Journal of Archaeological Science (2001) 28, 331–340doi:10.1006/jasc.2000.0606, available online at http://www.idealibrary.com on

Chemical and Textural Characterization of Tin Glazes inIslamic Ceramics from Eastern Spain

Judit Molera and Mario Vendrell-Saz

Dept Crystallography and Mineralogy, University of Barcelona, c/Martı i Franques s/n, 08028 Barcelona, Spain

Josefina Perez-Arantegui

Dept Analytical Chemistry, Faculty of Sciences, University of Zaragoza, 50009 Zaragoza, Spain

(Received 20 November 1998, revised manuscript accepted 25 May 2000)

Several productions of Islamic tin glazed pottery from eastern Spain have been studied under the chemical andmicrostructural points of view by means of WDS, SEM/EDX, XRD and XRF analyses. Samples of Islamic potteryfrom the workshops of Murcia 10th, Zaragoza 11th, Mallorca 11th, Denia 13th, Granada 14th and Cordoba 10th,which represent a wide range of local productions from medieval eastern Spain, have been studied in order toobtain the trends of the technical and compositional evolution. From the experimental data, some commonfeatures can be established, as well as some differences. All the Islamic Spanish opaque glazes are lead glazeswith PbO contents from 37 to 56%, opacified with tin oxide in the range 4–15%. In all the cases, they were appliedon a previously biscuited body made with a Ca-rich clay, probably to produce a buff colour less apparent throughthe glaze. The thicknesses range from 100 to 150 microns and the opacification is achieved by small crystals ofSnO2 (under a micron of size). The main differences are the size and distribution of such small crystals, beingsmaller in the early Islamic productions (Zaragoza and Murcia) and bigger in the late productions.

� 2001 Academic Press

Keywords: ISLAMIC, POTTERY, TIN GLAZE, SPAIN, OPACIFICATION, GLAZE CHEMICALCOMPOSITION.

Al-Andalus In The Early Medieval Age

A l-Andalus (or the Muslim area of the IberianPeninsula) became part of the Islamic world in711, when a Muslim army crossed the Straits of

Gibraltar, conquered the Visigoths, and establishedmost of the peninsula as a western province under therule of the Umayyad caliph in Damascus (Arie, 1984;Reilly, 1993). Perched at the edge of the Islamic west,al-Andalus took an autonomous existence under therule of the early emirs.

Political developments in the 10th century broughtan end to the somewhat subservient status ofal-Andalus in relation to the eastern Islamic world.When the Umayyad ruler Abd al-Rahman III declaredhimself caliph in 929, Muslim Spain became a majorplayer on the stage of Mediterranean politics andculture (Figure 1).

After nearly two and a half centuries of relativelystable centralized Umayyad rule in al-Andalus, thedynasty crumbled in the early 11th century (usuallydated to 1031) in the face of disputes over succession

3310305–4403/01/030331+10 $35.00/0

and civil war. In its place, smaller independent statesemerged called Taifas or Mulllluk al-tawa’if (‘‘PartyKingdoms’’), as Toledo, Badajoz, Sevilla, Cordoba,Granada, Murcia, Zaragoza, Valencia, Denia,Mallorca, etc. This period corresponds to a flourishingcultural and economic development of these smallkingdoms.

Because of the need for military assistance afterChristian pressure to the south and the loss of Toledoin 1085, Almoravid armies (a Berber dynasty based inMarrakech) arrived at the Iberian Peninsula in 1086and shortly established Almoravid control over boththe peninsula and North Africa. Unified Almoravidrule marked the end of Taifa diversity, and establishedal-Andalus as part of a huge single kingdom. Internaldissent gradually led to the disintegration ofAlmoravid power in the early 12th century and thebrief reappearance of Taifas before another Berberdynasty, the Almohads, incorporated al-Andalus intotheir North African empire in 1147. Christian con-quests during the first half of the 13th century broughtan end to Muslim rule in most of the south.

� 2001 Academic Press

332 J. Molera et al.

The Production of Tin Glazes in the IslamicWorldThe beginning of glazed pottery can be dated backto the Middle Assyrian period (13th century ) inMesopotamia, though glazes on small objects ofquartz-frit body—Egyptian faience—are much older(Kaczmarczyk & Hedges, 1983; Kleinmann, 1986). Butthe recipes of the old traditional glazes were modifiedto produce a new type, an opaque white alkaline–leadglaze, with tin oxide as opacifier to get a white ware.However, the exact nature and timing of the firstdevelopment of tin-opacificaiton in the Islamic world isstill being explored.

According to Allan (1991), the import in the 9thcentury of Chinese white wares and high-fired porce-lains into Iraq led the local Arab potters to experimentin reproducing the whiteness of the Chinese originals intheir own low-fired pottery. The Iraqi pottery bodies(earthenware) exhibit a yellowish buff colour afterfiring and one of the ways developed by the potters toachieve the white colour was to cover it with tin glaze(a glaze containing suspended particles of tin oxidewhich make the glaze opaque white). The same author(Allan, 1971) says that the potters of the Parthianperiod commonly using alkaline glazes often added tinoxide to the recipe to make them white and opaque,anticipating the extensive use of these materials in theIslamic period.

Mason & Tite (1997) propose a progressive develop-ment in the technology to produce opaque glazes in theIslamic world, from the 8th to the 10th centuries.Building initially upon the traditional glaze-makingtechnology in Iraq, this development involved anincreasing reliance on tin oxide as opacifier, togetherwith an increase of the lead content in the glaze.So that, keeping in mind the link between theformer tin-opacified wares and earlier white opaque-glazed wares of pre-Islamic Iraq, these authors

question the development of Islamic tin-opacifiedglazes attributed to the need to copy imports ofChinese whitewares.

The technology to produce tin-opacified glazesspread from Middle East to Iran, Egypt, northernAfrica, Spain, and ultimately into later medievalEurope. In this context, al-Andalus produced andexported ceramics, as well as imported ceramics fromother areas of the Islamic world and beyond. Thus, aconstant trade of objects was established throughoutall Muslim’s world. The evidence for trade of Andalusiceramics comes both from documents and existingexamples (Constable, 1994: 167–168, 189–191, 222–223).

There were no tin mines in the Islamic Near East andthe tin oxide had to be imported by sea from southernBurma and Malaysia (Allan, 1991). Meanwhile, theIberian Peninsula was rich in metals and several ofthe Andalusi ores and minerals were exported to otherMediterranean regions, among them tin was one of theexports (Constable, 1994: 186–187). Accordingly, thedevelopment in the making of tin-glazed ceramicscould be provided by the chance of availability of rawmaterials.

LEON

TOLEDO

ZARAGOZA

CÓRDOBA

GRANADA

MURCIADENIA

PATERNA

MALLORCA

0 50 100150 km

Figure 1. Map of Islamic Spain in Mediaeval ages. —— Islamicborder in the 11th century; – – – Islamic border at the beginning ofthe 13th century.

Tin Glazes in Islamic Ceramic From EasternSpainTwo different styles were used for decorating objectswith white glaze in Islamic ceramics. In the beginning,designs were painted over the glaze with metal oxides:blue (cobalt), green (copper), purple (manganese) andyellow (antimony forming a lead-antimoniate—Pb2Sb2O7—stable in the glaze). The second and moreimportant style of decoration found on tin-glazedwares is lustre painting, often associated to bluedecoration.

The tin-glaze technology developed in the IberianPeninsula since the 10th century, started with greenand brown designs painted on white glazes (the so-called ‘‘verde y morado’’), which flourished during theCaliphal period, a period of political and culturalsummit (Rossello, 1992; Rossello, 1995). The former‘‘green and brown’’ pottery was a palatial productionassociated to the splendour of Madınat al-Zahra’(‘Abd al-Rahman III palace, 5 km from Cordoba) andit was probably made in workshops attached to thecaliphal palace. Although this ceramic productionfrom Madınat al-Zahra’ involved an extraordinaryadvance in Andalusi pottery production, the ‘‘greenand brown’’ style was spread to other areas ofal-Andalus, as proved by its wide diffusion (Retuerce &Zozaya, 1986) which over time reached all the Taifakingdoms and continued after the Christian conquest.

Lustre pottery appeared quite early (around the 11thcentury) but few workshops have been found, mean-while 13th century and subsequent workshops havebeen documented. The Nasrid kingdom of Granada

Characterization of Tin Glazes from Eastern Spain 333

(14th century) was famous for its joinery, metalwork,carpet textiles, leather, architectural plaster relieves,but also for its golden pottery. And finally, hispano-moresque (mudejar) lustre pottery from Valencia waslargely exported around the Mediterranean coast.

Objectives and Sampled SitesThe aim of this paper is to study and characterize thetin-glazes produced in the Islamic period in easternSpain, in order to determine the composition, techno-logical features (such as thickness, whiteness, opacity,etc.) and microscopic characteristics, and to establishchronological and geographical evolution of thesecharacteristics and of the techniques of production.This study forms part of a scientific programme de-voted to give a full picture of the Islamic ceramicproduction in Spain and its evolution to the mudejarand Christian productions (13th to 17th centuries)after the conquest of the Muslim territories.

Ceramic samples belonged to Islamic productionsdated from the 10th to the 14th centuries, all of themcoated with white glaze and often decorated with‘‘green and brown’’ designs. As far as it has beenpossible, the selected sites were ceramic workshops andthe samples studied in this paper include pieces fromMurcia, Zaragoza, Mallorca, Denia, Granada andCordoba (see Figure 1). This collection of sites coversa wide geographical area and a large chronologicalperiod, which seem adequate to give a complete char-acterization of the tin-glaze productions in easternSpain. Sherds were part of small jars, pots, jofainas(small plates) and ataifores (big plates).

The following sections give a short description ofeach of the sites and the ceramic production associatedto each particular workshop or archaeologicalexcavation.

MurciaThe workshop of San Nicolas is located in an Islamicsuburb of the mediaeval Murcia (Munoz, 1993) andwas found during the excavations of an Islamic cem-etery (Navarro Palazon, 1990). Two periods may beidentified: one dated between the end of the 9th and the10th centuries, and another one at about the 11thcentury, after which an Islamic cemetery occupied thearea. At that time, Murcia was an important city of theCordoba Caliphate and at the beginning of 11thcentury it became the capital of a Taifa kingdom untilthe Christian conquest by Jaime the First in 1226.

Only the early workshop is studied here, since theremains of the 11th century workshop were destroyedby the Islamic cemetery. Different types of pottery wereproduced, namely jars, orzas (unglazed cooking pots),jars, ataifores, jofainas, oil lamps, little jars, etc., someare lead glazed and some tin glazed. About 50 ceramicfragments that were found in ash deposits close to thekilns have been studied.

ZaragozaIn 1989, an important Islamic potters’ suburb wasexcavated on the eastern side of the ancient town,which produced pottery during the end of the 10th andthe 11th centuries (Aguarod & Escudero, 1991;Aguarod et al., 1991; Mostalac, 1995). Thousands ofceramic objects (jars, ataifores, cooking pots, oil lamps,etc., both non-glazed and glazed) and potter’s tools(bars, stilts, etc.) were found associated with ten kilnsand some waster spots. Several decoration techniqueswere used on the glazed ceramics from site, includingmonochrome (green or yellow) glazes, black on yellowglaze, green on yellow glaze, green and brown on whiteglaze and cuerda seca (Perez-Arantegui, 1997; Perez-Arantegui et al., 1999; Perez-Arantegui & Castillo, inpress). At this time, Zaragoza was an important townin al-Andalus, which became the capital of a Taifakingdom (between 1018 and 1118, after the fall ofthe Cordoba government) before its conquest by theChristian king Alfonso of Aragon (Arie, 1984; Reilly,1993).

MallorcaIn different excavations in the city of Palma deMallorca, there appeared waster spots dated to the11th century, which suggested local productions,although no remains of the kilns or the workshops hasbeen found (Rossello, pers. comm.). During the 11thcentury Mallorca was a Taifa kingdom with goodrelationships with other Taifas such as Denia, Murciaand Valencia. The found pottery consisted of someunglazed and glazed jars (deep green on the outer sideand white on the inner side).

DeniaThe current centre of Denia was an important potters’suburb in medieval time (Gisbert, Azuar & Burguera,1991). The remains of workshop wastes with pottertools (bars, stilts, ganchos), waster spots, frits and lotsof ceramic objects were found there, as well as 11 largerectangular kilns (Gisbert, 1990; Gisbert, Azuar &Burguera, 1991). The ceramic production was datedbetween the end of the 12th century and the Christianconquest in 1244, when the workshop was destroyed.The ceramic fragments analysed correspond to asampling of all of the types produced: different jars,ataifores, jofainas, candle-lamps, cooking pots andtiles. The tin-glazed ceramic corresponds to jars,ataifores and jofainas.

GranadaThe Muslim kingdom of Granada survived until 1492.Since 1238 until 1492 Granada remained the onlyintegral Islamic territory in Spain and the AlhambraPalace was famous for its cultural significance. Nasridlustre had dense and sometimes intricately involved

334 J. Molera et al.

designs, witness of a certain horror vacui, in blue andgolden colours. Granada also became an importantexport centre and pottery was regularly shippedbetween the 13th and throughout the 14th centuries toValencia, Mallorca, Barcelona and more distant ports(Caiger-Smith, 1985). Since any workshop from thisperiod has been excavated in Granada, only Nasridlustre ataifores exported to Mallorca have beenstudied. Unfortunately, they were quite weatheredby the burial conditions (septic pit) and only sevenfragments could be properly analysed.

Experimental Methodology

Small fragments of glazed ceramics were cut perpen-dicular to the glaze–body interface with a diamond sawand prepared as polished sections for examination byoptical microscopy and Scanning Electron Microscopy(SEM). The distribution and morphology of the tinoxide crystals were studied on BSE (Back ScatteredElectrons) images.

The bulk chemical compositions of glazes weredetermined using an attached wavelength-dispersiveX-ray spectrometer (WDS, Cameca SU30) and anenergy-dispersive X-ray spectrometer (EDS, LINK,ANALYTICAL eXL-10 system). X-ray analyses werecaried out by scanning a large area (�2000 or �4000magnification) of the glaze and by acquiring datathrough several points (between 10 and 15) following aline across the glaze. When only EDS (with ZAFcorrection) were used to determine the glaze compos-ition, the correlation between WDS and EDS resultswas previously studied in different samples and stan-dards in order to be able to compare analyses obtainedby both techniques.

In order to check the accuracy of the SnO2 contentvalues, some samples have been prepared by scratchingpowder of the glaze with a diamond tool and then thecomposition determined by X-ray fluorescence spec-trometry by preparing a pearl by fusing 100 mg ofsample with Li tetraborate.

The chemical and mineralogical compositions of theceramic bodies were carried out by X-ray fluorescenceand X-ray diffraction analyses, respectively.

Results and DiscussionIslamic tin glazes usually appear very weathered andsometimes the surface looks as if a powdered layer hascoated it because the shine and vitreous aspect havebeen lost. This is why they have been largely related towhite slips by some archaeologists, being lead–silicaglazes with tin oxide as opacifier. In all of the cases,glazes were applied on the ceramic body and no slipappeared between body and glaze.

All the samples studied were glazed on the two sides;usually the less important one (interior in jars and potsor exterior in plates) was coated with a yellow to honeyglaze while the main side was tin glazed. The thicknessof the tin-glaze ranges between 100 and 200 m (Table 1,Figure 2), thinner than eastern Islamic glazes from thesame period (Mason & Tite, 1994). Tin glaze wasapplied on a previously fired body, which is inagreement with the very thin reaction layer (<10 m)observed between glaze and body (Molera, 1996;Tite et al., 1998) and with archaeological evidence suchas many unglazed fired sherds (Navarro Palazon,1990).

Major chemical components of glazes are alwayssilicon (30–46% SiO2), lead (38–56% PbO) and tin(4–15% SnO2) with small quantities of alkalis (sodiumand potassium) and calcium (probably diffused fromthe paste during firing) (see Table 2). However,samples from different sites showed specific contents ofsome elements.

Tin appears as small crystals of cassiterite (SnO2)scattered in the glaze, which are responsible of theopacity by the scattering of light (Vendrell, Molera &Tite, 2000). As tin oxide re-crystallizes during the firing(Molera et al., 1999), the crystal size, distribution andconcentration could be a fingerprint of each produc-tion centre, suggesting slightly different processes ofhandling and firing.

Table 1. Microstructure of tin-opacified glazes

Sites ofproduction

Tin-glazethickness

(m)

Interfacethickness

(m)K-Pb

feldspars InclusionsSnO2 crystal

size (nm) Distribution

Murcia 10th 100–200 <10 Very few Few 200–700 HomogeneousZaragoza 11th 100–200 <10 Very few Many/big <800 HomogeneousMallorca 11th 150 <10 Very few Few <900 HomogeneousDenia 13th 150 <10 None None 300–900 HomogeneousGranada 14th 200 <10 None Many 300–800 HeterogeneousCordoba 10th 200–250 <10 Very few Few <800 Homogeneous

MurciaSamples from San Nicolas workshop of Murcia havehigh PbO content (�54%), some 33% of SiO2 andSnO2 ranges between 7 and 10%, all as cassiteritecrystals (there is no tin dissolved into the glaze), andthe thickness ranges between 100 to 200 m. Most of the

Characterization of Tin Glazes from Eastern Spain 335

Figure 2. SEM images of tin glazes from Murcia (a, b), Zaragoza (c, d), Mallorca (e), Denia (f) and Granada (g, h).

tin-glazes are greatly altered by the burial conditions(Islamic cemetery) and present large amounts of leadphosphates and carbonates as weathering by-products.

Bubbles appear near the interface between body andglaze, and they are often filled by cerussite as aweathering. The weathered glazes show less PbO

336 J. Molera et al.

Table 2. Composition of the tin-opacified glazes (%wt)

Site Na2O MgO Al2O3 SiO2 K2O CaO Fe2O3 SnO2 PbO

Murcia (N=13) 1·46 0·24 0·48 33·1 1·10 1·48 0·31 7–10 53·80·16 0·14 0·35 1·84 0·34 0·68 0·24 2·62

ZaragozaGroup 1 (N=9) <1·0–2·08 <0·36–0·87 0·61 46·1 3·74 3·73 0·70 5–7 38·0

0·38 1·30 0·66 0·89 0·40 1·65Group 2 (N=4) <1·0–1·98 <0·36–0·84 1·02 41·5 3·26 3·72 0·77 9–14 37·8

0·64 0·70 0·50 0·55 0·09 2·71Second size (N=9) <1·0–1·73 <0·36–0·91 2·60 39·5 3·85 4·28 1·54 4–10 40·0

1·07 1·31 1·30 0·86 0·61 2·09Mallorca (N=5) 1·00 0·23 3·23 31·1 1·18 2·40 1·21 7–9 53·9

0·20 0·08 1·56 1·38 0·50 0·66 0·54 2·70Denia (N=9) 1·80 0·22 0·99 30·9 0·55 0·46 0·14 6–7 56·2

0·62 0·05 0·14 1·13 0·11 0·23 0·11 1·14Granada (N=5) 1·26 0·30 0·88 40·8 2·69 1·67 0·49 7–9 43·7

0·31 0·05 0·81 2·89 0·30 0·38 0·26 1·92Cordoba (N=3) <1·0–1·40 <0·36–0·50 2·15 35·2 1·22 2·39 1·14 8–15 47·2

0·42 2·10 0·34 0·67 0·28 1·55

Table 3. Composition of ceramic bodies (%wt)

Site SiO2 Al2O3 Fe2O3 MgO CaO Na2O K2O

Murcia (N=13) 40·57 10·89 4·05 3·64 18·27 0·36 2·521·87 0·61 0·34 0·62 2·79 0·04 0·32

ZaragozaGroup 1 (N=6) 49·73 15·86 5·51 3·68 16·55 0·93 2·85

1·89 0·47 0·13 0·44 0·83 0·19 0·42Group 2 (N=4) 50·43 16·74 5·79 3·95 16·70 1·88 1·62

2·94 0·87 0·28 0·11 1·52 0·29 0·53Mallorca (N=6) 49·67 13·16 5·01 2·40 13·30 0·80 2·65

1·78 0·58 0·19 0·23 1·51 0·11 0·29Denia (N=9) 45·78 11·41 3·78 1·17 18·78 0·40 1·92

2·03 1·55 0·24 0·07 2·63 0·15 0·39Granada (N=7) 49·58 13·18 5·09 2·43 13·56 1·10 1·93

1·28 0·21 0·15 0·22 0·57 0·34 0·52Cordoba (N=3) 54·98 11·57 10·21 <1·00 18·54 n.d. 4·23

2·55 0·71 2·55 1·90 0·76

content than those non-altered glazes, suggesting thatsome lead was washed out. The results shown in Table2 correspond to those non-altered glazes with greenand brown decoration. Jars have transparent honeyglaze in the inner part, while ataifores are tin-glazed onboth sides.

Tin appears as very small crystals of cassiterite(mean size of 300 nm), very well distributed in theglaze (Figure 2(a),(b)). There are small rounded quartzgrains, probably as remains of the original rawmaterials. This microstructure is quite common inglazes prepared from frits as in Denia and Paterna(13th century) workshops (Molera, 1996). Unfortu-nately, the frits found in this workshop do not containtin and thus we cannot be sure if the tin glazes analysedwere made with lead–tin–silica frit.

The alkaline content (Na2O+K2O) is under 3% andthe alumina content is really low (<1%), which sug-gests the use of relatively pure quartz sand for making

the glazes (sand of quartz with low feldspars and clayscontent). The low interaction between ceramic bodyand glaze has been interpreted as a result of applyingthe glaze on a previously fired body—double firingprocess (Tite et al., 1998; Molera, 1996; Molera et al.,forthcoming paper). Moreover, double firing is clearlyassessed by the large amounts of unglazed biscuitedpieces found in the workshop (Navarro Palazon, 1990).Green and brown decoration was painted on theunfired glaze with CuO and MnO.

In all of the cases, the ceramic of tin glazed pots isCa-rich (�18% CaO) (Table 3), indicating a rawmaterial specially used for the glazed ceramic produc-tion, meanwhile the workshop tools and cooking potswere made with a non-calcareous coarser clay (<5%CaO). This differentiation suggests some degree ofspecialization in the San Nicolas workshop: coarse claywas used for kitchen pottery and tools, which neededto be resistant to the thermal shock, and a calcareous

Characterization of Tin Glazes from Eastern Spain 337

clay for making table pottery, including those whichwere glazed.

The use of a calcareous clay for tin glazed pots isconvenient for bleaching the body red colour of clay(Kreimeyer, 1987). By using a Ca-rich clay it results ina lighter colour because some of the Fe atoms will beincorporated to the crystal structures of pyroxenesand thus, the amount of Fe oxides formed is lower.Consequently, it is necessary for less tin to opacify theglaze. Besides, calcareous clay has a thermal expansionmore adequate for tin-glaze than non-calcareous clay(Tite et al., 1998).

2860

48

PbO

SiO

2

30 5040

44

40

36

32

Saragossagroup 1group 2second. side

Granada 14th

Córdoba 10thMurcia 10thMallorca 11thDenia 13th

Figure 3. Scatter plot SiO2-PbO (%wt).

ZaragozaThirty fragments were sampled for this study, althoughsome of them could not be analysed because theirglazes were very altered. When important weatheringwas observed by SEM, the sample was not consideredfor chemical analysis. White glazes from Zaragozashow low contents of PbO (�38%) and SiO2 between41% and 46% (see Table 2 and Figure 3). Two groupscan be distinguished in this site, one of them withslightly lower concentration of SiO2 and higher con-tents of SnO2. Other characteristic elements arealuminium (Al2O3�1%) and iron (Fe2O3<1%), whoselow contents indicate a very small contribution of clayin the glaze recipe, and the use of very pure compo-nents. The presence of alkaline elements is quite con-stant in this workshop, about 3% K2O and 1·5–2%Na2O. This concentration of alkalis could compensatethe lower percentage of PbO acting as a flux.

Opacity was provided by the presence of tiny tin-oxide crystals, very well dispersed in the glaze (Table1). Tin contents of the first subgroup are 5–7% SnO2

(high enough for providing a well opacified whiteglaze) but higher in the second group (9–14% SnO2).

Another characteristic of glazes from this workshopis the presence of big quartz inclusions in all of the tinglazes (see Figure 2(c)–(d)), sometimes as large as thewhole glaze thickness, and which are probably due tothe way of preparing these glazes (less finally groundedsand, for example). These inclusions also contribute tothe opacification of glazes, so they provided a tinsaving in the final recipe. The presence of quartz inglazes as an opacifier has been also reported in Islamicglazes from Near East and Mesopotamia (Kleinmann,1986; Perez-Arantegui et al., 1996; Mason & Tite,1997). Most of the glaze samples showed strong weath-ering, with development of lead phosphate and leadcarbonate as weathering by-products and a loss of leadby leaching.

Green and brown designs were drawn on the whiteglaze with copper and manganese oxides as colourant.The only difference between brown and white zones ofthe glaze is the manganese (2·7% MnO) dissolved in theglaze. For the green, the only chemical difference is theaddition of copper (1·8% CuO) as colouring elementand a slightly higher proportion of aluminium(Al2O3�2%). This characteristic suggests that thepreparation of green was different from that of whiteglaze, with a small contribution of clay because eithercomponents were not as pure as for the white glaze orclay could help to a better adherence of the design onthe raw glaze surface.

Islamic tin-glazed objects always had the secondaryside (exterior in plates and interior in jars) also coatedwith glaze, usually honey or yellow glaze, but some-times we found some examples coloured in whiteamong objects from Zaragoza. The microstructure ofthe secondary white glazes is similar to that of themain-side. Nevertheless secondary white glazes show adifferent chemical composition (see Table 2 and Figure3), silicon content is lower (�39% SiO2) and thepresence of 2·6% Al2O3 and 1·5% Fe2O3 suggests theuse of some clay. These differences prove that a distinctrecipe was used for secondary and main sides. Prob-ably the components were less carefully chosen or theywere of a lower quality.

All of the white-glazed objects from Zaragoza havecalcareous clay bodies. A chemical and petrologicalstudy (Lapuente & Perez-Arantegui, 1999) proved thattin-glazed bodies are a group of vessels which requiredhigh temperature firing conditions: there are noremains of clay minerals (illite is totally destroyed at950�C), complete decomposition of calcite, and thedevelopment of mineral phases such as diopside,esseneite, gehlenite, anortite, etc. They exhibit twodifferent textural distributions: a group formed bysamples with high amounts of inclusions, suggesting atemper addition; and a second group with a finestfraction of inclusions. A summary of the main chemi-cal compositions (analysis by ICP-AES) of ceramicbodies is given in Table 3).

338 J. Molera et al.

MallorcaMallorca samples correspond to jars with greentin-glaze on the outer part and white tin glaze on theinner part. The chemical analyses of glazes show thatPbO content was about 54%, SiO2 about 31%, SnO2ranges between 7–9% and Al2O3 was relatively high(3·2%) as compared with the rest of productions. Thealkaline content is low (1·2% K2O and 1% Na2O)corresponding to impurities of raw materials and/orchemical diffusion from the body paste. The greenglazes are coloured by about 1·6% CuO.

The thickness of the glaze is around 150 m. Theinterface between body and glaze is very thin (about5 microns) suggesting a double firing process forthe glaze (Figure 2(e)). Cassiterite crystals are smallerthan 1 m and they are homogeneously distributedin the glaze. There are few inclusions in the glaze andvery few lead feldspars formed in the interface body-glaze. The glazes look shiny and well preservedbut under the SEM they showed some signals ofweathering.

DeniaMost of the white glazes from Denia were quiteweathered having lost their glassy shine. They show ahigh content of PbO (56%) and low SiO2 (31%) (seeFigure 3 and Table 2), while tin ranges between 6–7%SnO2. The low contents of Al2O3 (<1%) and CaO(<1%) indicate a double firing, in agreement with thelittle interaction between body and glaze. The doublefiring is also documented by the large amounts ofbiscuited pieces found in the workshop (Gisbert, Azuar& Burguera, 1991). Tin crystals are between 300–900 nm, and well distributed in the glaze as is shown inFigure 2(f). The ceramic bodies were Ca-rich (CaOaround 19%) and made with refined clays with someinclusions of quartz and metamorphic rocks as temper(Table 3).

GranadaSeven samples of Nasrid ceramics found in excavationsat Palma de Mallorca were studied for this paper.These ceramics are damaged by burial conditions (sep-tic pit) and they correspond to Granada productionsfrom the 14th century, decorated in lustre and blue.

The tin-glaze composition has a relatively low leadcontent (PbO �44%), 41% of SiO2, 7–9% of SnO2, lessthan 3% of alkalis and less than 1% of Al2O3 (Table 2).The tin oxide crystal size ranges between 300 and800 nm and its distribution is very heterogeneous,grouped in clusters (see Table 1 and Figure 2(g),(h)).This distribution would indicate a non-fritting processfor making the glaze or at least, a less accuratetechnique of preparation. There are many inclusions,quartz and feldspars grains and bubbles, which indi-cate an incomplete melting. The interaction betweenbody and glaze is very low, indicating a double firing

process. Ceramic bodies have medium calcium content(13·5% of CaO, see Table 3).

The characteristics of these southern 14th centurysamples are different from the preceding ones but arein good correlation with contemporary Hispano-Moresque (mudejar) samples of Teruel (Perez-Arantegui & Castillo, 1997) and Paterna (Molera,1996).

CordobaThree southern samples from Cordoba (two fromMadınat al-Zahra’ and one from the city of Cordoba)were studied. They are white-glazed ceramics withgreen and brown decoration and dated to the end ofthe 10th century. As compared with other productionsshown here, glazes have a microstructure similar tothose samples from Murcia and Zaragoza, but they arethicker (200–250 m) and the quartz inclusions (whenthey exist) are smaller. Chemical compositions of thesetin-glazes (see Table 2 and Figure 3) show a distinctPbO/SiO2 rate, a lower proportion of potassium (1·2%K2O) and higher aluminium content (2·2% Al2O3). Inone of the samples, the secondary side was coated withwhite glaze, but it also had a different compositionfrom secondary white-glazes of Zaragoza pottery, inthis case, it was very simiolar to main white glazesin southern samples.

The characteristics of these early southern samplesproved that tin glazes were made since the beginning ofthe tradition of white glazes in Muslim Spain (10thcentury). These tin glazes had some common featuresin all areas of al-Andalus, but there are some particu-larities in each zone, and even potters did not use thesame procedure if they made a main decorated whiteglaze or if they prepared a glaze for the secondaryside.

Comparison of the results of tin glazes from differentareasThe scatter plot SiO2–PbO (Figure 3) shows the differ-ences among productions. The lead-richest tin glazesare from Denia, Murcia and Mallorca, while Saragossatin glazes have a lower lead content (but more alkalinecomponents). According to chemical analyses of tin-opacified high-lead glazes published by several authors(Kleinmann, 1986; Mason & Tite, 1994; Mason & Tite,1997), Islamic tin glazes produced in al-Andalus showsimilar compositions to estern glazes. Samples fromFustat, 10th–11th century (Egypt), Baghdad, 9th–10thcentury, and Samarra, 9th century (Iraq) also havesilicon, lead and tin as main glaze components (25–49% PbO, 39–59% SiO2 and 5–15% SnO2) and smallquantities of alkalis.

From archaeological evidence and from analysescarried out by these authors (forthcoming paper), tinIslamic glazes from 10th to 13th century were pre-viously prepared by fritting the raw materials before

Characterization of Tin Glazes from Eastern Spain 339

applying them to the ceramic body. Frits have beenfound in the workshops of Murcia (10th century) andDenia (13th century)—analyses to appear in a forth-coming paper—thus, we may also suppose their use inother workshops between the 10th and 13th centuries,even if any archaeological find supports this hypoth-esis. The fact that samples of the 14th century show thetin oxide crystals with a heterogeneous distribution (tinremains in clusters), would suggest the possibility thatthey are remains of the original grains (raw material) oftin oxide. The complete explanation, if this indicates anon-fritting process of raw materials, is still open andthe authors are working in that direction. This techno-logical change would be in agreement with the mudejarcontemporary tin glazes (Molera, 1996), which werefritted during the 13th century and not fritted in the14th and 15th centuries.

ConclusionsFrom the experimental results shown here after thestudy of samples from a wide range of differentarchaeological sites (workshops and production areas),several points can be established concerning theIslamic opaque-glazes production:

(i) All the Islamic white glazes produced in Spain arelead–silica glaze opacified with tin oxide, at leastsince the 10th century. Accordingly, they do nothave a slip, as it has been largely referred. Neitheralkaline nor mixed lead–alkaline glazes have beendetected in any site, which is coherent with thefact that the Iberian Peninsula was one of themain galena producers at this epoch. Thus, thereare not any reasons to make alkaline glazes in aregion where lead is so abundant.

(ii) Slightly different recipes were used to producetin-glazes in each production centre, with differentlead/silica ratios. A very low contribution of clay(low aluminium contents) was used in almost allof the white glazes. It seems that very purecomponents were taken for preparing glazes (ironpercentage is always very low) and perhaps ap-plied to the ceramic body with some kind of gum.

(iii) The tin oxide contents ranges from 5 to 10%,which are the typical limits used traditionally forthese glazes. Glazes containing less than 5% arenot opaque enough and the colour of the ceramicpaste would be seen through the glaze and wouldgive a pinkish appearance, while contents over10% are not necessary and we must keep in mindthe high price of this component.

(iv) Copper and manganese were used for colouringthe green and brown designs on the white glaze.

(v) The thickness of tin glazes ranges between 150–200 m thinner than Islamic glazes from the Nearand Middle East.

(vi) In all of the cases, tin glazes were applied on apreviously biscuited body and then fired again. As

a result, a thin interface of reaction between bodyand glaze was developed.

(vii) Furthermore, tin glazes of the 10–13th centurieswere previously prepared as frits and then appliedto the ceramic body. Meanwhile, the tin glazes of14th century would be applied without friting rawmaterials.

(viii)A lower lead proportion in Islamic glazes(Zaragoza), as compared with the other sites, iscompensated for a constant quantity of potassium(about 3% K2O). Ground quartz was alwaysadded in the frit, but this glaze preparation stagewas distinctly done in other Islamic glazes than inZaragoza Islamic glazes, which seem to containbigger grains than glazes from other sites.

(ix) The ceramic paste is in all of the cases and sites acalcium-rich paste, which is coherent with twofacts: (i) the development of calcium silicatesincorporates iron in their structure and thus, alow development of iron oxides produces a buff-colour paste and less tin is necessary to opacifyglaze than for a red paste (the colour of the pasteis not seen through the glaze) (Vendrell et al.,2000); and (ii) the expansion coefficient of Ca-richpastes is similar to that of the lead glazes, avoid-ing cracks during heating and cooling (Tite et al.,1998).

AcknowledgementsWe are very grateful to the Research Laboratory forArchaeology and the History of Art of the OxfordUniversity for its collaboration. A grant of the‘‘Europa’’ Program of the CAI-CONSI+D (Diputa-cion General de Aragon, Spain) awarded to J. Perez-Arantegui helped support some aspects of this work.Our thanks are given also to Dr Antonio VallejoTriano, Dr Francisco Godoy Delgado, D. FranciscoEscudero, Dr Guillem Rossello-Bordoy, Dr JosepGisbert and Dr Julio Navarro-Palazon for the supplyof archaeological samples (Medina al-Zahra, Cordoba,Zaragoza, Mallorca, Denia and Murcia, respectively).

ReferencesAguarod, M. C. & Escudero, F. (1991). La industria alfarera del

barrio de San Pablo (siglos I–XIII). In Zaragoza: Prehistoria yArqueologıa. Zaragoza: Ayuntamiento de Zaragoza, pp. 44.

Aguarod, M. C., Escudero, F., Galve, M. P. & Mostalac, A. (1991).Nuevas perspectivas de la arqueologıa medieval urbana del peri-odo andalusı: la ciudad de Zaragoza (1984–1991). Aragon en laEdad Media IX (Universidad de Zaragoza), 445–491.

Allan, J. W. (1971). Medieval Middle Eastery Pottery. Oxford:University of Oxford.

Allan, J. W. (1991). Islamic Ceramics. Oxford: University of Oxford,pp. 6.

Arie, R. (1984). Espana musulmana (Siglos VIII–XV). In (M. Tunonde Lara, Coord.) Historia de Espana. Vol. III. Barcelona: Labor.

Caiger-Smith, A. (1985). Luster Pottery. Technique, Traditionand Innovation in Islam and the Western World. New York:New Amsterdam Books, pp. 84–99.

340 J. Molera et al.

Constable, O. R. (1994). Trade & Traders in Muslim Spain.Cambridge: Cambridge University Press.

Gisbert, J. A. (1990). Los hornos del alfar islamico de la Av.Montgo/Calle Teulada, casco urbano de Denia (Alicante). InFours de potiers et ‘testares’ medievaux en Mediterranee Occiden-tale. Madrid: Publicaciones de la Casa de Velazquez, SerieArcheologique XII, pp. 75–91.

Gisbert, J. A., Azuar, R. & Burguera, V. (1991). La produccionceramica en Daniya. El alfar islamico de la Av. Montgo/CalleTeulada (Denia-Alicante). In Actas del IV Congreso A CeramicaMedeival do Mediterraneo Occidental, Portugal, 1987. Mertola, pp.247–262.

Kaczmarczyk, A. & Hedges, R. E. M. (1983). Ancient EgyptianFaience. Warminster: Aris & Phillips, Ltd.

Kleinmann, B. (1986). History and development of Early Islamicpottery glazes. In (J. S. Olin & M. J. Blackman, Eds) Proceedingsof the 24th International Archaeometry Symposium, 1984.Washington: Smithsonian Institution Press, pp. 73–84.

Kreimeyer, R. (1987). Some notes on the firing colour of clay bricks.Applied Clay Science 2, 175–183.

Lapuente, M. P. & Perez-Arantegui, J. (1999). Characterisation andtechnology from studies of clay bodies of local Islamic productionin Zaragoza (Spain). Journal of the European Ceramic Society 19,1835–1846.

Mason, R. B. & Tite, M. S. (1994). The beginnings of Islamicstonepaste technology. Archaeometry 36, 77–91.

Mason, R. B. & Tite, M. S. (1997). The beginnings of tin-opacification of pottery glazes. Archaeometry 39(1), 41–58.

Molera, J. (1996). Evolucio mineralogica i interaccio de les pastescalciques amb els vidrats de plom: Interaccions arqueometriques.Doctoral Thesis. University of Barcelona (unpublished).

Molera, J., Pradell, T., Salvado, N. & Vendrell-Saz, M. (1999).Evidence of tin oxide recrystallization in opacified lead glazes.J. Am. Ceram. Soc. 82, 2871–2875.

Mostalac, A. (1995). Les four islamiques de Saragosse. In Le vert &le brun, de Kairouan a Avignon, ceramiques du Xe au XVe siecle.Marseille: Reunion des musees nationaux, pp. 31–32.

Munoz, P. (1993). Nuevos datos sobre urbanismo y alfarerıamedieval en Murcia. Verdolay 4, 175–184.

Navarro Palazon, J. (1990). Los materiales islamicos del alfarantiguo de San Nicolas de Murcia. In Fours de potiers et ‘‘testares’’medievaux en Mediterranee occidentale. Madrid: Publicaciones dela Casa Velazquez, Serie Archeologique XIII, pp. 29–43.

Perez-Arantegui, J., Querre, G., Eveno, M. & Kaczmarczyk, A.(1996). Chemical, SEM and petrographic study of Early Islamicglazed ceramics from several specific sites in Syria, Iraq and Iran.In (S. Demirci, A. M. O} zer & G. D. Summers, Eds) Archaeometry94, Proceedings of the 29th International Symposium on Archaeom-etry. Ankara: TU} BITAK, pp. 219–226.

Perez-Arantegui, J. (1997). Les glacures et les premiers emaux surla ceramique islamique en al-Andalus (Espagne). TECHNE 6,21–24.

Perez-Arantegui, J. & Castillo, J. R. (1997). Islamic tin glazedceramics in Spain and their later evolution in Christian areas.Ceramic Technology and Production Conference, London, 20–22November 1997, abstracts books. London: British Museum (un-published).

Perez-Arantegui, J. & Castillo, J. R. (in press). Chemical characteris-ation of clear lead glazes on Islamic ceramics, produced inNorthern al-Andalus (Muslim Spain), Proceedings of the 31stInternational Symposium on Archaeometry, 1998, Budapest(Hungary), BAR.

Perez-Arantegui, J., Soto, M. & Castillo, J. R. (1999). Examinationof the ‘‘cuerda seca’’ decoration technique on Islamic ceramicsfrom al-Andalus (Spain). Journal of Archaeological Science 26,935–941.

Reilly, B. F. (1993). The Medieval Spains. Cambridge: CambridgeUniversity Press, pp. 51–128.

Retuerce, M. & Zozoya, J. (1986). Variantes geograficas de laceramica omeya andalusı: los temas decorativos. In La ceramicamedievale nel Mediterraneo Occidentale. Firenze: Edizioniall’Insegna del Giglio, pp. 69–128.

Rossello, G. (1992). La ceramica en al-Andalus. In (J. D. Dodds,Ed.) Al-Andalus. Las Artes islamicas en Espana. New York: TheMetropolitan Museum of Art, pp. 97–103.

Rossello, G. (1995). La ceramique verte et brune en al-Andalus duXe au XIIIe siecle. In Le vert & le brun, de Kairouan a Avignon,ceramiques du Xe au XVe siecle. Marseille: Reunion des museesnationaux, pp. 105–117.

Tite, M. S., Freestone, I., Mason, R., Molera, J., Vendrell-Saz, M. &Wood, N. (1998). Lead glazes in antiquity. Methods of productionand reasons for use. Archaeometry 40, 241–260.

Vendrell, M., Molera, J. & Tite, M. S. (2000). Optical properties oftin opacified glazes. Archaeometry 42, 325–340.