206 Richards — Transition Temperature of &odic Sulphate. 



Third Problem — The Effect of Varying the Mass of the Phases. 



Theoretically, of course, a change in the relative amount of 

 the phases should cause no difference in the n on variance of the 

 point under consideration provided that no one phase totally 

 disappears. But since the superficial area of either solid 

 present is a factor in determining the speed of the adjustment 

 of the equilibrium, it was thought that varying proportions 

 might cause a slightly varying " temperature lag." Neverthe- 

 less, as a matter of fact, no reasonable increase in the amount 

 of either of the solids or of the liquid seemed to cause the least 

 effect, provided that the process was one of melting and not of 

 solidifying, and the mixture was properly stirred. The Beck- 

 mann thermometer remained always at 4*760 as long as the 

 barometer stood near to the normal height. In order to insure 

 the rapid establishment of the equilibrium it was our custom 

 to have always some anhydrous powder present in the first 

 place. This precaution also renders harmless drops of water 

 introduced on the washed thermometer or the stirrer. It is 

 well to powder also the hydrated salt, or at least to use it in 

 the form of fine crystals, so as to increase its surface. 



Since the point in question seems then to fulfil all the 

 requirements of an accurate standard of reference, manifestly 

 its relations to the international standard should be established. 



II. Reference to the International Standard. 



The only good thermometers indicating 33° at hand in the 

 chemical laboratory are two of a set (one of which has already 

 been mentioned) made by Baudin, and these have not been 

 accurately calibrated. On one occasion thermometer Baudin 

 No. 9,389 indicated 32*570° as the point in question, and 

 immediately afterward came to 0*097° in pure melting ice, a 

 difference of 32*473°. Subsequently the same thermometer 

 indicated a change of 32*560- 0*086= 32*474 ; while the other 

 thermometer ^No. 9,390) gave on two successive trials separated 

 by perhaps a week 32*560-0*084=32*476 and 32*550-0*083= 

 32*467. The mean of all these values is 32*472, and to this 

 about 0*009" must be added to correct for the column of 10° 

 exposed to the temperature of the room (about 26°). No other 

 corrections are possible, and there is no further guarantee of 

 the accuracy of the number 32*481° than the evident care used 

 in the construction of the thermometers and the fact that they 

 agree very closely. It is probable that the instruments were 

 made to be used vertically, hence no correction to the horizon- 

 tal position is needed. 



*See tables by Rimbach m Landolt und Bornstein, page 143. 



