acid and digested immediately with sulphuric acid. The digestion with sulphuric acid only is extremely slow.

There seems to be quite a difference in the various tannic acids (c. p.) on the market. Some produce a very voluminous precipitate, others comparatively little. No account is taken in the method of the volume (bulk) of this precipitate. If it is considerable (as it might be in some extracts), it would introduce the same error that one finds in some of the present official methods.

Kreatinin: A criticism pertinent at this point, but also applicable to the methods as a whole, is they are not specific enough. I could readily observe in our own laboratory that the personal equation would be quite a factor. As this is the first year's work on this subject, just such points should be brought out. The method for kreatinin states, “take 20 cc of A, coagulate, filter, and wash." As the sample in question contains no coagulable proteid, we made the kreatinin determination both with and without dilution (washing), and found that the volume of the solution is quite a factor. Since the proposed method is a modification of Folin's method, in which he uses approximately 10 cc of urine, we thought that the more concentrated the solution the better would be the results obtained, and our results seem to confirm this view. Consequently we avoided transferring and washing as much as possible. Hehner states that 15 ce of picric were not sufficient for meat extracts, and also gives the amount of picric acid in solutions saturated at different temperatures, and this ought to be mentioned. Time, unfortunately, did not permit us to confirm Hehner's results, and in the work reported the quantities given, in the method by the referee were used. I have heard from chemists who used 25 cc of picric acid, however, that they did not get results nearly so high as Hehner reports.

Kreatin: To determine whether the time of heating the solution on the steam bath was a factor, one determination was left on the steam bath for fifteen hours. The results checked very satisfactorily with the three-and-onehalf-hour test.

The Grindley-Woods method was not tried because of lack of time, but we would suggest that after converting into kreatinin the sodium hydroxid and picric acid be added to the original Erlenmeyer flask and not transferred to the 500-cc flask until ready to dilute at the end of five minutes.

One of the principal objections to reporting these results was the age of the sample at the time of analysis. From practical experience it is known that extracts change with age, in fact, seem to improve in flavor in the case of solid or paste extracts, and we did not know whether the changes would be appreciable nor how much they would affect the content of meat proteids and bases and the acidity test.

Mr. Holmes C. Jackson, of Albany, N. Y., writes:

(1) As regards the acidity determination, it would seem to me that the 10 cc portions which were required to be used were a somewhat small fraction of the total, since in calculating the final figures (cc N/10 NaOH per 100 grams) it is necessary to multiply the titration figures by 500. Obviously very slight variations in the titration will give large variations in the final results. As stated last year, I am decidedly in favor of phenolphthalein as an indicator as compared with litmus, since, unless one makes his own paper, and neutral paper at that, I can not see how the necessary accuracy can be obtained.

(2) Concerning coagulable proteid, it is suggested that some mechanical feature in the method be introduced so that during the boiling and heating in the steam bath evaporation does not take place. Although not mentioned in the

50056-Bull, 116-08- 4

method sent out, some such scheme was found to be absolutely necessary. I used a simple unjacketed reflux condenser.

(3) As regards the amido nitrogen, it seems to me that the results can not be absolutely correct, and the cooperative work shows that the results obtained by the different analysts are not comparable. There must be a variable factor introduced as the result of the presence of such a volume of precipitate. I therefore question again the accuracy of multiplying by 2 the results obtained from 50 cc of the filtrate from the original 100 cc of the mixture of filtrate and precipitate. In determining this point, it is suggested that the double dilution method used in the determination of lactose in milk be employed or some similar procedure which takes into account the volume of the precipitate.

The first sample sent me was lost after I had made the coagulable nitrogen analysis, and it gave a much higher result than the second sample sent. If the second sample was the same as the first and had been standing about during that time, is it not possible that a change had taken place in the character of the coagulable proteid by which it was transformed into noncoagulable? Might not this help to explain the great variations in the cooperative results for this factor?

Mr. Arthur W. Dox, of Storrs, Conn., says: "The precipitation of coagulable proteid seems to be incomplete. I would suggest that a greater concentration of acid be used and that the precipitate be washed with water that has been slightly acidulated. With sodium chlorid a higher figure is obtained, although the concentration is not sufficient to cause a precipitation of proteoses. If the latter method is adopted, the precipitate should be washed wih a solution of sodium chlorid, so that none of it will be redissolved."

Mr. E. W. Rockwood, of Iowa City, Iowa, says: "The principal difficulties encountered were the indistinctness of the end reaction with litmus and in the colorimetric determination of the kreatin and the kreatinin. The latter was very probably due to lack of experience in judging these shades. The work was verified, and I can not explain the discrepancy in the nitrogen on any other basis than that it was not easy to sample accurately a substance with so high a nitrogen content. Both chemists have had experience in Kjeldahl determinations and worked independently."

Some analyses of commercial samples of meat extract by the cooperative methods are given in Table IV, including analyses by Mitchell and Grindley, of the University of Illinois, and W. D. Richardson, of Swift & Co., of Chicago.

TABLE IV.-Analyses of commercial meat extracts by referee's methods.

Number of

sample.

H. H. MITCHELL AND H S. GRINDLEY.

[Each figure represents the average of three determinations.]

"Beef extract from the Bureau of Chemistry, U. S. Dept. Agr.

TABLE IV.—Analyses of commercial meat extracts, etc.—Continued.

(1) That the following method for the determination of the acidity of meat products be adopted as a provisional method:

Determine the acidity by titrating a dilute solution of the meat, or meat preparation, with N/10 alkali, using phenolphthalein as indicator. But if the color prevents the use of this indicator, use litmus paper with the understanding that the results are lower than when phenolphthalein is employed.

(2) That the modified tannin-salt method for the separation of proteoses and peptones from the simpler amido bodies be adopted as provisional. The method reads as follows:

Employ 50 cc of the filtrate from the determination of the coagulable proteids, transfer to a 100 cc graduated flask, add 15 grams of sodium chlorid, 10 cc of cold water, shake well and place in the cold at 12° C. (an ice box is a convenient place.) Prepare and filter a 24 per cent solution of pure tannin, and keep at the same temperature, 12° C., also a flask of distilled water. After the temperature of the contents of the flask has reached 12° C. add 30 cc of the tannin solution, shake thoroughly and fill to the mark with cold water, shake again and let stand at 12° C. for twelve hours (over night). Filter off 50 cc at 12° C. and determine the nitrogen therein as follows:

Transfer 50 cc of the tannin-salt filtrate to a Kjeldahl digestion flask and dd a few drops of sulphuric acid. Place the flask in the steam bath, connect with the vacuum, and evaporate to dryness. In the digestion process about 30 cc of sulphuric acid are added, but no potassium sulphate. The remainder of the process is carried out in the usual manner.

A blank must be run simultaneously. The nitrogen in 50 cc of the filtrate above, minus the nitrogen in the 50 cc of the blank filtrate multiplied by 2, gives amido nitrogen. The nitrogen of the proteoses and peptones is determined by difference.

(3) That, inasmuch as the best modification of the Folin colorimetric method for the determination of kreatin and kreatinin in meat products has not been determined upon, this point be studied another year.

(4) That the referee for 1908 investigate the xanthin base method and a method for estimating organic phosphorus.

REPORT ON THE SEPARATION OF VEGETABLE PROTEIDS.

By H. SNYDER, Associate Referee.

The work on the separation of vegetable proteids has been confined to a study of the literature on the subject. No attempt has been made to correlate the literature, and only a general summary is given. No samples were distributed for the cooperative work, as previous investigations have shown that in the case of cereals, as in wheat, closely agreeing results can be secured when skilled analysts work on the same sample and follow a definite course of procedure.

There is such an extensive overlapping of the proteids as to solubility that it is doubtful whether accurate analytical methods can be devised for the absolute determination of any one proteid. The water-soluble, dilute saltsoluble, and alcoholic-soluble proteids can be determined with a reasonable degree of accuracy, but each solvent has a tendency to yield a mechanical mixture rather than an individual proteid. In the case of some of the cereals, however, a solvent may extract an individual proteid with a high degree of purity.

In the studies that are being made of the synthetic and ultimate composition of proteids, pure proteids are secured by special treatment not admissible in quantitative work. The object is to secure a pure proteid, and the mechanical losses due to the overlapping of the solvent can be disregarded.

At the present time it would seem best, in the determination of vegetable proteids, to designate the products as water-soluble, salt-soluble, or alcoholsoluble proteids, designating the proteid by the solvent rather than by the specific names suggesting individual proteids, except in special cases, as in wheat, where the alcohol-soluble proteid is mainly gliadin.

There is much need of an improved method for the determination of proteid and nonproteid nitrogen. While the determination of total nitrogen has reached a high degree of perfection, that for the determination of proteid nitrogen is defective. Stutzer's albuminoid nitrogen method does not make a sharp distinction between proteid and nonproteid nitrogen in many substances. Instances occur where the copper proteid is soluble owing to secondary reactions. In the presence of a large amount of starch the filtration and washing of the precipitate often require hours, rendering the accuracy of the results questionable.

It would seem best first to direct attention to the improvement of the methods for the determination of proteid and nonproteid nitrogen rather than of individual proteids.

The analytical and synthetical work that is being done bids fair to give, in the near future, important results, and then it will be possible to devise rational methods for the separate determination of the various component proteids. To analyze accurately a complex compound, its composition must first be determined.

It is therefore recommended that the referee on vegetable proteids give special attention to the study of methods for the determination of proteid and nonproteid nitrogen.

The association adjourned to meet at 2.30 p. m.

THURSDAY-AFTERNOON SESSION.

On the suggestion of Doctor Frear, the recommendations in the president's address calling for changes in the constitution were referred to the committee on amendments to the constitution for recommendations.

In the absence of Mr. Woll, Mr. Bartlett, the associate referee, presented the report on dairy products.

REPORT ON DAIRY PRODUCTS.

By F. W. WOLL, Referee.

The referee was requested, by a vote of the association, to continue the study of the analysis of condensed milk with special reference to the determination of lactose, sucrose, and fat in the sweetened product.

The following explanatory letter, together with two carefully prepared samples of sweetened and unsweetened condensed milk, were sent to 22 different stations where chemists had signified their willingness to cooperate in the work outlined by the referee.

MADISON, WIS., March 28, 1907.

DEAR SIR: As stated in my circular letter of the 12th instant, the cooperative work on dairy products, Association of Official Agricultural Chemists, for this year will be along two lines:

I. Analysis of condensed milk, with special reference to the determination of sucrose and lactose in the sweetened product.

The provisional methods for analysis of condensed milk adopted tentatively at the last convention of the association are as follows:

Preparation of sample: Mix thoroughly by transferring the contents of a can to a large evaporating dish and stirring with a pestle. Weigh 40 grams into a 100 cc flask, transfer thereto by washing, and make up to the mark with water.

Total solids: Dilute a measured portion of the above 40 per cent solution with an equal amount of water, use 5 cc of the diluted mixture, and proceed as in the case of milk analysis.

Ash: Ignite the residue from the total solids, cool, and weigh.

Protein: Determine nitrogen by the Kjeldahl or Gunning method in 5 ce of the 40 per cent solution and multiply it by 6.25.

Lactose: Dilute 5 cc of the 40 per cent solution to about 40 cc and add 0.6 ce of Fehling's copper solution; nearly neutralize with sodium hydrate solution and make up to 100 cc. Filter through a dry filter and determine the lactose in an aliquot by the gravimetric and polariscope methods.

Sucrose: Determine (a) gravimetrically and (b) by difference, by deducting the milk solids (lactose, protein, fat, ash) from the total solids.

Fat: Determine by (a) double-extraction method (Circular 32, page 6, Bureau of Chemistry, U. S. Department of Agriculture) and (b) Babcock centrifugal method, using either Leach's modification (Food Inspection and Analysis, page 49) or Farrington's (Wisconsin Station Report 17, page 86). II. The Gottlieb method for the determination of fat in milk.

The directions for the Gottlieb method are as follows (Landw. Vers. Sta., 1892, 40: 1-27):

Ten cc of milk are measured into a glass cylinder inch in diameter and about 14 inches long (see Land. Vers. Sta., 40: 6; a 100 ce burette or a eudiometer tube will do); 1 cc concentrated ammonia is added and mixed thoroughly with the milk. The following chemicals are next added in the order given: 10 cc of 92 per cent alcohol, 25 cc of washed ether, and 25 cc petroleum ether (boiling point below 80° C.), the cylinder being closed with a moistened cork stopper and the contents shaken several times after the addition of each. The cylinder is then left standing for six hours or more. The clear fat solution is next pipetted off into a small weighed flask by means of a siphon drawn to a fine point (see Fig. 6, loc. cit.), which is lowered into the fat solution to within 0.5 cm of the turbid bottom layer. After evaporating the ether solution in a hood, the flasks are dried in a steam oven for two to three hours and weighed. This method is applicable to new milk, skim milk, buttermilk, whey, cream, cheese, condensed milk, and milk powder, but has been found of special value for determining fat in skim milk, buttermilk, cheese, and condensed milk. In the case of products high in fat a second treatment with 10 cc each of ether and petroleum ether is advisable in order to recover the last trace of fat.