(2001) 14, 101}109
doi:10.1006/jfca.2000.0951
Available online at http://www.idealibrary.com on
JOURNAL OF FOOD COMPOSITION AND ANALYSIS
ORIGINAL ARTICLE
Comparison of Methods for Determining Moisture Content
of Citrus and Eucalyptus Brazilian Honeys
by Refractometry
C. B. Cano*, M. L. Felsner-, J. R. Matos-, R. E. Bruns?, H. M. WhatanabeA,
and L. B. Almeida-Muradian#
*Instituto Adolfo Lutz, Sec7 aJ o de Bebidas, Av Dr. Arnaldo 355, 01246-902 SaJ o Paulo, SP, Brazil; -;niversidade
de SaJ o Paulo, Instituto de Qun& mica, Departamento de Qun& mica Fundamental, Av Prof. Lineu Prestes 748 bloco
8, 05508-900 SaJ o Paulo, SP, Brazil; ?Universidade Estadual de Campinas, Instituto de Qun& mica, bloco H, CP:
6154, 13083-970 Campinas, SP, Brazil; AInstituto de BotaL nica, Sec7 aJ o de DicotiledoL neas, Av. Miguel Stefano
3687, 04301-002 SaJ o Paulo, SP, Brazil; and #Universidade de SaJ o Paulo, Faculdade de CieL ncias FarmaceL uticas, Departamento de Alimentos e Nutric7 aJ o Experimental, Av. Prof. Lineu Prestes 580 bloco 14,
05508-900 SaJ o Paulo, SP, Brazil
Received July 19, 1999, and in revised form July 17, 2000
The o$cial method in Brazilian legislation for determining moisture content in honey is the one
established by the Association of O$cial and Analytical Chemists (AOAC). Recently, the
European Honey Commission (EHC) proposed a modi"cation of this method suggesting a
pre-treatment for crystallized honey samples. The objectives of the present work were: to study
by factorial design the in#uence of pre-treatment and physical state of sample on moisture
content determined by refractometry; to choose the better procedure and determine the average
value of moisture content of 50 citrus and eucalyptus honey samples. Samples were collected in
Sa o Paulo State, Brazil during 1996}1998. An AbbeH Refractometer thermostatted at 203C was
used and the pre-treatment was done in a water bath at 503C until crystal dissolution. The e!ects
of the variables of factorial design (pre-treatment and physical state) were signi"cant at 95%
level. Results of this design showed that the use of the pre-treatment, suggested by EHC, is
superior for analysis of liquid and crystallized honey samples. The moisture values obtained for
eucalyptus were higher than those for citrus honey samples independent of the analysis method
used.
2001 Academic Press
Key =ords: honey; analytical methods; moisture content.
INTRODUCTION
Honey is basically composed of carbohydrates like monosaccharides and oligosaccharides (60}85%) which vary in regular forms with moisture content (12}23%), low
proportions of inorganic and organic materials, such as proteins and polysaccharides,
and high #oral pollen contents (White, 1969). Moisture content has an in#uence on
honey color, viscosity, #avor, density and refractive index and is one of the most
important physical}chemical parameters for the analysis of conservation and stability
To whom correspondence and reprint requests should be addressed. Fax: 55-11-815-4410; E-mail:
ligiabi@usp.br
0889}1575/01/010101#09 $35.00/0
2001 Academic Press
102
CANO E¹ A¸.
of foods in general (Felsner et al., 1998; Mateo and Bosch-Reig, 1997; Sancho et al.,
1991). The physical properties of honey (viscosity, refractive index and density) are
di!erent when compared with an inverted sugar solution with the same water content,
because of the varieties of oligosaccharide compositions corresponding to each #oral
source (White, 1969; Crecente and Latorre, 1993).
Among many kinds of methods for analyzing moisture content in honey, the AOAC
(1995) recommends the refractometric method. Also, Brazilian laws use the same
AOAC procedure for analysis of the moisture content of honey (Brasil, Portaria no.
367, 1997). This method is simple, fast and accurate. The refractive index is converted
to moisture content using the Chataway Table with Wedmore's corrections (White,
1969).
The measurement of the refractive index of honey by the AOAC method is made
directly without any sample pre-treatment. In the literature, one "nds the refractive
index measurement used for liquid, translucid and viscous samples as well as for those
having small quantities of insoluble solids. Recently, the European Honey Commission (EHC) proposed a modi"cation of this method, suggesting pre-treatment for
crystallized honey samples in a water bath at 503C until sample dissolution (Harmonised Methods of the European Honey Commission, 1997).
Physical and chemical properties of honey produced in di!erent countries have
been investigated in other studies, because the composition and properties of honey
vary with botanical origin (Andrade et al., 1999; Guyot et al., 1999; Shing and Bath,
1997; Ferreres et al., 1996, 1994, 1993; Gomez et al., 1993; BonhevmH , 1989).
The #oral sources of honey have great in#uence on consumer preference, owing to
#avor and aromatic properties. Mono#oral honeys are more costly to buy than
multi#oral ones (Andrade et al., 1999).
Recently, there has also been a tendency to classify honey regarding its #oral source.
The most common #oral sources in warm regions are eucalyptus, citrus, rosemary,
lavender, sun#ower and heather (Andrade et al., 1999; Costa et al., 1999; Guyot et al.,
1999; Mendes et al., 1998; Anklam, 1998; BonhevmH and Coll, 1995; BonhevmH et al., 1987;
Ferreres et al., 1996, 1994, 1993). In Brazil, especially in Sa o Paulo State, there is
a predominance of mono#oral honeys from eucalyptus and citrus sources (Ramalho et
al., 1991). These kinds of honeys are investigated here by factorial design experiments
to study the in#uence of pre-treatment and physical state of samples on analyzed
moisture contents determined using both the AOAC and EHC methods with the
objective of choosing the more adequate procedure. An application is then made to
determine the average values of moisture content of 50 citrus and eucalyptus honey
samples from Sa o Paulo State, Brazil.
MATERIAL AND METHODS
Sample Collection
Fifty samples of honey in crystallized and liquid states (Table 1) were collected to
represent two di!erent sources, viz., Eucalyptus ssp. and Citrus ssp. from the 1996 to
1998 season. In Sa o Paulo State, forestry is predominated by two species, Eucalyptus
ssp. and Pinus ssp., that contribute to the development of eucalyptus honey. Citrus
groves and #owers abound, therefore citrus honey production is also highly developed.
The honey used in the present study was harvested from hives that were located in
areas where these constitute the major #ora. All the samples were stored at !183C in
plastic jars for further analysis.
103
MOISTURE CONTENT OF BRAZILIAN HONEYS
TABLE 1
Botanical origins and physical states of honey samples
Botanical origins
Eucalyptus
Citrus
Physical state
n
Liquid
Crystallized
Liquid
Crystallized
0
31
6
13
Analytical Methods
The analytical methods used are described below.
Pollen analysis. Pollen spectrum was obtained by a combination proposed by
Iwana and Melhem (1969) of the Louveaux et al. and Erdtman methods. Brie#y,
a sample of 10 g of crude honey was dissolved in 20 mL of distilled water and
centrifuged (2000 rpm;g) for 10 min. To the dry sediment, 5 mL glacial acetic acid
was added. The solution was again centrifuged (2000 rpm;g) for 10 min. To the
obtained dry sediment, 10 mL of acetolysis solution (1 mL sulphuric acid#9 mL
anhydrous acetic acid) was added, and the mixture was put in a warm bath (at 1003C)
for 2 min. This solution was centrifuged and to the resulting dry sediment 10 mL of
glycerine (50%) was added and centrifuged (2000 rpm;g) for 10 min. To the dry
sediment 10 mL alcohol was added. With a 100 lL pipette all of the polinic content
encountered at the bottom of the test tube was pipetted and placed on a microscope
slide. Then a piece of glycerine/gelatine was placed on the slide and sealed with
para$n. Slides were microscopically observed and compared with the reference for
identi"cation. Botanical classi"cation was done when the pollen spectrum contained
over 45% of the corresponding dominant pollen.
Moisture analysis. The moisture content of honey was detected by the refractometric method of the Association of O$cial and Analytical Chemists (AOAC*method
969.38b) and also by the European Honey Commission (EHC) method described in
Harmonised Methods of the European Honey Commission (1997). A unique di!erence of the latter method regards sample pre-treatment, when it is crystallized.
An AbbeH Refractometer thermostatted at 203C was used. Pre-treatment was done in
a water bath at 503C until crystal dissolution.
Factorial design study. The factorial design study (Box et al., 1978; Barros Neto
et al., 1995) used two liquid and crystallized samples. The variables studied were
physical state of the sample (liquid and crystallized) and sample pre-treatment (with
and without heating until crystal dissolution).
Moisture content data were obtained in random order from analyses made using all
four combinations of physical state and sample pre-treatment levels.
The factor e!ect values were calculated by
Ef"RM !RM ,
>
\
where RM and RM are averages of results with high and low levels, respectively, of the
>
\
factors involved.
The factor levels of the design study are reported in Table 2.
104
CANO E¹ A¸.
TABLE 2
Factorial design study for moisture content analysis in honey samples
Factors
Physical state (PS)
Sample pre-treatment (SP)
Low level (!)
High level (#)
Liquid
Without heat treatment
Crystallized
With heat treatment
TABLE 3
Classi"cation of honey samples according to the respective label
No. samples
Honey type
Plant family
Genus
31
19
Eucalyptus
Citrus
Myrtaceae
Rutaceae
Eucalyptus ssp
Citrus ssp
TABLE 4
Signs for calculating e!ects from 2 factorial design for moisture content analysis of honey samples
Physical state
!(liquid)
#(crystallized)
!(liquid)
#(crystallized)
2
Sample pre-treatment
Moisture content averages$S.D.
!(without heat)
!(without heat)
#(with heat)
#(with heat)
2
15.96$0.05
17.44$0.18
15.90$0.13
16.52$0.15
16.45$0.68
Ten replicate determinations each of two honey samples.
Statistical analysis. The e!ects of factors on moisture content were tested for
statistical signi"cance by standard error analysis on each model parameter. Means
per factor level were obtained to facilitate the interpretation of the e!ects. Data are
presented as the mean$standard deviation. An analysis of variance (ANOVA) was
used in order to determine signi"cant di!erences among moisture content of honeys.
Also, a t-test was used to perform a comparison of means. The statistical analysis of
data was carried out using STATISTICA for Windows [Statsoft, Inc. (1995), Tulsa,
U.S.A.].
RESULTS AND DISCUSSION
The microscopic examination of #ora in honey types con"rmed the identity of the
honey sources, listed in Table 3, which had been indicated by the manufacturers.
Based on t-test evidence, the factorial design variables, pre-treatment and physical
state were signi"cant at the 95% con"dence level (Table 4). The most important e!ect
on moisture content determined by the refractometric method was the physical state
of the honey samples (Table 5). Pre-treatment also showed an in#uence on the
moisture content but to a lesser degree than the physical state. Figure 1 suggests that
moisture content with pre-treatment application (with heat) in crystallized honey
samples had lower values ($1.0%) than values obtained without pre-treatment.
However, liquid honey samples had no signi"cant variation ($0.1%) in moisture
105
MOISTURE CONTENT OF BRAZILIAN HONEYS
TABLE 5
Calculated e!ects and their S.E.s for the 2 factorial design in moisture content analysis of honey samples
Estimate$S.E.
E!ect
Average
Main e+ects:
Physical state (PS)
Sample pre-treatment (SP)
1.05$0.09
!0.49$0.09
¹wo-factor interaction:
Physical state;sample pre-treatment (PS;SP)
!0.43$0.09
S.E. calculated from the
S.D.s
16.45$0.04
in Table 4.
FIGURE 1. Diagram of results for interpretation of the 2 factorial design.
content. This fact was related to the higher e$ciency in the transmitted and re#ected
light of the refractive index apparatus when the analyzed samples were transparent
liquids, without dirty and irregular crystals. Furthermore, a signi"cant physical
state;sample pre-treatment interaction e!ect was also found with a value of almost
the same size as the sample pre-treatment main e!ect. This can be seen in Figure 2.
The sample pre-treatment e!ect, given by slopes of the lines shown there, is seen to
depend on whether the sample is liquid or crystallized. In the liquid state the analyzed
moisture content did not depend on pre-treatment whereas in the crystallized state it
decreased by about 1% when heating pre-treatment was used. Therefore, the e!ect of
the physical state or pre-treatment conditions on moisture content results cannot be
completely understood by treating these factors separately, viz., the physical state
e!ects depend on pre-treatment factor levels and vice versa.
As such, both methods, AOAC and EHC, can be used for liquid samples. The
refractive index apparatus has a higher e$ciency for transmitted and re#ected
106
CANO E¹ A¸.
FIGURE 2. Graphs of the interaction e!ect between physical state and pre-treatment procedure:
, physical state, crystallize;
, physical state, liquid.
TABLE 6
Statistical estimates of moisture content data obtained in this work by the two methods
Average$S.D.
Botanical origin
of honey
Method EHC
Method AOAC
Eucalyptus
Citrus
17.36$0.74
16.68$1.30
18.84$1.34
17.27$1.75
Averages and
Averages and
S.D.s
S.D.s
of 31 eucalyptus honey samples.
of 19 citrus honey samples.
light when the analyzed samples are transparent and liquid, without the presence of
irregular crystals. On the other hand, the above observations show that the refractive
index measurement is in#uenced by honey crystallization. For crystallized samples the
EHC method with heating pre-treatment can be recommended. As such, we have
adopted this method for crystallized and liquid honey samples and the AOAC was
used here for comparison purposes.
Table 6 shows the results expressed as mean$S.D. for the moisture content of
honey samples analyzed by the two methods (AOAC and EHC).
Comparison between the average values and standard deviations for moisture
content of eucalyptus and citrus honeys analyzed with and without honey pretreatment, obtained for the two methods, is illustrated in Figure 3 and Table 6. The
average values, S.E.s and S.D.s obtained by pre-treatment application to honeys samples
were lower than those obtained without it. These di!erences may be explained by
crystallization of honey, which interferes with refractive index measurements of
samples.
107
MOISTURE CONTENT OF BRAZILIAN HONEYS
FIGURE 3. Comparison of the S.E.s, S.D.s and averages of samples of citrus and eucalyptus honey samples,
, $Std. Dev.;
, $Std. Err.; , Mean.
with heat (EHC) and without heat (AOAC):
TABLE 7
Analysis of variance (ANOVA) for moisture contents of honeys from di!erent botanical sources
Source of variation
Sum of squares
Degrees of freedom
Mean square
Ratio
Between treatments
Within treatments
Total about the
grand average
S "11.21
2
S "20.47
0
S "31.68
"
1
48
49
S"11.21
2
S"0.426
0
S/S"26.30
2 0
In fact, the di!erences in S.D.s observed here and reported in Table 6 for the citrus
and eucalyptus honey samples could be related to the fact that the citrus samples
included both crystallized and liquid samples whereas the eucalyptus ones involved
only crystallized samples (Table 1).
An analysis of the variance for the moisture content results showed that the average
values are signi"cantly di!erent well above the 95% con"dence level. The results are
shown in Table 7 and show a 26.30 F-test ratio. This corresponds to a 2.36 t-test value
also indicating signi"cance in the di!erence of the average values at the 95%
con"dence level. One might explain the signi"cant di!erences in moisture content as
owing to the di!erent botanical origins of honey. However, other factors such as
varying climatic and geographical conditions and di!erent beekeeping practices are
also expected to a!ect the moisture contents.
The nature of this di!erence for these honeys can be seen in histograms of moisture
content presented in Figure 4. The results for eucalyptus honey appear to be unimodal
approximating a normal distribution whereas those for citrus honey appear to be
bimodal. This is consistent with the fact that all eucalyptus samples were crystallized
whereas the citrus ones were in both liquid and crystallized states.
108
CANO E¹ A¸.
FIGURE 4. Histogram of moisture contents of citrus and eucalyptus honey samples:
,
eucalyptus.
,
, citrus;
CONCLUSION
Results of this factorial design showed that the use of the heating pre-treatment,
suggested by the EHC, is better for crystallized honey samples. For liquid honeys both
methods, AOAC and EHC, can be used. This design also showed evidence of
interaction between the physical state of the sample and its pre-treatment.
The di!erences between the standard deviations obtained by use of the EHC
method (with pre-treatment) and the AOAC method (without pre-treatment) for
crystallized samples may be explained because refractive index is in#uenced by the
physical state of the sample.
These results suggest that the EHC method can be used for moisture content
analysis of honey samples in the two physical states.
The moisture content values obtained for eucalyptus honeys were higher than those
for citrus honey samples. The F ratio observed and t-test values indicated that there
was a signi"cant di!erence between the averages of moisture contents of these kinds of
honey. This di!erence may be in#uenced by botanical origins of honeys and other
factors such as varying climatic and geographical conditions and di!erent beekeeping
practices.
REFERENCES
Association O$cial Analytical Chemists. (1995). O.cial Methods of Analysis of AOAC International, 16th
edn., Vol. II. Method 969.38b.
Andrade, P. B., Amaral, M. T., Isabel, P., Carvalho, J. C. M. F., Scabra, R., and Cunha, A. P. (1999).
Physicochemical attributes and pollen spectrum of Portuguese heather honeys. Food Chem. 66, 503}510.
Anklam, E. (1998). A review of the analytical methods to determine the geographical and botanical origin of
honey. Food Chem. 63(4), 549}562.
Barros Neto, B de; Scarminio I. S., and Bruns, R. R. (1995). Planejamento e Otimizac7 a o de Experimentos.
Editora da Unicamp, Campinas, SP.
MOISTURE CONTENT OF BRAZILIAN HONEYS
109
BonhevmH , J. S. (1989). Caracteristicas "sico-quimicas. Composicion de la miel de eucalipto (Eucalyptus spp.)
producida em Espan a. Anal Bromatol. XLI-1, 41}56.
BonhevmH J. S. and Coll, F. V. (1995). Characterization of citrus honey (Citrus spp) produced in Spain.
J. Agric. Food Chem. 43, 2053}2057.
BonhevmH , J. S., Pajuelo, A. G., and Galindo, J. G. (1987). ComposicioH n, propiedades fmH sico-qumH micas
y espectro polmH nico de algunas mieles mono#orales de espan a. Alimentaria 28 (185) 61}84.
Box, G. E. P., Hunter, W. G., and Hunter, J. S. (1978). Statistics for Experimenters. Wiley Press, New York.
Brasil, Leis e Decretos. (1997). MinisteH rio da Agricultura e Abastecimento Portaria no. 367 de 4 de
Setembro de.
Costa, L. S. M., Albuquerque, M. L. S., Trugo, L. C., Quinteiro, L. M. C., Barth, O. M., Ribeiro, M., and
De Maria, C. A. B. (1999). Determination of non-volatile compounds of di!erent botanical origin
Brazilian honeys. Food Chem. 65, 347}352.
Crecente, R. P. and Latorre, C. H. (1993). Pattern recognition analysis applied to classi"cation of honeys
from two geographic origins. J. Agric. Food Chem. 41, 560}564.
Felsner, M. L., Cano, C. B., Matos, J. R., and Almeida-Muradian, L. B. (1998). Determinac7 a o de umidade
e cinzas em meH is comerciais por termogravimetria e meH todos convencionais- Parte II. In Congresso
Brasileiro de Cie( ncia e tecnologia de Alimentos X<I, Anais, Rio de Janeiro, SBCTA, Vol. 1, pp. 453}456.
Ferreres, F., Andrade, P., and TomaH s-BarberaH n, F. A. (1996). Natural occurrence of abscisic acid in heather
and #oral nectar. J. Agric. Food Chem., 44, 2053}2056.
Ferreres, F., GarcmH a-Viguera, C., TomaH s-Lorente, F., and TomaH s-BarberaH n, F. A. (1993). Hesperetin:
A marker of the #oral origin of citrus honey. J. Sci. Food Agric., 61, 121}123.
Ferreres, F., Giner, J. M., and TomaH s-BarberaH n, F. A. (1994). A comparative study of hesperetin and methyl
anthranilate as markers of the #oral origin of citrus honey. J. Sci. Food Agric., 65, 371}372.
Gomez, M. E. M., Hernandez, E. G., Gomez, J. Y. M., and Marin, J. L. M. (1993). Physicochemical analysis
of Spanish commercial Eucalyptus honeys. J. Apic. Res., 32(3/4), 121}126.
Guyot, C., Scheirman, V., and Collin, S. (1999). Floral origin markers of heather honeys: Calluna vulgaris
and Erica arborea. Food Chem. 64, 3}11.
Harmonised Methods of the European Honey Commission. (1997). Bogdanov, S.; Martin, P.; Lullman, C.
(FAM, Bee Department, Bern 3003, Switzerland). Apidologie (extra issue), 1}59.
Iwana, S. and Melhem, T. S. (1969). The pollen spectrum of the honey of Tetragonisca Angustula La crielle
(Apidae, Mellipolinea) Apidologie 10(3), 275}295.
Mateo, R., and Bosch-Reig, F. (1997). Sugar pro"les of Spanish uni#oral honeys. Food Chem. 60(1), 33}41.
Mendes, E., Proenc7 a, E. B., Ferreira, I. M. P. L. V. O., and Ferreira, M. A. (1998). Quality evaluation of
Portuguese honey. Carbohydrate Polym. 37, 219}223.
Ramalho, M., Guibu, L. S., Giannini, T. C., Kleinert-Giovannini, A., and Imperatriz-Fonseca, V. L. (1991).
Characterization of some southern Brazilian honey bee plants through pollen analysis. J. Apic. Res. 30(2),
81}86.
Sancho, M. T., Muniategui, S. B., Huidobro, J. F., and Simal. (1991). Mieles del pais Vasco. III: agua
y azucares. Anal. Bromatol. XLIII-1, 101}112.
Shing, N. and Bath, P. K. (1997). Quality evaluation of di!erent types of Indian honey. Food Chem. 58(1}2),
129}133.
White, J. W. (1969). Moisture in honey: Review of chemical and physical methods. J. AOAC. 52(4), 729}737.