N.QN CIRCULATING CHECK FOR UNBOUND CIRCULATING COPY UNIVERSITY OF ILLINOIS Agricultural Experiment Station BULLETIN No. 263 RELATION OF SOLIDS IN MILK TO FAT AND SPECIFIC GRAVITY OF THE MILK BY 0. R. OVERMAN, F. A. DAVIDSON, AND F. P. SANMANN URBANA, ILLINOIS, APRIL, 1925 RELATION OF SOLIDS IN MILK TO FAT AND SPECIFIC GRAVITY OF THE MILK By O. R. OVERMAN, Assistant Chief in Dairy Chemistry, F. A. DAVIDSON, First Assistant in Dairy Husbandry, and F. P. SANMANN, Instructor in Dairy Chemistry in the College of Agriculture The use of formulas for computing the percentage of total solids or of solids-not-fat in milk has been common both in the United States and in Europe for many years. These formulas are based upon the specific gravity and the percentage of fat in the milk. The idea that a relation exists between specific gravity, fat, and solids in milk seems to have occurred first to Behrend and Morgen. 1 Clausnizer and Mayer, 2 and Hehner J> published formulas in the attempt to show this relation. These formulas were based on inaccurate data and have been abandoned. Fleischmann and Morgen 4 published a formula which was later corrected by Fleischmann. 5 This formula is T = .2665 S -f- 1.2 F, in which T = percentage of total solids, S = specific gravity of milk at 15 C., and F = percentage of fat in milk. Q Richmond has developed the formula T = .262 -f- 1.2 F, in which G = Quevenne lactometer reading, D = specific gravity, and F = percentage of fat, and has found that the simpler formula C^ f T= 1 F-J--14 gives results which correspond very closely with 4 5 it if the specific gravity is between 1.020 and 1.036. This formula is commonly used in England. Babcock 7 published the formula total solids = 1- f, in 3.8 which L = Quevenne lactometer reading and f = percentage of fat. Babcock 8 later stated this relation as total solids = - + 1.2 F. With the addition of .14 this is the same as Richmond's formula. This formula of Babcock is most commonly used in the United States and for that reason was selected for making all computations of total solids which are recorded in this bulletin. Babcock's formula for solids-not-fat as used at the present time is S.N.F. = f- .2 F -f- .14. 4 The purpose of this investigation was to show, by means of a statis- tical analysis, the relation existing between the percentages of total solids as determined by weight (A.O.A.C. method) and the correspond- 263 264 BULLETIN No. 263 [April, ing percentages of total solids as computed by the formula T.S. = [-1.2 F when applied to a large number of milk samples; also the 4 effect upon the results of the size of the lot of milk sampled. The relation between the percentages of solids-not-fat as determined by difference and the corresponding percentages of solids-not-fat as computed by the formula S.N.F. = \- .2 F also was studied. PLAN OF INVESTIGATION SOURCE OF MILK SAMPLES This investigation involved a statistical study of the percentages of total solids and of solids-not-fat by weight, and the corresponding per- centages of total solids and of solids-not-fat by formula, as determined from three different groups of milk samples, namely: (1) 1158 Samples from Individual Cows. Most of these samples were composites of the milk produced during three-day periods selected at regular intervals thruout the lactations of the cows.- A few of these samples represent only one milking. The cows used were all in the dairy herds at the University of Illinois and represent the Ayrshire, Guernsey, Holstein and Jersey breeds, and Guernsey-Holstein crosses. (2) 134 Random Samples of Mixed Milk. These samples were taken from cans of milk delivered at milk plants, from weigh-tanks, and from storage and pasteurizing vats. No record was kept of the sizes of the lots of milk from which the samples were taken; they varied, how- ever, from less than 10 gallons to 150 gallons or more. (3) 40 Samples Taken from Large Lots of Milk. These samples were all taken from storage or pasteurizing vats in plants which handle milk in large quantities. Records were kept of the sizes of these lots, which varied from 135 gallons to 3,000 gallons. In every case care was taken to make certain that the sample ob- tained was representative of the lot of milk from which it was secured. CHEMICAL ANALYSIS All samples were put into glass jars and securely sealed to prevent evaporation of the water from the samples. The samples were also pre- served with formaldehyde in approximately the quantity recommended by Palmer. 9 Determinations were made of specific gravity, percentage of fat, ancf percentage of total solids. The specific gravity was obtained at 15.5 C. with a chainomatic specific-gravity balance. At least two adjustments and readings of the vernier scale on the balance were made in determining the specific gravity of each sample. 1925] MILK. SOLIDS, FAT, AND SPECIFIC GRAVITY 265 percentage of fat was determined by the Roese-Gottlieb method, about 5 grams of milk being weighed into a Rohrig tube. The volumes of reagents used were reduced from those given in the methods of analysis 10 of the A.O.A.C. to correspond to the weight of milk used. Duplicate determinations were made in each case and the average of the duplicates reported as the percentage determined. The percentage of total solids was determined by weighing 2 to 3 grams of the sample into a weighed flat bottom lead dish and heating to constant weight at the temperature of boiling water. Duplicate deter- minations were made and the average of the duplicates reported as the percentage determined. The percentage of solids-not-fat was determined by subtracting the percentage of fat by weight from the percentage of total solids by weight. To avoid confusion, this percentage of solids-not-fat will be spoken of as percentage of solids-not-fat by weight. STATISTICAL ANALYSIS The percentage of total solids was computed from the specific gravity and percentage fat content of each sample according to the formula T.S. = -- 1- 1.2 F. The percentage of total solids by weight 4 for each sample was subtracted algebraically from its corresponding percentage of total solids by formula. In this way 1,158 differences were determined for the first group of samples, 134 differences for the second group, and 40 differences for the third group. 3 The mei/S*rences (Cta-ss Jnid -Points ) FIG. 1. FREQUENCY DISTRIBUTION OF THE DIFFERENCES BE- TWEEN TOTAL SOLIDS AND SOLIDS-NOT-FAT FOR THREE GROUPS OF SAMPLES: INDIVIDUAL, RANDOM, AND VAT in Fig. 1. Table 1 includes the mean, the standard deviation, and the limits at odds of 30:1 of the differences determined for each group of samples. Themeanof the differences 3 for thefirst groupof samples (samples of milk from individual cows) is .173 percent. In other words, the percent- "Differences between the percentages of total solids by weight and the correspond- L ing percentages of total solids by the formula T.S. = 1- 1.2 F. 4 19251 MILK SOLIDS, FAT, AND SPECIFIC GRAVITY 267 ages of total solids determined by weight for samples of milk from individ- ual cows are on the average .173 percent greater than the percentages of total solids computed for the same samples by the formula T.S. = -f- 1.2 F. The mean of the differences for the second group of samples (random samples from milk of more than one cow) is .105 percent. The mean of the differences for the third group of samples (samples from large vats of milk) is also .105 percent. Hence, the percentages of total solids determined by weight in milk from two or more cows and in milk from many cows, are on the average .105 percent greater than their corresponding percentages of total solids computed by the formula T.S. = + 1.2 F. 4 TABLE 1. MEANS AND STANDARD DEVIATIONS OF DIFFERENCES BETWEEN SOLIDS BY WEIGHT AND SOLIDS BY FORMULA Samples Number of differences Mean percentage Standard deviation, percent Limits at odds of 30:1 percent Individual 1158 1 73 -+- 0067 340-*- 0048 .173 .727 or +.554 and .900 Random 134 105 + 0014 242-*- 0010 .105.519 or +.414 and .624 Vat 40 105 - 1 - 0011 10Q-+- 0008 .105.214 or +.109 and .319 Altho the means of the differences for the three types of samples differ very little from each other, it does not follow that the formula is equally as accurate in computing the percentages of total solids within them, as will be shown later on. The standard deviation of the differences for the three groups of samples are .340, .242, and .100 percent respectively. The standard ^deviation is a 'measure of variability; hence, as the number of cows contributing to the milk samples is increased, the variability in the differences between the percentages of total solids by weight and their corresponding percentages of total solids by formula, is markedly de- creased. This relation is illustrated very clearly in Fig. 1, wherein the differences for each group of samples are shown graphically. In Fig. 1 it will be found that the differences for the first group of samples range from -f- 1.045 to 1.355 percent, for the second group of samples from -(-.845 to .555 percent, and for the third group of samples from -J-.145 to .255 percent. 268 BULLETIN No. 263 [April, Putting this variability into practical terms we have in Table 1 limits such that the odds, or chances, are 30:1 that any single difference determined by the above methods for each group of samples will fall within them. The limits for the first group of samples are -(-.554 and .900 percent; i. e., for samples from individual cows the chances are 30:1 that the percentages of total solids by the formula (T.S. = + 1.2 F) will lie within +.554 and .900 percent of their correspond- ing percentages of total solids by weight. The limits at odds of 30:1 for the second group of samples are +.414 and .624 percent, and for the third group of samples +.109 and .319 percent. Hence, as the number of cows contributing to the samples of milk are increased the limits are decreased within which the chances are 30:1 that the percentage of total solids computed by the above formula will deviate from its corresponding percentage of total solids by weight. L If the formula (T.S. = 1- 1.2 F) be corrected for samples of 4 milk from individual cows by adding the mean of the above differences for these samples, it will read T.S. = 1- 1.2 F +.173. In accord- 4 ance with this formula the limits at odds of 30:1 will also be corrected to .727 percent. The limits of .727 percent were determined by adding .173 percent to the above limits of +.554 and .900 percent. In like manner the formula T.S. = \- 1.2 F will be corrected for the 4 L second and third groups of samples to read T.S. = 1- 1.2 F + .105. 4 Altho the corrected formulas for the second and third groups of samples are the same, the limits at odds of 30:1 are much different and will be .519 percent and .214 percent respectively. Hence, it can readily be seen that the standard deviation or variability within the above dif- ferences determines the accuracy of the formula T.S. = 1- 1.2 F in" computing the percentages of total solids in milk from the respective sources. The mean of the above differences has a bearing on the accu- racy of the formula when combined with the standard deviation, but without the latter it has very little meaning. The probable errors of the means and the standard deviations of the differences as reported in Table 1 are all many times less than their constants. Hence, it may safely be assumed that the samples from which these constants were derived are representative of the general popula- tions of samples of the respective types. As the differences between the percentages of solids-not-fat by weight and of solids-not-fat by formula are identical with the corre- 1925] MILK. SOLIDS, FAT, AND SPECIFIC GRAVITY 269 spending differences for the total solids, the means, standard deviations, and limits at odds of 30:1 are the same. T The formula for solids-not-fat (S.N.F. = 1- .2 F) corrected for . 4 L samples of milk from individual cows will read S.N.F. f- 4 .2 F -f- .173; for the second and third groups it will read S.N.F. = (- .2 F + .105. SUMMARY The formula T.S. = 1- 1.2 F -{-.173 in computing the percent- age of T.S. in milk from individual cows is accurate in so far as the chances are 30:1 that the percentages of total solids computed by it will lie within .727 percent of their corresponding percentages of total solids by weight. In other words, it may be expected that, on the average, the percentages of total solids by the above formula for 30 samples out of every 31 samples will lie within .727 percent of their correspond- ing percentages of total solids by weight. Likewise, the percentage of total solids by the above formula for one sample out of every 3 1 samples will lie without the range of .727 percent of its corresponding per- centage of total solids by weight. The formula T.S. = \-l.2F-\- .105 in computing the percent- 4 age of total solids in random samples of milk from two or more cows is accurate in so far as the chances are 30:1 that the percentages of total solids computed by it will lie within .519 percent of their correspond- ing percentages of total solids by weight. Hence, it may be expected that on the average the percentages of total solids computed by the above formula for 30 samples out of every 31 samples will lie within the range of .519 percent of their corresponding percentages of total solids by weight. Likewise, the percentage of total solids by the above formula for one sample out of every 31 samples will lie without the range of .519 percent of its corresponding percentage of total solids by weight. T The formula T.S. = \- 1.2 F -|- .105 may also be used in com- puting the percentage of total solids in milk from large vats and storage tanks, i. e., milk from many cows. In using this formula to compute the percentage of total solids in such milk, it may be expected that on the average the percentages of total solids computed by it in 30 samples out of every 31 samples will lie within .214 percent of their corre- sponding percentages of total solids by weight. In like manner the per- centage of total solids by the formula for one sample out of every 31 samples will lie without the range of .214 percent. The formula S.N.F. = -|- .2 F +.173 in computing the per- 4 270 BULLETIN No. 263 {April, centage of solids-not-fat in the milk from individual cows is accurate in so far as the chances are 30:1 that the percentages of solids-not-fat computed by it will lie within .727 percent of their corresponding percentages of solids-not-fat by weight. L The formula S.N.F. = \- 2 F -f- .105 may be used to compute 4 the percentage of solids-not-fat in the mixed milk from two or more cows. For such samples, taken at random, the chances are 30:1 that the percentages of solids-not-fat computed by it will lie within .514 per- cent of their corresponding percentages of solids-not-fat by weight. For samples taken from large vats and storage tanks, the chances are 30:1 that the computed percentages of solids-not-fat will lie within .214 percent of their corresponding percentages of solids-not-fat by weight. CONCLUSIONS 1. The accuracy of the formula T.S. = \- 1.2 F in computing 4 the percentage of total solids in milk increases as the number of cows contributing to the milk is increased. 2. The formula T.S. = 1- 1.2 F -(-.173 may be used in com- puting the percentage of total solids in milk from individual cows, but the variability within the results is too great to make it of any prac- tical use. 3. The formula T.S. = (- 1.2 F -f- .105 when used to compute 4 the percentage of total solids in milk from many cows gives results which very closely approximate those determined by direct chemical analysis, and in plants handling large quantities of milk may be used with relative satisfaction. y 4. The accuracy of the formula S.N.F. = -\- .2 F in comput- ing the percentage of solids-not-fat in milk increases as the number of cows contributing to the milk is increased. 5. The formula S.N.F. = \- .2 F -f .173 may be used in com- 4 puting the percentage of solids-not-fat in milk from individual cows, but the variability within the results is too great to make it of any prac- tical use. j 6. The formula S.N.F. = (- .2 F -j- .105 may be used in com- 4 puting the percentage of solids-not-fat in large lots of milk. The results obtained approximate closely enough to those obtained by chemical analysis to be of practical value. 1925] MILK SOLIDS, FAT, AND SPECIFIC GRAVITY 271 LITERATURE CITED 1. BEHREND, PAUL, AND MORGEN, AUGUST Ueber die Bestimmung der Trockensubstanz in der Milch nach dem specifischen Gewicht derselben. Jour. Landw. 27, 249-259. 1879. 2. CLAUSNIZER, F., AND MAYER, ADOLF Forschungen auf dem Gebiete der Viehhaltung. 6, 265, 1879. Thru Tour Landw. 30, 293. 1882. 3. HEHNER, OTTO On the relation between the specific gravity, the fat, and the solids-not-fat in milk. Analyst, 7, 129-133. 1882. 4. FLEISCHMANN, WILHELM, AND MORGEN, AUGUST Ueber die Beziehungen welche zwischen dem specifischen Gewicht der Milch einerseits und dem procentischen Gehalt derselben an Fett und Trock- ensubstanz andrerseits bestehen. Jour. Landw. 30, 293-309. 1882. 5. FLEISCHMANN, WILHELM Beitrage zur Kenntnis des Wesens der Milch. Jour. Landw. 33, 251-267. 1885. 6. RICHMOND, H. D. Dairy Chemistry, 84-89. 1920. 7. The relation between specific gravity, fat, and solids-not-fat in milk. Analyst 20, 57-58. 1895. 8. BABCOCK, S. M. The estimation of the total solids in milk from the percent of fat and the specific gravity of the milk. Wis. Agr. Exp. Sta., Ann. Rpt., 1891, 292-307. 1892. 9. FARRINGTON, E. H. Report on dairy products. U. S. Dept. of Agr., Div. of Chem., Bui. 47, 122- 125. 1894. 10. PALMER, L. S. The preservation of milk for chemical analysis. Mo. Agr. Exp. Sta., Res. Bui. 34, 29. 1919. 11. ASSOCIATION OF OFFICIAL AGRICULTURAL CHEMISTS Official and tentative methods of analysis adopted by the Association of Official Agricultural Chemists, 227. 1920. ACKNOWLEDGEMENT The authors wish to express their appreciation of the courtesy shown them by the companies named below. The random samples and the samples from large lots of milk were for the most part obtained in plants operated by these companies: Blue Banner Dairy Company, Danville; Borden Farm Products Company, Chicago; Bredehoft Dairy Company, Danville; Champaign Sanitary Milk Company, Champaign; Horne- man-Cossey Company, Danville; Illinois Dairy Company, Springfield; Snow and Palmer, Bloomington; Wieland Dairy Company, Chicago. UNIVERSITY OF ILLINOIS-URBAN*