12 PROCEEDINGS OF THE AMERICAN ACADEMY. 



cipitating that crystalline salt " silver emetic," discovered by Wall- 

 quist * and further studied by Professor Cooke.f 



However, if care is taken to have only a slight excess of silver 

 nitrate over the calculated amount, and to precipitate from dilute solu- 

 tions, the probable error is reduced to a minimum. 



Calcium was determined by igniting calcic oxalate precipitated in 

 the usual way, and weighing as calcic oxide. This method was very 

 satisfactory. Magnesium was weighed as magnesic pyrophosphate 

 (MgoPoO;) formed by the ignition of the ammonic magnesic phos- 

 phate {NH4MgP04) precipitated in the usual manner. 



The presence of tartaric acid evidently interferes to some extent 

 with the perfect working of this determination ; but as the per cent of 

 magnesium in the salt was so small the error was not prominent. 



The determination of the water of crystallization presented the first 

 serious difficulty, for, drying the salt in an air bath at 98°-100° 

 showed rapid decomposition. In fact, no temperature was found at 

 which the water would all go off and the salt remain undecomposed. 



On placing the salt in a vacuum desiccator the loss was considerably 

 greater than theory would require, till finally the residue amounted to 

 but about 24% of the whole amount of salt used. Even in an ordi- 

 nary desiccator there was a gradual loss in weight, and constancy was 

 only reached when the residue was about 24% of the total mass, as 

 before. At the suggestion of Professor Cooke the salt was intimately 

 mixed with a quantity of ignited oxide of magnesia (heavy) in a 

 crucible and heated to constant weight. In this operation it was 

 necessary to get the upper layer of magnesic oxide hot before heating 

 the lower part of the crucible containing the compound. Should any 

 antimony haloid be volatilized, it would be decomposed upon reaching 

 the hot layer of oxide of magnesium. This method served excellently 

 for the first salt, i. e. SbClg . CaClo . 8 HoO . The oxide of magnesia 

 seemed to retain the antimonious chloride readily, but when the same 

 process was applied to the analysis of the bromides it was found that 

 the magnesic oxide would not retain all of the bromide of antimony. 

 The escape of this salt was clearly observed by the color of the flame 

 held over the mouth of the crucible. 



After several trials, a modification of the ajiparatus described by 

 Jannasch and Locke t was decided upon as giving the most satisfac- 



* Gmelin, Handbook, Cavendish ed., X. 326. 



t These Proceednigs, XVII. 5. 



J Zeitschrift fiir Anorgan. Chemie, VI. 174. 



