356 PROCEEDINGS OF THE AMERICAN ACADEMY. 



result in the measurement of the diflference of temperature between the 

 calorimeters would be attained only when the mercury was sufficiently 

 agitated to make the temperature of the part of the bar surrounded by 

 the calorimeter uniform and the Thomson Effect consequently zero. It 

 also follows that the value obtained for the Thomson Effect would depend 

 on the rate of stirring of the mercury. The variation between the 

 several values found for the difference in temperature between the calo- 

 rimeters sometimes, as in the specimen quoted above, amounts to 50%. 

 It would seem from this that the difficulties mentioned above are real. 



Two points of less importance are : (1) No correction was made for 

 the change in the heat liberated by the Joule Effect due to the change in 

 temperature, and hence in resistance of the part of the rods in the calo- 

 rimeters. (2) No correction was made for the change in the rate of heat 

 conducted into or out of the calorimeters, due to the slight change in 

 the temperature gradient caused by the action of the Thomson Effect. 



General Description of Method of this Paper. 



The ends of the copper wire, in which the Thomson Effect was to 

 be measured, were fastened into copper blocks provided with a water 

 circulation to keep them at a constant temperature. The copper wire 

 was heated by an electric current. There was then a flow of heat from 

 the middle of the wire to each end. In one half of the wire this flow 

 of heat was opposite in direction to the flow of electricity, in the other 

 half it was in the same direction. Consequently in one half of the wire 

 the Thomson Effect liberated heat and in the other half absorbed it. 

 On reversing the electric current the effect was of course reversed. By 

 measuring the change of resistance of short sections of the wire on 

 reversal of the current, it was possible to determine the change of tem- 

 perature. By a separate experiment, with the same apparatus, the 

 relation between the heat lost by radiation and convection from any 

 part of the wire and the temjierature of that part was determined. 

 With these data, considering any part of the wire made up of a whole 

 number of sections, the following quantities could be determined. (1) 

 The change in the heat generated by the Joule Effect on reversal of 

 the electric current. (2) The change in the heat lost by radiation and 

 convection on reversal. (3) The change in the heat conducted out of 

 the cold end of the part of the wire under consideration on reversal. 

 (4) The change in the heat conducted in at the hot end on reversal. 

 These quantities suffice for the absolute measurement of the Thomson 

 Effect. 



