378 Allen and Crenshaiv — Stokes Method for the 



(x—a) unless it is shaken with quartz. Here, to be sure, there 

 is no pyrite to compete with the inarcasite in the reduction of 

 the ferric salt, but it is not improbable that the sulphur which 

 is set free might act as a competitor to the marcasite, so that 

 less iron would dissolve in case the marcasite surface were 

 diminished by flocculation. Stokes found, it is true, that ordi- 

 nary powdered sulphur had scarcely any effect on the ferric 

 solution and we have confirmed his results. From this it 

 appears that such oxidation of the sulphur as normally occurs 

 in the Stokes reaction is probably contemporaneous with the 

 solution of the iron (see equations a and ft, p. 372). Still the 

 particles of separated sulphur are doubtless at first very small, 

 and if the surface of the marcasite should become much 

 reduced by flocculation, the slow reducing action of the sulphur 

 might cease to be negligible. At any rate, our value for 

 (x — a), the quantity of iron dissolved, was only 10*30 without 

 quartz, and this value rose to 10*57 and 10*58 when quartz was 

 used. 



(b) The necessary excess of the sulphide. — We have found 

 one other important source of error in the application of the 

 Stokes' reaction which must be discussed before our final 

 results are presented, viz., the quantity of sulphide which must 

 be used. It will be recalled that 250 cc standard ferric solution 

 is taken in each test, and that the quantity of the sulphide must 

 be such as to reduce completely all the iron (250 mg.) present. 

 If reproducible results are to be obtained, the excess of sul- 

 phide must be considerable. 0*75 g. gives virtually the same 

 results as 1*0 g. for all mixtures containing 25 per cent of 

 pyrite or more, but for other mixtures and for pure marcasite 

 as much as 1 g. should be used. It will be noted on p. 373 

 that it takes much more marcasite than pyrite to reduce a given 

 quantity of ferric iron. The ratio as determined by us is 

 2*37:1,* while Stokes found 2*26:1. The discussion of the rela- 

 tion between weight per cent and surface per cent given on p. 

 381 shows why the use of weights of the sulphides which are 

 but little in excess of the quantity actually decomposed by the 

 solution, would result in error, for in that case the surface 

 percentage, which is the all important matter, would vary 

 decidedly during the reaction, while in the ideal case it should 

 be constant. It may be seen from the table (Table II) that the 

 same results were obtained whether 1*5 g. or 1.0 g. marcasite 

 was taken, and the same is also nearly true for pyrite. Stokes 

 states that the fineness of the sulphide makes no appreciable 

 difference in the results. We have never found any systematic 



*This is the ratio between the constants for marcasite from Joplin, Mo., 

 and pvrite from Eoxbury, Conn. ; the ratio for the purest natural minerals is 

 about 2*5:1. 



