and qf some of its Compounds. 119 



The analysis of cafFein, as well as that of the platinum 

 compounds, agree so perfectly with the numbers of Professor 

 Liebig's formula, that there can be no doubt about its accu- 

 racy. 



Assuming 1 equiv. of oxygen less in the equivalent of caf- 

 fein, as is proposed by M. Pay en, the platinum compound 

 should contain not less than 24-46 per cent, of carbon and 

 25-12 of platinum. Now three determinations by Dr. Sten- 

 house, and six which I have made, never gave more than 24-64 

 per cent., that is, 0-6 per cent, less of platinum. 



Not satisfied, however, with these proofs, I have tried to 

 find some other compounds by which the atomic weight of 

 cafFein could be determined with equal accuracy. 



In what follows a description of several new double salts 

 of cafFein will be given, the analyses of which correspond 

 equally well with the original formula of this substance. 



Caffein and Nitrate of Silver. 



This compound is obtained when a solution of nitrate of 

 silver is added in excess to an aqueous or alcoholic solution 

 of cafFein. If the solutions are concentrated it falls down in 

 white hemispherical nodules, which adhere firmly to the side 

 of the vessel. 



When washed with water and crystallized from alcohol it is 

 absolutely pure. This compound is indistinctly crystalline, 

 of a perfectly white colour, and if dry undergoes no change 

 when exposed to light, but if moist acquires a purplish hue. 

 It is very soluble in hot M'ater and alcohol, sparingly soluble 

 in cold, and may be boiled in either solvent without vmder- 

 going decomposition. It loses no weight in the water-bath, 

 but at a higher temperature it is decomposed, cafFein sublimes, 

 and metallic silver is left. 



Analysis. — When burnt with chromate of lead — 



I. 0-4514 grm. of substance gave 0-4345 grm. of carbonic 

 acid, and 0'1162 grm. of water. 



