OP ARTS AND SCIENCES. 6 



4. Monoclinic and triclinic forms predominate very largely, but 

 owing to rapid efflorescence it -is very difficult to make good measure- 

 ments of crystals. The resemblance between the forms of different 

 compounds is frequently very close. 



5. Many salts are decomposed by boiling, or even by hot, water, 

 yielding two or more different salts in solution. These usually re- 

 combine in the act of crystallization or on cooling the solution, but the 

 reactions of hot and cold solutions are often, as I shall show, very 

 different. 



In determining the tungstic oxide in these compounds, I have em- 

 ployed the method of Berzelius with mercurous nitrate almost exclu- 

 sively, but I have modified the process slightly so as to gain materially 

 in accuracy. To the hot solution of the tungstate mercurous nitrate is 

 added until in small excess. Mercuric oxide, prepared by precipitat- 

 ing the chloride by sodic hydrate, is then added until the yellow mer- 

 curous tungstate takes a reddish hue which is persistent after boiling. 

 If the solution is boiled before filtering it clears rapidly, and the pre- 

 cipitate becomes rather more compact. The filtration and washing 

 are then very easy and expeditious. The precipitate must be ignited 

 as long as it loses weight. By this process Dr. Gooch, my assistant, 

 obtained results which in two successive analyses of the same prejia- 

 ration rarely differed by 0.1%. The water determinations were al- 

 ways made by simple ignition. They almost invariably agree within a 

 few hundredths. In the greater number of cases the alkaline base was 

 determined from the loss, as the results obtained in this manner are 

 far more accurate than those yielded by the direct method. But in 

 some doubtful salts the alkali was determined directly. Ammonia 

 was always estimated by boiling the compound with an excess of 

 sodic hydrate, collecting the ammonia in chlorhydric acid, and weigh- 

 ing it as chloride. 



The ouly objection to the method of determining tungstic oxide 

 above given is, that the precipitate of mercurous tungstate is rather 

 voluminous, so that it is necessary to work with quantities of alkaline 

 tungstate not mucli exceeding one gramme in weight. I endeavored 

 to overcome this difficulty by omitting the mercuric oxide and evapo- 

 rating the solution and precipitate, after the addition of mercurous 

 nitrate, to perfect dryness in a water bath, continuing the heat until 

 all the free nitric acid was expelled. This method gave results which 

 corresponded very closely with those obtained by the use of mercuric 

 oxide to neutralize the free nitric acid, and in consequence of the 

 extremely compact form of the dry mercurous tungstate, permitted the 



