The commonest impurity in commercial cyanide is carbonate of sodium or potassium. This may be tested for by dissolving, say, 2 grams in a little water and adding barium chloride. There may be formed a white precipitate of barium carbonate, which if filtered off, washed and treated with acid, will dissolve with effervescence. Cyanate may be tested for in the solution from which the barium carbonate has been filtered by adding a little soda and boiling; if cyanates are present they decompose, giving off ammonia (which may be tested for in the steam) and yielding a further precipitate of barium carbonate.[37] If the soda alone gave a further precipitate of barium carbonate, this may, perhaps, be due to the presence of bicarbonates. Alkaline sulphides may be present in small quantity in commercial cyanide. Their presence is shown at once when the sample is being tested for its strength in cyanide, inasmuch as the first few drops of silver nitrate solution produce at once a darkening of the liquor. A special test for sulphide may be made by adding a drop or two of solution of acetate of lead to four or five c.c. of soda solution and adding this to a clear solution of the suspected cyanide. This will cause a black precipitate or colour, if any sulphide is present. The cyanides of the heavier metals combine with the alkaline cyanides to form double cyanides. Some of these, ferrocyanide and ferricyanide of potassium for example, have such characteristic properties that the fact that they are cyanides may be overlooked. Others, such as potassium zinc cyanide (K_{2}ZnCy_{4}), have much less distinctiveness: they behave more or less as a mixture of two cyanides and are, moreover, so easily decomposed that it may be doubted if they can exist in dilute alkaline solutions. In reporting the cyanide strength of a cyanide liquor as equivalent to so much per cent. of potassium cyanide, there is a question as to whether the cyanide present in the form of any of these double cyanides should be taken into account. It must be remembered that the object of the assay is not to learn how much of the cyanide exists in the solution as actual potassium cyanide; reporting the strength in terms of this salt is a mere matter of convenience; what is really desired is to know how much of the cyanide present in the liquor is "free" or "available" for the purposes of dissolving gold. Every one is agreed as to the exclusion of such cyanides as the following: potassium ferrocyanide (K_{4}FeCy_{6}), potassium ferricyanide (K_{3}FeCy_{6}), potassium silver cyanide (KAgCy_{2}), and potassium aurocyanide (KAuCy_{2}); and the double cyanides with copper or nickel. But with cyanide liquors containing zinc the position is less satisfactory. One method of assay gives a lower proportion of cyanide when this metal is present; and the loss of available cyanide thus reported depends, though in a fitful and uncertain way, upon the quantity of zinc present. The