C. A. Perkins — Magnetic Permeability of Nickel. 221 



method used by him was suggested and developed by Kirchhoff* 

 in 1870, and consisted in magnetizing a ring of metal by a coil 

 about its entire length and then measuring the current pro- 

 duced in a secondary coil when the primary current was re- 

 versed. This method possesses decided advantages over all 

 previously used in important particulars and is capable of giv- 

 ing results of great regularity. He arrived at results similar to 

 those of Icilius, but found certain irregularities for which he 

 was not„able to account. 



The same year Prof. Rowlandf had made experiments using 

 the same general method, but introducing various modifications. 

 An earth inductor was placed in the secondary circuit by which 

 the induced currents could be directly compared with those 

 produced by a coil rotated under the influence of the earth's 

 action and thus reduced immediately to absolute measure. In 

 the same circuit was included a small coil of wire, sliding upon 

 a magnet. ~Bv the motion of this coil the galvanometer needle is 

 under perfect control and can be instantly stopped at the middle 

 of the scale. He wound the primary coil directly upon the 

 ring, while in Stoletow's experiments wooden rings were placed 

 inside the coil. 



Rowland also suggested plotting the results with respect to 

 the magnetization and the magnetic permeability as coordinates, 

 thus obtaining a curve which gave a definite value for the maxi- 

 mum magnetization possible. 



The results reached by him in the experiments on iron were 

 as follows : 



For weak forces all the magnetism is temporary. 



The value of the permeability rises very rapidly at first, 

 reaching a maximum when the magnetization has the value 

 33=6000 and then diminishes, approaching as 35=17500. 

 This, therefore, is about the maximum magnetization of ordi- 

 nary iron, giving a sustaining weight of 354 lbs. per square 

 inch of least cross section. 



(This is in C. Gr. S. units. The original results were on the 

 M. Gr. S. system, and as the dimensions of 33 are [IriM^T -1 ], 

 the original figures are to be divided by ten. fjt is of zero 

 dimensions.) 



The curve in form resembles a slightly inclined parabola, but 

 is more accurately represented by the equation 



ja = 2475 sin 



/ 33 + l-094^ + 500 \ 

 \ ioo " )' 



In some experiments the curve finally became curvex toward 

 the axis, which was ascribed to unhomogeneity in the ring but 



. Ergz. Bd., vol. v. f Phil. Mag., August, 1873. 



