RICHARDS. — ATOMIC WEIGHT OF BARIUM. 63 



at a dull red heat. The residue was dissolved in a platinum dish and 

 crystallized, and the crystals were dissolved in pure water. After 

 filtration the solution was acidified by hydrochloric acid and again 

 brought to crystallization. A portion of the salt thus formed (Sample 

 Il.a) was used for Analyses 19, 26, 30, and 36, while the remainder 

 was a^ain ignited in platinum, and was recrystallized twice more 

 after filtration and very faint acidification with hydrochloric acid. 

 The last mother liquor was evaporated to dryness in platinum, and 

 served for Analyses 23, 27, and 37 (Il.b). The purest crystals 

 were partly dried over a steam bath, powdered in an agate mortar, 

 and exposed to a moist atmosphere until constant in weight. These 

 crystals (Sample IT.c) served for Analyses 28, 29, 38, 39, 40, and 41. 

 An essentially similar preparation, containing somewhat more occluded 

 water, was used for Analyses 42 and 43. 



A third sample of baric chloride was prepared from a part of the 

 carbonate used for the preparation of baric bromide, and hydrochloric 

 acid which had been distilled twice in platinum. This was purified 

 in the usual fashion, and served for Analysis 24. 



These preparations each gave every evidence of being pure. After 

 most careful fractional separation of the greater part of the barium 

 no trace of calcium or strontium could be found, even in large samples, 

 by means of the spectroscope. The trace of sodium visible undoubt- 

 edly crept in during the fractional treatment used to prepare the 

 specimen for spectroscopic examination, for the salt itself before the 

 fractionation showed no more sodium than the surrounding air. The 

 effort was made wholly to free the salt from silica by means of re- 

 peated ignition and solution, and the use of platinum vessels ; but no 

 proof can be given that it did not contain one or two parts in a 

 hundred thousand of this impurity. The quantitative comparison of 

 these preparations which is made further on, furnished conclusive 

 proof of their similarity, at any rate. If the results are reduced to 

 the basis of 100.000 parts of argentic chloride, and arranged accord- 

 ing to the sample of baric chloride used, the following averages are 

 obtained.* 



* From these determinations all those made according to Stas's first and 

 second methods have been omitted, since these methods proved to be unsatis- 

 factory. Hence Preparation I.b is not represented in the table. The high 

 result from Sample Il.a (Experiments 26 and 36 are parts of the same analysis) 

 was due to a slight mechanical error. 



