336 Lord Kelvin on the 



§ 7. This produces an electric effect which, for great 

 distances from the molecule, is calculated by the same 

 formula as the magnetic effect of an infinitesimal bar-magnet 

 whose magnetic moment is numerically equal to the product of 

 the quantity of electricity of a single atom into the sum of the 

 diminutions of the two distances between the atoms of the two 

 pairs under consideration. Hence, denoting by N the number 

 of crystalline molecules per unit bulk of the crystal ; by b the 

 radius of the circle of each crystalline molecule ; by q the 

 quantity of electricity on each of the six atoms or double- 

 atoms, whether positive or negative ; by 3 the change of 

 direction of each of the two diameters through atoms which 

 experience change of direction ; and by //, the electric 

 moment* developed per unit volume of the crystal, by 

 the strain which we have been considering and which is 

 shown in tig. 2 ; we have 



/* = ISr2.4&$cos30 o =N.262$v/3 . . (1). 



It is of course understood that $ is a small fraction of a 

 radian. 



§ 8. To test the sufficiency of our theory, let us first 

 consider quantities of electricity which probably, we may 

 almost say certainly, are present in the atoms in nature. 



Instead of the silicon atoms marked + in the diagram let 

 us substitute globes of polished zinc ; and instead of the 

 double-oxygen atoms marked — let us substitute little globes 

 of copper well oxidized (polished copper, heated in air till it 

 becomes of a dark slate-colour). Let us suppose all the six 

 atoms of each compound molecule to be metallically connected, 

 and all the molecules insulated from one another. We are 

 not concerned with conceivable permeation of electricity by 

 conductance through the crystal ; and therefore we must 

 suppose the total quantity of electricity on each crystalline 

 molecule to be zero. Let the circle of each compound mole- 

 cule in the diagram be a real exceedingly thin stiff ring of 

 metal, no matter what kind of metal, and let each of the six 

 atoms be a bead (a perforated spherule) , whether of zinc or of 

 copper, moviDg frictionlessly on it. Thus we have, in idea, 

 a working model of an electrically eolotropic crystalline 

 molecule, 



§ 9. I have found by experiment f that the difference of 



* I do not know if this designation lias hitherto been used. I introduce 

 it with precisely the same significance relatively to electricity, as the 

 well-known " magnetic moment " in reference to magnetism. 



t ' Electrostatics and Magnetism,' § 400, and experiments not hitherto 

 published, by another method, 



