**3. Results and Discussion**

#### *3.1. Body Compositions*

Chemical compositional data of ceramic bodies obtained by ICP-OES were treated statistically by hierarchical clustering analysis using eight variables (Na, Mg, Al, K, Ca, Ti, Mn, and Fe) and Euclidean distances between samples (Ward method) to group them. This multivariate statistical treatment (Figure 3) revealed several differentiated groups whose compositions are summarized in Table 1 (oxides were calculated by stoichiometry). Calcium was one of the elements whose content varied between groups and, together with other components, allowed the characterization of each group. The groups presented calcareous matrix whose composition ranged from 5 to 18% CaO. Group 3 presented lower average calcium proportion (7.5% CaO) than the other calcareous matrixes, but potassium, magnesium, and aluminum contents were higher. The other two groups with calcareous bodies (group 1 and 2) had similar calcium quantities, but different minor element contents. All studied decorated ceramics were produced with calcareous clay, only a less frequent group of slip-ware decorated with red/brown painted designs (not included in this paper) was made with non-calcareous clay (0.5–1% CaO) [10]. The use of calcareous clays for manufacturing glazed pottery was expected in order to obtain light-colored bodies [11].

The reference groups obtained by cluster analysis, according to the chemical composition of the bodies, could be related to the decoration types manufactured in Albarracin. Groups 1 and 2 corresponded to ceramics decorated with white or green glazes. Most of the *cuerda-seca* fragments were included inside group 1. Group 3 was mainly formed by honey-glazed objects, although some yellow fragments were also included in groups 1 and 2. This fact suggested that yellow-glazed ceramics could be manufactured with different raw materials or in different workshops.

**Figure 3.** Dendrogram from hierarchical clustering analysis of the body composition data (CS: *cuerda seca*; GB: green and brown decoration on white glaze; G: green glaze; H: yellow/honey glaze).


**Table 1.** Chemical composition of ceramic bodies, determined by ICP-OES (data as %wt, except Ba and Sr results given as μg.g<sup>−</sup>1, *s*: standard deviation; *n*: number of samples).

> According to chemical results, it seems that selection of the raw materials was performed depending on the decoration. In groups 1 and 2, it can be assumed that the same type of illitic clay was employed for the manufacture of the pottery. However, the raw materials used in the ceramic production of group 3 differed in the chemical composition from those employed for groups 1 and 2. Therefore, ceramics included in this last group can belong to another production workshop or even another area.

> Comparing the results obtained from Albarracin ceramics with analysis of clay bodies of Islamic productions from other areas of the Iberian Peninsula [4,12–16], it can be pointed out that all the clays used in decorated ceramics were also Ca-rich (8–22% CaO) and had similar contents of iron (4–6% Fe2O3) and fluxing agents (4–5% Na2O+K2O). The use of calcareous clays for decorated pottery justified the buff colors of the bodies due to the growth of calcium silicates, which incorporated some iron atoms to their structure and decreased the content of iron oxides, responsible for red color [11]. A great similarity was found when we compared these objects from Albarracin with other pottery productions from Muslim centers, such as Cordoba, Denia, Granada, Mallorca, Murcia, Pechina, or Zaragoza [4,13–16]. For instance, the Islamic production from Zaragoza during the 11–12th centuries CE was studied [13] and four reference groups were described, based on the type of body clays employed. In these workshops from Zaragoza, differences among the calcareous groups were also based on the amounts of alkalis and the groups were related to the final decoration and to the utility of the pottery. The selection of the raw materials implied great knowledge of their characteristics and properties.

> In the south of al-Andalus, the production of ceramics was very popular with a great number of workshops. Cordoba, Denia, Granada, Murcia, and Pechina were important pottery centers during the 11th–13th centuries CE [4,14,16]. The body composition of these glazed productions from Islamic workshops was very similar to ceramics from Albarracin with high contents of calcium (groups 1 and 2), although ceramics manufactured in some of them were richer in calcium content than Albarracin. The study carried out in the workshop of San Nicolas [14], situated in Murcia and dating to the 10th century CE, revealed some differences with the pastes from Albarracin, but three different references groups can be also distinguished; these three groups were also related to the decoration to be applied. Again, this fact demonstrated the specialization of the potters on the selection of the material.

#### *3.2. White Tin-Opacified Glazes*

Different types of decoration on the Islamic pottery from Albarracin *Taifa*: white tin-opacified glazes, colored glazes, and *cuerda seca* were studied for their characterization. In the case of white tin-opacified glazed samples, the compositions of white glazes on the main side and the yellow-honey glazes on the secondary side of these fragments were studied. Further, brown and green decorations on the white glazes were analyzed in most ceramic samples. Tin-opacified glazes presented an average chemical composition that was very homogeneous in the main side, as can be seen in Table 2. The major elements found in the white glazes were silicon (SiO2 42.8%), lead (PbO 36.2%), and tin (SnO2 7.0%), mixed with contents below 5% of aluminum, iron, alkalis (potassium and sodium) and calcium oxides. Glazes produced during the medieval period in the Iberian Peninsula were characterized by lead-rich contents. The low percentage of aluminum (Al2O3 < 4.5%)

could be attributed to the use of small amounts of clay to produce the glaze, or also to the diffusion of aluminum from the paste to the glaze during firing [17]. Iron and calcium were not added to the recipes of glazes as main components. Their very low proportions (FeO < 1% and CaO 2.86%) were also due to the migration of iron and calcium into the glazes. Higher contents of iron would induce yellow or brown colors in the glaze. Therefore, raw materials employed in the production of tin-opacified glazes were carefully chosen in order to obtain a white color of glaze.


**Table 2.** Chemical composition (%wt) of white tin-opacified glazes and their colored decoration by EDS.

Green and brown decorations on tin-glazed samples from Albarracin *Taifa* did not differ in their compositions from the white ones, except for the presence of metallic oxides responsible for the colors (Table 2). Analyses carried out on green areas were rich in copper oxide, and brown zones had higher contents of manganese oxide. The content of CuO was in the range of 1.4–3.4% and MnO varied between 1.2 and 6.0%. The presence of copper in brown decorations can be explained due to the thin brown design that made it difficult to determinate only the composition of brown areas without including green zones.

Opacity was accomplished by the presence of small crystals of cassiterite (SnO2) within the glaze that reflected and scattered the light [18]. These crystals were achieved by the addition of tin oxide to the raw materials of the glazes. During the heating (above 700 ◦C) of the mixture, the tin oxide was dissolved and began to recrystallize as cassiterite [19]. The amount of tin, the size of tin oxide crystals, their distribution in the glaze, and the thickness of the glaze were factors with great influence on the grade of opacification [12,18,19] and they were related to the process of manufacture. The average amount of tin oxide (7% SnO2) employed in the studied samples was in good agreement to obtain enough opacity (5–10% SnO2) without increasing too much the price of the product. The size of tin oxide crystals in Albarracin samples varied from 500 to 800 nm. These dimensions of the crystals indicated that high temperatures (above 800 ◦C) were reached during the firing process and the cooling occurred fast, as it was demonstrated in the literature [19]. The homogeneous distribution of the cassiterite crystals suggested that the tin glazes were prepared by fritting the raw materials before applying them to the ceramic body [15,20]. Fritting was an extended method used since ancient times to obtain glasses and glazes, which consisted in melting the raw materials (lead, tin, and silica) previously to use them as components of the glazes. Finally, these fired materials were ground to powder before preparing the glaze suspension that was applied on the surface of the body. Besides the homogeneity of the fritted material, frits presented some advantages with respect to raw materials: they reduced the risk of contraction, avoiding cracking the surface, improved the maturation of the glaze, avoided bubbles, and allowed better control of the process [19].

Only one tin-opacified glazed sample had also white glaze on the secondary side; the others had honey glazes on the secondary side, as it has been explained before. The composition of these secondary glazes revealed a higher heterogeneity between specimens (Table 3), but it was possibly to classify them in two different subgroups. The first one included samples with contents of SiO2 and PbO about 37% and 46% respectively, and the other group was composed by glazes with higher SiO2 (45%) and lower PbO (37%) proportions. In both cases, the elements added as fluxes were lead and some alkalis with contents around 3–4%. Tin oxide was not present and the proportions of aluminum and iron oxides were increased regarding main tin-opacified glazes, which evidenced the use of

clays and sand not as pure as those used to manufacture white glazes. The honey/yellowish color of the glazes was achieved by iron (>2% FeO) dissolved in the glaze.



Microstructure of glazes observed by SEM differed depending on the fragment side (Figure 4). Tin-opacified glazes on the main side of the objects (Figure 4a) had few inclusions, while on the secondary side (transparent glazes) were abundant inclusions with large sizes (often bigger than 60 μm) and well distributed through the glaze (Figure 4b,c). In both cases, the inclusion compositions were the same: potassium feldspars and quartz. Those microstructural differences reflected that the materials for the main decoration were carefully chosen and probably fritted for producing the main glazes [14].

**Figure 4.** Tin-opacified glazed ceramics: (**a**) Secondary electron-SEM image of the main-side glaze with cassiterite crystals scattered in the glaze; (**b**) Secondary electron-SEM image of the secondary side with non-plastic inclusions; (**c**) Optical Microscopy image of the secondary-side glaze.

Concerning thickness, it is outstanding that glazes of the main side were thicker than the coatings applied on the secondary side in all studied samples. On average, the thickness of the layers ranged from 100 to 200 μm for tin-opacified glazes, and from 50 to 100 μm for transparent glazes.

In the contact zone between clay body and glaze, several reactions occurred during firing that favored the migration of atoms between both phases. Diffusion of the elements produced the growth of Pb-rich feldspars crystals. The development of these crystals in the interface and the reaction degree between clay body and glaze were related to the type of clay body and the firing/cooling process [17]. Not only in tin-opacified glazes from Albarracin but also in the transparent ones, the amount of crystals and the interaction between clay paste and glaze (interface thickness < 10 μm) was minimum. It was described that, when the glaze was applied on an already fired clay body, the diffusion between both phases was reduced, and consequently the number of crystals formed and the thickness of the interface were lower [17,21]. This feature indicated that tin-opacified glazes manufactured in the *Taifa* of Albarracin were applied on a previously fired paste (biscuited) and a second firing process was carried out to produce the decoration.

If the results obtained for tin-opacified glazes of Albarracin production are compared with other workshops of the Islamic period in the Iberian Peninsula [4,14], it can be pointed out that the technology (firing process and opacity) seemed to be very similar and the glaze recipes differed only slightly. In all Iberian production centers, white opaque glazes were composed by a mixture of lead–silica with tin oxide contents from 5 to 10%. Each workshop had a characteristic lead/silica ratio that could help to differentiate the productions. Islamic ceramic centers in the south and the SE of the Peninsula had higher contents of lead (45–60% PbO and 30–42% SiO2, in general; then, PbO/SiO2 ratios from 2 to 1.1) than Albarracin (PbO/SiO2 = 0.8) and Zaragoza samples (in the NE), except for some examples from the Vega of Granada [16]. The low lead contents of these northeastern groups were compensated by higher amounts of alkaline components (2–5% K2O); this fact allows the definition of these tin-opacified glazes as potassium-lead glazes. These differences were also revealed in the results obtained with the same methodology in our laboratory in samples from Pechina (9th–10th centuries CE), given by the Laboratoire de Céramologie (Lyon, France) (unpublished data). Those characteristics proved that white tin-opacified glazes manufactured in Albarracin and Zaragoza *Taifas* could be differentiated by their chemical composition from other production areas.
