method of titration after driving off the volatile acids by steam distillation. Volatile acids were determined by the indirect method of Windisch and by the proposed new direct method of distilling by steam. In the Windisch methods the titrations were made using litmus paper as an indicator and in the method of Sellier, as in the proposed new methods, phenolphthalein was used.

Comparison of methods for the determination of total, fixed, and volatile acids in wines. [Results expressed as cc tenth-normal acid in 100 cc of sample.]

The results obtained by the California Wine Association method were not satisfactory, the end-point of the titrations being, in most instances, very uncertain. In the angelicas, ports, and sherrys especially, much difficulty was experienced

in carrying out the titrations, and the results were scarcely better when litmus tincture was used. In titrating according to the Windisch methods the point of neutrality was judged to be attained when a small drop of the liquor placed on delicate blue litmus paper just ceased to produce a perceptible red. There appeared to be decided disadvantages in using litmus paper, and the use of litmus tincture even in a clear distillate is open to serious objections, which will be stated presently. In colored wines especially the difficulties were very great, and it was found wellnigh impossible at times to devise a means whereby to judge with reasonable certainty the true end-point of the titration. It was found, however, after considerable practice, that fairly concordant results were obtainable by this method in the majority of instances. Phenolphthalein, on the other hand, while not entirely unobjectionable, was found to give far greater satisfaction. While it was not always convenient to titrate on the undiluted sample, especially in the case of wines containing more or less natural coloring matter, it was found to be entirely permissible, as in the titration of cider vinegars, to dilute with boiled distilled water in order to carry out a successful titration with phenolphthalein. It has been shown that the end-point of a titration can be very accurately judged, even in a deeply-colored wine, and that the addition of water to the extent of 100 or 200 cc does not introduce a serious error in the result. As in a cider vinegar, the change in the color of a wine occurs at a much earlier stage than the change in the indicator and there is never a serious difficulty in safely judging the end point.

As already pointed out, the results shown in the first column of figures are at best only rough approximations. In the majority of instances it was observed that when litmus paper was used the titrations were carried somewhat beyond the point of neutrality which seemed to be indicated by the change in the natural coloring matter of the wine. It is also noted that the results obtained by the titrations employing litmus were uniformly much lower than the results obtained with phenolphthalein. This is true not only in the titrations of total and fixed acids, but also in the direct titrations of the volatile acids. On the basis of the results obtained with phenolphthalein, litmus indicates approximately from 77 to 92 per cent of the total acids and from 58 to 85 per cent of the fixed acids. Doubtless there are theoretical reasons underlying these differences, and the question may well be raised as to whether chemists have given due attention to these considerations in choosing indicators for titrating the acids in wines.

In the first place, there appears to be little justification for the practice adopted by some chemists of employing the natural coloring matter as a correct, indicator in titrating either the total or fixed acids. Little of value is known regarding the action of the oenocyanin or other coloring substances in the presence of acids or alkalies, and it is certain that such substances have not been recommended in the titration of any of the common acids. In the case of litmus also there are some important considerations which should bar it as an indicator for wines as well as fruit products in general. Litmus is not recommended for titrating such acids as tartaric, acetic, tannic, succinic, or malic. In titrating tartaric acid with this indicator, the change is gradual and the end-point indistinct, while in titrating acetic acid, the acetate of sodium formed is alkaline to litmus and tends strongly to hasten the end-point. On titrating solutions of tannic acid, a change takes place almost immediately on beginning the titration, and only a small proportion of the actual acid is indicated. Phenolphthalein, on the other hand, is a very satisfactory indicator with all these acids, and, with the exception of tannic acid, the theoretical amount of acid is obtained. About 80 per cent of tannic acid is indicated, but the total acid is obtained after boiling with a measured small amount of tenth-normal hydrochloric acid.

As a means of shedding some light on the differences occurring in titrating wines with the two indicators, the determinations shown in the following table have been carried out:

Comparison of litmus and phenolphthalein as indicators in titrating some of the organic acids existing in wines.

The results shown for volatile acids by the Windisch method (p. 22) are somewhat higher than those obtained by the proposed new method, using phenolphthalein. Such discrepancies, however, lose their significance when it is considered that in the determination of volatile acids by the indirect method not only are the results of the titrations employing litmus as indicator incorrect, but the titrations of total and fixed acids are not made under comparable conditions. While it is unquestionably true that the volatile acids may be completely driven off by repeated evaporation in an open dish, it does not follow that the results obtained by means of the two titrations are correct. It is conceivable that important changes may occur during the prolonged heating of the wine in order to reduce the material a third time to a pasty consistency. At any rate, we have no positive knowledge that the so-called fixed acids occurring in the final residue represent the actual fixed acids in the original wine. A titration of the residue may suffice as an indication of the acids remaining after driving off the volatile constituents by prolonged heating, but to employ the result of such a titration as a factor in the calculation of the actual volatile acids appears to be an unwarranted proceeding.

In expressing results of, analysis the orthodox custom appears to be to calculate the fixed and total acids as tartaric and the volatile acids as acetic. It is impossible to concede any advantages in favor of this custom. It may be safe to assume that in wines the fixed acids are in the main tartaric and the volatile acids acetic; but, even on such assumptions, the results are strictly erroneous and not readily comprehended. Such a method applied to the various fruit juices and ciders would fail to give significant results in practically all cases, and the case is still worse when one adopts the method of calculating the acids as sulphuric. Instead of these conventional methods it has been found better to adopt the plan of expressing all results for total, volatile, and fixed acids in terms of the number of cubic centimeters of normal acid in a definite measure, say 100 cc, of wine. There will then be afforded results which can be readily compared and comprehended. Furthermore, in case it be required to calculate results in terms of any particular acid, such an operation can easily be carried out.

RECOMMENDATIONS.

(1) The standard temperature for the determination of specific gravity should be changed to 20° C. A statement of reasons for this change seems to be unnecessary, as the matter has been fully discussed by others, and many chemists have for some time

adopted the custom of making determinations at a temperature not far from that of the average laboratory. If the alcohol tables can be revised in accordance with a standard temperature of 20° C. for specific gravity determinations, a very useful service will be performed, especially in the interest of industrial and food chemists.

(2) The method for glycerol should be made a subject for special study. Experience has shown that it is possible not only to increase the accuracy of the method but to shorten the time of the operation. As the provisional method now stands, it appears to be rather tedious, and there are too many opportunities for error. A large error undoubtedly occurs during the evaporations as well as during the repeated extractions. Also, it appears that the residue weighed as glycerol is far from being pure.

(3) The present methods for determining total, fixed, and volatile acids are exceedingly faulty. The method for volatile acids, especially, fails to give results anywhere near the truth. The difficulty lies not only in the collection of 200 cc distillate but in the operation, which is cumbersome and unreliable. The use of litmus in the titrations of total and fixed acids is open to criticism, as that indicator fails to show all of the acids. A study of the proposed new methods is recommended.

(4) A more comprehensive scheme for the examination of the natural coloring matter of wines is required. Attention is called to the use of standard color charts as a means of obtaining comparable results in the hands of different persons. It is recommended that the association make a special study of the character and properties of the coloring matters existing in genuine wines.

REPORT ON BEER.

By H. E. BARNARD, Associate Referee.

Mr. Barnard, referee on beer, reported that no cooperative work on the subject had been done, and made the following statement in regard to the condition of the methods:

Two years ago I presented beer methods which have since been adopted as provisional. I have been working with those methods since that time and find no special necessity for changing them. For that reason I have not made a special report on beer. Much work, however, seems to be necessary if we must determine the different kinds of beer, and I would only suggest to you the necessity for a careful study of methods of beer analysis with special reference to the adoption of some method which will enable us to tell more accurately than is at present possible whether or not beer is brewed from all malt, or part malt, or from malt substitutes.

REPORT ON DISTILLED LIQUORS: COOPERATIVE TEST OF METHODS FOR THE DETERMINATION OF FUSEL OIL.

By L. M. TOLMAN, Associate Referee.

The cooperative work undertaken this year was a comparison of the present AllenMarquardt method, as given in Bulletin 107, revised, page 98, and a proposed modification worked out by the associate referee and his assistants. The modification was based on the determination of the amount of bichromate reduced in the oxidation of the higher alcohols. This method eliminates the distillation of the acids, which the experiments made have shown are not completely distilled off. In order to test this modified method (for details see paper, p. 206) a series of samples was prepared containing varying amounts of pure amyl alcohol (boiling point 131° C.) in approximately 50 per cent by volume ethyl alcohol, and the samples sent to eighteen different laboratories, asking for a comparison of the modified method with the present method as given in Bulletin 107. Eleven reports were received, and the following table gives

the results, the percentage yields being calculated from the grams of amyl alcohol per 100,000 of 100-proof alcohol, as determined by each method.

Comparison of the Allen-Marquardt method and the proposed modification for the determination of fusel oil, using varying amounts of amyl alcohol.

0.050 gram. 0.100 gram. 0.150 gram. 0.200 gram. 0.250 gram. 0.350 gram.

Detroit laboratory.

New York laboratory.
Philadelphia laboratory.
Portland laboratory.

P. ct. P. ct. P. ct. P. ct. P. ct. P. ct, P. ct. P. ct. P. ct. P. ct. P. ct. P. ct. 99.8 102.0

Sunnybrook Distilling Co.. 120.0 125.9

St. Paul laboratory.
Washington laboratory.
Galveston laboratory a.

85.14 88.3

78.03 111.9 77.74 97.69 100.0 130.4 74.1 78.34 62.7 85.69 62.3 85.4 78.5 88.06 74.5 87.7 79.7 87.24 96.6 83.4 95.02 79.4 104.0 76.7 87.7 114.9 87.06 108.6 87.86 85.64 80.6 81.68 78.0 83.22 69.06 91.5 65.2 86.2 68.3 88.5 68.8 84.6 80.0 79.90 81.36 78.60 72.06 75.52 42.8 63.1 50.44 74.2 53.74 80.0 50.9 73.22 61.62 66.60 70.16 62.14 61.86 68.19 76.04 124.5 76.36 131.4 110.3 76.2 91.5 76.7 85.8

103.0

100.0 122.0 78.7 105.0 110.0
82.0 113.0 72.11 86.0 73.8
56.0 106.4 99.70 105.1 80.4
108.8 104.5 100.0
82.64
67.3 134.2 72.1 97.1
87.20 87.24 79.28 82. 14

62.2

San Francisco laboratory a..

85.3

103.7 77.7 94.7

a Excluded from average.

Boston laboratory a.

Average.

121.0

Some of the results obtained at laboratories which had not had experience in operating the method varied markedly from the other figures and are omitted from the average. The averaged results on the various amounts by both methods are plotted, using as the abscissa of the curves the amount of amyl alcohol used in grams per 100,000 of

proof spirit and as ordinates the average percentage yield.

This curve (fig. 2) shows that the new modification gives uniformly higher results, indicating a regular loss in the old method. This loss is undoubtedly largely due, as is shown by the experiments, to the failure to drive over all of the volatile acids in the distilling method unless a much larger amount of water is distilled than that. prescribed in the present provisional method. A very much higher yield of acids

FIG. 2.-Graphic of collaborators' results on amyl alcohol by the
Allen-Marquardt method and the proposed modification.

is obtained by carrying the distillation much further, as is pointed out in the supplementary paper submitted on this subject (p. 206). There is also shown a uniformly increasing loss by both methods as the per cents of amyl alcohol increase. This is doubtless due to the method of extraction, as a 100 per cent yield can be obtained in the oxidation part of the method, as was demonstrated in the experimental work on the determination of the factor 0.001773 (see p. 210). From these results it is evident that a higher yield is due to the more correct estimation of the higher alcohols present in the