste, while the other 
worked through a hole in the cover of the tank. The whole arrangement was fixed 
so that when in the central position the knife-edge was inch vertically below the 
nozzle at the end of the discharge pipe. 
This switch worked exceedingly well, diverting the stream of water almost instan¬ 
taneously, without making any perceptible splash. 
In the later trials this switch was connected by a chain of links with the revolution 
counter, so that when the latter was pushed into gear with the engine shaft the 
switch simultaneously directed the water into the tank, and vice versa. 
Weighing Machine and Tank .—(Part I., par. 13.) 
12 . To facilitate the weighing, the stream of water was led during each experiment 
into a galvanized iron tank which stood on the platform of a weighing machine. 
The tank was 4 feet long by 2 feet 9 inches deep, by 2 feet 9 inches wide. During 
the experiments it was kept covered by a lid of thin boards, steeped in paraffin 
wax. These boards were always weighed with the tank, so that any water they 
might absorb was accounted for. A. 24-inch valve in the tank bottom was used for 
discharging the water after weighing. 
The weighing machine was graduated up to 2200 lbs., and was supplied with 
three rider weights. 
No. 1, the largest, was provided with a knife-edge which fitted into grooves cut 
in the lever of the machine, each division representing 100 lbs. 
No. 2 worked on another scale on the lever, each division representing 1 lb., and 
graduated up to 100 lbs. 
No. 3 was made by Mr. Foster, in the laboratory, and indicated O’Ol lb. per 
division of the second scale. The lever was 32^ inches long, and readings wmre 
taken only when the middle of the swing of a pointer fixed to the end of the lever 
coincided with a line marked on a brass plate alongside it. 
It was quite easy in each individual weighing to set the machine to O'Ol lb., but 
owing, no doubt, to shifting of the platform, levers, &c., I do not think the readings 
taken were reliable beyond the i^th of a lb. 
Tliis machine was not at first quite as sensitive as was necessary to attain the high 
degree of accuracy required for the purposes of the research. On examination this 
was found to be due to the slightly imperfect adjustment of the knife-edges attached 
to the graduated lever. The fault was rectified by Mr. Foster, and since then the 
performance of the machine has been highly satisfactory. 
The Rubber Pij)e Connections to the Brake. 
13. On account of the very considerable pressure to which all the fittings of the 
3 A 2 
