70 HENRY A. KOWLAND 



In Table VI we have results for cobalt at a high temperature, and 

 see how greatly the permeability is increased by rise of temperature, 

 this being for the vertex of the curve about 70 per cent. But on plot- 

 ting the curve I was much surprised to find an entire departure from 

 that regularity which I had before found in all curves taken from iron 

 and nickel when the metal was homogeneous. At present I am not able 

 to account for this, and especially for the fact that one of the measure- 

 ments of 33 is higher than that which we have taken for the maximum 

 of magnetization, at, however, a lower temperature. The curve is 

 exactly of the same nature as that which I have before found for a 

 piece of nickel which had been rendered unhomogeneous by heating 

 red-hot, and thus burning the outside. The smaller cobalt ring gives 

 a curve of the same general shape as this, but has the top more rounded. 

 I will not attempt without fresh experiments to explain these facts, but 

 will simply offer the following explanations, some one of which may be 

 true. First, it may be due to want of homogeneity in the ring; but it 

 seems as if this should have affected the curve of Table IV more. 

 Secondly, it may be at least partly due to the rise in temperature of the 

 ring at high magnetizing-powers ; and indeed we know that this must 

 be greater in paraffin than in alcohol for several reasons : there is about 

 twice as much heat generated in copper wire at 230 C. as at with 

 the same current; and this heat will not be conducted off so fast in 

 paraffin as in alcohol, on account of its circulating with less freedom; 

 it probably has less specific heat also. Thirdly, it may be due to some 

 property of cobalt, by which its permeability and maximum of magneti- 

 zation are increased by heat and the curve changed. 



The experiments made with the small ring confirm those made with 

 the large one as far as they go; but as it was so small, they do not 

 possess the weight due to those with the larger one. But, curious as 

 it may seem, although they were turned from the same button side by 

 side, yet the permeability of the larger is about 45 per cent greater than 

 that of the smaller. I have satisfied myself that this is due to no error 

 in experiment, but illustrates what extremely small changes will affect 

 the permeability of any metal. 



We have now completed the discussion of the results as far as they 

 refer to the magnetic permeability, leaving the discussion of the tem- 

 porary and permanent or residual magnetism to the future, although 

 these latter, when discussed, will throw great light upon the nature 

 of the coercive force in steel and other metals. The whole subject 

 seems to be a most fruitful one, and I can hardly understand why it has 



