DIMENSIONS OF EINGS AND BODS OF IRON AND OTHER METALS. 225 



There is only one instance in which a point appears to deviate from the curve by a 

 distance equal to as much as half the height of one of the small square spaces, namely, 

 when the magnetic force is about 425 units. Now the height of one of these spaces 

 corresponds to five ten-millionths of the length of the rod ; or, since the rod is four 

 inches (10 cm.) in length, to an absolute length of two-millionths of an inch. Half 

 the height of a space, therefore, represents a length of one-millionth of an inch,* and 

 this is the widest deviation which occurs in the whole of the three series of obser- 

 vations. It is clear that in the great majority of cases the deviation from the mean 

 curve is very much less, not often being so great as one-fifth of a space height, or one 

 two-and-half-millionth part of an inch. 



The close agreement of columns E and K also shows how perfectly the demag- 

 netising apparatus fulfilled its object. It appears from columns H and M that the 

 elongating effect of the second current after a demagnetisation was generally different 

 from that of the first, the difference being of course accounted for by permanent 

 magnet ism .t 



The figures giving the remarkable cobalt curve are just as regular as those which 

 relate to the iron. In the results of the 14 pairs of observations in columns E and K 

 there is not a single discrepancy of more than 1 scale division. Observations were 

 made both with ascending and with descending currents, all of which are plotted in 

 fig. 4. It was not until after the experiment was nearly completed that my attention 

 was called to the fact of the retraction reaching a maximum with currents of medium 

 strength, and falling off when stronger currents were applied. Thinking that this 

 might possibly be an effect of heat upon susceptibility, I put on all the 30 cells at 

 once, the metal having been first allowed to become quite cool. The deflection was, 

 however, just the same as that which the 30 cells had produced before. 



There can be no doubt that the curve in fig. 4 correctly represents the effect of 

 magnetisation upon the length of my rod of cobalt. Whether all specimens of the 

 metal would behave in exactly the same manner is not so certain. Professor BARRKTT 

 found { that cobalt was elongated when magnetised. From the analogy of iron it 

 seems not unlikely that a softer rod of cobalt than mine would in the earlier stages of 

 magnetisation undergo some elongation ; but it is more than probable that with 

 sufficiently high magnetic forces such elongation would be ultimately converted into 

 retraction of the kind indicated by the curve in fig. 4. 



In the experiments with ring No. the observations were only made once, so that 

 their individual accuracy cannot be tested by comparison. 



* About one-twentieth part of the wave-length of green light. 



t [It appears from a comparison of the figures in columns O and P of Tables III. and IV. that the 

 permanent elongation which remains when the magnetising force has been withdrawn increases until 

 the magnetising force which produced it reaches about 130 units, after which it is nearly constant. It 

 has been pointed out that this fact tallies well with the known constancy of residual magnetism when 

 the fields are strong. May 14th, 1888.] 



$ ' Nature,' vol. 26, 1882, p. 585. 

 MIX i i I. \\XVIII. A. 2 G 



