You are on page 1of 8

V A N E R E M E T A L . : JOURNAL OF A O A C INTERNATIONAL V O L . 81, No.

1,1998 25

AGRICULTURAL MATERIALS

Determination of Water in Forages and Animal Feeds by Karl


Fischer Titration
TERRI VAN EREM and NANCY THIEX 1
South Dakota State University, Department of Chemistry and Biochemistry, Box 2170, Brookings, SD 57007-1217
JORG POHMER
Allied Signal, Inc., 101 Columbia Rd, PO Box 1053, Morristown, NJ 07962-1053

Downloaded from https://academic.oup.com/jaoac/article/81/1/25/5684171 by guest on 02 October 2021


W. MICHAEL POFFENBARGER
Hoechst Celanese Corp., U.S. Highway 43 North, PO Box 64, Bucks, AL 36512
VICTOR SMITH and ELENA PATEL
Brinkmann Instruments, Inc., PO Box 1019, Westbury, NY 11590-0207

Oven methods for determining moisture (volatiles) enced by the presence of volatiles other than water. An
in forages and other animal feeds are empirical. analytical tool is needed to evaluate the appropriate-
The moisture concentration obtained depends ness of various oven methods for different types of
upon the time and temperature the sample was animal feeds and forages. A method specific for water
dried and is influenced by the presence of other could be considered a reference method. The goal of
volatiles than water. A validated reference method this study was to establish an accurate, reproducible,
to measure water in forages and animal feeds yet practical method for determining water in dry,
could be used to evaluate the appropriateness of ground forages and animal forages and feeds. The
oven methods for various types of animal feeds method should have reasonable throughput so that it
and forages. Karl Fischer titration is a well-estab- could be used routinely in feed and forage laboratories.
lished method for determining water. However, A second objective was to make a preliminary compari-
thorough extraction of water from forages and son of the method to various oven methods for forages
feeds is a challenge because they often contain and animal feeds.
cellular structures that release water slowly. Water Karl Fischer titration is a well-established general
was successfully extracted into methanol-forma-
method for determining water (1,2). However, a tech-
mide (50 + 50) by high-speed homogenization and
nique to efficiently extract water from these types of
then titrated directly at 50 C with a one-component
materials and to subsequently analyze the extract by
Karl Fischer reagent based on imidazole. The
Karl Fischer titration was not available. We tried un-
method is described in detail, results of day-to-day
successfully to extract samples with methanol in a Spex
repeatability and laboratory-to-laboratory repro-
ducibility are reported, and preliminary compari- Model 8000 Mixer/Mill with 3-1.27 cm diameter steel
son data between oven methods are provided. balls. Methods using this approach for different materi-
als were reported (3,4). We experienced problems with
leaky seals leading to loss of solvent and dilution error.
here is a great deal of discussion and disagree- Error also was introduced during transfer of solvent

T ment in the analytical community as to which


oven method is the best indicator of the moisture
contents of forages and animal feeds. Because of the
into the titration flask, especially when humidity was
high. Throughput was very low because of the laborious
procedure. This method was not desirable for routine
empirical nature of oven methods, moisture concentra- (non research) applications. We considered another
tions obtained with a given method depend on the time approach, which involves heating samples in N,N-di-
and temperature the sample was dried and are influ- methylformamide (DMF) at 90°C for 60 min (5). We
found this approach undesirable because of potential
Received January 23, 1997. Accepted by RN August 11, 1997. error introduced by atmospheric moisture because
Author to whom correspondence should be addressed. DMF is hygroscopic, and because it involves direct
Published as South Dakota Agricultural Experiment Station Jour- handling of DMF.
nal Series 2971. A number of advantages are associated with high-
This paper reports results of research only. Mention of a trade-
mark, proprietary product, or vendor does not constitute a guarantee
speed homogenization of forage and animal feeds sam-
or warranty of the product and does not imply approval to the ples directly in the Karl Fischer titration vessel in
exclusion of other products or vendors that may also be suitable. methanol-formamide (50 + 50) combined with subse-
26 VAN EREM E T AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 81, No. 1, 1998

quent titration at elevated temperatures. The tech- glycol monoethyl ether, and hydriodic acid—and should
nique had been reported before (6). It provides quanti- be handled with care. Methanol: Highly flammable;
tative extraction of water from cellular structure in a toxic by inhalation, in contact with skin, and if swal-
short time. Other components coextracted with water lowed. Formamide: Moderately irritating to skin, mu-
(e.g., oils and carbohydrates) do not interfere because cous membranes.
the Karl Fischer method is based on a specific chemical
reaction consuming water. Sample Preparation
Grind samples to pass a 1 mm, or smaller, opening
Proposed Karl Fischer Method to ensure homogeneity.
Scope Drying or Conditioning the Cell
The proposed method is applicable for determina- Dry titration cell by completing a pretitration as

Downloaded from https://academic.oup.com/jaoac/article/81/1/25/5684171 by guest on 02 October 2021


tion of water in dry, ground animal feeds and forages. follows: Dispense 80 mL methanol-formamide solvent
into titration vessel. Close cell to minimize addition of
Reagents atmospheric moisture. Heat to 50° ± 1°C. Dry cell (in-
(a) Karl Fischer reagent.—One component, based on cluding solvent, cell walls, electrode walls, generator,
imidazole, with titer of ca 5 mg H 2 0 / m L reagent, cell atmosphere) by performing a blank run, including
Hydranal Composite 5 (Riedel-de Haen 34805, St. homogenization and titration. Homogenize at
Louis, MO), or equivalent. 24 000 RPM for 60 s. Begin titration immediately. Be-
fore reaching end point, homogenize momentarily to
(b) Methanol.—Anhydrous, for moisture determina-
rinse untitrated methanol that may have been sitting
tions, water content not to exceed 0.05% (EM Science,
under the cap into solution. The end point is reached
Gibbstown, NJ).
when no change in potential is observed for 10 s (titra-
(c) Formamide.—ACS reagent grade (Fisher Scien- tion system programmed for stop criterion: time; delay:
tific, Fair Lawn, NJ). 10 s). A dried titration cell has a maximum drift con-
(d) Solvent.—50% methanol and 50% formamide. sumption of 5-10 uX Karl Fischer reagent/min.
(e) Sodium tartrate dihydrate.—Primary standard
(water content, 15.66 + 0.05%), Hydranal standard Standardization
sodium tartrate-2-hydrate (Riedel-de Haen 34803, St. Heat cell to 50° ± 1°C. Dry cell as described above.
Louis, MO), or equivalent. Immediately after drying the cell, quickly weigh
150-250 mg sodium tartrate dihydrate standard into
Apparatus
glass weighing spoon and record weight of spoon and
(a) Karl Fischer titration-homogenization system.— standard to nearest 0.1 mg (S). Quickly transfer
Metrohm 720 KFS Titrino (Brinkman Instruments, Inc., weighed sample into the titration flask through septum
Westbury, NY), includes titration vessel LP with water stopper. Reweigh empty sample spoon to obtain tare
jacket (50-150 mL), snap-in buret unit (10 mL), weight (T). Obtain weight of standard material added
703 titration stand with pump, cable with timer by subtracting tare weight (T) from weight of spoon
(Titrino-Polytron), or equivalent. plus standard (S). Record weight of standard material
(b) Circulating water bath.—Capable of maintaining (S — T) in mg to the nearest 0.1 mg. Homogenize at
50° ± 1°C (Fisher Scientific, Pittsburg, PA). 24 000 RPM for 60 s. Begin titration immediately. Be-
fore reaching end point of titration, homogenize mo-
(c) Homogenizer.—Brinkmann Polytron homoge-
mentarily to rinse down moisture that may have been
nizer with PTA 20TSM foam-reducing generator with
sitting under the cap. This ensures that moisture of
saw teeth and knives. Assembled with generator ex-
particles sitting under the cap or hanging on the glass
tending into the titration vessel and adjusted to be ca
walls above the liquid surface is titrated. Titrate to
1 in. from the bottom of the titration vessel. Set to
same endpoint as described in the conditioning step,
provide a homogenization speed of 24 000 RPM.
recording volume of reagent required for titration (mL
(d) Balance.—Analytical, electronic, sensitive to reagent) to the nearest 0.001 mL. Repeat a total of
0.1 mg (Mettler Instruments, Hightstown, NJ). 5 times. Calculate titer and then average the 5 values.
(e) Glass weighing spoon.—With opening for dis- The relative standard deviation should be < 2%.
pensing sample into the titration flask through the
septum stopper (Brinkmann 6.2412.000 or equivalent)
(Brinkman Instruments, Inc., Westbury, NY). mL reagent

Safety mg N a 2 C 4 H 4 0 6 • 2 H 2 0 X 0.1566
mL reagent
Hydranal Composite 5 contains 5 hazardous compo-
nents—iodine, sulfur dioxide, imidazole, diethylene where mg N a 2 C 4 H 4 0 6 • 2 H 2 0 is S — T, in mg.
VAN EREM ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 81, No. 1,1998 27

Determination different days. To test method comparability, the same


5 samples were analyzed in triplicate by 4 oven meth-
Heat cell to 50° + 1°C. Dry cell as described above ods on the same 3 days as the repeatability tests. The
before running a sample or each time solvent is replen- oven methods chosen were selected because of their
ished in titration cell. Check drift in titration cell. A frequent use in laboratories analyzing these types of
dried titration cell has a maximum drift consumption of samples: Oven Method I (7), drying in a forced draft
5-10 |JLL Karl Fischer reagent/min. oven at 135°C for 2 h; Oven Method II (8), drying in a
Analyze sample as follows: Immediately after drying forced draft oven at 104°C for 3 h; Oven Method III
cell, quickly weigh ca 0.5 g sample (to contain ca 2 5 - (9), drying in a vacuum oven at 95°C for 5 h; and Oven
50 mg water) into glass weighing spoon and record Method IV (10), drying in a forced draft oven at 105°C
weight of spoon plus sample (W). Quickly add weighed for 16 h. Methods I and III are AOAC official methods
sample into titration flask through septum stopper. for animals feeds and plants, Method II is an AOAC
Reweigh empty sample spoon and record tare weight official method for malt, and Method IV is an AOAC

Downloaded from https://academic.oup.com/jaoac/article/81/1/25/5684171 by guest on 02 October 2021


(T). Obtain sample weight by subtracting tare weight official method for peat. To test method reproducibil-
(T) from weight of spoon plus sample (W). Record ity, the 5 samples were distributed to 3 different labo-
sample weight (W — T) in grams to the nearest 0.1 mg. ratories for analysis.
Homogenize at 24 000 RPM for 60 s. Begin titration
immediately. Before reaching end point of titration, Oven Methods
homogenize momentarily to rinse down moisture that Apparatus
may have been sitting under the cap. This ensures that
moisture of particles sitting under cap or hanging on (a) Drying oven, forced draft convection.—Capable of
glass walls above the liquid surface is titrated. The end maintaining temperature at ±1°C. Oven should be
point is reached when no change in potential is ob- equipped with a metal rod shelf to allow circulation of
served for 10 s (stop criterion: time, delay: 10 s). Re- air. Exhaust vent on top of oven should be wide open.
peat determination in triplicate. Relative standard de- Gallen Kamp Model Plus II, Leicester, UK.
viation of replicates should be < 5 % . The cell need not (b) Drying oven, vacuum oven.—Equipped with
be emptied between each sample. Usually about pressure gauge and needle-type air valve. Oven should
3 titrations can be performed before the cell needs to be equipped with system to dry input air. Precision-
be emptied and replenished. Thelco Model 19, Chicago, IL.
(c) Vacuum pump.—Connected to vacuum oven via
Calculations vacuum tubing, Welch Model 1404, Skokie, IL.
(d) Moisture pans.—Aluminum, 50 id X 20 mm
deep; lids are optional, Fisher Scientific, Itasca, NY.
H 2 0 , m g = VX titer (e) Analytical balance.—Electronic, sensitive to
V X titer 0.1 mg. (Mettler Instruments).
H
20, % = — : — (f) Desiccator. (Corning, Corning, NY).
10 X sample weight
Determination
where V is volume of titrant (mL) and sample weight is Dry moisture pans at > 100°C for at least 2 h. Trans-
W- T(g). fer moisture pan to a desiccator and cool. (A desiccator
full of containers should cool for at least 1 h but not
Method Repeatability, Reproducibility, and more than 2-3 h.) Weigh moisture pan; record tare
Comparability weight (T) in g to nearest 0.1 mg. Tare pan to zero
while on balance. Add 1.5-2 g well-mixed sample to
Five samples representing various types of forages, pan. Record sample weight (S) in g to nearest 0.1 mg.
animal feeds, and pet food were selected for repeatabil- Repeat each analysis in triplicate. Gently shake pan to
ity testing, reproducibility testing, and for comparisons uniformly distribute sample and expose maximum area
with oven methods. Samples included corn silage, tex- for drying. Place samples in preheated oven. Leave
turized feed, baked dog food, total mixed ration (TMR), space around each pan to allow free air circulation.
and legume hay. Corn silage, TMR, texturized feed, Dry at the temperatures and times indicated in Table 1.
and legume hay were dried at 60°C overnight. Samples Oven should come to temperature ca 5 min after sam-
were ground to pass a 0.75 mm screen in a Brinkmann ples are put into it.
Retch Mill Model ZM-1. Samples were split and placed Transfer dried samples to desiccator and cool.
in air-tight, heat-sealed plastic containers to minimize Weigh; record weight of pan and dried sample (D) in g
moisture change. Analysts were instructed to open bags to nearest 0.1 mg. Moisture pans can be cleaned by
on the day of analysis. blowing with air. Note: Official moisture methods call
To test method repeatability, 5 samples were ana- for use of a lid with moisture pan. Place lid on bottom
lyzed at South Dakota State University (SDSU) on 3 of pan. Dry and weigh empty pan and lid together. As
28 V A N E R E M E T A L . : JOURNAL OF A O A C INTERNATIONAL V O L . 81, No. 1, 1998

Table 1. Oven conditions


Oven
method Temperature, °C Drying time, h Oven type

135 ± 2 2 Forced draft


104 + 1 3 Forced draft
95-100 5 Vacuum oven
(<100 mm Hg)
IV 105 ± 1 16 Forced draft

dried sample is being removed from oven, close mois- Results and Discussion

Downloaded from https://academic.oup.com/jaoac/article/81/1/25/5684171 by guest on 02 October 2021


ture pan with its lid.
Method Repeatability
Calculations
Five samples representing various types of forages,
(D - T) X 100 animal feeds, and pet food were analyzed at SDSU on
Dry matter, % 3 different days. Means, standard deviations (SDs), and
coefficients of variation (CVs) for each day are re-
Moisture, % = 100 - % dry matter ported by product in Table 2. Combining results from
3 days, intralaboratory CVs ranged from 1.41 for dog
food to 5.09 for TMR. There were no statistical differ-
where D is weight of pan and dried sample (g), T is ences between days (method repeatability) for corn
tare weight of pan (g), and S is sample weight (g). silage, TMR, dog food, or legume hay. Only the mean
Data were statistically analyzed by using PROC for day 1 differed from those of day 2 and day 3 for
ANOVA in SAS (11). texturized feed (P = 0.012). On the basis of these

Table 2. Results of repeatability studies 2

Water found

Feed Day n Mean, % SD CV

Corn silage 1 3 4.63 b 0.05 1.08


2 3 4.72b 0.13 2.75
3 4 4.74 5 0.21 4.40

Combined days 10 4.70 0.16 3.31


b
Texturized feed 1 4 7.21 0.29 4.05
2 3 7.65 c 0.07 0.87
3 3 7.82 c 0.08 1.02

Combined days 10 7.52 0.20 2.64


b
Dog food 1 3 6.94 0.07 1.04
2 3 6.94 fa 0.09 1.30
3 3 7.05 b 0.13 1.78

Combined days 9 6.97 0.10 1.41


b
Total mixed ration 1 4 4.83 0.27 5.60
2 7 5.24 fa 0.29 5.50
3 3 5.08 b 0.09 1.83

Combined days 14 5.09 0.26 5.09


b
Legume hay 1 4 6.89 0.26 3.83
2 6 6.93" 0.31 4.51
3 3 6.58 fa 0.08 1.14

Combined days 13 6.84 0.27 3.90


a
Means within feeds with the same superscript letter are not statistically different (P = 0.05).
V A N E R E M E T A L . : JOURNAL OF A O A C INTERNATIONAL V O L . 81, No. 1,1998 29

results, method precision is adequate, and the method practical differences and were attributed to differences
is considered to be repeatable. in intralaboratory variability. All laboratories matched
on 2 of 5 samples, and 2 of 3 laboratories matched on
Method Reproducibility the other 3 samples. Interlaboratory variability was ac-
ceptable, and the method was sufficiently reproducible.
Five samples were distributed to 3 different labora- Comparison with Oven Methods
tories for analysis. Means, SDs, and CVs for each
laboratory are reported by product in Table 3. Combin- Five samples were analyzed in triplicate by each of
ing results from 3 laboratories, interlaboratory CVs 4 oven methods on the same 3 days as repeatability
ranged from 1.64 for dog food to 4.46 for TMR. There tests were conducted. Means, SDs, and CVs are re-
were no statistical differences between laboratories ported Table 4. For corn silage, Karl Fischer and oven
(method reproducibility) for TMR or legume hay. method II with respective means of 4.70 and 4.55%

Downloaded from https://academic.oup.com/jaoac/article/81/1/25/5684171 by guest on 02 October 2021


Means for laboratory 3 were different from those of were similar. All other methods were different (P =
laboratories 1 and 2 for corn silage (P = 0.0015) and 0.05). For corn silage, oven methods I and IV gave high
texturized feed (P = 0.0001). Means for laboratory 1 means of 6.33 and 5.57%, respectively. Oven method III
differed from those of Laboratories 2 and 3 for dog gave the lowest mean, 4.00%.
food (P = 0.0002). Differences between laboratory 3 Karl Fischer, oven method II, and oven meth-
and laboratories 1 and 2 for corn silage and texturized od III gave similar means of 7.52, 7.08, and 7.45% for
feed were apparently due to differences in precision texturized feed and 5.09, 4.85, and 4.95%, respectively,
between the laboratories. Generally, laboratory 3 had for TMR. All other methods were different (P = 0.05).
the least experience with Karl Fischer methods and The means for texturized feed and TMR for oven
also had more variability on all samples. Laboratories 1 method I and oven method IV of 9.17 and 8.22%, and
and 2 generally had extremely low SDs. Single-day, 6.57 and 5.86%, respectively, were considerably higher
intralaboratory CVs for laboratories 1 and 2 were ex- than those from the other 3 methods.
cellent, ranging from 0.66 to 1.79. Therefore, although For dog food, Karl Fischer and oven method I gave
some statistical differences were found between 2 labo- similar means of 6.97 and 7.03%. All other methods
ratories on 3 of 5 samples, they were not considered were different (P = 0.05). Means for oven methods II,

Table 3. Results of reproducibility studies 9

Water found
Feed Laboratory n Mean, % SD CV

Corn silage 10 4.85 b 0.03 0.66


3 4.98fa 0.07 1.51
9 4.70 c 0.15 3.24
All 21 4.81 0.10 2.17
b
Texturized feed 1 5 8.37 0.12 1.45
2 3 8.06" 0.09 1.15
3 10 7.52 c 0.33 4.38
All 18 7.85 0.27 3.38
Dog food 1 10 7.25' 0.13 1.79
2 3 7.00c 0.10 1.48
3 9 6.97c 0.10 1.47
All 22 7.10 0.12 1.64
fa
Total mixed 1 10 5.11 0.08 1.50
ration 2 3 5.05 b 0.04 0.73
3 14 5.09 b 0.30 5.92
All 27 5.09 0.23 4.46
b
Legume hay 1 10 7.01 0.08 1.13
2 3 6.79 6 0.04 0.66
3 13 6.84" 0.28 4.17
All 26 6.90 0.21 3.08
Means within feeds with the same superscript letter are not statistically different (P - 0.05).
30 VAN EREM ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 81, No. 1, 1998

Table 4. Comparison of proposed Karl Fischer (KF) method with various oven methods*

Water found
Feed Method n Mean, % SD CV

b
Corn silage Proposed KF 10 4.70 0.15 3.10
Oven method 1 9 6.33° 0.34 5.30
Oven method II 9 4.55b 0.45 9.83
Oven method III 9 4.00cf 0.58 14.42
Oven method IV 9 5.57e 0.28 5.05

Texturized Proposed KF 10 7.52" 0.33 4.38


feed Oven method 1 9 9.17c 0.44 4.73
Oven method II 9 7.08b 0.36 5.08

Downloaded from https://academic.oup.com/jaoac/article/81/1/25/5684171 by guest on 02 October 2021


Oven method III 9 7.45fa 0.84 11.34
Oven method IV 9 8.22d 0.26 3.16

Dog food Proposed KF 9 6.97" 0.10 1.47


Oven method 1 9 7.03fa 0.12 1.71
Oven method II 9 6.17c 0.31 5.09
Oven method III 9 6.38 d 0.11 1.77
Oven method IV 9 6.61e 0.13 1.89

Total mixed Proposed KF 14 5.09fa 0.30 5.92


ration Oven method 1 9 6.57c 0.34 5.22
Oven method II 9 4.85b 0.21 4.25
Oven method III 9 4.95b 0.41 8.33
Oven method IV 9 5.86d 0.14 2.39

Legume hay Proposed KF 13 6.84b 0.29 4.17


Oven method 1 9 7.81c 0.20 2.54
Oven method II 9 6.00 d 0.16 2.61
Oven method III 9 6.21e 0.20 3.28
Oven method IV 9 6.69b 0.22 3.32

Means within feeds with the same superscript letter are not statistically different (P = 0.05).

Ill, and IV of 6.17, 6.38, and 6.61% were all lower than II and III of 6.00 and 6.21%, respectively, were lower
the 2 other methods. than those for other methods.
For legume hay, Karl Fischer and oven method IV A summary of CVs and variances for the Karl Fis-
gave similar means of 6.84 and 6.69%. All other meth- cher and various oven methods is provided in Table 5.
ods were different (P = 0.05). The mean for oven CVs averaged for all feeds were 3.81, 3.90, 5.37, 7.83,
method I of 7.81% was considerably higher than those and 3.16 for Karl Fischer and oven methods I, II, III,
for 2 other methods, and the means for oven methods and IV, respectively. Compared with oven methods, the

Table 5. Comparison of variability of proposed Karl Fischer method with various oven methods

Corn Texturized Dog Total mixed Legume


Method silage feed food ration hay Mean

Proposed Karl CV 3.10 4.38 1.47 5.92 4.17 3.81


Fischer Variance 0.021 0.109 0.010 0.091 0.081

Oven method 1 CV 5.30 4.73 1.71 5.22 2.54 3.90


Variance 0.112 0.190 0.014 0.118 0.040

Oven method II CV 9.83 5.08 5.09 4.25 2.61 5.37


Variance 0.200 0.130 0.099 0.042 0.024

Oven method III CV 14.42 11.34 1.77 8.33 3.28 7.83


Variance 0.321 0.712 0.013 0.171 0.042

Oven method IV CV 5.05 3.16 1.89 2.39 3.32 3.16


Variance 0.079 0.068 0.016 0.020 0.049
VAN EREM ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 81, No. 1,1998 31

proposed Karl Fischer method performed well. Be- Oven Method IV overestimated water in corn silage,
cause sample size (about 0.5 g) for the proposed method texturized feed, and TMR and underestimated water in
is small relative to sample sizes (about 2 g) for oven dog food. For all feeds, the bias for oven method IV
methods, the precision obtained with the proposed ranged from -0.37 to 0.87%, which is 95-120% of the
method would be most affected by sample heterogene- Karl Fischer result.
ity. Despite the smaller sample size, CVs for the pro-
posed method were lower than those for all oven Conclusions
methods for corn silage and dog food. For texturized
feed and TMR, CVs for the proposed method were The proposed Karl Fischer method is acceptable for
comparable with those of oven methods. The highest determining water in animal feeds and forages on the
CV observed for legume hay was for the proposed Karl basis of repeatability, reproducibility, and comparison
Fischer method. This sample type was probably the with results of 4 oven methods. Results of the method
most heterogeneous; thus the smaller sample size would are not influenced by the presence of other volatile

Downloaded from https://academic.oup.com/jaoac/article/81/1/25/5684171 by guest on 02 October 2021


have the greatest effect on precision for this sample. components of feed samples, as are results of oven
Average CV for the proposed Karl Fischer method was methods. Throughput for the proposed method is good,
3.81%, and the range was from 1.47 to 5.92%. Only relative to other Karl Fischer methods used in the
oven method IV had a lower average CV of 3.17%. author's laboratory. About 25-30 samples can be ex-
The differences between Karl Fischer means and tracted easily and titrated in an 8 h day after instru-
oven method means are given in Table 6. In this table, ment standardization.
the difference is also expressed as a percentage of the Comparison of the Karl Fischer method to oven
Karl Fischer mean. For all feeds except dog food, oven methods illustrates the inconsistency of oven methods.
method I overestimated water content. For all feeds, No one oven method is appropriate for all types of
the bias expressed as % H 2 0 for oven method I ranged samples. Further study is needed to evaluate the appro-
from 0.06 to 1.63, which is 101-135% of the Karl priateness of oven methods for specific types of feed.
Fischer result. These results are consistent with those
of Windham et al. (4,12) who reported biases of 151, Acknowledgments
121, 121, and 119% for silages, legumes, rations, and all
We thank Paul D. Evenson, professor of plant sci-
forages averaged, respectively.
ence and experiment station statistician, College of
Oven method II underestimated water content in
Agriculture and Biological Sciences, SDSU, for assis-
dog food, and legume hay. For all feeds, the bias for
tance with statistical analysis.
oven method II ranges from -0.15 to -0.84%, which is
88-97% of the Karl Fischer result.
References
Oven method III underestimated water content in
corn silage, dog food, and legume hay. For all feeds, (1) Mitchell, J., Jr, & Smith, D.M. (1980) Aquametry, Part
the bias for oven method III ranged from -0.14 to III. The Karl Fischer Reagent, John Wiley & Sons, New
-0.70%, which is 91-99% of the Karl Fischer result. York, NY

Table 6. Method bias—difference between oven methods and proposed Karl Fischer (KR) method

Corn Textured Dog Total Mixed Legume


Method Parameter silage feed food ration hay Mean

Proposed KF % H20 4.70 7.52 6.97 5.09 6.84 6.22

Oven method 1 % H20 6.33 9.17 7.03 6.57 7.81 7.38


Bias from KFa 1.63 1.64 0.06 1.48 0.97 1.16
% of KFb 135 122 101 129 114 119

Oven method II % H20 4.55 7.08 6.17 4.85 6.00 5.73


Bias from KFa -0.15 -0.44 -0.81 -0.23 -0.84 -0.49
% of KFb 97 94 89 95 88 92

Oven method III % H20 4.00 7.45 6.38 4.95 6.21 5.80
Bias from KFa -0.70 -0.08 -0.60 -0.14 -0.62 -0.43
% of KFb 85 99 92 97 91 93
Oven method IV % H20 5.57 8.22 6.61 5.86 6.69 6.59
Bias from KFa 0.87 0.70 -0.37 0.77 -0.15 0.36
% of KFb 119 109 95 115 98 106
a
Difference between oven method mean and Karl Fischer mean.
b
Difference between means divided by Karl Fisher mean times 100, or percent H 2 0 recovered based on Karl Fischer.
32 VAN EREM ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 81, No. 1, 1998

(2) Scholz, E. (1984) Karl Fischer Titration: Determination (8) Official Methods of Analysis (1995) 16th Ed., AOAC
of Water—Chemical Laboratory Practice, Springer, New INTERNATIONAL, Arlington, VA, sec. 935.29
York, NY (9) Official Methods of Analysis (1995) 16th Ed., AOAC
(3) Jones, F.E., & Brickenkamp, C.S. (1981) /. Assoc. Off. INTERNATIONAL, Arlington, VA, sec. 934.01
Anal. Chem. 64, 1277-1283 (10) Official Methods of Analysis (1995) 16th Ed., AOAC
(4) Windham, W.R., Robertson, J.A., & Leffler, R.G. (1987) INTERNATIONAL, Arlington, VA, sec. 967.03
Crop. Sci. 27, 777-783 (11) SAS Institute, Inc. (1989) SAS Procedures Guide ver-
(5) Official Methods of Analysis (1995) 16th Ed., AOAC sion 6, 4th Ed., Vol. 1, Cary, NC
INTERNATIONAL, Arlington, VA, sec. 967.19 (12) Windham, W.R., Barton, F.E., & Robertson, J.A. (1988)
(6) Riedel-de Haen Hydranal Manual, Allied Signal, /. Assoc. Off. Anal. Chem. 71, 256-262
Hydranal Technical Center, Morristown, NJ
(7) Official Methods of Analysis (1995) 16th Ed., AOAC
INTERNATIONAL, Arlington, VA, sec. 930.15

Downloaded from https://academic.oup.com/jaoac/article/81/1/25/5684171 by guest on 02 October 2021

You might also like