124 PROCEEDINGS OF THE AMERICAN ACADEMY 



Around the calorimeter on all sides was a water-jacket, nickel- 

 plated on its interior, to make the radiation perfectly definite. 



The calorific capacity of the thermometers, including the immersed 

 stem and the mercury of the bulb, was estimated as follows: 14 em - of 

 stem weighed about 3.8 gr -, and had a capacity of .8 gr ; lO 81- of 

 mercury had a capacity of .3 gr- ; total, l.l gr . 



Often the vessel B was removed, and the water allowed to flow 

 directly into the calorimeter. 



The following is the process followed during one experiment at low 

 temperatures. The vessel A was filled with clean broken ice, the open- 

 ing into the stopcock being covered with fine gauze to prevent any 

 small particles of ice from flowing out. The whole was then covered 

 with cloth, to prevent melting. The vessel was then filled with water, 

 and the two thermometers immersed to get the zero points. The 

 calorimeter being about two thirds filled with water, and having been 

 weighed, was then put in position, the holes corked up, and one ther- 

 mometer placed in it, the other being in the melting ice. An obser- 

 vation of its temperature was then taken every minute, it being 

 frequently stirred. 



When enough observations had been obtained in this way, the cork 

 was taken out of the aperture F and the spout inserted, and the water 

 allowed to run for a given time, or until the calorimeter was full. It 

 was then removed, the cork replaced, and the second thermometer 

 removed from the ice to the calorimeter. Observations were then 

 taken as before, and the vessel again weighed. 



Two thermometers were used in the way specified, so that one 

 might approach the final temperature from above and the other from 

 below. But no regular difference was ever observed, and so some 

 experiments were made with both thermometers in the calorimeter 

 during the whole experiment. 



The principal sources of error are as follows : — 



1st. Thermometers lag behind their true reading. This was not 

 noticed, and would probably be greater in thermometers with very 

 fine stems like Geissler's. At any rate, it was almost eliminated in 

 the experiment by using two thermometers. 



2d. The water may be changed in temperature in passing through 

 ■the spout. This was eliminated by allowing the water to run some 

 time before it went into the calorimeter. The spout being very thin, 

 and made of vulcanite, covered on the outside with cloth, it is not 

 thought that there was any appreciable error. It will be discussed 

 more at length below, and an experiment given to prove this. 



