311 
diurnal variation of the magnetic needle , &c. 
their contact being supposed to take place as between a bar 
and a wire connecting the two ends , heat being applied at a 
joint , we must first enquire, whether substances which ex- 
hibit magnetic pheenomena when thus joined, and heat is so 
applied, will also exhibit them when their contact is sym- 
metrical throughout, and heat is applied to any point ; and if 
so, whether the directions in which the forces elicited would 
act, according to the distribution of heat on the earth, will 
account for any of the phcenomena of terrestrial magnetism. 
In the first experiments which I made with these views, 
I employed a compound ring of bismuth and copper, the 
bismuth having been cast in a circular ring, the outer cir- 
cumference of which was formed by a slip of thin copper 
having the two ends rivetted together, and holes punched 
through at regular distances : the metals were thus in contact 
throughout. With this ring I immediately found, that, to 
whatever point heat was applied, magnetic phenomena were 
exhibited, a needle being affected differently, according to 
the different positions in which the ring was placed with 
regard to it. It is not my intention to detail the results 
which I obtained, as, although they were as uniform as I 
could expect, the experiments were made only for a first 
trial, to ascertain whether any sensible effect would be pro- 
duced on a needle, by copper and bismuth so united, to 
whatever point in the combination heat was applied. They re- 
moved all doubt on this part of the subject ; but after making 
them, I was so much otherwise engaged, that more than two 
years elapsed before I had sufficient leisure to repeat and 
extend them with a more efficient apparatus. 
In the experiments which I have more recently made, and 
