KING. — MEASUREMENT OF THOMSON EFFECT IN COPPER. 355 



applied, the mercury in the calorimeters being violently agitated mean- 

 while and throughout the whole experiment. (1) Readings of the 

 difference of temperature of the calorimeters were taken at half-minute 

 intervals for twenty minutes with no electric current flowing. (2) Read- 

 ings were taken at like intervals for the same time with a known electric 

 current flowing. (3) Readings were taken as before for twenty minutes 

 with no current flowing. 



From these three sets of observations, the actual difference in tempera- 

 ture between the calorimeters due to the heat liberated in one and 

 absorbed in the other could be found. This difference in temperature 

 between the calorimeters reached in 20 minutes, deduced from the three 



FiGDRE 1. 



sets of observations as described above, is multiplied by the wafer equiva- 

 lent of the calorimeters ; the product is divided by the number of seconds 

 in 20 minutes, by the difference in temperature between the ends of the 

 calorimeters, that is between the points a and I), Fig. 1, by the electric 

 current in c. g. s. measure, and by 2 to allow for the reversal of the 

 current. 



In a set of six experiments on Cadmium made as described above, the 

 difference in temperature obtained between the calorimeters ranged 

 between 0°.0122 C. and 0°.0085 C. The difference of temperature 

 between the ends a, b, of the calorimeters measured in the bar was 

 21°.0 C. 



There are two great difficulties in Batelli's method. (1) It is neces- 

 sary to measure accurately a mean difference of 0°.012 C. between boxes 

 of mercury, each of which surrounds a part of a bar in which there exists 

 a difference of temperature of 21°.0 C. It seems likely that in spite of 

 violent stirring considerable differences of temperature would be found 

 in the mercury. (2) The presence of the violently agitated mercury in 

 the calorimeter tends to destroy the temperature gradient in the rod, 

 on which the Thomson Effect depends. In fact, it seems that the best 



