28 report — 1877. 



In experiment 1 an excess of alkali was employed, and the liquid was 

 then titrated back with sulphuric acid. It was hoped in this way to ensure 

 the complete and speedy decomposition of the silicofluoride ; but the end of 

 the reaction was very dim cult to read, perhaps owing to the formation of 

 silicate. It was also found to be no advantage to add the acid in sensible 

 excess and again titrate with alkali. In some cases decinormal alkali was 

 employed towards the conclusion of the titration, but the end of the reaction 

 was not sufficiently defined to make the precaution valuable. No. 2 can 

 scarcely be considered a test experiment, for the precipitated silicofluoride 

 was dried on the filter and then scraped off. 



The next three experiments were made on about 1-5 gramme (twice the 

 former quantity) of potassium chloride, the precipitated silicofloridc being 

 dried on the filter, scraped off, and weighed. 



Expt. Wt. of Sola. Wt. of Precipitate. = KC1 = KC1 found per 



100 parts taken. 



7 15-0525 2-2150 1-5018 99-77 



S 15-0475 2-2115 1-4994 99-64 



9 15-0365 2-2070 1-4964 99-52 



Theso results do not show any great departure from tho truth, especially 

 as traces of the precipitate probably adhered to the filter and were thus 

 lost. The manipulation was very easy, filtration occurring rapidly, and the 

 precipitate being easily washed, dried, and separated from tbe filter. 



After weighing, the precipitates obtained in the last experiments were 

 suspended in boiling water and titrated with normal alkali, with the follow- 

 ing results : — 



j; f Wt. ofppt. C. c normal Kj,SiP 6 found 



J ' i ' taken. alkali used. per 100 parte taken. 



7 A.... 2-2150 39-60 P 98-4 



8 A.... 2-2115 31-60 8 98-6 



9 A.... 2-2070 40-10 S 99-95 



In the last experiment the titration was slightly overdone. It appears, 

 therefore, that tho volumetric method gives results sensibly below the truth. 

 Probably the error was greater in the last three experiments owing to the 

 precipitates baring been dried, and thus reacting less readily with alkali 

 than the undried silicofluoride. 



In these, as in all other experiments, the alkali was added very slowly 

 towards the end of the reaction, and the liquid was well boiled after each 

 addition. 



Three more experiments by direct titration of the silicofluoride with 

 alkali gave the following results : — 



jl^ Wt. of soln. C. c. of normal _ ,, pl KC1 found for 

 '*■"• taken. alkali used. " *^ L 100 parts taken. 



10.. ..15-0525 40-10 P 1-4956 99-4 

 n ... . 15-0475 39-90 S 1-4881 98-9 

 12.. ..15-0365 40-10 S 1-4956 99-4 



In these last experiments the large quantity of silica produced rendered 

 the end of the reaction difficult to observe. In fact the want of sharpness 

 in tho termination of the reaction is a serious defect of the process. A 

 porcelain basin was found preferable to a beaker for conducting the titration. 



Although in the above experiments the volume of alkali "used was read 

 to ^L of a cubic centimetre, the end of the reaction could not be defined 



