liquid is then poured into a flask, together with the rinsings, 
and oxidised with permanganate to a very slight excess, and 
then filtered so as to separate any precipitate, and also to 
destroy the excess of permanganate. The fluid thus obtained 
is next tested as before by adding the whole or a known 
part of it to one of the cylinders containing the ferrocyanide. 
When the water after being filtered has still a cloudy 
appearance, as is the case wflth sewage and polluted rivers, 
&c,, a known quantity of the filtered water must be evapor- 
ated to dryness and ignited, the residue dissolved in a small 
quantity of hydrochloric acid and filtered, washed with 
water, and the free acid in the filtrate as nearly as possible 
neutralized with ammonia and then Icc. of sulphuric acid, 
after which oxidised with permanganate, then filtered, if 
requisite, to destroy excess of permanganate, and the iron 
estimated as before. A green colour may be sometimes 
obtained instead of the pure blue ; this is owing to a slight 
quantity of unreduced permanganate being present; this, 
however, is of no consequence, as with a little practice the 
green tint may be compared with the blue and correct 
results obtained; still the comparison maybe rendered easier 
by adding 1 to 2 drops of permanganate to the cylinder to 
which the standard iron is run, and which by this means 
will also assume a green tint. Experiments were made 
with reference to this point and it was found that the 
presence of not more than a few drops of unreduced per- 
manganate has little or no effect on the results obtained, the 
only consequence being the change of tint but not of depth 
of colour. 
Potassium permanganate is employed as the oxidiser in- 
stead of the nitric acid, because (1) The oxidation is per- 
formed much quicker than it would be if nitric acid were 
used, and in the latter case the liquid would have to be 
heated. (2) It can be added to exactly the right point, 
which could not easily be done with nitric acid. (8) An 
excess of the latter is very detrimental to the accuracy of 
the method, for, from experiments made in relation to this 
point, it was found that when the amount of free acid pre- 
