X I B R.AR.Y

OF THE

U N IVERSITY OF 1LLI NOIS

G>30.T

MfilCULTIHI

NOTICE: Return or renew all Library Materials! The Minimum Fee for each Lost Book is $50.00.

The person charging this material is responsible for its return to the library from which it was withdrawn on or before the Latest Date stamped below.

Theft, mutilation, and underlining of books are reasons for discipli- nary action and may result in dismissal from the University. To renew call Telephone Center, 333-8400

UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN

L161— O-1096

10-39 — 4050— 17321-S

Accuracy of Methods of Sampling Milk Deliveries at Milk Plants

By P. H. TRACY and S. L. TUCKEY

University of Illinois

Agricultural Experiment Station

Bulletin 459

CONTENTS

PAGE

WORK OF OTHER INVESTIGATORS 47

PLAN OF PRESENT STUDY 50

COMPLETENESS OF MIXING AT THE FOUR PLANTS. . 53

Plant A 53

Plant B 53

Plant C 55

Plant D 55

Change in Plant D sampling 61

Decision concerning sampling technic 65

COMPARISON OF AVERAGES OF TESTS ON DAILY

SAMPLES AND ON COMPOSITES 65

Winter samples 65

Summer samples 65

COMPARISON OF TESTS ON DAILY SAMPLES AND

CALCULATED TESTS ON TRUE COMPOSITES.... 69

GENERAL DISCUSSION OF RESULTS 77

SUMMARY AND CONCLUSIONS 83

LITERATURE CITED . . 84

ACKNOWLEDGMENT

Acknowledgment is made of the cooperation of the Champaign County Milk Producers Association in the promotion of the studies reported in this bulletin.

Urbana, Illinois November, 1939

Publications in the Bulletin series report the results of investigations made or sponsored by the Experiment Station

Accuracy of Methods of Sampling Milk Deliveries at Milk Plants

By P. H. TRACY and S. L. TucKEY1

I HE IMPORTANCE of an accurate measurement of the fat contained in milk deliveries is fully appreciated by most pro- ducers and distributors. However, since the beginning of the present method of marketing milk according to fat content, the accuracy of the procedure used in sampling and testing the milk has often been questioned by either the buyer or the seller. At the sug- gestion of the Champaign County Milk Producers Association a study was begun in the fall of 1936 to determine the accuracy of the methods being employed to sample the milk delivered by members of the Asso- ciation to each of four milk plants in Champaign and Urbana. This bulletin is a report of that study.

WORK OF OTHER INVESTIGATORS

Investigators began studying the relative merits of the daily (fresh), periodic (fresh), and composite milk samples almost imme- diately after the introduction of the Babcock test for determining the fat content of milk in 1890. This same year G. E. Patrick8* proposed a plan whereby an amount of milk proportionate to that delivered was kept and placed in a receptacle containing a certain amount of a pre- servative. Later such a sample was called a "composite sample." In a later publication Patrick7* stated that if a patron's deliveries ran fairly uniform in amount from the beginning to the end of a composite period, the taking of uniform-size samples was correct enough; but that if there were wide variations in the weight of milk delivered daily, the amount of the sample should be taken in proportion to the amount delivered.

In 1891 E. H. Farrington2* of the University of Illinois reported that testing composite milk samples once each week gave results prac- tically as accurate as testing milk every day. He published the results

'P. H. TRACY, Chief in Dairy Manufactures, and S. L. TUCKEY, Assistant Chief in Dairy Manufactures.

*These numbers refer to literature citations on page 84.

47

48 BULLETIN No. 459 [November,

of an experiment in which daily samples and composite samples of twenty patrons' milk were tested. The results were as follows:

Average percent fat

7 daily tests for each patron 3.91

Composite (same amount) 3 . 96

Composite (aliquot sample taken each day) 3 . 93

Farrington further stated that a single sample and test of milk only once in a week might not be sufficiently accurate.

Hunziker3* in 1914 reported a remarkable uniformity of results when comparing the accuracy of different methods of sampling includ- ing daily samples, composite samples with aliquot portions, samples every fourth day and every fifth day.' Tests were made on 4,900 samples taken by these methods over a 14-day period.

Sanmann and Overman10* in 1926 studied the importance of proper storage of composite samples. They found that nearly all the samples stored in the receiving room tested lower than the samples stored in the refrigerator, the differences being greater when the samples were held for two weeks than when they were held for only one week. The following data compiled from their publication involve milk deliveries

Samples stored one week Samples stored two weeks

Receiving Receiving

room Refrigerator room Refrigerator

Fat test, percent 3.6 3.81 3.57 3.84

In a continuation of this study Sanmann and Overman9* made a comparison of tests secured on periodic, composite, and fresh daily samples of the milk delivered by twenty patrons. The composite samples were prepared by taking one milliliter of milk for each pound of milk delivered. At the same time a sample was taken for the daily test. The composite samples were mixed carefully each day after adding the fresh portion. The samples were kept in one-quart fruit jars sealed and stored in a refrigerator at about 44° to 50° F. They were preserved by corrosive-sublimate tablets and extended over a month's time divided into four periods — 7 days, 7 days, 8 days, and 8 days.

The following averages are compiled from their data:

Percent fat

Average of daily tests of samples taken by aliquot 4. 10

Average test of composite samples 4 . 08

Average test for 4 fresh milk samples* 4.19

Average test for 5 fresh milk samples8 4.12

("Taken at approximately equal intervals during the month.)

1939] ACCURACY OF METHODS OF MILK SAMPLING 49

From these averages it is very evident that the fat test of composite samples properly taken and kept is comparable to the fat test of fresh daily samples. The data also indicate that under the conditions of the experiment the average of four or five periodic tests made on fresh samples taken at approximately equal intervals during the month is comparable to the average daily test, there being a slightly closer correlation when the averages were based on five tests than when they were based on four tests.

C. F. Monroe6* in 1930 reported that the average fat test of 290 seven-day composite samples averaged 5.13 percent, while the fresh daily samples averaged 5.22 percent.

Marquardt and Durham4* in 1932 studied the milk sampling at milk plants to find out whether or not milk is sufficiently agitated in dumping to make it possible to secure an accurate sample without further mixing. They concluded that stirring the milk before or after dumping did not improve the uniformity of the sample. They recom- mended, however, that each weigh tank be checked for its correctness for proper sampling, since such things as shape of the tank and type of strainer vary from plant to plant. They concluded that natural varia- tions in the milk test cause some of the variations in tests obtained by the milk plant. The authors then explained the relation of certain factors to the fat content of milk. Among these factors the following were mentioned:

1. During the first part of the lactation period the milk tests higher.

2. The test is highest during the cold season of the year and lowest in lidsummer.

3. Short intervals between milkings raise the fat test.

4. Omitting the foremilk raises the fat test of the milk, while omitting the stripping lowers the fat test.

5. Some breeds (as the Jersey) produce richer milk than other breeds (as the Holstein).

6. Night's milk will test higher than morning's milk.

7. Exercise increases the fat test.

8. Low temperatures cause the milk to test higher.

9. Underfeeding results in an increased fat test in the milk.

10. As cows grow older, their milk becomes lower in fat content.

Bailey1* in 1934 reported a two-year study of the accuracy of sampling of the milk delivered by 19 patrons. He found that the milk did not mix adequately when dumped into the weigh tank ; and that, after such dumping, nine out of ten of the lowest testing samples were at the front end of the tank. He attributed the inadequate mixing to the dumping of milk that has creamed. The low-testing milk, being the

50 BULLETIN No. 459 {November,

last dumped, tends to remain on top. He also noted that low tests some- times resulted from the adherence of thick cream to the strainer box until after the milk was allowed to run out of the weigh tank. It was found that the inadequate mixing could be eliminated by stirring the cans before dumping and that low-testing pools in the dump tank could be avoided by the use of a mechanical weigh tank agitator.

In 1936 Meade and Leckie5* compared composite and fresh samples taken from milk delivered by nine patrons during a 151-day period. The composite samples covered a period of 10 days, and one fresh milk sample was taken during each 10-day period. The periodic fresh samples had an average test of .09 percent higher than the average test of the composite samples. Considerable variation in the test of the milk delivered by the individual patron was also observed. The range by composite samples was .60 to 1.20 percent and the periodic fresh samples, .55 to 1.35 percent.

From the foregoing survey of past work, it may be concluded that:

1. Composite samples will give accurate results provided they are: (a) taken in proportion to the amount of milk delivered (this is particularly important when there is a wide variation in the amount of milk delivered daily); (b) placed in closed containers; (c) held in the refrigerator; (d) preserved by a germicidal agent, such as corrosive sublimate, and properly mixed after the addition of each fresh sample; (e) kept for a period of time not exceeding two weeks but preferably one week.

2. Composite tests and the average of daily tests on the fresh milk will check within the range of experimental error, altho the composite tests tend to average slightly lower.

3. Periodic samples taken at least four times a month will give average results that will check reasonably close to the average of daily tests.

4. Improper mixing of the milk in the weigh tank is sometimes respon- sible for discrepancies in tests.

5. Natural variations in the composition of the milk as produced will account for some of the variable tests reported by distributors.

PLAN OF PRESENT STUDY

The standard sampling procedure at each of the four dairies in this study was as follows: The plant employees dumping the milk took the composite milk samples daily either directly from the milk cans or from the dump tanks. These samples were kept in Mojonnier sample bottles stored either in the milk-receiving room or in the refrigerator. They were tested four times each month by an operator employed and paid jointly by the producers and distributors. This test is called the Asso-

1939] ACCURACY OF METHODS OF MILK SAMPLING 51

elation test. To study the accuracy of the sampling procedure at these plants, the following steps were taken:

1. The completeness of mixing of the milk at each of the four dairies was determined.

2. Comparisons were made of the tests made on fresh daily samples, regular plant composite samples, and laboratory composite samples over seven-day test periods. The procedure of the testing was as follows:

a. During the period of December 22 to January 4 inclusive samples were taken daily at each of the four plants from the milk delivered by each patron. These samples were obtained by a representa- tive of the University, placed in half-pint bottles and taken to the University laboratory. In addition a composite sample was taken by the plant and tested by the Association tester in the regular manner. A test was later run on this plant composite at the University laboratory.

b. At the University laboratory composite samples were prepared from the fresh daily samples. The composites were kept in double- capped quart bottles stored at 60° F. Approximately 18 grams of milk were taken for the composite sample each day.

c. The fresh samples were tested daily in single tests at the University laboratory.

d. Owing to the large number of tests to be run, the labor was so divided that one man performed the same task each day. These tasks were: the preparation of composites, measuring of samples, adding of acid and mixing, operation of centrifuges and 130°-140° F. bath, reading and recording of results, and the washing of test bottles.

e. The standard Babcock method of testing was followed. The temperature of the acid and the amount used was such that the fat columns were free from charred fat or curd particles.

/. Additional studies were made later in the season, one in May and one in July. For these summer tests the same general procedure was followed as for the winter tests.

g. All glassware including test bottles (10 percent graduated to .1 percent) and pipets were checked for accuracy.

3. Tests were made to determine the importance of taking composite samples in aliquot portions.

U. OF ILL LIB.

52

BULLETIN No. 459

[November,

Fig. 1. — Receiving room, Plant A

1939]

ACCURACY OF METHODS OF MILK SAMPLING

53

COMPLETENESS OF MIXING AT THE FOUR PLANTS

To determine the completeness of the mixing of the milk sampled at the four dairies, tests were run on the milk delivered by a number of the patrons at each dairy. The number of patrons serving each dairy, together with the amount of milk delivered daily is shown in Table 1.

Plant A. At this plant samples were taken over a 3-hour period from a round weigh tank. The milk was poured from the cans at a height of about 30 inches. A sample was taken directly from the tank after the milk was poured in (the usual procedure). The milk was then stirred and as it flowed thru the discharge valve of the weigh tank, another sample was taken. The fat tests of the mixed and unmixed samples are given in Table 2.

From these data and a comparison of the averages of the tests on the unmixed and the mixed samples (4.77 percent fat and 4.76 percent), it is evident that the method of sampling at Plant A was satisfactory and that nothing would be gained by stirring the milk before sampling.

Plant B. Since this plant was not equipped with a weigh tank, the samples were taken directly from the cans after stirring. Com- parisons were made between samples taken with a lipped stirring rod (the usual procedure) and those taken with a milk thief, which should give a more nearly aliquot portion, as it takes the sample in proportion to the volume of milk in the can. The use of the thief would seem particularly advisable when the farmer delivered his milk in two or more cans with milk varying in amount and test in each can. The results of the sampling at Plant B are given in Table 3.

The summary of the data in Table 3 shows very plainly that the method of securing the sample at Plant B was not in error and that under the conditions of the experiment, the use of the lipped stirring rod dipper gave as accurate results as the use of a milk thief.

TABLE 1. — AVERAGE AMOUNT OF MILK DELIVERED DAILY TO EACH PLANT

Plant	Number of producers	Average amount of milk delivered daily per producer	Average amount of milk delivered to plant daily by all producers
A. .	136	Ib. 78	Ib. 10 608
B	64	77.1	4 934
C. .	63	81	5 103
D	170	79,8	13 566
Total	433	79	34 210

54

BULLETIN No. 459

{November,

Fig. 2. — Receiving room, Plant B

1939~\ ACCURACY OF METHODS OF MILK SAMPLING 55

TABLE 2. — TESTS OF MIXED AND UNMIXED SAMPLES: PLANT A

Sample No	Fat test of sample taken —	Approximate amount of milk delivered
Before mixing	After mixing
1	perct. 4.2 5.0 4.55 5.7 5.1 5.5 4.4 4.3 5.0 4.9 5.8 4.6 5.0 5.05 4.5 3.3 4.8 4.9 4.75 5.0 4.8 4.9 4.25 4.4 5.25 4.6 4.6 4.2 4.3 5.4 4.77	perct. 4.1 5.05 4.5 5.7 5.1 5.6 4.4 4.4 4.8 4.9 5.9 4.5 5.0 5.1 4.5 3.3 4.9 4.9 4.75 5.0 4.85 4.9 4.1 4.4 5.0 4.6 4.7 4.15 4.3 5.4 4.76	gal. 20 25 5 11 15 10 20 10 10 20 8 20 12 6 10 5 13 14 8 12 8 22 10 18 8 8 5 20 15 10
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22 ...
23
24
25
26
27
28
29
30

Summary

Number of times two tests were identical

Number of times two tests checked within .1 percent. . Number of times two tests checked within .2 percent. . Number of times two tests checked within .25 percent.

13 (43.3 percent) 27 (90.0 percent)

29 (96.7 percent)

30 (100.0 percent)

Plant C. As in Plant B, the usual procedure was to take the samples from the milk cans after mixing with a lipped stirring rod. Here again, comparisons were made between the tests secured on samples taken in this manner and those taken with a milk thief.

As in the other two plants, the method of sampling followed in Plant C was found to be entirely satisfactory. The data are given in Table 4.

Plant D. This plant used a weigh tank. Ordinarily, the receiv- ing man took the samples from the front end of the weigh tank, using a small sampling dipper. For the purpose of this study samples were taken at the front and rear before mixing. The milk was then mixed with a stirring rod and a third sample taken.

It is evident from the data in Table 5 that mixing resulting from

56

BULLETIN No. 459

[November,

Fig. 3. — Receiving room, Plant C

1939}

ACCURACY OF METHODS OF MILK SAMPLING

57

the dumping of the milk into the weigh tank was not sufficient to make it possible to secure an accurate sample from either end of the tank without additional agitation. Whereas the average test of the samples

TABLE 3. — TESTS OF SAMPLES TAKEN WITH LIPPED STIRRING ROD AND WITH

MILK THIEF: PLANT B

Sample No.	Fat test of sample taken —	Approximate amount of milk delivered
With rod	With milk thief
1	perct. 5.5 4.0 5.0 3.4 4.5 4.9 4.7 5.25 4.5 4.3 4.65 4.1 4.5 4.55 5.2 4.7 3.8 4.75 4.35 4.6 3.4 3.9 4.6 4.2 4.3 3.9 4.75 4.65 4.2 4.4 4.5 4.6 4.8 4.47	perct. 5.4 4.1 5.0 3.45 4.4 5.0 4.7 5.25 4.5 4.3 4.6 4.3 4.4 4.5 5.2 4.6 3.8 4.7 4.3 4.6 3.4 3.95 4.4 4.2 4.3 3.9 4.7 4.75 4.2 4.5 4.45 4.6 4.7 4.45	gal. Two 5-gal. cans % full One 5-gal. can % full Two 10-gal. cans Hfull Three 8-gal. cans full Two 10-gal. cans % full One 8-gal. can full One 8-gal. can Yz full One 10-gal. can % full Two 8-gal. cans % full Two 10-gal. cans %full One 10-gal. can full One 8-gal. can full One 5-gal. can full Two 8-gal. cans % full Two 10-gal. cans %full Two 10-gal. cans % full One 5-gal. can full Five 5-gal. cans full Two 5-gal. cans full and % full Two 10-gal. cans full and Y* full Two 5-gal. cans % full Four 5-gal. cans % full One 8-gal. can Y* full Two 5-gal. cans full One 5-gal. can J^ full One 10-gal. can full One 8-gal. can full One 5-gal. can Yz full Two 8-gal. cans % full One 10-gal. can full One 8-gal. can full One 5-gal. can Y% full Two 8-gal. cans % full One 10-gal. can Yi full One 8-gal. can Yz full One 8-gal. can % full One 5-gal. can % full One 10-gal. can Yz full One 8-gal. can % full Three 10-gal. cans % full One 8-gal. can % full One 10-gal. can Mfull Two 8-gal. cans full One 10-gal. can yz full One 5-gal. can J^ full One 5-gal. can Ys full Two 5-gal. cans full Two 8-gal. cans full Two 8-gal. cans Yt full
2
3
4 . . .
5
6
7
8
9
10
11
12
13
14
15.. .
16
17
18
19
20..
21
22...
23
24...
25
26
27
28
29
30
31...
32...
33
Average

Summary

Number of times two tests were identical

Number of times two tests checked within .1 percent. Number of times two tests checked within .2 percent.

14 (42.4 percent) 31 (93.9 percent) 33 (100.0 percent)

58

BULLETIN No. 459

[November,

Fig. 4. — Receiving room, Plant D

1939}

ACCURACY OF METHODS OF MILK SAMPLING

59

taken from the front of the tank was only .11 percent lower than the average test of the sample taken after mixing, the discrepancy between certain samples was much greater, as shown by the distribution of differences in tests of unmixed and mixed samples (Table 6).

From these differences it is very apparent that there is a definite trend towards lower tests in the front end samples, particularly when variations higher than .2 percent are considered. Above .2 percent it will be noted that 27 front samples tested less than the rear samples,

TABLE 4. — TESTS OF SAMPLES TAKEN WITH LIPPED STIRRING ROD AND WITH

MILK THIEF: PLANT C

Sample No.	Fat test of sample taken —	Approximate amount of milk delivered
With rod	With milk thief
1.. .	perct. 5.0 3.6 4.9 4.2 4.8 4.4 3.7 4.4 3.9 5.35 5.5 4.4 3.9 4.3 5.35 4.2 5.2 4.6 4.3 4.6 3.8 4.5 3.45 5.3 5.0 4.1 4.8 4.5 4.50	perct. 5.0 3.7 4.9 4.2 4.8 4.4 3.75 4.4 3.9 5.3 5.55 4.45 3.9 4.2 5.4 4.2 5.2 4.6 4.3 4.6 3.8 4.5 3.45 5.3 5.0 4.0 4.85 4.5 4.51	gal. Two 5-gal. cans ^ full Two 10-gal. cans }$ full Three 5-gal. cans % and Y* full Two 10-gal. cans y% full Two 5-gal. cans Vi full Two 5-gal. cans y% full Two 5-gal. cans Y$ and Yi full One 8-gal. can full One 5-gal. can full One 8-gal. can full One 5-gal. can yz full Two 8-gal. cans y? full One 5-gal. can full One 8-gal, can % full Two 8-gal. cans full One 10-gal. can % full Two 10-gal. cans ^ and %full Two 5-gal. cans full Two 8-gal. cans full One 10-gal. can full One 10-gal. can yz full One 10-gal. can y% full One 5-gal. can full Two 10-gal. cans full One 10-gal. can full Two 5-gal. cans full One 10-gal. can y2 full One 8-gal. can yz full Two 10-gal. cans full Two 10-gal. cans full One 5-gal. can % full One 10-gal. can full One 8-gal. can full Two 10-gal. cans % full Two 8-gal. cans $4 full Two 5-gal. cans % full Two 5-gal. cans full
2.
3
4.. .
5
6
^
8
9
10
11..
12..
13
14..
15..
16
17..
18..
19..
20
21
22...
23
24
25...
26
27
28
Average

Summary

Number of times two tests were identical

Number of times two tests checked within .1 percent.

(67.8 percent) (100.0 percent)

BULLETIN No. 459

[November,

TABLE 5. — TESTS OF SAMPLES TAKEN AT FRONT AND REAR OF WEIGH TANK BEFORE MIXING, AND TAKEN AFTER MIXING IN THE WEIGH TANK: PLANT D

Sample No.	Weight of milk	Fat test of sample before mixing taken —	Fat test of sample after mixing
From front of tank	From rear of tank
1	Ib. 110 167 137 308 165 100 185 66 30 140 50 170 80 75 95 60 97 84 100 100 48 101 85 130 95 95 170 35 40 25 80 130 140 190 55 60 90 60 140 100 50 65 60 190 215 60 150 70 80 90 95 125 60 75 60 40 110 160 60 45 60 75 140 65 60 100 65 15 110 90	perct. 4.0 4.8 3.9 3.6 4.3 3.9 4.2 4.5 4.95 4.4 4.3 3.6 3.35 3.75 4.8 4.4 4.7 4.2 3.25 3.6 5.05 6.5 4.8 3.3 4.6 4.65 5.1 4.4 3.75 4.3 4.7 3.7 3.95 4.7 3.85 5.2 4.8 4.3 4.4 4.0 5.1 5.2 5.2 3.9 5.4 4.0 3.35 3.6 4.2 3.95 3.85 3.15 4.4 3.1 5.0 5.0 4.4 5.1 5.0 4.35 4.5 5.8 4.75 4.6 4.0 5.2 5.0 4.95 5.0 4.5 4.5 4.4 4.40	perct. 4.4 4.95 3.95 3.75 4.35 3.8 4.15 4.55 5.5 4.75 5.0 3.5 3.8 3.7 4.85 4.35 5.1 .4.95 4.1 4.6 5.85 6.3 4.4 3.3 5.1 5.3 4.9 4.8 4.1 5.0 4.8 4.2 3.9 5.0 4.2 5.2 4.75 4.3 4.45 4.0 5.2 5.2 5.3 3.85 5.6 3.9 3.2 3.6 4.6 4.8 4.4 3.3 4.5 3.7 5.1 5.4 4.4 5.1 4.8 4.45 4.5 5.9 5.0 4.9 4.1 5.3 5.3 5.2 5.3 4.45 4.4 4.5 4.60	perct. 4.2 4.85 3.95 3.8 4.4 3.85 4.2 4.55 5.4 4.8 4.7 3.5 3.6 3.7 4.9 4.35 5.0 4.6 3.7 4.2 5.4 6.3 4.55 3.3 4.95 5.1 5.1 4.6 3.9 4.8 4.7 3.9 3.9 4.85 4.3 5.2 4.8 4.45 4.3 4.0 5.5 5.1 5.35 3.7 5.5 3.95 3.5 3.6 4.45 4.3 4.0 3.2 4.4 3.2 5.0 5.15 4.4 5.1 5.0 4.3 4.5 5.9 5.0 4.4 4.1 5.3 5.1 4.95 5.1 4.45 4.0 4.4 4.51
2
3 .
4
s
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35. .
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50 .
51
52. .
53. . .
54
55
56
57
58
59 ;
60. .
61
62 ...
63
64. ..
65
66 .
67
68
69
70. ..
71
72
Average

1939}

ACCURACY OF METHODS OF MILK SAMPLING

61

TABLE 6. — DISTRIBUTION OF DIFFERENCES IN FAT TESTS OF 72 SETS OF FRONT AND REAR SAMPLES, AND MIXED SAMPLES: PLANT D

Variation range	Times front sample tested less than —	Times front sample tested more than —
Rear sample	Mixed sample	Rear sample	Mixed sample
perct. 0 - .1	15 4 6 8 3 3 3 1 2 0 1 46	13 11 4 6 6 1 0 0 0 0 0 41	11 5 0 1 0 0 0 0 0 0 0 17	9 3 1 0 1 0 1 0 0 0 0 15
15- 2 . . .
25- .3
35- 4 .
45- .5
.55- .6
65- .7
. 7 - .75
8 - .85
.9 - .95
95-1 00
Total

Summary

Times front and rear sample were the same . . Times front and mixed sample were the same. Times rear and mixed sample were the same. .

while only one front sample tested more than the rear sample. In the same range, 17 of the front samples tested less than the well-mixed samples, while only 3 tested more. These results compare favorably with those of Bailey.1*

Examination of the weigh tank at Plant D (Fig. 5) revealed a pos- sible explanation for the improper mixing of the milk when dumped. The shape of the tank was such that the first milk out of the can, which was often higher in test than the remainder, rushed to the rear and was held there to a certain extent by the last milk from the can, so that there was very little back washing or mixing. The amount of milk dumped did not seem to be a factor of any consequence.

To further show that the discrepancies between tests on front and rear samples were due to improper mixing, a test was made on the milk from twenty patrons which was thoroly mixed in the can by stirring before it was dumped. The distribution of differences of the tests on front and rear samples is shown in Table 7. It is evident that when the milk was properly mixed, front and rear samples tested practically the same.

Change in Plant D sampling. Since the stirring of milk in the can or after dumping is not a practical procedure from the standpoint of plant costs, and since the results of the study of the accuracy of the weigh tank in Plant D had indicated the front end samples to test

62

BULLETIN No. 459

[November,

•20 (A) WEIGH TANK

FILTERS RETAINED FILTERS DISCARDED

(C) SIDE VIEW OF WEIGH TANK

STRAINER

MILK OUTLET

ORIGINAL COVER FOR WEIGH TANK REVISED COVER FOR WEIGH TANK

(b)	— /4|"-*|	± T H* J -H/KT ** _i_ JL '-* lie?
SECTION FOR STRAINER

		-HrfK-
SECTION FOR STRAINER
(a) *£
t

T 1

K

•45?-

Fig. 5.— Construction details of weigh tank and strainer at Plant D Dumping milk into this weigh tank (A) failed to mix the milk thoroly, and consequently the samples taken from the front of the tank differed from those taken from the rear by an average of .2 percent of butterfat. By changing the strainer (B) so that it had niters only on the bottom, and by changing the cover (D) so that all samples were taken thru one opening (a, revised cover) located at one side of the center in front of the strainer, more accurate sampling was obtained.

1939}

ACCURACY OF METHODS OF MILK SAMPLING

63

TABLE 7. — COMPARISON OF TESTS ON TWENTY SETS OF SAMPLES TAKEN FROM

FRONT AND REAR OF VAT WHEN MILK WAS MIXED IN

CAN BEFORE DUMPING: PLANT D

Variation range	Times front sample tested less than rear sample	Times front sample tested more than rear sample
perct. 01-. 1	2	1
15-. 2	1	0
No difference. ..	1	5

rather consistently lower than the rear end samples, a new sampling opening was cut in the top of the weigh tank half way between the two ends, but still convenient to the operator. To check the accuracy of the samples taken from this location, the milk from one load was sampled daily for a period of one week.

A sample of each patron's milk was taken from the center of the weigh tank before and after mixing. These samples were tested daily and experimental composite samples were prepared. The plant also took its usual composite sample. The results of the daily tests on the fresh samples are shown in Table 8, while the averages of tests on daily and composite samples are given in Table 9. In this experiment the sampling and testing was all done by the same operator, the tests being performed in duplicate, the average test being reported in each

TABLE 8. — DAILY TESTS BEFORE AND AFTER MIXING, WHEN SAMPLES WERE TAKEN FROM CENTER OF DUMP TANK: PLANT D

Patron

Fat test of samples taken on dates indicated

No.	5-17	5-18	5-19	5-20	5-21	5-22	5-23
	perct.	perct.	perct.	perct.	perct.	perct.	perct.
	(B)a (A)«	(B) (A)	(B) (A)	(B) (A)	(B) (A)	(B) (A)	(B) (A)
201.	4.75 4.80	4.85 4.90	4.65 4.80	4.75 4.75	5.00 5.10	4.30 4.30	4.70 4.70
202.	4.05 4.15	4.10 3.90	4.25 3.90	4.10 4.00	4.15 4.00	3.50 3.70	3.50 3.50
204.	3.70 3.70	4.20 4.20	3.30 3.35	3.60 3.55	3.90 3.85	3.60 3.70	3.50 3.55
205.	3.80 3.75	3.75 3.75	3.70 3.75	3.55 3.50	3.50 3.50	3.60 3.60	3.40 3.40
206.	4.70 4.70	4.80 4.85	4.70 4.70	5.00 5.00	4.60 4.60	4.50 4.50	4.40 4.40
209.	4.30 4.40	4.40 4.50	4.20 4.30	4.30 4.20	3.90 3.95	4.30 4.35	4.10 4.10
210.	3.10 3.10	3.00 3.00	3.00 2.85	3.10 3.00	3.00 2.95	2.80 2.80	2.85 2.80
211.	4.20 4.10	4.70 4.75	4.40 4.10	3.95 3.80	4.15 4.20	4.15 4.10	3.50 3.70
212.	4.00 4.20	4.20 4.10	4.55 4.70	4.20 4.25	4.40 4.35	4.30 4.30	3.90 4.10
214.	5.20 5.30	4.70 4.70	4.10 4.20	4.90 4.60	4.30 4.20	4.85 4.80	4.05 4.10
216.	4.75 4.85	5.00 4.80	5.15 5.10	5.50 5.00	5.15 5.10	4.75 4.70	5.10 4.90
219.	3.90 3.90	3.80 3.60	4.65 4.60	4.20 4.05	4.20 4.10	3.80 3.80	3.90 3.80
220.	3.80 3.80	2.90 3.10	3.50 3.65	3.35 3.30	3.30 3.30	3.00 3.10	3.40 3.40
222.	4.55 4.60	4.75 4.80	5.00 5.00	5.00 5.00	5.30 5.30	4.80 4.90	5.10 5.10
223.	3.85 3.80	3.50 3.50	3.60 3.75	3.85 3.85	3.60 3.60	3.90 3.70	3.80 3.80

"B = before mixing, A = after mixing.

64

BULLETIN No. 459

[November,

TABLE 9. — AVERAGE TESTS OF DAILY SAMPLES BEFORE AND AFTER MIXING COM- PARED WITH TESTS OF COMPOSITE SAMPLES: PLANT D

Patron No.	Average of daily tests	Test of experimental composite	Test of plant composite
201	Perct. (B)» (A)- 4.77 4.65	perct. (B) (A) 4.65 4.65	perct. 4.75
202	3 . 85 3 . 80	3.85 3.80	3.80
204 .	3.70 3.70	3.60 3.65	3.50
205	3.60 3.55	3.70 3.55	3.50
206	4 70 4.60	4.50 4.60	4.60
209	4.20 4.15	4.10 4.15	4.05
210	2.90 2.90	2.90 2.90	2.90
211	4.10 4.00	4.05 4.00	4.00
212	4.28 4.15	4.30 4.15	4.30
214	4.55 4.50	4.50 4.50	4.30
216	4.90 4.70	4.90 4.70	4.50
219 .	3.95 3.90	3.85 3.90	3.85
220	3.37 3.30	3.30 3.30	3.30
222 .	4 95 4.90	4.90 4.90	4.80
223	3.70 3.70	3.70 3.70	3.60
Average	4.10 4.03	4.05 4.03	3.98

•B = before mixing, A = after mixing.

case. In Table 10 will be found the distribution of differences between the tests on samples taken before and after mixing.

Tables 8, 9, and 10 show that sampling the milk from the center of the dump tank before mixing gave results comparable with those obtained after mixing. A close correlation between the average test of daily samples and that of the experimental composite was also obtained. While the average test of the plant composite was lower than that of the experimental composite taken before mixing, the difference was not great enough to be considered serious. A wide variation in the daily tests of the milk from the same farm is evident in several cases. For example, the milk delivered by patron 220 varied in fat content from 3.1 to 3.8 percent; that from patron 219 varied from 3.6 to 4.6

TABLE 10. — DISTRIBUTION OF DIFFERENCES BETWEEN TESTS OF FIFTEEN PAIRS

OF DAILY SAMPLES AND FIFTEEN PAIRS OF COMPOSITE SAMPLES

TAKEN BEFORE AND AFTER MIXING"

Variation range	Number of pairs of daily samples with tests in range indicated	Number of pairs of experimental composite samples with tests in range indicated
Perct. None	3	6
.01-. 05. .	6	5
.06-. 10	3	1
.11-. 15	2	2
.16-. 20	1	1

•One sample of each pair was taken before mixing, the other sample, after mixing.

1939]

65

percent; that from patron 214 varied from 4.1 to 5.3 percent; and the milk delivered by patron 211 varied from 3.7 to 4.75 percent.

Decision concerning sampling technic. For the remainder of the study pertaining to a comparison of tests of daily and composite samples, it was decided that the usual procedure followed in Plants A, B, and C would be accepted, but that in Plant D the milk would be stirred thoroly in the cans before being dumped, in order to enable the plant operator to sample at the front end while one of the investi- gators was sampling at the rear end, each one thereby obtaining a sample the accuracy of which could not be questioned.

COMPARISON OF AVERAGES OF TESTS ON DAILY SAMPLES AND ON COMPOSITES

Winter samples. From December 22, 1936, to January 4, 1937, samples were taken daily from all the deliveries at four plants, as explained on page 51. The composite samples were tested at the end of each seven-day period. The milk delivered by about 425 patrons was included in this experiment. The information was not complete on the

TABLE 11. — AVERAGES OF FAT TESTS OF FRESH AND COMPOSITE MILK SAMPLES: WINTER SAMPLES, ALL PLANTS

	First period (348 samples)	Second period (343 samples)
	perct. 4.702	perct. 4.739
Tests of experimental composite	4.641	4.713
	4.571	4.601
	4.539	4.564

milk of some patrons, owing to such uncontrollable factors as loss of sample or failure of the farmer to make delivery each day. In such cases the available data were not included in the calculated averages.

A summary of the data obtained on the samples taken at the four plants is given in Tables 11 and 12. The results of the tests on the various samples are compared in such a way as to show the distribu- tion of differences by .1 -percent intervals. The extent to which the test of the milk delivered by each patron varied within a period of one week is shown in Table 13.

Summer samples. From July 8 to July 14, 1937, samples were taken daily from the milk delivered by 50 patrons at each of two

66

BULLETIN No. 459

[November,

dairies (A and D). The same general procedure was followed as in the experiments conducted December 22 to January 4. The composite samples taken by the University representatives as well as those taken by the plants, with the exception of Plant D, were stored at 40° F. In Plant D the samples were stored in the receiving room.

A summary of the results of the tests made on the samples taken at Plants A and D during the summer are given in Tables 14, 15, and 16. These data are presented in the same manner as those given in Tables 11, 12, and 13. Laboratory tests of the Plant A composites were not available for comparison, however.

TABLE 12. — DISTRIBUTION OF DIFFERENCES BETWEEN AVERAGES OF FAT TESTS ON DAILY SAMPLES AND ON COMPOSITE SAMPLES: WINTER SAMPLES, ALL PLANTS

Variation range	First period	Second period
Number of tests	Percent	Number of tests	Percent

Tests on daily samples and experimental composites

perct. 0-.09	244	65 . 410	292	83.670
10- 19	102	27 340	52	14.900
.20-. 29	16	4.300	5	1.430
. 30- .39	6	1.600
.40- 49 .	3	800
.50-. 59	1	.270
.60-. 69	1	.270
Total	373		349

Tests on daily samples and laboratory tests of plant composites

0-.09	115	33.04	121	34.280
.10- 19 .	125	35 92	125	35.410
.20-. 29	78	22.41	67	18.980
30- 39	19	5 46	26	7.370
.40-. 49 . . .	9	2 59	9	2.550
.50-. 59	2	.58	1	.280
.60-. 69			1	.280
. 70-over			3	.850
Total	348		353

Tests on daily samples and association tests on plant composites

0-.09	80	19.950	90	25.000
.10- 19	110	27 430	117	32.500
.20-. 29	108	26 93	92	25.550
.30-. 39	59	14.710	43	11.940
. 40- .49 ... .	29	7 230	14	3.890
.50-. 59	8	1 990	2	.550
.60-. 69 .	4	990	1	.270
.70-. 79	2	498
.80-. 89	1	.249	1	.270
Total	401		360

1939]

ACCURACY OF METHODS OF MILK SAMPLING TABLE 12. — Concluded

67

Variation range	First period	Second period
Number of tests	Percent	Number of tests	Percent

Tests on experimental composites and association tests on plant composites

perct. 0- .09	58	12.16	75	21.490
10- .19	141	29.56	120	34.380
.20- .29	86	18.03	88	25.210
30- 39	50	10.48	42	12.030
40- .49	131	27.46	17	4.870
50- 59	4	.838	3	.850
60- .69	3	.628	2	.570
70- 79 .	1	.21	1	.280
80- 89
90- 99 .	1	.21
1 00-1 09
1 10-1 19	1	.21
1 20-1.29	1	.21	1	.280
Total	477		349

Laboratory tests on plant composites and association tests on plant composites

0-	09	104	30.670	132	38.260
.10-	19	135	39.820	132	38.260
20-	29 .	71	20.940	59	17.100
.30-	39	16	4.710	14	4.050
40-	49 .	7	2.060	4	1.150
.50-	59	2	.590	2	.580
60-	69			1	.290
.70-	79	1	.290
.80-	89	2	.590
.90-	99
1.00-1	09	1	.290	1	.290
Total .		339		345

TABLE 13. — VARIATION BETWEEN HIGHEST AND LOWEST DAILY TESTS OF MILK FROM SAME PATRON, WINTER SAMPLES, ALL PLANTS

Variation range	First period	Second period
Number of tests	Percent	Number of tests	Percent
Perct. 0- . 25 . ....	29 143 109 80 24 11 4 4 1	7.160 35.310 26.910 19.750 5.930 2.710 .990 .990 .250	6 75 108 90 32 30 10 5 1 2 359	1.670 20.890 30.080 25.060 8.910 8.350 2.780 1.390 .270 .550
.30- .50
.55- .75 .
. 80-1 .00
1.05-1.25 .
1.30-1.50
1.55-1.75 .
1.80-2.00
2.05-2.25 .
2 . 30-2 .50
Total	405

68

BULLETIN No. 459

[November,

TABLE 14. — DISTRIBUTION OF DIFFERENCES BETWEEN AVERAGES OF FAT TESTS OF FRESH AND COMPOSITE MILK SAMPLES: SUMMER SAMPLES, PLANTS A AND D

Variation range

Number of tests

Percent

Tests of daily samples and experimental composites

perct.

0-.09 90 90.90

.10-. 19 8 8.08

.20-. 29 1 1.01

Total 99

Tests of daily samples and laboratory tests of plant composites

0-.09.. 29 90.62

.10-. 19 3 9.38

Total 32

Tests of daily samples and association tests of plant composites

0-.09.. ..:.. 28 28.50

.10-. 19 29 29.50

.20-. 29 24 24.50

.30-. 39 11 11.20

.40-. 49 4 4.10

.50-. 59 0 0

.60-. 69 1 1.00

Total 97

Tests of experimental composites and association tests of plant composites

0-.09.. 18 18.50

.10-. 19 38 39.20

.20-. 29 26 26.80

.30-. 39 9 9.20

.40-. 49 4 4.10

.50-. 59 1 1.00

.60-. 69 0 0

.70-. 79 1 1.00

Total 97

Laboratory tests of plant composites and association tests of plant composites

0-.09 8 16.33

.10-. 19 22 44.90

.20-. 29 12 24.49

.30-. 39 3 6.12

.40-. 49 3 6.12

.50-. 59 1 2.04

Total... 49

19391

ACCURACY OF METHODS OF MILK SAMPLING

69

TABLE 15. — AVERAGES OF FAT TESTS OF FRESH AND COMPOSITE MILK SAMPLES: SUMMER SAMPLES, PLANTS A AND D

Percent

Both plants

Daily tests of fresh samples

Tests of experimental composites. . . . Association tests of plant composites.

Plant D

Daily tests of fresh samples

Tests of experimental composites. . . . Association tests of plant composites. Laboratory tests of plant composites.

4.15 4.13 3.99

4.10 4.07 3.92 4.07

TABLE 16. — VARIATION BETWEEN HIGHEST AND LOWEST DAILY TESTS OF MILK FROM SAME PATRON: SUMMER SAMPLES, PLANTS A AND D

Variation range	Number of tests	Percent
perct. 0- .25. .	2	2
.30- .50	24	24
.55- .75	21	21
.80-1.00	18	18
1.05-1.25	18	18
1.30-1.50	10	10
1.55-1.75	2	2
1 . 80-2 .00	2	2
2.05-2.25 ... .	1	1
2.30-2.50	0	0
2.55-2.75	2	2
Total	100

COMPARISON OF TESTS ON DAILY SAMPLES AND CALCULATED TESTS ON TRUE COMPOSITES

As previously stated, a true composite sample is one taken in proportion to either the volume or the weight of milk delivered. How- ever, the use of a dipper is so much simpler than the use of a milk thief or other means of taking a proportionate sample that many dairies use the dipper and take a sample of practically the same size from all deliveries regardless of variations in the amount of milk delivered. Since variations were apparent in both the weight and test of the daily deliveries, a comparison was made of the test of the true composite, as calculated from the weight and test of each daily delivery, with the mathematical average of the tests on daily samples taken with a dipper. Nearly 3,000 daily deliveries of milk were tested (Tables 17 and 18).

The data show a remarkably close correlation between the tests

70

BULLETIN No. 459

[November,

TABLE 17. — COMPARISON OF CALCULATED TRUE AVERAGE TEST AND

MATHEMATICAL AVERAGE OF DAILY TESTS DURING 116

TEST PERIODS OF SEVEN DAYS EACH: PLANT B

Patron No.	First week	Second week
Daily test*	True test>>	Difference	Daily test	True test	Difference
True test less	True test greater	True test less	True test greater
1	perct. 5.62 3.67 4.54 5.04 4.25 4.95 4.55 5.12 4.21 4.92 4.06 3.80 4.29 4.62 4.66 4.47 4.69 4.48 5.52 4.84 4.80 4.47 4.37 4.95 4.29 3.64 3.82 5.19 4.13 4.76 4.50 4.46	perct. 5.628 3.680 4.544 5.043 4.249 4.945 4.555 5.096 4.207 4.928 4.020 3.801 4.296 4.555 4.622 4.467 4.687 4.485 5.516 4.842 4.802 4.499 4.375 4.951 4.294 3.639 3.821 5.211 4.110 4.762 4.470 4.502	perct. .'66i .005 !624 .003 !646 !065 .038 .003 .003 !664 !66i !6io .030	perct. .008 .010 .004 .003 !66s loos !66i .006 !66s '.002 .002 .029 .005 .001 .004 !66i .021 '.002 !6i2	perct. 5.35 3.50 4.71 5.10 4.10 5.19 4.83 5.11 4.13 4.61 3.58 3.93 4.16 4.74 4.82 4.25 4.53 4.85 5.24 5.01 4.91 4.46 4.64 5.04 4.30 3.56 4.17 5.15 4.42 4.99 4.51	perct. 5.352 3.502 4.707 5.090 4.081 5.186 4.813 5.101 4.158 4.614 3.575 3.922 4.164 4.713 4.826 4.248 4.492 4.790 5.202 5.015 4.903 4.457 4.641 5.036 4.277 3.563 4.162 5.142 4.415 4.990 4.504	perct. '.003 .010 .019 .004 .017 .009 !<X)5 .008 '.021 .002 .038 .060 .038 !667 .003 !6o4 .023 ioos .008 .005 0 .006	perct. .002 .002 .028 .004 !604 ;6o6 ioos !66i !6o3 "o"
2
3
4
5
6
7
g
9
11
13
14 . .
15
16
18
19
20
21
22
23
24
26
27
28
29 ....
32
33
35
37
39
41
42
43	4.95 4.71 6.06 4.48 4.45 5.05 4.99 4.88 4.68 5.77 4.53 4.71 4.52 4.85 4.54 5.03 5.45 4.37 5.52 5.88 5.78 4.81 4.24 5.55 4.98 4.45 5.27 4.761	4.945 4.709 6.014 4.473 4.469 5.048 4.983 4.871 4.681 5.730 4.550 4.690 4.528 4.852 4.536 5.033 5.394 4.362 5.502 5.867 5.791 4.807 4.235 5.559 4.978 4.450 5.274 4.775	.005 .001 .046 .007 !6o2 .007 .009 !040 !6io !o6i !056 .008 .018 .013 ;6o3 .005 .002 0	!6i9 '.001 .020 .008 .002 !6o3 !6ii !6o9 b .004 .014
45	4.63 5.74 4.41 4.10 5.02 4.83 4.42 4.71 5.08 4.40 4.70 4.60 4.82 4.61 5.43 5.36 4.25 6.21 5.82 5.71 5.20 4.60 5.35 5.35 4.53 5.00 4.733	4.618 5.730 4.447 4.096 5.028 4.837 4.406 4.680 5.072 4.316 4.687 4.611 4.812 4.613 5.437 5.300 4.243 6.315 5.805 5.704 5.178 4.596 5.353 5.356 4.551 5.002 4.731	.012 .010 !6o4 !6ii .030 .008 .084 .013 ioos .060 .007 '.015 .006 .022 .004 .002	!637 ioos .007 !6ii !6o3 .007 !ios !6o3 .006 .021 .002
46
47
48
49
50
52
53
54
57
61
63
64 .
H30
H31
H32
H33
H34 . . .
H35
H36 . . .
H37
H38
H39
H40
H41
H42
Average . . .

•Arithmetical average of daily percentages as determined by the Babcock test on daily deliveries.

bThe true average test was determined by dividing the weight of the total amount of fat by the total weight of milk delivered and multiplying by 100.

JPJ9]

71

TABLE 18. — COMPARISON OF THE CALCULATED TRUE AVERAGE TEST AND MATHEMATICAL AVERAGE OF DAILY TESTS DURING 310 TEST PERIODS OF SEVEN DAYS EACH:* PLANT D

Patron No.	First week	Second week
Daily testb	True test"	Difference	Daily test	True test	Difference
True test less	True test greater	True test less	True test greater
1	perct.	perct.	perct.	perct.	perct. 4.95 5.00 3.82 4.59 4.99 4.28 4.15 3.93 4.82 4.15 4.32 3.97 3.05 4.92 4.88 4.45 4.55 3.88 4.74 4.54 5.31 4.31 4.66 3.99 4.44 4.37 5.23 4.95 4.68 5.05 4.09 4.90 4.76 4.65 4.04 5.02 4.45 6.52 4.61 4.73 4.86 4.28 4.34 4.60 3.97 4.67 4.32 3.79 5.25 4.30 4.05 4.94 4.67 5.87 4.52 4.46 5.09 5.95 4.70 4.94 5.09 6.57 4.70	perct. 4.948 5.008 3.825 4.581 4.998 4.271 4.146 3.935 4.822 4.151 4.301 3.979 3.056 4.921 4.900 4.425 4.550 3.869 4.761 4.471 5.301 4.326 4.663 3.962 4.441 4.367 5.232 4.967 4.688 5.049 4.070 4.891 4.769 4.650 4.040 5.013 4.487 6.454 4.588 4.727 4.858 4.290 4.348 4.580 3.973 4.672 4.319 3.800 5.246 4.322 4.056 4.909 4.676 5.897 4.525 4.465 5.085 5.945 4.726 4.945 5.084 6.570 4.699	perct. .002 .009 '.009 .004 '.019 '.025 0 .011 .'6<59 .009 !028 !663 .'66i .020 .009 6'" 0 .007 !666 .022 .003 .002 !6io !66i iooi '.031 '.005 .005 !6o6 0 .001	perct. '.obk .005 !66s !66s .002 .001 .009 .006 .001 .020 6 " .021 ioie .003 !66i iooi .017 .008 .'669 0 0 !037 !6io .008 !6o3 .002 !6io !622 .006 '.006 .027 .005 .005 '.026 .005 6
4	4.90 4.06 4.57 5.00 4.14 4.25 4.06 4.62 4.24 4.18 3.68 3.09 4.92 5.00 4.27 4.58 3.89 4.65 4.35 5.22 4.27 4.42 4.17 4.24 3.98 5.23 4.63 4.64 4.78 4.05 5.12 4.80 4.65 3.78	4.946 4.079 4.581 4.993 4.140 4.266 4.062 4.612 4.250 4.198 3.691 3.091 4.915 5.000 4.273 4.580 3.888 4.640 4.357 5.217 4.281 4.432 4.189 4.226 3.982 5.229 4.635 4.644 4.768 4.056 5.109 4.958 4.646 3.783	!6o7 0 !66s !66s 0 6'" .002 .01 .003 !6ii !66i !6ii !6ii !66i	.046 .019 .011 6' .016 .002 !6i' .018 .011 .001 6'" .003 0 !6o7 .oii .012 .019 !<x>2 .005 .004 !6o6 iiss iooi
5 ....
7
8
9
11
12
13
19 .. . .
20
22 .
200
201
202
203
204
205
206
207
208
209
210
211. .
212
213
214
215
217
219
220 .
221
222
223
226
227... .
228	4.46 6.60 4.27 4.95	4.458 6.582 4.289 4.940	.002 .018 '.010	!6i9
229 .
232
300
301
302	3.95 4.40 4.48 4.12 4.73 4.59 3.79 5.68	3.958 4.425 4.450 4.116 4.730 4.566 3.790 5.690	'.030 .004 0 .024 0	.008 .025 6'" 6 .010
303
305
306 .
307
308 .
309
310
311
312	3.81 5.15 4.39 6.06 4.59 4.33 4.99	3.860 5.180 4.390 6.050 4.580 4.330 5.000	o'" .010 .010 0	.050 .030 0 6'" .010
313
314
315
316
318... .
400
401
402	4.67 5.02 5.05 6.25 4.34	4.680 5.020 5.050 6.250 4.340	6 0 0 0	.010 0 0 0 0
403.
404
405 . . .
406

(Table 18 continued on following page)

72

BULLETIN No. 459 TABLE 18. — Continued

[November,

Patron No.	First week	Second week
Daily testb	True test«	Difference	Daily test	True test	Difference
True test less	True test greater	True test less	True test greater
408	perct. 3.90 4.97 4.85 4.22 5.07 5.70 4.56 4.75	Perct. 3.905 4.974 4.857 4.205 4.959 5.697 4.568 4.752	perct. !6is .111 .003	Perct. .005 .004 .007 ioos .002	perct. 4.03 4.99 4.79 4.54 5.02 5.51 4.66 4.81 4.53 4.74 4.72 5.50 3.56 4.72 4.17 4.01 4.47 4.32 4.78 4.01 4.69 4.68 4.97 4.78 5.21 3.61 5.06 5.12 5.70 5.66 5.89 5.86 4.49 4.83 5.27 4.30 4.86 5.14 4.49 4.65 4.52 5.37 5.22 4.93 6.46 4.84 5.12 5.34	perct. 4.036 4.989 4.794 4.547 5.023 5.503 4.689 4.837 4.528 4.758 4.717 5.808 3.481 4.700 4.007 4.017 4.483 4.323 4.751 4.010 4.691 4.675 4.980 4.772 5.184 3.608 5.061 5.125 5.694 5.621 5.870 5.799 4.501 4.846 5.280 4.296 4.892 5.140 4.497 4.632 4.505 5.351 5.249 4.895 6.379 4.830 4.983 5.348	perct. !66i !6o7 .002 .003 !679 .020 .163 !629 0 !66s .008 .026 .002 iooe .039 .020 .061 .004 6' " !6is .015 .019 ."635 .081 .010 .137	perct. .006 !664 .007 .003 !629 .027 .018 !308 !667 .013 .003 6 .001 !6io !66i .005 !6ii .016 .010 !632 0 .007 !629 .'<x>8
409 .
411
412
413
414
415
416
417
418	5.05 4.32 5.40	5.048 4.331 5.421	.002	!6ii .021
419
424
426
427	4.28 4.12 3.85 4.54 4.30 4.82 3.97 4.72 4.46 5.26 4.60 5.39 3.69 4.96 5.21 5.49 5.79 5.70 5.39 4.80 4.98 5.11 4.46 4.35 5.40 4.47 4.46 4.92 5.42 5.20 5.05 5.75 4.75 4.79 5.33 5.20 3.90 5.94 4.45 5.35 3.98 4.34 4.51 4.86 4.64 4.89	4.297 4.127 3.848 4.535 4.297 4.834 3.971 4.711 4.463 5.261 4.600 5.392 3.678 4.968 5.213 5.524 5.807 5.710 5.391 4.832 4.980 5.045 4.467 4.255 5.300 4.483 4.403 4.886 5.464 5.212 5.045 5.770 4.747 4.798 5.340 5.193 3.970 5.942 4.445 5.350 3.993 4.347 4.500 4.855 4.620 4.920	iooi .005 .003 !6o9 6'" .012 6 .065 !095 .010 .'6s 7 .034 !66s iooi !6o7 ioos 0 !6io .005 .020	.017 .007 !6ii .001 !6o3 .001 0 .002 ioos .003 .034 .017 .010 .001 .032 0 !6o7 .oii !044 .012 !620 ioos .010
428
429
430
431
432
434
435
436
437
438
439
440
441
442 .
443
444
445
446
503 .
504
506
507 . .
508
509
510
511
513
514
515
516
517
518
519
520
521 .
523	.070 .002 6 ' .013 .007 !630	4.45 5.83 4.35 5.41 4.34 4.29 5.32 4.81 4.96 4.66 5.36 4.99 3.63 4.21 3.74 3.53 3.56	4.436 5.819 4.340 5.410 4.321 4.289 5.323 4.781 4.934 4.628 5.292 4.989 3.665 4.193 3.735 3.544 3.561	.014 .011 .010 0 .019 .001 .029 .026 .032 .068 .001 !6i7 .005	6 '" !6o3 !635 !6i4 .001
600
601
602
604 .
607
608
609
610
612...
613
615	4.53 3.62 4.50 3.80 3.54	4.530 3.626 4.522 3.838 3.540	0 6 "	0 .006 .022 .038 0
618
619
900
901
903

1939~]

ACCURACY OF METHODS OF MILK SAMPLING TABLE 18. — Concluded

73

Patron No.	First week	Second week
Daily test>>	True testc	Difference	Daily test	True test	Difference
True test less	True test greater	True test less	True test greater
904. . .	perct. 3.43	perct. 3.430	perct. 0	perct. 0	perct. 3.87 5.67 4.18 5.17 4.09 5.11 5.09 5.29 4.07 4.67 4.20 3.86 4.69 3.64 4.83 4.04 5.64 5.86 4.34 5.14 5.09 4.45 4.54 4.51 4.61 4.20 6.83 4.79 4.15 4.72 4.51 4.42 4.70	perct. 3.869 5.660 4.182 5.160 4.079 5.126 5.059 5.287 4.069 4.682 4.195 3.862 4.675 3.625 4.826 4.037 5.675 5.862 4.330 5.126 5.080 4.444 4.528 4.493 4.598 4.171 6.736 4.787 4.148 4.698 4.519 4.419 4.704	perct. .001 .010 .010 .011 .031 .003 .001 '.005 !6is .015 .004 .003 .010 .014 .010 .006 .012 .017 .012 .029 .094 .003 .002 .022 !66i	perct. 002 016 012 002 035 002 009 .004
905
907	4.37 5.15 4.04 5.00 5.02 5.57 4.07 4.80 4.29 3.92 4.55 3.83 4.65 4.28 5.77	4.373 5.150 4.043 5.009 5.019 5.566 4.072 4.793 4.309 3.925 4.552 3.837 4.656 4.302 5.733	6 ' !66i .004 !6o7 !637	.003 0 .003 .009 !66i .019 .005 .002 .007 .006 .022
908
909
910
911
912
913
914
917
920
921 ..
922
1001 .
1002
1005
1006
1007	4.60 4.92 4.87 4.84 4.81 4.49 4.34 4.22 6.97 4.70 4.26 4.67 4.52 4.38 4.69	4.592 4.911 4.861 4.833 4.809 4.497 4.329 4.233 6.830 4.705 4.263 4.585 4.523 4.431 4.683	.008 .009 .009 .007 .001 !6ii !i46 .085 .007	!667 !6i3 !oos .003 !6o3 .051
1009
1011
1012 . .
1013
1014
1015
1016
1017
1018
1019... .
1020
1021
1022
Average

"Averages for the 426 test periods in Plants B and D, Tables 17 and 18. Daily test, 4.70 percent fat; true test, 4.6995 percent fat.

••Arithmetical average of daily percentages as determined by the Babcock test on daily deliveries.

cThe true average test was determined by dividing the weight of the total amount of fat by the total weight of milk delivered and multiplying by 100.

calculated on the true composites and the mathematical average of the tests on the daily samples. For only seven samples did the difference between the tests amount to .10 percent fat or more. Naturally, some differences would be expected because of the wide variations in the weight of the milk delivered during the seven-day test periods (Tables 19 and 20). However, under the conditions of these experiments these differences are not of sufficient significance to seriously affect the accuracy of the test on the composite samples.

74

BULLETIN No. 459

[November,

TABLE 19. — VARIATIONS IN WEIGHT OF MILK AND FAT DELIVERED BY PATRONS DURING EACH OF Two SEVEN-DAY TEST PERIODS: PLANT B

Patron No.	First week	Second week
Milk	Fat	Milk	Fat
Max.	Min.	Max.	Min.	Max.	Min.	Max.	Min.
1	Ib. 140 134 113 72 152 150 121 119 175 126 223 154 27 123 241 236 78 87 22 142 85 168 47 70 57 110 115 122 123 87 55 173 75 88 155 153 168 98 189 97 19 60 69 161 47 82 67 29 56 67 54 37 51 103 61 58 50 48	Ib. 117 88 100 68 140 100 107 97 151 74 165 140 23 32 120 201 64 82 10 139 72 42 38 62 43 89 100 95 88 51 35 163 65 72 118 122 157 82 168 79 13 47 14 128 38 76 60 23 40 17 45 31 32 90 54 52 37 39	Ib. 7.788 4.958 5.668 3.816 8.360 7.200 5.616 5.950 7.216 6.300 10.296 6.258 1.215 5.535 12.050 10.534 3.510 3.999 1.210 7.089 4.165 6.888 2.068 3.432 2.622 4.180 4.600 6.954 4.945 3.915 2.420 8.084 3.600 4.428 6.384 6.248 8.568 4.896 8.325 4.365 .936 2.580 3.024 7.900 2.350 3.772 3.640 1.426 2.352 4.355 3.036 2.164 2.703 4.686 3.355 3.190 2.350 2.400	Ib. 6.201 3.256 4.326 3.360 5.5825 4.950 4.905 5.044 6 . 3075 3.637 6.764 5.040 .966 1.664 5.400 8.610 2.772 3.654 .550 6.526 3.384 1.890 1.539 2.8615 1.786 3.293 3.636 4.224 3.988 2.346 1.855 7.138 2.970 5.104 5.324 5.002 8.007 3.772 7.308 3.792 .611 2.068 .742 5.760 1.920 3.465 3.120 1.352 1.740 .994 2.655 1.860 1.728 4.275 2.912 2.652 1.615 1.911	Ib. 144 113 110 83 152 137 137 156 167 85 218 147 26 85 228 258 71 87 50 143 80 172 48 82 45 103 111 112 110 115 46 177 78 117 126 155 162 88 180 98 16 63 54 146 47 79 68 37 53 49 58 40 56 142 58 56 52 50	Ib. 121 84 88 65 118 72 82 98 141 72 169 130 23 65 163 198 61 62 21 140 49 95 37 64 27 93 91 94 85 73 24 150 61 17 105 120 144 72 155 89 8 45 21 116 38 73 61 25 39 23 40 32 46 97 47 50 13 46	Ib. 7.848 3.955 5.060 4.233 6.192 7.260 6.371 8.112 6.847 4.080 7.848 6.235 1.092 3.792 11.350 11.374 3.185 4.158 2.800 7.722 3.760 7.912 2.208 4.018 1.866 3.648 4.394 5.618 5.225 5.750 2.156 8.704 3.510 7.313 5.607 7.285 8.215 4.481 8.910 4.802 .856 3.150 2.322 6.716 2.256 3.950 3.604 1.776 2.120 2.573 3.625 2.356 2.800 5.893 3.219 2.800 2.344 2.675	Ib. 6.292 2.656 4.092 3.430 5.428 3.852 4.018 5.096 6.063 3.384 6.422 5.180 .943 3.124 7.661 8.316 2.924 3.410 1.145 6.816 2.401 5.935 1.517 3.201 1.215 3.255 3.822 4.606 3.910 3.650 1.056 7.580 2.806 1.105 4.620 4.920 7.200 3.600 7.584 3.916 .496 1.980 .872 4.988 1.800 3.198 2.989 1.550 1.638 1.265 2.480 1.808 2.254 4.141 2.444 2.544 .546 2.346
2
3 ..
4
5 .
6
7
8 .. . .
9
11...
13
14 .
15
16 .
18...
19
20 ..
21
22 .
23
24
26... .
27
28 ..
29
32 .
33
35
37
39
41 ..
42
45 .
46
47
48...
49
50
52
53
54
57
61...
63
64 ..
H30.. .. H31 H32.. .. H33 H34 H35 H36 .... H37 H38 H39 H40 H41 H42

1939~]

ACCURACY OF METHODS OF MILK SAMPLING

75

TABLE 20. — VARIATIONS IN WEIGHT OF MILK AND FAT DELIVERED BY PATRONS DURING EACH OF Two SEVEN-DAY TEST PERIODS: PLANT D

Patron No.	First week	Second week
Milk	Fat	Milk	Fat
Max.	Min.	Max.	Min.	Max.	Min.	Max.	Min.
1 ...	Ib. 194 135 70 202 185 287 105 491 150 125 32 88 74 81 36 129 24 210 77 144 96 55 83 56 84 58 65 62 148 118 94 158 96 42 92 141 59 68 58 70 138 152 76 58 147 74 56 123 73 79 77 122 84 17 125 147 47 41 68 171 52 98 123 205 126 63 40	Ib. 162 105 52 156 150 222 92 415 122 110 19 72 60 55 32 118 21 178 63 98 76 44 45 43 63 44 51 54 128 96 84 148 63 35 80 106 51 53 44 50 119 124 54 34 122 63 38 99 62 51 58 98 80 12 109 125 40 34 52 152 44 80 78 183 98 52 28	Ib. 9.118 7.020 3.010 8.9745 9.250 11.480 4.725 20.131 6.600 5.750 1.456 3.432 2.263 3.888 1.819 5.658 1.128 8.151 3.542 6.336 4.814 2.530 3.652 2.436 3.696 2.320 3.445 2.914 6.882 5.724 3.948 8.532 4.840 1.953 3.496 7.825 2.726 4.216 2.726 3.500 6.901 6.080 3.648 2.378 6.1625 3.848 2.430 5.0215 4.615 3.510 3.157 6.588 4.000 1.054 5.900 6.174 2.397 2.500 3.196 8.208 2.782 6.370 5.488 8.5425 6.100 3.150 1.760	Ib. 8.415 5.145 2.028 6.972 8.256 9.990 3.312 17.248 5.246 4.256 .7885 2.520 1.830 2.750 1.504 4.543 .924 7.030 3.072 4.214 3.948 1.665 1.935 1.505 2.8035 1.716 2.470 2.508 5.952 4.512 3.375 7.488 3.213 1.620 3.080 4.982 2.346 3.520 1.738 2.425 5.160 4.788 2.160 1.666 5.104 2.479 1.653 3.465 3.520 1.836 2.030 4.300 3.360 .708 5.074 5.628 1.575 1.938 2.340 7.650 2.024 4.800 2.9625 6.825 4.900 2.552 1.505	Ib. 183 126 50 190 195 274 94 468 132 117 32 74 70 66 35 130 21 203 70 195 91 58 68 50 82 60 56 64 140 110 95 173 103 36 85 130 151 65 56 60 135 188 55 48 153 71 49 122 72 80 82 118 78 17 117 151 63 38 92 161 48 90 80 194 122 67 50	Ib. 155 110 42 136 161 226 75 411 111 103 21 56 56 54 31 105 18 183 54 50 78 43 63 32 70 41 48 48 123 95 60 153 52 28 78 100 52 44 40 42 114 156 44 34 114 52 36 90 65 63 64 83 50 10 100 130 46 30 53 147 37 78 78 173 108 42 44	Ib. 9.516 6.300 2.072 8.740 9.750 12.420 3.999 18.343 6.240 5.022 1.392 3.034 2.048 3.380 1.792 5.565 9.660 7.503 3.500 8.385 5.185 2.494 3.150 2.256 3.813 2.520 2.997 3.402 7.000 5.720 3.800 8.400 4.841 1.674 3.520 5.980 6.040 3.803 2.856 2.820 6.480 8.084 2.438 2.304 6.579 3.536 2.136 4.950 3.780 3.760 3.280 5.445 3.588 1.088 5.148 6.946 3.150 2.242 4.784 8.533 2.464 5.940 4.212 8.536 6.608 3.283 2.350	Ib. 7.968 5.390 1.596 6.435 7.590 10.130 2.847 16.068 5.340 3.914 1.012 2.240 1.860 2.619 1.395 4.826 7.200 7.030 2.241 2.450 4.056 1.763 2.880 1.344 2.975 1.764 2.538 2.184 5.658 4.750 2.600 7.605 2.288 1.316 3.108 4.692 2.314 3.059 1.863 1.848 5.415 6.162 1.782 1.530 4.526 2.418 1.548 3.060 3.380 2.457 2.464 4.183 2.200 .580 4.545 5.719 2.208 1.920 2.438 7.056 1.961 5.070 3.200 6.552 5.463 2.037 1.826
4
5 .
7
g
9
11 12
13 19
20
22 ..
200
201 .
202
203 .
204 205 .
206
207 208 209 210 211 212 213 214 215 217 219 220 221
222 223 .
226
227 .
228 229 232
300 301 302
303 305
306 307
308 309 . .
310 311 . .
312 313 314 315 316 318 400 401 402 403 404 405 406 408 409 411 412

(Table 20 continued on following page)

76

BULLETIN No. 459 TABLE 20. — Continued

[November,

Patron No.	First week	Second week
Milk	Fat	Milk	Fat
Max.	Min.	Max.	Min.	Max.	Min.	Max.	Min.
413... 414	Ib. 70 87 59 87 30 58 107 15 28 115 363 150 213 82 77 103 83 150 73 94 143 59 110 155 83 46 132 83 127 32 68 169 80 150 111 58 76 32 170 100 89 55 76 94 22 76 79 122 69 82 63 148 173 112 118	Ib. 51 73 48 73 10 50 98 11 22 48 335 125 160 59 68 91 63 127 63 83 123 42 98 143 47 29 86 71 96 28 48 144 21 25 78 22 30 24 130 30 28 43 62 30 10 37 65 97 52 67 53 93 147 72 58	Ib. 3.536 4.8025 2.8615 4.176 1.290 2.900 4.8685 .854 1.026 5.060 15.4275 6.000 10.011 3.731 3.8625 4.116 3.818 6.750 3.9195 4.324 7.865 2.124 5.830 8.208 4.9385 2.576 7.524 4.482 6.096 1.568 3.060 7.943 2.960 7.650 5.232 2.262 3.724 1.824 9.435 5.000 5.518 2.420 3.800 5.076 1.166 3.306 4.7795 5.551 3.657 3.773 3.024 6.216 8.131 5.152 6.018	Ib. 2.703 4.028 2.093 3.525 .465 2.400 3.360 .520 .770 2.1725 14.070 4.9375 7.600 2.655 3.134 3.720 3.096 5.461 2.925 3.320 6.765 1.734 4.606 7.007 2.809 1.590 4.730 3.763 4.800 1.248 2.514 6.336 1.008 1.475 3.354 1.056 1.605 1.236 6.175 1.455 1.650 1.980 2.816 1.800 .515 1.326 3.795 4.312 2.756 2.345 2.226 4.350 7.252 3.312 2.8125	Ib. 62 77 56 75 21 54 100 20 32 70 360 150 247 84 67 105 80 134 87 90 155 50 110 165 55 31 132 93 141 30 54 224 55 100 93 40 72 30 175 61 75 53 67 70 24 76 90 102 61 89 58 122 196 85 75 83 75 104 25 74 276 158 121 71 110 78 165 103	Ib. 45 68 46 56 16 45 80 7 20 34 330 128 190 73 42 90 54 111 73 70 97 40 92 145 40 15 96 60 107 24 35 137 45 78 80 28 60 18 97 20 20 37 53 45 17 54 64 80 52 75 42 102 147 65 45 47 67 82 16 64 250 144 100 50 97 64 133 74	Ib. 3.120 4.477 2.915 3.723 .935 2.754 4.680 1.044 1.104 3.150 15.225 6.300 11.609 3.744 3.015 4.242 4.134 6.200 4.698 4.200 7.776 1.838 5.720 8.498 3.108 1.798 7.590 5.208 6.839 1.470 2.754 9.408 3.025 5.843 4.263 1.760 3.402 1.530 9.625 2.928 4.350 2.491 3.250 3.675 1.416 3.572 5.040 4.464 3.294 3.838 2.622 7.320 9.300 4.234 3.230 3.901 3.863 3.811 1.088 2.680 10.212 5.846 4.780 4.114 5.040 4.017 6.765 5.044	Ib. 2.273 3.536 1.833 2.550 .720 1.845 3.740 .445 .780 1.598 10.562 4.800 8.360 2.993 2.226 3.690 2.646 5.332 3.285 3.360 5.626 1.440 4.830 6.536 2.240 .983 6.144 3.600 4.565 1.008 1.820 5.880 1.688 4.067 3.120 1.428 2.560 .999 5.044 .980 1.170 1.684 2.727 2.115 .901 2.106 3.610 3.480 2.886 2.880 1.857 5.457 7.350 2.860 2.205 2.820 3.162 3.116 .741 2.278 8.514 4.884 4.000 2.950 3.800 3.520 5.476 3.552
415 416
417
418
419 424 426
427 .
428 429
430 431 432 434
435
436.. .
437 438 . .
439
440 441
442 443 . .
444
445
446...
503 504 506
507 508 .
509
510 511
513 514 .. .
515
516 517 518
519 .
520 521 523
600
601 602 . . .
604 607 ..
608 609 610 612
613 . . .
615 618 619 900 901 903 904 905 907 908 909 910	85 120 34 84 271 159 116 63 120 85 173 150	70 100 19 73 260 110 103 39 104 76 153 94	3.818 4.200 1.716 3.360 9.756 6.123 3.811 3.480 5.452 4.293 7.093 7.650	2.993 3.400 .798 2.916 8.942 3.905 3.5535 2.436 4.368 3.800 6.314 4.888

19391

ACCURACY OF METHODS OF MILK SAMPLING TABLE 20. — Concluded.

77

	First week	Second week
Patron No.	Milk	Fat	Milk	Fat
	Max.	Min.	Max.	Min.	Max.	Min.	Max.	Min.
	Ib.	Ib.	Ib.	Ib.	Ib.	Ib.	Ib.	Ib.
911.	120	100	5.880	5.100	113	92	5.511	4.935
912.	116	98	6.322	5.562	112	94	6.380	5.194
913.	97	73	4.0255	2.9565	104	60	4.160	2.220
914.	58	36	2.697	1.773	65	46	3.016	2.001
917.	90	65	4.200	2.520	86	70	3.825	2.870
920.	83	65	3.311	2.665	82	68	3.239	2.698
921.	116	101	5.432	4.242	112	85	5.100	4.123
922.	106	80	4.028	2.880	120	51	4.380	1.913
1001.	70	65	3.536	2.665	68	60	3.283	2.790
1002.	51	40	2.275	1.560	51	40	2.040	1.554
1005.	55	18	3.245	1.121	46	25	2.737	1.428
1006.	41	33	2.394	1.716	40	30	2.379	1.705
1007.	118	102	5.100	4.664	134	97	5.786	4.268
1009.	S3	42	2.544	2.058	48	35	2.668	1.920
1011.	62	39	2.976	2.050	59	43	3.233	2.376
1012.	148	50	6.882	2.400	105	90	4.656	3.870
1013.	96	89	4.806	4.042	102	90	4.500	3.999
1014.	76	70	3.496	2.992	80	68	3.520	3.280
1015.	170	126	7.682	5.796	160	127	7.040	5.720
1016.	144	120	6.681	4.736	152	105	6.384	5.040
1017.	36	11	2.304	.897	40	10	2.620	.750
1018.	49	37	2.254	1.640	46	42	2.208	1.932
1019.	82	79	3.840	3.081	83	76	3.510	3.081
1020.	127	50	5.3975	2.650	86	30	3.913	1.485
1021.	64	45	2.880	2.040	52	44	2.525	1.890
1022.	38	23	2.090	.989	35	19	1.820	.836

GENERAL DISCUSSION OF RESULTS

The data obtained in this study confirm the findings of Sanmann and Overman9* and others that seven-day composites when properly taken and stored will test about the same as fresh milk samples. While the tests on fresh milk samples averaged somewhat higher than the average test of the laboratory composites, the difference was slight, being .061 percent the first period of the winter series ; .026 percent the second period of the winter series; and .020 percent for the summer series. Comparing all the samples, 75.64 percent of the daily and laboratory composite samples checked within .09 percent of each other, and 95.98 percent were within .20 percent of each other. The greatest variations were in the first period of the winter series.

In comparing the daily tests with the Association tests on the plant composites, it will be noted that agreement between tests is not so close as it is between the laboratory composite tests and the daily tests. The average variation for the first period of the winter series was .163 percent, for the second period .175 percent, and for the summer series .205 percent. However, 22.78 percent of the 708 comparisons show a difference of .3 percent or more.

78 BULLETIN No. 459 {November,

In comparing the laboratory tests on the plant composites with the Association tests on these same samples, it will be found that only 7.4 percent of the samples show a difference of .3 percent or more. In the comparison of the average of daily tests and the laboratory tests of plant composites, 18.3 percent of the samples differ .3 percent or more in fat content. It would seem, therefore, that there were more variations traceable to the plant composite samples themselves than to the testing of these composites. Possible causes for inaccurate plant composites are improper mixing in the bottle each day, improper refrigerating of the samples, and failure to take samples each day. It has been observed that sometimes composite samples are not taken by the plants on holidays, Sundays, or on days the regular receiving-room man is off duty. The occasional omission of a daily sample would not be serious except when the tests on daily deliveries varied widely. Since 67.71 percent of 864 seven-day delivery periods were found to have variations over .5 percent between the highest and lowest daily tests on the milk delivered within the period, failures to include samples from all deliveries likely affected the accuracy of the composite samples of such deliveries.

In general, the tests reported by the Association representative seem to have been accurately performed. As it is not humanly possible to prevent all errors, the question rises as to what degree of tolerance should be allowed. Examination of the data indicates errors either in the testing or in the recording of the Association tests on several of the plant composites. In such cases the tendency was for these tests to be low. With the average daily tests, laboratory composite tests, and the laboratory tests of the plant composites as a check, an attempt was made to select the Association tests of the plant composites that seemed in error.

The laboratory tests of the plant composites might be subject to some criticism because of the fact that by the time some of these samples reached the laboratory, they were churned, and occasionally there was only a small portion of sample left. However, whenever the average daily tests and the laboratory composite test agreed reasonably well with the laboratory test of the plant composite and all three tests were .2 percent or more higher than the Association test of the plant composite, it was assumed that there was some error in the performing of the test by the Association representative either thru faulty tests, incorrect reading of the fat column, or incorrect recording of the test. How best to prevent such errors, however, is rather difficult to determine.

1939] ACCURACY OF METHODS OF MILK SAMPLING 79

It is very likely that errors of this nature will occur wherever many tests are being performed at one time, and probably the only way in which the number could be held to a minimum would be by some system of checks. The person doing the testing should realize that his tests are likely to be checked at any time. Duplication of all tests is probably unnecessary, yet there is ample evidence in this study to sup- port the belief that a retest of at least part of the samples would be justified and practical. In milk delivered by a selected group of 117 patrons the errors evident in the test for fat totaled 31.30 percent (Table 21). Assuming the average weekly delivery was 600 pounds, the total loss to the producers of this group was the value of 187.8 pounds of fat. At 40 cents a pound this amount of fat would have a value of $75.12, a value that would take care of the extra cost of double-check- ing most of the composite milk samples on this market.

One of the most striking things brought out in this study was the wide variation between the highest and lowest test of the milk delivered by a large number of the patrons during a seven-day period. A total of 432 patrons made deliveries which were tested over two weeks time. Each week during which each patron's milk was tested was considered a separate period, so that there were 432 patrons and 864 test periods. Data on these 864 periods show that only 37 of them do not exceed .25 percent between the highest and lowest test. Considering .5 per- cent as a normal variation, 67.71 percent of the test periods would indi- cate an abnormal variation in the fat content of the milk. That 18.4 percent of the seven-day periods showed variations over one percent (some over 2.5 percent) is sufficient evidence that mechanical manip- ulation of the fat content of the milk took place in a number of cases. A possible explanation for this may be found in the plan followed in paying the farmers for their milk. Each patron had a base, which approximated 60 percent of the amount of milk he delivered from September 15 thru December 15. For this base, in December, 1936, he was paid a net price of $2.05 per hundred pounds. The price differ- ential was 3.5 cents a point. Since the only restriction on his base allotment was its weight, a farmer may have considered it good business to skim a reasonable amount of his surplus milk, place the cream he did not need for table purposes in with the remaining whole milk and utilize the skimmilk for feeding. For example, a farmer may have delivered 2,000 pounds of 3.8-percent milk in a seven-day period. With a base of 1,200 pounds, if he did not skim the milk, he would have

80

BULLETIN No. 459

[November,

TABLE 21. — TESTS OF DAILY AND COMPOSITE SAMPLES SHOWING PROBABLE ERROR IN ASSOCIATION TEST OF PLANT COMPOSITE

Patron No.

Average test of daily samples

Test of experimental composite

Laboratory test of plant composite

Association test of plant composite

Evident error

Plant A

First period 9	Perct. 4.19	perct. 4.2	perct. 4.2	perct. 4.00	Perct. .20
12	5.56	5.6	5.6	5 35	25
56	5.12	5.0	4.9	4.70	.20
63	4.72	4.6	4.6	4.40	.20
58a	4.30	4.3	4.4	4.10	.20
Second period 142	3.87	3 80	3 85	3 60	20

Plant B

First period 3	4.54	4.50	4.50	4.30	.20
4	5.04	5.10	5.10	4.80	.20
6	4.95	4.90	4.80	4.60	.20
9	4.21	4.20	4 05	3 80	.20
20	4.69	4.60	4.60	4.40	.20
22	5 52	5 40	5 40	5 20	.20
23	4.84	4.80	4.90	4.60	-.20
37	4.13	4.10	4.20	3.90	.20
41	4.50	4.50	4.50	4.30	.20
52	4.42	4.45	4.40	4.20	.20
54 ... .	5.08	5 00	4 80	4 60	.20
61	4.70	4.80	4.50	4.30	.20
64	4.82	4.95	4.80	4.00	.80
31	5.43	5 40	5 40	5 10	.30
Second period 3	4.71	4.70	4.60	4.30	.30
7	4.83	4.80	5.00	4.60	.20

Plant C

First period H26	5.80	5.80	5.80	5.60	.20
Second period 3	4 27	4 30	4 20	4 00	.20

Plant D (winter)

First period 1	4.77	4.80	4.75	4.60	.15
7	4.57	4.50	4.60	4.40	.20
19	4.24	4.20	4.20	3.80	.40
204	4.58	4.60	4.70	4.40	.20
205	3.89	3.90	3.90	3.50	.40
214	5.23	5.30	5.30	4.90	.40
228	4 46	4 60	4.55	4.20	.30
229	6.60	6.50	6.35	6.10	.25
302	3.95	3.90	3.80	3.50	.30
305	4.48	4.50	4.50	4.20	.30
312	3.81	3.95	3.90	3.50	.40
313	5 15	5.20	5.10	4.80	.30
315	6.06	6.00	6.10	5.70	.40
405	6 25	6 30	6 20	6.00	.20

1939]

ACCURACY OF METHODS OF MILK SAMPLING

81

TABLE 21. — Continued

Patron No.

Average test of daily samples

Test of experimental composite

Laboratory test of plant composite

Association test of plant composite

Evident error

Plant D (winter, first period, concluded)

First period 406	perct. 4.34	perct. 4.40	perct. 4.35	perct. 4.10	Perct. .25
408	3.90	4.00	3.95	3.50	.45
429	3.85	3.85	3.75	3.40	.35
432	4.82	4.65	4.70	4.40	.30
441	4.96	4.90	4.90	4.60	.30
444	5.79	5.70	5.65	5.40	.25
445	5.70	5.70	5.60	5.30	.30
516	5.05	5.00	5.00	4.80	.20
600	5.94	5.80	5.70	5.50	.20
602	5.35	5.30	5.30	5.00	.20
609	4.86	4.90	4.70	4.50	.20
901	3.54	3.50	3.50	3.30	.20
912	5.57	5.50	5.50	5.30	.20
914	4.80	4.75	4.80	4.60	.20
920	3.92	3.80	3.80	3.60	.20
1009	4.92	4.90	4.90	4.70	.20
401	5.80	5.70	5.90	5.40	.50
Second period 12	3.93	4.00	3.90	3.70	.20
13	4.82	4.75	4.70	4.50	.20
22	3.97	3.95	3.80	3.60	.20
200	3.05	3.10	3.00	2.80	.20
201	4.92	4.90	4.90	4.70	.20
202	4.88	4.80	4.80	4.60	.20
203	4.45	4.40	4.25	4.00	.25
206	4.74	4.70	4.60	4.30	.30
207 . ...	4.54	4.55	4.50	4.20	.30
208	5.31	5.30	5.10	4.70	.40
209	4.31	4.25	4.10	3.90	.20
210	4.66	4.60	4.60	4.30	.30
228	4 45	4 40	4 30	4.10	.20
301	4.86	4.70	4.70	4.40	.30
308	4 32	4 30	4 30	4 10	.20
313	4.67	4.70	4.50	4.30	.20
315	5.87	5.75	5.70	5.50	.20
405 . .	6.57	6.55	6.40	5.30	.10
406	4.70	4.80	4.60	4.40	.20
408	4.03	4 15	3 90	3.70	.20
411	4.79	4.90	4.80	4.50	.30
412	4 54	4 70	4 70	4.30	.40
414	5.51	5.60	5.40	5.10	.30
415	4 66	4 75	4 60	4 20	.40
443	5.70	5.70	5.60	5.40	.20
506	5 27	5 20	5 30	5 10	.20
507 . .	4.30	4 30	4.30	4.10	.20
516	4.93	4.85	4.70	4.50	.20
517.	6 46	6 40	6 25	6.00	.25
520	5.34	5.35	5.40	5.20	.20
523	4 45	4 40	4 20	4.00	.20
601	4.35	4.35	4.30	4.10	.20
612	4 66	4 65	4 70	4 40	.30
613	5.36	5.45	5.40	5.00	.40
618	3 63	3 60	3 75	3 40	.35
619	4.21	4.15	4.10	3.90	.20
900	3 74	3 65	3 60	3 40	.20
905	5.67	5 60	5.60	5.40	.20
909	4 09	4 00	3 95	3 60	.35
910	5.11	4.90	4 90	4.70	.20
920	3 86	3 90	3 75	3 50	.25
922	3.64	3 60	3 55	3.30	.25
1005	5 64	5 60	5 50	5 30	.20
1006	5 86	5 90	5 90	5.40	.50
1020	4 72	4 65	4 70	4 50	.20

(Table 21 concluded on following page)

82

BULLETIN No. 459 TABLE 21. — Concluded

[November,

Patron No.

Average test of daily samples

Test of experimental composite

Laboratory test of plant composite

Association test of plant composite

Evident error

Plant D (summer)

201	perct. 5.28	perct. 5.20	perct. 5.20	perct. 4.80	perct. .40
212	4.39	4.40	4.40	4.20	.20
213	4.33	4.30	4.25	4.00	.25
214	4.59	4.50	4.50	4.30	.20
216	4.52	4.45	4.40	4.20	.20
244	3.79	3.70	3.70	3.50	.20
601	3.75	3.70	3.75	3.40	.35
1002	3.41	3.40	3.40	3.20	.20
1005	4.20	4.10	4.10	3.90	.20
1012	4.13	4.20	4.20	3.70	.50
1013	4.32	4.30	4.40	4.00	.40
1014	4.94	4.95	4.90	4.70	.20
1015	4.59	4.60	4.60	4.40	.20
1017.	4.25	4.20	4.15	3.80	.35
1019	4.61	4.60	4.70	4.30	.40
1020	3 85	3.80	3.80	3.60	.20
1022	3.61	3.60	3.60	3.30	.30

received under the conditions of the Champaign-Urbana market in December, 1936, $36.50 calculated as follows:

Base allotment: 1200 Ib. at #2.05 per cwt. 3.8% = #24.60

Surplus: 800 Ib. at #1.50 per cwt. 3.8% = 12.00

Total = 36.60

If the farmer had skimmed half his surplus milk, he would have received $36.05 for the milk he sold and would have had about 339 pounds of skimmilk left for feeding. Further, he would have saved the shipping cost on 339 pounds of milk. The method of arriving at these values is shown by the following calculations:

400 X 3.8% = 15.2 pounds of fat in milk skimmed

Assuming that a 25-percent cream was skimmed, the weight of the cream skimmed would be equal to 60.8 pounds:

1600 + 60.8 = pounds of milk delivered

1600 X 3.8% = 60.8 pounds of fat in unskimmed milk

60.8 X 25% = 15.2 pounds of fat in added cream

Thus the 1660.8 pounds of milk delivered contained 76 pounds of fat. As it tested 4.58 percent, its value would be figured as follows:

Value of 1200 Ib. of 4.58% base milk at $2.323 per cwt = #27.88

Value of 460.8 Ib. surplus milk at $1.773 per cwt = 8. 17

Total value of 1600.8 Ib. milk testing 4.58% = 36.05

Assuming the skimmilk has a feeding value of 25 cents a hundred pounds and that hauling charges are 25 cents a hundred pounds, the farmer would gain $1.70 by not marketing the 339 pounds of skimmilk.

1939] ACCURACY OF METHODS OF MILK SAMPLING 83

His net gain, however, would be $1.70 minus $.55 ($36.60 minus $36.05) or $1.1 5.

Apparently the advantage to the farmer of skimming a portion of his surplus milk will depend upon:

1. Relative value of price differential used in determining the value of the milk produced in excess of the base test (3.8 percent in this case) per pound of fat, and the market price of butter (which is used as basis for determining the value of the surplus milk).

2. Value of skimmilk for feeding.

3. Hauling costs.

It seems hardly logical that all the evident skimming mentioned above can be explained by a desire on the part of the farmer to secure the slight financial gain that would result from such a practice. Since the majority of these farmers are small producers, it seems more logical to assume that they use a certain amount of their milk, cream or skimmilk for table purposes, and so the milk varies in test from day to day.

The wide variation in daily milk tests that were found would make the use of periodic tests undesirable. Under such conditions composite milk samples would be most satisfactory.

SUMMARY AND CONCLUSIONS

This study of the sampling procedure followed on the Champaign- Urbana milk market was made to determine the accuracy of the meth- ods used. The completeness of mixing before sampling was determined at each of the four milk plants purchasing milk from more than 400 members of Champaign Milk Producers Association. Comparisons were made between the daily tests on fresh milk samples, the weekly tests on laboratory composites, and weekly tests on plant composites, as well as between the laboratory tests and the Association tests on the plant composites. Comparisons were also made between the tests of com- posite samples taken in aliquot portions and the mathematical average of the tests on daily samples taken with a dipper. From the data secured the following conclusions are drawn:

1. Inaccurate tests may result from improper mixing of the milk when dumped in the weigh tanks.

2. To determine the accuracy of sampling from the weigh tanks, samples taken from each tank without previous stirring of the milk should be checked against samples taken when the milk has been thoroly stirred.

84 BULLETIN No. 459

3. Tests on composite samples properly taken and kept will give an accurate measurement of the fat content of the milk.

4. Periodic testing would not be satisfactory on a market where variations in daily tests are as wide as those on the Champaign- Urbana market.

5. Variation in daily tests on milk from the same patron was sufficiently great to indicate mechanical manipulation of the fat content.

6. The tendency for plant composite samples to test less than laboratory composite samples is thought to be due to variations from the accepted practice in the care of the samples.

7. A system of double-checking the Association tests of the plant composites would be desirable and possibly profitable to the milk producers. It should not be necessary, however, to recheck each patron's samples in each test period.

8. Composite samples need not be taken in aliquot portions to give results that will be sufficiently accurate for practical purposes.

LITERATURE CITED

1. BAILEY, D. H. Methods of sampling milk. Perm. Agr. Exp. Sta. Bui. 310.

1934.

2. FARRINGTON, E. H. Composite milk samples tested for butterfat. 111. Agr.

Exp. Sta. Bui. 16, pp. 504-515. 1891.

3. HUNZIKER, O. F. Report of the Dairy Husbandry Department. Ind. Agr.

Exp. Sta. Ann. Rpt. 27 (1913-14), pp. 37-47.

4. MARQUARDT, J. C. and DURHAM, H. L. Sampling milk for fat test at milk

plants. N. Y. (Geneva) Agr. Exp. Sta. Bui. 605. 1932.

5. MEADE, DEVOE, and LECKIE, J. N. To what extent do tests from composite

samples and fresh samples of milk agree? Milk Plant Mo. 25, No. 8, 28-30. 1936.

6. MONROE, C. F. Accuracy of composite milk samples. /;/ Ohio Agr. Exp. Sta.

Bui. 446, pp. 127-128. 1930.

7. PATRICK, G. E. The composite sample at creameries. Iowa Agr. Exp. Sta.

Bui. 22, pp. 833-844. 1893.

8. - - The "Relative value plan" at creameries. Iowa Agr. Exp. Sta.

Bui. 9, pp. 356-369. 1890.

9. SANMANN, F. P. and OVERMAN, O. R. Periodic sampling compared with

composite sampling and the true average tests of monthly milk deliveries. The Creamery and Milk Plant Mo. 15, No. 7, 39-41. 1926.

10. — - The importance of proper storage of composite milk

samples. 'The Creamery and Milk Plant Mo. 15, No. 6, 39-41. 1926.

6050—11-39—17324

UNIVERSITY OF ILLINOIS-URBANA

• .

.

-

• ••

•

•

• . • • • •

•

• •

•

.

•

•

• • ;

' '

.

- • '

.

.

.

it I.* 1

'

-

'

' *

-

•

.

•

•

.