16 
ME. J. REGINALD ASHIVOETH: 
or loss being greater the less the intensity ; a reversal of the direction of the 
magnetisation at a low intensity hy the operation of changes of temperature ; and 
also a change in the sign of a. But the magnitude of the temperature coefficient 
and of the irreversible change is decidedly larger in this table than in the former one. 
Hence, for the production of a magnet of constant intensity constructed of drawn 
steel wire, it would appear to be advantageous to magnetise step by step, heating and 
cooling at each step without intermediate demagnetisations up to maximum intensity, 
and then to remove, a small fraction of the magnetism by a reversed force ; ^ is then 
at its least value.On the other hand, this gives a larger value for a than if no 
reversed force had been applied. 
14. In order to determine how far these results are due to drawing, it is necessary to 
have a comparison with similar experiments ^^erformed on an iron wire, and this was 
subsequently done. The iron wire was that which is supplied in commerce as such, 
but probably it borders on very mild steel; it was carefidly annealed, and then 
submitted to a cycle of magnetisation, at first cold and afterwards hot. The curves of 
magnetisation intersect in this material when the intensity, for rising forces, is about 
800 units ; at higher intensities the susceptibility is less hot than cold, and the hot 
residual lies below the cold residual,! accordingly subsequent heating is found to 
diminish and cooling to increase the residual magnetism. The coercive force is about 
4-0 units cold and a little less when hot. 
In the next place the wire, after demagnetisation by reversal, was submitted to a 
very small force and the force withdrawn, then a series of heatings and coolings vms 
applied, and the })ermanent loss and coefficient were calculated in the usual way. 
Again, after demagnetisation, a stronger force was applied and withdrawn and a 
series of lieatings and coolings executed. Repeating these operations a step higher 
each time and demagnetising between each step, we get the result exhibited in 
Table VI. 
* J ide Hookham, ‘Journal Inst. Electr. Engineers,’ vol. 18, p. 688. 
t Baur, ‘Wied Ann.,’ vol. 11, 1880; EwiNG, ‘Phil. Trans.,’ vol. 176, Part IP, p. 637. 
