GRAVIMETRIC ESTIMATION OF POTASSIUM 
549 
The main objections to the perchlorate method at present are 
the time 8 required and the slight solubility of potassium 
perchlorate in 95 per cent alcohol. The use of aniline perchlorate 
in place of perchloric acid shortens the process to such an ex- 
tent as to make it decidedly advantageous over all the processes 
in use for the separation and estimation of potassium. More- 
over it affords the best means for direct quantitative separation 
and estimation of sodium in the alcoholic filtrate. 
As to the solubility of KC10 4 in 95 per cent alcohol, the writ- 
er’s experience, in applying’ this method to the estimation of 
sodium and potassum in the ash of forage plants and animal 
carcasses, has been that some potassium chloride is occluded in 
the perchlorate. This is shown by the fact that higher results 
are obtained if the precipitate is allowed to stand for about two 
hours before filtration. Three series of four experiments each 
were conducted to establish this point. When the precipitant, 
dissolved in the proper amount of alcohol, was added all at once 
and filtered within fifteen minutes, decidedly lower results were 
obtained, but when the precipitant was added drop by drop with 
constant shaking and allowed to stand about two hours before 
filtration, the results were decidedly better. If it were simply 
a matter of solubility, no better results could be expected under 
the latter conditions. The potassium chloride which seems to 
adhere persistently to the perchlorate, being soluble in alcohol, 
is of course washed off gradually with the alcohol wash. Jarrell, 
in summing up his. experience with regard to the solubility of 
potassium perchlorate, does not state whether he obtained the 
theoretical yield when he prepared the potassium perchlorate 
from potassium chloride. 
The writer is inclined to believe that under the conditions 
Jarrell’s precipitates were contaminated with potassium chlor- 
ide. A careful observation of the table shown in this paper 
will illustrate the fact more clearly. In experiments 9, 10, 11 
and 12, 0.1 gram of KC1 was used instead of 0.2 gram. Exactly 
the same amount of water and alcohol were used and the same 
procedure followed, but the negative errors in this case are low 
enough to be within experimental error. These results clearly 
tend to show that when sufficient precautions are taken to pre- 
vent occlusion during the conversion of the chlorides into per- 
chlorate, a complete precipitation may be expected. 
