SHUDDEMAGEN. — DEMAGNETIZING FACTORS FOR IRON RODS. 187 



in 1901 by Carl Benedicks.^ This investigator, while working on the 

 subject of pole-distances in cylindrical rods, interested himself in a 

 few careful experiments on the demagnetizing factors. He gets for a 

 hard steel rod of diameter 0.8 cm. and a length equal to 25 diameters, 

 hysteresis curves by means of both the magnetometric and the ballistic 

 methods. Then by turning it down on the lathe, he transforms the 

 same specimen of iron into an ellipsoid of revolution of length equal to 

 30 diameters, and gets a hysteresis curve magnetometrically. This 

 last curve is, by means of the known ellipsoid iV for m = 30, back- 

 sheared into the "normal" curve, which, according to Benedicks, can 

 then be used to determine the N for any point on either the ballistic 

 or the magnetometric curve for the cylinder. The result is that the 

 magnetometric N behaves qualitatively exactly like that of Mann, but 

 the ballistic iV^, after likewise remaining practically constant up to 

 / = 800, decreases rapidly as / is further increased. 



The present paper is an attempt to contribute to the subject a 

 discussion of the demagnetizing factor for cylinders as determined 

 ballistically. It will appear later that the curve on the B vs. ff' plane 

 (or the /vs. H' plane) which determines the back-shearing from a 

 magnetization curve of a finite cylinder to the limiting normal curve, 

 is quite different from the straight line which obtains in the case of 

 the ellipsoid of revolution. It has, in fact, two opposite curvatures : 

 one near the origin, and the other soon after the maximum value of 

 the susceptibility has been passed. The first curvature is not very 

 marked, and it turns out, as has been found before for the magneto- 

 metric N, that up to values oi B = 10,000 (or /= 800) the ballistic N 

 is not far from constant. The upper part of the curve, however, has a 

 violent turn toward the ^-axis (or /-axis) just as has been observed 

 by Benedicks for his short steel cylinder. Theoretical reasons can be 

 given to account in a general qualitative way for these experimental 

 results. 



Hitherto it has been the common custom, for lack of experimental 

 evidence on the subject, to regard the N for iron cylinders, leaving 

 out of consideration the variation of this coefficient with the /, as de- 

 pending only on the ratio m = L/D, and not on the absolute dimen- 

 sions of the rod. As practically all the previous results have been 

 obtained from experiments on iron cylinders having a diameter of less 

 than 1 mm., that is, mere iron wires, the question has naturally not 

 received any attention. In the present work the writer had at his 



6 Bih. Svenska Vet.-Akad. Handlingar, 27, (1), No. 4, 14 pp. (1902) ; Wied. Ann., 

 6, 726-761 (1901). 



