64 Boyal Society, 



be where the temperatures of the single metals are non-uniform, 

 since the thermal effect of a current in any homogeneous uniformly- 

 heated conductor is always an evolution of heat. Hence there must 

 be on the whole an absorption of heat, caused by the current 

 in passing from cold to hot in copper, and from hot to cold in iron. 

 When a current is forced through the circuit against the thermo- 

 electric force, the same reasoning establishes an evolution of heat to 

 an amount equivalent to the sum of the heat that would be then 

 taken in at the cold junction, and the value in heat of the energy 

 spent by the agency (chemical or of any other kind) by which the 

 electromotive force is applied. The aggregate reversible thermal 

 effect, thus demonstrated to exist in the unequally heated portions 

 of the two metals, might be produced in one of the metals alone, or 

 (as appears more natural to suppose) it may be the sum or difference 

 of effects experienced by the two. Adopting as a matter of form the 

 latter supposition, without excluding the former possibility, we may 

 assert that either there is absorption of heat by the current passing 

 from hot to cold in the copper, and evolution, to a less extent, in the 

 iron of the same circuit; or there is absorption of heat produced by the 

 current from hot to cold in the iron, and evolution of heat to a 

 less amount in the copper ; or there must be absorption of heat in 

 each metal, with the reverse effect in each case when the current is 

 reversed. The reversible effect in a single metal of non-uniform 

 temperature may be called a convection of heat ; and to avoid cir- 

 cumlocution, I shall express it, that the vitreous electricity carries heat 

 with it, or that the specific heat of vitreous electricity is positive, 

 when this convection is in the nominal " direction of the current," 

 and I shall apply the same expressions to " resinous electricity '* 

 when the convection is against the nominal direction of the current. 

 It is established then that one or other of the following three hypo- 

 theses must be true : — 



Vitreous electricity carries heat with it in an unequally heated 

 conductor whether of copper or iron ; but more in copper than in 

 iron. 



Or Resinous electricity carries heat witii it in an unequally heated 

 conductor whether of copper or iron; but more in iron than in 

 copper. 



Or Vitreous electricity carries heat with it in an unequally heated 

 conductor of copper, and Resinous electricity carries heat with it in 

 an unequally heated conductor of iron. 



Immediately after communicating this theory to the Royal Society 

 of Edinburgh, I commenced trying to ascertain by experiment which 

 of the three hypotheses is the truth, as Theory with only thermo- 

 electric data could not decide between them. I had a slight bias in 

 favour of the first rather than the second, in consequence of the 

 positiveness which, after Franklin, we habitually attribute to the 

 vitreous electricity, and a very strong feeling of the improbability of 

 the third. With the able and persevering exertions of my assistant, 

 Mr. M*^Farlane, applied to the construction of various forms of ap- 

 paratus and to assist me in conducting experiments, the research 



