RICHARDS AND WELLS. — TRANSITION TEMPERATURE. 443 



for the determination of the atomic weight of cadmium. Both speci- 

 mens of sodic bromide were well drained, desiccated in vacuum, fused in 

 nitrogen, weighed and analyzed by conversion into silver bromide as 

 usual. 



Of our sodic bromide 5.49797 grams yielded 10.03253 grams of 

 argentic bromide, and of the other sample 3.64559 grams yielded 6.G5248 

 grams, all the weighings being reduced to the vacuum standard. The 

 respective weights of argentic bromide corresponding to 1.00000 parts 

 of sodic bromide are 1.82177 and 1.82480, showing the essential 

 identity of the samples. 



These two analyses not only serve this purpose, however ; they like- 

 wise furnish new data for calculating the atomic weight of sodium, which 

 is thus found to be 23.008, if silver is taken as 107.930 and bromine 

 as 79.955. The new result from the chloride is also 23.008, an exactly 

 identical figure. Thus the comparison of the results furnishes another 

 reason for believing that the salt used in our transition temperature 

 experiments was pure. Only their small number and our uncertainty 

 concerning the exact atomic weight of bromine debars us from present- 

 ing these results in a special paper as a contribution to the literature on 

 atomic weights. 



Method of determining the Transition Temperature. 



The apparatus necessary for this work naturally divides itself into 

 two parts, — the thermometers on the one hand, and the apparatus for 

 containing the salt on the other. The thermometers employed have all 

 three been standardized with the greatest care by the International 

 Bureau of Weights and Measures in Sevres. They have already been 

 described in detail in the previous papers. 6 They were those bearing the 

 designations Tonnelot 11142 and Baudin 15200 and 15275. Further 

 description of them is unnecessary here. 



To contain the salt two different arrangements were used. In- one 

 case the salt was contained in a wide test-tube placed in the bottom of 

 a long still wider sealed tube sunk in a very deep thermostat until the 

 50° mark on the thermometer stem was almost level with the water. 

 The object of this device was obviously to maintain the thermometer 

 stem all at the transition temperature in order to avoid any considerable 

 correction for the cooled column. 



The more convenient apparatus employed in all the other determina- 



6 Am. Jour. Science, 6, 201 (1898) ; These Proceedings, 38, 431 (1902). 



