66 PROCEEDINGS OP THE AMERICAN ACADEMY. 



ing. The tubes containing the boat and weighing bottle were then 

 thoroughly washed out with a current of air dried in an apparatus simi- 

 lar to that used for drying the hydrochloric acid gas, as previously 

 described. After it was certain that all of the acid gas had been dis- 

 placed, and while the current of air was passing rapidly to prevent any 

 diffusion of moist air back into the apparatus, the bulbs were removed 

 fi'om the farther end of the ignition tube, and the boat was pushed into 

 the bottle in the manner already described. The boat itself remained 

 constant in weight during these operations, showing that the magnesic 

 chloride had not acted upon it. 



After weighing, the boat and its contents were placed in a large glass- 

 stoppered Erlenmeyer flask, and the magnesic chloride was dissolved in 

 pure water. The chlorine was precipitated with a dilute solution of 

 argentic nitrate ; * and after a thorough shaking the whole was allowed 

 to stand in the dark over night. The argentic chloride was washed by 

 decantation a number of times, with vigorous shaking, and was finally 

 collected upon a Gooch crucible in the usual manner. The precipitate 

 was dried from five to ten hours in an oven, carefully protected from 

 dust and dirt, and weighed. After weighing, the cake of jirecipitate, 

 together with some adherent asbestos, was removed to a tared porcelain 

 crucible and heated uutil it bes;an to fuse. The crucible was ajjain 

 weighed, and the loss of weight, if any, noted, and subtracted from the 

 weiu;ht of the Gooch crucible and contents. The filtrate, containing a 

 little dissolved argentic chloride, was evaporated down to small bulk 

 and filtered through a very small filter ; and the weight of the precip- 

 itate was added to the weight of the first portion. In some cases the 

 small amount of argentic chloride present was determined with the 

 nephelometer. t 



The wash water from the precipitate collected on the Gooch crucible 

 was also run through a small filter to make sure that no asbestos had 

 been carried away from the crucible in the process of washing ; and this 

 correction, when appreciable, was applied in the appropriate place. 



The washing and filtration were both performed in dim orange light, 

 which had been suitably tested as to its non-actinic properties. Even 

 after fusing the argentic chloride was almost colorless, showing that only 

 unessential traces had been decomposed by the light. 



The result of the first series of five experiments is given below. These 



* This solution contained never more than one per cent of silver, 

 t See Tliese Proceedings, XXX. 385. 



