13 



An examination of the methods in use for determining potassium 

 shows that the ordinary platinum method is very uncertain, and 

 affected by errors that make it undesirable apart from considerations 

 of the growing cost of platinum. 



A careful study was therefore made of the perchloric acid 

 method which has been widely adopted in Germany. The details 

 as now used in our work are as follows : — The solution of chlorides 

 is evaporated to dryness in a porcelain dish and ignited for about 15 

 minutes at a dull red heat so as to throw out iron, etc., as oxides, as 

 in Neubauer's simplified method (Landw. Versuchs. Stat. 1905, 63, 

 141). If sulphates are present in large amount, 5 to 10 c.c. of 

 saturated barium hydroxide solution are first added to precipitate 

 SO4. The soluble alkali salts are then extracted with boiling water 

 as completely as possible, the iron oxide residue being broken up 

 with a glass rod during the extraction. The aqueous extract is filtered 

 into a glass evaporating dish (Si ins. diam.), and the process is usually 

 complete when the dish is full. The aqueous extract should be quite 

 colourless and free from iron (if not, this shows that the ignitionwas not 

 carried on long enough). It is now treated with 2'5 c.c. of perchloric 

 acid solution, sp. gr. IT 25 (20%) and evaporated nearly to dryness 

 on a sand bath not too strongly heated ; the evaporation must be 

 carried to the point of vigorous evolution of heavy white fumes of 

 perchloric acid. The soluble perchlorates are now taken up by 

 stirring with 20 c.c. of 95-967o alcohol and, after settling, the clear 

 solution is poured off through a 9 cm. filter paper which has been dried 

 to constant weight at 100" in a stoppered weighing bottle (li in. 

 diam. x 2 in.). 10 c.c. of 95% alcohol saturated with potassium 

 perchlorate are now added and the insoluble potassium perchlorate 

 transferred as completely as possible to the weighed filter paper. 

 The last traces of precipitate are washed into the filter paper with 

 another 20 or 30 c.c. of the alcohol containing perchlorate, and finally 

 the perchloric acid itself is washed out of the filter as completely as 

 possible. For this purpose the washings are tested, until quite free 

 from acidity, with sensitive litmus paper. Care must be taken that 

 the top edge of the filter paper is washed well. The freedom of the 

 filter paper from perchloric acid is shown by its not blackening 

 during the subsequent drying in the oven at 100°. The use of a 

 glass dish for the evaporation of the solution in the early part of the 

 process greatly simplifies the complete removal of the last traces of 

 perchlorate precipitate to the filter paper, as these last traces are 

 then plainly visible. In washing, a total quantity of 120-150 c.c. of 

 95% alcohol can be safely used without causing any perceptible loss 

 of potassium perchlorate, although less usually suffices unless much 

 NaCl is present. After washing, the filter paper and precipitate 

 are dried in a steam oven for about 20 minutes, whilst still in the 

 funnel, the filter paper plus precipitate is then transferred to its 

 weighing bottle and the drying completed until the weight is constant. 

 1 mgrm. KCIO4 =0*3401 mgrm. K^O. 



A Gooch crucible or Soxhlet tube can also be employed for 

 collecting the potassium perchlorate, but if this is used, care must 

 be taken that the asbestos layer is sufficiently thick to prevent the 

 finely divided potassium perchlorate from passing through. With a 

 layer ^ in. thick perfectly accurate results can be obtained. In 



