Introduction

In ancient Mesopotamia, the clay provided by the Tigris and Euphrates alluvial plains was used for many purposes, such as for construction and for manufacturing votive objects, and as a writing medium1. Inscribed with cuneiform signs, clay artefacts record a variety of information, including administrative records, letters, medical text, and literature. A significant number of written documents date back to the so-called Ur III period (21st c. BC) because of a massive reorganization of state administration at that time2. With respect to the manufacturing of tablets, authors have suggested that these artefacts were mostly sun-dried, while only a small number were intentionally fired in antiquity3. Clay cones and clay bricks, inscribed with accounts of the ruler’s building activities, were supposedly baked and then placed in the foundations and walls of monumental buildings4. Nonetheless, determining which artefacts were treated intentionally is difficult, given that firing was a common practice developed by museums for conservation purposes5, especially from the 1950s to the early 1990s, until this treatment was deemed unnecessary in 1987 and 19936. Furthermore, historical events such as the sacking of royal archives following the siege of Nineveh in 612 BC7 complicate the correct determination of intentional firing as a pottery practice.

From an archaeological perspective, correctly addressing the origin of a tablet is not always possible considering that many cuneiform tablets were acquired from the antiquity market rather than being discovered during scientific excavations; in these instances, the archaeological context and, usually, the place of origin are lost. A contribution from the field of natural sciences is essential to better understand the manufacturing of these artefacts and their conservation until the present day. Among the many scientific methodologies implemented for the study of cuneiform clay tablets, X-ray fluorescence (XRF) spectroscopy is probably among the most recognized methods, allowing the determination of the chemical composition at the surface in a non-invasive way8. With this technique, an extensive study focused on investigating regional differences in the chemistry of clay tablets from Iraq and Turkey, and a total of 540 artefacts housed at the Yale Babylonian Collection of Yale University were analysed9. In other recent research, energy-dispersive X-ray spectrometry (SEM–EDX) and X-ray computed tomography (CT), together with a petrographic study, were conducted of thin sections of cuneiform clay tablets obtained from the British Museum collection10.

To better understand the raw material used for manufacturing these artefacts, an important method for determining the mineral composition is X-ray powder diffraction (XRPD). This technique has been implemented both invasively and non-invasively. In the cases where the analysis was destructive, a portion of the clay tablets was scratched or removed3,11, whereas in one instance, a micro X-ray diffractometer was used to identify the main minerals in reflection geometry12. Synchrotron radiation-based techniques have been widely exploited at facilities such as the European Synchrotron Radiation Facility (ESRF)13,14 and the Deutsches Elektronen-Synchrotron (DESY)13,15 for cultural heritage studies. A recent contribution conducted at the French national synchrotron facility SOLEIL performed non-invasive synchrotron X-ray fluorescence (SR-XRF) spectroscopy on salt-covered cuneiform clay artefacts16. Non-invasive analyses are important owing to the conservation of the artefacts and for the possibility to reproduce the experiments.

In this study, we characterized cuneiform clay artefacts on the basis of their mineral composition without damaging them. By understanding the mineralogy, the manufacturing process can be reconstructed; for example, the clay deposit from which the raw material was collected can be determined, important processes that affected these artefacts after manufacturing, including burial and mineral percolation, can be identified, and whether the artefacts underwent thermal treatment can be determined. For this reason, synchrotron XRPD measurements have been carried out at PETRA III, DESY, on cuneiform clay artefacts provided by the Museum für Kunst und Gewerbe (MK&G) and the Staats und Universitätsbibliothek (SUB). In detail, on clay tablets, including their envelopes in some cases, clay cones, and a unique leg-shaped artefact. This study represents the start of a dataset aimed at determining insights onto provenance of the raw material used for manufacturing these artefacts, together with expanding the knowledge of pottery practices used in antiquity. In fact, we show how the mineral composition can be used to distinguish between two main types of documents. The identification of high contents of diopside is suggested to indicate thermal treatment rather than the use of a materials from a source other than Tigris and Euphrates sediments.

Methods

Samples and provenance

The artefacts considered in this study encompass 11 clay tablets obtained from the collection of the SUB in Hamburg and 25 clay artefacts, including tablets, fragments of tablets, cones, and a leg-shaped artefact, obtained from the MK&G in Hamburg; all the derived artefacts were procured from sites in Iraq (Fig. 1). Representative artefacts arranged according to shape and size from the MK&G and SUB collections, together with the inferred provenance of all the objects measured in this study, are shown in Fig. 2. The thickness of the measured artefacts ranged from 10 mm to 30–40 mm. Occasionally, the sample surfaces were blackened (Fig. 2c).

Fig. 1
figure 1

Map showing the site provenance of the 36 written artefacts considered in this study, created using the free and open-source QGIS.

Full size image
Fig. 2: Representative artefacts from the MK&G and SUB collections.
figure 2

Clay tablets from Puzrish-Dagan, 1913.125 and Cod. Var. 3 (a, e); clay cones from Uruk, not altered, 1913.132 (b), and with a blackened surface, 1913.130 (c); a leg-shaped inscribed artefact, 1983.286 (d); and an artefact from Assyria, the northern region that comprises the Niniveh and Assur sites; a tablet with the corresponding envelope from Girsu, 1920.107 a + b (g). All the artefacts presented here are from the MK&G, except (e), which is from the SUB, Hamburg.

Full size image

Archaeological background: a case study

Clay cone 1913.131 from the MK&G collection bears a building inscription attributed to Sîn-kāšid, an Amorite ruler of Uruk in the Old Babylonian period (20–17th c. BCE) (Fig. 3). The same inscription has been observed on over a hundred written artefacts, including clay cones and tablets, some discovered through scientific excavations of the remains of Sîn-kāšid’s palace at Uruk and others acquired from the antiquities market. The inscription is edited in Frayne, D. R.17 as no. 4.4.1.3. For a detailed list of corresponding artefacts see Frayne, D. R.17 444−447. The transliteration and translation presented here follow the edition in this volume. Italics indicate Semitic elements in the Sumerian text, including the ruler’s name and the name of an Amorite tribe. The present artefact belongs to the latter category and lacks a documented excavation context. Nevertheless, given that Sîn-kāšid is known only from Uruk, the provenance of the artefact can be reliably assigned to this site. The chronological placement of Sîn-kāšid’s reign in the 19th c. BCE provides an approximate date for the object.

Fig. 3
figure 3

Hand copy and picture of clay cone 1913.131.

Full size image

The inscription comprises nine lines written in the ductus of Old Babylonian monumental inscriptions. The transliteration and translation of the Sumerian text inscribed on the cone are as follows:

  1. 1.

    dsuen-ka3-ši-id / Sîn-kāšid

  2. 2.

    nita kalag-ga / mighty man

  3. 3.

    lugal unugki-ga / king of Uruk

  4. 4.

    lugal am-na-nu-um / king of the Amnānum (tribe)

  5. 5.

    u2-a / provider

  6. 6.

    e2-an-na / of the Eanna (temple)

  7. 7.

    e2-gal / built

  8. 8.

    nam-lugal-la-ka-ni / his royal

  9. 9.

    mu-du3 / palace

Clay cones, along with other artefacts bearing short royal inscriptions, were commissioned by rulers in connection with construction and restoration projects. These objects, which include tablets, cones, figurines, and architectural elements such as door sockets, were produced en mass in workshops and placed within buildings as part of foundation or dedicatory assemblages. Clay cones were specifically designed as decorative elements inserted into the walls. In some cases, more than a thousand cones bearing the same inscription have been preserved, underscoring the scale of their production. Once embedded in the walls, the text on their shaft remained hidden from human sight, as their inscriptions were intended for a divine audience to affirm the ruler’s piety and immortalize his achievements in the presence of the gods. Furthermore, rulers of later times were also meant to discover foundation deposits and honour previous builders and restorers with their own inscriptions.

Royal inscriptions, when preserving the name of the ruler, are generally straightforward to date. In the case of Sîn-kāšid, both the dating and provenance are not problematic, as his rule is exclusively associated with Uruk. However, in many other cases, establishing a date and provenance is considerably more challenging, especially when artefacts lack a documented archaeological context or when excavation records are incomplete. Without such information, researchers must rely on internal textual features, palaeographic analysis, and comparative material to approximate the time and place of origin.

Beyond royal inscriptions, which are confined to a limited range of textual genres, additional clues for determining provenance can be found in geographical names, personal names and prosopography, dating formulae, and other textual elements. However, these indicators are rarely definitive on their own. Personal names may suggest a regional or cultural affiliation, but they are not exclusive to a single location. Similarly, dating formulae, while valuable for chronological placement, do not always reveal the geographical origin of an artefact.

Orthographic features, particularly diagnostic cuneiform signs that exhibit period- or region-specific characteristics, play a crucial role in refining both the date and provenance. Certain sign forms, spellings, or sign readings were used only in specific periods or scribal traditions, making them essential tools for classification18. These palaeographic and linguistic markers are especially important when dealing with artefacts from the antiquities market, where the lack of archaeological context necessitates a reliance on internal textual analysis to reconstruct an artefact’s historical and cultural background.

Instrumental and analytical methods

During the present study, X-ray powder diffraction patterns were collected in transmission geometry at the PETRA III synchrotron of DESY, Hamburg, at the Powder Diffraction and Total Scattering Beamline P02.1. Prior to the actual measurement of the original artefacts, three replicas were used to evaluate the method and prove the non-invasive treatment. The original samples were then placed at the beamline in a specially designed clamp system, where the clamps were covered with foam material so that the surface of the artefacts was not scratched. No prior treatment of the artefacts was conducted. Data collection was carried out at room temperature using an unfocused collimated X-ray beam with a diameter of ca. 1 × 1 mm2, a photon energy of approximately 60 keV (wavelength λ = 0.2073 Å) and a theoretical photon flux of 4 × 1010 ph/s. Diffracted X-rays were collected by a Varex XRD 4343CT detector (150 × 150 µm2 pixel size, 2880 × 2880 pixel area) mounted orthogonal to the beam path with a sample-to-detector distance of approximately 1100 mm and calibrated using LaB6 NIST SRM 660c19 as a standard reference. Each artefact was measured at three points to estimate its homogeneity. The irradiation time for each spot was 5 min. The 2-dimensional diffraction patterns collected were subsequently azimuthally integrated into one-dimensional powder diffraction patterns using the programme DAWN20. Owing to the relative thickness of the investigated objects, diffraction occurs along the penetration trajectory of the X-rays through the artefact. This leads to an intrinsic broadening of the diffraction peaks and, therefore, to a lower resolution than that of standard powder X-ray diffraction measurements for samples in small-diameter capillaries. Moreover, the resolution depends on the angular 2theta value; with increasing angle, the extent of broadening increases. Nevertheless, the diffraction peaks were sharp enough to perform accurate refinements for phase identification purposes. The historical objects were subsequently characterized according to their mineral composition. Mineral identification was conducted using the open source programme Profex21, and entries were selected from the programme’s internal database and from the Crystallographic Open Database (COD)22. Rietveld refinements were implemented to quantify the mineral phase content of the measured artefacts using the programme TOPAS 7.023. The peak profile was modelled using the fundamental parameters approach (FPA)24 in TOPAS, whereas Chebyshev polynomials were used to model the background of the diffraction patterns. The lattice parameters of the selected phases—illite, palygorskite, kaolinite and hornblende—were fixed during refinement for stability. The lattice parameters of the other mineral phases were refined. A pseudo-Voigt function was used to model the peak shape of each phase. Refined broadening parameters for the peak shapes were Gaussian and Lorentzian crystallite size parameters as well as phenomenological Gaussian and Lorentzian microstrain parameters, as included in the TOPAS software. Atomic positions, site occupancies, and displacement parameters were not refined. For the majority of the clay artefacts but not for the clay cones, the illite peaks were modelled using 4th order spherical harmonics for correct scaling of peak intensities affected by the preferred orientation. In these cases, the lattice parameters were fixed.

Results

Mineralogical composition

The experiments revealed considerable variability in the mineral composition among the artefacts. A mineral identified in all the cases is quartz, given its ubiquitous abundance and high thermal stability. Feldspars were also found in all the objects, represented by the plagioclase endmember albite, whereas other studies report plagioclase12 or plagioclase+K-feldspar11 as the main feldspar phases. Clay minerals were represented mainly by illite, both in terms of distribution and abundance, whereas lower concentrations of kaolinite, chlorite and, in some instances, palygorskite were also observed. Small amounts of amphibole hornblende were found in most of the artefacts. Calcite was also highly represented, with only two samples completely missing its characteristic peaks. In these two samples, namely, a clay tablet (2023.1) and a clay cone (1913.132) from Uruk and dated to the 19th c. BC, illite and hornblende were not observed, but the zeolite analcime was present (Fig. 4a). Other carbonates, such as aragonite and dolomite, were observed in some of the samples. Gypsum, found only in one clay cone, 1913.128 (Lagash, 22nd c. BC), has been identified in a study12 on tablets from Umma, Dilbat, Larsa, Ur, Babylon, Uruk, Sippar, and Nippur (Ur III and Early Achaemenid periods) using micro X-ray diffraction. This mineral could form on the surface of the artefact as an alteration product. The clinopyroxene diopside was found in some artefacts in conspicuous amounts, and samples 2023.1 and 1913.132 contained up to 70 wt. % diopside according to phase quantification with the Rietveld method (Fig. 5). Finally, we determined the occurrence of spinel-phase magnetite in a clay tablet whose envelope was from Girsu, 21st c. BC (Table 1). The occurrence of magnetite in other samples is not excluded but could not be determined given that the intense peaks overlap with the diopside peaks.

Fig. 4: Phase identification for representative artefacts.
figure 4

Clay cone 1913.132 shows diopside peaks but no carbonates or clays (a), whereas the leg-shaped inscribed clay artefact, 1983.286, shows the most frequent mineral assemblage (b). The corresponding pictures are also displayed.

Full size image
Fig. 5
figure 5

Phase analysis with the Rietveld method of artefact 1913.132. The difference between the observed and calculated diffraction patterns is shown.

Full size image
Table 1 Comparison of mineral phases identified in this study with existing research that applied XRD*
Full size table

Mineral content in relation to morphology, provenance and dating

The surfaces of some of the artefacts considered in this study, including clay cone 1913.130, were blackened. However, we could not identify any relevant mineral conferring this effect. Since alterations of the surface often consist of only a few atomic layers, surface diffraction methods in reflection geometry (e.g., grazing incidence) might provide greater insight into the nature of this blackening. However, such an investigation was beyond the scope of our study. Considering the individual documents according to morphology, the leg-shaped artefact displays the paragenesis of quartz, calcite, albite, aragonite, amphibole and illite (1983.286; Fig. 4b), shared with most of the other clay tablets, including their envelopes (Table 2). This composition is independent of the inferred provenance for tablets belonging to the Ur III period and for the leg-shaped artefact, which dates approximately 1000 years later. The objects from the 22nd c. BC and the 19th c. BC, consisting of two tablets (2023.1 and 2023.2) and all the investigated clay cones, contain various amounts of diopside. Nonetheless, according to existing research on clay tablets from the Old Babylonian (ca. 1800 BC), Neo-Babylonian (ca. 550 BC), and Early Achaemenid (ca. 500 BC) periods12 and from the inferred Neo-Babylonian11, pyroxene was not observed. In our study, the presence of diopside is independent of provenance and period. All the clay cones contain various amounts of diopside. This suggests a differentiation of manufacture according to the purpose of the clay cones, which confirms the description in Chiera, 19384. Sufficient statistics of measurements would clarify this finding.

Table 2 Mineral quantification for each artefact*
Full size table

With respect to a possible clustering of artefacts, in a study carried out by9, samples were grouped according to the chemical composition determined by X-ray fluorescence (XRF) in relation to the provenance, resulting in three distinct trends from Anatolia and the lower or upper stream of the Tigris and Euphrates. In our study, sites from Anatolia and the upper stream region, from which the tablets from Assur and Nineveh originated, were not sufficiently represented to draw conclusions and obtain similarities with the trends reported by other authors.

Discussion

The raw starting material used for manufacturing the measured objects resulted in illitic clay or silt with various contents of chlorite, kaolinite, and palygorskite. A comprehensive study in which XRPD, FTIR, and SEM−EDX methods were performed on small samples from cuneiform tablets also revealed palygorskite-based silt with inclusions of calcite, quartz, and amphibole3. In our case, palygorskite is not the main representative clay mineral in all the objects. Instead, the main clay fraction resulted in an illitic phase. The mineral composition of some of the tablets is comparable to that of deposits along the Euphrates from Hilla (Babylon) to Basrah, Iraq, with illite, chlorite, kaolinite, and palygorskite being the major representative minerals of the clay fraction25. The clay mineral contents in the sediments studied were, on average, 20%. However, in the samples considered here showing comparable mineralogy, the average clay mineral content increased to 40 wt. % after Rietveld quantification (Table 1). This difference could be attributed to the variability in the mineralogy of sediments from the Mesopotamian Plain according to the area distribution and depth. Among the heavy fraction, sediments near Nasiriyah, Iraq, also contain hornblende26. Compared with sediments from the Basrah area27, which is closer to the Persian Gulf, there is a substantial difference, namely, the lack of gypsum–except for the small amount in one sample, possibly due to percolation and precipitation during burial—or halite in our samples derived from estuarine fluctuations.

With respect to the determination of thermal treatments, firing tests conducted on materials used in the manufacturing of ceramics in the Algarve Basin in southern Portugal28 attributed the occurrence of diopside to the thermal decomposition of dolomite:

$$2\text{Si}{{\rm{O}}}_{2}\,\left({quartz}\right)+\mathrm{CaMg}\left({\rm{C}}{{\rm{O}}}_{3}\right)\,\left({dolomite}\right)\to \mathrm{CaMgS}{{\rm{i}}}_{2}{{\rm{O}}}_{6}\,\left({diopside}\right)+2{\rm{C}}{{\rm{O}}}_{2}$$
(1)

The findings were observed from temperatures ranging from 700–900 °C depending on the starting composition of the minerals. This pyroxene forms from either the vitreous phase derived from the decomposition of clay minerals and silicates or at the interfaces between minerals. Additionally, thermal decomposition of chlorite can lead to the formation of clinopyroxene and a spinel phase (magnetite)29,30 according to the following reaction:

$$(\text{Mg},\text{F}{{\rm{e}}}^{2+}{)}_{5}\text{Al}(\text{Al},\text{S}{{\rm{i}}}_{3}){{\rm{O}}}_{10}(\text{OH}{)}_{8}({chlorite})+\text{CaC}{{\rm{O}}}_{3}({calcite})+2\text{Si}{{\rm{O}}}_{2}\left({quartz}\right)\to \,4\text{Ca}(\text{Mg},\text{F}{{\rm{e}}}^{2+})\text{S}{{\rm{i}}}_{2}{{\rm{O}}}_{6}\left({fassaite}\right)+\left(\mathrm{Mg},\mathrm{Fe}\right){\rm{A}}{{\rm{l}}}_{2}{{\rm{O}}}_{4}\left({\text{spinel}}\right)+4{\rm{C}}{{\rm{O}}}_{2}+4{{\rm{H}}}_{2}{\rm{O}}$$
(2)

Our results indicate that dolomite can occur in some of the tablets in modest amounts, whereas the main carbonate comprised calcite in all the cases. To explain the high contents of diopside in artefacts 1913.132 and 2023.1, the high content of MgO available during a probable firing must be considered. In addition to dolomite, illite and chlorite are included. An important contribution to the reaction of MgO might also be derived from the thermal decomposition of palygorskite, a clay mineral containing high contents of MgO, which we observed in some cases and is found in raw materials from the Mesopotamian Plain, along the Tigris and the Euphrates25. The occurrence of magnetite, together with diopside, could be determined in only one sample, and owing to its low content and peaks in the diffractogram overlapping those of the more represented diopside, whether other samples might also contain small amount following high-temperature treatment is unclear. Iraqi sediments contain small amounts of magnetite26; however, magnetite is also a product of firing pottery. For this reason, it is unclear whether magnetite is derived from the sediment or a firing process.

Diopside and magnetite have also been observed as high-temperature products—1100–1200 °C—in illite-rich clays used for the production of black ceramics31. Studies on thermal treatments in pottery using thermogravimetry have already determined that Bronze Age kilns for pottery in Iraq could reach T values of 1070–1180 °C32. This, together with many petrographic studies finding signs of thermal treatments, including vitrified filaments10 and alteration of minerals such as hornblende and voids because of the former presence of vegetal materials33, indicate that many clay artefacts were thermally treated. The presence of diopside and absence of calcite have also been suggested to indicate thermal decomposition in the case of Bronze Age pottery shards from Shahr-i-Sokhta, Iran34. In this case, analcime was also observed.

Subdivision among the objects was possible according to the occurrence of diopside. All the clay cones from the 22nd to 19th c. BC in our study contain various amounts of pyroxene. In two unusual cases, very high contents of diopside and no carbonates or clays were found. The reasons could be the substantial variability in the deposits used or, more likely, a sign of thermal treatment according to mineral equilibria, as derived from existing petrographic studies on clay artefacts and commonly observed phases in pottery. Determining when this firing occurred is not possible unambiguously, not only for modern conservation practices but also in cases where archives have been burned in the past. Synchrotron XRPD is proven effective when investigating clay-written artefacts without damaging them. Further studies are needed to confirm whether the occurrence of pyroxenes and lower contents of clays and carbonates, until they are completely absent, or paragenesis with higher contents of clays could reflect variations in pottery practices.