Magnetisation in the Length of Metal Rods. 127 



Fig. 4, 



any elongation represented by the dotted portion would be too small 

 to be measurable. 



The results of this experiment of Joule's are thus shown to be 

 reconcilable with others, if we may make the assumption that in this 

 particular specimen of steel the elongations due to temporary and 

 permanent magnetisation followed different laws, and that while the 

 critical point of the former occurred at an unusually early stage, that 

 of the latter was not reached within the limits of the experiment. It 

 may be indeed that this is always the case, though under ordinary 

 circumstances the difference is too small to lead to the anomalous 

 result under discussion. Having confined my attention almost 

 entirely to the investigation of temporary effects, I have little experi- 

 mental evidence to bring forward bearing upon the point.* 



Mayer's results may be much more easily accounted for. The fact 

 that his steel rods were invariably shortened by magnetisation (after 

 the first magnetisation, the effect of which varied in different speci- 

 mens) clearly indicates that his magnetising force exceeded the 

 critical value, which was smaller for the steel bars than for the iron 

 which he had previously used. He apparently never magnetised his 



* Were it not for the proverbial accuracy of Joule's work, a simpler explanation 

 of the anomaly would have suggested itself. The lower of the two curves above 

 the horizontal axis represents the state of things while a current is passing, and the 

 fact that this curve does not coincide with the upper one migbt, perhaps, be 

 accounted for by the " solenoidal suction" which would occur if the rod were not 

 quite symmetrically placed with respect to the coil, or even if it were not perfectly 

 homogeneous throughout. Thus, the apparent elongations when the circuit was 

 broken would be really due to the cessation of the suction, while the elongations 

 indicated when the circuit was closed would be less than those which actually 

 occurred. Each of the vertical divisions in the diagram represents only one 

 thirteen-millionth part of the length of the rod : a very small variation in the 

 pressure of the end of the rod upon its support would, therefore, have a sensible 

 effect.— (February 23, 1886.) 



