656 Messrs. W. Morris Jones and J. E. Malam on the 



is to be expected in view of the fact that Heaps' wire was 

 bent back on itself several times, thus introducing a slight 

 transverse component. It seems probable that, as Heaps 

 suggested, the slight fall obtained is mainly due to non- 

 uniformity of field. The correction for polarity is small. 

 The length of the specimen was roughly 175 diameters, 

 giving a self-demagnetizing factor of less than *02. Taking 

 the saturation value of intensity of magnetization for nickel 

 (hard-drawn) as 400, the true mean value of the external 

 field will be less than that of the undisturbed field by roughly 

 8 C.G.s. units for fields above say 2000 c.G.s. (below which 

 value saturation is obtained) with the specimen in the longi- 

 tudinal position. This small constant correction would not 

 account for the steady fall in the curve. Tt is also evident 

 from the above that any small transverse disturbance intro- 

 duced by the polarity of the metal would, with the specimen 

 in the longitudinal position, be of such magnitude as to be 

 negligible. 



The specimen was soldered to the leads under a tension of 

 about 6 kilogrammes per square cm. Owing to this initial 

 stress the change of resistance for high fields would be 

 slightly diminished. The magnitude of the effect of the 

 changes in length and volume of the specimen when magne- 

 tized is, however, so small as to be negligible. 



The curves obtained for intermediate orientations are given 

 in fig. 5. The necessarily varying conditions of temperature 

 under which the curves were taken will account for anv 

 slight irregularities in spacing. It should be noted that 



values of -p- are in all cases plotted against the external 



field. The effect of polarity on the average value of the 

 induction inside the specimen will of course increase as the 

 specimen is rotated from the longitudinal to the transverse 

 position, and this must be taken into account in making 

 comparisons between the curves. It will be seen from the 

 upper curves of the series that an error of setting of 10°, 

 such as might arise when using wires 1 mm. long, would be 

 sufficient to account for the greatest drop obtained by Owen. 

 It seems evident that the phenomena observed are closelv 

 connected with the changes of molecular configuration in 

 the metal produced by the magnetic field. If in the case of 

 the ferro-magnetic metals we take the quantity intensity of 

 magnetization as some measure of these changes, we may 

 note that the changes of resistance attain a maximum at 

 about the strength of field at which the saturation value of 

 the intensity is reached. Following the analogy of the 



