286 Wisconsin Academy of Sciences , Arts and Letters. 
is different from the result obtained by Poloni, who found the propor¬ 
tional change sensibly constant throughout the magnet. It could not, 
be expected that the small difference noticed here could be detected by 
his method which consisted in measuring two large quantities and sub¬ 
tracting them in order to obtain a small difference. In the method em¬ 
ployed in this research the quantities measured differed but little in 
size from the quantities desired, and much greater accuracy is easily 
obtained. 
I would suggest the following explanation: Prof. Ewing has recently 
made an important addition to Weber’s theory of magnetism. He 
says the forces which hold the little molecular magnets in position are 
largely the mutual attractions and repulsions of these molecular magnets 
themselves. In applying Ewing’s theory to the case in hand, let us con¬ 
sider a row of magnetic molecules ABC, etc. 
ABC HIJKL 
J is held in position by the action of H I, etc., on the one side and 
K L. etc., on the other, while A has only B C, etc., to act upon 
it. It is evident that the force holding J in position is greater 
than that acting on A. Suppose the bar of which this line of 
molecules is a part is heated. If in this process the energy of 
vibration of A and J receive equal increments, it is evident that the 
increase of amplitude of A will be greater than that of B. Now the 
magnetic moment contributed by each molecule is the moment of the 
molecule resolved along the direction of magnetization of the bar. The 
moment contributed by A would suffer a larger proportional loss than 
that contributed by J, and so the loss would be greatest at the ends. 
There are other facts pointing in this direction, e. g., when a magnet 
is heated before it has reached the permanent state, Kupffer found, as 
already stated, that the proportional permanent loss was greatest at the 
ends. In some rough tests I have made on this point, heating the bar 
almost to redness, I have found the proportional loss at the ends nearly 
twice as great as at the center of the bar. This would naturally follow 
from the supposition made, for the force holding the end molecules in 
position being less, they are more easily set in such violent vibration as 
to swing out of one position of equilibrium into another. 
I am tempted to add a single remark on another part of the investiga¬ 
tion. In every case tried so far, the area of the cycle of magnetization 
(the largest magnetizing force being the same for all temperatures) is 
always smaller for the higher temperature. I have only experimented 
