444 PROCEEDINGS OF THE AMERICAN ACADEMY. 



tions was modelled after that described in the paper upon the sulphate, 

 and is illustrated iu the accompanying Figure 2. The tube containing 

 the inverting mixture was surrounded by an air-jacket (A), and the 

 latter by water maintained at the desired temperature above the transi- 

 tion temperature by an electrically heated resistance wire. The ther- 

 mometer itself was surrounded by a long jacket (D) of good glass 

 through which a current of pure water was maintained, flowing from a 

 large thermostat, in order to keep the stem of the thermometer at tho 

 same temperature as the bulb. Preceding experience had taught us to 

 heed the cooling of this stream of water as it flowed from an elevated 

 thermostat (C) ; of course, at the temperature of 50°, the cooling was 

 even more marked than in the case of sodic sulphate at 32°. To avoid 

 any error on this account, a small extra thermometer was inserted in 

 the long jacket around the standard thermometer to record the tempera- 

 ture of its stem. This extra thermometer is not shown in the diagram. 

 In order to be certain of the temperature of the transition to a thou- 

 sandth of a degree, it is necessary to be certain of that of the stem to a 

 tenth of a degree. The evenly distributed cooling of the flowing 

 column in the distance represented by the length of the mercury thread 

 amounted to no more than 0.4°, therefore the temperature of the water 

 in the jacket observed at about the middle point of the column could be 

 taken as the average temperature of the column. The thermostat (C) 

 was accordingly set so that the temperature of the flowing water at this 

 point was within a tenth of a degree of the temperature of the inverting 

 salt. In this way the conditions for accurate thermometery were 

 adequately obtained and all uncertain corrections were avoided. 



According to the degree of the exposure of the top of the large test- 

 tube containing the mixture it is necessary to maintain the water bath 

 outside its air-jacket at a temperature somewhat above that of the 

 mixture, in order to allow for the cooling effect of convection in the air- 

 jacket. For the particular apparatus employed in these experiments and 

 at the temperature of 50° this interval was found to be 2.5°. With 

 this iuterval the inflow of heat was so nicely regulated that the tempera- 

 ture of the sodic bromide was maintained for at least two hours with 

 wonderful constancy. 



The stirrer used in stirring the salt and solution during the observa- 

 tion of the transition temperature was made of glass and a large sealed-iu 

 platinum wire. Platinum alone was first tried, but it was deemed to 

 have too large a thermal conduction ; and as the containing tube was 

 glass, a further stirrer of glass could be accepted. Instead of powder- 



