Prof. Challis on a Theory of Magnetic Force. 67 
probably the internal gradation of density chiefly results, as in 
a body subject to the influence of electricity by induction. Ifa 
closed circuit of any metal be heated at any point, currents, 
according to this theory, will be generated in consequence of the 
different degrees of expansion of the metal at different points, 
and, flowing in opposite directions, will neutralize each other if 
they be equal. But if by any mechanical means they be made 
unequal, as by contortions of the metallic circuit, a galvanic 
current should result, as is found to be the fact by experiment. 
When the circuit is formed by two metals soldered together at 
their extremities, and heat is applied at one of the positions of 
junction, inequality of the opposite currents might be expected 
to arise from difference of the capacity of the two metals for 
generating currents, owing to difference of their atomic constitu- 
tion; and accordingly it is found that under these circumstances 
a galvanic current is produced. I proceed now to the theory of 
magnetism. 
3. In the preceding theories the generation of secondary 
etherial currents has been ascribed to a disturbance of the 
atomic condition of bodies by external agency,—ain electricity, by 
friction ; in galvanism, by the mutual molecular actions of dis- 
similar substances in contact; and in thermo-electricity, by 
heat. In the theory of magnetism it must be assumed that 
there are substances in which a gradation of interior density 
exists independently and permanently,—that iron, for instance, 
is found in this state in nature; that the same state may be 
induced in steel by mechanical means, with different degrees of 
permanence ; and that it may be momentarily induced in soft 
iron. Also it must be supposed that the direction of the grada- 
tion of density depends on the form of the magnetized body. 
These suppositions rest immediately on facts of experience, the 
explanations of which, since they relate to qualities of the bodies, 
and not to the agency by which magnetic phenomena are pro- 
duced, are not now under consideration. 
4. Let us take the case of a bar of magnetized steel of the 
form of a long rectangular parallelopiped, and let it be assumed 
that there exists permanently a uniform decrease of its atomic 
density from the end A to the end B. . By the argument in art. 
1, on the supposition of a steady and uniform primary current, 
there will be impressed on the ether within the bar a uniform 
accelerative force, acting throughout its length from B towards A. 
To the accelerative force in that half of the bar which lies towards 
B, may be ascribed the effect of overcoming the inertia of the 
ether in motion within and without the bar on the side of B, 
and to the accelerative force in the other half, the effect of over- 
coming the resistance opposed to the flow of the current by the 
