RICHARDS AND WELLS. — TRANSITION TEMPERATURE. 441 



of the several specimens of the various samples. These analyses need 

 not be detailed here, because all of the specimens used in the final work 

 to be recounted were as free as possible from chlorine, but they were 

 amply worth the trouble which they occasioned, because they not only 

 proved the slight impurity in the bromine, but also led to the subse- 

 quently confirmed suspicion concerning the atomic weights of sodium 

 and chlorine. 



The reason for this difficulty in preparing pure bromine is clear. 

 The elimination of chlorine from bromine by solution in calcic bromide 

 and precipitation with water is a question of the distribution of the 

 chlorine between two phases, much the greater part of the chlorine 

 going into the aqueous phase. When the amount of the chlorine in the 

 original bromine is small, a good product may be obtained by a single 

 treatment ; but when it is large, several successive treatments must be 

 used in order to eliminate every weighable trace of chlorine. One of 

 the portions of sodic bromide used in the transition temperature work 

 was made from bromine thus purified several successive times in addition 

 to repeated distillation. 3 This sample, perhaps as pure as may be 

 obtained by any method, is called III in the work which follows. A 

 fourth (called IV) was prepared by another method of procedure. 

 Potassic permauganate was recrystallized until free from chloride, the 

 tests being made in small portions after precipitation with alcohol and 

 filtration of the manganese dioxide. Sulphuric acid was boiled until 

 free from chloride. With the help of these two substances, bromine 

 was set free from a large sample of pure sodic bromide (sample II). 

 After shaking with a solution of the bromide, the bromine was twice 

 redistilled. This very pure preparation was run into carefully purified 

 ammoniacal sodic carbonate in a hard glass dish (for platinum might have 

 been slightly attacked) and the solution of sodic bromide was transferred 

 to platinum as soon as the operation was completed. The ammonia 

 used in this process had been prepared from ammonic sulphate purified 

 by treatment with nitric and sulphuric acids in the manner recommended 

 by Scott 4 to free it from amines. It was twice further distilled into 

 pure water. The reaction may be expressed as follows : 



3Na 2 C0 3 + 2NII 3 + 6Br = 6NaBr + N 2 + 3C0 2 + 3H 8 



3 This sample had been prepared by T. W. Richards for work on atomic 

 weights. 



* Journ. Chem. Soc, 79, 147 (1901). 



