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



of the above equation. Edlund* who later repeated these 

 investigations with the best methods then known obtained 

 thermal effects which were some 40 per cent smaller than the 

 theoretical. Taking Edlund's results as a basis of calculation 

 I found that in some cases these thermal effects could affect 

 my measurements by as much as "1 mm. The influence of 

 the after-effect was much less, as will be shown later. 



On account of the uncertainty in determining at any given 

 instant the exact temperature of the wire after it had been 

 stretched, I used in nearly all my experiments fine wires of 

 from - 2 mm. to 0*3 mm. diameter, for such wires assume quickly 

 the temperature of the surrounding air. In order to estimate 

 the rapidity with which the wire returned to its normal tem- 

 perature after a weight had been placed iipon the pan, I used 

 a thermo-pile with mirror-galvanometer and scale. The thermo- 

 pile consisted of two very fine wires, one of iron and the other 

 of german-silver, fastened to the main-wire with the least pos- 

 sible amount of solder. 



The sensitiveness as determined by the mean result of four 

 experiments was 73 scale-divisions for 1° Celsius. In every 

 case only the first deflection was recorded. The time of vibra- 

 tion of the needle was 6*5 sec. In determining the sensitive- 

 ness of the thermo-pile the junctions of the iron and german- 

 silver wires with the galvanometer wires were immersed in 

 petroleum of constant temperature, while the junctions with 

 the main-wire, which were some 15 mm. apart, were immersed 

 in a little rubber vessel containing distilled water. The temper- 

 atures of the water and of the petroleum were accurately 

 determined, the circuit closed, and the deflection of the needle 

 noted. 



In measuring the thermal effects produced by the stretching 

 of the wires the junctions of the thermo-pile wires with the 

 main-wire were protected against air currents by a little cotton. 

 Without this cotton the rapidity with which the wire near the 

 junctions reassumed its normal temperature would have been 

 even greater than it was. In a series of 8 consecutive deter- 

 minations the mean position of rest of the galvanometer-needle 

 was 635 "7, the cooling of the brass wire occasioned by a weight 

 of 1'5 kg. caused an average first deflection of 9 - l scale di- 

 visions toward the smaller numbers, and after 13 sec. or two 

 vibrations the mean position of rest was 635*9. 



A further proof of the rapid disappearance of the thermal 

 effects is given by this very deflection of 9'1 scale-divisions. 

 The first deflection was, as already mentioned, 73 scale-divis- 

 ions for 1°. The cooling of the wire produced by the weight 



* Edlund, Pogg. Adii., .cxxvi, p. 557, 1865. 



