AND STRAIN ON THE ACTION OF PHYSICAL FORCES. 
71 
the cylinder with an air-pump and barometer-gauge. The whole arrangement liaving- 
been placed in a box the cylinder was well packed up in sawdust, and after a sufficient 
time had been allowed to elapse to enable the temperature of the wires to become 
constant the experiments began. 
This plan entirely failed, the effect of the alteration of pressure on the resistance 
of the wire being completely masked by the change of temperature caused by the 
rarefaction or condensation of the air even when a considerable time had been allowed 
to elapse. The condensation and rarefaction of the air would cause a change of resistance 
the opposite to that looked for as the result of change of pressure ; and so slowly did the 
temperature alter after the first half-hour that I began to suspect that increase of pressure 
caused increase of resistance , and for some days actually tried to measure the increase of 
resistance apparently caused by increase of pressure. Finding, however, that the 
observations did not agree sufficiently with each other I filled the cylinder with water 
and immersed it in a large glass vessel, also filled with water, which was stirred from 
time to time. This plan also failed, the compression of the water causing an effect 
in the same direction though not to the same extent as the compression of the air 
had done. It was noticed, however, that after the air was let into the vessel there was 
no change of resistance till a second or so afterwards. 
Finally, the following method was adopted :—The vessel containing the wire to be 
tested consisted (fig. 14) of a brass cylinder closed at one end, and which could 
Fig. 14. 
I 
! 
be closed at the other by a brass cover A. A. The vessel was about 7 inches deep 
and 6 inches in internal diameter; the open end and the cover which closed it having- 
been carefully ground, so that with the help of a little grease it could be made 
perfectly air-tight. The wire under examination was wound double on the outside 
of a very thin brass cylinder B B, and the ends of the wire were soldered to the 
stout copper wires C, C, which latter passed through an indiarubber cork, fitting air¬ 
tight into a tube about 1 inch in diameter, which was soldered into the vessel and 
made an angle of 45° with the side. The comparison-wire was wound double round 
the outside of the large cylinder, and the two stout copper wires D, D, to which the 
