JULY 15, 1939
ANALYTICAL EDITION
tribute to the value of fusel oil as normally determined. However, the acetyl chloride method indicates the presence of considerably more alcoholic hydroxyl groups than the Allen-Marquardt method. The major portion of the difference lies in the distillate, and indicates that the oxidation procedure is not capable of oxidizing quantitatively ethyl alcohol to the corresponding acid. This fact was corroborated by another set of experiments in which carbon tetrachloride containing a small amount of ethyl alcohol was subjected to both procedures. By the Allen-Marquardt method 0.00074 mole of alcohol was recovered; by the acetyl chloride method, 0.000975 mole. This is in agreement with the conclusions of Schidrowitz and Kaye ( 2 1 ) , who indicated that ethyl a1coho1 was not oxidized quantitatively during the oxidation procedure. I n the Allen-Marquardt method a blank must be run to determine the acidity produced by the carbon tetrachloride in the oxidation procedure. TABLEVII. CCId Used Ml. 25 50 100
GENERATION OF ACIDS Acid Found Mole 0,000020 0.000025 0.000020
Several experiments were made to determine the true cause of the generation of acids during the oxidation procedure. The results, given in Table VII, indicate that the acids are not generated by virtue of impurities in the carbon tetrachloride but by the partial decomposition of the carbon tetrachloride itself. It is evident that the evaluation of fusel oil in distilled spirits
393
depends upon the method used for the determination. Each method produces accurate values which are relative, but not in agreement when compared with values obtained by different procedures. Each procedure, used previously, is not specific for alcoholic hydroxyl groups alone, but is affected by the presence of other reactive groupings such as aldehydes, unsaturates, and esters. The acetyl chloride esterification procedure is normally not affected by the presence of these groupings and tends to give values which are specific for hydroxyl groups. However, according to the method as outlined for the acetyl procedure, any alcohol below n-butyl will be excluded from the fusel oil value. I n view of these considerations, the results obtained by the acetyl chloride method are designated as an “amyl alcohol” value and not as a “fusel oil” value.
Literature Cited (1) Bleyer, B., Diemair, W., and Frank, E., 2. Untersuch. Lebensm., 66, 389 (1933). (2) Budagyan, F., and Ivanova, N., Ibid., 63,200 (1932). (3) Fellenberg, Th. von, Chem.-Ztg., 34, 791 (1910). (4) Komarowsky, Ibid., 27, 807, 1086 (1903). (5) Korenmann, J., 2. anal. Chem., 88, 249 (1932). (6) Leach, “Food Inspection and Analysis”, p. 780, New York, John Wiley & Sons, 1930. (7) Lunge, G., “Technical Methods of Chemical Analysis”, p. 726, New York, D. Van Nostrand Co., 1914. (8) Penniman, W. B. D., Smith, D. C., and Lawshe, E. I., IND. ENG.CHEM.,Anal. Ed., 9, 91 (1937). (9) Rose, Stutzer, and Windisch, Arb. kaiserl. Gesundh., 5, 391 (1889). (10) Ruppin, E., 2. Untersuch. Lebensm., 66,389 (1933). (11) Sohidrowitz and Kaye, Analyst, 30, 190 (1905). (12) Smith, D. M., and Bryant, W. M. D., J. Am. Chem. Soc., 57, 61 (1935).
Kniaseff Fat Test HORACE H. SELBY
AND
THELMA A. SELBY, Hage’s, Ltd., San Diego, Calif.
F
OR twelve years the writers have made comparative analyses of ice creams and ice milks, taking the methods
of Mojonnier and the Association of Official Agricultural Chemists as standard, and attempting to correlate the results of some twenty-five modifications of the Babcock method with the standard results. With only one method has it been possible to obtain even approximate results consistently in a series of comparisons when a suitably wide range (4 to 20 per cent) of fat content has been employed. The exceptional method is that of Kniaseff (W), which, in the authors’ hands, has yielded results of a high order of accuracy when suitable corrections have been applied. The method is lengthy, however, and the advantage of speed which Babcock-type methods inherently possess over the solvent-extraction methods is lost, if but one or two estimations are to be made. The purpose of this note is to outline minor changes in the Kniaseff procedure which appear to increase the accuracy and the speed attainable.
Reagents and Procedure REAGENT I. Distilled water, 1000 ml.; sodium hydroxide, A. C. S., 132 rams; ammonium sulfate, A. C. S., 42.8 grams. REAGENT Ethyl alcohol, U. S. P., 391 ml.; n-butyl alcohol, b. p. 116-118’ C., 24 ml.; ammonium hydroxide,
8.
A. C. S., 24 ml.; ethyl ether, A. C. S., 94 ml.; petroleum ether, b. p. 26-31’ C., 94 ml. PROCEDURE. 1. Weigh 9 * 0.02-gram duplicate samples into 9-inch, 9-gram, 50 per cent Babcock cream test bottles which have been proved accurate by mercury calibration. The writers discarded bottles with errors greater than 0.1 per cent in the ranges 50 to 40 per cent, 50 to 30 per cent, and 50 to 20 per cent. 2. Add 8 * 0.1 ml. of reagent I. Shake 5 seconds with rotary movement. 3. Add 5 * 0.1 ml. of reagent 11. Shake 10 seconds. 4. Digest 5 minutes at 84 * 1’ C., shaking 5 seconds after 1.5, 3.5, and 5 minutes. If foam rises above 40 per cent mark, remove bottle for an instant, then reimmerse. 5. Centrifuge 30 * 5 seconds at relative centrifugal force of 165 X gravity a t base of bottle at 65 * 5‘ C. 6. Add tap water at 70 * 5” C. to base of neck. Do not agitate. 7. Centrifuge 30 * 5 seconds as before. 8. Add water t o 50 per cent mark. 9. Centrifuge 60 * 5 seconds. 10. Place in water bath at 60 * 2’ C., so adjusted that temperature falls to 57” C. in 10 t o 20 minutes. 11. When temperature reaches 57 * 0.5” C., add oil and read. 12. Multiply reading by appropriate factor, K , obtained from Figure 1 .
I. CHANGES IN KNIASEFF PROCEDURE. Reagent I1 under-
goes some change in composition due possibly to amine forma-
INDUSTRIAL AND ENGINEERING CHEMISTRY
394
VOL. 11, NO. 7
FIGURE1
tion. It is recommended, therefore, that the solution age one week before being used. 11. It is recommended that reagent I1 be stored under screw caps with liners of metal foil instead of under groundglass stoppers. The ordinary ground-glass closure permits the low-boiling- fractions to evaporate. 111. As originally devised, the method required a digestion neriod of 15 minutes. In the mesent work a 5-minute aeriod &as proved to be sufficient. IV. Completed tests should be read from a water bath the temperature of which is falling, in order that the menisci may have constant volumes. V. The results should be multiplied by a factor which is a function of the fat content. Figure 1 was derived as follows: Thirty mixtures, varying in size from 1000 to 15,500 pounds, were prepared under commercial conditions, pasteurized, homogenized, cooled, and thoroughly mixed. One sample 'was taken from each batch and was analyzed for fat by one of the authors, using the Mojonnier method. At the same time, the other worker estimated the fat content of the same sample, using the modified Kniaseff procedure. The fat contents varied from 4.2 to 18.3 per cent (Kniaseff). The mean Kniaseff result was multiplied by the factor necessary t o equal the Mojonnier in each case and the factors found were laid out as circles on the graph. The smooth curve which would yield the minimum variation in the extreme cases was then drawn. This seemed wiser than the usual statistical procedure, for the number of observations was much too small to justify least squares treatment, No analyses were suppressed or discarded.
Discussion A large number of comparisons have been made, using frozen products of several flavors, with apparently the same order of accuracy, but the results are not tabulated because frozen samples cannot be prepared and analyzed invariably with the accuracy requisite to a comparison of this nature. This is well known and has been reported by many, notably by Bird and Johnson (1). I n the writers' experience duplicate Mojonniers on mixes have differed no more than 0.05 per cent, while carefully sampled frozen products have shown differences between duplicates as great as 0.26 per cent. Because the Kniaseff method is empirical, balancing fat losses from hydrolysis and saponification with gains from dissolved solutes, the physical treatment accorded the sample is of the utmost importance. For this reason, all variables were controlled as closely as was practicable. However, it appears that operations 2 and 3 can be performed more roughly (3). Also, operation 5 is probably satisfactorily done at a relative centrifugal force of 100 to 200 x gravity, for the density differential between the fat and the surrounding liquid is large (0.82/1.02 at 60" C., a t end of test). The data pertinent to this paper are presented in Table I. Columns 1,2, and 3 give the Kniaseff results, column 4 gives
-
PER C E N T . F A T
the circle values of Figure 1, the fifth column contains the K values represented by the curve, the sixth column represents the "corrected" Kniaseff results-i. e., column 3 values multiplied by those of column 5-and the last column gives variations of the Kniaseff results from the Mojonniers. TABLEI. KKIASEFF AND MOJONKIER RESULTS Xniaseff Kniaseff I I1 Mean
4.10 4.20 4.90 4.20 4.40 6.30 10.35 10,60 10.75 10.90 11.20 12.20 12.35 12.80 13.15 13.30 13.40 13.50 13.40 13.30 13.30 13.50 14.40 15.80 16.10 16,OO 17.80 18I O 0 18.25 18.15
4.30 4.30 5.00
4.25 4.80 6.40 10.55 10.70
10.75 11.00 11.40 12.30 12.35 12.80 13.25 13.40 13.40 13.70 13.50 13.60 13.50 13.50 14.50 15.80 16.10 16.20 18.10 18.00 18.25 18.45
4.20 4.25 4.95 4.22 4.60 6.35 10.45 10.65 10.75 10.95 11.30 12.25 12.35 12.80 13.20 13.35 13.40 13.60 13.45 13.45 13,40 13.50 14.45 15.80 16.10 16.10 17.95 18.00 18.25 18.30
K
K Actual
Curve
Correoted
0.932 0.931 0.923 0.917 0.927 0.926 0.923 0.913 0.915 0.903 0.906 0.915 0.917 0.912 0.916 0.917 0.915 0.915 0.914 0,911 0,910 0.910 0,912 0.914 0,909 0,909 0.913 0,909 0.907 0.899
0.927 0.927 0.923 0.927 0.924 0,920 0.913 0.913 0.913 0,913 0.913 0.912 0.912 0,912 0,912 0.912 0.912 0.912 0.912 0,912 0.912 0.912 0.911 0.909 0,909 0,909 0.907 0.907 0.907 0.907
3.89 3.94 4.57 3.91 4.26 5.84 9.54 9.72 9.81 10.00 10.32 11.17 11.26 11.67 12.04 12.1s 12.22 12.40 12 27 12.27 12.22 12.31 13.16 14,36 14.63 14.63 16.28 16.33 16.55 16.60 I
Mojonnier
Dif-
ferenoe
3.91 -0.02 3.96 -0.02 4.57 .... 3.87 -0.04 4.26 -0.01 5.88 -0.04 9.64 -0.10 .... 9.72 9.84 -0.03 9.89 so.11 10.24 $0.08 11.21 -0,04 11.32 -0.06 11.67 .... 12.09 -0.05 12.24 -0.06 12.26 -0.04 12.44 -0.04 12.29 -0.02 12.25 $0.02 12.19 +0.03 12.29 $0.02 13.18 -0.02 14.44 -0.08 14.63 14.63 .... 16.39 -0.11 16.36 -0.03 16.55 .... 16.45 +0.15
....
Although these results indicate that 90 per cent of the corrected Kniaseff tests agree within 0.1 per cent of the Mojonnier values and that no variation exceeds 0.15 per cent, the fact that Kniaseff duplicates differed by as much as 0.4 per cent indicates that this excellent state of affairs is partly adventitious. A larger series would doubtless prove the point. In conclusion, it is suggested that any individual Kniaseff f a t test, run as outlined above, will probably be within 0.4 per cent of the truth and that the mean of duplicates will be within 0.25 per cent.
Literature Cited (1) Bird and Johnson, Iowa Agr. Expt. Sta., Bull. 287 (1931). (2) Kniaseff, Ice Cream Trade J., 30,No. 12 (Dec., 1934). (3) Nelson, Ice Cream Trade J.,32,No.11,19-20 (Nov., 1936).