216 PROCEEDINGS OF THE AMERICAN ACADEMY 



shall see that in special cases there is another heat effect at this plane ; 

 it is probable, therefore, that this effect exists in every case. 



An abrupt variation of temperature can be produced practically in 

 various ways. Le Roux and others have produced the currents by 

 cutting a strip of metal in such a shape that its section varied abruptly 

 from the broad to the narrow part. When a flame is aj^plied to the 

 narrow part, it becomes heated so much more rapidly than the broad 

 portion that the temperature changes abruptly. 



Maxwell says, in regard to these cases, that a current is produced 

 in a homogeneous conductor when " at any part of the conductor a 

 sensible variation of temperature occurs between points whose distance 

 is within the limits of molecular action." 



Le Roux attempts to explain the phenomena by a difference of 

 pressure caused by the heat, the current being due to heating the 

 contact of compressed and free parts of the same metal. 



Becquerel produced an abrupt variation of temperature by suddenly 

 uniting the hot and cold ends of a wire. The current in this case is, 

 as Becquerel pointed out, at least partly due to a coating of oxide. 

 In the case of lead the current is probably due to a coating of oxide, 

 as the Thomson effect for lead is zero. 



In the experiments made by us, with the object of observing whether 

 there was any law connecting the direction of the current with that of 

 the Thomson effect, the method of varying the section abruptly was 

 employed. This method gets rid of all complications due to impuri- 

 ties at the surface of contact ; but in cutting the metal to the neces- 

 sary shape, the parts were necessarily more or less strained, and the 

 results sought were always masked by a permanent current between 

 the strained and unstrained parts. The direction of this current de- 

 pended upon the position of the point where the heat was applied. 

 The direction could be reversed by moving the lamp a few centimeters 

 to either side, beneath the narrow part. The best method to avoid 

 these complications is to pass a fine wire through the non-conducting 

 partition separating the two compartments of a vessel, each compart- 

 ment being filled with some good thermal conductor, and kept at 

 different temperatures. This method has not yet been tried. 



We have seen that the equations do not determine the direction of 

 the thermo-electric current, and consequently that it is probable that a 

 current exists in both directions. The Thomson effect, then, cannot 

 alone explain the current that experiment shows to exist ; there must 

 be some other effect at the plane of abrupt variation. That this effect 

 does take place in certain special cases there can be no doubt. 



