76 The N.Z. Journal of Science and Technology. [Mar. 
asbestos layer is sufficiently thick to prevent the very finely divided potas¬ 
sium perchlorate from passing through. With a layer Jin. thick perfectly 
accurate results can be obtained. In rapid working, when a large number 
of analyses have to be made, the Gooch crucible is preferable to a filter- 
paper.” 
Mr. Davis, who points out that discrepancies in reputable English 
analysts’ results of the same sample are by no means common, and that 
in Germany the platinum method has been almost superseded by the 
perchlorate, concludes his paper with the following 
Summary. 
“ It is shown that whereas the platinum chloride method of estimation 
is uncertain and liable to give varying results, the perchlorate method is 
at once more simple in manipulation and more uniform and exact in its 
results. An improvement has been introduced which consists in washing 
the perchlorate precipitate with 95 per cent, alcohol saturated with potas¬ 
sium perchlorate, by means of which any error due to the solubility of the 
precipitate is obviated. This is of importance when dealing with small 
quantities of precipitate. The following other advantages may here be 
enumerated 
“ (1.) Economy: In view of the very large excess of platinum required 
when sodium salts are present, and the great cost of the metal at the present 
moment (£13 per ounce), this is a very, real advantage. The troublesome 
working-up of large quantities of platinum residues, with the attendant 
risks of poisoning, are avoided. 
“ (2.) The presence of barium, magnesium, and calcium chlorides and 
sodium phosphate is without prejudice to the method, and these salts need 
not be removed. Potassium sulphate can be estimated direct, using a 
sufficient excess of perchloric acid, without conversion into chloride by 
means of barium chloride ; if in the analysis of commercial sulphate of 
potash the sulphate is converted into chloride by the Stassfurt method, 
the exact precipitation of the sulphate is not imperative, as in using the 
platinum method. There can be excess of either potassium sulphate or 
barium chloride. 
“ (3.) All uncertainty such as exists as to what value shall be taken 
for the atomic weight of platinum is avoided. The calculation is made 
KC10 4 KC10 4 „ 
K 2 0 ’ KC1 5 &C ' 
In conclusion, the writer can assure others that he has examined the 
perchlorate method and quickly adopted it in his branch as being far 
superior to the platinum method for commercial purposes. The perchloric 
acid is somewhat difficult to obtain owing to the unwillingness of shippers to 
carry it. Two methods of preparing the acid from sodium or potassium 
chlorate and hydrochloric acid are given in Wiley’s Agricultural Analysis , 
vol. 2, “ Insecticides and Fertilizers,” one of which, the method of Kreider 
(p. 582), will make it easy for every analyst to make and keep a supply of 
this admirable yet unappreciated reagent. 
Postscript. 
Since the above paper was written Thin and Gumming* have rein¬ 
vestigated the method and have confirmed the conclusions of Davis. If 
a high degree of accuracy is required they, however, recommend as a 
washing solution alcohol saturated with potassium perchlorate. 
from the simple molecular ratios 
* Jour. Chem. Soc., vol. 107, pt. 1, 1915, p. 361. 
