Dr. Draper 07i the Electro-motive Power of Heat. 459 



These results explain the anomalies observed by some of 

 those who investigated the course of thermo-electric currents 

 by means of small metallic fragments. 



It would therefore seem, that when wires of the same 

 metal are used as electromotors, the quantity of electricity 

 evolved depends on the quantity of caloric that can be com- 

 municated in a given time. Time, therefore, under these 

 circumstances must enter as an element of thermo-electric 

 action. In the case of a single metal, the maximum eifect 

 would be produced, when the hot element is a mass, and the 

 cold one a point. 



And lastly, " That the quantities of electricity evolved in a 

 pile of pairs are directly px'oportional to the number of the 

 elements." 



In the first trials which I made, to determine the effect of 

 increasing the number of pairs in a pile, the results obtained 

 were contradictory ; by operating however in the following 

 way, the proposition was at last satisfactorily determined. 



1st. The resistance to conduction was made nearly con- 

 stant by uniting all the pairs intended to be worked with, at 

 once. The current, therefore, whether generated by one, 

 two, three, four, or more pairs, had always to run through the 

 same length of wire, and experienced in all cases an uniform 

 resistance. 



2ndly. By making each individual element of considerable 

 length, the liability of transmission from the hot to the cold 

 extremity was diminished. 



Having, therefore, taken six pairs of copper and iron wires, 

 Y^-g-th of an inch thick and each element 38 inches long, I 

 formed them, by soldering their alternate ends^ into a con- 

 tinuous battery. Then I successively immersed in boiling 

 water one, two, three, &c. of the extremities, their length al- 

 lowing freedom of motion, and the other extremities not dif- 

 fering perceptibly from the temperature of the room. 



The following table exhibits one of this series of experiments. 



Table IV. 



No. of pairs. 



Calculated Deviations. 



Observed Deviations. 



1. 



55 



55 



2. 



110 



111 



3. 



165 



165 



4. 



220 



220 



5. 



275 



272 



6. 



330 



332 



