clusion is not in harmony with that of Lendrich and Nottbohm. Results of their investigation of the action of permanganate on caffein in water solution indicate no action at room temperature for 30 minutes. However, the procedure adopted by them differed from the one used in this investigation, as they calculated their caffein recovery on the nitrogen determination only and did not weigh the dry residue. It is possible that the action of permanganate is such that the nitrogen content is not changed, in which case the caffein calculated from nitrogen would indicate a theoretical recovery. Moreover, the loss reported in this article is of a magnitude such as Lendrich and Nottbohm, judging from their results, considered within the limits of error of the nitrogen determination.

Although the error introduced by the purification of the caffein at room temperature is well within the experimental error of the method, it seemed advisable to try the modification providing for the purification at the temperature of the ice bath, as at this temperature no loss was detected. The value of this modification will be discussed with the discussion of the collaborative results.

EXTRACTION OF THE CAFFEIN FROM THE COFFEE.

The method as adopted in 1917 recommends the extraction of 10 grams of coffee after the addition of 10 grams of 10 per cent ammonium hydroxid with 200 grams of chloroform by shaking for 30 minutes with subsequent filtration and weighing of 150 grams of the filtrate. Fendler and Stüber give no data to show that the 150-gram aliquot is equivalent to 7.5 grams of the sample. To accomplish the weighing of 150 grams of the filtrate without evaporation and coincident concentration of the filtrate is tedious. Evaporation of the filtrate, of course, would cause high results. It appeared to be simpler and more accurate to catch and weigh the filtrate in a flask previously weighed with stopper. In order to test the accuracy of this procedure, two portions of caffein of 0.2 gram each were dissolved in 200 grams of chloroform and the entire solution in each case was poured, after chilling in ice, on a 24 cm. folded filter and covered with a watch glass, the filtrate being collected in a tared flask surrounded by ice. After the continuous running of the filtrate ceased, the flask was stoppered and weighed. The caffein, after evaporation of the chloroform, was dried and weighed. The caffein as weighed and as calculated from the weight of the chloroform aliquot is given in Table 3 with the difference between the theoretical and practical results. Also, two samples of coffee were extracted with 200 grams of chloroform and the filtrate collected as just described. The residue, with filter paper in each case, was transferred to a Soxhlet extractor, care being exercised to transfer all of the excess of chloroform which was not

collected as filtrate. The residue was then extracted for 12 hours with chloroform. The caffein in the weighed filtrate and in the extract was then determined as directed in the method. Results were obtained which gave the total caffein in the coffee, the portion in the filtrate and the portion in the extract. From the weight of the aliquot filtrate the proportion of caffein which should be in the filtrate could be calculated and compared with the quantity actually found. As the residues were of doubtful purity, nitrogen was determined by the Kjeldahl-GunningArnold method and the caffein calculated. The results are given in Table 4.

TABLE 3.

Recovery of caffein from chloroform solution by the modified method of collecting the chloroform filtrate.

Recovery of caffein from coffee by the modified method of collecting the chloroform filtrate.

Consideration of the results given in Tables 3 and 4 leads to the conclusion that the method of extraction, filtration and manipulation of the filtrate, as given in the preceding paragraph, gives a representative aliquot of the total caffein present. Moreover, no appreciable error will be caused through concentration of the chloroform extract by evaporation if the procedure is carefully followed. The method was accordingly modified to provide this procedure.

COLLABORATIVE RESULTS.

Two samples of coffee, a roasted Bogata and a green Santos, were sent to the collaborators with a copy of the method modified as indicated. An explanatory sheet explaining the changes, with suggestions on manipulation, accompanied the method. The manner of filtration of the

excess of manganese dioxid after the purification of the caffein was changed to provide for filtration on a Gooch crucible with suction instead of on paper. This was found to be quicker and more satisfactory for washing the precipitate. Caution was requested with respect to having the hydrogen peroxid acetanilid free. This chemical is soluble in chloroform and also contains nitrogen. High results would be caused by the introduction of acetanilid with the hydrogen peroxid. If an acetanilidfree reagent can not be obtained, an extraction with chloroform should be made to remove the objectionable substance. The chloroform used in this investigation was found to leave considerable residue on evaporation. Distillation removed this objectionable feature.

These results show conclusively that the purity of the caffein when judged from the nitrogen determination is not equal to that of the caffein obtained in 19171 when the temperature of the purification with permanganate was not reduced with ice. One of the chief objects of the collaborative work this year was to determine if the modification calling for lowered temperature would produce a caffein of high purity. The error of 0.01 to 0.015 per cent caused by carrying out the purification at room temperature is within the error of the method itself and well within the personal equation as shown by the results of the collaborators. In view of this fact and of the relatively low purity of the caffein obtained by the cold process, it is believed advisable to retain the details of the method in this respect as adopted tentatively in 19171. However, no reason is apparent why the other modifications on the extraction and filtration of the extract and on drying the caffein should not be included to insure greater accuracy and ease of manipulation.

The method as presented to the association this year would then read as follows:

FENDLER-STÜBER METHOD FOR CAFFEIN IN COFFEE (MODIFIED).

Pulverize the coffee to pass without residue through a 1 mm. sieve. Treat a 10-gram sample with 10 grams of 10% ammonium hydroxid and 200 grams of chloroform in a glass-stoppered bottle, shake continuously by machine or hand for 30 minutes and chill in an ice bath. Pour the entire contents of the bottle on a 24 cm. folded filter, covering immediately with a watch glass. Collect the filtrate with the funnel resting directly in the neck of a flask (previously weighed with stopper) having the flask surrounded with ice. Stopper as soon as the solution ceases to run from the funnel in a continuous stream and weigh. Evaporate on the steam bath, removing the last chloroform with a current of air. Digest the residue with 80 cc. of hot water for 10 minutes on the steam bath, with frequent shaking, and let cool. Treat the solution with 20 cc. (for roasted) and 10 cc. (for unroasted) of 1% potassium permanganate and let stand 15 minutes at room temperature with occasional shaking. Add 2 cc. of 3% hydrogen peroxid (containing 1 cc. of glacial acetic acid in 100 cc.). If the liquid is still red or reddish add hydrogen peroxid, 1 cc. at a time, until the excess of potassium permanganate is destroyed. Place the flask on a steam bath for 15 minutes and add 0.5 cc. portions of hydrogen peroxid until the liquid ceases to become lighter. Cool and filter by suction through a Gooch crucible, washing with cold water. Transfer the filtrate to a separatory funnel and extract six times with 25 cc. portions of chloroform. Evaporate the combined chloroform extracts to a small volume, transfer to a tared beaker, finish evaporation, dry at 100°C. to constant weight (30 minutes is usually sufficient) and weigh the residue as caffein. The weight of the caffein, multiplied by 2000 and divided by the weight of the chloroform aliquot obtained from the first filtration, equals the percentage of caffein in the 10-gram sample. Test the purity of the residue by determining nitrogen and multiplying by the factor 3.464.

RECOMMENDATIONS.

It is recommended

1 J. Assoc. Official Agr. Chemists, 1920, 4: 211.

(1) That the Fendler-Stüber method, as modified, page 533, be continued as a tentative method.

(2) That further study be made of the method before it is adopted as official or superseded.

F. B. Power (Bureau of Chemistry, Washington, D. C.) called attention to work conducted by him and V. K. Chesnut, also of the Bureau of Chemistry, which resulted in an improved method for the quantitative determination of caffein in vegetable material, and also to ilex vomitoria as a native source of caffein.

REPORT ON TEA.

By E. M. BAILEY (Agricultural Experiment Station, New Haven, Conn.),

Referee.
CAFFEIN.

In 1915 the referee on tea and coffee confined his attention to methods for the determination of caffein. Data for this determination were presented by the Fuller, Stahlschmidt, and modified Stahlschmidt methods in case of tea, and by the Fuller, Gorter, and modified Stahlschmidt methods in case of coffee1. In 1916 the referee confined his attention to coffee, and in 1917 the report on tea referred only to microscopical methods as applied to tea examination.

In the work on coffee during the last two or three years particular attention has been given to a comparison of the Fendler-Stüber method with the modified Stahlschmidt method. This suggested to the present referee the advisability of comparing these two methods in the case of tea. Unfortunately, promised collaboration did not materialize, so that the data now presented are confined to results obtained in the writer's laboratory.

Twelve samples of the United States standard teas for 1918 to 1919, furnished through the courtesy of the Collector of Customs of the Port of New York, were used in the comparative trials. The results therefore are comprehensive so far as different types of teas are concerned.

The details of the modified Stahlschmidt method, as given in the present tentative method2, were followed. They are the same as those given by J. M. Bartlett3 (Agricultural Experiment Station, Orono, Me.), except that the initial boiling is for 2 instead of 3 hours, and the drying of the caffein residue is done at 75°C. instead of at 100°C. The FendlerStüber method as directed for coffee in 19174, was followed without modification.