172 Scientific Proceedings, Royal Dublin Society. 



In order to simplify matters by observing the current when 

 only one side of the flame changed, a row of burners were so 

 arranged that one after the other could be lighted beneath the 

 wire. The current, on turning on the burners in succession, was in 

 the direction that the ignition travelled, or to the side of the steepest 

 gradient in temperature. On turning the burners out in reverse 

 order, one after the other, the current now flowed in the opposite 

 direction to that indicated in the first instance. And it was in- 

 creased when the cooling was hastened by applying water. The 

 steep gradient is, in both cases, on the same side ; that is, both 

 while the flame is spreading out, and again while it is going back. 

 Yet the currents are in opposite directions. However, though the 

 steep gradients are on the same side, it is to be remarked that 

 everything is not in the same condition in both cases ; for in one 

 the temperature is everywhere rising, while in the other, that is as 

 the flame goes back, the temperature is everywhere falling. So 

 that it would be insufficient to consider the electromotive force 



e = I — J, but must rather be £ = f — -, — j, 6 being the tempera- 

 ture at any point at a distance x along the wire from a fixed point, 

 and t denoting the time. 



Direct experiments were made to determine if a difference in 

 the flow of heat in opposite directions in a wire was sufficient alone 

 to produce a current. Thus, along a wire on one side of a heated 

 place a moist thread was laid and kept moistened. On coming to 

 a permanent state no appreciable current was observed, though 

 there must be a very great difference in the rate of flow of heat to 

 either side, due to the great difference in the temperature gradients. 

 I find a similar experiment was made by Le Roux, 1 who came also 

 to the conclusion that no current whatever was produced by a dif- 

 ference in the flow of heat in opposite directions in a wire. In 

 another experiment no current was observed in a wire kept heated 

 at the place where it came up out of a vessel of water. So that 



,, , idB dd\ , . . dd . 



there can be no term in e = — , — ) containing — alone ; Dut 



K dx dt J ° dx 



d 2 d 

 it probably consists principally of — dt 



ClOO 



1 Annates de Chemie et de Physique, quatrieme series, tome x., p. 208. 



