Archaeometric Analyses of Medieval Pottery from the Lower Danube Region, Romania

Abstract

This study discusses the pottery manufacturing in the Lower Danube region during the Early Medieval period. Optical microscopy, Particle induced X‐ray emission (PIXE), and X‐ray diffraction (XRD) are performed on 32 ceramic shards unearthed at Pantelimonu de Sus, Constanţa County, Romania, dated to the 8th–10th c. AD. The petrographic observations show an important variability in terms of mineral composition, homogeneity, and porosity, documented by the presence of four types of ceramic paste, three indicating the use of alluvial clays and one suggestive for kaolinitic clays. The mineralogy of selected ceramic fragments is further refined by powder XRD. The principal component analysis of the PIXE data separates the fragments made of kaolinitic clays from the rest. The archaeometric investigations show that during the Early Middle Age, the potters from Pantelimonu de Sus use alluvial and kaolinitic clays, most likely of local or regional origin to manufacture various kind of vessels. Most potteries are fired in an oxidizing atmosphere—complete or incomplete—at temperatures ranging from 600 °C to 900 °C. The analytical data are compared to those previously obtained on coeval ceramic finds from some nearby archaeological sites; a certain degree of similarly of the results is evidenced.

This study is about the pottery manufacturing in the Lower Danube region during the Early Medieval period (8th–10th c. AD). Optical microscopy, Particle induced X‐ray emission, and X‐ray diffraction analyses performed on 32 representative ceramic shards excavated at Pantelimonu de Sus, Constanţa County, Romania, provide clues on the raw materials and craftsmen skills.

1. Introduction

In the last two decades, the archaeological excavations in southeastern Romania (historical province Dobrudja) have significantly advanced our understanding of the Early Medieval material culture, particularly through the systematic recovery and analysis of ceramic assemblages dated from the late 8th to the 11th c. AD.^{[ 1 ]} However, despite the growing corpus of ceramic material, scientific studies capable of distinguishing local from imported wares—based on technological, compositional, and functional attributes—remain insufficient. This lack of precise provenance data has limited the ability of reconstructing the trade dynamics and cultural interactions in the Lower Danube region during this period. The present study aims to address this gap by applying archaeometric techniques to ceramic fragments excavated at Pantelimonu de Sus, a site previously unexamined from this perspective.

The good quality of the paste for certain vessels (amphorae, jugs, bowls, etc.) and their sporadic presence in the archaeological records suggested their attribution to some sophisticated workshops, either from Constantinople or from other vital centers of the Byzantine Empire.^{[ 2 ]}

Several archaeometric studies of ceramic finds from the Early Medieval period discovered in archaeological sites from southeastern Europe revealed the features of the local—versus imported—potteries, pointing toward the existence of contacts between communities living at large distances, evidencing the role of commercial hubs played by certain settlements, and bringing valuable contributions on ancient ceramic provenance issues.^{[ 3} , ⁴ , ^{5 ]}

Building on this framework, a multidisciplinary research initiative was launched to systematically investigate the Early Medieval pottery manufacturing in Dobrudja, aiming to identify the raw materials, to unravel the employed technologies and to establish some criteria to distinguish the locally manufactured potteries from the supposedly imported ones.

To date, the project has yielded significant results disseminated through publications focused on ceramic materials from five key sites located along the Black Sea coast and Danube banks: Hârşova, Oltina, Păcuiul lui Soare, Castellum 22, and Constanţa, Romania.^{[ 6} , ⁷ , ⁸ , ⁹ , ^{10 ]} The present study extends the geographical scope of this research to the central part of Dobrudja, commenting on the outcomes of the archaeometric characterization of an assemblage of 32 ceramic fragments from Pantelimonu de Sus, Constanţa County (see Figure  1 ). This site is significant as it lies inland, providing a new axis of comparison for the coastal and fluvial sites previously studied. By analyzing ceramics from Pantelimonu de Sus, the study introduces new data to a growing regional database and strengthens the interpretative framework concerning the social, economic, and technological fabric of the Early Medieval communities in the Lower Danube region.

Figure 1.

Map of Romania showing the location of all archaeological sites mentioned in the text.

This impact of this research extends beyond ceramic studies, as it contributes to a broader understanding of the local production systems, trade interactions, and cultural exchange in a frontier zone of the Byzantine world. Moreover, it provides scholars with comparative datasets necessary for reevaluating the connectivity and the self‐sufficiency of rural settlements during the Early Middle Ages.

Through the application of optical microscopy (OM), Particle‐induced X‐ray emission (PIXE), and X‐ray diffraction (XRD), the project also showcases the value of interdisciplinary collaboration between archaeologists, chemists, geologists, physicists, and historians. These methods validate the archaeological hypotheses about production and trade and strengthen the analytical foundation for future investigations in southeastern Europe.

The study brings a fresh contribution to the relatively meager corpus of archaeometric publications on Early Medieval potteries from this part of Europe.^{[ 3} , ⁵ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ^{21 ]} At the same time, focusing on the analytical side, this research offers a different perspective on the pottery finds excavated in Romanian archaeological sites.

2. Experimental Section

2.1. Archaeological Site

Between 2010 and 2018, the Museum of National History and Archaeology in Constanţa conducted extensive excavations near the village of Pantelimonu de Sus, Constanţa County, Romania. Coordinated by a team of archaeologists and local specialists, these campaigns revealed a stratified occupation sequence spanning from the Roman to the Ottoman period, offering critical insight into the site's long‐term significance. This archaeological site from the continental part of Dobrudja is located 200 m south‐west from the Roman fortress Ulmetum, on a hill slope dominating the valley of Pantelimon River, a small tributary of the Casimcea River (see Figure 1).

The archaeological excavations performed east of Pantelimonu de Sus village revealed the existence of several successive layers of occupation, the oldest dated to the Roman period (2nd–4th c. AD, vicus Ulmetum). During the Early Middle Ages (the end of the 8th c. to the first half of the 10th c. AD), a rural settlement was located here, which was replaced during the 17th–18th c. AD by an Ottoman seasonal settlement.^{[ 22 ]} The Early Medieval settlement from Pantelimonu de Sus had a regional importance, as demonstrated by the fact that this zone continued to be inhabited after the abandonment of the province by the Byzantine administration in the first decades of the 7th c. AD.

On an area of roughly 800 m², the excavations uncovered seven dwellings and two areas for domestic waste. The findings indicate a modest settlement, without any fortifications, with the dwellings situated far apart from one another. The archaeological material—among which the ceramic finds prevail—indicates similarities with the discoveries from other coeval settlements from the western coast of the Black Sea, excavated in archaeological sites from the present‐day Ukraine, Romania, and Bulgaria.

The site raised particular interest among the archaeologists due to its inland position, unfortified character, and proximity to Ulmetum. The recovered ceramics—both local and possibly imported—offered a rare opportunity to investigate questions of production and distribution in a non‐urban, rural setting. The expedition team prioritized this site as a test case for identifying how peripheral communities participated in wider trade networks and technological exchange.

The discovery of distinctive ceramic types prompted new hypotheses regarding the extent of local innovation and external influence. Was the pottery production here autonomous, primarily shaped by the local resources and needs? Or was Pantelimonu de Sus part of a wider commercial circuit that extended across the western Black Sea basin? These questions guided the analytical phase of the study.

2.2. Sample Description

The present study reports on the archaeometric investigations on thirty‐two ceramic fragments from Pantelimonu de Sus (see Figure  2 ). For a detailed archaeological description of each sample, the reader is referred to Table S1, Supporting Information.

Figure 2.

Photos of the ceramic shards/vessels from Pantelimonu de Sus reported in this paper.

The samples were chosen to be representative of all types of ceramic vessels, stemming from pots without handles, pots with handle(s) of various shapes, pitchers, bowls, buckets, supply vessels, and a cauldron (see Table S1, Supporting Information and Figure 2). The selected shards originated from closed stratigraphic complexes with well‐dated contexts and layers, indicative of the end of the 8th c. and the 10th c. (see Table S1, Supporting Information).

Within the present vessel types, fragments of various types of paste were selected for the archaeometric analyses: common paste, fine, slightly sandy semi‐fine, or coarse (with coarse sand and very fine gravels), but also very fine paste with pure white clays, possibly kaolinitic (see Table S1, Supporting Information).

Based on the color of the vessel surfaces, firing took place in a reducing or in an oxidizing atmosphere (see Table S1, Supporting Information). The very fine paste vessels are generally indicative of firing in a reducing atmosphere, except for samples P31 and P32, clearly fired in an oxidizing atmosphere. The vessels made of kaolinitic clays are also suggestive of a reducing firing (see, e.g., shard P8). However, most of the samples originate from oxidizing fired vessels. Some fragments are strongly smoked, indicating their direct exposure to fire while cooking or their discovery in burned complexes.

While certain vessels have burnished surfaces, incision is the most often encountered decoration technique (see Figure 2). The bottom of vessel P30 features a pottery mark.

3. Methods

3.1. OM

To characterize the paste types of the ceramic fragments, 25 samples from Pantelimonu de Sus were analyzed at a stereomicroscope, with magnifications varying from ×10 to ×30. Having established the main categories of fabric, 10 thin sections, considered representative of the entire batch of analyzed shards, were used to detail the types of paste. The thin section characterization was performed using an Olympus BX 60 polarizing petrographic microscope, at ×50–×500 magnification. The microscopic description was based on textural characteristics (size and shape of the grains, sorting, and frequency), composition (nature of constituents), microstructure, porosity, firing, and surface treatment.^{[ 23 ]} The fabric types were also described according to the coarse/fine (c/f)‐related distribution.^{[ 24 ]}

3.2. PIXE

PIXE compositional analyses were performed at AN2000 accelerator of Laboratori Nazionali di Legnaro (LNL), Istituto Nazionale di Fisica Nucleare (INFN), Italy, with a 2 MeV proton beam, macrobeam settings (3 × 3 mm²), and a preset charge of 4 μC. The characteristic X‐rays were detected with an IGLET‐X™ HPGe detector from ORTEC covered with a 100 μm thick Al funny filter, that is, an Al foil with a small central hole with the ratio of the hole area to the detector area equal to 0.007.

Small fragments from each of the 32 ceramic shards were transformed into powders using an agate mortar and pelletized. To obtain quantitative results, the PIXE spectra were treated with GUPIXWIN software (version 2.2.4),^{[ 25 ]} considering all detected elements (Mg, Al, Si, K, Ca, Ti, Cr, Mn, Fe, Ni, Cu, Zn, Ga, Rb, Sr, Zr, and Pb) as oxides, with oxygen tied in stoichiometric ratios, and normalizing the concentrations to 100 mass%.

Basalt BHVO‐2 Certified Reference Material (CRM) was used to assess the accuracy of the quantitative results. The overall uncertainties were estimated to be ≈5% for the major elements, 10% for the minor elements, and 20% for the trace elements.

3.3. XRD

The XRD measurements were performed using a Panalytical X’Pert PRO MPD Diffractometer on seven samples. The data acquisition took place in continuous mode in Bragg–Bretano geometry using a Cu K_α X‐Ray radiation source (λ = 0.15418 nm). The working parameters were 45 kV accelerating voltage and 40 mA current intensity. The 2θ scans were taken in 15°–80° angular range, with steps of 0.02° using 10 s/step acquisition times. A divergence slit (1/2°) was used on the incident beam and a Ni filter and a curved graphitic monochromator on the diffracted beam. The aim of the XRD measurements was to determine the crystallographic phases and to corroborate the XRD results with those resulting from the mineralogical analyses executed by OM.

For structural characterization—phase identification and lattice constant determination—the measurements were assisted by the database developed by the International Center for Diffraction Data, formerly known as Joint Committee on Powder Diffraction Standards (JCPDS), and the High Score Plus software.

4. Results

4.1. OM

OM documented four types of fabric for the ceramic paste within a subset of twenty‐five shards. The petrographic observations are presented in the lines below and illustrated by Figure  3 . The detailed description of the hand specimen for each type of paste fabric is given in the Supplementary Information SI 1.

Figure 3.

OM micrographs of thin sections from several shards from Pantelimonu de Sus. Images in cross‐polarized light (XPL) or plane‐polarized light (PPL). Image width is 4 mm, except for C and G where is 2 mm. A) P10, XPL. B) P11, XPL. C) P11, XPL. D) P32, XPL. E) P13, XPL. F) P16, XPL. G) P16, XPL. H) P19, XPL. I) P23, XPL. J) P24, PPL. K) P29, XPL. L) P28, XPL.

Fabric Description

I. Fine silty paste with homogeneous fabric

I. 1. Oxidizing firing silty fabric: P1, P2, P3, P7, P9, P10, P31

Fine‐grained silty clay matrix, c/f = 20 μm, well/very well sorted, with rare to frequent (5%–25%) grains of fine sand–quartz and feldspar, 50–200 μm, rarely 150–400 μm, 50–150 μm frequent mica (predominant muscovite; Figure 3A), fine micrite and very rare sparite calcite, 50–100 μm, and rare to frequent fine ferruginous vegetal debris, 20–100 μm.

Rare inclusions: daub grains and fine sandy ceramic fragments with oxidizing firing (Figure 3A), 400–500 μm, very rarely 1–1.5 mm, and rare fine, rounded/subrounded, few angular, fine carbonate grains 300–400 μm, very rarely 1 mm.

Fine porosity (5%–15%), frequent isolated rounded voids and rare chambers (larger, irregular voids), 50–250 μm, very rarely 400 μm to 1 mm with areas of calcite recrystallization.

Complete or almost complete oxidizing firing.

I. 2. Fine silty fabric with banded structure: P4, P11, P32 (with gray, fine slip)

Paste with fine silty clay matrix, well sorted, homogeneous, and more heterogeneous in the central part, with frequent fine vegetal debris (decayed plant fragments, some including phytoliths). The texture is composed of silt with clay, with frequent (15%–25%) fine sand grains, 50–100 μm, rarely 150 μm, quartz and feldspars (few plagioclases with polysynthetic twins noted), rare mica flakes (5%), and 50–150 μm (Figure 3B). The birefringence fabric includes areas of oriented clay with muscovite.

Inclusions are represented by rare carbonate grains and grog (daub and crushed ceramic fragments, 1%–2%), 100–220 μm, sporadic up to 2 mm, and rare areas of calcitic recrystallization in voids and fine cracks (Figure 3C) and frequent mm size areas with iron oxides, especially in the central zone.

Well‐developed porosity (10%–30%), with fine circular isolated voids, 100–300 μm, microfissures 1–1.5 mm, and mm size irregular chambers in the central part. The banded structure is determined by the incomplete firing, with zones with diffuse boundaries (P32, Figure 3D).

The paste fabric is similar to I.1., without notable difference between the zones of I.2.

II. Fine clay fabric

II. 1. Oxidizing firing fine clay fabric: P13, P14, P15, P16, P17

Fine matrix of pure clay or silty clay with rare (5%–10%) silt grains, 20–50 μm, well sorted, with birefringence fabric presenting yellowish‐brown oriented clay domains (Figure 3G).

Includes rare to frequent (5%–25%) fine sand grains, 100–250 μm, rarely 300–750 μm, and very rare 1–2.5 mm, subangular and subrounded quartz and feldspar, very rare (<1%) mica flakes (Figure 3E), 50–100 μm and rare opaque grains (1%–2%), 100–150 μm (Figure 3F). Includes very rare ceramic and burnt daub fragments, 100–400 μm.

Low to medium porosity (5%–15%), with fine isolated voids, 50–100 μm, and rare microfissures 0.5–1 mm.

Complete oxidizing firing.

II. 2. Fine clay fabric with banded structure: P18, P19, P27

Fine, well sorted, oriented silty clay matrix, with pure, oriented, possibly kaolinitic clay (Figure 3H), with moderate sorting, homogeneous, with dark core. Birefringence fabric with oriented clay domains.

Includes rare (3%–5%) silty grains, 20–50 μm and frequent (15%–20%) sand grains, 50 μm–2 mm, quartz, feldspars, with very rare (<1%) muscovite flakes, maximum 100 μm, and frequent fine opaque grains, 50–100 μm (in the clay), rarely 300–700 μm. Very rare grog fragments, 0.5–1 mm, and rare fine vegetal debris.

Fine porosity, 10%–15%, 100–200 μm.

Incomplete oxidizing firing with core comprising very fine organic grains and fine channel voids.

III. Semifine fabric

III. 1. Oxidizing firing semifine fabric: P22, P23

Silty clay matrix, moderately sorted, with rare silt grains (3%–5%), 20–50 μm, and 5%–20% fine sand, 50–200 μm, rarely 400–500 μm, very rare 1 mm, and rare (2%–3%) fine mica, 50–100 μm (Figure 3I). Frequent opaque grains and fine vegetal debris, 50–100 μm, rarely 200–300 μm. Includes rare (1%–2%) ceramic fragments, 100–500 μm, rarely up to 2 mm, and more frequent sandstone fragments (5%) and rare (1%–2%) micrite limestone grains, 200–300 μm, rarely 400–600 μm. It is a fine paste, but with frequent inclusions, with a heterogeneous and poorly sorted appearance (Figure 3I).

Fine porosity, 10%–25%, with isolated voids, 100–200 μm, and channels, chambers and microfissures, 500–700 μm, rarely 1–1.5 mm.

Complete oxidizing firing.

III. 2. Semifine fabric with banded structure: P24

Poorly sorted matrix, fine silty clay with frequent (20%–25%) fine and very fine sandy grains, quartz and feldspars, 50–100 μm, rarely 200–400 μm, with rare (5%) fine mica flakes, 50–120 μm.

Includes very rare shell and grog fragments 1–1.5 mm, two grains of fine gravel, 2–3 mm.

Fine porosity, 10%–15%, with isolated voids, 50–100 μm, rarely 200–300 μm.

Semifine paste with thin (1–2 mm) oxidizing firing zones (Figure 3J) and organic core, with frequent fine vegetal fragments, 50–100 μm.

Incomplete oxidizing firing.

IV. Coarse fabric

IV. 1. Oxidizing firing coarse fabric: P29

Coarse paste, poorly sorted matrix with fine silty clay matrix with 5%–20% fine sandy grains, 50–200 μm, quartz, feldspar and rare fine mica (2%–3%), 50–200 μm.

Frequent subangular to subrounded inclusions: quartzite (5%–15%), micaschist (5%), and limestone (5%–30%), micrite but also with recrystallizations, frequent bioclasts (5%), 400 μm–3 mm, rare (1%) opaque grains, 50–150 μm (Figure 3K).

Fine, well‐developed porosity (up to 25%), with frequent isolated voids, 100–200 μm, very rarely 400–500 μm and 1–2 mm, and microfissures obliquely oriented.

Complete oxidizing firing.

IV. 2. Coarse fabric with banded structure: P25, P26, P28 Silty clay poorly sorted matrix, with rare (1%–2%) silty grains, 20–50 μm, and fine sandy (5%), 50–200 μm, rarely 200–500 μm, with banded structure; quartz and feldspar, frequent (10%–25%) carbonate grains, and rare (1%) fine mica, 120–150 μm, very rarely 200 μm.

Relatively frequent inclusions of medium to coarse sand, 0.5–2 mm, and rarely 2.5–3 mm gravel of quartzite (5%–10%), limestone (5%–10%), irregular angular but also rounded, 50–700 μm, rarely 1–2 mm, with frequent bioclasts, and very rarely silty shale (Figure 3L). Rare organic inclusions, 100–300 μm, and more frequent (2%–3%) fine opaque grains, 50–100 μm.

Medium porosity, 10%–20%, with frequent fine isolated voids, 50–500 μm, rare to frequent microfissures (Figure 3L), 500–600 μm, rarely 1–1.5 mm.

Incomplete oxidizing firing.

4.2. PIXE

PIXE data obtained on pelletized powders extracted from all 32 shards are given in Table S2, Supporting Information.

All samples in this assemblage feature relatively low calcium concentrations, well below 10 mass% CaO. The OM analysis showed that some shards contain fine shell fragments or carbonate grains; however, these components, present with variable frequencies and in variable amounts, did not induce a marked increase in their calcium content.

To separate the ceramic shards according to their chemical composition, given the large number of samples and variables, principal component analysis (PCA) was performed using STATISTICA software (version 8.0). The pretreatment of the PIXE data consisted of a log‐ratio transformation, as recommended by Baxter.^{[ 26 ]} The input table for the runs of PCA consisted of the logarithms of the ratios of the concentrations of several oxides (MgO, Al₂O₃, K₂O, CaO, TiO₂, Cr₂O₃, MnO, Fe₂O₃, NiO, ZnO, Ga₂O₃, Rb₂O, SrO, and ZrO₂) to SiO₂ concentration (the major component). PCA was iteratively performed, and several outliers (P8, P15, P20, and P27) were removed from the analysis until a clear clustering of the samples emerged (see Figure  4 ).

Figure 4.

Bi‐plot of the first and second principal components resulting from the PCA of log‐ratio PIXE data. The input for statistical analysis were the logarithms of the ratios of concentrations of several oxides to silica concentrations. The oxides of the following chemical elements, all determined with good precision, were taken into account: Mg, Al, K, Ca, Ti, Cr, Mn, Fe, Ni, Zn, Ga, Rb, Sr, and Zr. Some outliers (P8, P15, P20, and P27) were excluded during the preliminary runs of PCA.

The result of the PCA of the PIXE data shown in Figure 4 indicates that the analyzed ceramic samples fall into two distinct groups. The group consisting of samples P12–P14 and P16–P19 is characterized by a high content of aluminum (≈22.3 mass% Al₂O₃ on average) and low amounts of calcium (≈0.7 mass% CaO on average) and iron (≈2.7 mass% Fe₂O₃ on average). The fragments in this cluster pertain to type II of fabric resulting from petrographic observations. Considering the statistical analysis based on chemical composition, the rest of the samples are dissimilar from these above listed fragments, regardless of their petrographic characteristics. The samples from this second cluster feature lower amounts of alumina (≈16.5 mass% Al₂O₃ on average) and higher amounts of lime (≈4.2 mass% CaO on average) and iron oxide (≈6.9 mass% Fe₂O₃ on average).

Some explanations about the significance of the outliers (P8, P15, P20, and P27) deriving from the PCA were sought. All of these shards were characterized as being made of kaolinitic clays during the visual examination initially performed by the archaeologist.

The relatively high alumina content of sample P8 as reported in Table S2, Supporting Information (20.8 mass% Al₂O₃), might have suggested its inclusion in the cluster with the samples made from kaolinitic clays, which are roughly superposing with shards from OM fabric type II. However, the high concentrations of MgO, CaO, and SrO, as well as that of Fe₂O₃, differentiate the fragment P8 from those included in this group, as indicated in Figure 4. Unfortunately, this shard was not subjected to the OM study. Similar considerations apply to sample P27, included in the type II. 2 of fabric from OM: relatively high amounts of Al₂O₃, MgO, CaO, SrO, and Fe₂O₃ compared to the rest of samples from the cluster in the right side of the diagram in Figure 4.

Fragment P15 (type II. 1 of fabric from OM) features relatively high amounts of K₂O and Fe₂O₃ (slightly >3 mass% of both oxides) compared to the rest of fragments from the cluster made of the samples pertaining to type II of OM fabric. The same apply findings can be mentioned for sample P20—originating from a shard not subjected to petrographic observations, too.

A possible explanation behind these slightly off chemical compositional patterns might be the presence of grog evidenced by OM in most fragments pertaining to type II of fabric. These crushed potteries might have changed the chemical signature of the outliers, despite of their manufacturing mainly from kaolinitic clays.

4.3. XRD

The crystalline structure and phase composition of the seven powdered samples determined by XRD are shown in Table S3, Supporting Information; several diffractograms are presented in Figure  5 . All samples are relatively similar in phase composition, the differences stemming just from the intensities of the main peaks, reflecting different percentages of various crystalline phases.

Figure 5.

XRD patterns for powders obtained from selected ceramic shards unearthed at Pantelimonu de Sus.

Some small shifts of the peak positions on the 2θ(°) axis, as well as variations in the measured number of counts on the intensity axis (arbitrary units), might be a consequence of the ball milling process that can strongly influence the degree of crystallinity, the mean crystallite size (calculated using Debye‐Scherrer formula), a process that might even conduct to the loss of certain mineral phases evidenced by OM observation.

All powders contain quartz as major phase (SiO₂—pdf file JCPDS 01‐087‐2096), with the most intense diffracted peaks corresponding to a hexagonal system of space group P ₃₃₂₁(154).^{[ 27 ]}

Another crystalline phase frequently identified in the analyzed samples is muscovite, a mineral of the aluminosilicates class with the chemical formula KAl₃Si₃O₁₁ (JCPDS 00‐046‐0741), described by the monoclinic system, with well‐defined main peaks at (110), (006), and (131) and with the calculated interplanar distance d very close to that from JCPDS data base, where the reference was prepared by heating it at 850 °C for 5 h—more or less, similar to most ceramic samples reported here.

The difference between muscovite and illite is given by the presence and arrangement of Al atoms in the crystalline structure. Muscovite and illite are indistinguishable by XRD; on the other hand, muscovite can be easily evidenced by OM, while illite is not.

The presence of muscovite/illite in the XRD patterns indicates that the firing temperature for the ceramic fragments containing those minerals did not exceed 900 °C.^{[ 14 ]}

Mineralogically speaking, calcite is very similar to dolomite, the main difference being that in dolomite some Mg atoms replace Ca atoms, increasing the volume of the unit cell and modifying a bit the interplanar distance, thus making the distinction between these two phases very challenging to analytical methods other than XRD.

XRD scans evidenced the presence of calcite (CaCO₃) in samples P1 and P23, while dolomite (CaMg(CO₃)₂) is present in samples P24 and P28. The latter finding agrees with the compositional results provided by PIXE, as samples P24 and P28 contain the highest amounts of calcium in this assemblage (6.6 mass% CaO in P24 and 8.1 mass% CaO in P28, respectively). Sample P24 pertains to group III. 2 in which fragments of shells (2–3 mm) were observed by OM, while sample P28 belongs to group IV. 2 in which OM evidenced frequent carbonate grains. Sample P1 pertains to an OM type I. 1 characterized by the presence of some 1 mm carbonate grains—possibly fine shell fragments, while sample P23 belongs to OM type III. 1 characterized by OM as containing rare limestone fragments (1–2 mm). Thus, the OM observations for these two samples might be correlated with the presence of calcite detected by XRD. On the compositional side, PIXE data, samples P1 and P23 turned out to contain moderate amounts of calcium (≈2.9 mass% CaO each).

The presence of calcite in ancient pottery occurs in two situations: low firing temperatures or postburial depositional processes. Calcite was reported to survive temperatures up to ≈800 °C.^{[ 28 ]} Considering the selection and preparation of these samples, postburial depositional processes could be eliminated as a possible cause of calcite presence. Calcite detection by XRD suggests firing temperatures below 800 °C for at least two of the analyzed fragments, namely P1 and P23. It is possible that all samples from OM types I. I and III. 1 might have been fired at low temperatures, namely <800 °C.

Kaolinite is an alumino‐silicate clay with a layer structure of 1:1 type. The basic structural unit of kaolinite consists of one tetrahedral (Si–O) sheet and one octahedral (Al–O) layer; the stoichiometric formula is Al₂Si₂O₅(OH)₄. Kaolinite consists of trioctahedral minerals such as cronstedite, chrysotile, chamosite, and antigorite, as well as similarly dioctahedral minerals such as halloysite, kaolinite, nacrite, and dickite. This clay is usually white and has a soft plastic‐like appearance, as it is composed of hydrated aluminum silicate and mineral kaolinite. Depending on the arrangement of the atoms of silicon and oxygen, different minerals named halloysite, dickite, nacrite, and kaolinite have the d‐spacing varying from 3.35 Å to 3.58 Å. It is noteworthy that kaolinite, nacrite, and dickite are generally formed through the alteration of feldspar and muscovite.^{[ 29 ]}

The XRD patterns of samples P18 and P19 samples show significant reflections of kaolinite with a monoclinic structure, in good agreement with the results in Zhomartova and colleagues,^{[ 10 ]} who reported neutron diffraction data where the same mineral phase was identified in a coeval Medieval ceramic find excavated at Constanţa, a city on the Black Sea Coast. The authors commented then about the difficulty of distinguishing halloysite from kaolinite using diffraction techniques.

At temperatures above 600 °C, kaolinite transforms into the amorphous phase meta‐kaolinite.^{[ 30 ]} This suggests that the firing temperatures of the vessels in OM type II (P12…P20) were below 600 °C.

5. Discussions

The OM data identified four main types of fabric in the assemblage of ceramic fragments from Pantelimonu de Sus.

The fine silty fabric was made of fine alluvial sediments, being characterized by a good sorting of the sedimentary matrix, the presence of organic elements (vegetal fragments), and the occasional inclusion of shell fragments. This type of fabric was also identified in coeval ceramic fragments from two other sites from Dobrudja, namely Oltina and Castrul 22, as well as in some pottery fragments from Hârşova and Păcuiul lui Soare dated to the 11th c. AD.^{[ 6} , ⁷ , ⁸ , ^{9 ]}

All these four neighboring sites also yielded pottery made of a paste consisting of a mixture of sediments, silty clay and pure clay, the last component oriented and birefringent, possibly kaolinitic—similar to type II fabric described in this study.

A third type of paste fabric was made of semifine silty clay, very probably alluvial, containing sand of variable grain sizes.

The coarse fabric presents a fine matrix, but coarser than those previously described, as well as frequent inclusions of limestone and carbonate sand.

Two types of paste identified in the case of coeval pottery from Hârşova and Oltina, namely the paste made of pure kaolinite clay and that made with coarse sand, were not identified in the ceramic shards from Pantelimonu de Sus. Thus, the comparison to other coeval potteries from the Lower Danube region, namely those from Hârşova and Oltina, resists up to a certain point—at least with respect to the types of paste fabric.

The OM results reported in this study indicate that the most likely source of clay used to manufacture the Early Medieval pottery at Pantelimonu de Sus came from the vicinity of the settlement. The location of the site on the most important river valley in Dobrudja, that of the Casimcea River, makes it probable that the local sediments represented the source of the alluvial clay used by the Medieval potters.^{[ 31} , ^{32 ]}

The OM indicated that most potteries were fired under oxidizing conditions, either complete or incomplete. However, the dark gray color of the outer surface of the samples P1–P11, different from that of the core, suggested their firing in a reducing atmosphere. A possible explanation was found in some ethnographic sources that mentioned the apparition of different shades of gray induced by the final firing atmosphere in the kiln. For example, if some potteries are fired in an oxidizing atmosphere for 10–12 h, and at the end of the process, the kiln is sealed and remains closed for 48 h until the fuel quenches, after sealing, certain chemical reactions occur. Depending on the amount of various iron compounds, granulation of the burnt clay, and the levels of CO₂, gray and black hues on the vessel surface might appear. In the presence of CO₂, the ferric oxide from the red‐fired pottery turns into ferrous oxide, the surface of pottery becomes dark gray, while the core remains red.^{[ 33 ]}

Three types of ceramoclasts were identified in the Pantelimonu de Sus potteries: grog, shells, and vegetal material. If the shell fragments and the vegetal material were likely naturally present in the alluvial clays used for pottery making, the inclusion of grog indicates a deliberate choice of the craftsmen. The presence of grog, indicative of recycling small ceramic shards from broken shards, was evidenced through OM in most of the analyzed fragments—mainly those from types of fabric I, II, and III. This complicates the interpretation of the PIXE and XRD results, as these analyses were performed on homogeneous powders obtained by milling small fragments of each ceramic sample, that possibly also included grog fragments.

The correlation between the OM and XRD data suggested that the vessels from Pantelimonu de Sus were produced in kilns at temperatures below 900 °C. In several cases, potteries were fired at much lower temperatures, i.e., not higher than 600 °C. This may have been the case of the vessels made from kaolinitic clays, i.e., those pertaining to OM type II. 2.

The statistical analysis of the PIXE data clustered the samples in two groups. PCA indicated that the chemical signature of ceramic vessels made of kaolinitic clays is completely different from that of the remaining samples. Similar results were reported for some Medieval assemblages originating in some nearby archaeological sites in Dobrudja.^{[ 6} , ⁷ , ⁸ , ^{9 ]}

The presence of kaolinite (clearly evidenced by XRD in two ceramic fragments) is indicative of the use of kaolinitic clays by the potters from Pantelimonu de Sus. A well‐known kaolinite clay source, exploited since the Roman period, is located nearby the present‐day city of Medgidia—see Figure 1.^{[ 8 ] and the references therein} Another possible explanation is that these vessels arrived to Pantelimonu de Sus by trade/exchange with other settlements where the potters used kaolinitic clays.

The analytical data did not evidence any ceramic item with particular chemical and/or mineralogical signature suggestive for long‐distance trade connections.

6. Conclusions

This study comments on the raw materials and manufacturing techniques used for pottery manufacturing in the Early Medieval settlement from Pantelimonu de Sus, Constanţa County, Romania. This archaeometric approach provided a glimpse on the life in a village from a territory under the influence of the First Bulgarian State, and indirectly, on that of the Byzantine Empire.

The analyses OM, PIXE, and XRD analyses singled out four types of fabric and different chemical or mineralogic characteristics for the analyzed shards; several types of ceramoclasts were evidenced in the clay matrix.

The reported results suggest that during the Early Medieval period, the potters from Pantelimonu de Sus used mainly alluvial clay sources, likely of local origin, adding—or naturally including—various tempers to produce ceramic vessels of different appearances and functionalities, such as kitchenware, tableware, and storage containers.

The presence of kaolinite in some shards suggested that some of the vessels from Pantelimonu de Sus were possibly produced using clays brought from nearby, considering that a source of kaolinitic clays exploited since Roman period is located some tens of km away from this settlement. Alternatively, these potteries arrived here as a result of the regional trade. Most of the analyzed potteries were fired in an oxidizing atmosphere in a temperature range varying from 600 °C up to 900 °C.

The analyses did not provide any clear evidence to support the hypothesis of imported potteries at Pantelimonu de Sus, suggesting the existence of weak economic relations of this settlement with the neighboring ones during the 8th–10th c. AD.

The investigations revealed how pottery was manufactured in the central part of Dobrudja during the Early Middle Ages, suggested the raw materials and their sources, and described the technological level reached by the local potters, thus contributing to unravel the daily living of the small communities from the Lower Danube region and the western shore of the Black Sea.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supplementary Material

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.