TRANSACTIONS OF SECTION A. 503 



The author has, with the assistance of one of his students, Mr. D. Low, looked 

 for the effect in question and has found it, not only iu iron but in steel, brass, and 

 copper wires. lie has not yet e.\amined other metals. 



The experiment consisted iu observing, with much optical magnification, the 

 e.xtension of a very long piece of wire, directly loaded with lead weights. The 

 wire was hung from a rigid support in a testing flue or recess, built in the wall of 

 the laboratory, and extending up through four storeys. At a distance of 806 cms. 

 below the top a small clamp was flxed on the wire, and this formed the support of 

 the back foot of a little tripod, the feet of which consisted of three needle points 

 about an inch apart. The two front feet rested in a slot and hole in a fixed shelf 

 which stood in front of the long wire. The tripod carried a plane mirror which 

 became tilted forward or backward as the wire extended or retracted. Readings 

 were taken by a telescope of the reflected scale of a levelling staff placed vertically 

 at a distance of some 5 metres from the mirror. The start' was graduated to jj.jth 

 of a foot, and it was easy to read by estimation to lyjjjth of a foot, which corre- 

 sponded to 0-000000102 of the length of the wire. At first a fixed sheif was used 

 to support the two front feet of the mirror, but the effects of temperature were 

 found to be excessive, although the greatest care was taken to shield the wire from 

 draughts; and the plan was resorted to of hanging the shelf from two adjacent 

 wires of the same material, suspended from the same support as the wire which 

 was to be stretched. 



In this form the method of optical multiplication was nearly the same as one 

 used by Mr. Bottomley in recent experiments on the extension of loaded wires by 

 heat.' 



The first wire tested was of iron, in the hard-drawn state, 1 '08 mm. in dia- 

 meter. A weight of 14 lbs. was kept permanently on it, and an additional weight 

 of 20 kilos, was applied and removed many times. Then readings were taken at 

 intermediate loads (the extreme load caused so much tilting of the mirror that 

 readings could not be taken throughout the whole range). At 10 kilos, (the 

 middle of the cycle) on the way up the scale reading was 2,.340 ; with the same 

 weight on the way down it was 2,305. The difference, due to hysteresis, is 35, or 

 0'0OOO0357 of the length. The whole extension per kilo, was 533 scale divisions, 

 so that the change of length due to the full load of 20 kilos, was 20 x 533 = 10,GG0 

 scale divisions. Hence the amount of difference in length at the middle position, 

 namely at 10 kilos., in the loading and unloading processes was j^'f^o, or, say, 

 ^J^th of the whole extension between the extreme load. Another way of 

 stating the result would be to say that the extension had the same value at 10-OGO 

 kilos., while the load was being increased, as it had at 10 kilos., while the load 

 was being diminished : the eff'ect of hysteresis is equivalent to a difference of 60 

 grammes in the load. 



Further experiments were made to test how far this hysteresis depends on the 

 time rate of loading and unloading. In the above case, the cycle of load had been 

 performed as quickly as was practicable, the only pauses being at 10 kilos, on the 

 way up, and at 10 kilos, on the way down. It was found that even when the 

 pause at 10 kilos, lasted for two hours no distinct change took place in the read- 

 ings. So far as this experiment showed, the hysteresis is persistent. It appeared, 

 however, that pauses at the extremes of load have the effect of increasing the 

 difference of readings at the middle. In a slowly performed cycle where a pause 

 of about three minutes was allowed at each step (of 2 kilos.) iu the processes of 

 loading and unloading, the difl'erence of the readings at 10 kilos, was about 50 

 scale divisions instead of the 35 divisions observed in the former case. 



A similar trial was made with mild steel wire, previously annealed and then 



hardened by the imposition of a load much greater than the load afterwards used 



in the testing cycle. Here the wire was 0'94 ram. in diameter. There was, as in 



the former case, a constant load of 14 lbs., and the cycle was made by applying and 



removing 12 kilos. At G kilos, the readings were 4,695 and 4,726. " Difference 31. 



31 1 



The extension per kilo, was 750. Hence the hysteresis is - — 7F7x = ;,a,v of ^^^ 



' Phil. Mag. August 1889. 



