62 Professor CuMMING on the Developement 
each taken separately. If it were, it would be affected by altering 
the internal structure. As bismuth is a metal easily cast, and which 
has a strong tendency to crystallization, I cast two bars of the 
same dimensions; one in a charcoal mould, which was cooled as 
slowly as possible; the other in iron placed in cold water, and 
therefore cooled almost instantaneously: but I could discover no 
difference either in the nature or quantities of the deviations they 
produced, when heated to the same temperature. Again, a glass 
tube eight inches long was filled with mercury, and copper 
wires being passed through corks at each end it was placed, 
the one half in hot sand at 170°, and the other in water at 60°*. 
The deviation was 9° positive, which at 150°, and at 115°, be- 
came 6° and 3°. It appears then, that, in the same metal, the 
magnetic effects are not varied, whether the crystallization be 
more or less perfect; and that they may be exhibited in a fluid 
metal, where, of course, there can be no crystallization. When, 
however, two metals are combined, the change of structure thus 
produced, is attended with a change of electro-magnetic proper- 
ties. The alloy of bismuth and tin in Table I. is negative with 
copper wires, though each metal separately is positive, and one 
of them to a high degree; in other mstances, as zine with lead, 
the deviation of the predominating metal seems to be increased 
rather than diminished, by its union with another, whose deviation, 
separately considered, is opposite. 
Leaving these theoretical considerations, I wish to call the 
attention of those who may be disposed for a further investigation 
of this subject, to the facilities which this mode of exciting electro- 
magnetism affords, for examining some points, which are as yet 
undecided. 
* In this and other instances it appeared to be immaterial whether one half of the 
bar were heated, or merely the extremity. 
