STEEL, NICKEL, AND COBALT TUBES IN THE MAGNETIC FIELD. 473 



as the bore of the tube is widened is a new experimental fact ; but it is one which 

 might almost have been predicted from our previous knowledge of the magnetic 

 properties of iron (see above, p. 465). 



In the case of the narrow bored Nickel Tubes 1 and I. the radial elongation is 

 greater numerically than the longitudinal dilatation, at least in the higher fields. 

 This feature is slightly shown in Tube II., and is associated with a tangentia] elongation 

 of the same sign as the longitudinal elongation. In these tubes, accordingly, the 

 internal radius shortens, and the external radius lengthens, so that we are led to the 

 conception of a cylindrical surface within the metal which experiences no displacement 

 outwards or inwards. 



A similar feature is characteristic of the narrower bored iron tubes in high fields, 

 as a glance down the columns X and v in Table VII. will show. 



Now, if we compare the measured quantity A + 2// for a nickel tube of very narrow 

 bore with the quantity S obtained for the bar from which the tube was formed, we 

 find that the former quantity is numerically much the greater. This shows that the 

 removal of a comparatively small amount of material from the core of a nickel bar 

 alters, in a remarkable degree, the character of the strain which accompanies powerful 

 magnetization. In the bar, therefore, the molecules must be subject to considerable 

 mechanical constraint. The mere formation of a second free surface of small extent 

 in the heart of the metal produces, so to speak, a relief, whose effects extend appreciably 

 to the outer surface. There is, indeed, more resemblance between the narrowest and the 

 widest bored tubes as regards their condition of strain in a magnetic field than between 

 the narrowest bored tube and the original bar. 



On the other hand, if we compare the quantity A + 2// for an iron tube of very 

 narrow bore with the quantity S, we find that the former is the smaller, or, at least, is 

 never the larger. A comparison of the quantities &v' and 8V for A I. and B I. in 

 Table VI. or in Plate II. indicates this clearly enough. That is to say, the removal of 

 a small amount of material in the heart of the bar alters, to a comparatively slight 

 extent, the displacement of the outer surface under magnetization. The alteration is, 

 nevertheless, quite unmistakable, the formation of a second free surface in the heart 

 of the metal distinctly modifying its behaviour. The tendency, in both iron and 

 nickel, is for 8V to be algebraically less that SV ; but numerically the difference is 

 very much less in iron than in nickel. 



Another point of interest is the form of ellipsoid into which a small spherical 

 element at either surface is changed ; and here again the comparative simplicity of the 

 results for nickel is noteworthy. In every case the shortest axis of the magnetic strain 

 ellipsoid is in the direction of magnetization, and in the great majority of cases the 

 longest is in the direction of the radius. Occasionally in the tubes of narrowest bore 

 the third axis, that which corresponds to m, the tangential elongation, is one of contrac- 

 tion, but generally it is, like the radial axis, one of elongation. Thus, in all cases of the 

 bored tubes, a circular element in any plane perpendicular to the axis of the cjdinder 



VOL. XXXIX. PART II. (NO. 15). 4 B 



