192 Gooch and Austin — Constitution of the 



made to determine the silver chloride precipitated from the 

 solution in nitric acid of an unignited precipitate thrown down 

 by microcosmic salt in presence of sixty grams of ammonium 

 chloride, confirms this conclusion : the precipitate was unweigh- 

 able. If ammonium chloride present in the solution to so 

 great an amount is not included in the precipitate in signifi- 

 cant quantity, it would seem to be unnatural that the micro- 

 cosmic salt should be included mechanically in any very great 

 amount. But unless the microcosmic salt was mechanically 

 included, the increase in weight must be due to the chemical 

 influence of the reagents — that is, to the production of a phos- 

 phate rich in ammonium and deficient in magnesium. Ber 

 zelius* recognized the existence of such a phosphate of 

 magnesium ; but Wach,f in following the work of Berzelius, 

 failed to find it. No attention seems to have been given since 

 to the existence of such a salt. It would be natural to expect 

 its formation, if ever, when the precipitating phosphate is in 

 excess and ammonium salts are present in abundance, with free 

 ammonia. Obviously the natural effects of all these reagents 

 would be toward the production of a salt holding more ammonia 

 and more phosphoric pentoxide for a given amount of mag- 

 nesium. The results of the table seem to point strongly to 

 such tendencies, and, by inference, toward the existence of 

 such a compound. Thus in experiments (11) and (12), in 

 which the greater part of this excess of microcosmic salt was 

 removed by decantation before the second precipitation, while 

 no ammonium chloride was present excepting the small amount 

 made by the solution and reprecipitation of the first precipi- 

 tate, the error is practically nothing. In experiments (13) and 

 (14), (15) and (16), (IT) and (18), all similar to (11) and (12) 

 excepting that ammonium chloride was present, the average 

 errors — +0*0036 grm. in terms of magnesium phosphate, 

 0*0064 grm., +0*0074, respectively — increase as the ammonium 

 chloride is increased in the final precipitation. Jn experiments 

 (19) and (20), in which the ammonium chloride amounted to 

 sixty grams in the first precipitation and to five grams in the 

 second in addition to the amount that would naturally remain 

 after decanting the strong solution of the former precipita- 

 tion, the similarity of this error (+0*0082 in the mean) to 

 that of the experiments in which smaller amounts of the 

 ammonium chloride were used throughout goes to show that 

 only the amount of ammonium salt present in the final pre- 

 cipitation counts. Further, a comparison of corresponding 

 experiments of A and B shows very plainly that the treatment 

 which involves the removal of the large part of the micro- 



* Berzelius, Jahresbericht 3 Jahrgang (1824) iibersetzt von C. G. Gmelin s. 92v 

 fSchweigger, 1830, Band 29, s. 265. 





