OCTOBER, 1915. Vol. XL., NO. 476. THE ANALYST. PROCEEDINGS OF THE SOCIETY OF PUBLIC ANALYSTS AND OTHER ANALYTICAL CHEMISTS. THE ESTIMATION OF STARCH IN COCOA BY MEANS OF TAKA-DIASTASE. BY CECIL REVIS AND H. R. BURNETT. (Read at the Meeting, June 2, 1915.) THE estimation of starch in cocoa has been carried out hitherto either by direct acid conversion or by the use of diastase, or of diastase followed by acid conversion.Direct acid conversion carried out on the fat-free dry cocoa matter is the worst possible method, as there are present in cocoa large quantities of pentosans and other bodies leading to the production of reducing substances when boiled with acid. The magnitude of the error is easily seen from the results of such conversions published by Winton and others (“An.Rep. Conn. Agric. Exp. St.,” 1902, p. 270), who give 23-66 per cent. crude starch in nib and 12.59 per cent. in shell as mean figures; while from microscopical examination it is practically certain that shell does not actually contain any starch at all. Bolton and Revis (see “Fatty Foods,” p. 298) found 19 to 23 per cent. in nib and 9 to 10 per cent. in shell by the use of a modified form of Ewer’s method.Better results are obtained by the use of diastase followed by acid conversion, in which case Winton and others record 17 per cent. in nib and 4.47 per cent. in shell as mean figures; but Davis and Daish (J. Agric. Sci., 1913, 5, 437; ANALYST, 1913,38,504), in an investigation into the various methods of estimating carbohydrates in plant substances, have shown that there is a distinct loss of carbohydrate during the acid conversion which follows the diastase treatment, and that low results are obtained.These observers have also shown that the simple diastase method of O’Sullivan fails in the case of plant extracts when a clearing agent has to be employed after the starch conversion, owing to the mechanical removal of dextrin from solution by the precipitate formed during the clearing process.On account of these difficulties they experimented with taki-diastase as the starch converter, since, with this substance the dextrin stage is rapidly passed through, All of these methods lead to erroneous results, but for different reasons.430 REVIS AND BURNETT: THE ESTIMATION OF STARCH IN COCOA and there are then present only maltose and dextrose, which may be easily deter- mined by combined polarimetric and copper reduction methods, and there is no loss of carbohydrate in such a solution when a clearing agent is used.The proportions of maltose and dextrose vary with the time of action of the takh-diastase, but dextrin completely disappears after a few hours at blood-heat temperature.As the problem presented by cocoa is very similar to that of plant extracts, we hrve applied the use of taka-diastase to the determination of starch in cocoa with success. 4 few preliminary experiments showed that the conclusions of Davis and Daish as to the exactness of the method were fully justified, and we also found that other observers obtained identical results on the same sample.A clearing agent is necessary in the caSe of cocoa, and we have found that a, solution of acid mercuric nitrate, made up on the formula of Wiley for use in clearing milk, behaves admirably for this purpose. Five grms. of fat-free dry cocoa matter are weighed out into a beaker and thoroughly stirred with 50 C.C. of 10 per cent. alcohol (by volume) and filtered under pressure-for preference in a large Buchner funnel as the process is thereby considerably hastened.The cocoa is then washed with two. further quantities of 50 C.C. of 10 per cent. alcohol, and finally with 10 C.C. of 95 per cent. alcohol (by volume), taking care that at no stage the cocoa sucks dry, as it is then more difficult both to wash and to remove from the filter.This alcohol washing is very important, as otherwise great difficulties occur later on and erroneous results are obtained. The moist cocoa is then scraped and washed out of the funnel into a 250 C.C. flask with about 125 C.C. of boiling water, well mixed, and the flask placed in boiling water, with constant shaking, for fifteen minutes to gelatinise the starch. The contents of the flask are then cooled to 38' C., mixed with 0.05 grm.of takh-diastase, rubbed up with a little water, 2 C.C. of toluene added, and the flask, after shaking and stoppering, placed in an incubator at 38" C. for twenty-four to thirty-six hours, shaking at intervals. At the end of this period 10 C.C. of & NaOH are added to stop the action, the flask is cooled to 15' C., about 100 C.C.of water added, then 10 C.C. of acid mercuric nitrate solution run in on the surface, and finally water to the mark below the toluene. The contents of the flask are then mixed and filtered. To 100 C.C. of the filtrate (which should be perfectly colourless and bright), carefully measured into a flask, are added 0.5 grm. of sodium phosphate (Nat2HP0,.12H20), and, after this has dissolved, 10 C.C.of sodium hydroxide solution are run in, the Bolution being agitated as this is done. The contents of the flask are mixed and filtered, and, in the case of 5 grms. of cocoa matter, 50 C.C. are used for the gravinietric copper determination, the polarimetric readings being also made on this filtrate. The solution of sodium hydroxide used above must be so adjusted that 10 C.C.just neutralise 4 C.C. of acid mercuric nitrate solution, and should on no account be in excess, it being preferable to leave the solution slightly acid after the addition of the alkali. I n our case the solution contained 55 grms. of ground caustic soda (97 per cent. NsOH) per litre. We carefully determined that this treatment with acid mercuric nitrate caused no loss of starch conversion products.I n The proeess is as follows : We have examined a number of COCOBS, both nib and shell, by this method.BY MEANS OF TAKA-DIASTASE 431 all cases the partition into shell and nib was made by ourselves from the whole roasted bean i n order to be sure that the differentiation was a true one. Shell.-Four samples of shell were examined, using 10 grms.in each case, and only negligible reductions and polarimeter readings were obtained. The results are given in the table below. The copper oxide in each case was that obtained from the 50 C.C. of final solution used, and the polarimeter figures are the actual Venteke saccharimeter degrees obtained on polarising the filtrate after the acid mercuric nitrate and phosphate treatment.They have not been calculated to he original shell. Sample of Shell CuO Weighed. 1. Blend ... ... 0.038 grm. 2. Guayquil ... ... 0.031 ,, 3. Trinidad ... ... 0.022 ,, 4. Grenada ... ... 0.024 ,, Veiitzke Degrees. + 0.04 + 0.01 + 0.03 + 0.05 From these results it is evident that cocoa shell contains no true starch, and it is Nib.-Five samples were examined, after careful extraction of the fat, with the probable that this is the first time that it has been demonstrated analytically.following results : I’erucutage of Starch in Fat-Free Dry Nib. Cocoa. ... Accra (low grade cocoa) ... ... 12.8 Gunyquil ... ... ... ... ... 8.0 Trinidad ... ... ... . . . ... 14‘5 Grenada, ... ... ... ,. . ... 12.3 San Thomi! ... ... ... . . . . . . . 12.4 I n all the above cases estimations were made with 5 and 10 grrns. in order to eliminate the effect of the volume of the precipitate caused by the acid mercuric nitrate, but it is probable that, if an allowance of 3 C.C. be made for the volume of precipitate from 5 grms.of cocoa matter, that such allowance is within the error of experiment. We had hoped to find that the percentage of starch in cocoa would prove to be a, fairly constant quantity, and so provide a means of estimating ‘( cocoa matter ” in mixtures.The number of samples examined is not sufficient to allow of any very definite statement, but, with the exception of Guayquil, the percentage appears to be about 13 per cent., and probably in blends might be taken to approximate to this figure. It might be noted that the figure obtained by Winton and others, after deducting the figure for shell, is 12.5 per cent.(diastase and acid) and 11-1 per cent. (crude acid conversion), and by Bolton and Revis is 11.5 per cent. The acid mercuric nitrate solution is made as follows: Pure mercuric oxide (not mercury as recommended by Wiley) is dissolved in twice its weight of nitric acid (sp. gr. 1-42), and the resulting solution diluted to five times its volume with water.The powder keeps well in the dark, but should be tested before use to see if it has any reducing power or is optically active, i n which case a control must be run side by side with the experiment. Taka-diastase can be obtained from Messrs. Parke, Davis and Co.432 REVIS AND BURNETT: THE ESTIMATION OF STARCK IN COCOA In conclusion, we desire to express our thanks to Mr.A. W. Stokes and to Mr. T. Macara for obtaining reliable samples of cocoa bean for us. DISCUSSION. The PRESIDENT remarked that the method of acid hydrolysis was well known to give erroneous results in the case of substances like cocoa, the extent of the error depending on the nature of the cellulose and pentosans present. He could not quite agree with the authors, however, in passing censure on the ordinary diastase method, provided that it was properly carried out.A blank determination was, of course, necessary, and this must include hydrolysis with acid. Failing this, the results would obviously be too high, to an extent possibly as great as in the case of direct acid hydrolysis. Malt extract had the advantage of being eaeily obtainable any- where, whereas in some places there might be some little difficulty in obtaining tak&-diastase.Mr. J. L. BAKER said that he had seen the aubhor’s process carried out by Mr. Hulton, and had been impressed by its smoothness of working. He had tried the ordinary method (i.e., hydrolysis with acid and precipitation with phosphotungstic acid), but, as had been said, this was useless in the case of products like cocoa. He regarded the use of taki-diastase as a distinct step forward, and he must say that he should prefer it to malt extract.I t was quite readily obtainable in a dry condition, and its hydrolytic activity remained fairly constant even after several months’ keeping. He did not see any particular necessity for continuing the action for thirty-six hours.Davis and Daish had shown that, so long 8s the dextrin stage was passed, the relative proportions of dextrose and maltose were immaterial. Mr. REVIS, in reply, said that the use of ordinary diastase was quite satisfactory as long as the solution did not require clearing; but if any clearing agent was used, dextrin was liable to be carried down mechanically.Moreover, Davis and Daish had shown that, on hydrolysis of the results of the diastase conversion with acid, dextrose disappeared to an appreciable extent. Personally he had always thought that such loss of dextrose was small, but according to Davis and Daish it might be rather considerable. The PRESIDENT remarked that the solution might be made up to a known bulk and a portion filtered off. Mr. REVIS said that even then some clearing agent would have to be used, and some dextrin would thereby be thrown down which should g? into the filtrate. That was quite clear from the results obtained by Davis and Daish. Tak&-diastase was quite readily obtainable. One must be careful to see that it had no reducing or polarising power, as some samples had, in which case the necessary allowances had to be made. The time of standing was merely a question of convenience. At first they had thought that the conversion might perhaps not be complete after the shorter period, but they now thought that eighteen hours would probably be sufficient. I t would probably be simplest to allow twenty-four hours.