48 Dr. E. Taylor Jones. 



In order to make allowance for the slow time-change in the pro- 

 perties of the wire, the determinations of magnetisation and effect 

 of change of tension on magnetisation were then repeated, the 

 nickel wire being rewound with the same number of turns of insu- 

 lated copper wire. 



Finally, the contraction of the wire was once more determined at 

 the temperature 10 C. 



Values of the expression (1) were calculated from both sets of 

 magnetisation observations, and the mean of the two sets compared 

 with the actual contraction observed between them. 



The results are shown in Tables II, III, IY, Y, VI, in which values 

 of all the quantities (obtained from the corresponding curves) are 

 given for the same set of field strengths. 



In Tables II, IV, the first column contains values of the field 

 strength H, the second the magnetisation I at load 4*9 kg., the third 

 the change of magnetisation 1 accompanying an increase of load 

 of 7 kg., the fourth the corresponding values of expression (1), i.e., 

 the calculated value of 5Z/Z. The numbers in the second, third, 

 and fourth columns were determined in November, 1897. The fifth, 

 sixth, and seventh columns contain values of I, 51, and 5Z/Z, deter- 

 mined in January, 1898. 



Tables III and V contain values of the field, the mean values of I 

 and SI 1 1 calculated from Tables II and IV, the actual change of 

 length a in millionths, observed about the end of December, 1897, 

 and the difference (txBl/l) between the observed and mean calculated 

 elongations. 



Finally, in Table VI are given the values of I and 3Z/Z, measured in 

 January ; the mean a of the actual changes of length observed in 

 December and in January, the latter being determined after the 

 second set of magnetisation measurements ; and the difference, 

 (a 5Z/Z), representing the corrected elongation at the time when the 

 second set of measurements of I and el/SP was made. 



Tables II and IV show the nature of the slow time-change in the 

 magnetic behaviour of the nickel wire, the magnetisation at any 

 field and load being less in January than in November. 



The effect of this change on l/<)P is very marked at low fields, 

 though slight at higher fields ; 5I/&P appears to diminish rapidly at 

 low fields, as time goes on. This is remarkable, because the magnetic 

 contraction at low fields seemed to change but slowly with time, and 

 more rapidly at higher fields. 



The change of length observed in January was nearly the same as 

 in December (about four weeks earlier). At medium fields it was 

 rather greater, but this may be due to a slight annealing caused by 

 the repeated warming and cooling during the determinations of the 

 magnetisation earlier in January. 



