Faraday -Maxwell Mechanical Stress. 651 



fundamental import in the electromagnetic scheme that 

 without it no propagation of electromagnetic waves could 

 take place. Over and above this, we are met by essential 

 difficulties. 



20. One of these relates to the question of polarization. 

 For if an essential feature of electromagnetic wave-motion 

 is a motion of the aether parallel to the direction of propa- 

 gation, it would hardly seem possible for a beam of radiation 

 to be fully polarized. 



21. Another difficulty arises from the conclusion that, in 

 a progressive train, the motion of the aether is persistently 

 in the direction of propagation. Such uni-directional flow 

 involves something which cannot properly be described as 

 wave-motion at all ; while on the view now considered a 

 body such as the sun, which continually emits more radiation 

 than it receives, must be regarded as a source of aether. 



22. We might here evade the conclusion that an actual 

 creation of aether is involved in the process of radiation by 

 supposing a hot body to be a centre of afitherial condensation, 

 as compared with a colder body of like composition. But 

 the difficulty of the position becomes more evident when we 

 consider the conditions under which radiation is emitted by 

 a glow-lamp filament, say of tungsten. Let AD (fig. 3) be a 



o 



portion of such a filament, and let BCEF be a closed surface 

 made up of the narrow cylindrical surface FE, BC coaxal 

 with the filament, and two plane ends BF, CE perpendicular 

 to the filament. If we suppose that a bodily flow of aether 

 takes place in the direction of the Povnting vector, then so 

 long as the filament is maintained at a high temperature by 

 an electric current passing through it, there will be a flow 

 of aether outward through the cylindrical surface FE, BC. 

 Through the plane surfaces BF, BC there will be on the 

 whole no flow of aether, the circuital character of the electric 

 current precluding the idea that anything directly associated 

 with the current can flow into our closed surface (say) through 

 BF, except what inevitably flows out again through CE*. 

 Moreover, if the filament AD is in motion with respect to 

 the sether, we may suppose it accompanied by the ideal 



* We could even use an alternating current. 



