J. 0. Thompson — Law of Elastic Lengthening. 37 



of 1*5 kg. amounted to 225° according to Edlund, and this 

 difference of temperature, if it had remained constant during 

 the time of the first deflection, would have produced a deflec- 

 tion of 164 scale-divisions instead of the actual 9*1. From 

 this one can draw conclusions as to the rapidity of the disap- 

 pearance of the thermal effects. 



In all my experiments, even in extreme cases, I uniformly 

 found that after a weight had been placed on the pan the 

 needle would be deflected so as to indicate a certain cooling 

 and then return with scarcely perceptible delay to its former 

 position. Thus it was clear that after 13 sec. no visible trace 

 of the cooling effect remained. If a difference of tempera- 

 ture of -^°, enough to produce a contraction of 0*005 mm. in 

 the length of the wire, had remained, the galvanometer would 

 surely have indicated it. 



The indications of the galvanometer were confirmed by 

 noting the gradual contraction or lengthening of the wire 

 immediately after unloading or loading the scale pan. In the 

 first 10 or 12 sec. there would be a change of say O07 mm. in 

 the length of the wire, and then in the following minute a 

 further change of O01 mm. The first change one would natu- 

 rally attribute chiefly to the thermal changes, the second to the 

 after-effect. Since it is hardly possible in less than 12 sec. to 

 place a weight on the pan and set the cathetometer with accu- 

 racy, it was to me a matter of indifference how much quicker 

 the wire returned to its normal temperature. 



Here it may be mentioned that Wertheim did not make his 

 measurements until 5 or 10 min. after loading the pan. He 

 waited, as he says, until the position of the mark became con- 

 stant. In other words he did exactly what he should have 

 avoided, he allowed his measurements to be affected by the 

 most active part of the after-effect. Miller on the contrary 

 went too far in the opposite direction and made measurements 

 2 sec. after stretching the wire. His results are obviously 

 affected by temperature errors which he in a later article* 

 seeks to correct. 



The After-effect. 



The complete calculation of the influence of the after-effect 

 on my results would have required a large amount of time 

 aud trouble. 



In the first place these phenomena have been studied chiefly 

 in the cases of torsion and bending, and the laws of the after- 

 effect in the case of stretching, at least as far as the metals are 

 concerned, are unknown. Further in the present case we have 



* Miller, Wied. Ann., xx, p. 94, 1883. 



