508 PROFESSOR KNOTT ON SOME RELATIONS BETWEEN 



the effects of tension on the elongations of these metals, as recently established by Mr 

 Bidwell. If these coincidences do not demonstrate that the elongation and twisting 

 phenomena are related as cause and effect, they at all events demonstrate a very close 

 connection between them — a connection which should be an essential feature in any 

 theory that would satisfactorily co-ordinate any one group of these magneto-strain 

 phenomena. 



Part III. 



1 5. I now pass to the discussion of certain results which at first sight seem to belong 

 to a reciprocal set of phenomena. The experiments form a detailed investigation into 

 certain magnetic changes produced by twist. Some four j^ears ago I planned a series of 

 researches bearing on this subject, with the aim of filling up the many gaps in our 

 knowledge of the relations of magnetism and twist. Professor Wiedemann's paper of 

 1886* especially suggested many lines of inquiry. Accordingly, in the spring of 1887, 1 

 set Mr Imagawa, one of our students of physics, to work at one of the simplest of these 

 problems. Other scientific work prevented me working up these results or in any way 

 following them out till 1889. I then published, in the Journal of the College of Science 

 (Imperial University, Japan), vol. iii., 1889, a short account of the most striking 

 peculiarity observed by Mr Imagawa, under the title " On Magnetic Lagging and 

 Priming in Twisted Iron and Nickel Wires." This peculiarity was that, whereas for 

 small twistings the changes in the longitudinal magnetic moment take place during 

 untwisting from either limit more slowly than during the immediately preceding 

 twisting, nevertheless for sufficiently large twistings the reverse holds true. If we speak 

 of the former effect as a lagging of the magnetic change behind the strain that produces 

 it, the result may be simply expressed thus : For small cyclic twistings of an iron or 

 nickel wire magnetised either circularly or longitudinally, there is magnetic lagging in the 

 changes of the longitudinal magnetisation ; for large cyclic twistings there is magnetic 

 priming. Or otherwise thus : If we draw the graph representing the march of magnetic 

 change with twist, we shall find that below a certain range of twisting the closed graph 

 is gone round (say) counter-clockwise, while above the critical range it is gone round 

 clockwise. The area, in fact, changes sign ; and there is a particular range of twisting for 

 which the area vanishes. In the earlier experiments, as conducted by Mr Imagawa, the 

 wire was circularly magnetised by a current passing along it ; and nearly all the experi- 

 ments were made with nickel wire. A few experiments were tried with the same wire 

 longitudinally magnetised, with the view of finding if the same curious reversal of 

 magnetic lag was obtained with it. The effect was also found to exist for longitudinally 

 magnetised nickel wire. By a strange oversight, Mr Imagawa did not observe at the 

 time that circularly magnetised iron wire showed the same peculiarity as nickel, conse- 

 quently he did not search for an analogous effect in longitudinally magnetised iron. It 

 was only when I came to collate the results of all the experiments that the reversal effect 



* See Wiedemann's Annalen, vol. xxvii. p. 377, translated in the Philosophical Magazine (1886). 



