sensory analysis applied to honey: state of the art

HAL Id: hal-00891892https://hal.archives-ouvertes.fr/hal-00891892

Submitted on 1 Jan 2004

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Sensory analysis applied to honey: state of the artMaria Piana, Livia Oddo, Antonio Bentabol, Etienne Bruneau, Stefan

Bogdanov, Christine Declerck

To cite this version:Maria Piana, Livia Oddo, Antonio Bentabol, Etienne Bruneau, Stefan Bogdanov, et al.. Sensoryanalysis applied to honey: state of the art. Apidologie, Springer Verlag, 2004, 35 (Suppl. 1), pp.S26-S37. �10.1051/apido:2004048�. �hal-00891892�

https://hal.archives-ouvertes.fr/hal-00891892

https://hal.archives-ouvertes.fr

26Apidologie 35 (2004) S26–S37© INRA/DIB-AGIB/ EDP Sciences, 2004DOI: 10.1051/apido:2004048

Sensory analysis applied to honey: state of the art1

Maria Lucia PIANAa*, Livia PERSANO ODDOa, Antonio BENTABOLb, Etienne BRUNEAUc, Stefan BOGDANOVd, Christine GUYOT DECLERCKc

a Istituto Sperimentale per la Zoologia Agraria, Sezione di Apicoltura, Roma, Italyb Casa de la miel, Tenerife, Canarias, Spain

c Cari, Louvain-la-Neuve, Belgiumd Federal Dairy Research Station, Swiss Bee Research Centre, Liebefeld, Switzerland

(Received 20 February 2004; revised 11 May 2004; accepted 17 May 2004)

honey / sensory analysis / organoleptic / analytical methods

1. INTRODUCTION

Sensory analysis is the examination of aproduct through the evaluation of the attributesperceptible by the five sense organs (orga-noleptic attributes), such as colour, odour,taste, touch, texture and noise.

Used in many fields, sensory analysisallows to establish the organoleptic profile ofdiverse products (foods, cosmetics, pharma-ceuticals, textiles, household products), andcan be useful in knowing how they are per-ceived by the consumer. Until the 1960s, sen-sory analysis techniques basically relied on thepersonal experience of expert assessors. Thistechnique was simple and low cost, but itlacked the essential requisite of reproducibilityin order to be considered a fully fledged ana-lytical method.

In the second half of the twentieth century,new and improved sensory evaluation meth-ods were developed using panels of assessors,well defined and controlled experimental pro-tocols and statistical techniques for processingthe results (Pangborn, 1964; Stone and Sidel,

1985; ISO 6658, 1985; SSHA and ISHA, 1990;Meilgaard et al., 1991). The advantage ofthese methods over traditional ones is that theresults are reproducible, but the complexityand high costs limit their use to the field ofresearch and development of new products,rather than the routine use in the framework ofmonitoring processes and quality control.

In the case of honey, sensory analysis wasfirst used in France with traditional techniques,by the Gonnet team (Gonnet and Vache, 1979,1985, 1992, 1998). In Italy, Gonnet’s ideas weretaken up with particular enthusiasm; mucheffort was dedicated to training activities andan Italian Register of Experts in the SensoryAnalysis of Honey was founded, which estab-lished a standard traditional methodologyincluding harmonized terminology, evaluationforms, tasting methods, methods for trainingand selecting assessors and sensory descrip-tions of the principal Italian unifloral honeys(Persano Oddo et al., 1995, 2000; IstitutoNazionale di Apicoltura, 1999). Similarly,Gonnet’s legacy was also taken up and devel-oped in other European countries (such as

1 This research was carried out within the national programme of the Ministero delle Politiche Agricole e Fores-tali (EC Regulation 1221/97).* Corresponding author: [email protected]

Sensory analysis of honey S27

Spain: Serra Bonvehí and Gómez Pajuelo,1988).

Only in recent years has there been a moveto develop and apply modern techniques(Piana, 1993a, b; Bogdanov et al., 1998;Bruneau et al., 2000; Bentabol, 2002; NationalHoney Board, 2002a, b; Vejsnaes et al., 2003).

In 1998 a working group was set up by theInternational Honey Commission of Api-mondia (IHC) to study the sensory analysisapplied to honey (IHC, 1998, 1999, 2000,2001a, b, 2002).

This article describes the work carried outin this field under the auspices of the IHC:

1. collection and re-organization of thelarge amount of information previously pro-duced by traditional methods;

2. proposal of a harmonized routine methodbased on internationally recognized ISOstandards for evaluating sensory defects andconformity of unifloral honeys.

2. USE OF SENSORY ANALYSIS IN HONEY EVALUATION

Sensory evaluation enables us to distin-guish the botanical origin of honey and toidentify and quantify certain defects (fermen-tation, impurities, off-odours and flavours). Italso plays an important role in defining prod-uct standards and in the related controls,regarding botanical denominations or other

specific labels. Moreover it is an essential partof consumer preference/aversion studies.

Some of the characteristics that can berevealed by sensory analysis can also be deter-mined by laboratory analysis (for example,fermentation can be identified by testing forfermentation products or yeasts), but for othercharacteristics there are currently no alterna-tive analytical methods. In particular, sensoryevaluation is important in verifying the con-formity of unifloral honeys, since it can revealthe presence of botanical components notpicked up by other analytical systems (physic-ochemical and melissopalynological), but thatnonetheless alter the typical sensory character-istics, sometimes to such an extent that thehoney cannot be marketed as unifloral.

To illustrate this point the results of a testcarried out by the Italian Register of Experts inthe Sensory Analysis of Honey are reported.

Two groups of samples were set up: onecomprising 5 Robinia honeys and the otherone 5 Eucalyptus honeys. Only one sample ineach group corresponded to a perfectly uniflo-ral product, while the other four samples wereobtained by mixing the unifloral sample withdifferent quantities of other unifloral honeys.The samples were tested by a panel of 30assessors who were asked to evaluate thedegree of conformity by giving a score from1 to10 (10 being the maximum conformity and 5the limit for marketing the honey as unifloral).

The results are reported in Figure 1: themedian score awarded by the assessors for

Figure 1. Classification of samples according to the percentage of unifloral honey in samples and sensoryevaluation.

S28 M.L. Piana et al.

each sample is compared to the percentage ofunifloral honey in the blend; the threshold forconsidering honey unifloral according to theorganoleptic score (score = 5) is indicated.

This example demonstrates that smallquantities of a highly aromatic honey (that areusually hardly detected in blends by commonlaboratory analysis) can considerably alter theorganoleptic characteristics of a unifloralhoney (Robinia samples 2 and 3; Eucalyptus 2and 3). On the contrary, much larger quantitiesof a delicately flavoured honey have no or lit-tle effect (Robinia 4 and 5; Eucalyptus 4). Thetest also confirms the importance of sensoryevaluation as a basic criterion in selecting uni-floral honeys for commercial purposes.

3. HARMONIZED TERMINOLOGY AND DESCRIPTION OF EUROPEAN UNIFLORAL HONEYS

In the study and control of unifloral honeys,sensory evaluation is commonly used to completeand interpret the overall analytical results.Therefore in the work of characterization ofthe principal European unifloral honeys under-taken by the IHC (Persano Oddo and Piro,2004) it was considered indispensable to inte-grate the descriptions with the available sen-sory data.

The problem was how to tap into the largeamount of knowledge accumulated by variousspecialists in the subject, but hardly accessiblebecause produced with traditional methodsand without any systematic standardizedmethodology. While aware of the limitationsof a less than rigorous procedure from thepoint of view of modern sensory methodology,we considered it important to preserve thisinformation collected from thousands of samples.

The first step involved compiling a uniformglossary, drawing on the experience of theauthors who had worked on sensory descrip-tion of honey (IHC, 2000). The harmonizedglossary (reported in the online version of thispaper, see appendix I) refers to all theattributes and terms used in sensory descrip-tions of European unifloral honeys (PersanoOddo and Piro, 2004), except odour and aromadescriptors. For these latter, the Belgian teamfrom CARI gave a valuable contribution

developing a standardized terminology. Theycreated an “odour and aroma wheel for honey”(Bruneau et al., 2000) similar to the modelspreviously developed for wine (Peynaud, 1980;Guinard and Noble, 1986), beer (Meilgaardet al., 1979a, b) and hard and semi-hardcheeses (Issanchou et al., 1995; Bérodier et al.,1997; Lavanchy and Bütikofer, 1999).

A key aspect of this approach is the defini-tion of standardized terminology, based onactual references, so that the product can beconsistently and unequivocally described. Itshould contain a sufficiently wide range ofterms to describe all the possible variations inthe product. While the terms are conventionaland should not necessarily evoke a sensationin an untrained taster, they should do so une-quivocally and with precision when two peo-ple with the same training are talking to eachother. The terms are usually arranged on awheel divided into sectors (families) and sub-sectors (sub-families) correspondent to one ormore actual references. Experience has shownthat this device can also have a positive impacton communication and product perception, inthat the terms defined for technical purposesare gradually adopted by the consumers.

The first experimental honey wheel devel-oped by the Belgian group was tried out by asmall working group of IHC (IHC, 2001a) andconsequently modified by adding someattributes for the differentiation of certainMediterranean honeys. The resulting harmo-nized “wheel” is presented in Figure 2.

The next step was a review of the descrip-tive sensory data on unifloral honeys availablein the literature (Persano Oddo et al., 1995,2000; Gonnet and Vache, 1998; Bruneau et al.,2000; Sáinz Laín and Gómez Ferreras, 2000)and their integration, together with the inputfrom the members of the IHC working group,into unified descriptions, based on the harmo-nized terminology.

Finally, to complete and confirm thesedescriptions, about 40 Italian experts took partin an experiment in which they were asked touse the harmonized terminology to describethe different unifloral honey samples.

The results of all this work are the descrip-tions contained in the characterization sheetsof European unifloral honeys (Persano Oddoand Piro, 2004) and summarised in Table I.


4. HARMONIZED ROUTINE METHOD FOR HONEY QUALITY EVALUATION

In honey quality control a method for clas-sifying or grading honey according to defectsor conformity with unifloral profiles is needed.The method here proposed is the result ofyears of experience acquired by the laborato-ries that work within the IHC. It sets down theapplication to honey of a sensory analysismethodology for which general guidelinesalready exist regarding the design of test rooms(ISO 8589, 1988), general sensory analysisguidance (ISO 6658, 1985) and training of

assessors (ISO 8586-1, 1993; ISO 8586-2,1994). It is mainly based on the experienceattained in the olive oil industry (InternationalOlive Oil Council, 1996).

The aim of this method is to evaluate theolfactory-gustatory characteristics of honey. Ithas been designed in particular to verify theabsence of defects and the conformity of thesensory profiles of unifloral honeys, referringto previously memorized standards.

Visual and tactile characteristics, whileimportant in the overall evaluation of the prod-uct, are not included because of the need toseparate the visual and tactile stage from the

Figure 2. Odour and aroma wheel (IHC, 2001a).


olfactory-gustatory stage, as there is much evi-dence for the direct influence of the former onthe results of the latter. Moreover, in the eval-uation of unifloral honeys, some visual and

tactile characteristics (physical state, consist-ency, crystals) are not strictly dependent onthe origin of the product, since they are alsoinfluenced by the processing methods. Finally,

Table I. Description of main European unifloral honeys (Persano Oddo and Piro, 2004).

Bra

ssic

a

Cal

luna

Cas

tane

a

Citr

us

Euc

alyp

tus

Hel

iant

hus

Lav

andu

la

Rho

dode

ndro

n

Rob

inia

Ros

mar

inus

Tar

axac

um

Thy

mus

Tili

a

Hon

eyde

w

Met

calf

a h.

dew

Vis

ual

Colour intensity (from 1 to 5) 2 4–5 4–5 1 3–4 3 2 1 1 1 3 3–4 2–3 4–5 5

Tone

Normal honey colour + + + + + + + + + + + + +

Other tones + + + + + + + +

Olf

acto

ry

Intensity of odour (from 0 to 3) 2 2–3 3 2 2–3 1 2 1 1 1-2 3 2 3 2 2

Odo

ur d

escr

iptio

n (s

ee “

Hon

ey a

rom

a w

heel

”, F

igur

e 2)

Woody + + + + + + + +

Chemical + + +

Fresh + + +

Floral – fresh fruit + + + + + + + + +

Warm + + + + + + +

Spoiled + + +

Vegetal + + + + +

Tast

ing

Sweetness (from 1 to 3) 2 1 1 2 2 2 2 2 2-3 2 2 2 2 2 2

Acidity (from 0 to 3) 1 2 1 1 1 3 2 1 1 1 2 2 1 1 1

Bitterness (from 0 to 3) 0 2 3 0 0 0 0 0 0 0 0–1 0 0–2 0 0

Intensity of aroma (from 0 to 3) 2 3 3 2 2 1–2 2 1 1 1–2 3 2–3 3 2 2

Aro

ma

desc

riptio

n (s

ee “

Hon

ey a

rom

a w

heel

”, F

igur

e 2)

Woody + + + + + + + + +

Chemical + + +

Fresh + + + + + +

Floral – fresh fruit + + + + + + + + + +

Warm + + + + + + + + + +

Spoiled + + + +

Vegetal + + + +

Persistence (from 0 to 3) 1 3 3 1–2 2 1 2 1 1 1 3 2 3 2 2

After-taste +/– +/–

Oth

er

Astringency + + +/– +/–

Refreshing +/– +/– +/–

Crystallisation rate (from 1 to 3) 3 2 1 2 2 3 2 2 1 2 3 2 2 1 1

Other physical characteristics + + + + + + +


some attributes, such as colour, can be esti-mated using instrumental analysis.

4.1. Principle

The method is based on the evaluation ofthe olfactory-gustatory characteristics ofhoney by assessors trained to identify sensorystimuli on the basis of previously memorizedstandards (ISO 8586-1, 1993; ISO 8586-2,1994) and to quantify them on an unstructuredscale of 10 cm (ISO 4121, 1987). The evalua-tion is carried out according to the conditionsand general methodology set down in ISO6658 (1985).

4.2. Test room

The room where the sensory analyses arecarried out must conform to ISO 8589 (1988),at least as regards the minimum requirements(lighting, temperature, noise, odours). Individ-ual booths should be set up, so that the asses-sors can work on their own without distraction(modular mobile units can be used).

4.3. Setting up the panel: selection and training

A panel consists of a minimum of 7 asses-sors. ISO 8586 standard should be followedfor selection and training of assessors. Theassessors must have memorized the character-istics they have to evaluate in their long-termmemory. There should also be a panel leaderresponsible for coordinating the trials and col-lecting and processing the results. The panelleader also oversees the selection, training andmonitoring of the performance of the membersof the panel.

4.4. Preparation of sample

4.4.1. Sampling beakers

The beakers or containers used to presentthe samples to the assessors should meet thefollowing requirements:

– enable the samples to be presented in ahomogeneous and anonymous way (identicalcontainers with no distinguishing marks apartfrom the identification code);

– be odourless;

– mask the colour of the honey (coloured oropaque non-white containers; alternativelycoloured light can be used (e.g. red) with whitecontainers);

– help the liberation and concentration ofthe honey odours retarding their dispersion(for instance by having a covering lid).

Oil sampling beakers defined in the COI/T.20/Doc. No. 5 norm (International Olive OilCouncil, 1987) meet these requirements. Insome tests 30 ml red plastic beakers with tightcovers (N. 4196, www.semadeni.com) wereused.

4.4.2. Preparation of sample

Samples can be prepared in two differentways.

4.4.2.1. Crude sample

This preparation should be used for evalu-ating the olfactory-gustatory characteristics inthe way in which they are perceived by theconsumer. The samples are prepared as follows:each sample is given a random three-figurecode; about 30–40 g of the sample are put intothe sampling beaker (one for each assessor),which is then covered with a suitable lid (e.g.petri dish, aluminium foil, cling film). Thehoney should be transferred in such a way asto minimize any alteration due to manipulationand to ensure that the sub-samples look thesame. To guarantee anonymity, the technicianpreparing the samples for testing should notperform the sensory analysis. The sample tem-perature must be between 18 and 25 °C. Theevaluation should take place within 24 h aftersample preparation. If honey is covered with atight lid, samples can be evaluated within alonger period of time. The 30–40 g quantity isrecommended for a testing beaker with a totalcapacity of about 130 ml. Containers of differ-ent dimensions can be used, maintaining thesample/volume ratio near 1/4–1/5, with a min-imum of 10 g of honey per assessor.

4.4.2.2. Preparation of dilute sample

Samples should be diluted when it is felt thatinterference from secondary characteristics(physical state and type of crystallization,water content) may negatively affect the


reproducibility of the method. Dilution is par-ticularly advisable in evaluating olfactory char-acteristics and in evaluating conformity with abotanical profile. The samples are prepared asfollows: a portion of the sample is diluted withodourless water (distilled or low in mineralcontent) in proportions of 1 part water to 5 partshoney (in weight), so that the final water con-tent of the mixture is about 30%. The mixtureis homogenized; if some crystallized parts arestill present it is possible to heat the solution ina closed container in a water bath at 40 °C untilsugar crystals are completely dissolved. Thesample is then divided up among the test beak-ers and the same procedure followed asdescribed in Section 4.4.2.1.

4.5. Evaluation

4.5.1. General conditions

Assessors must abstain from smoking, eatingand drinking anything except water for 30 minbefore the evaluation. They should also refrainfrom using toothpaste or highly aromaticmouthwashes, perfumed toiletries and any-thing else that may give rise to odours of anykind being introduced into the tasting room.

Each assessor has the material necessary forsensory analysis and evaluation forms (seeappendix II in the online version of this paper).The assessors work individually filling in theform that will be handed in by the panel leader.

The number of samples per session shouldbe limited to a maximum of 7. Between to ses-sions at least 30 min have to pass and assessorsmust be rested. The sessions should be timedto take place at least 2 h after main meals. Thebest sensitivity is obtained in the middle hoursof the morning and afternoon. The order inwhich the samples are presented to the asses-sors should vary.

4.5.2. Evaluation of the olfactory characteristics

The olfactory characteristics are evaluatedat first. In the case of crude samples of honey,the odour is evaluated immediately after thehoney has been spread on the surface of thebeaker with the plastic spoon, to encourage theliberation of volatile substances and to pro-duce an evaporation surface area that is the

same for all samples. In the case of dilute sam-ples, it is sufficient to swirl the sample roundin the beaker to encourage evaporation. Thepanellist has to breath in for some secondsover the top of the beaker. The odour has to beevaluated both immediately after havingspread or swirled the honey and after 10 or20 s. Before taking a second sniff, the panellisthas to wait for 5–20 s, or even longer, so thatthe full force of the odour is perceived. Theintensity of any defect perceived and, ifrequired, conformity with the unifloral profileare immediately noted on the form.

4.5.3. Evaluation of the olfactory-gustatory characteristics (tasting)

To evaluate mouth sensations a small quan-tity (1 or 2 g) of honey is sampled with a dis-posable (or stainless steel) spoon. The honey isallowed to dissolve in the mouth before beingslowly swallowed, so that the taste (sweet,salty, acid, bitter), the aroma (intensity andquality), the persistence, any after-taste andother mouth sensations can be perceived.

When the honey is presented as is, the pan-ellist should concentrate on the chemicalaspects of the mouth sensations and should notbe distracted by tactile characteristics. At least1 or 2 min should be allowed to lapse before asecond tasting, much longer in the case of per-sistent honeys, to give the taste buds time torecover before tasting again, in order to detectdetails that may not have been detected the firsttime. The evaluation takes into considerationmany descriptors that are summarized by theassessor into a single evaluation regarding theintensity of each defect identified and, whenrequired, conformity with the unifloral profile.Between samples, the panellist has to wait fora few minutes and eat a piece of apple (prefer-ably juicy and slightly acidulous but not bitteror astringent) or low-in-salt bread. Alterna-tively the panellist has to rinse his mouth withwater or rose-hip tea.

4.5.4. Evaluation of defects and unifloral profiles

4.5.4.1. Defects

Defects evaluation is based on the identifi-cation of the correspondent reference stand-ard, previously memorized by assessors, and


on the quantification of the intensity, com-pared with the references, on an unstructured10-centimetre scale. Reference samples areprepared specifically for this purpose. Off-odourand off-flavour of fermentation and foreigntastes and odours (such as smoke and thymoltaints) are considered as defects.

4.5.4.2. Unifloral profiles

Evaluation of unifloral conformity is car-ried out by a panel of specialized expert asses-sors who have been trained to recognize thedifferent unifloral types, having memorizedtheir typical characteristics as well as everypossible variation of the product. During thisevaluation the assessor must interpret a com-plex profile comprising various componentsthat may vary independently of each other. Sothe task is not to evaluate the intensity of amemorized sensation, but rather to give anoverall assessment that takes into considera-tion all the components perceived, and toquantify them on the unstructured 10-centimetrescale in comparison with memorized referencesamples. The criterion for selecting referencesamples should be their perfect conformitywith the physicochemical and melissopalyno-logical characteristics (Persano Oddo andPiro, 2004; this paper also reports the sensorycharacteristics of the principal unifloral honeys).

4.5.5. Evaluation form

The assessors note the score for eachdescriptor on an evaluation form (see appen-dix II in the online version of this paper).

4.6. Processing and interpreting the results

When the assessors have completed theirevaluations, the panel leader collects theresults (measuring in centimetres the distancebetween the left extreme of the scale and thesign put by the assessor) and proceeds to cal-culate the following statistical indices: median(M), and robust standard deviation (S):

where:N is the the number of responses.

IQR is the the interquartile range, that is thespread between the 75th and 25th percentile(= 75th percentile – 25th percentile).

The median values are used to classify thesamples according to the presence of sensorydefects and conformity in the case of uniflo-rals. Robust standard deviation is an index ofthe reliability of the test. Only results forwhich the value of this parameter is equalto, or lower than, 1 should be taken intoconsideration. Appendix III in the onlineversion reports an example of the use ofthis type of results (Bentabol, unpublisheddata).

4.7. The “yes/no” classification method

A very simple yes/no classification system(ISO 6658, 1985) is used in some laboratoriesof the IHC group and may represent an alter-native to the routine method. In this case thedistribution of the results can be evaluated bymeans of a simple χ2 test (at a level of confi-dence of 95%): results are valid if there is, inone category (yes or no), a given number ofreplies, depending on the number of assessors(Appendix IV in the online version).

An intrinsic drawback of the yes/no methodis the difficulty of defining with precision thelimit between the two categories (accepted/notaccepted) by means of reference samples. Infact this would imply a previous decision ofwhich level of defects (or uniflorality) isacceptable or not: for evaluation of defects,reference samples with a low level of a defectcould be perceived or not by the assessors; onthe other hand, for evaluation of uniflorality,border line reference samples may be verydifferent to each other. Therefore, whendefinitely good or bad samples are tested,results are generally valid, but when interme-diate samples are assessed it is very difficult toobtain valid results, because assessors per-ceive or interpret them in different ways.

With the proposed harmonized routinequantitative method, the use of a grading sys-tem allows training of the panel with very typ-ical reference samples (near to the rightextreme of the scale) and shifts the problem ofthe acceptability of the samples to the numericresults (see Sect. 4.8).

S 1.25 · IQR1.35 · N-------------------------=


4.8. Results of some preliminary tests with the harmonized routine method

Some preliminary tests were carried out toverify the suitability of the proposed routinemethod (quantitative method) and to compareit with the yes/no method.

In a first trial 10 samples were tested for 3characteristics: uniflorality, fermentation andsmoke taint. Three panels of 8 to 16 assessorseach were asked to evaluate the samplesaccording to the harmonized routine methoddescribed above and to classify them accord-ing to the yes/no classification system (corre-spondence/not correspondence to the unifloralprofile and presence/absence of the 2 defects).

Results were elaborated separately for the 3panels (in total 90 answers: 10 honeys × 3characteristics × 3 panels) and all together (asa whole big panel: in total 30 answers: 10 hon-eys × 3 characteristics × 1 panel).

The validity of results was evaluated throughthe robust standard deviation (not higher than 1)for the quantitative routine method, and throughthe χ2 test (see appendix IV in the online ver-sion) for the yes/no method.

When the answers of the 3 panels wereelaborated separately, the valid results were 75out of 90 (83%) with the quantitative routinemethod, and 62 out of 90 (69%) with the yes/no method.

When answers were elaborated altogether,all results were valid with the quantitative rou-tine method, while with the yes/no method 7out of 30 (23%) remain not valid.

A second series of trials was performed inorder to obtain more information on the repeata-bility of the harmonized routine method. Twoseries of unifloral honeys of different puritywere prepared: series A = 4 chestnut honeysand series B = 4 Robinia honeys. The twoseries were presented (anonymously and ran-domly) to two groups of 12 assessors: series A- chestnut, to panel A and series B - Robinia topanel B. Assessors were asked to evaluate theuniflorality of the samples. In the afternoon ofthe same day the same samples were presenteda second time for the evaluation to the sameassessors, always anonymously and randomly.

The results of the two consecutive evaluationswere processed both separately (12 responses)

and together (24 responses). Median and con-fidence intervals at 95% were calculated andresults were considered valid if the robuststandard deviation was not higher than 1.

Results are presented in Figure 3: the fullelaboration (24 answers) is identified with thecodes 1,0, 2,0, 3,0 etc.; the two repetitions(12 answers each) are identified with the codes1,1, 2,1, 3,1, etc., and 1,2, 2,2, 3,2 etc. Invalidresults are highlighted in grey (Castanea 2,2and Robinia 4,1).

A greater degree of dispersion among theresults were recorded for samples with inter-mediate characteristics, whose correspondencewas interpreted and quantified less uniformlyby the assessors (perhaps because they had notreceived enough training in the use of theunstructured scale). For one sample (Robiniasample 4) the 2 repetitions gave not compara-ble results, but one of the 2 repetitions (4.1)also resulted not valid.

Samples were later presented for a thirdtime to the same assessors, who were asked toperform a simple yes-no classification. In thiscase results related to Robinia sample 2 and 4were not valid (Fig. 4).

These results confirm the suitability of theproposed routine method for the evaluation ofhoney defects and uniflorality, the main problembeing the organization of a numerous andskilled panel (in our trials we considered insuf-ficient training as the main cause of someunsatisfactory results).

The yes/no system can have a useful appli-cation as a screening method, but for problem-atic samples (with intermediate characteristics)the quantitative routine method is recom-mended even if more training and, sometimes,more replies are needed to get valid results.

5. CONCLUSIONS

The traditional sensory evaluation of honey,widely used throughout the honey-producingworld, has proved an important instrument forquality control and improvement of honeyquality.

Nevertheless, the use of sensory analysis inthe fields of science and quality control neces-sarily demands a shift to modern, more relia-ble and reproducible techniques, even if theirapplication to honey requires what might seem


to be a disproportionate effort in comparisonwith the economic value of the product.

The procedures described above go someway towards meeting the urgent need for astandardized method, albeit with a lower degreeof reproducibility, taking into account theobjective fact that at the present time, most ofthe assessors are trained as “experts” and are

all involved to a some extent in apiculture andhoney production.

Some variability factors in the evaluation ofunifloral honeys could be avoided by using a“profile method” (as defined in ISO 6564,1985). While the routine method only requiresan overall assessment that takes into consider-ation all the components perceived (and

Figure 3. Results of the repeatability trials carried out with the harmonized routine method. In A the resultsof the 2 evaluations carried out by the panel A on the 4 Castanea samples (expressed as median and confi-dence intervals at 95%) are reported and compared with the same results processed all together. In B theresults of the panel B on the 4 Robinia samples are reported. 1,0 = sample #1 both evaluations elaboratedtogether; 1,1 = sample #1 first evaluation; 1,2 = sample #1 second evaluation; 2,0 = sample #2 both evalu-ations, etc. For Castanea sample #2 second evaluation and for Robinia sample #4 first evaluation the resultshave a robust standard deviation >1 and are not valid. The two repetitions of Robinia sample #4 gave differentresults.

Figure 4. Results of the yes/no evaluations. In the histograms the number of the answers in each categoryis indicated. Results of Robinia samples #2 and #4 are not valid (with 12 assessors at least 10 replies in thesame category are needed, see Appendix IV in the online version).


quantifies them on the unstructured 10-centi-metre scale in comparison with memorizedreference samples), the profile method wouldimply selecting, for each unifloral honey type,the most significant attributes, with the respec-tive reference standards, and to use them todraw a complete reference profile.

Work on the development of these moderntechniques to describe the sensory profile ofhoney and evaluate conformity with specificsensory requirements has barely started butseems promising.

It is important that in fields where moderntechniques cannot be applied because they areeconomically unviable, traditional methodscontinue to be used so that the experience andresults gained in around 25 years of activityare not lost.

It is also important that works in the twobranches of sensory analysis of honey con-tinue to progress in tandem, thus enabling thetraditional branch to progress and improve andthe modern one to be receptive to the changesand needs of honey producers.

REFERENCES

Bentabol A. (2002) Aportaciones metodológicas alanálisis sensorial descriptivo de las mieles.Descripción de olores y aromas, Alimentaria 339,49–52.

Bérodier F., Lavanchy P., Zannoni M., Casals J., HerreroL., Adamo C. (1997) Guida per la valutazioneolfattogustativa dei formaggi a pasta dura e semi-dura, G.E.CO.TE.F.T., Poligny, France.

Bogdanov S., Kilchenmann V., Fluri P., Bühler U.,Lavanchy P. (1998) Influence of organic acids andcomponents of essential oils on honey taste, Am.Bee J. 1, 61–63.

Bruneau E., Barbier E., Gallez L.M., Guyot-DeclerckC. (2000) La roue des arômes des miels, Abeilles& Cie 77, 16–23.

Gonnet M., Vache G. (1979) Technique de dégustationdes miels et recherche d’un système de notation etde classification objectif pour apprécier leur qual-ité par l’analyse sensorielle, 27th Apimondia Int.Apic. Congr., Athènes, pp. 499–506.

Gonnet M., Vache G. (1985) Le goût du miel, Ed.U.N.A.F., Paris, France.

Gonnet M., Vache G. (1992) The taste of honey, Api-mondia, Bucarest.

Gonnet M., Vache G. (1998) Analyse sensorielledescriptive de quelques miels monofloraux deFrance et d’Europe, Ed. Abeille de France, Paris,France.

Guinard J.X., Noble A.C. (1986) Proposition d’uneterminologie pour une description analytique del’arôme des vins, Sci. Aliment. 6, 657–662.

International Honey Commission (IHC), Minutes ofthe IHC meetings: Toledo (1998), Dijon (1999),Celle (2000) Louvain-la-Neuve (2001a), Atene(2001b), Celle (2002). Online at: http://www.apis.admin.ch/host/honey/minutes.htm(accessed on 12 August 2004).

International Olive Oil Council (1987) Sensory anal-ysis of olive oil. Standard. Glass for oil tasting.COI/T.20/Doc. No. 5, online at: http://www.inter-nationaloliveoil.org/downloads/orga2.pdf(accessed on 14 June 2004).

International Olive Oil Council (1996) Sensory anal-ysis of olive oil. Method. Organoleptic assessmentof virgin olive oil. COI/T.20/Doc. No. 15/Rev. 1,online at: http://www.internationaloliveoil.org/downloads/orga6.pdf (accessed on 14 June 2004).

ISO 6658 (1985) Sensory analysis – Methodology –General guidance, International Organization forStandardization, Geneva (http://www.iso.org).

ISO 6564 (1985) Sensory analysis – Methodology –Flavour profile methods, International Organiza-tion for Standardization, Geneva (http://www. iso.org).

ISO 4121 (1987) Sensory analysis – Methodology –Evaluation of food product by methods usingscales, International Organization for Standardi-zation, Geneva (http://www.iso.org).

ISO 8589 (1988) Sensory analysis – General guidancefor the design of test rooms, International Organ-ization for Standardization, Geneva(http://www. iso.org).

ISO 5492 (1992) Sensory analysis – Vocabulary, Inter-national Organization for Standardization, Geneva(http://www.iso.org).

ISO 8586-1 (1993) Sensory analysis – General guid-ance for the selection, training and monitoringof assessors - Part 1: Selected assessors, Interna-tional Organization for Standardization, Geneva(http://www.iso.org).

ISO 8586-2 (1994) Sensory analysis – General guid-ance for the selection, training and monitoringof assessors – Part 2: Experts, InternationalOrganization for Standardization, Geneva(http://www.iso.org).

Issanchou S., Lesschaeve I., Koster E.P. (1995)Screening individual ability to perform descrip-tive analysis of food products: basic statementsand application to a Camembert cheese descriptivepanel, J. Sens. Stud. 10, 349–368.

Istituto Nazionale di Apicoltura (1999) Albo nazionaledegli allevatori di api regine. Albo nazionale degliesperti in analisi sensoriale del miele. Albo nazi-onale degli esperti in melissopalinologia, IstitutoNazionale di Apicoltura, Bologna.

Lavanchy P., Bütikofer U. (1999) Caractérisation sen-sorielle de fromages à pâte dure ou mi-dure fab-riqués en Suisse, Mitt. Lebensm. Hyg. 90, 670–683.


Meilgaard M.C., Dalgliesh C.E., Clapperton J.F.(1979a) Terminologie de flaveur de la bière, Bios10, 2.

Meilgaard M.C., Reid D.S., Wuborski K.A. (1979b)Reference standards for beer flavour terminologysystem, J. Am. Soc. Brew. Chem. 40, 119–128.

Meilgaard M., Civille G.V., Carr B.T. (1991) SensoryEvaluation Techniques, CRC Press, Boca Raton.

National Honey Board (2002a) Carbohydrates andsweetness of honey, Technical information andapplications for honey, National honey board,http://www.nhb.org/download/factsht/4tastes.pdf(accessed on 14 June 2004).

National Honey Board (2002b) Sensory attributesof honey, Technical information and applica-tions for honey, National honey board,http://www.nhb.org/download/factsht/sensory.pdf(accessed on 14 June 2004).

Pangborn R.M. (1964) Sensory evaluation of food:a look backward and foreward, Food Technol. 18,1309.

Persano Oddo L., Piro R. (2004) Main European uni-floral honeys: descriptive sheets, Apidologie 35(Suppl. 1), S38–S81.

Persano Oddo L., Piana L., Sabatini A.G. (1995) Con-oscere il miele. Guida all’analisi sensoriale, Ave-nue Media, Bologna, Italia.

Persano Oddo L., Sabatini A.G., Accorti M., ColomboR., Marcazzan G.L., Piana M.L., Piazza M.G.,

Pulcini P. (2000) I mieli uniflorali - Nuove schededi caratterizzazione, Ministero delle politicheAgricole e Forestali, Roma.

Peynaud E. (1980) Le goût du vin, Dunod Ed.,Paris.

Piana M.L. (1993a) Aromi del miele: gli assaggiatoriin grado di rilevarli, Apitalia 20, 22–25.

Piana M.L. (1993b) Contenuto in acqua e cristalliz-zazione variano l'intensità dell'odore del miele,Apitalia 20, 20–23.

Sáinz Laín C., Gómez Ferreras C. (2000) Mielesespañolas. Características e identificación medi-ante el análisis del pollen, Ed. Mundi-Prensa,Madrid.

Serra Bonvehí J., Gómez Pajuelo A. (1988) La califi-cación de mieles mediante el análisis organolép-tico, Apiacta 23, 103–108.

SSHA (Societé Scientifique d’Hygiène Alimentaire),ISHA (Institut Scientifique d’Hygiène Alimentaire)(Eds.) (1990) Évaluation sensorielle. Manuelméthodologique, Lavoisier, Paris.

Stone H., Sidel J.L. (1985) Sensory evaluation prac-tices, Academic Press, San Diego.

Vejsnaes F., Theuerkauf R., Wienberg L. (2003) Thepractical application of sensory analyses in theevaluation of honeys, 38th Apimondia Int. Apic.Congr., Ljubljana, CD Rom, Book of Abstract,pp. 784–785.

To access this journal online:www.edpsciences.org

sensory analysis applied to honey: state of the art

Documents