RICHARDS. — ATOMIC WEIGHT OF BARIUM. 81 



serration of the point at which a drop of one solution or the other 

 absolutely ceases to produce a precipitate is a very trying matter. 

 The physical condition of the observer, and his practice in observation 

 of this sort, have much to do with his determination of the presence 

 or absence of a faint cloud at the top of the vessel under observation. 

 The phenomenon is one which gradually fades away as more and 

 more silver or hydrochloric acid is added, the last opalescence ap- 

 pearing only after some time. Often when no cloud is visible at the 

 end of an hour, a very evident one is to be seen in two hours. The 

 amount of light at hand is also an important factor in the determina- 

 tion. Temperature produces a very important effect on the solubility 

 of argentic chloride ; and this effect is tripled when it is referred to 

 the solution which is being added. The results are thus far less satis- 

 factory than similar results obtained with the bromide. 



For preliminary experiments the method used by Stas in his early 

 analyses * was adopted. The method consisted in the addition of an 

 excess of silver to the solution of baric chloride, a standard solution 

 of hydrochloric acid being used to determine this excess. At first the 

 atomic weight thus obtained corresponded very closely to that ob- 

 tained by the last ratio, but as the work proceeded the apparent 

 atomic weight of barium steadily rose. The reason for this anomaly 

 was not clear for a long time ; but in the end it was evident that the 

 apparent accuracy of the first results had been due to the mutual 

 elimination of two opposite errors. The occlusion of argentic nitrate 

 by argentic chloride tended to require too much silver, and the method 

 of titration required almost equally too much hydrochloric acid. When 

 more care was taken with regard to the dilution of the argentic nitrate, 

 and as the observer became more and more practised in the delicate 

 observation of the end point, the first error steadily decreased, and 

 the second steadily increased. Hence, they no longer counterbalanced 

 one another. A long time was spent in finding that this first method, 

 used by Stas in his early analyses of lithium, sodium, potassium, and 

 ammonium chlorides, was quite false. Even relative results by this 

 method, such as Stas's comparison of the different varieties of silver, 

 can have no important value, because it is wellnigh impossible to 

 make all the conditions absolutely constant. It is almost needless 

 to say that these remarks apply only to this early method, and not to 

 Stas's later work. Naturally he made errors at first. It is much to 

 be regretted, however, that he did not point out mrre clearly after- 



* " Untersuchungen," etc. Leipzig, 1867. 

 vol. xxix. (n. s. xxi.) 6 



