194 THE EARL OF BERKELEY ON SOME 



quarters of an inch from the top, was passed through the top plate of the thermostat and 

 soldered in position. The body of the thermostat was of copper and held about three 

 litres of water, and the expansion and contraction of this was enough to actuate the 

 flexible diaphragm (the iron plate of the receiver, of a telephone) sufficiently to keep 

 the temperature of the thermostat constant to about 0'2 C. ; that is to say, that for 

 2 or 3 hours before taking an observation, the temperature in the beaker would only 

 show, at the utmost, a change of 0'05 C., but in the course of 24 hours after setting 

 the temperature of the thermostat, and consequently that of the solution, might slowly 

 rise or slowly fall to the extent mentioned, and then remain constant to 0'05 C. 



Means employed for Determining the Point of Saturation. 



After numerous experiments the following general method was found to be the 

 most satisfactory. The thermostat was set at 90 C. , and water, together with a 

 quantity of salt more than necessary to saturate it, placed in the platinum beaker, 

 and the mixture stirred very rapidly until it was thought that saturation had been 

 attained ; an observation of the density was then made, and the stirring continued for 

 a further period of 2 or 3 hours, and then another density determination made. If 

 the two observations agreed within the experimental errors, it was considered that 

 saturation had been practically reached ; if the two observations did not agree, the 

 stirring was continued and the density taken at intervals until it became constant. 

 The temperature of the thermostat was then lowered 2 degrees, and after stirring 

 2 or 3 hours the density again taken. The mean of this and of the constant 

 density previously mentioned was considered to be the density of a saturated solution 

 at the mean of the respective temperatures. The temperature of the thermostat was 

 then lowered to the next point of observation, and after 2 or 3 hours' stirring the 

 density was taken ; water was then added to the solution and the stirring continued 

 until the density, taken at intervals of from 4 to 12 hours, was constant the mean of 

 the first and the last observations, which usually differed by an amount slightly 

 greater than the experimental errors, was taken a giving the true density. The 

 process was then repeated for the other temperatures. It should be noted that 

 whether working with supersaturated or an unsaturated solution, the liquid is always 

 stirred in contact with a large excess of solid salt. 



In the case of salts whose solubilities decrease with an increase of temperature, 

 the process is reversed ; with Na 2 S0 4 , for example, which has a maximum solubility 

 at 32 - 5 C., the thermostat was set at 33 C., and stirring continued until constant 

 density was obtained ; the temperature was then raised 1 C. and the density again 

 determined, and the mean of this and of the constant density above mentioned was 

 taken as the true density of a solution saturated at the mean of the respective 

 temperatures. The temperature of the thermostat was then raised to the next point 

 of observation, the density taken, boiling water added and a constant density 



