258 H. G. SMITH AND J. READ, 
three mojecules of water, calculated from the formula 
Cse6H30016. An analysis of the anhydrous substance gave 
© = 59°82 and H = 4°31 per cent.: CsgH30O16 requires C = 
60°18 and H = 4°45 per cent. A specimen which had been 
dried at 130° for 1$ hours gave C = 59°15 and H = 4°00 
per cent. 
The substance readily formed a crystalline potassium 
salt when treated with potassium acetate in hot alcoholic 
solution. After drying at 140°, it was found to contain 
K = 10°12 per cent. A specimen of the potassium salt 
prepared by mixing cold alcoholic solutions of the glucoside 
and potassium acetate and allowing the mixture to stand 
overnight contained K=10°17 per cent.: CsgHosOigKe re- 
quires K=9°85 per cent. The potassium salt was insoluble 
in boiling alcohol. 
An ammonium: salt was also formed by dissolving the 
glucoside in a large excess of concentrated ammonia solu- 
tion and keeping for some hours. This derivative was a 
crystalline powder, insoluble in boiling alcohol. It melted 
with evolution of ammonia at 125—126°, and gave an alka- 
line reaction when dissolved in water. It contained N=3'5 
per cent.: OseHes (NH 4). O16, or Cz¢6Hs6O16N 9, requires N = 
3°7 per cent. 
It thus appears that the molecule of the glucoside contains 
two fisetin nuclei united with one sugar nucleus, and that 
only one phenolic group in each fisetin rest is effective in 
giving rise to metallic derivatives. It may be recalled 
that quercetin, which has five phenolic groups in its mole- 
cule, yields a mono-potassium derivative, and many similar 
instances have been recorded. 
Acetylation of the glucoside was accomplished in the 
customary manner, the product being finally purified by 
crystallisation from alcohol., It was a colourless crystalline 

