Professional Documents
Culture Documents
5, 2001 1499
FOOD COMPOSITION AND ADDITIVES
1500 KLEYN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001
Reference Testing
The Barbano Laboratory at Cornell University (Ithaca,
NY) conducted the Mojonnier ether extraction reference tests
for both studies using AOAC Method 989.05 (7). Ether extraction analyses were conducted in duplicate within 24 h of
receiving the test samples.
Statistical Analysis
In the original collaborative study protocol, laboratories
were instructed to analyze each test sample in duplicate by the
Gerber method. However, current AOAC guidelines for collaborative studies state that each test sample be analyzed once
and that blind duplicate test samples be used to determine
method repeatability parameters (8). To conform to current
AOAC guidelines, only the first result (uncensored) from the
duplicate analyses was used in the statistical analyses. Repeatability parameters were determined from the 2 materials (one
raw and one pasteurized homogenized) tested in blind duplicate in each study.
Using data from the first test of the duplicate analyses,
interlaboratory study results were determined by AOAC
guidelines (8). Test results from the Gerber method were
compared with ether extraction results using paired t-tests.
The significance level for outlier identification was set at =
0.025. The significance level for all other statistical analyses
was set at = 0.05.
KLEYN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001 1501
A. Principle
Milk is weighed into Gerber butyrometer containing
H2SO4. Isoamyl alcohol is added and butyrometer contents
are vigorously mixed to dissolve curd and release fat. Fat is
isolated by centrifugation and quantified in the graduated portion of the Gerber butyrometer.
B. Reagents
(a) Sulfuric acid.Specific gravity 1.8201.825 at
15.5C. Technical grade.
(b) Isoamyl alcohol.Specific gravity 0.8140.816 at
15.5EC, bp 128 to 133EC, free of water, acid, fat, and furfural.
(Caution: Vapors are poisonous.)
C. Apparatus
(a) Standard Gerber milk-test butyrometer (graduated
08%) with lock stopper and key.(See Figure 2000.18.)
Clear transparent, colorless, resistant borosilicate glass, annealed and free from defects; neck, large bulb, flat tube, and
small bulb on a straight median axis, joined smoothly to permit free flow of liquid; wall thickness adequate to provide sufficient strength but not less than 0.9 mm at any point. Total
length 190 2.5 mm.
(1) Small bulb.Tapered, with matte surface on side
above graduations for sample identification; #15.0 mm od;
capacity, measured between top graduation line and internal
end of bulb $1.5 mL.
(2) Large bulb.#25.0 mm od; capacity, 21.5 0.4 mL.
Table 2000.18A. Interlaboratory study results for percent fat in milk (g/100 g) determined by Gerber by weight
methoda
Sample ID
Mean
No. of labs
sr
sR
RSDr, %
RSDR, %
Raw
4.431
1011
0.023
0.053
0.52
1.20
0.065
0.148
Pasteurized, homogenized
3.677
1011
0.029
0.040
0.78
1.08
0.080
0.111
Mean based on analysis of 7 raw whole milk and 4 pasteurized homogenized whole milk, respectively. Standard deviations (sr, sR) and
prediction values (r, R) based on analyses of 8 raw milks (2 in blind duplicates) and 8 pasteurized homogenized milks (2 in blind duplicates),
for a total of 200 tests. Of the total of 200 tests, statistical outliers were identified for 5 raw milk tests and 5 pasteurized homogenized milk
tests. Relative standard deviations (RSDr, RSDR) were calculated as (sr/mean) 100 and (sR/mean) 100, respectively.
1502 KLEYN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001
240
1140
250
1120
260
1100
270
1080
300
1020
D. Determination
(a) Milk sample preparation.Place milk test sample in
H2O bath maintained at 39 1EC. Level of H2O should be at
or above milk level. Mix milk 10 by inversion. If fat line remains on inside surface of container, run hot tap H2O (ca
5060EC) over outside surface for 1520 s. Mix thoroughly
by inversion and weigh test portion immediately. Do not allow
milk to remain in H2O bath more than 15 min after reaching
38EC.
(b) Testing.Add 10.0 0.2 mL of H2SO4 at 1521EC to
butyrometer. Tare butyrometer containing H2SO4 on analytical balance. Weigh 11.13 0.03 g tempered milk test sample,
D(a), into butyrometer, adding milk slowly at first to prevent
charring and violent reaction with acid. Add 1 0.05 mL
isoamyl alcohol to butyrometer containing test portion. Insert
lock stopper securely using hand-held key. Wearing insulated
gloves, grasp butyrometer at graduated neck with stoppered
end up. Without allowing small bulb to empty, shake until all
traces of curd disappear. Holding butyrometer by both
stoppered end and graduated neck, invert at least 4 times to
mix acid remaining in the small bulb and graduated neck with
the contents of larger bulb.
Place butyrometers in centrifuge, small bulb pointing up,
and counterbalance. Centrifuge 4 min after proper speed is
reached. Transfer butyrometers to H2O bath maintained at
6063EC and immerse leaving only small bulb exposed. Let
fat column equilibrate for $5 min.
Remove one butyrometer from water bath and wipe dry.
Apply gentle pressure to lock stopper to bring bottom line of
fat column upward to coincide with nearest whole percentage
graduation mark. Promptly read scale at bottom of upper meniscus to nearest 0.05%.
E. Repeat Analysis
Repeat analysis if fat column is turbid or dark in color, or if
there is white or black material at bottom of fat column. Ac-
KLEYN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001 1503
Table 1. Fat in milk (g/100 g) determined by Gerber by weight method
Laboratory
Typea
Replicate
Raw
3.65
3.65
3.70
3.70
3.70
3.65
3.65
3.60
3.65
3.65
3.65
Raw
3.85
3.75
3.80
3.85
3.85
3.80
3.80
3.75
3.80
3.75
3.80
3.80
3.75
3.80
3.80
3.85
3.70
3.80
3.75
3.75
3.75
3.80
Raw
5.25
4.80b
5.30
5.25
5.25
5.10
5.15
5.20
5.15
5.20
5.15
Raw
5.55
5.30
5.50
5.55
5.60
5.50
5.40
5.40
5.45
5.50
5.50
PH
0.95
1.00
0.95
0.95
1.05
0.90
0.95
0.95
0.95
0.90
1.00
PH
1.95
1.95
1.90
2.00
2.00
1.95
1.95
1.90
2.00
1.95
1.90
PH
3.25
3.20
3.15
3.20
3.30
3.20
3.20
3.00b
3.20
3.15
3.20
Material
8
a
b
c
PH
3.20
3.20
3.20
3.25
3.25
3.20
3.20
3.10
3.20
3.20
3.25
4.30
4.30
4.30
4.25
4.30
4.25
3.20b
4.10c
4.25
4.20
4.00c
Fat in Milk
F. Calculations
Calculate % (w/w) fat content in milk as follows:
Fat, % =
reading at upper meniscus reading at lower meniscus
Ref.: J. AOAC Int. 84, XXXXXX(2001)
10
Typea
Replicate
Raw
1.35
1.45
1.40
1.40
Raw
1.45b
1.35
1.40
1.35
1.40
1.40
1.35
1.45
1.35
1.40
1.35
1.35
1.35
1.35
1.40
1.40
3.45
3.45
3.45
3.50
3.50
3.40
3.45
3.50
3.40
3.45
11
Raw
4.55
4.50
4.50
4.55
4.60
4.50
4.45
4.05
4.50
4.50
12
Raw
5.00
4.90
5.00
4.90
4.95
4.90
5.35c
4.85
4.95
4.80
13
PH
0.95
1.00
0.95
0.95
1.05
0.95
0.95
0.95
0.95
0.95
14
PH
1.95
1.95
1.90
1.95
2.00
1.95
1.85
1.90
1.95
1.90
15
PH
3.25
3.25
3.20
3.20
3.25
3.20
3.20
3.20
3.20
3.25
3.25
3.20
3.20
3.25
3.25
3.25
3.20
3.20
3.20
3.15
4.10
4.00
4.00
4.00
4.10
4.00
3.95
3.95
4.00
4.00
16
a
b
c
PH
1504 KLEYN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001
A. Principle
See Method I.
D. Determination
B. Reagents
See Method I.
C. Equipment
See Method I, and in addition:
(j) PipetTo deliver 10.77 mL at 20EC. Made with clear,
transparent, colorless, resistant borosilicate glass, well annealed and free from defects. The suction tube shall be cylindrical. The graduation mark shall be finely etched, of uniform
width #0.25 mm, durable, and carried completely around the
suction tube. The graduation mark shall be at right angles to
the axis of the pipet. The end of the suction tube shall be
firepolished and be at right angles to the axis of the pipet. Bulb
is cylindrical and evenly drawn out in both directions, with no
E. Calculation
See Method I.
Table 3. AOAC statistical parameters by test material for fat in milk (g/100 g) determined by Gerber by weight
method
Statistic
Material
Typea
No. of labs
No. of tests
Mean, %
sr
sR
RSDr, %
RSDR, %
No outliers removed
1
Raw
11
11
3.659
0.030
Raw
11
22
3.789
Raw
11
11
5.164
0.134
2.602
0.376
Raw
11
11
5.477
0.085
1.547
0.237
PH
11
11
0.959
0.044
4.556
0.122
PH
11
11
1.950
PH
11
22
3.196
PH
11
11
0.028
0.041
0.824
0.744
0.039
0.034
4.132
0.061
1.086
0.084
0.079
1.986
1.055
0.323
1.896
0.115
0.108
0.094
7.829
0.170
0.906
Raw
11
11
3.659
Raw
11
22
3.789
Raw
10
10
5.200
0.062
1.199
0.175
Raw
11
11
5.477
0.085
1.547
0.237
4.531
PH
11
11
0.959
PH
11
11
1.950
PH
10
20
3.210
PH
0.028
0.041
0.058
0.824
0.744
0.744
0.044
0.035
2.597
0.027
3.564
0.028
1.086
1.290
0.084
0.079
0.079
4.556
0.039
0.027
4.269
Summary of PH materials
0.030
0.028
1.095
0.039
0.853
0.050
0.780
0.037
0.165
0.122
1.986
0.853
0.115
0.108
0.077
0.098
1.499
0.077
0.110
1.393
0.079
0.140
0.871
0.104
KLEYN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001 1505
Table 4. AOAC statistical parameters by test material for fat in milk (g/100 g) determined by Gerber by pipet method
Statistic
Material
Typea
No. of labs
No. of tests
Mean, %
sr
sR
RSDr, %
RSDR, %
0.037
1.977
2.677
0.077
0.104
No outliers removed
9
Raw
10
20
1.385
0.027
10
Raw
10
10
3.455
0.037
1.068
0.103
11
Raw
10
10
4.470
0.153
3.425
0.429
12
Raw
10
10
4.960
0.151
3.035
0.422
13
PH
10
10
0.965
0.034
3.497
0.094
14
PH
10
10
1.930
15
PH
10
20
3.218
16
PH
10
10
0.042
0.030
4.010
0.030
2.185
0.919
0.052
0.919
0.118
0.083
1.288
0.083
0.145
Raw
18
1.383
10
Raw
10
10
3.455
0.037
1.068
0.103
11
Raw
4.517
0.043
0.959
0.121
12
Raw
4.917
0.066
1.345
0.185
3.568
PH
10
10
0.965
14
PH
10
10
1.930
15
PH
10
20
3.218
PH
10
10
4.010
16
Summary of PH materials
0.017
0.017
0.035
0.047
1.205
1.205
0.034
0.030
1.315
0.047
0.047
3.497
0.042
0.030
2.538
0.052
0.919
0.133
0.094
2.185
0.919
0.098
0.118
0.083
1.288
0.083
0.145
2.531
0.030
0.040
0.919
1.593
0.083
0.114
3.049
0.024
0.044
0.788
1.435
0.068
0.124
elimination of outliers was similar to that of all other materials. As a result, the decision was made to include all data from
material 13 in the analysis.
Interlaboratory study results, with and without statistical
outliers removed, for Gerber by weight and Gerber by pipet
methods are presented in Tables 3 and 4, respectively. There
were no meaningful differences in interlaboratory study results (statistical outliers removed) between raw and pasteurized materials when parameters were summarized by material
type. When performance parameters for raw and pasteurized
homogenized materials were pooled within method (statistical
outliers removed), interlaboratory study results for Gerber by
weight and Gerber by pipet were similar.
The milk fat content of the 8 materials ranged from 0.959
to 5.477% for Gerber by weight method and from 0.965 to
4.917% for the Gerber by pipet method. There was a statistically significant (p < 0.05, linear regression) increase in RSDR
as fat content decreased for both methods. No significant
trends for sR were observed. Only 2 estimates of repeatability
1506 KLEYN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001
Table 5. Comparison of milk fat test results by the ether extraction method with the Gerber by weight and Gerber by
pipet methods
% Fat, Gerber by weight vs ether extraction
Gerber by
pipet
Ether
extraction
Differenceb
13
PH
0.965
1.038
0.073
Raw
1.383
1.457
0.074
Typea
Gerber by
weight
Ether
extraction
Differenceb
Material
PH
0.959
1.012
0.053
PH
1.950
1.994
0.044
Material
PH
3.210
3.213
0.003
14
PH
1.930
1.987
0.057
Raw
3.659
3.672
0.013
15
PH
3.218
3.273
0.056
Raw
3.789
3.797
0.008
10
Raw
3.455
3.498
0.043
PH
4.269
4.263
0.006
16
PH
4.010
4.054
0.044
Raw
5.200
5.200
0.000
11
Raw
4.517
4.579
0.063
Raw
5.477
5.503
0.025
12
Raw
4.917
4.998
0.081
3.564
3.582
0.017
Mean
3.049
3.111
0.061
Mean
a
b
performance were available for each method; therefore, statistical tests to determine the relationship between fat level and
repeatability parameters were not conducted.
Because Gerber interlaboratory study results were unaffected by material type (raw or pasteurized, homogenized) or
method of test portion addition (weight or pipet), performance
parameters determined for all materials and methods were
pooled to estimate overall Gerber interlaboratory study results
and increase the number of materials used to determine repeatability (Table 6). Repeatability and reproducibility standard deviations, as well as the r and R values, were not correlated with fat level, and, thus, each is summarized by a single
parameter. Because the sR was fairly constant, relative
reproducibility (RSDR) was greater at lower fat levels (12%)
Table 6. Comparison of interlaboratory study statistics for determination of fat in milk (g/100 g) by Gerber, Babcock,
and ether extraction methods
Interlaboratory study results, % fat in milk (g/100 g)
AOAC
Method No.
Method
Material type
Current study
Gerbera
Raw, PH
Mean
sr
sR
4.156
0.026
0.047
1.437
RSDr, %
RSDR, %
0.074
0.132
989.04
Babcockb
Raw
4.110
0.037
0.047
0.901
1.147
0.105
0.133
989.05
Ether extractionc
Raw, PH
3.886
0.015
0.020
0.396
0.512
0.044
0.056
b
c
sr and sR values for the Gerber method were calculated by variance averaging individual material performance statistics (16 materials) with
outliers removed. RSDs were calculated from grand sr and sR using means from 11 milk materials within 26% fat (14, 7, 8, 1012, 15, and
16) and the 5 milk materials within 12% fat (5, 6, 9, 13, and 14).
Ref. 1.
Ref. 2.
KLEYN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001 1507
than fat levels found in whole milk. Trends in RSDr could not
be identified because of the limited number of materials examined, but the nature of the test (readability maximum
0.05% fat) suggests that RSDr would increase proportionally
in lower (12%) fat milk. For this reason, relative repeatability
and relative reproducibility are summarized by fat level.
Table 6 presents the summary method performance statistics and compares them with other AOAC chemical methods
used to determine fat content of milk. The within- and between-laboratory performance of the Gerber method is very
similar to that of the Babcock method (AOAC Method
989.04; 1). Because the Babcock method is limited to raw
milk testing, the Gerber method has the advantage of application to pasteurized homogenized milk. The interlaboratory
study results of ether extraction (AOAC Method 989.05; 7),
the reference test, are clearly superior to either the Babcock or
the Gerber methods, but for routine application where the authority and precision of reference testing is not required, both
the Gerber and Babcock tests have acceptable and similar performance when conducted properly.
Compared with the ether extraction reference method, fat
test results were slightly lower (0.017% absolute fat) for the
Gerber by weight method and significantly lower (0.061%)
for the Gerber by pipet method. Of course, these differences
are crude estimates only, as the amount of data available was
limited. The original justification for selecting a weight addition of 11.125 g (11.13 0.03 g) and a pipet that delivered
10.77 mL was evidence of good agreement with ether extraction in an in-house comparison (4). However, the in-house
Mojonnier ether extraction reference method in use at that
time had not been collaboratively studied and specified 2 extractions (5). The current Mojonnier ether extraction method
(AOAC Method 989.05; 7), collaboratively studied, requires
3 extractions to maximize fat recovery and improve precision
(7, 9). In the Gerber collaborative study, the participation of
many laboratories to arrive at Gerber test values and the use of
Mojonnier Method 989.05 with 3 extractions to optimize fat
recovery and precision increase the ability to detect biases and
define differences between methods.
Compared with ether extraction, a trend towards underestimating fat content at lower fat concentrations was observed
for the Gerber by weight method but not for the Gerber by
pipet method. This difference may be due to the effect of density on the amount of sample actually delivered by the pipet.
The density of milk decreases as fat content increases, and the
weight of the test portion delivered increases with decreasing
milk fat content when a constant volume is used.
A weakness in the collaborative study was that only one
laboratory conducted the ether extraction reference test. This
was somewhat mitigated by the expertise of this laboratory,
which is under the direction of the AOAC Associate Referee
for the Mojonnier ether extraction method (9). Additionally,
this laboratory participates in a formalized interlaboratory
proficiency testing program (7 materials in blind duplicate every 2 months). This program, begun in 1990, documents that
method bias for this laboratory is small, averaging 0.004%
lower in fat than all laboratory means (10). In terms of
within-laboratory precision, the repeatability standard deviation for the ether extraction reference testing for the Gerber
studies (20 laboratory samples, tested in known duplicate, representing 16 materials) was 0.007% fat, which is well within
the sr of 0.015% fat for the method (7).
Recommendations
The Associate Referee recommends that the Gerber by
weight method be adopted Official First Action with applicability limited to whole milk because of limited evidence of a
bias between Gerber by weight and ether extraction results at
fat levels of 1 to 2%. At the request of AOAC INTERNATIONAL, the interlaboratory study results accompanying the
method description are listed separately for the raw whole
milk and pasteurized homogenized whole milk materials, although performance for the 2 types of whole milk materials
was equivalent.
At this time, the Associate Referee does not recommend
adopting the Gerber by pipet method using the 10.77 TD pipet
because of a significant bias between this method and ether
extraction results at all fat levels tested. When a suitable pipet
volume is identified, the Gerber by weight method can be
modified to include the option of either weighing or pipeting
the test portion.
Acknowledgments
The authors thank Dr. Kleyns family, including his brothers Carl and John, his children Frank and Allison, and sister-in-law Mary, for giving us permission to publish his material. The associates of Dr. Kleyn felt it important to bring his
work to completion. Jeffrey Bloom at Weber Scientific organized the effort. Researchers at Cornell University, Joanna M.
Lynch and David M. Barbano, evaluated the data and conducted the statistical analysis. Martin W. Mitchell of Certified
Laboratories, Inc. assumed the role of Associate Referee to
oversee the method now and in the future. This report was prepared as a tribute to Dr. Kleyn for his lifelong contributions to
the dairy industry and the field of dairy chemistry.
The authors thank the following collaborators for their participation in the study:
Lynda S. Cooper, Velda Farms, Miami, FL
Elizabeth Cusak, Certified Laboratories, Corona, NY
Mike Fick, Crowley Foods, Inc., Weeks Division, Concord, NH
Tom Hanks, Wyoming Department of Agriculture, Division of Laboratories, Laramie, WY
Mary K. Hesen, Valley Farms Dairy, Inc., Williamsport,
PA
Julie Johnson, Dairy Quality Control Institute, St. Paul,
MN
Dick H. Kleyn, Rutgers University, Cook College, New
Brunswick, NJ
Fayette Mercer, Florida Department of Agriculture, Central Dairy Laboratory, Winter Haven, FL
Denise Monahan, Welsh Farms, Long Valley, NJ
1508 KLEYN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001
(4) Kleyn, D.H., Trout, J.R., & Weber, M. (1988) J. Assoc. Off.
Anal. Chem. 71, 851853
(5) Standard Methods for the Examination of Dairy Products
(1985) 15th Ed., G.H. Richardson (Ed.), American Public
Health Association, Washington, DC
(6) Vester, C.F. (1959) Neth. Milk Dairy J. 13, 5972
(7) Official Methods of Analysis (1995) 16th Ed., AOAC INTERNATIONAL, Gaithersburg, MD, sec. 33.2.26
(8) AOAC INTERNATIONAL (1997) AOAC Official Methods
Program Manual on Development, Study, Review, and Approval Process for AOAC Official Methods, AOAC
INTERNATIONAL, Gaithersburg, MD
(9) Barbano, D.M., Clark, J.L., & Dunham, C.E. (1988) J. Assoc.
Off. Anal. Chem. 71, 898914
(10) Lynch, J.M., Barbano, D.M., Healy, P.A., & Fleming, J.R.
(1994) J. AOAC Int. 77, 976981