A STUDY IN MORBID AND NORMAL PHYSIOLOGY. 



21 



Trial No. 2. — Small Calorimeter. 



Extended comment upon these tables does not seem necessary. They appear 

 to show conchisively that tlie nearer the apparatus is in temperature to the air the 

 less the chance of serious error; also that in comparative experiments, when the 

 calorimeter is above the temperature of the air, the least chance of serious error is 

 to be obtained by maintaining tlie conditions uniform during the whole experiment, 

 rather than by attempting to calculate the amount of heat lost by the calorimeter 

 under varying conditions. Besides the trials reported, a number of others were 

 performed in which the calorimeter was J>i'Io)o the temperature of tlie air. These con- 

 clusively proved what is, a jn'iori, probable, that, when the difterence of temperature 

 is not more than three degrees, the calorimeter works with almost exact accuracy; 

 there is of course no loss of heat from th(" calorimeter, and its power of absorbing 

 caloric from the air seems to be nill. Unfortunately I did not appreciate the im- 

 portance of having tlie calorimeter cooler tlian the air until late in the investiga- 

 tion, but relied for accuracy upon maintaining similar conditions throughout each 

 single experiment, and keeping the difference between the temp(>ratures of the air 

 and calorimeter as little as possible. The chief safeguard against error has seemed 

 to me to be found in having a number of experiments. If in such a series of 

 similar relative experiments upon animals the result is at tlie same time fairly 

 uniform and very decided in one direction, it is practically demonstrated that the 



