ON THE ESTIMATION OF POTASH AND PHOSPHORIC ACID. 



19 



From these considerations it appears almost certain that the deficiency is 

 due to the solubility of the precipitate in platinum solution containing sodium 

 chloroplatinate. 



If a loss of about four milligrammes produces an error of -37 by Tatlock's 

 method, the discrepancy would be much greater by Frank's, in which a 

 smaller weight of the sample is employed. This fact, and the very strong 

 alcohol required, renders this process less satisfactory than that of Fresenius. 

 Why the latter process should give results in excess of the truth, even when 

 the modified method (II.) was used, seemed difficult to explain. With a 

 view of ascertaining the cause, three quantities of pure chloride of potassium, 

 with equal weights of chloride of sodium, were treated by method I. After 

 weighing, the precipitates were dissolved in hot water, 10 or 12 drops of 

 platinum solution added, and the process of evaporation, &e. repeated. The 

 following results were obtained : — 



Table VI. 



In these and in all previous experiments the precipitates were dried at 

 100° C. In the last edition of his ' Quantitative Analysis,' Fresenius directs 

 the precipitate to be dried at 130° O. To ascertain if this difference of treat- 

 ment was the cause of the error, some pure potassium chloroplatinate was 

 prepared by rapidly cooling a saturated solution of the salt in boiling water. 

 In this way it was obtained as a Sue crystalline powder. By the slow 

 evaporation of the mother liquor another sample was obtained in the form of 

 " scales." The products were dried for half au hour at 100° C, and 3 grammes 

 of each exposed to a higher temperature, with the following results : — 



Table VII. 



It will bo seen that no loss occurred on further drying at 100° C, and a 



very trilling loss at 130° 



After heating to 140° there was a slight change 



c2 



