154 PROFESSOR W. THOMSON ON THE 



magnetization ; and may be said to be inductively crystalline. Or again, minute 

 fragments of non-crystalline substances may be put together, so as to constitute 

 solids, which, on a large scale, possess the general characteristic of homogeneous 

 crystalline substances ; and such bodies may be said to possess the crystalline 

 characteristic by structure, or to be structurally crystalline. 



148. As regards thermo-electric currents, the characteristic of crystalline sub- 

 stance must be, that bars cut from it in different directions would, when treated 

 thermo-electrically as linear conductors, be found in different positions in the 

 thermo-electric series ; or that two bars cut from different directions in the sub- 

 stance would be thermo-electrically related to one another like different metals. 

 This property has been experimentally demonstrated by Svanberg, for crystals of 

 bismuth and antimony ; and there can be no doubt but that other natural metallic 

 crystals will be found to possess it. I have myself observed, that the thermo- 

 electric properties of copper and iron wires are affected by alternate tension and 

 relaxation in such a manner, as to leave no doubt but that a mass of either 

 metal, when compressed or extended in one direction, possesses different thermo- 

 electric relations in different directions. Fragments of different metals may be 

 put together so as to form solids, possessing by structure the thermo-electric 

 characteristic of a crystal, in an infinite variety of ways. Thus, a structure con- 

 sisting of thin layers alternately of two different metals, possesses obviously the 

 thermo-electric qualities of a crystal with an axis of symmetry. I have inves- 

 tigated the thermo-electric properties in all directions of such a structure, in terms 

 of the conducting powers for heat and electricity, and the thermo-electric powers, 

 of the two metals of which it is composed ; and bars made up of alternate layers 

 of copper and iron, one with the layers perpendicular, another with the layers 

 oblique, and a third with the layers parallel, to the length, illustrating the theo- 

 retical results, which were communicated along with this paper, were exhibited to 

 the Royal Society. The principal advantage of considering metallic structures 

 with reference to the theory of thermo-electricity is, as will be seen below, that 

 we are so enabled to demonstrate the possibility of crystalline thermo-electric 

 qualities of the most general conceivable type, and are shown how to construct solids 

 (whether or not natural crystals may be ever found) actually possessing them. 



149. The following two propositions with reference to thermo-electric effects 

 in a particular case of crystalline matter are premised to the unrestricted treat- 

 ment of the subject, because they will serve to guide us as to the nature of the 

 agencies for which the general mathematical expressions are to be investigated. 



Prop. I. If a bar of crystalline substance, possessing an axis of thermo-electric 

 symmetry, has its length oblique to this axis, a current of electricity sustained in 

 it longitudinally will cause evolution of heat at one side, and absorption of heat 

 at the opposite side, all along the bar, when the whole substance is kept at the 

 same temperature. 



