The mean per cent of nitrogen determined is 1.446 for the corn meal and 6.749 for the cotton-seed meal. The various determinations range from 0.12 per cent below and 0.19 per cent above the mean for corn meal and for the cotton-seed meal 0.40 per cent below and 0.41 per cent above. If the mean of the several determinations be taken as the commercially correct figure, it appears that the average error is 0.035 per cent of nitrogen for corn meal and for cotton-seed meal 0.11 per cent. On the corn meal 41 out of 52 analysts, or 80 per cent, estimated within 0.05 per cent of the mean, and in the case of the cotton-seed meal 29 analysts, or 56 per cent, estimated within 0.10 per cent. This is a most satisfactory result, so far as these particular analysts are concerned, i. e., the 80 per cent and the 56 per cent, respectively. It seems probable that the limits just mentioned, 0.05 per cent for corn meal and 0.10 per cent for cotton-seed meal, above or below the accepted average, are the best that can be expected of a large number of analysts working in different laboratories and following a general description of a method. It would be desirable to make the limits of permissible error narrower, for mistakes are costly; but this represents the present state of accuracy and about one-fifth of our cooperating colleagues fail to attain even this limit in their corn meal determinations and over two-fifths that for cotton-seed meal. The fact that a large number of analysts do agree closely is a most gratifying improvement over the experience of former years. It is doubtful if any further definition can be made to bring about a closer agreement. The only possible solution, in my opinion, is in concrete standards such as the samples sent out have now come to be. The errors giving too low results may be explained by (1) a false weight (a most improbable source of error); (2) incomplete digestion, leaving some unreduced nitrogen; (3) excessive digestion, sometimes attended by the decomposition of ammonium sulphate; (4) mechanical loss from spurting; (5) incomplete distillation; (6) escape of unabsorbed ammonia in distilling, either through loose fittings or an unsubmerged outlet tube; (7) a false standard acid. Errors in excess are harder to explain. They may be due to (1) a false weight; (2) the incrustation of flasks with ammonium chlorid; (3) the carrying over of caustic alkali; (4) a false standard acid; (5) impure reagents; (6) with a glass condenser a possible solution of alkali. A common fallacy in connection with this subject is that the differences in moisture contribute much to the variation in percentage of nitrogen. But a simple calculation will show that the extreme variations in moisture, as estimated, would cause a difference of 0.06 per cent nitrogen in the corn meal, and of 0.24 per cent in the cotton-seed meal. Furthermore, as it happens, the greater part of the analyses of both samples show that the moisture and nitrogen deviate from the mean in the same direction, i. e., both greater or both less; whereas if variation in moisture caused or contributed to the variation in nitrogen content the two quantities should vary in opposite directions. The only reasonable inference is that personal carelessness or inaptitude causes the widest deviations from the mean. After a consideration of many analyses, therefore, it would seem that 0.05 per cent of nitrogen in corn meal and 0.10 per cent in cotton-seed meal, are fair allowances for deviation. It is believed that the two samples submitted to the association have been analyzed accurately. The one can not be far from 1.45 per cent and the other not far from 6.75 per cent of nitrogen. They are hermetically sealed and nearly enough alike (as regards loss or gain of moisture) to serve as useful standards of comparison. The mean figures may not be absolutely free from error when the sulphuric acid method is used (the Kjeldahl method with its various amend ments and modifications), but as this method is followed almost exclusively in commercial work and investigations, they are at least conventionally correct. Determinations by the soda lime method were made by one analyst, with results considerably higher than the other averages. But the data are so meager that no reliable conclusion can be drawn. Four recommendations are made. The first one calls for from 20 to 30 cc of sulphuric acid instead of merely 20 cc, on the ground that there is danger of the flame striking the bare wall of the flask and decomposing some of the ammonium sulphate. The third one requires the condenser tube to dip below the surface of the standard acid on the ground that when there is much ammonia some may easily escape unless the tube is submerged. The second and the fourth permit the use of copper sulphate, in both the Kjeldahl method proper and in the Gunning modification of it. The claims made for the katalytic action of copper sulphate seem to be justified by experience, and there is reason to suppose that this reagent will do much to shorten the time of digestion. It is well nigh inconceivable that its presence can be in anyway hurtful, and there seem to be many analysts who are very desirous of using it and at the same time of conforming strictly to the official methods. It is recommended that (1) Bulletin 107, p. 6, under (3) Determination, third line of first paragraph, be changed from "20 cc " to " 20 cc-30 cc” (i. e., of sulphuric acid). (2) Ibid., line 4, after "acid," add "From 0.1 to 0.3 gram of crystallized copper sulphate may also be used in addition to the mercury, or in lieu of it.” (3) Ibid., p. 6, line 6 from bottom, after standard acid,” add: “During at least the first ten minutes of distillation the lower end of the condenser tube should dip beneath the surface of the standard acid.” (4) Ibid., p. 7, (b) Gunning Method, (3) Determination, line 4, after “sulphuric acid," add: “From 0.1 to 0.3 gram crystallized copper sulphate may also be added." (5) Ibid., p. 7, (3) Determination, line 4, after "sulphuric acid," write: "Approximately 0.7 gram of mercuric oxid or its equivalent in metallic mercury may also be added, previously to the addition of the potassium sulphate, but if mercury be used potassium sulphid must be used, as in the Kjeldahl method, in the distillation." A general discussion of the third recommendation made by the referee followed, showing a wide variation in practice and opinion regarding this detail of manipulation. The recommendation made was said to be in harmony with the practice in the Bureau of Chemistry, while Mr. Lipman stated that on a thousand determinations made by the New Jersey station without having the tube dip beneath the acid, almost perfect agreement was obtained. Mr. Patrick and Mr. Haskins spoke of the satisfactory results obtained by using a trap if the procedure recommended by the referee were not followed, and Doctor Wiley and Mr. Hopkins called attention to the bearing of temperature on the question. [It is to be noted that in the case of soils, fertilizers, and other low-grade materials, the variation introduced by this loss of ammonia is much less marked than when such materials as meat, cheese, and milk are under consideration. This recommendation was not adopted when presented to the association by Committee A (see page 129).] REPORT ON THE SEPARATION OF MEAT PROTEIDS. By F. C. Cook, Associate Referee. Samples of commercial meat extract and the following letter of instructions were sent to the seven collaborators, who expressed a willingness to cooperate on meat proteids this year: SEPTEMBER 4, 1907. DEAR SIR: I am sending you, under separate cover, a sample of meat extract. The sample should be kept cool until ready for work, then opened and well mixed. DIRECTIONS. Total nitrogen. (1) Determine the total nitrogen in not over 0.5 gram sample. Soluble nitrogen. (2) Dissolve 10 grams of the extract in water and make up to 500 cc. Filter on a plain filter paper, pouring back the first 10 or 20 cc. This is filtrate A. Determine the soluble nitrogen in 25 cc of filtrate A. For the acidity determinations use two 10 ce portions of filtrate A. In one case use phenolphthalein as indicator, diluting the solution before titrating. In the second case employ delicate litmus paper, testing by taking a drop outside by means of a small capillary tube. In determining the coagulable proteid take two 50 cc portions of filtrate A. To No. 1 add 9 grams of sodium chlorid, heat to boiling, add 1 cc N/10 acetic acid and boil three minutes. Let stand on steam bath ten minutes, filter, and wash with 50 cc hot water. Heat No. 2 to boiling and add 1 cc of N/10 acetic acid, boil three minutes, let stand on steam bath ten minutes, filter, and wash with 50 cc hot water. In both cases determine the nitrogen on the filter paper. Use nitrogen-free filter paper. The amido nitrogen is determined by placing 25 cc of filtrate A in a 100 ce flask. Add 15 grams of sodium chlorid and 25 cc of water. Shake and keep cool (15° C. or less) for two or more hours. Make a 24 per cent tannic-acid solution, filter and keep at same temperature. Add 30 cc of the 24 per cent tannic-acid solution to the flask, fill to mark with water, shake, and keep cool (15° C. or less) for twelve hours (usually overnight). Filter off 50 cc of the solution and determine nitrogen therein by adding a few drops of sulphuric acid and evaporating to dryness on the steam bath with help of vacuum if at hand. Add 25 or 30 cc of sulphuric acid but no potassium sulphate and proceed as in the Gunning method. It is necessary to run a blank, as the tannic acid often contains nitrogen. The nitrogen figure minus the blank nitrogen figure multiplied by 2 equals amido nitrogen in the filtrate from 25 cc of the original sample (0.5 gram). Kreatinin determination. Coagulate 20 cc of filtrate A, as in No. 2 under coagulable proteid, filter, and wash with hot water. Place the filtrate in a 500 ce flask, add 15 cc of saturated picric acid solution and 5 cc of 10 per cent sodium hydroxid, shake well. After standing 5 minutes dilute to mark with water and compare the color with an N/2 potassium dichromate solution in a Dubose colorimeter, the scale being set at 8 mm; 81 divided by reading x on the other scale equals milligrams of kreatinin. This is a modification of Folin's method as applied to the estimation of kreatin and kreatinin in the urine. For the determination of kreatin use 15 cc of filtrate A, coagulate as above, filter, and wash. Place in an Erlenmeyer flask. Add 5 ce of N/2 hydrochloric acid and attach a reflux condenser, place on steam bath for 34 hours, cool, transfer to a 500 cc flask, add 5 ce of N/2 sodium hydroxid and proceed as above, adding 15 ce picric acid, etc. This reading gives the total kreatinin. From the total kreatinin subtract the original kreatinin and multiply the difference by 1.16 to obtain the kreatin originally present. Report results as follows: Total nitrogen, per cent. Soluble nitrogen, per cent. Amido nitrogen, per cent. Kreatin, per cent. a Zts. physiol. Chem., 1904, 41: 223; Amer. J. Physiol., 1905, 13: 48. Acidity (1) using litmus cc N/10 sodium hydroxid per 100 grams. Coagulable nitrogen, using sodium chlorid, per cent. Duplicate determinations should be made. Any criticisms and suggestions will be gladly received. DISCUSSION OF RESULTS. In the sample sent out there was but a small amount of insoluble proteid present, consequently no conclusion can be drawn under that head. The cooperative results as given in Table I show a wide variation for total nitrogen, and consequently for the different nitrogenous constituents, as the nitrogen is the basis of the determination. The tannin-salt method gave fairly satisfactory results, although there is still considerable variation in the results obtained by the different collaborators. The trouble often experienced by the foaming in the Kjeldahl process, due to the presence of the tannin, is eliminated by evaporating the solution, after adding 2 or 3 drops of sulphuric acid, almost to dryness, then adding about 30 ce of sulphuric acid and no potassium sulphate, and digesting as usual. The proteose and peptone figures reported were obtained by difference. Cooperation was also obtained on the determination of coagulable proteids. There has been considerable discussion as to the use of sodium chlorid for this purpose, and the collaborators were instructed in one case to add 9 grams of sodium chlorid to 50 cc of the meat extract solution, while in the other case no sodium chlorid was to be added. As would be expected, practically all of the results were higher when sodium chlorid was used, but as the amount of coagulable proteid present in the sample was small the results are not striking. In Table II some results obtained by the referee are given, showing the influence of varying amounts of sodium chlorid on the coagulation of proteid in a commercial meat juice. TAELE II.-Coagulable proteid tests using varying amounts of sodium chlorid and 50 cc of the meat extract solution. These figures show that the amount of nitrogen precipitated increases with the amount of sodium chlorid used, and Mr. Lowenstein, chemist for Morris & Co., Chicago, has obtained similar results for coagulable proteids on samples of meat extracts. It is evident, therefore, that to obtain correct figures for coagulable proteid nitrogen no sodium chlorid should be employed. The two indicators, phenolphthalein and litmus, used in determining the acidity of meat products were compared, and the results obtained by the various chemists show that in practically all cases higher results were obtained where phenolphthalein was used. In a good many cases the difference is marked. The reason for this is that many of the organic acids have no action on litmus, but do act on phenolphthalein. The referee finds that phenolphthalein inevitably gives higher results than litmus, is easier to handle, and the end point is more accurate, but in certain dark-colored solutions of meat preparations litmus paper must be used. In Table III the results of acidity tests on solutions of meat extract and on similar solutions after removing the insoluble proteid by filtering, and after coagulating and filtering, are given: TABLE III.-Acidity determinations. [Comparing the original solution, the solution after filtering, and the solution after coagulation and filtering, using two indicators.] As the amount of insoluble proteid present in most of the samples was small, the results on this point are not conclusive, but the two latter sets of figures show lower results than those obtained by titrating the original solution, and in two cases the removal of the coagulable proteid before titration gives the lowest results. This is true only when litmus is used, the phenolphthalein |