98 
DE. JOULE ON SOME THEEMO-DYNA^OC PEOPEETEES OP SOLIDS. 
Carbon. 
SiHcium. 
Cast iron 
. . 2-275 
2-720 
Puddled bar 
. . 0-296 
0-120 
Iron wire . 
. . O-III 
0-088 
Sulphur. 
Phosphorus. 
Iron. 
Total. 
0-301 
0-645 
94-059 
100 
0-134 
0-139 
99-311 
100 
0-094 
0-II7 
99-590 
100 
From the above digression I now return to the main subject of the present paper, b} 
describing my 
Experiments on the Thermal Effects of Tension on Solids. 
18. All the metals employed, with the exception of lead, were in the form of cylin- 
drical bars about a foot long and a quarter of an inch in diameter. The upper end of 
the bar was screwed into a piece of metal supported by a wooden framework, the lower 
end was attached to a lever, to the extremity of which weights could be hung without 
approaching the apparatus. In the first method the thermo-electnc junction was made, 
as represented in the adjoining sketch, by binding to opposite sides of the bai- copper 
and iron wires 4 ioth of an inch in diameter, hammered flat at the ends. In the second 
method, the junction of fine wires was inserted in a hole ^ofh of an uich in diameter, 
bored through the centre of the bar, a small portion of nme being bound to the bar by 
means of cotton thread, but metallic contact prevented by an intervening slip^ of paper. 
The terminals of these wires were immersed in deep mercui-y cups formed in a solid 
block of wood, whence thick copper wires proceeded to the thermo-multiplier. 
19. Immediately after each experiment on the effect 
of tension, the thermometric value of the deflections was 
ascertained by immersing the bar, to within one-third of 
an inch of the junction, in water of different tempera- 
tures. The deflections thus produced were about two- 
thirds of those occasioned by the same changes of tem- 
perature when the junction was completely immersed. 
The diminished effect in the former case is oiving for 
the most part to the conduction of heat from the air by 
the thermo-electric wires. The experiments on tension 
were liable to be affected in the same way, but they were 
not subject to the loss arising from the conduction of heat from the surface of the bath 
to the junction. The error intervening from this latter circumstance could not be great, 
and was moreover in all probability almost exactly neutralized by a small error in the 
tension experiments, arising from the escape of f^th. of the thermal effect from the 
quarter-inch bars during the 40" occupied by the swing of the needle. 
20. Iron . — The weight of the bar per foot was T568 of a lb. The lever alone gave 
it a constant tension of 70 lbs. The additional tensions successively given were 194 lbs., 
388 lbs., 583 lbs., and 1166 lbs., or as I, 2, 3, and 6. The mean of five trials gave 20'-6 
as the deflection indicating cold on applying the tension of 194 lbs., and 20'-2 as the 
