204 Transactions of the Boyal Society of South Africa. 
of any colour at all — this is probably due to the gold being rendered 
unavailable, by conversion into KAu(N02)2, analogous to aurocyanide, in 
which the gold is part of the anion — it being well known that SnCl^ 
has no action on gold compounds of this nature. Excess of caustic alkali 
appears to behave similarly by converting the gold into KAuOg, which 
does not react with stannous compounds. Iodine solution followed by 
strong acid and SnCla gives a good purple. 
Some experiments were tried to get the purple shade without the 
presence of tin. Addition of hydrazine sulphate to the faintly acid gold 
solution has no immediate effect, but a blue-purple shade develops in 
half a minute or so. Hydrazine hydrate (or its sulphate followed by 
alkali) gives the same blue-purple shade immediately ; this shows that 
the tin in purple of Cassius is merely a vehicle for the finely-divided gold 
and does not enter into chemical combination with it. 
If aqua regia or chlorine water be added to the very dilute gold solution, 
and then hydrazine sulphate, the action is much inhibited, but eventually 
a pale pink coloration appears. This is evidently the shade produced by 
gold in its finest state of division, as was also proved by Bredig in his 
electrically-prepared colloidal gold solutions. The purple shade is pro- 
duced by larger particles and the blue shade by still larger ones. Addition 
of salts to the pink solutions causes coagulation and change of colour 
through purple to blue and grey ; conversely, when a gold leaf is thinned 
by cyanide its transmission colour changes from green to violet. The 
pinkest shade is obtained by still slower precipitation, e.g., by adding an 
alkaline solution of glycerol to HAuCl^ — but however slow, the shade is 
never yellow-brown like the SnClz product, which latter is therefore 
unique and not due simply to colloidal gold. It would be interesting to 
apply the ultramicroscope to this problem (if there were one in S.A. !). 
The yellow substance may possibly be colloidal AuCl, and the effect of 
strong HCl in slowly turning it pink may be due to its keeping it in solu- 
tion and enabling it to be reduced to metallic gold. Curiously enough, 
Johannesburg tap-water when added to HAuCl^ generally produces a 
brown coloration of similar nature from the action of contained organic 
matter. 
Neither hydrogen peroxide nor nitrous acid reduce very dilute HAuCl^ 
solutions, and hydroxylamine acts very slowly unless alkaline. 
Good imitations of solid purple of Cassius are obtainable by adding 
MgCl^ or ZnCL to HAuCl^ solutions and then adding caustic soda contain- 
ing a trace of hydrazine. Alumina on the contrary gives a dirty grey and 
BaSO^ does not combine at all. 
