STRESSES IN THE DIELECTRIC 141 



dielectric be solid or fluid. Consider a small cube with four vertical 

 edges parallel to the lines of force, Fig 93. The system of stresses 

 is obviously equivalent to a shear stress in the plane of the figure and 

 to pressures on the two faces parallel to that plane. Now, in a fluid 

 in equilibrium ordinary elastic shear stresses cannot exist and can- 

 not contribute to maintenance of equilibrium. Indeed, hydrostatic 

 equilibrium is only possible in a fluid when the pressures about a 

 point are equal in all directions. Further, even in a solid the 

 elastic system of a shear stress and a perpendicular pressure would 

 produce on the whole a decrease in volume. For the shear would 

 not affect the volume, while it is easy to show that the pressure P 



1 2 cr 

 would decrease it bv P , where or is Poisson's ratio and Y 



is Young's modulus. But experiments to be described in the next 

 chapter show that, at any rate in the case of glass, the presence 

 of electric strain is accompanied by a uniform dilatation. 



We must therefore suppose that the electric stresses are not 

 elastic forces accompanied by elastic changes of shape, but that 

 they are called into play in some other way. Where the medium 

 is material we may probably account for the stresses by molecular 

 arrangement, on the supposition that the molecules are electric 

 doublets. 



We may illustrate the idea by considering the corresponding 

 case of a magnetic system. Suppose that we have a number of 

 little magnets pivoted on points arranged in rows 

 and columns, as in Fig. 94. If the magnetic axes 

 are equally distributed in all directions there will 

 be no continuous lines of force going in any one 

 direction. But if the axes are arranged as in the 

 figure the opposite poles in the columns will 

 attract each other, forming tensions along the 

 lines of axes. The like poles in the neighbouring 

 columns will repel each other, forming pressures 

 perpendicular to the lines of axes. Some such 

 arrangement of electric doublets in a material 

 medium may account for the electric stresses 

 accompanying electric strain. We cannot say, 

 u />riori, what effect such a rearrangement should YLU. 94. 

 have on the dimensions of the system. We only 

 know by experiment that it appears to lead to a uniform 

 dilatation. 



When the matter is exceedingly attenuated it may be doubted 

 whether we can account for the stresses by the molecules present, 

 and we may have to imagine some structure in the ether before 

 we can frame a hypothesis to supply the forces. 



