of various Iron Bodies. 201 



infinitely long cylinder and is uniformly magnetized longitudi- 

 nally. I have named this coefficient (in that essay denoted by 

 k) the. magnetizing -function of the given irou, since it depends 

 on the quantity of the magnetizing force. An analysis, namely, 

 of the experiments of Quintus Icilius (with ellipsoids) and of 

 my own (with a ring) showed that the function k at first in- 

 creases rapidly as the decomposing force rises, and then again 

 diminishes. This behavour seems to take place with all sorts of 

 iron ; yet the absolute numerical values of k, with the same val ue 

 of the argument, are very different, according to the quality of 

 the iron. These results have been corroborated by a thoroug h 

 investigation by Mr. H. A. Rowland*. He shows that the 

 course of the function k is precisely similar for steel and nic kel 

 as well as iron, and can be represented by the same empiric 

 formula, but that the constants of the formula, even for two 

 varieties of one and the same metal, come out very different f- 



Professor Eiecke, in his " Contributions to the Knowledge of 

 the Magnetization of Soft Iron - " (Pogg. Ann. vol. cxlix. p. 433), 

 proposes, instead of the magnetizing-function of the infinite 

 cylinder, to consider another function p, which has the same 

 signification in reference to the sphere. 



The two quantities, referred to the same decomposing force, 



* " On Magnetic Permeability, and the Maximum of Magnetism of Iron, 

 Steel, and Nickel," Phil. Mag. August 1873, p. 140. The term "mag- 

 netic permeability " is used, alter Sir W. Thomson, to denote the quantity 

 {j. = \-\-47rk, which, as h is here generally much greater than unity, varies 

 nearly proportionally with k. 



f Professor Wiedemann, when discussing my work (in Galvanismus, 

 2nd ed. vol. ii. p. 518), regards the function which is calculated from ex- 

 periments with the ring as another magnetization-function, not to be con- 

 founded with that obtained from experiments with "unclosed systems." 

 There does not seem to me sufficient reason for this distinction. Residual 

 magnetism, which is here in question, is present in bfirs also. If we con- 

 sider a very thin and long bar and a ring, both magnetized uniformly, the 

 difference between them in relation to the residual magnetism is hardly to 

 be reckoned considerable. The demagnetizing force proceeding from the 

 mass of its iron will in the ring be equal to nil; in the bar it is a small 



(O 



quantity, of the order of y 2 , where a> is the cross section, and I the length 



of the bar. (Maxwell, ' Treatise on Electricity and Magnetism/ vol. ii. 

 p. 67.) In both cases an external force is requisite in order to expel the 

 residual magnetism. If we always observe the reversal of the magnetism 

 of the iron, the calculation of k is only to this extent vitiated by the residual 

 magnetism, that a certain portion of the reversed decomposing force is ex- 

 pended in discharging it. But M. Wiedemann's own experiments with 

 bars, and those of Poggendorff with closed systems (Wiedemann, /. c.p.519), 

 show that that portion is only very little. 



A survey of the numbers obtained by Mr. Rowland, partly with bars, 

 partly with rings, establishes that the most essential cause of their differ- 

 ence is not the form, but the quality of the material. 



