Results
Optical microscopy
Using OM, four main types of ceramic paste were identified for the Păcuiul lui Soare shards. Their description is given in the lines below; their assignation to a certain type of paste is summarized in Table 1.
-
I.
Fine paste with silty clay matrix, oriented microstructure and micro-fissures: P15, P16, P17, P18, P19, P20, P21, P22, P23, P24, P25, P26, P27, P28, P29, P30, P42.
Fine, homogeneous paste with fine silty clay matrix with rare (5–10%) silt grains, 10–50 μm, and more frequent (10–15%) fine sand grains, 100–250 μm, rarely medium sand, 300–500 μm, and accidental coarse grains, 1.0–1.4 mm; mono- and polycrystalline (for coarse sand fraction) quartz and feldspar, frequent amorphous ferruginous stains and concretions, 10–30 μm, and rarely (1–2%) much larger, i.e. 100–300 μm. Low porosity (5–15%), with fine isolated pores, 100–300 μm, and more frequent micro-fissures, 0.5–3.0 mm. The clay is birefringent, the sand grains are sub-rounded and rounded, being intentionally added to the silty clay matrix. This paste shows a good to moderate sorting; mica flakes are absent (Fig. 5a). These vessels were fired in oxidizing conditions, complete or incomplete (“banded” structure). In some cases, secondary firing is present, as some of these shards were fragments of cooking pots or because some fragments were discovered in burned dwellings.
-
II.
Fine silty paste with porphyritic microstructure: P5, P8, P9, P10, P11, P12, P13, P14, P31, P32, P41, P43, P44, P45.
Fine, homogeneous paste with fine grained silty matrix, well sorted, with frequent (15–25%) grains of fine sand; mono- and polycrystalline quartz and feldspar, 50–100 μm, rarely 150–200 μm and frequent mica (predominant muscovite), 50–200 μm, frequent fine vegetal debris, 20–150 μm, ferruginous stains and concretions, 50–150 μm, rare fine carbonate grains, 100–400 μm, and accidental shell fragments. Fine porosity (5–10%), with frequent isolated and rounded voids, 50–200 μm, very rarely up to 1 mm. Rare elongated voids originating from vegetal temper are present. This paste was most probably made from alluvial sediments, well sorted silt with fine sand and clay; sand grains are sub-rounded and rounded (Fig. 5b). These shards were fired in complete or incomplete (with dark organic “core”) oxidizing conditions.
-
III.
Fine carbonate paste with porphyritic microstructure: P1, P2, P3, P4, P6, P7, P36, P39.
Fine, homogeneous paste with fine carbonate matrix, with frequent (10–15%) fine silty grains, 10–50 μm, rare fine sand, 50–200 μm, and very rare (2–3%) medium to coarse sand grains, 400–750 μm; sand fraction with mono- and polycrystalline quartz and feldspars, rare (5%) fine flakes of mica, 50–150 μm, rare calcite grains, 100–200 μm and opaque grains, 50–100 μm (Fig. 5c). This paste includes areas of calcite recrystallization on voids and fine cracks. Low porosity (5–10%), with fine circular voids, 50–300 μm, and rare mm-size micro-fissures are visible. This paste was made from carbonated sediments (unconsolidated marl type), featuring a well-sorted matrix, with possible mixture of fine sand with medium to coarse sand grains, that are generally sub-rounded and rounded. The shards were fired in oxidizing conditions, featuring a banded structure and an organic “core”.
-
IV.
Medium granular paste with porphyritic microstructure: P33, P34, P35, P37, P38, P40.
Semi-fine, homogeneous paste, with silty clay with fine sand matrix, moderately to poorly sorted; rare silty grains (5%), 20–50 μm, and frequent very fine sand (10–20%), 50–200 μm. It includes rare coarse sand grains (1–5%), 500–1500 μm. This paste is similar with the fine silty paste (type II), with frequent fine mica flakes, 50–200 μm, but with frequent quartz and quartzite sand inclusions, with heterogeneous appearance (Fig. 5d). Fine porosity, 10–15%, isolated voids, 50–200 μm, circular and elongated, irregular, and micro-fissures, 100–300 μm, rarely 0.5–1.5 mm. The sand grains are sub-angular, sub-rounded and rounded. These shards were fired in reducing conditions, with an organic “core”, being oxidized at the exterior.
The four main types of ceramic paste identified for the ceramic fragments from Păcuiul lui Soare are similar to the ones evidenced for the coeval potteries discovered in the nearby archaeological sites Hârșova and Oltina that were previously analysed in the frame of the on-going project mentioned in “Introduction” section [7, 8].
The fine paste with silty clay matrix (type I) is made from a mixture of sediments, silty clay and fine sand, with oriented and birefringent clay, possibly kaolinitic, such as the one identified in the two aforementioned sites.
The silty paste with porphyritic structure (type II) is made of fine alluvial sediments. As in the other two archaeological sites studied until now (Hârşova and Oltina) this is indicated by the good sorting of the sedimentary matrix, the presence of the vegetal debris (partially decayed fragments of wood), and accidentally, of the shell fragments.
The fine carbonate paste (type III) is most probably made of fine alluvial sediments deposited in low energy sedimentary environments, such as marshy areas or lakes.
The medium granular paste (type IV) is made of silty fine sands with clay, very probably alluvial, mixed with coarse alluvial sands with variable granulometry.
PIXE
To get a clear picture about the possible grouping of the ceramic shards according to their composition, taking into account the relatively large number of samples and variables, PCA (Principal Components Analysis) of the standardized PIXE data—Mg, Al, K, Ca, Ti, Cr, Fe, Ni, Cu, Zn, Ga, Rb, Sr and Zr oxides—was performed using STATISTICA software (version 8.0). Figure 6 shows the result of this analysis indicating the separation of the analyzed samples into two main compositional groups. A column in Table 1 also indicates to which of these two groups each sample belongs.
One of these groups (conventionally indicated with “I” in Fig. 6) is made out of 18 samples with relatively high Al2O3 (~ 24.9 mass% on average) and low CaO (~ 1.0 mass% on average) contents. This first group separates quite well from the rest of the other 27 ceramic fragments that cluster in a second group (marked with “II” in the same figure), regardless of their characteristics (decoration, granulation, firing etc.). This second group is characterized by relatively low Al2O3 (~ 17.7 mass% on average) and high CaO (~ 4.0 mass% on average) concentrations. As also visible from the loading plots shown in Fig. 6, the higher contents of Al are correlated with higher concentrations of Ti (group I), while the higher contents of Ca are correlated with higher concentrations of Mg, K and Sr (group II).
The first compositional group superposes relatively well with type I resulting from the petrographic analysis, except for sample P14 (containing 23.2 mass% Al2O3), attributed to type II from OM and sample P38 (with 24.6 mass% Al2O3) that belongs to type IV resulting from OM. Other discrepancy appears in the case of sample P42 (18.8 mass% Al2O3) that belongs to type I resulting from OM, and, at the same time, pertains to compositional cluster II, the one made out of the low-Al samples. Explanations for these discrepancies are still sought, but in any case, there seems to be a pattern that can be recognized from the analysis of the data obtained using these two different analytical techniques. Thus, paste I resulting from OM superposes quite well with the samples attributed to group I resulting from PCA; more or less, this reflects the use of a distinct type of clay that was chosen on purpose—most likely, to obtain vessels with different properties. On the other hand, the samples belonging to group II resulting from the PCA of PIXE data correspond to three types of paste (II, III and IV) revealed by OM. The most plausible explanation for this finding is that the shards belonging to these groups contain more or less the same minerals/compounds, hence the compositional similarity of the pellets (see also PIXE section). The differences between types II, III and IV of paste mainly consist in textural parameters, homogeneity and porosity, etc., and do not result from geochemistry.
A similar sub-division of potteries into two large groups, mainly separated by their Al and Ca contents, was also evidenced in the case of previously analyzed coeval ceramics from Hârșova and Oltina [7, 8].
Micro-PIXE scans of the interfaces between the decorated surfaces and the ceramic bodies were performed to identify the compound(s) responsible for the green glaze, as well as to estimate the thickness of these layers. A clear change in the chemical composition were observed for this type of decoration characterized by a large lead content (~ 64.4 mass% PbO on average, PbO content ranging from 50.4 to 76.8 mass%) compared to the ceramic body (~ 0.1 mass% PbO on average)—see Fig. 7 and Tables 3 and 4.
There are two ways of producing a Pb glazing: either by applying a Pb compound (e.g. litharge) as a suspension in water or by applying a mixture of PbO and quartz. In the former case, the glaze forms through direct reaction of PbO with the clay body; in the latter, the two components of the suspension react during firing, form a glaze that interacts with the body, and certain chemical elements (Al, Ca, Fe, Na, Mg) diffuse from the body into the glaze. These two glazing methods can be distinguished by subtracting the PbO content from the glaze composition and renormalizing the recipe to 100% and subsequently comparing this adjusted glaze composition with the one of the body [25,26,27]. Figure 8 shows that the Al2O3 content in the recast glaze composition is roughly the same as in the corresponding ceramic body, indicating that glazing was made through the application of lead oxide onto a non-calcareous ceramic body.
The micro-PIXE Pb Lα maps were used to estimate the thickness of the green glaze layer that turned out to range from 50 μm up to 300 μm. Some shards feature decorative layers with variable thickness.
Relatively small amounts of CuO (~ 1270 ppm on average) were found in the green glaze—see Table 4. The most likely explanation for the olive-green hues of the glaze is that they were induced by iron ions in reducing state [3, 4, 26, 28]. It must be mentioned here that certain amounts of Sb were determined in the composition of the green glaze covering samples P1 and P2—see Table 4. However, these particular two samples do not exhibit yellow hues, color that might have resulted from the presence of lead antimonate [3]; they are also olive-green, similar to the other five glazed samples analyzed in this study. The relatively small amounts of Sb from these two particular samples did not lead to any change in the appearance/coloring of their decorations. A possible explanation for the presence of Sb in the glaze is that this chemical element is a contaminant of the raw materials employed for making the glaze (in particular, of the lead compound).
The micro-PIXE scans of the interfaces did not provide any hint about what causes the shining of the golden engobe. However, the explanation was provided by OM: fine flakes of muscovite embedded in the slip produce the glittering golden shine of these shards. This decoration is the result of a different preparation manner and it did not imply the use of other raw materials.