172 PROCEEDINGS OF THE AMERICAN ACADEMY 



the air is the greatest. The ordinary formula for the correction is 

 .000156 n (t — I"). But, in applying this correction, it is difficult to 

 estimate n, the number of degrees of thermometer outside the calo- 

 rimeter and at the temperature of the air, seeing that part of the stem 

 is heated by conduction. The uncertainty vanishes as the thermometer 

 becomes longer and longer, or rather as it is more and more sensitive. 

 But even then some of the uncertainty remains. I have sought to 

 avoid this uncertainty by placing a short tube filled with water about 

 the lower part of the thermometer as it comes out of the calorimeter. 

 The temperature of this was indicated by a thermometer, by aid of 

 which also the heat lost to the water by conduction through the ther- 

 mometer stem could be computed ; this, however, was very minute 

 compared with the whole heat generated, say 1 in 10,000. 



The water being very nearly at the temperature of the air, the stem 

 above it could be assumed to be at the temperature of the air indicated 

 by a thermometer hung within an inch or two of it. The correction 

 for stem would thus have to be divided into two parts, and calculated 

 separately. Calculated in this way, I suppose the correction is per- 

 fectly certain to much less than one hundredth of a degree : the total 

 amount was seldom over one tenth of a degree. 



Among the uncertain errors to which the measurement of temper- 

 ature is subjected, I may mention the following : — 



1. Pressure on bulb. A pressure of 60 cm - of water produced a 

 change of about 0°.01 in the thermometers. When the calorimeter 

 was entirely closed there was soon some pressure generated. Hence 

 the introduction of the safety-tube, — a tube of thin glass about 

 jQcm. loDg^ extending through a cork in the top of the calorimeter. 

 The top of the safety-tube was nearly closed by a cork to prevent 

 evaporation. Had the tube been shorter, water would have been 

 forced out, as well as air. 



2. Conduction along stem from outside to thermometer bulb. To 

 avoid this, not only was the bulb immersed, but also quite a length of 

 stem. As this portion of the stem, as also the bulb, was surrounded 

 by water in violent motion, there could have been no large error from 

 this source. The immersed stem to the top of the bulb was generally 

 about 5 cm " or more, and the stem only about .8 om - in diameter. 



3. The thermometer is never at the temperature of the water, be- 

 cause the latter is constantly rising ; but we do not assume that it is 

 so in the experiment. We only assume that it lags behind the water 

 to the same amount at all parts of the experiment, and this is doubt- 

 less true. 



