in Terms of the Mechanical Equivalent of Heat. 7 



taken to ]/. The former was about 13°, the latter 6°. T was 

 obtained by observing ten transits with a seconds-clock, allow- 

 ing the coil to vibrate for several minutes, and then taking ten 

 more transits. The difference between the mean times of the 

 two series divided by the number of vibrations gives T very 

 exactly. The difference between the values before and after 

 the experiment never exceeded 1 part in 3000. The mean 

 value is about 2*42 seconds. H was determined before and 

 after the main experiment. 



The quantities in the formula for J remaining to be dis- 

 cussed are t, h, and the deflections. To treat these intelligibly, 

 I proceed to describe the method of experiment exactly. 



First, a determination of H was made. The calorimeter was 

 then weighed, filled with distilled water at a temperature 

 usually 2° or 3° below that of the air, carefully wiped with a 

 towel to remove moisture, again weighed and placed in the 

 water-jacket. Its amalgamated electrodes were placed in the 

 mercury-cups with the terminals of the two circuits, the main 

 circuit being broken at the commutator. The water-jacket was 

 kept permanently filled, and stood in a room of fairly constant 

 temperature, so that its temperature changed little during the 

 experiment. The thermometer was placed in position and the 

 stirrer started. During a few minutes readings were taken of 

 the thermometer and of three auxiliary thermometers, giving 

 the temperatures of the jacket, the 30,000-ohm coil, and the 

 air near the stem of the principal thermometer, the time of 

 each reading being noted by a seconds-clock. The circuit was 

 then closed and a galvanometer-reading taken, one of the com- 

 mutators was reversed and another reading taken, the time of 

 each reading being noted. The time of passage of the mercury 

 of the thermometer over several successive scale-divisions was 

 then taken, also readings of the other thermometers. Two more 

 commutator-reversals and galvanometer-readings followed, 

 then another set of thermometer -readings; and this alternation 

 was continued for about 40 minutes, during which time the 

 thermometer rose about 12° C. Usually sixteen galvano- 

 meter-readings were taken and seven groups of thermometer- 

 readings comprising 35 or 40 readings of the principal ther- 

 mometer. Then the circuit was broken and the calorimeter 

 allowed to cool for two or three hours, during which time 

 groups of readings were taken as before, the stirrer being kept 

 in motion. While this radiation-experiment was in progress 

 another determination of H was made. Finally, the thermo- 

 meter was removed and the calorimeter taken out and weighed. 



The mean of each group of thermometer-readings, corrected 

 for stem error, gives very exactly the temperature of the ther- 



