Review article

Characterization of unifloral honeys

L Persano Oddo MG Piazza AG Sabatini M Accorti

1 Istituto Sperimentale per la Zoologia Agraria, Sezione di Apicoltura, Via L Rech 36, 00156 Rome;2 Istituto Nazionale di Apicoltura, Via S Giacomo 9, 40126 Bologna;

3 Istituto Sperimentale per la Zoologia Agraria, Via Lanciola, Cascine del Riccio, 50126 Florence, Italy

(Received 15 February 1995; accepted 8 August 1995)

Summary — The characterization of 14 types of Italian unifloral honeys was carried out on the basisof the organoleptic, microscopic (qualitative and quantitative melissopalynological analysis) andphysicochemical properties (colour, moisture, ash, HMF, diastase, pH, total acidity, electrical con-ductivity, specific rotation and sugars). The botanical origins of the examined honeys were Arbutus unedoL, Castanea sativa Mill, Citrus spp, Erica arborea L, Eucalyptus camaldulensis Dehnh, Hedysarum coro-narium L, Helianthus annuus L, Rhododendron spp, Robinia pseudoacacia L, Taraxacum officinale Web,Thymus capitatus Hofmgg et LK, Tila spp, honeydew honey from Abies and honey from Metcalfa pru-inosa honeydew. The synoptic picture emerging from the results can be used as a reference frameworkfor the diagnosis of unifloral honeys.

honey / Italy / melissopalynology / physicochemical characterization / organoleptic analysis

INTRODUCTION

The very first problem that arises whenapproaching the subject of unifloral honeysis when to define a honey as unifloral. Acontinuous series of intermediate possibili-ties exists between a multifloral and a uni-floral honey. At what point and on what basisdoes the unifloral/multifloral discriminationtake place?

International standards specify that"honey may be designed according to flo-ral or plant source if it comes wholly ormainly from that particular source and hasthe organoleptic, physicochemical and

microscopic properties corresponding withthat origin" (CAC, 1989, 6.1.4; EEC Direc-tive, 1974, 7.4.a), but this does not meanmuch unless these properties are definedthrough the prescription of given limits forprecise analytical parameters.

An extensive study was carried out onthe most common parameters used inhoney analysis, to identify a characteristicrange of values for the different unifloral

types. Numerous contributions have beenpublished on this subject (Persano Oddo and Amorini, 1985; Accorti et al, 1986; Per-sano Oddo et al, 1986, 1988a, b, 1990,1991; Sabatini et al, 1989, 1990; Piazza et

al, 1991). The aim of the present paper is tocomplete the results presented in previouspapers by adding new data obtained aftertheir publication, and to provide a synopticpicture, which can be used as a referenceframework for the diagnosis of unifloral hon-eys.

Data are related to Italian productions,but bibliographic comparisons generallyshow that the honey characteristics areaffected more by their botanical origin thanby their geographic provenance. In somecases discrepancies were noted betweenour results and those presented by otherauthors on the same honey types, but theycould mostly be accounted for by differencesin analytical methods or by the botanicalorigin of the honey, which was not exactlythe same (same genus, different species,for instance). Some geographic aspects (cli-mate, soil acidity or other pedologic condi-tions) may, however, influence the nectarand honey composition. Therefore, cautionmust be taken in applying the ranges char-acteristic of Italian unifloral honeys in othercountries.

MATERIALS AND METHODS

More than 2 000 honey samples of differentbotanical origin, declared as unifloral by the pro-ducer, were collected and analyzed over a periodof 10 years.

Three analytical approaches were used, forthe characterization of the different honey types,with reference to organoleptic properties (Gonnetand Vache, 1985), qualitative and quantitativemelissopalynological analysis (Louveaux et al,1978) and physicochemical parameters. Theseparameters included colour (Aubert and Gonnet,1983), moisture, ash, hydroxymethylfurfurale(HMF) and diastase (CAC, 1989), pH, total acid-ity (White et al, 1962), electrical conductivity (Lou-veaux et al, 1973), specific rotation (Battagliniand Bosi, 1973; Junk and Pancoast 1973) andsugars (Sabatini et al, 1984).

The criteria for selecting samples and definingunifloral groups were based on the organoleptic

evaluation and on the uniformity of behaviour withregard to the various analytical parameters. Four-teen honey types from the following botanical ori-gins were thus selected and described: Arbutusunedo L; Castanea sativa Mill; Citrus spp; Ericaarborea L; Eucalyptus camaldulensis Dehnh;Hedysarum coronarium L; Helianthus annuus L;Rhododendron spp; Robinia pseudoacacia L;Taraxacum officinale Web; Thymus capitatusHofmgg et LK; Tilia spp; honeydew honey fromAbies; and honey from Metcalfa pruinosa hon-eydew. (M pruinosa (Say) is a Flatid planthop-per which has spread in north-eastern Italy overthe last few years. This polyphage insect para-sites on many different plant species. Its honey-dew is actively foraged by bees which then pro-duce a characteristic honey (Barbattini et al,1991 ).) Further details on materials and meth-ods and on bibliographic comparisons with otherstudies can be found in the quoted literature.

RESULTS AND DISCUSSION

Organoleptic properties

The organoleptic properties are reported intable I. Descriptions can only provide somegeneral information, due to the difficulty ofgiving an exact ’translation’ of sensorial per-ceptions in words. On the other hand, thepossibility of correctly diagnosing the botan-ical origin of the honey through the sensorialanalysis is strictly bound to the personalability and experience of the taster.

Melissopalynological characteristics

The melissopalynological data reported intable I include the minimum percentage ofthe specific pollen, the absolute content ofpollen grains in 10 g honey and the repre-sentative class (according to Louveaux etal, 1978) found in the selected samples.

Honeys from Robinia, Citrus, Tilia,Rhododendron, Arbutus and Thymus haveunder-represented pollen (class I), so the

percentages of the respective pollens in uni-floral honeys are generally low to very low.The most typical ranges are 10 to 20% forCitrus and Arbutus, 15 to 30% for Robiniaand Thymus and 30 to 60% for Rhododen-dron. Honey from Tilia can even have noTilia pollen at all, because of the sterility ofmany cultived varieties of this plant.

Castanea and Eucalyptus honeys belongto the over-represented class (III and IV)and in unifloral samples their respective pol-lens are greater than 90% (in most of Cas-tanea honeys higher than 95%) and canmake about 100% of the pollens.

Erica and Hedysarum behave like ’nor-mal’ honeys, for which 45-50% of the spe-cific pollen is required to be considered uni-floral. The most typical range is 60 to 90%for Hedysarum, with a PK/10 g of 10 000 to50 000, and 50 to 80% for Erica, which isusually richer in pollen (PK/10 g from 50 000to 150 000). Hedysarum honey could beassumed to be affected by a secondary con-tamination, since the species is actively vis-ited by bees for pollen. This would explainthe high percentage associated with a rela-tively low absolute pollen content.

Helianthus can behave as a normal or

under-represented honey. In most of theunifloral samples, it composed 15 to 50%of the pollen found, with an absolute numberof pollen grains not exceeding 20 000/10 g honey. In some other samples, both thepollen percentage and the PK/10 g werehigher (60 to 90% and 25 000 to 50 000,respectively).

Taraxacum honey is characterized bylow percentages (usually 5 to 15%) and arelatively high absolute pollen content (classII), because of its almost constant contam-ination with Salix (which is over-repre-sented). This feature seems to be peculiar toItalian Taraxacum honeys.

Honeydew honeys show great variabil-ity in their palynological characteristics. Boththe absolute pollen content and the honey-

dew elements/pollen (HDE/P) ratio arespread over a very large range of values. Itis to be noted that in Abies honeydew theHDE/P ratio rarely complies with the value of3 given as the minimum limit for diagnos-ing honeydew honeys from central Europe(Louveaux et al, 1978).

Physicochemical characteristics

The physicochemical parameters (tables IIand III) have a different diagnostic value forthe various groups. Generally speaking, thehigher or lower values are a more charac-teristic feature than medium values, becausewider overlapping occurs in the mediumarea. By means of analysis of variance, elec-trical conductivity proved to have the highestdiscriminatory power (Stefanini, 1988). Mois-ture and HMF are more important for qual-ity evaluation, but also show a certain vari-ability according to botanical origin, probablydue to the production period (water content)or to other composition factors. For theseparameters the range of variability is notgiven in table II, because honeys with highwater and HMF content were rejected, toavoid including low quality samples.

Robinia, Hedysarum, Rhododendron andCitrus honeys are characterized by theirvery light colour and low conductivity andash values; Robinia and Citrus have a lowdiastase content. Robinia also shows a par-ticular sugar composition, with a high fruc-tose and a low glucose content and conse-quently a high fructose/glucose (F/G) anda low glucose/water (G/W) ratio, whichdetermine its liquid physical state.

Hedysarum is further distinguished by itslow pH and relatively high acidity; Rhodo-dendron has particularly low negative spe-cific rotation values.

Castanea and the 2 honeydew honeysare the darkest. They also have the high-est pH, electrical conductivity and ash val-ues. In addition, Castanea honey, a typi-

cally liquid honey, is characterized by a highfructose and a low glucose content, with ahigh F/G and a low G/W ratio. Honeydewhoneys have many other distinctive fea-tures. They are the only honeys with positivespecific rotation values and present a typicalglucide spectrum, poor in monosaccharidesand rich in di- and trisaccharides (in partic-ular, melezitose in Abies honeydew andmaltotriose in honeydew from Metcalfa).Metcalfa honeydew also shows the highestacidity and diastase values.

Both Helianthus and Taraxacum honeyshave a very high monosaccharide (partic-ularly glucose) content, giving rise to a veryhigh G/W and a very low F/G ratio. This istypical for honeys which granulate rapidly.Taraxacum is also characterized by its lowacidity and diastase content; Helianthus hasa very typical bright yellow colour.

Erica and Arbutus honeys have highmoisture and acidity, low diastase and quitehigh ash and electrical conductivity, mid-way between the honeys from nectar andthose from honeydew and Castanea. Ericais also characterized by its uncommonlyhigh HMF content, exeeding that of all theother honey types, and by its dark colourwith orange hues.

For Thymus honeys the most diagnos-tic parameters are acidity, diastase and spe-cific rotation, all 3 showing very high val-ues.

Eucalyptus and Tilia honeys are less eas-ily identifiable on the basis of their physico-chemical values, because of their ’average’behaviour. Tilia also shows great variabil-ity, probably due to the contaminations thatfrequently occur with Tilia honeydew. Themicroscopic examination is not very help-ful either, because Eucalyptus is over-repre-sented, and many cultivated varieties of Tiliaare extremely under-represented. For thesehoneys, therefore, the organoleptic analy-sis is the most significant diagnostic crite-rion.

An important remark concerns the com-pliance of these unifloral honeys with inter-national standards. Some of the types stud-ied have characteristics that go beyond theprescribed limits. It would be appropriatefor regulations to be modified taking into

account these particular features, ie highash value for Castanea, high acidity for Thy-mus, Metcalfa honeydew, Arbutus andErica, low diastase values for Robinia,Taraxacum, Arbutus and Erica, high sucrosecontent for Hedysarum, and low reducingsugars for honeydew honeys.

CONCLUSIONS

The data summarized in the tables give afairly good description of 14 Italian unifloralhoneys. Some of them are more easily iden-tifiable due to the presence of particularcharacteristics, whereas others have an

’average’ behaviour.In summary, it is not possible to evalu-

ate exactly how much nectar of one or otherbotanical species there is in a honey and, inthe last analysis, this may not even be soimportant. Indeed, what is really of interestis that a unifloral honey complies with whatit is expected to be, ie that an orange honeylooks, smell and tastes like an orangehoney. So, if, for example, it contains a cer-

tain quantity of Hedysarum nectar that doesnot substantially alter its organoleptic andanalytical picture, this is less important thana low contamination from a ’stronger’ nectarcapable of modifying its characteristics.

In general, we believe that only a globalexamination, which takes into account allthe analytical aspects, can lead to a rea-sonably correct judgment, because no sim-ple parameter can assure an exact diagno-sis.

Even melissopalynological analysis givesrise to important interpretation problems;often the sole presence of a dominant pollen

(> 45%) was considered the key to ’uniflo-rality’, whereas it is now currently acceptedthat microscopic analysis alone is not suffi-cient to establish the botanical origin of hon-eys and that, in any case, each honey typehas its own palynological pattern which mustbe known before drawing any conclusions.

As far as the physicochemical charac-teristics are concerned, a statistical studycarried out by cluster analysis (Stefanini,1984, 1988; Persano Oddo et al, 1988b)showed that this analytical approach has asignificant discriminating power. However, amore or less widespread overlapping of thehighest values of one group with the low-est values of the next group occurs in almostall the parameters examined. On the otherhand, the variability of the product makesit difficult to establish precise limits.

An organoleptic analysis carried out byan expert can provide a fairly precise eval-uation. It should not to be forgotten that thefinal judge, the consumer, can only appre-ciate the organoleptic properties. However,strictly it depends on the taster’s ability, andit is difficult to express the results in an

objective and repeatable way that can beused by others.

To conclude, we think that a reliable diag-nosis will only be achieved by means of anorganoleptic examination integrated bymicroscopic and, physicochemical data and,above all, correctly interpreted by an analystwho has a good knowledge of the product.

Résumé — Caractérisation des miels uni-floraux. Une étude approfondie a été effec-tuée sur 14 types de miel unifloral sur labase des propriétés organoleptiques, micro-scopiques (analyse mélissopalynologiquequalitative et quantitative) et physico-chi-miques (couleur, humidité, cendres, HMF,diastase, pH, acidité totale, conductibilitéélectrique, rotation spécifique et sucres).Les miels analysés proviennent des originesbotaniques suivantes : Arbutus unedo L,

Castanea sativa Mill, Citrus spp, Erica arbo-rea L, Eucalyptus camaldulensis Dehnh,Hedysarum coronarium L, Helianthusannuus L, Rhododendron spp, Robiniapseudoacacia L, Taraxacum officinale Web,Thymus capitatus Hofmgg et LK, Tilia spp,miel de miellat de sapin et miel de miellatproduit par Metcalfa pruinosa. La descriptiondes caractéristiques organoleptiques(tableau I) fournit des renseignementsessentiellement orientatifs, à cause de ladifficulté à traduire en paroles les percep-tions sensorielles. Ce type d’analyse estétroitement lié à l’expérience du dégusta-teur. Les analyses mélissopalynologiques(tableau I) montrent que les miels de Robi-nia, Citrus, Tilia, Rhododendron, Arbutusand Thymus sont sous-représentés ; ceuxde Castanea et Eucalyptus sur-représen-tés ; Erica et Hedysarum appartiennent àla classe normale ; Taraxacum, Helianthuset les miels de miellat ont un comportementvariable. Les propriétés physico-chimiques(tableaux II, III) des différents miels uniflo-raux sont les suivantes : Robinia : couleur

claire, valeurs basses de conductibilité élec-trique, cendres, diastase, glucose et rap-port glucose/eau (G/W) ; valeurs élevéesde fructose et du rapport fructose/glucose(F/G). Hedysarum : couleur claire, valeursbasses de conductibilité électrique, cendres,pH ; valeurs élevées d’acidité. Rhododen-dron : couleur claire, valeurs basses deconductibilité électrique, cendres et rotationspécifique. Citrus : couleur claire, valeursbasses de conductibilité électrique, cendreset diastase. Castanea : couleur foncée,valeurs élevées de pH, cendres, conducti-bilité, fructose et rapport F/G. Miel de miel-lat de sapin : couleur foncée, valeurs éle-vées de pH, cendres, conductibilité, di- ettrisaccharides (mélézitose) ; valeurs bassesde monosaccharides, valeurs positives derotation spécifique. Miel de miellat produitpar Metcalfa pruinosa : couleur foncée,valeurs élevées de pH, cendres, conducti-bilité, acidité, diastase, di- et trisaccharides(maltotriose), valeurs basses de monosac-

charides, valeurs positives de rotation spé-cifique. Helianthus : couleur jaune vive,valeurs élevées de monosaccharides (glu-cose) et du rapport G/W, valeurs basses durapport F/G. Taraxacum : valeurs élevées demonosaccharides (glucose) et du rapportG/W, valeurs basses de diastase, acidité etdu rapport F/G. Erica : couleur foncée avecdes nuances orange, valeurs élevées d’aci-

dité, eau, HMF, cendres et conductibilité,valeurs basses de diastase. Arbutus :valeurs élevées d’acidité, eau, cendres etconductibilité, valeurs basses de diastase.Thymus : valeurs élevées d’acidité et dias-tase, valeurs fortement négatives de rotationspécifique. Eucalyptus et Tilia ont un com-portement moyen par rapport à tous lesparamètres examinés et sont donc moinsfacilement identifiables. Par conséquentl’examen organoleptique est particulière-ment important pour ces miels. D’une façongénérale, nous pouvons conclure qu’un dia-gnostic fiable des miels unifloraux requiertune analyse organoleptique intégrée parles données mélissopalynologiques et phy-sico-chimiques et, surtout, interprétée cor-rectement par un analyste ayant une bonneconnaissance du produit.

miel / Italie / mélissopalynologie / carac-térisation physico-chimique / analyseorganoleptique

Zusammenfasung — Charakterisierungvon Sortenhonigen. An 14 Arten italieni-scher Sortenhonige wurde eine ausführli-che Studie zu ihrer Charakterisierung durch-geführt. Diese umfasste organoleptische,mikroskopische (qualitative und quantita-tive melissopalynologische Analysen) undphysikalisch-chemische Eigenschaften (Fär-bung, Wassergehalt, Asche, HMF, Diastase,pH, Säuregehalt, Leitfähigkeit, optische Dre-hung und Zucker). Die botanische Herkunftder untersuchten Honige war: Arbutusunedo L, Castanea sativa Mill, Citrus spp,Erica arborea L, Eucalyptus caldulensis

Denh, Hedisarum coronarium L, Helianthusannuus L, Rhododendron spp, Robiniapseudacacia L, Taraxacum officinale Web,Thymus capitatus Hofmgg et LK, Tilia spp,Honigtauhonig von Abies und Metcalfa prui-nosa. Die Beschreibung der organolepti-schen Merkmale (Tabelle I) dient nur zurgenerellen Information, da Sinneseindrückeschwer in Worte gefaßt werden können.Diese Analysen sind sehr stark von Fähig-keiten und Erfahrung der Schmecker abhän-gig. Die melissopalynologische Analyse(Tabelle I) zeigt, daß die Honige von Robi-nia, Citrus, Tilia, Rhododendron, Arbutusund Thymus unterrepräsentiert sind; Honigevon Castanea und Eucalyptus sind überre-präsentiert; Erica und Hedysarum liegen immittleren Bereich während Taraxacum, Heli-anthus und die Honigtauhonige sehr varia-bel sind. In den physikalisch-chemischenEigenschaften (Tabellen II, III) wurden dieverschiedenen Sortenhonige wie folgt cha-rakterisiert: Robinia: helle Färbung, nied-rige Werte bei Leitfähigkeit, Asche, Dia-stase, Glucose sowie ein niedrigesVerhältnis von Glukose/Wasser; hoher Fruc-tosegehalt und hohes Verhältnis von Fruc-tose/Glucose. Hedysarum: helle Farbe, nied-rige Werte bei Leitfähigkeit, Asche, pH;hoher Säuregehalt. Rhododendron: helleFarbe, niedrige Werte bei Leitfähigkeit,Asche und der optischen Drehung. Citrus:helle Farbe, niedrige Werte bei Leitfähig-keit, Asche und Diastase. Castanea: dun-kle Farbe, hohe Werte bei pH, Asche, Leit-fähigkeit, Fructose und dem Verhältnis vonFructose/Glucose. Abies Honigtau: dunkleFarbe, hohe Werte bei pH, Asche, Leit-fähigkeit, Di- und Trisacchariden (Melezi-tose); niedriger Gehalt an Monosacchari-den und positive optische Drehungswerte.Honigtau von Metcalfa pruinosa: dunkleFarbe, hohe Werte bei pH, Asche, Leit-fähigkeit, Säuregehalt, Diastase, Di- undTrisacchariden (Maltotriose); niedrigerGehalt an Monosacchariden und positiveoptische Drehungswerte. Helianthus: hell-gelbe Farbe, hoher Gehalt an Monosac-

chariden (Glucose) und hohes Verhältnisvon Glucose/Wasser; niedriges Verhältnisvon Fructose/Glucose. Erica: dunkle Farbemit oranger Tönung; hohe Werte bei Säu-regehalt, Wassergehalt, HMF, Asche undLeitfähigkeit; niedriger Diastasewert. Arbu-tus: hohe Werte bei Säuregehalt, Wasser-gehalt, Asche und Leitfähgigkeit; niedrigeDiastasewerte. Thymus: hoher Säure- undDiastasegehalt, stark negative optische Dre-hungswerte. Eucalyptus und Tilia nehmen inallen untersuchten Parametern eine Mittel-

stellung ein und können daher weniger leichtidentifiziert werden. Bei diesen Honigen istdie organoleptische Analyse besonderswichtig. Letztendlich kann eine verläßlicheSortendiagnose nur durch die Integrationvon organoleptischer Überprüfung mit mikro-skopischen und physikalisch-chemischenWerten erreicht werden, für deren korrekte

Interpretation ein Untersucher mit gutenKenntnissen des Produkts erforderlich ist.

Honig / Italien / Melissopalynologie / phy-sikalisch-chemische Charakterisierung /organoleptische Analyse

REFERENCES

Accorti M, Persano Oddo L, Piazza M, Sabatini A (1986)Schede di caratterizzazione delle principali qualit_ dimiele italiano. Apicoltura 2, supplement

Accorti M, Piazza M, Persano Oddo L (1983) Electricalconductivity and ash in honeys. XXIX Inter Congr ofApiculture, Apimondia, Budapest, 343

Accorti M, Piazza M, Persano Oddo L (1987) La con-ductivité électrique et le contenu en cendre du miel.Apiacta XXII, 19-20

Aubert S, Gonnet M (1983) Mesure de la couleur desmiels. Apidologie 14, 105-118

Barbattini R, Greatti M, lob M, Sabatini A, MarcazzanGI, Colombo R (1991) Osservazioni su Metcalfapruinosa (Say) e indagine sulle caratteristiche delmiele derivato dalla sua melata. Apicoltura 7, 113-135

Battaglini M, Bosi G (1973) Caratterizzazione chimico-fisica dei mieli monoflora sulla base dello spettroglucidico e del potere rotatorio specifico. Scienza eTecnologia degli Alimenti III 4, 217-221

Codex Alimentarius Commission (1989) Codex stan-dard for sugars (honey). Worldwide standard. FAO-WHO, CAC Vol 3, ed 1, suppl 2, Rome

European Economic Community (1974) EEC Councildirective of 22 July 1974 on the harmonisation ofthe laws of the members states relating to honey.Off J Eur Comm No L221, 10-14

Gonnet M, Vache G (1985) Le goût du miel. UNAF, Paris

Louveaux J, Maurizio A, Vorwohl G (1978) Methods ofmelissopalynology. Bee world 59, 139-157

Louveaux J, Pourtallier MP, Vorwohl G (1973) Méth-odes d’analyses des miels. Conductivité. Bul TechApic XVI, 2

Junk WR, Pancoast HM (1973) Handbook of Sugars.Avi Publ Co, USA, 295

Persano Oddo L, Amorini MT (1985) Analisi melis-sopalinologica quantitativa dei principali tipi di mieleitaliano. Apicoltura 1, 105-122

Persano Oddo L, Baldi E, Piazza MG (1986) Acidità epH nei principali mieli uniflorali italiani. Apicoltura 2,145-154

Persano Oddo L, Piazza MG, Accorti M (1988a) Diag-nosis of unifloral boneys. I. Present knowledge andproblems. Apicoltura 4, 1-12

Persano Oddo L, Stefanini R, Piazza MG, Accorti M(1988b) Diagnosis of unifloral honeys. III. Applicationof a statistical approach to honey classification. Api-coltura 4, 27-38

Persano Oddo L, Baldi E, Accorti M (1990) Diastatic activ-ity in some unifloral honeys. Apidologie 21, 17-24

Persano Oddo L, Piazza MG, Sabatini AG (1991) Schededi caratterizzazione del miele di tiglio e del miele dimelata di Metcalfa pruinosa. Apicoltura 7, 151-159

Piazza MG, Accorti M, Persano Oddo L (1991) Electricalconductivity, ash, colour and specific rotatory powerin Italian unifloral honeys. Apicoltura 7, 51-63

Sabatini AG, Nanetti A, Maurizi M, Lercker G (1984)Studio sull’origine botanica dei mieli attraverso il pro-filo gascromatografico dei componenti neutri. Rivmerceol 23, 71-81

Sabatini AG, Persano Oddo L, Piazza MG, Accorti M,Nanetti A (1989) Glucide spectrum in the main Ital-ian unifloral honeys. I. Fructose and glucose. Api-coltura 5, 35-46

Sabatini AG, Persano Oddo L, Piazza MG, Accorti M,Marcazzan G (1990) Glucide spectrum in the mainItalian unifloral honeys. II. Di- and trisaccharides.

Apicoltura 6, 63-70Stefanini R (1984) Variability and cluster analysis of Ital-

ian honeys. Apiacta XIX, 109-114

Stefanini R (1988) Diagnosis of unifloral honeys. II.

Numerical taxonomy of honey and statistical diag-nostic test of its botanical origin. Apicoltura 4, 13-26

White JW Jr, Riethof FL, Subers MH, Kushnir I (1962)Composition of American Honeys. Tech Bull USDA,1261, Washington, DC, USA