1890.] upon Magnetic Changes of Length in Wires. 477 



Discussion of the Tables and Curves. 



Iron. — The curves in fig. 2 clearly show the effect of tension in 

 diminishing the maximum elongation and hastening the contraction. 

 With the two greatest loads used there was no preliminary elonga- 

 tion at all. The curve for an unstretched iron rod is generally 

 found to cut the axis at about 300 ; probably, therefore, if the wire 

 used in these experiments could have been tested with no load, its 

 curve would lie a little above that for the load of 351 kilos. 



Nickel. — The results for nickel are of great interest. In fields 

 below about 140 units increase of load is always accompanied by 

 decrease of magnetic contraction, the earlier portions of the six curves 

 in fig. 3 following one another in inverse order of the magnitude of 

 the several loads. But, although the initial slope of the curves 

 diminishes with increasing loads, the " turning points," where the 

 ratio of the contraction to the magnetic field is a maximum, occur 

 later with great than with small loads, so that in a field of 360 the 

 order of the relative values of the contractions for the three smallest 

 loads is actually reversed, the contraction being greatest with the 

 heaviest load and least with the lightest. It appears probable that, 

 if the experiment had been carried far enough, the curve for 

 1400 kilos, would have crossed one if not all of the three curves lying 

 below it. Whether the two remaining curves for 1890 and 2310 kilos, 

 would behave similarly is more doubtful. Possibly they would 

 have become parallel to the horizontal axis before the others were 

 reached. 



I think it may fairly be assumed that if the experiment could have 

 been made with the wire quite unstretched, we should have obtamed 

 a curve having a steeper initial slope than any of those in the 



2 n 2 



