WHITE: J. ASSOC. OFF. ANAL. CHEM. (VOL. 62, NO.

3, 1979) 515

Methods for Determining Carbohydrates, Hydroxymethylfurfural,

and Proline in Honey: Collaborative Study
JONATHAN W. WHITE, JR 1
U.S. Department of Agriculture, Federal Research, Science and Education Administration,
Eastern Regional Research Center, Philadelphia, PA 19118

A modification of the official selective absorp- A chemical procedure and an ultraviolet

Downloaded from https://academic.oup.com/jaoac/article-abstract/62/3/515/5702221 by guest on 10 March 2020

tion method for honey carbohydrates, 31.124— (UV) absorption procedure for HMF were de-
3 1 . 1 3 3 , was studied collaboratively; the deter- scribed by Winkler (5); the former has been
minations of sucrose, total monosaccharides, adopted by the Codex Alimentarius (6). Neither
disaccharides, and higher sugars by this proce-
method has been collaboratively studied; the
dure were satisfactory and were adopted by the
simplicity of the UV method would make it
AOAC. High pressure liquid chromatography of
glucose, fructose, and sucrose in honey showed preferable if concordant results could be ob-
better precision than the modified official meth- tained.
od and gave concordant results; it was also Proline is a natural constituent of honey and
adopted. Two methods for hydroxymethylfur- has been useful in distinguishing honey and
fural do not qualify. A method for proline was sirups. The method of Ough (7), in which the
also adopted. predominant free amino acid reacting with acid
ninhydrin solution is proline, is specific. Inter-
The official selective adsorption (SA) method ferences from other amino acids is < 5 % , which
for carbohydrates in honey (1) uses copper re- is insignificant when their relative occurrence in
duction and hypoiodite oxidation for determin- honey is considered. This method has been ap-
ing individual sugars (fructose, glucose, sucrose, plied to honey samples (8): The distribution of
reducing disaccharides as maltose, and higher proline content was surveyed in 740 samples of
sugars) after class separation by adsorption on United States honey.
activated charcoal columns. Although this
method led to a considerable revision of knowl- Collaborative Study
edge of the composition of honey (2), the some- Six 10-lb containers of processed honey from
what cumbersome wet methods have limited its different lots were obtained from a commercial
acceptance as a regulatory procedure. Com- packer. Two containers each represented 3 dif-
parable results have been obtained with simpli- ferent colors.
fied procedures (3) in the laboratory of the After preliminary analysis for sucrose and
Associate Referee. A high pressure liquid chro- proline by the methods under study, aqueous
matographic (HPLC) method has also been re- solutions of these compounds were added to 4
ported for determining glucose, sucrose, and of the samples; 2 samples already contained the
maltose in honey (4). concentrations desired. This produced 3 pairs of
The presence of enough hydroxymethylfur- samples, 2 each low, average, and high in su-
fural (HMF) in honey to respond to the aniline crose and proline content. Each sample was
chloride or resorcinol test (31.138-31.139) has then evaporated in a rotary vacuum evaporator
long been considered indicative of adulteration to the original density.
with commercial invert sirup. However, it has HMF was determined spectrophotometrically
been recognized that honey may generate in each sample; 4 were selected to represent
enough HMF for a positive test by excessive average and high values for this constituent.
exposure to heat from improper processing or Two composite samples were prepared from
storage. A quantitative method is needed for available unprocessed honeys to provide a pair
HMF, since amounts produced by handling or with low HMF content. Each sample was thor-
storage abuse, although providing equivocal or oughly mixed to ensure homogeneity. The low
positive color tests, are generally lower than HMF samples (Samples G and H) were not
those arising from adulteration. heated. Aliquots were shipped in 2 oz wide-
mouth screw-cap polypropylene bottles with in-
1
Present address: 217 Hillside Dr, Navasota, TX 77868. structions to refrigerate the samples which were

5756-0004/79/6203-0515/12$01.00
© Association of Official Analytical Chemists, Inc.
516 WHITE: J. ASSOC. OFF. ANAL. CHEM. (VOL. 62, NO. 3, 1979)

Table 1. Composition of collaborative samples for Procedural differences between the official
determining carbohydrates, hydroxymethylfurfural,
and proline in honey
method for honey carbohydrates and the modi-
fied SA method tested in this study are sum-
Component"
marized in Table 2. Results from the specific
mple Sucrose Proline HMF glucose oxidase method for glucose in honey
A average high — (9), when applied to whole honey solutions, do
B average low — not differ from those by the official method.
C high average average Details of the HMF procedures appear else-

Downloaded from https://academic.oup.com/jaoac/article-abstract/62/3/515/5702221 by guest on 10 March 2020

D low average average
E high high high where (5).
F low low high
G — — low Hydroxymethylfurfural
H — — low
The collaborators were instructed to deter-
° No effort was made to adjust concentrations of the mine the purity of their standard HMF by
remaining components, which ranged as follows:
glucose 29.4-33.31; fructose 35.7-41.1; maltose 7.9-9.6; measuring the absorbance at 284 nm and using
monosaccharides 67.2-72.6; disaccharides 9.7-13.2; an absorptivity of 16,830 L/mol cm (10).
higher sugars 1.4-2.1.
The calibration factor of Winkler (5) was
provided for calculating the HMF content from
the lowest in HMF. See Table 1 for composition
the UV absorbance values as follows:
of collaborative samples.
p-Toluidine, although not on the Occupational HMF(mg/100g honey) = 43.1 X [A285
Safety and Health Administration list of car- - (A245 + A 3 2 5 )/2]/light path in cm
cinogens, carries a warning label stating that it
Later calculations, using data from Winkler's
produces cancer in animals and listing precau-
paper, showed the HMF which had been used
tions in its use. This information was brought
by Winkler had a molar absorptivity of 15,600.
to the attention of the collaborators.
Hence, the factor should have been 43.1 X
As a convenience, standardized charcoal
15,600/16,830 = 39.95. Therefore, all collabora-
(Darco G60) and filter aid (Dicalite 4200)
tive results reported by this method were multi-
(1 + 1), (31.125), enough for 2 adsorption col-
plied by 15,600/16,830 = 0.927 to compare
umns, instructions for an alternative wet pack-
these results with results from the chemical
ing procedure (2), and a practice sample of
method.
known composition were sent to the collabora-
tors. The collaborators were requested to report Proline
one result for each constituent determined by
In the procedure supplied, analysts were in-
each method.
structed to dilute 2.5 g honey to 100 mL with
The HMF and sucrose determinations were water. This results in rather low absorbance
required; the remaining carbohydrate and pro- values, more useful for samples with proline
line analyses were optional but were recom- content of 100 mg/100 g. Dilution to 50 mL
mended because the column fractions in which would have been preferable as a routine.
the sucrose is determined would be available as Only the methods recommended for adoption
a result of the sucrose determination. are described below; some suggestions from the
A high pressure liquid chromatographic collaborators have been incorporated in the
(HPLC) method (4) for glucose, fructose, and methods.
sucrose was also included as an option, to be
used on the same samples by laboratories pos- Chromatographic Separation of Sugars
sessing adequate instrumentation, to provide Alternative Method
comparative analyses.
31.133 Principle
Experimental
Carbohydrates For use when sucrose is sugar of primary interest.
Sugars are sepd by charcoal column, 31.124-31.125.
Storage precautions were unnecessary to pre- Glucose is detd on disaccharide fraction 2 by glucose
vent reduction of sucrose content and change of oxidase before and after invertase hydrolysis and calcd
monosaccharides by action of honey invertase to sucrose. Other sugars are detd by weighing residues
because the samples had been heat-processed. of sepd fractions.
 W H I T E : J. ASSOC. OFF. ANAL. CHEM. (VOL. 62, NO. 3, 1979) 517

Table 2. Differences between the official AOAC method and the proposed method for carbohydrates in honey

Item Official method Modified method

Preparation of fractions adsorpt. on stdzd charcoal filteraid

column, desorpt. by dil. ale. soln
to provide 3 fractions*
Glucose (fraction 1) hypoiodite oxidn under controlled glucose oxidase reagent
conditions
Fructose (fraction 1) reducing value with Shaffer-Somogyi by diff. betw. dry wt of fraction 1 and

Downloaded from https://academic.oup.com/jaoac/article-abstract/62/3/515/5702221 by guest on 10 March 2020

reagentafter hypoiodite destruct. glucose value
of glucose
Sucrose (fraction 2) increase of reducing value with increase in glucose by glucose oxidase
Shaffer-Somogyi reagent caused after yeast invertase hydrol.
by mild acid hydrol., with cor-
rection for reducing disacch.
Maltose (reducing disacch.) reducing value by Shaffer-Somogyi by diff. betw. dry wt of fraction 2 and
(fraction 2) reagent calibr. against maltose sucrose value
Higher sugars (fraction 3) reducing value, as glucose, after 1 dry wt of fraction 3
hr hydrol. in boiling 1/V HCI

° For example, 1% alcohol elutes monosaccharides, 7% alcohol elutes disaccharides, and 50% alcohol elutes
higher sugars.

31.134 Reagents glucose (SRM 41, NBS) in 25 mL H 2 0 in 250 mL vol.

flask. Boil 2 min and dil. to vol. or dil. to vol. and hold
(a) Column.—Prep, as in 31.124. Alternatively, use
final soln 2 hr before use.
slurry prepn: Place glass wool plug at bottom of column
and add ca 1 cm dry filter aid (Dicalite 4200, or equiv.). 31.135 Preparation of Fractions
Wet filter aid layer from below. With outlet open, add
slurry of 20 g adsorbent mixt. in 200 mL H 2 0 from Proceed as in 31.126.
top. Let drain 5 min and apply 4 psi (27.6 kPa) pressure
31.136 Determination of Sucrose
until surface is stabilized. Then apply 10 psi (69 kPa)
pressure, release, and remove excess adsorbent be- Pipet 2 mL fraction 2 into each of four 18 x 150 mm
yond 17 cm depth by suction from above. Add ca 1 cm test tubes. Prep. 2 series, one control, other inverted.
filter aid. Wash column as in 31.124. For each series, arrange in rack tube with 2 mL H 2 0 ,
(b) Acetate buffer soln.—0.1M, pH 4.5. Add 5.72 2 sample tubes, tube with 2 mL glucose std, 2 sample
mL HO Ac to 500 mL H 2 0 , adjust to pH 4.5 with \M tubes, etc., finishing with 2 mL glucose std. To all
NaOH soln, and dil. to 1 L. tubes in control series add 0.50 mL H 2 0 ; to all tubes
(c) Tris buffer soln.—pH 7.6. To 48.44 g in inverted series add 0.50 mL invertase reagent (or
tris(hydroxymethyl)aminomethane (available as 0.50 mL pH 4.5 acetate buffer may be added to std
Trizma base, No. T 1503, Sigma Chemical Co.) in 500 tubes). Hold all tubes 30 min at room temp.
mL H 2 0 , add 384 mL 0.8M HCI, adjust to pH 7.6 if At intervals appropriate to measuring system to be
necessary, and dil. to 1 L. used (i.e., 30 or 60 sec with flow-thru cells; longer
(d) Glucose oxidase-peroxidase reagent (GOP).— with manual cells), add 5.00 mL room temp, glucose
Dissolve 120 mg glucose oxidase (Type II: purified, oxidase reagent to each tube, beginning with inverted
15,000-20,000 units/g; Sigma Chemical Co. G 6125, or series followed by control series. After 60 min, add
equiv.) and 32 mg peroxidase (Type I: from horse- 0.15 mL 47V HCI to first tube and mix thoroly (vortex
radish, salt-free powder; Sigma Chemical Co. P 8125, mixer). Continue adding 4N HCI at same intervals as
or equiv.) in 400 mL tris buffer, (c). Add soln of 270 previously established. One min after first addn, det.
mg o-tolidine.2HCl (available from Fisher Scientific A at 530 nm.
Co. as Fisher certified T-320) in 520 mL H 2 0 . Refrig- Av. A for each pair of sample tubes and use as std
erate in brown bottle. Filter before use, if necessary. A av. of stds read before and after corresponding
Stable 5=6 weeks. sample tubes.
(e) Invertase reagent.—Dissolve 12.5 mg invertase /xg Glucose = (fig glucose in std tube) x (A of
(Grade VI, from baker's yeast, essentially melibiase- sample tube/A of std tube)
free, activity ca 200 units/mg; Sigma Chemical Co. % Sucrose in honey = 0.02375 Ong glucose in in-
I 5875, or equiv.) in 50 mL pH 4.5 acetate buffer soln, verted tube - /Ag glucose in
(b). control tube)/g sample,
(f) Glucose std soln.—0.1 mg/mL. Dissolve 25.0 mg where 0.02375 = pig glucose x 1.9 x 10"6 x (1/2) x 250
518 WHITE: J. ASSOC. OFF. ANAL. CHEM. (VOL. 62, NO. 3, 1979)

x 100; fig glucose x 1.9 = /u.g sucrose; 10~6 = /ng/g; 31.141 Preparation of Sample
Vz = 2 mL analyzed; 250 = mL diln of sample; 100 = Weigh 5.000 g sample in small beaker and transfer
to convert to %. to 50 mL vol. flask with 25 mL H 2 0 . Immediately dil.
to vol. with CH 3 CN and filter thru 0.45 fim filter, using
31.137 Distribution of Sugars sample clarification kit.
Filter fractions if filter aid is visible. Evap. to
dryness, on steam bath with current of air or N, 50.0 31.142 Chromatography
mL fraction 1, 100 mL fraction 2, and entire fraction
Inject 10 (JLL std soln into chromatograph. Establish

Downloaded from https://academic.oup.com/jaoac/article-abstract/62/3/515/5702221 by guest on 10 March 2020

3, finally transferring each fraction to sep. weighed 50
retention times, measure peak hts, and check repro-
mL beakers. Dry to const wt in vac. oven at =£95°.
ducibility. Repeat for sample soln. Calc. glucose,
% Monosaccharides = g fraction 1 x 500/ g sample fructose, and sucrose from integrator values or from
% Disaccharides = g fraction 2 x 250/ g sample peak hts as follows:
% Higher sugars = g fraction 3 x 100/g sample
Wt % sugar = 100 x (PHIPH1) x (V'/V) x (W'/W)
High Pressure Liquid
where PH and PH' = peak hts (or integrator values)
Chromatographic Method of sample and std, resp.; V and V = mL sample and
std (50 and 100) solns, resp,; and W and W = g sample
31.138 Apparatus (5.000) and std, resp.
(a) Chromatograph.—Waters Associates Model Reference
ALC/GPC, or equiv., with Model 6000 solv. delivery
system and Model U6K injector. JAOAC 60, 838-841(1977)
(b) Detector.—Waters Associates R401 refractive
index detector, or equiv. Proline
(c) Recorder.—Varian Aerograph Model A-25 dual
pen recorder, or equiv. 31.116 Principle
(d) Column.—300x4 (id) mm /A-Bondapak/Carbo- Proline, predominant free amino acid of honey, is
hydrate (Waters Associates, No. 84038). reacted with acid ninhydrin soln. Interference from
(e) Magnetic stirrer.—Fisher Versamix stirrer No. other amino acids is negligible, *s5%.
14-511-90 (Fisher Scientific Co.), or equiv.
(f) Sample clarification kit.—Available in kit form 31.117 Reagents and Apparatus
from Waters Associates (No. 26865), or equiv.; 0.45
(a) Ninhydrin soln.—3%. Dissolve 3.0 g ninhydrin
ixm filters stable in org. solvs are suitable.
in 100 mL peroxide-free ethylene glycol monomethyl
(g) Syringes.—10 /il No. 701-N point style No. 1,
ether. Store solv., not reagent, over Zn metal in amber
2x0.020" od, 25 gage needle (Hamilton Co.).
bottle.
31.139 Reagents (b) L-(-)-Proline.—Eastman No. 2488; dry in vac.
oven and store in desiccator. Prep std solns as follows:
(a) Mobile phase.—Nonspectro acetronitrile dild
(7) Stock soln.—0.5 mg/mL H 2 0 . Dil. 25 mg proline to
with H 2 0 (83 + 17). Degas mobile phase daily by mag.
50 mL with H 2 0 . Refrigerate stock soln. (2) Working
stirring 15 min under vac.
soln.—50 /Mg/mL. Dil. 10 mL stock soln to 100 mL
(b) Sugar std soln.—Place 3.804 g fructose, 3.010 g
with H 2 0 . Prep, fresh daily.
glucose, and 0.602 g sucrose into 100 mL vol. flask,
(c) Reaction tubes.—18x 130 mm borosilicate screw-
dissolve in 50 mL H 2 0 , and add CH 3 CN to vol.
cap tubes with Teflon liners.
Composition of std approximates 5 g honey dissolved
i n 5 0 m L a q . CH 3 CN(1 + 1).
31.118 Determination
31.140 Operating Conditions Weigh 2.500 g honey, transfer to 50 mL vol. flask,
Fructose, glucose, and sucrose are baseline sepd and dil. to vol. with H 2 0 . Pipet 0.5 mL into each of
and quantitated in 20 min under following conditions: 3 reaction tubes, add 0.25 mL HCOOH and 1.00 mL
flow rate, 1.0 mL/min (ca 500 psig, 345 kPa); temp., ninhydrin soln. Cap tightly, shake well, and place in
ambient (ca 23°); detector (R401), 8x (fructose and boiling H 2 0 bath 15 min. Cool 5 min in 22° H 2 0 bath,
glucose) and 2x (sucrose); attentuation, 10 mv on remove cap, and pipet 5 mL aq. isopropanol (1 + 1)
recorder, detector set so that 380 fig fructose gives into each. Mix well and det. A at 520 nm against blank
full-scale deflection of pen; and chart speed, 0.17min. of H 2 0 carried thru method. Read all tubes within 35
Mono-, di-, and trisaccharides are eluted from column min of cooling.
in order of MW. Correct for color of honey by detg A of soln contg
Received June 20, 1978. Accepted June 26, 1978. 0.5 mL prepd honey soln, 1.25 mL H 2 0 , and 5.00 mL
 WHITE: J. ASSOC. OFF. ANAL. CHEM. (VOL. 62, NO. 3, 1979) 519

isopropanol (1 + 1). Subtract value from that of reacted monosaccharide fraction and the glucose analy-
sample before calcg. sis. This is obvious from the magnitude of the
Prep, calibration curve as in detn, using proline std ^-values for fructose, significant for 2 of the 3
soln instead of honey. A of 0.5 mL of soln of 50 /ig pairs. Significant systematic error was not pres-
proline/mL is ca 0.35 in 10 mm cell. ent in the glucose determinations. An alterna-
Calc. mg proline/100 g honey.
tive to the use of the official wet method for
Reference glucose and fructose analysis on the column
J. Food Sci. 34, 228-230(1969) fraction may be polarimetric analysis of the
J. Apicul. Res. 17, 89-93(1978) evaporated fraction, which gave values concord-

Downloaded from https://academic.oup.com/jaoac/article-abstract/62/3/515/5702221 by guest on 10 March 2020

Results and Discussion ant with the official method in a limited study
(13).
Results were received from 17 collaborators,
about half of whom conducted the optional Determination of sucrose by this procedure is
analyses. All data are shown in Tables 3-6, with satisfactory; no significant F-value for syste-
a statistical analysis of each group. The single matic error was obtained. The sd for maltose
analyses were treated in pairs, as Youden (11) (actually nonsucrose disaccharides) is roughly
recommends, to estimate precision and accuracy. double that for sucrose, again reflecting its cal-
Before analysis, Youden's ranking test was ap- culation by difference. Distribution of carbohy-
plied to identify collaborators reporting results drates in the monosaccharide, disaccharide, and
for all samples that were greatly different (P = higher sugar categories is useful in identifying
0.05) from the others and therefore showed a falsified materials. The precisions reported in
pronounced systematic error. These are identi- Table 3 for the measurements are adequate for
fied in the tables and the results for all 6 sam- the purpose of the analysis. A significant system-
ples of an identified collaborator for that analy- atic error for the disaccharide measurement is
sis were eliminated from the statistical calcula- shown by 2 significant P-values. It is possible
tions. In addition, Dixon's test for outliers (12) that the systematic error and the precision of
was used once in each set to identify the few the disaccharide measurement could be im-
single values greatly differing (P = 0.05) from proved sufficiently by evaporating 100 mL rather
the remaining data. Because the unit block pro- than the 50 mL portion of the column eluate.
cedure was used, the value paired with the out-
lier was also eliminated. Seven of the 490 val- Carbohydrates by High Pressure Liquid
ues reported for all analyses (except HPLC) Chromatography
were eliminated from the calculations after this Five collaborators had suitable equipment for
test was applied, as well as the 7 values paired this analysis. While this number may be mini-
with them. The presence of significant (P = mal, collaborative results (Table 4) indicate the
0.05) systematic error for each pair of samples method is promising for simplifying honey anal-
for each constituent determined is shown in the ysis. Collaborator 18 did not use the column
table by the F-value, calculated as recom- specified in the procedure, but used Amino-Sil-
mended by Youden (11). X-I®, 13 /Am, instead. Although this could dis-
qualify the results from that collaborator, in
Carbohydrate by Modified Selective Adsorption view of the small number reporting, the data
(SA) Procedure were retained and subjected to Youden's rank-
Although the procedures tested here have ing test and Dixon's test for outliers. These tests
given acceptable precision in the Associate eliminated all the glucose results and 4 of the
Referee's laboratory, it must be concluded that, sucrose results reported by Collaborator 18 from
under the conditions of the collaborative study, the statistical calculations shown in the table.
they do not provide an acceptable level of inter- In comparing the standard deviations paired
laboratory agreement for measuring glucose and for the HPLC and modified SA methods (Table
fructose in honey. The sd for fructose is sev- 5), the calculated values are lower for the
eral times that for glucose. Since fructose is cal- HPLC procedure in 7, 6, and all 9 comparisons
culated by difference between 2 other measure- for glucose, fructose, and sucrose, respectively.
ments, variations in its determination derive Systematic error was not significant for 8 of the
from additive variations—those of weight of 9 pair comparisons. Differences between the
o "So5 •? £ S o "So5 ? 2 s: a,5'?'|'2 2 D 71 & <? &

CO H CO O l O l -t> H O. U H t O O l » H

J> OJ OJ -b co J> co OJ o j o j o j o j e o r o c o
0O M 00 M 0O 0 0 00 co r o J> r o O J O J cn r o
o t o cn o si J > 00
H O l d J i l O W M l O I - ' O W O l I—» OJ J> I—* O O »-» co J> J > 00 cn 00 co
J > r o r o 0 0 s i en t o cn t o O J t o CJI t o i->

si J > J S O J J > O J J ^ O J co o j o j o j o j e o r o c o
t o ro o o o ro H r o r o r o H H H M H N j i - ' M CJl O ) O O O 0O CO OO CO 00 0 0 0 0 CO J> J> [\5 O J> I - 1 H Ol H SI Ol IO Ul M O H H rO H CO W M * S| M
M P P N O O O O M O I S I S S U U I U I O O H r o c u N en 00 00 en o J> 00 00 0 ^ tD ^ H O M O) H Ol O CO to J > o H 00 c n c n t O h - ' O t o c o
N ) O) 0 0 t O U ) O O t O H C O - ^ U I - ^ t O S t o l u i r O S I CO
OJ O I-*
H* O— C71 J> SI * l-» 0 0 t j l S I O l S I O0 O l H OJ cn r o 0 cn co 00 s i co cn i-» to
CO CJl J>

OJ OJ OJ CO OJ CO CO OJ OJ OJ
00 I D S I CO S I CO O IO si c n cji co cn s i 00 o 00 i-1 0 o
CJl O O N O I U U I O O N I D M C n t O Cn M I O N I U O 0 0 O cn 0 rsj r o O J 00 00 j > r o cn 4^ cn s i H-»
cn co cn o cn CJI 00
co u i o i ^ c u r o o o M o w r o M C o en r o cn r o t o co o o J> H H J > OO OJ O l

OJ CO CO CO CO CO OJ CO COCOCOCOCOOJCO
c n t - * o o o c o t o o o c o c o c o c o c o >
j>l_>O>- l-'00 sJCOCOtOCJlCO O l l - ' O O O t - ' U N H O O N H
o o o o o o o o o o o o
M N M O0 H coco w 0 0 0 o n o o r o j > c j i 0 1 — » s i s i co M J > trt r o 0 1 J > 0 1
CO CO CJl CJl tO t-tsi tjioo c j i o c n c o o t o cocjicjicocn c n c T i c o t o r o t j i o

OJ OJ 0J OJ CO OJ ro co co O J r o r o O J r o
J>J>J>J>COJ>J>J>COSICOJ> co o c o 10 co co co o co o o o s i s i c n c o
to OJ o o o J> cn J> 00 cn o cn
h-* OJ OJ CJl J> cn cn J > o t o 1—< 0 0 cn 0 1 00 I-" M Ol W Ol O s j H
cn
~ o o *J»> s i cjicocjicncjicncjii-»sjcJicD SI 4>OID J » M O S J> cn cn ro cn si
cn cn J > co co 1-" j> si cn

l-i CO OJ OJ J> OJ N> OJ OJ O J 00

co r o co co co
CO CO O O O O O O O O O O O O O l - ' O O cn r o o o o co o c o o o c o o o o o J> O OJ H-> CJl O I CO N O O l s i 0 0 O I N O O P H O N H I O W s i o
c o o o i - i c n o i v i c n j i t n c f » o > s j f f l w s s i M J> 0 1 00 0 1 i-"cnsicjir\jcjioj co t o 0 0 C O M co co 0 0 o o ro cn 0 0 t o t n H cncjiJ>
*
r" o' 4> 00 cn
0 1 r" o' 0- 0
. c .n 0- 0 s i 0 1 0 0 H C O H - ' O I - ' O o J> ro OJ OJ ro s i cn cn J> 0 0 c ' t o cn —1 s i
O C n c j I O O J U O U l M t D O M H O l O M U l O CO S I
cn CJI s i cn

Downloaded from https://academic.oup.com/jaoac/article-abstract/62/3/515/5702221 by guest on 10 March 2020

 W H I T E : J. ASSOC. OFF. ANAL. CHEM. (VOL. 62, NO. 3, 1979) 521

Table 3. (Continued)
Sample"
Coll. A B C D E F
Monosaccharides
1 71.84 72.16 68.54 71.09 66.44 69.09
4 70.61 73.32 66.15 70.28 67.20 64.80
5 72.09 72.87 69.52 71.76 69.90 71.89
6 71.62 71.62 68.04 69.80 65.96 67.28
9" 72.90 7515 69.80 70.75 69.80 71.75
11" 70.05 70.35 66.50 65.65 65.30 65.45

Downloaded from https://academic.oup.com/jaoac/article-abstract/62/3/515/5702221 by guest on 10 March 2020

13 72.35 72.89 67.47 70.21 66.59 69.50
Meand 71.70 72.57 68.00 70.62 67.22 68.52
Sd 0.577 1.464 2.669
Sr 0.755 0.629 1.524
Sb 0i 0.935 1.550
P 0.58 5.42 3.07
DF 4 4 4
Disaccharides
1 10.71 11.14 13.85 9.96 14.84 11.06
4 10.20 10.38 13.73 9.73 12.37 9.24
5 11.02 11.21 13.29 10.03 12.98 10.96
6 9.92 10.19 12.26 9.17 13.49 10.19
9 10.90 11.15 12.85 9.75 13.45 9.25
11 9.80 10.85 12.20 9.35 13.05 10.30
13 10.77 11.33 14.17 10.16 14.93 10.88
Mean d 10.47 10.89 13.19 9.73 13.59 10.27
Sd 0.626 0.794 1.104
Sr 0.220 0.349 0.546
Sb 0.414 0.505 0.629
P 8.12* 5.19* 4.10
DF 6 6 6
Higher S u g a r s
1 2.08 1.53 1.86 1.89 2.22 2.32
4 1.47 1.41 1.58 1.78 1.96 2.03
5" 1.76 1.90 1.95 2.56 2.42 2.43
6 1.38 1.05 1.39 1.93 1.83 1.82
9 0.73 1.45 0.57b 1.72c 2.13 1.95
11 1.31 1.28 1.59 1.65 1.89 1.99
13 1.51 1.47 1.73 1.74 2.18 2.17
Meand 1.55 1.37 1.63 1.80 2.04 2.05
Sd 0.355 0.140 0.227
Sr 0.301 0.086 0.075
Sb 0.133 0.079 0.152
P 1.39 2.68 9.13*
DF 5 5 5
Solids and Water
1 100.47 99.91 100.30 100.14 99.58 99.51
4 98.12 100.19 97.50 98.99 97.61 93.11
5 99.95 101.06 100.80 101.55 101.38 102.32
6 98.76 97.94 97.73 98.10 97.36 96.33
9 100.37 102.83 99.26 99.42 101.46 99.99
11» 97.00 97.56 96.33 93.85 96.32 94.78
13 100.47 100.77 99.41 99.31 99.78 99.59
Meand 99.69 100.33 99.17 99.59 99.53 98.48
Sd 1.630 1.716 3.450
Sr 0.958 0.439 1.332
Sb 0.932 1.17 2.250
P 2.89 15.3** 6.71*
DF 5 5 5
a
See Table 1 for description of samples.
6
Excluded as outlier by Dixon's test (12).
e
Value not included in statistical analysis; paired with excluded value.
d
Without excluded values.
e
For presence of systematic errors.
f
Degrees of freedom.
» Excluded from calculations by Youden's ranking test (11).
h
Paired samples for sucrose are D and F (low), A and B (average), and C and E (high).
* Negative value for Sb2.
522 WHITE: J. ASSOC. OFF. ANAL. CHEM. (VOL. 62, NO. 3, 1979)

Table 4. Collaborative results for carbohydrates in honey by HPLC

Sample"

Coll. A B C D E F

Glucose

6 34.95 33.39 31.406 33.07 c 30.23 31.15

9 33.3 33.3 31.0 30.9 29.8 31.5
10 33.62 33.53 30.67 30.78 28.00h 29.92 c
16 31.9 33.3 29.9 30.8 27.4 31.1
18d 37.1 37.1 31.5 36.2 31.3 31.6

Downloaded from https://academic.oup.com/jaoac/article-abstract/62/3/515/5702221 by guest on 10 March 2020

Mean 33.44 33.38 30.52 30.83 29.14 31.25
Sd 0.921 0.724 1.160
Sr 0.855 0.377 1.013
Sb 0.242 0.437 0.399
F 1.16 3.67 5.39
DF» 3 2 2

Fructose

6 40.86 39.90 37.65 6 40.06 c 36.22 37.20

9 38.1 38.3 37.1 37.2 35.7 37.6
10 39.39 39.40 37.13 37.28 33.94 36.76
16 37.7 38.9 36.4 37.8 34.4 37.3
18 38.1 37.7 35.0 40.1 35.4 37.4
Mean 38.83 38.84 36.41 38.10 35.13 37.25
Sd 1.460 0.288 0.821
Sr 0.566 1.664 0.515
Sb 0.952 0' 0.452
F 6.66 0.03 2.54
DF 4 3 4

Sucrose'

6 2.73 2.04 5.01 0.40 4.49 0.48

9 2.7 2.2 5.1 0.6 5.1 0.7
10 2.89 2.34 5.24 0.48 4.51 0.57
16 2.54 2.12 4.85 0.55 4.39 0.57
18 4.4C 3.8C 5.0 0.6 6 4.4 1.9C
Mean 2.72 2.18 5.04 0.51 4.58 0.58
Sd 0.175 0.265 0.123
Sr 0.080 0.194 0.025
Sb 0.012 0.128 0.085
F 4.74 1.87 23.4*
DF 3 4 3

a
See Table 1 for description of samples.
6
Value not included in statistical analysis; paired with excluded value.
e
Excluded as outlier by Dixon's test (12).
d
All 6 values were excluded from statistical analysis by Youden's ranking test (11).
e
Degrees of freedom.
f Negative value for si, 2 .
e
Paired samples for sucrose are D and F (low), A and B (average), and C and E (high).

average values of the 6 samples for glucose, though additional study is needed for final
fructose, and sucrose for the 2 methods (Tables acceptance.
3 and 4) were analyzed by the i-test. t-Values
of 0.81, 1.08, and 1.60 for glucose, fructose, and Hydroxy methyl furfural
sucrose, respectively, did not exceed £0 0 5 (5DF) Collaborative results for the 2 HMF methods
= 2.57, indicating agreement in the results by were disappointing. Significant (P = 0.05) F-
the 2 methods for the 3 sugars. The HPLC pro- values for systematic error resulted for 2 of the
cedure is thus suitable for the 3 analyses, al- 3 comparisons for each method. The 2 methods
 WHITE: J. ASSOC. OFF. ANAL. CHEM. (VOL. 62, NO. 3, 1979) 523

Table 5. Comparison of precision and systematic error for carbohydrate analysis by the SA and HPLC methods

Glucose Fructose Sucrose

Statistic Pair SA HPLC SA HPLC SA HPLC

Sd 1 1.168 0.921 4.154 1.460 0.297 0.175

2 0.786 0.724 1.850 0.288 0.544 0.265
3 1.542 1.160 5.819 0.821 0.107 0.123
Sr 1 1.001 0.855 0.482 0.566 0.180 0.080
2 0.755 0.377 1.375 1.664 0.306 0.194
3 0.962 1.013 1.805 0.515 0.055 0.025

Downloaded from https://academic.oup.com/jaoac/article-abstract/62/3/515/5702221 by guest on 10 March 2020

Sb 1 0.426 0.242 2.918 0.952 0.167 0.012
2 0.153 0.437 0.876 — 0.306 0.128
3 0.852 0.399 3.912 0.452 0.065 0.085

studied do not give comparable results, in con- F-value for the one pair (Samples B and F)
trast to Winkler's conclusion (5). A £-test on resulted from the data reported by Collabora-
the first 18 values in Table 3 in Winkler's tors 2 and 5; although the values were low,
paper comparing the 2 methods yields t = 1.96; they were not rejected as outliers.
£ 005 (17DF) = 2.11. After rejection of data by
the ranking and outlier tests, 21 pairs of values Collaborators' Comments
remain in Table 6 in which the same collabora- Collaborator 1 strongly objected to the use of
tors used both methods on the same samples. p-toluidine. Collaborator 2 questioned the HMF
The same calculation with differences between standardization procedures for both methods,
totals for these 21 pairs of analyses yields t = the need for analyzing a proline standard for
9.81; £ o o l (20DF) = 2.845, which indicates a each day, and the failure to recommend appro-
significant between-methods difference. This dif- priate concentrations for a standard curve; he
ference may be accounted for by the dissimilarity said that the sucrose procedure is a satisfactory
in types of honey used in the 2 studies. This is and sensitive method but believed that the col-
explained further in another publication (14). umn procedure required too much time. Col-
The UV method is superior in both precision laborator 4 preferred the sucrose procedure to
and systematic error (Table 6). Unfortunately, the official titrimetric procedure. Two of the col-
the values from the chemical method are prob- laborators stored the ninhydrin reagent rather
ably more accurate because of the specific nature than the methyl Cellosolve solvent over Zn metal
of the reaction involved and the empirical nature and reported high and erratic values for pro-
of the baseline correction in the UV method. line. The cause of the difficulty was discussed
The validity of Winkler's correction has been with each before this report was written and
questioned by Gautier et al. (15) and Romann both re-analyzed the samples for proline with
and Staub (16). A major problem with the the proper reagent; the second set of results are
chemical method is its use of p-toluidine. For given in Table 7. Collaborator 6 noted that the
these reasons, neither method can be considered UV spectra of the low HMF samples did not
suitable for adoption. A new method which has match the spectrum for the standard; he also
the precision of the UV method and the accu- provided comments to clarify several method
racy of the chemical method has been developed descriptions and noted that inclusion of concen-
by the Associate Referee (14) and was sub- trations for calibration curves would have saved
jected to collaborative study in 1978. time. Collaborator 7 found the timing require-
ments for the glucose oxidase reagent cumber-
Proline some, detracting from the value of the method,
Collaborative results for determining proline and objected to the use of reagents with short
are quite satisfactory (Table 7): the coefficients shelf-life; he requested concentration values for
of variation for precision are 2.32, 2.56, and preparing standard curves for proline and
3.10%, in order of increasing concentration, HMF, objected to the use of p-toluidine, and
with an average value of 2.66%. The significant noted that the 2 HMF methods did not give
524 WHITE: J. ASSOC. OFF. ANAL. CHEM. (VOL. 62, NO. 3, 1979)

Table 6. Collaborative results for hydroxymethylfurfural in honey"

Sample

Coll. C D E F G H
6
UV Method

1 5.6 6.1 9.5 15.9 2.1 2.4

2 6.0 6.7 9.8 16.5 2.3 2.5
3 5 8 6.2 9.5 15.5 2.1 2.5
4 6.20 6.36 9.80 16.72 2.28 2.44

Downloaded from https://academic.oup.com/jaoac/article-abstract/62/3/515/5702221 by guest on 10 March 2020

5 5.8 6.3 10.1 16.9 2.2 2.5
6 6.12 6.74 9.8 16.8 2.76 2.38
7 6.1 6.7 9.8 16.4 2.1 2.4
8 6.24 6.86 9.92 16.8 2.52 2.70
9C 6.60 7.08 10.20 16.83 2.84 2.84
10 6.6 7.1 9.6 16.3 2.2 2.3
11 5.36 5.96 9.28 15.64 2.08 2.76
12° 5.47 5.48 7.83 13.64 2.01 2.20
13c 5.41 6.01 9.53 15.44 1.45 1.16
Meand 5.98 6.50 9.71 16.35 2.26 2.49
Sd 0.500 0.516 0.182
Sr 0.156 0.217 0.189
Sb 0.336 0.331 0e
F« 10.2** 5.66* 0.93
DF/ 9 9 9

Chemical Method
c
l 4.0 3.8 5.9 9.0 0.42 0.0
2 4 6 4.7 7.9 13.8 0.8 0.5
3 4.4 2.3 10.7 11.3 1.0 0.6
4c 3.48 3.48 5.93 12.11 0.57 0.47
5 4.8 4.8 7.7 13.8 1.7 1.5
6 4.25 4.48 8.0 14.9 0.59 0.17
7 4.9 4.8 8.9 14.4 0.9 0.5
9 5.36 5.36 8.9 13.8 0.57 0.57
10 4.9 4.1 6.5 9.7 1.5 1.5
11 4.25 4.25 7.16 11.5 0.95 1.22
12 3.85 4.35 7.1 12.4 0.85 0.5
13 5.16 5.08 8.0 12.5 0.71 0.52
17 4.73 4.70 7.83 14.35 0.76 0.5
Meand 4.66 4.45 8.06 12.95 0.94 0.74
Sd 0.766 1.516 0.572
Sr 0.319 1.254 0.154
Sb 0.491 0.602 0.389
F« 5.74** 1.46 13.64**
DF 10 10 10

° See Table 1 for description of samples.

b
All collaborative values were multiplied by 0.928 (see text).
c
Excluded from statistical analysis by Youden's ranking test (11).
d
Without excluded values.
e
Sb2 slightly negative.
1
Degrees of freedom.
0
For presence of systematic errors.

comparable results. Collaborator 9 commented dure recommended would tend to strip the more
on the need for concentration data for stand- volatile component. He also felt that flow rates
ardizations; he thought that the HMF time for HPLC should be given as a range. He ob-
schedule (due to unstable color) was too de- tained baseline separations at 3 instead of the 1
manding and requested clarification of direc- mL/min recommended with a proportional sav-
tions for proline analysis. Collaborator 10 pro- ings in time and reported glucose and fructose
posed that water and acetonitrile be degassed values both by peak height and integrator; the
separately and then only briefly after being latter showed a mean bias of + 2 . 5 % . (The peak
combined, contending that the degassing proce- height values were used for this report.) Col-
 WHITE: J. ASSOC. OFF. ANAL. CHEM. (VOL. 62, NO. 3, 1979) 525

Table 7. Collaborative results for proline in honey (mg/100 g sample)

Sample"

Coll. A E B F C D

1 99.2 101 39.8 43.0 76.1 80.8

2 96.4 99.6 35.4 37.3 68.1 6 67.6C
3 89 98 40 45 72 76
4 95.4 99.2 40.0 44.2 68.8 76.0
5 50.5C 101.I 6 37.2 38.3 68.0 80.3
6 93.1 100.4 39.7 43.3 72.2 77.5

Downloaded from https://academic.oup.com/jaoac/article-abstract/62/3/515/5702221 by guest on 10 March 2020

7 92.3 94.6 38.5 41.7 69.8 76.7
8d 198.4 190.4 66.4 69.6 125.6 140.0
gd 85.6 91.2 26.4 35.6 62.4 65.2
10 92.56 100.8 37.5 42.5 69.9 76.0
11 86.9 101.5 38.2 44.0 69.8 78.4
15 97.1 104.7 39.2 43.6 75.6 78.5
Mean e 93.55 99.98 38.55 42.29 71.36 77.80
Sd 3.82 2.77 2.79
Sr 2.90 1.03 1.98
Sb 1.76 1.82 1.96
F' 1.73 7.25** 1.98
DF" 8 9 8

" See Table 1 for description of samples.

6
Value not included in statistical analysis; paired with excluded value.
c
Excluded from statistical analysis by Dixon's test (12).
d
Excluded from statistical analysis by Youden's ranking test (11).
6
Without excluded values.
f
For presence of systematic errors.
g
Degrees of freedom.

laborator 11 thought that the sucrose procedure Acknowledgments

had major advantages over the official titri- The Associate Referee acknowledges with ap-
metric procedure and that the proline method preciation the statistical counsel of John G.
had considerable merit, being easy and quick, Phillips, consulting statistician, and statistical
and yielding repeatable results. He recom- assistance by Sandra P. Graham, Northeastern
mended standardized shaking time for the HMF Region, U.S. Department of Agriculture
chemical procedure, and observed that the re- (USll)A), and the cooperation of the following
sults from the 2 HMF methods did not agree. collaborators: Janet L. Booth, Food and Drug
Administration (FDA), Seattle, WA; L. J. Bur-
Recommendations
ton, Agriculture Canada, Calgary, Alberta, Can-
Based on the collaborative results, it is rec- ada; L. W. Doner, A. P. Hoban, 0. N. Rudyj,
ommended— and J. Siciliano, U.S. Department of Agriculture
(1) That the modified selective adsorption (USt)A), Philadelphia, PA; Ronald E. Draper,
procedure for sucrose, monosaccharides, disac- FBA, San Francisco, CA; Vincent Franco,
charides, and higher sugars be adopted as in- F ^ A , New York, NY; Rosemary Bottcher,
terim official first action. Rose Ann Brannen, Walter Funderburk, Sylvia
(2) That the method for proline in honey be Kresel, David Lorenz, and James E. Thean,
adopted as interim official first action. Florida Department of Agriculture, Tallahassee,
(3) That the HPLC method for glucose, fruc- FL; G. Kuhn and L. Zygmunt, Quaker Oats
tose, and sucrose in honey be adopted as interim Co., Barrington, IL; Robert Meloy, Sioux
official first action. Honey Association, Sioux City, IA; Robert
(4) That the method for HMF in honey be Mipro, U.S. Customs Laboratory, New Orleans,
further studied. LA; R. J. Rlina, FDA, Boston, MA; Walter
The recommendation of the Associate Referee was ap-
Schmidt, FD&, Philadelphia, PA; Patricia J.
proved by the General Referee and by Subcommittee D Schneider, FDA, Kansas City, MO; John J.
and was adopted as interim first action by the Association.
The Association subsequently adopted the method as official Stamp, FDA, Los Angeles, CA; Donald W.
first action at the 1978 Annual Meeting. See / . Assoc. Off. Thompson, FDA, Atlanta, GA.
Anal. Chem. (1979) 62, 412.
526 WHITE: J. ASSOC. OFF. ANAL. CHEM. (VOL. 62, NO. 3, 1979)

REFERENCES (8) White, J. W., Jr, & Rudyj, O. N . (1978) J.

Apicul.Res. 17,89-93
(1) Official Methods of Analysis (1975) 12th Ed., (9) White, J. W., Jr (1964) J. Assoc. Off. Anal.
AOAC, Washington, DC, sees 31.124-31.133 Chem. 47, 488-491
(2) White, J. W., Jr, Riethof, M. L., Subers, M. (10) Turner, J. H., Rebers, P. A , Barrick, P. L., &
H. & Kushnir, I. (1962) Composition oj Cotton, R. H. (1954) Anal. Chem. 26, 898-
American Honeys, U.S. Department of Agri- 901
culture, Agricultural Research Service, Wash- (11) Youden, W. J., & Steiner, E. H. (1975) Sta-
ington, D C , Tech. Bull. 1261 tistical Manual of the AOAC, AOAC, Wash-
(3) White, J. W , Jr (1979) Submitted to / . Assoc. ington, D C
(12) Dixon, W. J. (1953) Biometrics 9, 74
(4) Thean, J. E , & Funderburk, W. C , Jr (1977) (13) White, J. W , Jr, & Subers, M. H. (1960) J.
J. Assoc. Off. Anal. Chem. 60, 838-841 Assoc. Off. Anal. Chem. 43, 774-777

Downloaded from https://academic.oup.com/jaoac/article-abstract/62/3/515/5702221 by guest on 10 March 2020

(5) Winkler, 0 . (1955) Z. Lebensm. Unters. (14) White, J. W., Jr (1979) J. Assoc. Off. Anal.
Forsch. 102, 161-167 Chem. 62, 509-514
(6) Codex Alimentarius Commission (1969) Rec- (15) Gautier, J. A., Renault, J., & Julia-Alvarez,
ommended European Regional Standard for M. (1961) Ann. Fals. Fraudes X X , 397-411
Honey, C A C / R S 12-1969, F A O / W H O (16) Romann, E., & Staub, M. (1961) Mitt. Geb.
(7) Ough, C. A. (1969) J. Food Sci. 34, 228-230 Lebensm. Hyg. 52, 44-58

&^Z) CT^g,