produced in Soft Iron by Magnetism. 505 



to vary the ratio between the diameter and the length, not by 

 diminishing the diameter but by changing the length. This 

 I did, as has already been mentioned, by taking bar III. and, 

 after testing its expansion, cutting off a portion of it. The 

 result is shown in Tables III., IV., and Y. Lest, perchance, 

 bar III. of less diameter be compared with bar VI., which 

 although larger gives less expansion, I took the precaution to 

 turn bar VI. down again until it was about the same size 

 as JSo. III. As a result, the expansion was then less 

 than 5 fringes, while bar III. gave 7*4 fringes, and this 

 although the iron in the two bars was very nearly alike in 

 magnetism. 



The following table will show clearly that the expansion 

 varies directly as some function of the ratio between the 

 diameter and length, possibly the square root. Joule's bars 

 were rectangular, and I have computed a mean diameter. His 

 tests were confined to four bars, and I have taken the mean 

 value of the temporary elongations in the four bars. 



I have dealt entirely with the temporary elongations, or the 

 elongation produced by the second or following contacts, 

 because both Mr. BidwelPs observations and my own deal 

 with these maxima. 



Prof. A. M. Mayer tested six different bars, and in this case 

 I have taken the mean value on the second contact. 



Shelford Bidwell tested three bars of different diameters. 

 I have used the mean of the two observations which he 

 himself quotes in a comparison between his results and those 

 formerly obtained. I have taken the mean of the diameters, 

 and likewise the mean expansion. 



Mr. Barrett's observation in ' Nature ' (1882) I do not 

 give, as he has furnished so few details of his experiment that 

 one can form no judgment in regard to value. 



Although I have quoted M. Alfonse Berget's single experi- 

 ment, because the ratio in this case w r as exceedingly large, 

 yet, as the length of his bar was so short, and as he tells us 

 nothing as to whether he has given us the elongation from 

 the first or second contact, one must feel compelled to give it 

 little weight except for showing the main point in the 

 discussion. 



It must also be borne in mind that the iron used by different 

 observers could of itself produce variations of as much as 25 per 

 cent., due to a difference of permeability. 



