Conduct a 11 cm, blue ribbon, S. and S. paper. Repeat this treatment until no more coarse material remains. After the ether has evaporated from the filter, transfer the fat-free residue to the mortar by means of a jet of cold water and rub to an even paste, filtering on the paper previously employed. Repeat this process until all sugar is removed. In the case of sweetened products the filtrate should measure at least 500 cc. the hydrolysis of the residue as directed for “Starch” under “VI. General Methods," 8 (a), page 53, Bul. 107, Rev., except that after neutralizing with sodium hydroxid, add 5 cc of basic lead acetate solution (prepared as directed under "VI. General Methods, " 6 (b), (1), page 40) before completing the volume to 250 cc. To 100 cc of the filtrate add 1 cc of 60 per cent sulphuric acid, filter off the lead sulphate and determine reducing matters in 25 cc of the filtrate as directed under "VI. General Methods, for Reducing Sugars, 7, (b), (2), page 49. Determine copper by the direct weighing of cuprous oxid, 7, (c), (6), page 53. SUGARS (SAMPLE B). Determine the lactose and sucrose as described by Dubois.a The amount of work requested was purposely made small in order that it should not prove burdensome, but in spite of this results were received from only four chemists, two on cocoa products and two on tea and coffee. TEA AND COFFEE. The results on tea and coffee are shown in the following table: Krug method for caffetannic acid.—(1) The water must be kept to constant volume during thirty-six hours' digestion. (2) Unless great care is used, the addition of lead acetate to the hot alcohol solution will cause violent ebullition and partial loss of contents. A safety tube helps to overcome this difficulty. (3) In determining the lead content of the caffetannate it is advisable to filter the caffetannate through a tared gooch. This will allow of digestion of contents in nitric acid and precipitation of the lead with sulphuric acid without using a filter paper, the carbon of which does not completely oxidize and produces a blackening of the lead sulphate. The final weighing of the sulphate should also be made in a gooch. a J. Amer. Chem. Soc., 1907, 29: 560. Caffein method.—It is suggested that the caffein iodin precipitate does not form immediately and that the low results are due to filtering and titrating the solution too quickly. Other work indicates that after the iodin is added the flask should be allowed to stand in an ice chest overnight before titrating. Much better results can be secured by Gomberg's original method for caffein, as given in the Journal of the American Chemical Society (1896, 18: 331), and modified as follows: Extract 2 grams some time with four portions of water, cool, and make to 1,000 cc. Treat 500 cc with 15 cc of saturated lead acetate solution, let settle, filter, remove lead with hydrogen sulphid, boil off excess of hydrogen sulphid, divide filtrate into two parts, concentrate each to 50 cc, add 0.2 cc of concentrated hydrochloric acid to one and 0.5 cc of acetic acid to the other, cool to 15° C., add 20 cc of tenth-normal iodid solution, stopper flask, and let stand in ice two hours, filter on a gooch. Caffein does not precipitate unless mineral acid is present, so the acetic acid portion shows if any other materials are present which would precipitate with the iodin solution. If any absorption of iodin is found in the acetic portion, it must be deducted from the titration containing mineral acid. The difference represents the iodin used up in the formation of the periodid of caffein: 1 cc of tenth-normal iodin equals 0.00485 gram of caffein. Using this method, 0.78 per cent of caffein was obtained from the coffee reported. Krauch method for extract in tea.-The bulk of sample (20 grams) makes complete removal of water-soluble substances almost impossible. The absorption of water by large filter paper and on surface of flask during weighing is also a serious objection to the method. If sample is ground, filter paper is clogged and filtration prevented. Doolittle and Woodruff method.-Care should be used to keep the entire sample in the boiling liquid during extraction or low results will be obtained. Any loss of water by evaporation should be replaced. NOTES BY REFEREE. The discrepancy in the results obtained by the two analysts with the Trillich and Göckel method is due principally to the fact that Mr. Woodruff used 5 cc of basic lead acetate in the precipitation instead of 1 cc, as prescribed in the method. Determinations made by the referee on the same sample gave 10.08 per cent where 2 cc of basic lead acetate was used and 12.04 per cent when 4 cc was used. The lower results obtained by Mr. Woodruff in the caffein estimation may have been due to the greater volume of solution in which the caffein periodid was precipitated, he using a volume of 100 to 150 cc, while the referee employed a volume of 20 cc. Experiments made by Mr. W. C. Taylor in the writer's laboratory have shown the necessity for concentrating the caffein solution to small bulk. The determinations made of extract in tea by the referee convinced him of the great superiority of the Doolittle and Woodruff modification over the Krauch method as regards convenience, time, and liability to error. COCOA PRODUCTS. The following results were obtained from the collaborating chemists on cocoa products: COMMENTS BY ANALYSTS. G. M. Bartlett: The conversion of the starch was carried out as outlined, no difficulty occurring in the procedure. The aliquot for precipitation was obtained as follows: After converting, neutralizing, and adding basic lead acetate the sample was made up to volume at about 35° C. To 100 cc at this temperature was added the 60 per cent of sulphuric acid, cooled so that the volume of liquid contracted to 100 cc. It was necessary to cool only to about 18° C. The sample for precipitation was taken when the liquid had contracted to the mark. Two determinations were made-one by precipitating by the Walker-Munson method (J. Amer. Chem. Soc., June, 1906,) and the other following the method in Bulletin 107. The latter gave 11.83 per cent of starch. In determining crude fiber the electric stove was used for boiling the 1.25 per cent sulphuric acid and caustic soda. There was but little frothing. The filter paper and crude fiber were dried at 100° C and over sulphuric acid. During weighing the filter paper gained in weight. I do not care for this method of getting the weight of the crude fiber, even though it is not to be ignited, and would prefer filtering on a weighed platinum gooch filter. In calculating the sugar in chocolate by Dubois's method a it seems illogical to multiply (a—b) by 1.05x (x equaling the volume obtained by dissolving sugar in 100 cc of water) rather than by 105 plus the increase in volume due to the solution of the sugar. This actually makes but little difference in the result, but the following statement of the formula seems preferable: In calcu Where increase of volume owing to the solution of the sugar in water. lating the lactose the complete formula reads: Per cent lactose-CX4×1.11×1.05x X1.264, where x=volume of solution when the sugar is dissolved in 100 cc. R. W. Hilts: The samples on arrival were immediately placed in glass-stoppered bottles. Before removing portions for analysis they were rubbed down to a coarse powder in a large porcelain mortar and mixed as well as possible. This was done quite rapidly, both to avoid possible changes in moisture content and to avoid formation of a pasty mass. Crude fiber: First filtration was made on closely woven linen in a 4-inch Büchner funnel with light suction. Second filtration was on a 11 cm B & A ashless filter paper without suction. Both filtrations were rapid and satisfactory. Starch: Results are multiplied by the factor 1.01 to correct for the dilution of 100 cc of the solution by the 1 cc of sulphuric acid. Sugars: The method of Dubois was followed exactly. It was necessary in extracting with water to break up with a glass rod the compact cake left after centrifuging the last time with gasoline. Inversions were made in the cold (50 ce+5 cc of hydrochloric acid, being allowed to stand over night. All volumes were adjusted at 20° and all polarizations were made in jacketed tube at exact temperatures. The actual polariscope readings (averages of four to five close readings) illustrate the very great influence that small differences in readings have upon the results, in these dilute solutions. In spite of this fact, the method seems to be satisfactory and convenient for judging milk chocolates. The methods are, in my opinion, in as simple a form as possible, and can not well be improved. RECOMMENDATIONS. In view of the small number of collaborators it is hardly possible for the referee to make any formal recommendations based on collaborative work. It is evident, however, that the study of certain of these methods should be continued by the association, especially the caffein estimation and the determination of sugars in chocolate. There would appear to be no reason why the determination of extract in tea as outlined by Doolittle and Woodruff should not be substituted for the cumbersome Krauch method. The experience of the referee on numerous samples of cocoa products suggests that the requirement for filtering and weighing the crude fiber on a a J. Amer. Chem. Soc., 1907, 29: 556; see also Bul. 107, Rev., p. 256. 6 paper should be omitted, as the determination can be made more conveniently on a gooch crucible as ordinarily used. Attention is also called to the accompanying paper involving some of Mr. W. C. Taylor's work on caffetannic acid and caffein. ESTIMATION OF CAFFETANNIC ACID AND CAFFEIN IN COFFEE. By A. G. WOODMAN and W. C. TAYLOR. In connection with an examination of the methods for coffee analysis the writers have made a study, in the limited time available, of the provisional methods for determining caffetannic acid and caffein, especially the former. CAFFETANNIC ACID. Experience has shown that with the directions as given at present it is practically impossible to obtain concordant results or a lead caffetannate of constant composition. It has been the general experience of those who have worked with the Krug method that it is tedious in the extreme, and, furthermore, that the composition of the so-called lead caffetannate obtained varies with the conditions of precipitation. It was our purpose to ascertain if possible the source of some of these difficulties. It was seen early in the work that variations in the amount of lead acetate used for precipitation gave variations in the proportion of lead caffetannate obtained, as well as in its content of lead. This is shown in the following table, in which the determinations were made on aliquot portions of a coffee infusion and varying amounts of saturated lead acetate were used, all other conditions being kept constant. Determination of caffetannic acid, using varying amounts of lead acetate. The averages of several results are stated in each case, although the results showed very considerable variation. While too much reliance can not be placed on these figures, owing to variations among themselves, they show the necessity of using a definite amount of lead acetate for the precipitation. Another source of error is the difficulty of washing the lead caffetannate free from lead acetate. Those who have attempted it know the tediousness and the difficulty of washing the precipitate on the filter. It is of course necessary to use alcohol of 90 per cent strength in washing, on account of the solubility of the precipitate in water or dilute alcohol. On the other hand, the lead acetate which is to be removed is only slightly soluble in 90 per cent alcohol. Hence it will be readily seen that it is practically impossible to wash the bulky precipitate on the filter. It is true, also, that when the wash water no longer reacts for lead the precipitate is not necessarily free from it, since owing to its character the wash water easily forms channels and does not wash it thoroughly. After numerous experiments, washing by the centrifugal machine was tried, giving several treatments with 90 per cent alcohol in the tubes of the centrifugal before trans ferring to filter paper. This method gave results which were in much closer agreement, as shown by the following results on the same sample: Per cent caffetannic acid..... ... 9.69 48.28 9.69 48.35 9.57 9.40 48.01 48.71 Tests made on a considerable quantity of the lead caffetannate washed in this way showed it to be free from fat and nitrogen. It would seem as if the long process of digestion with water and with alcohol prescribed by Krug could be materially shortened. In much of our work extracts of the coffee were prepared by the use of a shaking machine, shaking the sample for an hour with water and half an hour with alcohol. Results obtained in this way agree very well with those obtained by the official method of digestion, although there is evidence to show that neither method extracts all of the caffetannic acid. Regarding the vexed question of the composition of caffetannic acid, we seem not much nearer a settlement. The views previously held, which seem to lead to the formula for a di-glucosid, have been clearly set forth in Bulletin 105 by Mr. Howard. Lack of time prevented any extended investigation of this problem, but an endeavor was made to confirm the work of Cazeneuve and Haddon in regard to the di-glucosid formula for caffetannic acid. We were unsuccessful, however, in preparing more than traces of the osazone prepared by them, although carrying out the experiments exactly in the manner prescribed. In this connection the paper recently published by Görter a is of interest, in which the correctness of the Cazeneuve and Haddon formula is questioned. Görter states that he was unable to form more than a few s.nall crystals of the osazone, which he was unable to isolate and considered that it was due to some impurity in the caffetannic acid. The caffetannic acid is considered by Görter to be a mixture of chlorogenic and coffalic acids. Numerous derivatives and salts of these acids have been prepared and are described by the author to support his contention. The method for carrying out the Krug test which we found to work most satisfactorily may be summed up as follows: To 2 grams finely ground coffee (passing 0. 5 mm sieve), add 10 cc of water and shake for an hour in a mechanical shaking device. Add 25 cc of 90 per cent alcohol and shake again for half an hour. Filter and wash with 90 per cent alcohol. Bring the united filtrate and washings, about 50 cc, to boiling and add 6 ce of saturated lead acetate solution. Separate the precipitated lead caffetannate by means of a centrifuge, decanting the supernatant liquid through a tared filter. Repeat the centrifugal treatment twice with 90 per cent alcohol, decanting each time through the filter. Transfer the precipitate to the filter and wash free from lead. Wash with ether, dry at 100°, and weigh. The weight of precipitate multiplied by 0. 51597 gives the weight of caffetannic acid. CAFFEIN. In the work on caffein a comparison was made of three methods: The official method (Bul. 107, p. 154); the titration of caffein with iodin, according to Gomberg, in the filtrate from the lead caffetannate; and the method proposed by Görter in the paper previously mentioned. Our attention has been directed by Mr. C. D. Howard to a source of error in the provisional method, arising from the fact that the extraction with dry chloroform of the sand-magnesia mixture does not yield the whole of the caffein. Mr. Howard says in his letter: My practice has been to add to the concentrated filtrate, contained in a tin-foil dish, about 10 grams of sand and I gram of magnesium oxid, evaporate and dry in the water oven for a short time. The brittle mass, easily stripped from the dish, I grind finely, place in a paper extraction cartridge, and extract for ten to twelve hours in the usual way. a Annalen, 1908, 359: 217. |