218 MB. S. BIDWELL ON CHANGES PRODUCED BY MAGNETISATION IN 



or 74 ten-millionths of the length ; i.e., the contraction of the rod when longitudinally 

 compressed by the weight which it could support in a field of 600 is only 8 ten- 

 millionths greater than when it is compressed by the weight which it carried in a 

 field of 250. 



Now it appears from the tables and curves contained in the present paper that the 

 magnetic contraction of an iron rod (or ring) is greater in a field of 600 than in one of 

 250 by as much as 40 ten-millionths. It follows, therefore, that what I have called 

 the mechanical contraction only accounts for a fifth part of the total observed effect, 

 or even less if it should be the fact, as is not unlikely, that the elongating influence 

 upon which the contracting influence is superposed, does not altogether cease to 

 increase after the net elongation has reached its maximum. 



The main cause of the retraction, therefore, still remains a mystery. 



It is proposed to continue these experiments, which must certainly tend to throw 

 some light upon the molecular effects of magnetism. 



APPENDIX. 



The value to be attached to the results given in this paper depends greatly upon 

 evidence of the precision with which measurements can be made of lengths commonly 

 regarded as infinitesimal. In order that an estimate may be formed of the degree of 

 accuracy arrived at, I have thought it desirable to give in the form of an appendix 

 certain details of the experiments, which if contained in the paper itself would have 

 been tedious and cumbersome. 



As elsewhere stated, the optical arrangements are so perfect, and the edge of the 

 focussed image of the wire is so sharp that, under favourable conditions, it is possible 

 to read to a quarter of a scale division (1 scale division = 0'64 mm. = 4 J Q inch) even 

 when the reflected beam of light is 732 cm. (24 feet) long. But during an experi- 

 ment it unfortunately happens that, in consequence of the heating effect of the 

 currents, which no amount of care can completely obviate, the image is in constant 

 movement up or down the scale, a change of one degree in the temperature of the 

 magnetised ring or rod, together with its brass connections, causing a rise or fall of 

 more than 150 scale divisions, This not only renders it difficult to make an accurate 

 reading, but it is the source of another and more serious inconvenience, namely, that 

 there can be no fixed zero. Every observation of a deflection, therefore, involves two 

 scale readings, in each of which there is a chance of error. The following was the 

 method adopted. At the moment when the upper edge of the wire appeared in the 

 course of its wanderings to coincide exactly with a certain previously determined 

 division which it had been seen to be approaching, and the number of which was 

 recorded as the temporary zero, the contact key was depressed, and the division 



