the Theory of Electrons. 273 



straight the electrons are in equilibrium. Contrary to the 

 assumption implicitly made in thus appealing to the principle 

 of virtual work, such electron-models are not " freely mobile"; 

 when the ends of the originally straight wire are moved from 

 A, B to A', B', by bending it after the electrons have been 

 formed, the resulting equilibrium-strain in the aether is not 

 precisely the same as that due to two real point-charges at A',B'. 

 In the former case, at every point P of the wire the com- 

 ponents along the normal to its surface of the vectors which 

 represent the aethereal strain and the rotational displacement 

 of the element of surface of the wire are the same ; in the 

 latter the strain-vector is the resultant of e/4:ir . A'P -2 along 

 A'P, and —e/ATr* B'P -2 along B'P. The strains cannot then 

 in general be the same. 



The strain-distribution in each of the above cases, more- 

 over, is different from that which would exist if the wire 

 were first bent into the new position, then connected to the 

 aether, and afterwards rotated by the imposed torque. 



The specification of this illustrative model can however, I 

 believe, be amended so as to permit the existence, and show 

 the transmission, of attraction between the electrons. The 

 following mode of doing so, though perhaps not very elegant, 

 would serve this purpose. Suppose the wire replaced by an 

 indefinitely long hollow tube, having only slight resistance to 

 torsion, and riding loose on a concentric fixed rigid shaft ; 

 let keyways parallel to the axis be cut on the inside of the 

 tube and on the shaft, in such positions that in order to key 

 the tube and shaft together the former must be rotated 

 against the reacting torque of the aether ; and let the rigid 

 key, of finite length, which fits in these ways, terminate in 

 rounded or spherical ends. If the tube be now thus forcibly 

 rotated around its own axis, and keyed to the shaft, the 

 equilibrium-strain in the aether will, as the torsional resistance 

 of the unkeyed portion of the tube is only slight, be almost 

 identical with that due to two conjugate electrons coinciding 

 Mith the spheres. Also the portions of the keyway on the 

 tube beyond the spheres will assume the form of spirals of 

 some kind very flat near the spheres, and will therefore 

 touch the latter at some point or points so situated that the 

 resulting pressures will not be at right angles to the axis of 

 the shaft, but will have components urging the spheres 

 towards' each other. And, further, these electrons are " freely 

 mobile " along the line joining them ; an alteration of the 

 length of the key merely alters the distance between the 

 electrons. An endlong thrust in the key at the same time 

 prevents this mobility coming into play, and resists the 



