Introduction

Jebel Moya is a granite massif that rises from the Gezira Plain in Central Sudan, south of the capital Khartoum and halfway between Sennar on the Blue Nile and Rabak on the White Nile (Fig. 1a). In the northeast corner of the range is a valley 10.4 hectares in size whose official designation is Site 100 but is more commonly referred to by the name of the massif. It was first excavated by Henry Wellcome (1911–1914) and more recently by The University College London–University of Khartoum-National Museum of Antiquities and Museums (UCL–UoK-NCAM) Expedition to the Southern Gezira (Brass, 2016; Brass & Vella Gregory, 2021; Brass et al., 2019, 2020; Vella Gregory, 2020). It is one of the few sites currently being excavated south of Khartoum.

Fig. 1
figure 1

a. Geological map of Sudan showing the location of Jebel Moya in the Gezira Plain and b. Jebel Moya with the locations of the ten excavated trenches from 2017 to 2019. Photograph was taken facing south from the House of the Boulders

Full size image

The site has so far yielded 3141 human burials, most excavated by Henry Wellcome’s team (Addison, 1949). The present expedition is shedding light on the previously unknown history of southern Gezira. The earliest known habitation layers in this large site are currently attributed to the late sixth millennium BC (Brass & Vella Gregory, 2021; Brass et al., 2018). However, the ongoing excavations have yet to reach sterile layers or bedrock. Domesticated cattle, sheep or goats, and early domesticated sorghum were present at Jebel Moya by the early-mid third millennium BC. The earliest known human burial is AMS-dated to C. 2350 BC (Brass et al., 2019). Located outside the southernmost boundary of the Napatan and Meroitic states (Brass, 2015), the final phase of occupation is currently placed between the first millennium BC and the early first millennium AD (Vella Gregory et al., 2022) .

Due to the lack of visual microstratigraphy, excavating in small 10 cm spits within the microgeological strata provided a good measure of control (Brass et al., 2020). Three macro-level phases of occupation are discernible so far from radiocarbon dates (Brass et al., 2019; Brass & Vella Gregory, 2021; Vella Gregory et al., 2022), and these can be correlated with the macro-geological strata termed A–D:

  • Phase 1. This late sixth millennium BC phase encompasses Stratum D and pottery Assemblage 1.

  • Phase 2. It begins in the third millennium BC and continues into the second millennium BC. It is represented by pottery Assemblage 2 and is predominantly in Stratum C. Burial activity occurred in this phase.

  • Phase 3. This first millennium BC to early first millennium AD phase encompasses strata A and B and pottery Assemblage 3. Considerable burial activity occurred in this phase.

Pottery forms a significant part of the material culture at Jebel Moya and contains key evidence for potential connections between the site and the surrounding areas (Brass & Vella Gregory, 2021). It also has the potential to inform the subsistence strategies and the nature of mobility of the local agro-pastoral community (Gosselain, 2000, 2010; Gosselain & Livingstone Smith, 2013). The excavated pottery has been subdivided into three distinct assemblages based on single and co-occurring attributes (Brass & Vella Gregory, 2021). These attributes are the variety and types of tools used, their motor actions, and the resulting décor on different rim types and sherd bodies.

In the present study, a sample of recently excavated pottery sherds predominantly originating from Assemblage 3 (first millennium BC) was analyzed in order to shed light on their raw materials, production location(s), manufacturing technology, and use(s). The study is the second of its kind south of Khartoum (Del Sasso et al., 2014) and the first for the time period, which means there are no comparative samples from southcentral Sudan. This is an area where detailed localized geological data needs to be built up.

The inorganic composition of the paste was investigated using thin-section ceramic petrography and instrumental geochemistry. At the same time, organic residue analysis was conducted using gas chromatography (GC), gas chromatography-mass spectrometry (GC–MS), and gas chromatography-combustion-isotope ratio mass spectrometry (GC-C-IRMS). The results are used to tackle several research questions: Firstly, was pottery locally produced or brought to the site from other areas? Second, can one or more pottery-making traditions be detected at the site, and how far back do they date? Third, what functions did the ceramic vessels serve at the site? By addressing these topics, the study aims to shed light on how the agro-pastoral communities of Jebel Moya and the Gezira Plain functioned outside the direct political control of the Napatan and Meroitic states.

Materials and Methods

Ceramic Compositional Analysis

A total of 62 sherds were selected from the 2017 season when the vast majority of Assemblage 3 sherds were excavated for the purpose of inorganic compositional analysis (Fig. (2; Table 1). The sherds come from in-situ habitation spits and not burial contexts; there was no sherd in unequivocal association with the one burial excavated during this season. Sherds with decorative motifs (Fig. (2) were selected to permit comparisons between the tools and motor actions used and the paste composition from which the sherds were manufactured. These sherds came from secure contexts from Trenches 1, 2, and 4 (Fig. 1b) and three sherds from surface finds. Of the 62 sherds, three earlier sherds from Assemblages 1 and 2 were also included as outliers.

Table 1 Details of 62 pottery sherds from Jebel Moya, analyzed by thin-section petrography and instrumental geochemistry
Full size table

The small size of the sherds prevented accurate reconstruction of the shape and size of their parent vessels. Decorated sherds number 32 (Fig. 2), of which 29 were rim sherds and one was a handle fragment. The rim shape is predominantly thin and straight (n = 21), with eight being everted. Fifteen rim sherds were angled at 90° or less, and three were greater than 90°. Very few signs of primary forming were visible on the sherds, perhaps because of their small size and the predominance of smoothing or burnishing on the exterior. A stylus or comb were the most commonly used tools for decoration (Fig. 2). Comb stamping (n = 10) (Fig. 2D, E) is the dominant decoration, with stylus incised lines (n = 7) and stylus incised chevrons (n = 5) also frequent. One sample was decorated using a seedpod, and another had fingernail impressions. The two sherds from Assemblage 2 had a thick, everted rim with incised chevrons, a rocker comb-stamped body, and a rim with stylus incisions. The single body sherd ascribed to Assemblage 1 was decorated with rocker comb stamping.

Fig. 2
figure 2

Selected decorated pottery sherds from Assemblages 1, 2, and 3. A Stylus incised diagonal lines, sample JMC046, Assemblage 3. B Diagonal comb stamps, sample JMC023, Assemblage 3. C Rolled rim, sample JMC062, Assemblage 2. D Comb stamp, sample JMC061, Assemblage 1. E Comb stamp framed by dragged lines, sample JMC008, Assemblage 3. F “Grooved” pottery, sample JMC012, Assemblage 3

Full size image

All 62 sherds were characterized and classified compositionally using a combination of thin-section petrography and instrumental geochemistry. Thin sections were made using a modification of the standard geological technique (Humphries, 1992; Quinn, 2022, p. 23–36) and analyzed under a polarizing light microscope at magnifications of × 40, × 100, and × 200. The first step in the petrographic analysis was to visually divide the thin sections into “fabrics” based on similarities and differences in their inclusions, ceramic matrix, and texture, representing distinct “recipes” characterized by specific raw materials and paste preparation techniques. The resulting fabrics were described using the system of Quinn (2022, p. 98–124). The petrographic composition of the sherds was contextualized via published geological information (Adamson et al., 1987; Ministry of Energy & Mines, Geological & Mineral Resources, 1981; Vail, 1982) in order to determine likely raw material sources and production location or provenance of the ceramics. In addition, three sand samples collected within Trenches 1 and 2 from strata dated to Assemblage 3 were mounted in epoxy, thin-sectioned, and compared to the ceramic sherds under the petrographic microscope.

Geochemical data were recorded using an Olympus Innox-X Delta Premium portable X-ray fluorescence spectroscopy (pXRF) machine with a rhodium (Rh) source and a 2-mm aluminum filter. Each measurement was taken for 120 s of live time at 40 kV. In order to address concerns about the quality of pXRF data (e.g., Speakman & Shackley, 2013), an in-house calibration developed especially for ceramics was used (Burton et al., 2019, 2021; Lewis et al., 2020; Quinn et al., 2020; Sorresso & Quinn, 2020). The pXRF spectra were deconvoluted in Brukker ARTAX software to correct for the interferences, including Rb Kß/Y Kα, Y Kß/Nb Kα, and Sr Kß/Zr Kα. The disturbance from the varying surface shape of the pottery was accounted for with a Rayleigh scatter distance correction. The in-house calibration was developed using a series of homogeneous fired spiked clay samples with four concentrations of the elements Fe, Ga, Nb, Rb, Sr, Ti, Y, and Zr (Wilke et al., 2016). These samples represent a mass absorption of mid-Z elements in a broad range of clay and other aluminosilicates, with more than 90% of the matrix composed of Al, O, and Si. In total, the calibration measured 15 elements: Ca, Co, Cu, Fe, Ga, K, Mn, Nb, Pb, Rb, Sr, Ti, Y, Zn, and Zr. In order to account for the small (c. 1 cm2) beam of the pXRF and possible heterogeneity in coarse ceramics (Tykot, 2016), at least three spots were analyzed on every sherd, and their measurements were averaged after calibration. The accuracy of the instrument and the “UCL pXRF < 10% Ca calibration 2” applied to the Jebel Moya ceramics was assessed using 14 certified reference materials (CRMs) of rock, ore, sediment, soil, and ceramic (OSM 1). Each CRM was analyzed five times and calibrated using the above protocol. The averages of the five measurements were compared to the certified values for the standards that occur within the range of composition found in earthenware archaeological ceramics, as determined using the data in several published geochemical studies (Day et al., 2011; Quinn et al., 2010; Quinn & Burton, 2015; Travé Allepuz et al., 2014), and accuracy was calculated as percentage relative difference using the formula: (measured – certified/certified) × 100 (OSM 1).

The multivariate geochemical data collected on the 62 sherds were explored using principal component analysis (PCA) both in its raw state and after transformation to base 10 logarithms and the geometric mean (Baxter, 2015, p. 41). Bivariate scatter plots of the PCA scores and pairs of specific elements were used to detect similarities and differences in the bulk composition of the sherds. The resulting geochemical groupings were compared to the petrographic classification of the pottery to check for correspondence and reconcile the two in terms of the mineral and rock sources of specific elements.

Lipid Residue Analysis

Analytical Methods

Lipid residue analysis was performed on a separate set of 50 sherds randomly selected (but based on size criteria) from Assemblage 3, originating from trenches 1, 2, and 4. Lipid analysis and interpretation were performed using established protocols described in earlier publications (Correa-Ascencio & Evershed, 2014). All solvents used were HPLC grade (Rathburn), and the reagents were analytical grade (typically > 98% of purity). Briefly, ~ 2 g of potsherd was sampled, and surfaces were cleaned with a modeling drill to remove exogenous lipids. The cleaned sherd powder was crushed in a solvent-washed mortar and pestle and weighed into a furnaced culture tube (I). An internal standard was added (20 µg n-tetratriacontane; Sigma-Aldrich Company Ltd.) together with 5 mL of H2SO4/MeOH 2–4% (δ13C value measured), and the culture tubes were placed on a heating block for 1 h at 70 °C, mixing every 10 min. Once cooled, the methanolic acid was transferred to test tubes and centrifuged at 2500 rpm for 10 min. The supernatant was then decanted into another furnaced culture tube (II), and 2 mL of DCM-extracted double distilled water was added. In order to recover any lipids not fully solubilized by the methanol solution, 2 × 3 mL of n-hexane was added to the extracted potsherds contained in the original culture tubes, mixed well, and transferred to culture tube II. The extractions were transferred to clean, furnaced 3.5 mL vials, and blown down to dryness. Following this, 2 × 2 mL n-hexane was added directly to the H2SO4/MeOH solution in culture tube II and whirl mixed to extract the remaining residues and then transferred to the 3.5 mL vials and blown down until a full vial of n-hexane remained. Aliquots of the TLE’s were derivatised using N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) containing 1% trimethylchlorosilane (TMCS; Sigma-Aldrich Company Ltd.; 40 μL; 70 °C, 1 h), excess BSTFA was removed under nitrogen, and the derivatized FAMes were dissolved in n-hexane prior to GC, GC–MS, and GC-C-IRMS.

Firstly, the samples underwent gas chromatography using a GC fitted with a high-temperature non-polar column (DB1-HT; 100% dimethylpolysiloxane, 15 m × 0·32 mm i.d., 0.1-μm film thickness). The carrier gas was helium, and the temperature program comprised a 50 °C iso-thermal hold followed by an increase to 350 °C at a rate of 10 °C min−1 followed by a 10 min isothermal hold. A procedural blank (no sample) was prepared and analyzed alongside every batch of samples. Further compound identification was accomplished using GC–MS. FAMEs were introduced by autosampler onto a GC–MS fitted with a non-polar column, 50 m × 0.32 mm fused silica capillary column coated with an Rtx-1 stationary phase (100% dimethylpolysiloxane, Restek, 0.17 μm). The instrument was a ThermoScientific Trace 1300 gas chromatograph coupled to an ISQ single quadrupole mass spectrometer. Samples were run in full scan mode (m/z 50–650), and the temperature program comprised an isothermal hold at 50 °C for 1 min, followed by a gradient increase to 300 °C at 10 °C min−1, followed by an isothermal hold at 300 °C (15 min). The MS was operated in electron ionization (EI) mode at 70 eV. Data acquisition and processing were carried out using the HP Chemstation software (Rev. C.01.07 (27), Agilent Technologies), and Xcalibur software (version 3.0). Peaks were identified based on their mass spectra and GC retention times compared with the NIST mass spectral library (version 2.0).

Carbon isotope analyses by GC-C-IRMS were carried out using a GC Agilent Technologies 7890A coupled to an Isoprime 100 (EI, 70 eV, three Faraday cup collectors m/z 44, 45, and 46) via an IsoprimeGC5 combustion interface with a CuO and silver wool reactor maintained at 850 °C. Instrument accuracy was determined using an external FAME standard mixture (C11, C13, C16, C21, and C23) of known isotopic composition. Samples were run in duplicate, and the average was recorded. The δ13C values are the ratios 13C/12C, expressed relative to the Vienna Pee Dee Belemnite, calibrated against a CO2 reference gas of known isotopic composition. The instrument error was ± 0.3‰. Data processing was carried out using Ion Vantage software (version 1.6.1.0, IsoPrime).

Results

Pottery Compositional Classification

In the thin sections, all but two of the 62 sherds have a related petrographic composition characterized by angular inclusions of plutonic igneous origin in a non-calcareous clay matrix. They can be subdivided in terms of the abundance of the mineral intergrowth perthite, which is common in two-thirds of the analyzed ceramics (perthite-rich granitic fabric) but not in the others (granitic fabric). The granitic fabric sherds contain poorly sorted angular inclusions of microcline, plagioclase, and weathered iron-containing alkali feldspar together with quartz and ferruginous grains (Fig. 3A, B). Argillaceous inclusions, iron-rich rock fragments with quartz inclusions, and small grains of zircon, biotite, and amphibole are also present in low abundance. The clay source used to produce the ceramics of this fabric is likely to have derived from the weathering of coarse-grained granitic igneous rock, such as granodiorite or quartz diorite, which could have been used without significant modification. The perthite-rich granitic fabric contains similar poorly sorted angular inclusions derived from granitic igneous rock. However, it differs in terms of the presence of significant quantities of perthitic feldspar (Fig. 3B, C) and a lower proportion of microcline. This suggests the use of coarse untempered clay derived from the weathering of a related but compositionally distinct, acid-intermediate plutonic rock. Samples JM011 and JM022 of the grog fabric are characterized by poorly sorted inclusions of grog, quartz, and feldspar (Fig. 3D, E), as well as rare amphibole, biotite, and chert. While the presence of quartz, feldspars, biotite, and amphibole suggest that granitic rock was the source of the mineral inclusions, their more rounded nature and lower concentration indicates that these originated from a sedimentary clay source. Grog inclusions testify to the addition of temper of crushed sherds of a similar ceramic composition.

Fig. 3
figure 3

Thin section photomicrographs of Jebel Moya pottery sherds belonging to the granitic fabric (A and B), the perthite-rich granitic fabric (C and D), and the grog fabric (E and F) detected in this study. Images were taken in crossed polars (XP). Image width = 2.9 mm

Full size image

The accuracy of the “UCL pXRF < 10% Ca calibration 2” over the standards that fall within the range of composition found in earthenware archaeological ceramics revealed that 11 of the 15 measured elements (Ca, Fe, Ga, K, Mn, Nb, Rb, Sr, Ti, Zn, Zr) had an average error of ≦25% relative difference between the certified and measured values (OSM 1). The Prolene film covering the powdered CRM samples will have absorbed X-rays and could have affected the counts of some or all of the elements, meaning that the performance may be slightly better than determined here. The element Ca was removed from the dataset because several samples had visible calcite encrustation, which was registered in the pXRF data, even after abrading the analyzed surface. The remaining ten elements (Table 2) were subjected to PCA before and after transformation to base 10 logarithms and the geometric mean. The latter two transformations did not improve the PCA in terms of the discrimination of chemical groups, so the raw non-transformed data was used (Table 2).

Table 2 Geochemical composition and classification of the 62 pottery sherds as determined by pXRF, including data on elements with ≦25% relative difference used in multivariate statistical classification and corresponding thin section petrographic classification
Full size table

A plot of the first two principal components, which explain 66% of the total variance in the multivariate dataset, reveals the presence of two clear chemical groups and a single outlier (Fig. 4A). Sherds of chemical group 1 are defined by their higher values for the elements K, Mn, Nb, and Rb, and of chemical group 2 by their lower values for these elements, as well as its high Sr compared to chemical group 1 (Fig. 4B; Table 2). The two chemical groups correspond well to the granitic fabric (chemical group 1) and the perthite-rich granitic fabric (chemical group 2; Fig. 4C; Table 3). The higher abundance of K and Rb in the chemical group 1/granitic fabric sherds may be explained by their proportion of microcline in thin sections. The chemical group 2/perthite-rich granitic fabric samples have a higher K/Rb ratio supporting the idea that they were made from clay derived from a distinct, though related, igneous source and/or the clay was less weathered. One of the grog fabric sherds plots in chemical group 1 in the PCA, while the other occurs as an outlier due to its higher values for the elements Fe and Ti (Fig. 4C; Table 2). This suggests that the two samples may not share the same base clay despite containing grog. Plotting the elements Rb and Fe against each other makes it possible to discern a compositional split within chemical group 1, which is not readily apparent in the granitic fabric in thin-section despite the sherds containing some heterogeneity (Fig. 4D; Table 2). This is likely to indicate the use of a slightly different clay source, as no difference between these sherds was visible in the thin section.

Fig. 4
figure 4

Geochemical classification of the 62 pottery sherds. A Scatterplot of principal components analysis (PCA) scores using raw data from 10 elements (Fe, Ga, K, Mn, Nb, Rb, Sr, Ti, Zn, Zr) showing the existence of two chemical groups plus one outlier. B Loading plot, showing the influence of the 10 elements on the first two principal components. C PCA scatterplot with samples labeled by petrographic fabric. D Scatterplot of values for the elements Rb and Fe revealing compositional split in chemical group 1. Granitic fabric = closed circles; perthite-rich granitic fabric = open triangles, grog fabric = closed squares. For chemical group membership of individual samples and elemental characteristics of chemical groups and outliers, see Table 2

Full size image
Table 3 Lab number, context, lipid concentration (µg g−1), δ13C and Δ13C values, and attributions of pottery lipid residues from Jebel Moya
Full size table

Organic Residue Analysis

Analysis of the total lipid extracts (TLEs, n = 50), using GC and GC–MS, revealed that nine of the 50 sherds (18%) contained sufficient concentrations (> 5 µg g−1) which can be reliably interpreted (Evershed, 2008; Table 3). Lipid concentrations were generally low. The extracts comprised two main lipid profiles. The first was dominated by free fatty acids, palmitic (C16:0) and stearic (C18:0), typical of a degraded animal fat (n = 6, Fig. 5a) (e.g., Berstan et al., 2008; R. P. Evershed et al., 1997) with the second (n = 3) also containing a sequence of abundant long-chain fatty acids, in distributions generally indicative of plant processing.

Fig. 5
figure 5

Partial gas chromatograms of acid-extracted FAMEs from pottery extracts. a Sample JMO2331; b Sample JMO2317. Circles, n-alkanoic acids (fatty acids, FA); IS, internal standard, C34 n-tetratriacontane. Numbers denote carbon chain length

Full size image

GC-C-IRMS analyses were carried out on six samples typical of degraded animal fats (Table 3, Fig. 6) to determine the δ13C values of the major fatty acids, C16:0 and C18:0, and ascertain the source of the extracted lipids. The δ13C values of the C16:0 and C18:0 fatty acids reflect their biosynthetic and dietary origin, allowing non-ruminant and ruminant adipose and ruminant dairy fats to be distinguished (Copley et al., 2003; Craig et al., 2012; Dunne et al., 2012). Ruminant dairy fats are differentiated from ruminant adipose fats when they display Δ13C values of less than − 3.1 ‰ (Dunne et al., 2012; Salque, 2012).

Fig. 6
figure 6

Graph showing the Δ13C (δ13C18:0 –δ13C16:0) values. The ranges represent the mean ± 1 s.d. of the Δ.13C values for a global database comprising modern reference animal fats from Africa (Dunne et al., 2012), UK (animals raised on a pure C3 diet; Dudd & Evershed, 1998), Kazakhstan (Outram et al., 2009), Switzerland (Spangenberg et al., 2006), and the Near East (Gregg et al., 2009)

Full size image

Lipid residue results for sherds, displaying typical degraded animal products, show that two vessels, JMC2330 and JMC2359, plot in the ruminant dairy region with Δ13C values of − 5.0 and − 3.2 ‰, respectively (Table 2). However, it should be noted that vessel JMO2359 plots at the extent of the range, suggesting some mixing of dairy and ruminant carcass products. Three vessels (JMC2331, JMC2335, and JMC2351) plot in the ruminant adipose region with Δ13C values of − 2.2, − 0.9, and − 0.9 ‰, respectively (Table 3). The remaining vessel (JMO2360) plots between the ruminant and non-ruminant/plant regions, with a Δ13C value of − 0.6 ‰, suggesting either the mixing of ruminant and non-ruminant animal fats (or, possibly, plants) contemporaneously or during the lifetime of use of the vessel (Mukherjee, 2004; Mukherjee et al., 2005).

Notably, the second lipid profile, seen in three sherds (JMC2317, JMC2342, and JMC2354), differed from a typical animal product, comprising low abundances of C16 and C18 fatty acids but sequences of even-numbered long-chain fatty acids (LCFAs) in much higher abundance, containing C20 to C30 carbon atoms, generally dominated by the C26 (Fig. 5B). These LCFAs are strongly indicative either of an origin in leaf or stem epicuticular waxes (Bianchi, 1995; Kunst & Samuels, 2003; Tulloch, 1976) or possibly suberin (Kolattukudy, 1981; Pollard et al., 2008; Walton, 1990), an aliphatic polyester found in all plants. Although primarily found on the surface of plant leaves, sheaths, stems, and fruits, epicuticular waxes are also associated with other plant organs, i.e., seed oils and coats, flowers, bark, and husks (Bianchi, 1995). Fatty acids can be found in high abundance in some plants; for example, waxes from the leaves of the sorghum plant comprise nearly 40% very long-chain fatty acids (Bianchi, 1995). However, these LCFAs are not diagnostic to families of plants and so cannot be used as anything other than a general indicator for plant processing.

Discussion

The geology of the Gezira Plain is dominated by unconsolidated Tertiary and Quaternary sedimentary rocks, Pre-Cambrian basement, including schists, and igneous intrusions (Fig. 1a). The latter, which form the rocky hills of Jebel Moya and neighboring Jebel Saqadi, are composed of granite, containing the quartz, perthite, orthoclase, albite-oligoclase, biotite, and minor amphibole. Other igneous rocks include sills and plugs of basalt that are exposed at Sennar and west of Omdurman (Vail, 1982, p. 54). Cretaceous Nubian sandstone, siltstone, and mudstone occur to the northeast and northwest, where the Blue and White Nile rivers converge.

The petrographic composition of the majority of the analyzed sherds suggests the use of clay derived from acidic plutonic igneous rock such as granite. The mineral assemblage detected correlates well with the general description above of the granite bodies that form the Jebel Moya and Jebel Sagadi massifs, particularly the presence of perthite, which is characteristic of the perthite-rich granitic fabric/chemical group 2 (Fig. 3). With this in mind, it seems plausible that the raw materials for these perthite-rich granitic fabric/chemical group 2 sherds were sourced locally from coarse clay formed by the weathering of the granite, as suggested by the previous analysis of three sherds from the site in the study of Brass (2015). The sand samples collected from Jebel Moya confirm the presence of degraded igneous material of the composition reported for the area and that seen in thin-section in the pottery, including perthite.

The significance of the two distinct but related granite-derived fabrics, as well as the two chemical subgroups of the granitic fabric/chemical group 1, is not clear given the lack of further information on the geology of Jebel Moya and other igneous outcrops in the Gezira Plain and the absence of field samples from other locations. However, they seem likely to represent several separate clay sources. Large granite bodies typically contain mineralogical and chemical variations resulting from several intrusion events and/or the fractionation of magma during its slow cooling underground. With this in mind, it is possible to imagine potters sourcing compositionally similar clay from several places around Jebel Moya and the surroundings, including Jebel Saqadi, albeit containing mineralogical and chemical differences that were not perceptible to the eye or the touch. Equally, one or more of the three granitic compositions could represent pottery made at another location, for example, the site of Jebel Saqadi (Fig. 1a). Whether the granite bedrock of this massif is distinct from that of Jebel Moya is not clear.

Archaeological evidence exists for the occupational activity at Jebel Saqadi (Crawford & Addison, 1951), including mudbrick architecture and pottery sherds. The latter is claimed to bear similarities with the pottery of Assemblage 3 of Jebel Moya, as well as the agro-pastoral village Abu Geili, along the Blue Nile (Fig. 1a). The presence of two main granitic compositions within the ceramics analyzed from Jebel Moya may therefore indicate the existence of local ceramics, as well as non-local material produced at Jebel Saqadi.

The site of Abu Geili is located c. 30 km east of Jebel Moya along the banks of the Blue Nile (Fig. 1a). Also, it contains pottery with stylistic links to that of Jebel Moya (Fig. 2D). The latter includes grooved sherds that belong to the granitic fabric/chemical group 1 that could not have been produced at Abu Geili due to its different underlying geology and may therefore be tentatively taken to suggest the transport of ceramics from Jebel Moya. Abu Geili is situated in an area of Cenozoic sedimentary rock, overlain by recent alluvium. While data on the exact composition of these strata were not available and pottery from the site has not yet been studied compositionally, it may be that the two petrographically and chemically distinct sherds of the grog fabric analyzed in this study could have been made here and thus represent the transport of material in the other direction to Jebel Moya. However, the chemical correspondence of one of these samples with chemical group 1.1 might not support this suggestion. An alternative source area is the Northern Gezira, where Cretaceous sedimentary rocks, including mudstone outcrop, occur (Fig. 1a). The potential small-scale movement of pottery detected here could have occurred as a result of savanna herding, for which ceramics may have been essential tools for processing milk, meat, and/or blood (e.g., Grillo, 2012, p. 195–203), with such movements tied into relationships with neighboring communities.

The stylistic variability of the analyzed ceramics does not correlate with their compositional characteristics, as defined in this study. Multiple decorative motifs made using different tools are on the sherds belonging to all petrographic and chemical groups. Equally, no clear pattern exists in terms of the distribution of the various petrographic and chemical groups across the different spits in Assemblage 3 of Jebel Moya based on the samples analyzed in this study. Sherds of the granitic fabric/chemical group 1 and its subgroups and the perthite-rich granitic fabric/chemical group 2 are present in the lower, middle, and upper spits. Only three Assemblage 1 and 2 sherds were included for comparative purposes, all made of the granitic fabric/chemical subgroups 1.1 and 1.2. The absence of material belonging to the perthite-rich granitic fabric/chemical group 2 and the grog fabric among these samples cannot be considered significant.

The low lipid recovery rate from the ceramics (Table 2), which underwent organic residue analysis, could suggest that the vessels were not subjected to sustained use. Alternatively, it may be a feature of the preservation conditions at the site. Nonetheless, the results demonstrate that two vessels were routinely used to process ruminant dairy products, and another three vessels were used to process ruminant carcass products. A sixth vessel was likely used to process a mixture of ruminant and non-ruminant carcass products. The non-ruminant signal could originate from wild pigs or boar, if present in the area, other non-ruminant wild fauna, or possibly, plants.

The δ13C16:0 values of the fatty acids extracted from the potsherds provide valuable information on environmental conditions and possible animal management strategies. Here, the δ13C16:0 values from the animal products range from − 23.8 to − 16.0 ‰, possibly suggesting the processing of products from animals subsisting on varied C3/C4 diets or that animals may have been managed differently, i.e., through seasonal movements. The somewhat enriched values suggest locally available C4 plants, confirmed by domesticated sorghum at the site. Certainly, cattle eat wild sorghum and other C4 grasses (Snow, 1948).

As a substantial faunal assemblage is developed from ongoing excavations (see Brass et al., 2019, for a preliminary analysis from the first 2017 season), then it would be useful to compare the lipid results to the faunal assemblage to determine what animals may have been processed in the vessels and gain further insight into possible animal husbandry strategies. The bones of cattle (Bos sp.), goats (Capra hircus), and probably sheep (Ovis aries) are found in the sampled stratigraphic sequences, confirming, together with the chemical evidence, that the exploitation of domesticates for their meat and milk products was a key feature of subsistence strategies at the site. It should, however, be noted that it is difficult to make broad generalizations on animal management based on small sample sizes. Also noteworthy is the likely importance of dedicated plant processing at the site, a tradition with a long history to the north, i.e., at Khor Shambat (Dunne et al., 2021), Sphinx (Garcea et al., 2020), and al-Khiday in Central Sudan (Dunne et al., 2022).

Conclusion

Our research set out to answer three questions: the origin of clay sources, the identification of pottery traditions and their longevity, and the function of vessels during the period encompassed by Assemblage 3 at Jebel Moya. Our method of spit excavation has resulted in a finely detailed breakdown of changing pottery attributes and decoration motifs (Brass & Vella Gregory, 2021). This study shows that during the first millennium BC, pottery was overwhelmingly made using locally available clay. Over the course of a millennium, people do not appear to have changed their clay sources or recipes, based on the analyzed samples. No correlation exists between the motifs and tools used to decorate the locally manufactured ceramics and their petrographic and chemical composition. This suggests a strong manufacturing tradition.

Some sherds were made from a source with related but distinctive geological characteristics, such as Jebel Saqadi, 20 km to the north (Brass, 2015). Rare sherds of a very different paste composition, made from raw materials that may not have been available at Jebel Moya or Jebel Saqadi, suggest the existence of pottery from a third location. One potential candidate is the contemporaneous site of Abu Geili, c. 30 km to the east, implying the transport of pottery and communication between the people of Jebel Moya and those inhabiting the banks of the Blue Nile. Jebel Moya Assemblage 3 sherds have been identified at Abu Geili (Brass, 2016; Crawford & Addison, 1951). This and the connection with Jebel Saqadi should be tested by future compositional analysis of pottery and raw material samples from these sites and elsewhere in southern Gezira.

Seen only through a compositional lens, traditions appear to be fairly static. People did not change their clay sources or recipes; instead, they focused on the sheer diversity of visual elements of pottery: while motifs do persist, they also eventually change (see Brass & Vella Gregory, 2021 for the detailed study of how visual decoration of pottery reveals aspects of identity and communities of practice).

Finally, with regard to vessel function, while it should be noted that this is a small dataset, some preliminary observations on subsistence practices at Jebel Moya can be made. Our results demonstrated that at least some pottery was used to process animal products, including ruminant dairy and carcass products and non-ruminant carcass products. This is verified by the zooarchaeological data, which confirms the presence of domesticated cattle, sheep, and goats at the site. The importance of vessels dedicated to plant processing is noteworthy, suggesting the importance of a broad range of resources to people at Jebel Moya.

Together, these results illustrate the importance of integrating various strands of archaeological data. The petrographic results are also the first of its kind for the first millennium BC, south of Khartoum. While pottery production from Al Khiday, immediately south of Omdurman along the White Nile, has been analyzed, it focused on Mesolithic and Neolithic pottery (Del Sasso et al., 2014). In particular, this study provides a way to address complex questions relating to mobility and change. Thus, while the composition of clay did not change much across a millennium, our sequence enabled us to trace other changes. Equally, the zooarchaeological data must be read with residue studies. This was a time of fairly rapid climatic change, and a better understanding of how the people were moving in the landscape, what foods they were consuming, how they were consuming, and how they were constructing social identities and traditions through their pottery in a region where today there are a number of different social cultures interacting are vital questions only partially answered by this study. This integrated approach to material culture will enable us to expand the research to materials and remains from subsequent field seasons. It will also assist in answering questions on mobility in the southern Gezira in finer detail.