ON ELECTEIC ENDOSMOSE AND OTHER ALLIED PHENOMENA. 503 



in the neighbourhood. For instance, in the circumstances of Quincke's 

 experiments we havo 



and, therefore, for a suspended particle of the same nature as the walls of 

 the tube we should have for the absolute velocity the value 



_3 crJ cE 

 2 7rR2 * ^ 



when the particle is in the axis, and 



_1 jtJ cE 

 2 7rB,2 * J 



when it is near the walls.' 



4. We may next consider the electromotive forces produced by the 

 passage of a liquid through a capillary tube or a porous diaphragm. 

 This subject has been studied experimentally by Quincke, Edlund, Haga, 

 Clark, and more recently by Dorn,^ the general result being that the 

 potential is higher on the side where the pressure is least by an amount 

 proportional to the difference of pressure. The phenomenon is ascribed 

 to a sort of electric convection, the superficial electrified layer of fluid 

 carrying its charge with it as it slides over the walls of the channels. 

 In the case of a straight uniform tube, for instance, there is in this way 

 a transfer of positive electricity along the walls, from the near to the 

 farther end, which is compensated, if no other path is open, by conduction 

 backwards through the column of liquid in the tube. If the tube be of 

 varying section there will be a tendency also to convergence of positive 

 or negative electricity by convection at intermediate points, and a conse- 

 quent establishment of ' sources ' and ' sinks ' as regards the conducting 

 mass of fluid in the interior. 



Taking the case of a tube of circular section, through which fluid is 

 forced by an excess of pressure P, and using the same notation as in § 2, 

 we find by the ordinary theory of Poiseuille's experiments 



Hence the total quantity of electricity carried per second along the wall 

 of the tube is 



If no other conducting channel is open the electricity thus carried forward 

 will return by ordinary conduction through the column of liquid in the 

 tube. Since the resistance of this column is ctL/ttR^, the difierence of 

 potential between the ends of the tube is 



|.cE .... (27) 



If E is positive (as it appears to be in most cases) the higher potential is 



' It is to be noticed that one of Quincke's observations remains unexplained, 

 viz., the fact that in sufficiently wide tubes the direction of motion of particles near 

 the walls varied with the strength of the current. 



2 Kor references see Wiedemann, EleUricitdt, ii. pp. 153 et seq. 



