392 Canadian Record of Science. 



exist in the published statements as to the amount of caffeine in 

 raw coffee: — 



Robiquet 0.32 to 0.64 per cent. 



Liebig 0.23 to 0.46 " 



Zenneck 0.75 



Graham, Stenhouse & Campbell 0.86 to 1.00 " 



Dragendorf 0.99 to 1.22 



Squibb 1.00 to 1.03 " 



Bell 1.08 to 1.11 " 



Allen 0.50 to 2.00 « 



The discrepancy between the data given as applying to roasted 

 coffee is still greater, and in the Allgemeiner Kaffee Zeitung for 1884, 

 the amount of caffeine in roasted coffee is stated to range from 2.00 

 to 3.64 per cent. The first striking result obtained by Dr. Paul 

 and Mr. Cownley on carrying out a number of experiments with 

 several different samples of raw beans, was the very narrow range 

 within which the amount of caffeine appeared to vary. Instead of 

 being a varying amount, it was more nearly a constant quantity, 

 as follows : — 



Coorg 1.10 per cent. 



Guatemala 1.18 " 



Travancore 1.16 " 



Liberian 1.20 " 



Liberian 1.28 



The above determinations were all made withundried, raw coffee, 

 taken just as it came to hand and powdered. A difference in the 

 amount of water might therefore alter, to some extent, the percen- 

 tage of caffeine in the dry material, and a new set of determina- 

 tions were made, with 14 different berries, all carefully dried at 

 212°, when the amount obtained varied from 1.20 per cent, in Coorg 

 to 1.39 in Liberian ; average for the 14 samples 1.26 per cent. It is 

 evident from these results that the discordant statements hitherto 

 published in reference to the amount of caffeine in coffee must be 

 ascribed to defective methods of analysis, and that, in reality, the 

 determination of the amount of caffeine in a sample of coffee would 

 be one of the most conclusive data to rely upon in any question of 

 adulteration, as, in experimenting with roasted coffee, a similar 

 uniformity was found. 



Dr. Paul stated that there was no loss of caffeine in roasting 

 coffee. Roasters had stated that there was a considerable percen- 

 tage of loss, and, if printed authority was to be taken, the loss was 

 18 per cent. In his experiments he had failed to detect any loss. 



Ph. S. J. 



