116 Mr. S. Bidwell. On the Changes produced by 



The relative lengths of the pieces of brass were so adjusted that when 

 the compound rod was fixed in position, the centre of the magnetic 

 portion of it coincided as nearly as possible with the centre of the 

 coil. This arrangement was, however, found not to be entirely satis- 

 factory. It was difficult to secure the required coincidence with 

 perfect accuracy, and it was necessarily somewhat disturbed during 

 the adjustment of the image of the wire upon the scale ; while, even 

 supposing that the geometrical coincidence was perfect, it might well 

 happen that owing to inequalities in the magnetic properties or 

 physical condition of the rod, the source of error might still exist. 

 A simple experiment showed conclusively the immense importance of 

 guarding against any trace of this interaction between the rod and 

 the coil. A compound rod of iron and brass was prepared, such that 

 when it was placed with one end uppermost the centre of the iron 

 was somewhat below that of the coil, and when the other end was 

 uppermost the centre of the iron was about 5 mm. above that 

 of the coil. A current was passed through the coil when the 

 iron was in the first position and a certain elongation was 

 indicated. The position of the rod was then reversed and the same 

 current passed. It was expected that the apparent elongation would 

 be diminished, but in point of fact an actual retraction equivalent to 

 two or three scale divisions was indicated. The sucking action of the 

 coil caused the lower end of the rod to press upon the base with 

 increased force ; the base yielded a few hundred thousandths of a 

 millimetre, and this was sufficient, in spite of the real elongation of 

 the rod, to cause the image of the wire upon the scale to move in the 

 direction of retraction. 



It appeared that the only method of avoiding with certainty the 

 misleading effects of this attraction between the coil and the rod 

 would be to attach the two together. The pressure upon the base 

 would then depend simply upon the joint weights of the coil and the 

 rod, and would not be varied by any interaction between them. The 

 coil was therefore detached from the table, and its ends were fitted 

 with corks through which the experimental rod was passed, care 

 being taken that it fitted tightly at both ends. The arrangement was 

 then exactly as shown in fig. 1, the coil being supported solely by 

 the rod ; and it was so used in all the experiments described in this 

 paper. 



Before giving an account of the new effects which I have obtained, 

 it may be well to state how far the maximum elongations and retrac- 

 tions of iron and nickel bars, as indicated by my apparatus, accord 

 with those published by previous experimenters. This is done in the 

 subjoined table. 



