SHUDDEMAGEN. — DEMAGNETIZING FACTORS FOR IRON RODS. 193 



netizing solenoid, the L may be enormously large. There are two ways 

 of realizing the condition of the smallness of the time-constant as com- 

 pared with the quarter-period: (1) We may use a storage battery of 

 high E.M.F. in the primary circuit, which will necessitate large it's in 

 the circuit in order to give magnetizing fields of the desired intensity ; 

 (2) It is quite possible to increase the moment of inertia of the needle- 

 suspension so as to give a complete period of several minutes. Several 

 of the experimental series obtained in this investigation by means of 

 the reversal and step methods illustrate very forcibly how these two 

 different methods may lead to various determinations of the suscepti- 

 bility. Finally, the magnetometric methods are often very useful, 

 especially in accurate determinations of magnetic moment of short iron 

 magnets. With none of these magnetometric methods can we measure 

 the /at any particular part of the iron bar, but get instead a mean 

 value of /(moment/volume of bar) for the whole rod. Plotting /vs. 

 H' curves for various lengths of soft iron cylinders, we can find mean 

 demagnetizing factors X, by means of which a " normal " curve can be 

 constructed. But it will be seen, after a little reflection, that the curve 

 Mean / vs. Mean H which we get here is not necessarily the same, or 

 even approximately the same, as the " normal " curve of / vs. //, which 

 gives corresponding values of / and H in the middle of the bar imme- 

 diately surrounded by the secondary coil, and which may be regarded 

 as an extremely close approximation to the /and H at a single point 

 in the iron. It is this fact which accounts for the wide difference which 

 has been found between the 2V r as determined ballistically and the iYas 

 determined magnetometrically. It is hardly likely that the process of 

 back-shearing a magnetometric magnetization curve will yield a curve 

 from which anything like the true susceptibilty can be found. 



Returning now to our iron ellipsoids of revolution, we see that if we 

 know the ratio of the length to the diameter of one of them, we can 

 calculate exactly what the demagnetizing factor X will be. Ewing and 

 Du Bois, in their texts on magnetism, give tables of values of X (see 

 page 204) for various ratios a/b. It follows from a paper by Lord 

 Rayleigh, 11 that if we magnetize any iron ellipsoid of revolution 

 having a known ratio a/b, from zero magnetism to full saturation, 

 measuring the / ballistically by means of a small secondary coil around 

 the middle part of .the rod, and plot out the curve / vs. H' , we can 

 " back-shear " this curve parallel to the ZT-axis by the amount 

 Hi = A/T = XI, and thus construct the " normal magnetization " 

 curve, for which H = W, and from which alone the true susceptibility 

 can be found for everv / 



11 Phil. Mag., 22, 175-183 (1886). 



VOL. XLIII. — 13 



