DRIFFIELD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 88, NO.
1, 2005 121
FOOD COMPOSITION AND ADDITIVES
Single Laboratory Validation of a Method for the Determination
of Hydroxymethylfurfural in Honey by Using Solid-Phase
Extraction Cleanup and Liquid Chromatography
MALCOLM DRIFFIELD, DANNY CHAN, ROY MACARTHUR, SUSAN MACDONALD, and PAUL BRERETON
Central Science Laboratory, Sand Hutton, York, YO41 1LZ, United Kingdom
ROGER WOOD
Food Standards Agency, Aviation House, 125 Kingsway, London, WC2B 6NH, United Kingdom
A method is described for the determination of
hydroxymethylfurfural (HMF) in honey. The
method, which is based on solid-phase extraction
cleanup followed by liquid chromatography (LC)
with UV absorbance detection, was tested on a
variety of different honey types: liquid, set,
blended, filtered, crystalline, and comb honey. A
sample of honey fortified with a known amount of
HMF acted as an in-house reference material. LC
with diode-array detection showed that the HMF
peak did not contain any peaks of coeluting
interfering species. Stability studies showed that
honey samples should not be repeatedly frozen
and thawed because the temperature changes
caused a gradual increase in the HMF
concentration. It was also shown that aqueous
HMF standard solutions should be kept in the dark
at 4°C to avoid degradation of the HMF. The
method was internally validated, and the
measurement uncertainty was estimated to be ±9.0
at 40 mg/kg, the legal limit. A comparison of the
relative standard uncertainty with the Horwitz
relative standard deviation showed that the method
was suitable for its purpose and should be
validated by a collaborative trial.
I
nternational honey standards are specified in a European
Honey Directive and in the Revised Codex Standard for
Honey (1–3). The standards include criteria for quality
factors such as moisture, ash, acidity, hydroxymethylfurfural
(HMF) concentration, apparent reducing sugars, and apparent
sucrose. HMF is a cyclic aldehyde produced by self-catalytic
degradation of sugars, starting mainly from the dehydration of
fructose and glucose in an acidic medium (4). It is a major
quality factor in honey, and its presence was once considered
to be indicative of adulteration with commercial invert syrup.
However, it was found that HMF can occur naturally in honey
Received February 12, 2004. Accepted by SG June 3, 2004.
Corresponding author's e-mail: m.driffield@csl.gov.uk.
Downloaded from https://academic.oup.com/jaoac/article/88/1/121/5657399 by guest on 05 October 2020
through exposure to heat from improper processing or during
extended storage. The amount of HMF produced in honey
during storage depends on the pH of the honey and the storage
temperature, with greater amounts of HMF produced in
honeys from countries with higher ambient temperatures (5).
The recent European Union (EU) Directive 2001/110 and
the Honey (England) Regulations 2003 (SI 2003/2243) state a
maximum HMF level of 40 mg/kg for all honeys except
Baker’s honey, which is suitable only for industrial use or as
an ingredient in other foods that are then processed and must
contain HMF at £80 mg/kg. The Revised Codex Standard also
specifies an HMF limit of 40 mg/kg for honey after processing
and blending. For honeys from countries with tropical
ambient temperatures, and for blends of these honeys, both
standards specify a limit of 80 mg/kg.
Spectroscopic and chromatographic methods can be used
for the determination of HMF in honey, but problems have
been reported. The colorimetric method first described by
Winkler (6) measures HMF defined as constituents of honey
that combine with barbituric acid and p-toluidine under test
conditions, and the color of the resulting solution is measured
against a blank at 550 nm. This procedure was collaboratively
tested in 1979 by AOAC INTERNATIONAL (along with a
UV method) and again in 1988 in the United Kingdom (7, 8).
The latter collaborative trial incorporated calibration with
standard solutions of HMF to calculate the results. Neither
trial produced satisfactory results and, in both cases, further
work to investigate the method or find an alternative was
recommended. In the case of the AOAC trial, comparison of
the colorimetric and UV methods revealed that the 2 methods
did not give comparable results, and neither was adopted as an
official method. The Winkler method has also been criticized
for use of the carcinogen p-toluidine, the instability of the
developed colored solution, and the temperature dependence
of the technique.
To overcome these problems, a spectrophotometric
method (the White method) was developed that claimed to
have retained the accuracy of the chemical method but also to
have incorporated the precision of the UV method (9). The
method is based on the determination of the UV absorbance of
HMF at 284 nm. To avoid interference from other
components, the difference between the absorbance of a clear
122 DRIFFIELD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 88, NO. 1, 2005
aqueous honey solution and the absorbance of the same
solution after the addition of bisulfite is determined. The HMF
content is calculated after subtraction of the background
absorbance at 336 nm (10). This method is the current AOAC
Official Method for the determination of HMF in honey (11).
Interference problems can arise for samples showing a natural
strong absorbance at 284 nm, giving rise to incorrect
determination of HMF. Attempts to remove interferences by
clarifying the samples or subtracting background absorbance
decrease them somewhat, but because these procedures do not
use any separation technique or isolation of the analyte, there
is no assurance that HMF is the only compound being
measured. In addition, no calibration or reference standard is
used. Sample cleanup followed by liquid chromatography
(LC) has also been reported, and there are 3 main cleanup
procedures: filtration of an aqueous honey solution,
clarification, and solid-phase extraction (SPE; 12, 13).
A simple reliable method is needed to support both the EU
Regulation and the Codex Standard. This paper reports a
method for the determination of HMF in a range of honey
types by using SPE cleanup to remove interfering compounds,
followed by LC determination using UV detection.
Experimental
Apparatus
(a) Spectrophotometer.—Hitachi Model U-2001
(London, UK).
(b) LC system.—Consisted of a Gilson 307 pump, a
Gilson 232 Bio sample injector, a Gilson 401 dilutor, a Gilson
System Interface module, and a PC with Gilson Unipoint
software (Anachem, Luton, UK).
(c) Absorbance detector.—Gilson UV/Vis 151
(Anachem) fitted with a 12 mL, 5 mm flow cell set at 280 nm.
(d) LC apparatus.—HP1090 Series II (Agilent,
Bracknell, UK) with a diode-array detector. Used to determine
the peak purity of the HMF. HP ChemStation for LC and
LC/MS systems (Rev A.06.03 [509]) software was used for
data processing. Peak purity values were calculated from
spectra over the range 250–320 nm with a noise threshold of
0.04.
Materials and Reagents
(a) HMF (99%).—Sigma (Dorset, UK).
(b) Methanol.—LC grade (Sigma).
(c) Acetonitrile.—LC grade (Sigma).
(d) Water.—Deionized before use.
(e) Filter papers.—Whatman 113V wet strengthened,
18.5 cm (SLS, Hull, UK).
(f) SPE cartridges.—Bond Elute C18, 3 mL capacity,
500 mg sorbent (Varian, Walton-on-Thames, UK).
Samples of blended liquid, pure liquid, set (creamed),
filtered, crystalline, lime blossom, and comb honey were
purchased from local supermarkets and health food shops or
obtained from the Bee Unit at the Central Science Laboratory
(York, UK).
Sample Preparation
The honey samples were individually homogenized by
using a domestic blender to ensure uniform HMF distribution.
Care was taken to avoid heating the honey: each sample was
stirred with a bowl whisk attachment for 15 min and then with
a K-piece attachment for 15 min on the lowest possible
setting. Each sample was then subsampled and stored at
–20°C.
A 500 g sample of comb honey was fortified with HMF at
approximately 40 mg/kg, for use as an in-house reference
material. The material was then homogenized as described
above and subsampled into 10 g portions. Ten subsamples
Downloaded from https://academic.oup.com/jaoac/article/88/1/121/5657399 by guest on 05 October 2020
were taken at random and run in duplicate, and the
concentration of HMF was determined by using the SPE
method. These aliquots were shown to be homogeneous by
using F-test statistics, and the mean HMF value of 42.7 mg/kg
was used during the method validation. The within-batch
variation of 3.75 was similar to that determined in the method
validation (3.605).
Stability Study
Honey samples were initially stored in 1 pot at –20°C.
These were tested at weekly intervals over 3 months. The
stability of aqueous HMF standard solutions was investigated
spectrophotometrically with a stock solution calibrated at
7.55 mg/mL. This stock solution was stored for 3 months at
room temperature on the laboratory bench, at 4°C in a
refrigerator, and at –20°C in a freezer.
METHODS
The White method and the clarification method were
carried out as described in the literature (11, 12). For the
filtration method, 5 (±0.3) g honey was dissolved in 50 mL
deionized water, the solution was filtered through paper, and
the first 10 mL filtrate was discarded; 0.5 mL filtrate was
transferred to an amber vial suitable for solutions for LC
injection.
SPE Procedure
Dissolve 5 (±0.3) g honey in 25 mL deionized water, and
dilute by taking a 1 mL aliquot and adding it to 9 mL deionized
water. Condition the SPE cartridge with 1 mL methanol,
followed by 3 mL deionized water. Apply 2 mL diluted
sample to the SPE cartridge at approximately 1–2 drops/s.
Wash cartridge with 2 mL deionized water, and dry cartridge
by passing 25 mL air through the cartridge. Elute the HMF
with 4 mL acetonitrile–water (1 + 9) into a 4 mL amber vial,
and pass 25 mL air through the cartridge to ensure that all the
solvent has been eluted. Cap the vial and mix the contents for
10 s, using a Vortex mixer. Dilute 1.5 mL eluate with 0.5 mL
deionized water, and mix on a Vortex mixer for 10 s. Transfer
0.5 mL to an amber vial suitable for solutions for LC injection.
 DRIFFIELD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 88, NO. 1, 2005 123
Figure 1. Z-scores from FAPAS series 28, round 01. Black bars used LC methods. White bars used
spectrophotometric methods.
LC Analysis
LC analysis was carried out by using a stainless steel, 250 ´
4.6 mm id column packed with Synergie HydroRP 4 mm
packing material (Phenomenex, Macclesfield, UK) with a
mobile phase of acetonitrile–deionized water (8 + 92),
degassed before use, at a flow rate of 1 mL/min. A volume of
50 mL was injected, and a retention time of 14 min was
observed for the HMK peak. HMF standard solutions at
suitable concentrations were used in each analysis, and spiked
honey samples (40 mg/kg) were also included to allow
correction of the HMF concentration for recovery.
Validation Experiments
Two batches of a series of fortified honey samples were
analyzed by using the SPE LC method. The HMF was added
to comb honey to provide samples with the following
concentrations: 10, 25, 50, 75, and 100 mg/kg. These were run
in duplicate and used to investigate the linearity of the method
over a calibration range of 2–120 mg/kg. All other validation
data were determined by using 8 batches of the following
honey samples with naturally occurring levels of HMF:
blended, liquid, set, filtered, crystalline, and comb honey,
along with comb honey spiked at 40 mg/kg and the fortified
sample. These samples were run in duplicate by using the SPE
LC method over a calibration range of 2–50 mg/kg. Statistical
treatment of the data included a t-test of the concentration
values determined for the fortified samples, analysis of
variation (ANOVA) testing for within- and between-batch
variation, and calculation of measurement uncertainty and
HORRAT values according to standard methods (14–17).
Downloaded from https://academic.oup.com/jaoac/article/88/1/121/5657399 by guest on 05 October 2020
Results and Discussion
Stability Studies
A short stability study was undertaken to determine
optimum storage conditions; honey samples were initially
stored in 1 pot at –20°C. Upon repeat analysis, the HMF
concentration was found to gradually increase most likely
because of repeated thawing and refreezing of the samples,
with the temperature increases causing HMF to form. To
overcome this effect, all honeys were subsampled into smaller
portions, and only the desired amount was removed from
storage each time.
The stability of aqueous HMF standard solutions was also
investigated. The standards stored at 4° and –20°C showed no
degradation trends over the time period by UV or LC analysis,
but the standard solution stored at room temperature was
found to rapidly degrade after 1 month. It is likely that the
HMF standard stored at room temperature was
photochemically or thermally broken down, but those stored
at 4° and –20°C avoided the degradation process. All the
standard solutions used in the remainder of the work were
stored at 4°C, in the absence of light, and were assigned a shelf
life of 3 months.
Performance of Established Methods–Proficiency
Testing
The results of a Food Analysis Performance Assessment
Scheme® (FAPAS) proficiency testing report on the
determination of HMF in honey were published in July
2001 (18). The 56 participants used a wide variety of methods,
with only 11 laboratories (20%) producing satisfactory
Z-scores. The histogram of the Z-scores for this analysis
124 DRIFFIELD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 88, NO. 1, 2005

Table 1. HMF concentrations as determined by 2 different methods: White spectroscopic method, and SPE cleanup
and LC

HMF concn, mg/kg

Honey type White method SPE Mean difference between White and SPE method, %a

Blended 36.7 28.1 27

Liquid 10.4 7.9 27
Set 32.5 26.5 20
Comb 3.0 1.5 67

a
Calculated by taking the normalized difference between the 2 results and dividing by the mean value.

Downloaded from https://academic.oup.com/jaoac/article/88/1/121/5657399 by guest on 05 October 2020

(Figure 1) showed that the data did not have a normal
distribution, with the uncertainty of the assigned value being
much higher than would be expected. From examination of
the methods reported, it appeared that the results fell broadly
into 2 data sets: those with Z-scores less than zero, obtained by
those who mainly used LC analysis (24 of 31 laboratories),
and those with Z-scores above zero obtained by those who
mostly used spectroscopic methods (20 of 25 laboratories). Of
the 11 laboratories with satisfactory Z-scores, 7 used LC and
4 used spectroscopy. The results support the hypothesis that
spectroscopic methods can provide erroneously high results
due to the presence of compounds that interfere at the UV
wavelength.
HMF, causing complications in determining the HMF
concentration. The chromatogram for the sample cleaned up
by SPE has an absorbance value of approximately 1/10 that of
the filtered sample because of the dilution step in the method;
however, there are very few impurity peaks and none near the
HMF peak.
HMF peak purity.—To ensure that the HMF peak was pure
and there were no coeluting impurities, LC analysis with
diode-array detection (DAD) was used to assess the peak
purity. Spectral software was used to compare normalized and
overlaid UV spectra recorded during the formation of the
peak. A purity factor is calculated based on the similarity in
the shape of the spectra and, if the peak is pure, the purity

LC Methods

Three LC-UV methods were tested with strategies

involving: (1) a simple cleanup using filtration; (2) an
adaptation of the White method; and (3) an SPE cleanup
method. In this work, a previously reported method for
furfural and HMF determination in apple juice and puree was
adapted for honey (13). This adaptation involves SPE cleanup
using a C18 cartridge, followed by LC with a specially
endcapped silica C18 stationary phase. The polar endcapping
allows the C18 phase to be used in highly aqueous mobile
phases without any phase collapse, which would cause
breakdown of the chromatography.
Table 1 compares the HMF concentrations in 4 honey
samples as determined by the White and SPE methods. Each
value is the average of 3 batches, with each sample analyzed
in duplicate within the batch. Table 1 shows that the White
method gives higher concentrations of HMF than does the
SPE method, with a mean difference of 35% between
methods. The differences in values determined were shown to
be significantly different from zero by using a
paired-comparison t-test. It seems likely that the White
method produces an overestimation of the HMF concentration
because of interfering species that also absorb at 284 nm.
Figure 2 shows chromatograms for a blended honey
sample cleaned up by filtration and SPE. The chromatogram
for the filtered sample shows peaks for impurities, most of Figure 2. Liquid chromatograms of blended honey
which are separated from the HMF peak. However, should the (HMF concentration = 42.5 mg/kg). (A) Filtered cleanup;
chromatography fail, these impurities could coelute with the (B) SPE cleanup.
 DRIFFIELD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 88, NO. 1, 2005 125

Table 2. Purity factors and threshold values obtained 100 mg/kg in 2 batches. A plot of concentration versus peak
for diferent honey types during LC analysis with area was produced for each with equations for the lines: y =
diode-array detection
1.109x – 0.89 for Batch 1; y = 1.047x – 0.06 for Batch 2. A plot
HMF concn, Threshold of residuals versus concentration showed no obvious pattern,
Honey type mg/kg Purity factor value
with points randomly distributed about the x-axis. This
showed that the method is linear over the range of
Comb 0.9 794 356
1–100 mg/kg.
Filtered 4.1 878 594 Precision and limit of detection (LOD).—Between- and
Liquid 14.7 965 729 within-batch variation was investigated by ANOVA testing.
Crystalline 2.5 950 646 The standard deviations (between and within) are given in
Blended 42.5 985 838 Table 3.
These values were plotted (Figure 3), and an estimate of
Set 39.7 977 801
the standard deviation (uvc) associated with low-level

Downloaded from https://academic.oup.com/jaoac/article/88/1/121/5657399 by guest on 05 October 2020

Fortified 40.3 971 823
concentrations was determined from the average of the
low-level values (uvc = 0.69). An estimate of the standard
deviation associated with higher-level concentrations (uvp)
was determined from the gradient of the line for the higher
value should be greater than the threshold value, calculated by values (uvp = 0.104). These values were used in the following
taking into account the system baseline noise. Table 2 shows equation to give a measure of the method variation (un), where
the values calculated for the range of honey samples tested. x is the given concentration:
All the peak purity values were greater than the threshold
uv = uvc + ( uvp x ) 2
2
values, proving that the HMF peak obtained for each honey
type was pure and did not include peaks of interferences.
Lime blossom honey.—In addition to the interferences The LOD was also determined by using this plot and was
common to all honeys, lime blossom honey was found to calculated as 1.4 mg/kg (by using a coverage factor of 2).
contain other compounds that caused complications when the Measurement uncertainty.—Although no significant bias
White method was used. These compounds caused an was determined (P < 0.05), there is a degree of uncertainty
increase in the noise of the absorbance spectra, which resulted associated with this estimate. The size of this uncertainty was
in higher uncertainties in the concentration values calculated. estimated by calculating the amount of bias that could be
Using SPE followed by LC overcame these difficulties and present but still be undetectable by the t-test. The following
gave a more accurate HMF determination because all the equations were used:
impurities were either removed during SPE or separated from
the HMF. t crit S
x av = y +
n
Method Validation
x av -100
The results of the single-laboratory method validation are u bias =
2
shown in Table 3.
Chromatographic performance.—HMF standards were
used with each batch, and the calibration curves were shown
to be linear by correlation coefficients of ³0.9995. Stable
retention times were obtained with a retention time drift
[(maximum time – minimum time/maximum time) ´ 100] of
£2.5%. The plate count was good, with values ranging
between 15 040 and 16 370, and the HMF peaks had
asymmetry values of <2 in all cases. Baseline resolution was
achieved between HMF and other constituents in all cases.
Trueness.—Two 40 mg/kg spikes were included in each
batch for the purpose of recovery correction, and the recovery
values were in the range of 74–97%. Statistical treatment
(t-test) of the concentration values for the fortified samples
(assigned value, 42.7 mg/kg) showed that there was no
significant difference (P > 0.05) between the mean of the
Figure 3. A plot of between-batch standard deviation
fortified recoveries (94.5%) and 100%.
versus HMF concentration in honey obtained by using
Linearity.—The linearity of the method was investigated the SPE method (uvc = standard deviation associated
over the concentration range of 1–100 mg/kg by using comb with low-level concentrations; uvp = standard deviation
honey samples spiked with HMF at 10, 25, 50, 75, and associated with higher-level concentrations).
126 DRIFFIELD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 88, NO. 1, 2005

Table 3. Method validation data: recovery-corrected HMF concentrations (mg/kg), recovery values (%), mean corrected HMF concentrations (mg/kg), and standard
where xav is the average recovery for the fortified samples, y is

83
81
84
82
74
82
82
82
100% recovery, n is the number of measurements used, and S
Recovery is the standard deviation of the recovery from the fortified

—
—
—
samples. The value of ubias was determined to be 0.0382
77
76
92
93
87
97
91
94
(3.82%).
Combined standard uncertainty (uc) was estimated at the
given concentration x by using the following equation:
43.9
45.3
36.5
34.7
42.1
26.5
49.2
36.9
Fortified

u c = uvc + ( uvp + u bias )x 2

2 2 2

6.160
5.556
40.3
45.7
41.0
46.8
44.6
31.3
37.0
43.7
40.2
The standard uncertainty associated with the standards was
negligible because the solid purity was reported as >99% by
the supplier.
38.2
41.6
39.6
41.6
38.0
42.6
31.4
33.5

Downloaded from https://academic.oup.com/jaoac/article/88/1/121/5657399 by guest on 05 October 2020

Thus, the measurement uncertainty associated with this
method at the legal limit (40 mg/kg) is ±9.0 mg/kg at the 95%
3.605
3.605
Set

39.7

confidence level.
35.7
44.3
40.4
43.4
40.9
41.1
42.9
40.4

Suitability for collaborative study: comparison of

measurement uncertainty with HORRAT values.—HORRAT
values (Ho) for the results were calculated by using the
following equations (16, 17):
42.3
42.6
46.2
43.9
45.1
47.0
47.6
38.3
Blended

3.802
2.943

H = 21- 0. 5 log x
42.5
42.2
40.8
45.6
42.5
38.8
43.7
40.3
32.9

(100u c / x )
Ho =
H

One fitness-for-purpose criterion for analytical methods is

2.8
2.5
2.9
3.0
2.9
2.6
2.3
0.6
Crystalline

that they should produce Ho values of <2. Figure 4 shows the

0.650
0.325

results for this study; this method was shown to be fit for the
2.5

analysis of honey samples containing natural levels of HMF at

2.7
2.7
2.5
2.6
2.5
3.0
2.9
1.4

concentrations between 3 and 50 mg/kg. The linearity of the

method was shown to be good for a series of fortified samples
at HMF concentrations of 2–100 mg/kg.
13.9
14.4
16.5
15.0
14.4
13.9
14.9
15.8
Liquid

1.005
0.732

Conclusions
14.7
12.9
13.1
15.0
14.5
16.0
14.8
14.8
15.7

This paper describes a method for the determination of

HMF in honey by using SPE cleanup and LC analysis. The
method overcomes the limitations of current methods by
3.7
4.2
4.6
4.1
3.9
4.2
3.9
3.7
Filtered

0.388
0.228
4.1
3.9
4.4
4.3
4.4
3.5
4.6
4.2
3.3
1.5
1.3
0.0
0.8
0.7
0.7
0.7
0.1
Comb

0.670
0.443
0.9
1.7
2.5
1.0
0.8
1.0
0.9
0.0
0.3

Between-batch SD
Within-batch SD
deviations (SD)

Mean
Batch

Figure 4. Plot of HORRAT value (Ho) versus

concentration for the SPE method.
1
2
3
4
5
6
7
8
 DRIFFIELD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 88, NO. 1, 2005 127
removing most interferences and chromatographically
separating the remaining ones from the HMF. This was shown
to be particularly useful for lime blossom honey. The purity of
the HMF peak was investigated by using DAD, and it was
shown that the peak contained no peaks of coeluting
components. Method validation was carried out, and it is
recommended that this method be collaboratively tested to
enable it to be used for official purposes.
References
(1) EU Directive 2001/110 Relating to Honey
(2) The Honey (England) Regulations 2003 (SI 2003/2243) of 25
September 2003
(3) Revised Codex Standard for Honey, CODEX STAN 12-1981,
Rev. 1 (1987), Rev. 2 (2001)
(4) Ghoshdastidar, N. (1992) J. Food Sci. Technol. 29, 399–400
(5) Ramirez Cervantes, M.A., Gonzalez Novelo, S.A., & Sauri
Duch, E. (2000) Apiacta. 35, 162–170
(6) Winkler, O. (1955) Z. Lebensm. Unters. Forsch. 102,
160–167
(7) White, J.W. (1979) J. Assoc. Off. Anal. Chem. 62, 515–526
(8) Lord, D.W., Scotter, M.J., Whittaker, A.D., & Wood, R.
(1988) J. Assoc. Pub. Anal., 26, 51–76
(9) White, J.W., Kushnir, I., & Doner, L.W. (1979) J. Assoc. Off.
Anal. Chem. 62, 921–927
(10) Bogdanov, S., Martin, P., & Lullmann, C. (1997) Apidologie,
5 extra issue, 25–27
(11) Official Methods of Analysis (1995) 16th Ed., AOAC
INTERNATIONAL, Gaithersburg, MD, Method 980.23
(12) Bogdanov, S., Martin, P., & Lullmann, C. (1997) Apidologie,
5 extra issue, 23–24
(13) Blanco Gomis, D., Gutiérrez Alvarez, M.D., SopeZa Naredo,
L., & Mangas Alonos, J.J. (1991) Chromatographia 32,
45–48
(14) ISO 5725-2 (1994) Accuracy (Trueness and Precision) of
Downloaded from https://academic.oup.com/jaoac/article/88/1/121/5657399 by guest on 05 October 2020
Measurement Methods and Results, Part 2, Geneva,
Switzerland
(15) ISO 5725-3 (1994) Accuracy (Trueness and Precision) of
Measurement Methods and Results, Part 3, Geneva,
Switzerland
(16) Horwitz, W. (1982) Anal. Chem. 54, 67A–76A
(17) Horwitz, W. (1995) Pure Appl. Chem. 67, 331–343
(18) FAPAS(2001) FAPAS Series 28 Round 01, Report 2801,
Central Science Laboratory, York, UK