18 ROYAL SOCIETY OF CANADA 



showed the same phenomenon with silver chloride as before. It was 

 concluded therefore that sufficient calcium sulphate goes into solution 

 to produce this effect. Since barium sulphate is almost completely in- 

 soluble, it was thought that by neutralizing the solution with barium 

 carbonate better results might be secured. In the presence of the pre- 

 cipitate of barium sulphate the titration was still unsatisfactory, but 

 "when the solids were removed by filtration a finely divided precipitate of 

 silver chloride was produced and a sharp end point secured. 



The difficulty of filtering off the barium sulphate, however, and the 

 small losses of potassium chloride involved, rendered the process other- 

 .wise unsatisfactory, and it was necessary, therefore, to follow the orig- 

 inal plan of using calcium carbonate. By working with known quanti- 

 ties of chlorate in test cases, it was found possible to secure results in- 

 volving an error of less one per cent, and in order to reduce this error 

 as much as possible, duplicate tests for the amount of chlorate were made 

 in all cases, and two readings for each duplicate, the second being made 

 after the addition of a known quantity of sodium chloride solution. The 

 average of the four readings so obtained was used in subsequent calcula- 

 tions. 



Thus in one case: — 



Titration 1. (a) AgNO^ (corrected) 5.63''« 

 (b) " " 5.64^'' 



Titration 2. (a) AgNO^ (corrected) 5.59«« 



(b) " " 5.62^^ 



Average reading 5.62^* 



A number of sulphates were tested and all showed the power of co- 

 agulating silver chloride. Sodium phosphate was found to act in the 

 same manner. 



The amount of chlorate changed to perchlorate was obtained by 

 difference and the weight of oxygen necessary for the formation of the 

 perchlorate was calculated. No determinations of the alkalinity of the 

 residue were made. 



Since the total oxygen from the decomposed chlorate either escapes 

 as free gas or combines with other chlorate molecules to form perchlorate, 

 there should be a correspondence between the sum of the loss of weight 

 of the bulb and the weight of oxygen required for j)erchlorate, on the 

 one hand, and on the other, the weight of oxygen from the decomposed 

 chlorate. In the experiments it was found that the latter was always 

 slightly in excess, except in a few cases at high temperatures where a 

 small quantity of the salts had been carried over mechanically. 



