84 



Now my recent experience has been that if the extracted residue be shaken with 

 water and the latter further extracted, an additional quantity of caffein, sometimes 

 equivalent to 10 per cent of the whole, is thus obtained. 



Our experience has been a similar one. To illustrate by a specific instance, the 

 residue from twenty hours' extraction with chloroform was shaken with water, filtered, 

 and the aqueous solution extracted four times with chloroform, One-half the chloro- 

 form extract was tested for caffein, and gave positive tests with Wagner's reagent and by 

 the "murexid" test. The other portion showed by a Kjeldahl determination 0.0035 

 gram of caffein, corresponding on the whole sample to about 10 per cent of the amount 

 present. 



Gorter finds that a considerable proportion of the caffein in coffee is present as a 

 double salt, the potassium caffein chlorogenate, from which the caffein is extracted 

 by dry chloroform only with great difficulty. Whether or not this be the cause of the 

 incomplete extraction, it is evident that the official method needs revision. 



Preliminary experiments with the Gomberg method showed it to be practicable for 

 the small amount of caffein (approximately 20 mg) that would be present in the 

 filtrate from the lead caffetannate, providing the volume of solution were not over 25 

 cc. On account of the slight solubility of the caffein periodid in the wash water it 

 was found best in working with this small amount to suck the precipitate as dry as 

 possible on the gooch filter and not to wash it. Determinations made on 20 mg of 

 caffein in 25 cc of water in this way gave from 98 to 99 per cent of the caffein present 

 Numerous experiments made on the filtrate from the lead caffetannate precipitate by 

 precipitating the lead with hydrogen sulphid and evaporating the filtrate gave fairly 

 concordant results, which were uniformly lower than those given on the same coffee 

 by the other methods for caffein. It was observed that the variations in amount of 

 caffein as determined in this way corresponded roughly with the variations in amount 

 of caffetannic acid as found by the Krug method. Whether these variations and low 

 results are due to incomplete extraction of caffein by the process of digestion employed 

 in the Krug method is a matter which we expect to investigate further. 



Gb'rter's method was not given a thorough trial. As far as the work goes it has been 

 -uti-factory, and the method is worthy of further trial by the association. It reads 

 briefly as follows: 



Eleven grams of the finely powdered coffee are moistened with 3 cc of water and 

 after standing a half hour extracted for three hours in a Soxhlet extractor with chloro- 

 form. The extract is evaporated, the residue of fat and caffein treated with hot water, 

 filtered through a cotton plug, and washed with hot water. The filtrate and washings 

 are made up to 55 cc, 50 cc pipetted off and extracted four times with chloroform. 

 This chloroform extract is evaporated in a tared flask and the caffein dried at 100 

 and weighed. 



In the determinations made it has never been possible to weigh the caffein directly 

 on account of impurities, the caffein having been calculated from a determination of 

 nitrogen in each case. From the work done there seems to be a strong probability that 

 a combination of the Gomberg and Gorter methods will prove to be the best and most 

 convenient process for determining caffein in coffee . 



