Introduction

Arslantepe is a multi-stratified mound of approximately 4.5 ha located in south-eastern Anatolia a few kilometers west of the Euphrates river (Malatya, Turkey) (Fig. 1). The excavations conducted continuously since 1961 by the Italian Archeological Expedition in Eastern Anatolia from Sapienza University of Rome (MAIAO) allowed the discovery of a long and detailed sequence, mostly anchored to high precision 14C dating stretching from the end of the fifth millennium BC to the Byzantine era (Frangipane 2019; Vignola et al. 2019; Manuelli et al. 2021).

Fig. 1
figure 1

The site of Arslantepe and the Malatya plain from north-west. The Late Bronze and Iron Age sectors are visible on the left (photo ©MAIAO). In the upper box, the geographical position of the site is highlighted (map courtesy of Maps for Free, elaborated by the authors)

Full size image

Investigations into the Late Bronze and Iron Age levels have been conducted intermittently at the site. The first activities of the Italian expedition were carried out on the northern sector, in the area where a French team directed by Louis Delaporte had brought to light during the 1930s the remains of the famous “Lions Gate” (Delaporte 1940). Here, a sequence of monumental structures, which spans from the sixteenth to the tenth centuries BC, were investigated for over 10 years (Pecorella 1975; Palmieri 1973: 65–80; Palmieri 1978a; Manuelli 2013: 39–48). In 1971, the beginning of the activities on the southern sector marked the interruption of the large-scale investigations of the historical levels; nonetheless, a significant series of Late Bronze Age 1 domestic structures, dated to the seventeenth and sixteenth centuries BC, have been unearthed over the years (Palmieri 1973: 65–66; Frangipane and Palmieri 1983: 288–290; Manuelli 2013: 48–71). After more than 40 years of inactivity, a new round of investigations of the northern sector started in 2008 with the aim of finally providing answers about the development of the site during the second and first millennium BC (Liverani 2012). Excavations carried out until 2010 and later in 2015 and 2016 have been conducted over a large area, leading to the discovery of an uninterrupted sequence from the mid-thirteenth to the seventh century BC (Manuelli et al. 2021). In 2016, a new area was opened with the aim of connecting the northern sector with the inner part of the settlement. Ongoing excavations have brought to light monumental remains that can be dated between the tenth and the seventh century BC (Frangipane et al. 2020).

The pottery repertoire from Arslantepe has been for decades an object of archaeometric analysis, mainly consisting in thin-section petrography and wavelength dispersive X-ray fluorescence. The ceramic samples analyzed so far cover a chronological span of almost 4 millennia from the Late Chalcolithic to the Middle Iron Age. Coupled with raw material surveys and technological observations, these analyses evidenced diachronic variations in the organization of production in relation to different levels of socio-economic complexity (Fragnoli and Palmieri 2017; Fragnoli 2018; Fragnoli and Frangipane 2022).

In this paper, we present the results of petrographic and geochemical analyses recently carried out on 99 samples (Table 1) from well-contextualized Late Bronze and Iron Age pottery assemblages from the site. Considering the wide time-frame covered by this research as well as the manifold historical implications related to the abovementioned periods, this analysis is not meant to deal comprehensively with all the questions that might be raised by the study of each ware class involved. Rather, it intends to provide an overview of the results obtained so far and new food for thought for future discussions. Physical and chemical analyses have been widely carried out on pottery assemblages at many Late Bronze and Iron Age key-sites of the eastern Mediterranean world. However, the specific situation of the Upper Euphrates is instead very fragmented, and comparisons need to be found in the wider framework of the Syro-Anatolian region.

Table 1 List of the ceramic samples analyzed in the paper with respective phases, wares, descriptions, contexts and petrographic groups
Full size table

With this in mind, the main aim of this study is to highlight elements of continuity and discontinuity in both paste recipes and material supply as well as potential diachronic changes in manufacturing techniques and the organization of the production at the Late Bronze-Iron Age transition. We also seek to better comprehend the impact that certain exogenous ware classes have had on the local production over the centuries, with a specific focus on the Middle Iron Age.

Archeological and historical overview

During the Late Bronze Age Arslantepe was affected by the Hittite imperial military expansion. The campaigns of the Hittite kings succeeded in subjugating the whole Upper Euphrates region, and Arslantepe gradually assumed the role of a frontier site at the margin of the empire (de Martino 2012).

Period VB (Late Bronze Age 1: VB1 = ca. 1700–1600 BC / VB2 = ca. 1600–1400 BC) is characterized by the presence of an impressive rampart that surrounded the entire mound (Palmieri 1978a: 35–37; Manuelli 2013: 41–43; Alvaro 2012: 350–352). The fortification was connected with a gate system flanked by protruding towers and enclosed a series of dwellings arranged for domestic good storage (Palmieri 1973: 58–71; Frangipane and Palmieri 1983: 289–290; Manuelli 2013: 48–66). Period IV (Late Bronze Age 2, ca. 1400–1250 BC) presents some interesting changes. The southern part of the site was gradually abandoned, and a new fortification system associated with a chambered gateway was built (Pecorella 1975: 3–5; Alvaro 2012: 353–355; Manuelli 2013: 404–409).

The downfall of the Hittite civilization represented the push factor for the emergence of a series of independent powerful reigns (Osborne 2021). The kingdom of Malizi/Melid had its capital at Arslantepe and its domain extended over a vast area westwards of the site up to the Euphrates river (Hawkins 2000: 282–329).

Period IIIA (Late Bronze-Iron Age transition: IIIA.1 = 1250–1200 BC / Early Iron Age 1: IIIA.2 = ca. 1200–1000 BC) is characterized by the presence of structures that directly overlap the destruction of the early gateway. A massive fortification wall keeps on enclosing the northern portion of the mound (Manuelli and Mori 2016). Period IIIB (Early Iron Age 2, ca. 1000–850 BC) in contrast marks an interesting change in the occupation pattern of the site. The former fortification wall was reused after its destruction, but associated structures are now exclusively represented by pits and silos and the entire area was devoted to grain storage (Frangipane et al. 2020: 81–86). Period IIA (Middle Iron Age, ca. 850–700 BC) shows a succession of three monumental-pillared halls built on top of the silos level. The building sequence finds a direct stratigraphic connection with the structures belonging to the abovementioned “Lions Gate” (Liverani 2010).

The pottery repertoire

Because of the long history of investigations and the dissimilarities of the excavated contexts as well as the wide chronological framework, the Late Bronze and Iron Age pottery production at Arslantepe has been studied through many approaches and with different aims. The Late Bronze Age repertoire has been comprehensively examined and ultimately published (Manuelli 2013), while the Early Iron Age material has already been fully processed but its study is still ongoing (Manuelli 2018). On the other hand, the Middle Iron Age pottery assemblage has been only given a preliminary analysis, considering a restricted amount of better-preserved shapes and the most renowned ware classes (Frangipane et al. 2020: 92–98). This heterogeneity of the analyses has of course implied differences in the way the single wares have been processed, identified, and labeled (Table 1).

The Late Bronze Age and Early Iron Age wares have been macroscopically identified thanks to the co-occurrence of their main attributes, including pastes composition (mineral, vegetal, mixed), inclusions texture (coarse, semi-coarse, medium, semi-fine, fine, extremely-fine), surface treatments, colors of pastes and surfaces, decorations, manufacturing technologies, firing characteristics, and associations with specific shapes and their functional attributes (Manuelli 2013, 2018; Frangipane et al. 2020). In contrast, the Middle Iron Age wares have been macroscopically defined exclusively because of their cross-cultural correspondence with well-known coeval productions, and their classification is mostly based on the presence of specific decorative elements or surface treatments.

The Late Bronze Age (Fig. 2 )

Fig. 2
figure 2

Late Bronze Age ceramic wares groups: a Ware group A; b Ware group B; c Ware group C; d Ware group D; e Ware group E; f Ware group F (Red Lustrous Wheel-made Ware) (photos ©MAIAO)

Full size image

The Late Bronze Age (ca. 1700–1250 BC) pottery production is characterized by 11 ceramic classes (Manuelli 2013: 82–89): mineral fabric and medium texture kitchen ware (A1), mineral fabric and coarse texture kitchen ware (A2), vegetable fabric and medium texture storage ware (B1), mineral-gritty fabric and medium texture common ware (C1), mineral-gritty fabric and medium texture orange common ware (C2), mineral-gritty fabric and semi-coarse texture common ware (C3), mineral-sandy fabric and fine texture ware (D1), mineral-sandy fabric and semi-fine texture ware (D2), mixed fabric and semi-coarse texture storage ware (E1), mixed fabric and coarse texture storage ware (E2), and mineral fabric and extremely-fine texture red ware (F1). The production is in general wheel-made and fired at well-controlled atmospheres. Pale color clays are mostly used, and surfaces are mainly characterized by a light and hasty smoothing. Decorations are rather common: painting with geometric patterns, red-decorative slips, and irregular drip marks have been identified on many of the abovementioned wares (Manuelli 2022; Fragnoli and Rodler 2022). In general, the shape assemblage is characterized by the presence of elements that originate from the local Bronze Age tradition and the appearance of aspects related to the Hittite central Anatolian repertoire of the fourteenth and thirteenth century BC. A few more words should be said about Ware F: this is comparable to the so-called Red Lustrous Wheel-made Ware, a well-known and widely studied production mostly attested in northern Cyprus as well as southern and central Anatolia between the fifteenth and the thirteenth century BC (Mielke 2007; Kibaroǧlu et al. 2019).

The Early Iron Age (Fig. 3 )

Fig. 3
figure 3

Early Iron Age ceramic wares groups: a Ware group K; b Ware group P; c Ware group S; d Ware group F (photos ©MAIAO)

Full size image

The Early Iron Age (ca. 1250–850 BC) pottery production is characterized by 8 ceramic classes (Manuelli 2018): mineral fabric and medium texture kitchen ware (K1), mixed fabric and coarse texture kitchen ware (K2), mineral fabric and semi-coarse texture storage ware (P1), vegetable fabric and coarse texture storage ware (P2), vegetable fabric and medium texture common ware (S1), mineral fabric and semi-fine texture common ware (S2), mineral fabric and semi-coarse texture common ware (S3), and mineral fabric and fine texture ware (F1). The production is in general still characterized by the use of pale color clays, but firing must have now often taken place in a not very well-controlled atmosphere. Moreover, despite the fact that wheel marks are visible on nearly every sherd, it is possible that at least part of the production was only wheel-finished. Surfaces are exclusively poorly smoothed or self-slipped, decorations are virtually absent, and the assemblage is in general characterized by the total lack of any supra-regional ware. The Early Iron Age pottery shapes show on the one hand an interesting continuity of the Late Bronze Age tradition and on the other hand the introduction of new aspects typical of northern Syrian and the Levantine regions (Manuelli 2020: 118–121).

The Middle Iron Age (Fig. 4 )

Fig. 4
figure 4

Middle Iron Age exogeneous ceramic wares: a Ware RS (Red Slip Ware); Ware GW (Grooved Ware); c Ware IN (Incised Ware); d Ware RB (Red Burnished Ware); e Ware AL (Alişar IV Ware) and Ware AB (Bichrome Alişar IV Ware); f Ware CP (Cypro-Phoenician Ware)

Full size image

The Middle Iron Age (ca. 850–700 BC) repertoire is composed of 7 exogenous productions (Frangipane et al. 2020: 92–98). Red Slip Ware (RS) consists of mineral fabric, semi-fine texture, and orange color paste with a surface characterized by a red/orange layer accurately burnished. It is considered one of the hallmarks of the Levantine region from the ninth to the seventh century BC (Soldi 2013). Grooved Ware (GW) is a mixed fabric, semi-coarse texture, and gray color paste production with a set of grooved lines deeply incised. It is typical of the Iron Age in eastern Anatolia that appears along the western side of the Euphrates from the ninth century BC (Blaylock 2016: 15–20). Incised Ware (IN) shows mineral fabric, medium texture, and buff color paste with oblique or crossed lines incised on top of a red-slipped band. Again, in the Upper Euphrates region, it is mainly attested in the Middle Iron Age (Ökse 1988: 105–108). Red Burnished Ware (RB) consists of mineral fabric, semi-fine texture, and buff color paste with surfaces covered by a bright glossy red burnished slip. It is typical of the Urartian settlements in eastern Anatolia and found all the way to the Euphrates river (Batmaz 2020). Alişar IV Ware (AL) presents mineral fabric, medium texture, pale color paste, and a brown linear painted decoration applied over a thick white slip. Bichrome examples of Alişar IV Ware (AB) have mineral fabrics, semi-fine textures, and pale color pastes with white slipped surfaces and red and black-brown geometric paints. Alişar wares are typical of the Middle Iron Age in south-central Anatolia (d’Alfonso et al. 2022). Cypro-Phoenician Ware (CP) shows mineral fabric, semi-fine texture, pale color paste, and white slipped surfaces decorated with black-brown and/or red linear painted motifs. It suggests affinities with the so-called White Painted and Bichrome wares attested in the Cypro-Phoenician world from approximately the mid-ninth century BC (Gilboa 2015).

The geological setting

Arslantepe (Fig. 5) rests upon sediments from the Miocene era that were deposited in a lake setting and primarily comprise calcareous clays, sandstones, and limestones (Palmieri AM 1978b). Less than 1 km to the northeast, we find the Orduzu volcanics of the Middle Miocene with quartz-micromonzonites, basaltic trachyandesites, trachyandesites, and rhyolites (Önal 2008; Önal et al. 2008). The Baskil magmatics (Late Cretaceous) and the Yüksekova/Elazığ units (Maastrichtian to Early Eocene) crop out 5 to 6 km eastward. This area is dominated by a variety of intrusive and volcanic rocks, such as rhyolites, dacites, andesites, basaltic andesites, monzonites, tonalites, diorites, and gabbros (Yazgan and Mason 1988; Yiğitbaş and Yılmaz 1996). The extensive formations of the Anti-Taurus emerge 7 to 10 km south of Arslantepe. The Malatya metamorphics (Carboniferous-Triassic) occupy the western part of these units with meta-cherts, meta-clastic rocks, slates, phyllites, mica schists, and meta-carbonate rocks (Bozkaya et al. 2007; Robertson et al. 2006). The dominant features in the eastern portion consist of the Ispendere ophiolites (Late Cretaceous) and the Maden Complex (Middle Eocene). The former showcases an undisturbed sequence of ophiolites that have been intruded by granites (Parlak et al. 2012), whereas the latter represents a volcano-sedimentary sequence encompassing andesites, altered basalts, cherts, radiolarites, spilitic lavas, mudstones, limestones, sandstones, and conglomerates (Şaşmaz et al 2014).

Fig. 5
figure 5

Geological map of the Malatya-Elazığ region with indicated location of Arslantepe (based on the Turkish Geology Map by MTA (2002) at scale of 1:500,000)

Full size image

Materials and methods

The analysis has mostly taken into account ceramic vessels belonging to the Late Bronze Age (n = 36) and Early to Middle Iron Age (n = 45+18) in order to inspect elements of continuity and discontinuity between the two periods (Table 1). Sampling has been done to cover all the ware classes identified macroscopically, using a strategy that aimed to balance the number of samples per wares or, if not possible, per main group of wares according to their recurrence, availability, and possibility of sampling. The samples were analyzed through thin-section petrography (n = 96) and XRFWDS geochemistry (n = 24). As in previous papers, the emphasis has been put on thin-section petrography as more indicative of raw material procurement and processing due to both the predominance of coarse fabrics and the variety of the local geology (Fragnoli and Palmieri 2017; Fragnoli 2018). On the other hand, geochemical analyses have been more consistently performed on samples dating to the Middle Iron Age, considering the presence of exogenous productions in this period and the necessity to determine ceramic provenience and potential imports. In order to avoid biases related to the general limited number of samples, especially those analyzed geochemically, the data from the previous chronological phases have been considered in the study for comparison (Fragnoli and Palmieri 2017; Fragnoli 2018, 2019a, 2019b; Fragnoli and Liberotti 2019).

Thin sections were examined under the polarizing microscope LEICA DM 2700P and grouped according to textural, mineralogical, and technological features. Comparative charts were utilized to quantify the main components (Rice 1987: 348). Geochemical analyses were conducted with the wavelength dispersive X-Ray fluorescence spectrometer PANnalytical AXIOS of the Archea Laboratory in Warsaw. Samples were treated with a corundum polishing machine to remove any contaminated surfaces, washed using distilled water and ultrasounds, and subsequently stored in a drying oven for 24 h before being transferred to a desiccator. Subsequently, the samples were ground using an agate mill (Fritsch Pulverisette Null) and left to dry once more. Afterward, a quantity of 1.5–2 g of powder was taken from each sample and subjected to ignition at 900 °C. The resultant material was then melted using a lithium-borate mixture (Merck Spectromelt A12) and cast into small discs measuring 32 mm in diameter. The concentrations of all major elements and of 14 trace elements were determined.

Thin-section petrography

The analyzed samples were classified into 16 petrographic groups based on several criteria, including the presence or absence of vegetal temper, the nature of the clay matrix (low- or high-calcareous content), and the geological origin of inclusions (Fig. 6, Table 1). Table 2 reports the detailed mineralogical and textural features of each group, named according to an acronym that indicates the predominant components as explained in the caption of the table. Most of the samples (89.6%) exhibit a high-calcareous matrix, while about half of them (44.8%) are vegetal tempered. The shape of voids left by the burned-out vegetal temper is generally not assignable to agricultural by-products, such as spikelets, glumes, caryopses, or straw. Mineral and rock inclusions are predominantly of intrusive (62.5%) and to a lesser degree of volcanic (36.5%) and metamorphic origin (32.3%). Intrusive and volcanic rocks range from basic to acidic, while metamorphic rocks refer to either ocean-floor or low to high-grade regional metamorphism. Sedimentary rocks, such as sandstones, mudstones, and limestones, appear only in a negligible number of samples. The textural features are quite variegated as well, with maximum grain sizes and incidences between 0.3 and 7 mm and 7 and 30% respectively. Most of the ceramic pastes were tempered, as demonstrated by the polymodal grain sizes, clustered distribution, and angular shapes of inclusions.

Fig. 6
figure 6

Thin-section microphotographs under XP of petro groups: a VCEm-a + Ib; b CEm-a + Ib; c CIb; d VCEm-a; e VCEa; f CEa; g CEb-m; h CIa + Ib; i VCIa; j Ia + Ib; k VCMgne + metag + Ib; l and m CMgne + metag + metaub; n Mgne + serp + metag + sc; o Q. Field of view = 6.85 mm (a; b; c; e; f; g; l; and o) and 4.8 mm (d; h; m; and n)

Full size image
Table 2 Inclusions, clay matrixes and pores of the petrographic groups related to the Late Bronze to Middle Iron Age vessels from Arslantepe
Full size table

As in the prehistoric phases, the Late Bronze and Iron Age assemblage from Arslantepe displays a wide textural and geological variety (Fragnoli and Palmieri 2017). Among the large spectrum of recipes, 9 out of 17 (46%) were already used in the Late Chalcolithic and Early Bronze Age phasesFootnote 1. Most of them—i.e., petro groups CEam + Ib, CEm-a, CIb, VCEa-m + Ib, VCEm-a, and VCIb—typically occur in the light-colored wheel-finished vessels dating to the Late Chalcolithic 3 to 5 (periods VII and VIA of the site sequence). Clay matrixes are calcareous and tempered with trachytic lavas and/or microgabbros from the Baskil and Yüksekova complexes available 6 to 15 km east of the site. Both vegetal-tempered and non-vegetal-tempered variants exist. Two other calcareous recipes, both tempered with rhyolitic lavas from the Orduzu suite (900 m east from the site) but differing in the use or non-use of vegetal temper (VCEa and CEa), first appeared in the Early Bronze Age 1 Handmade Red-Black Burnished Ware (Periods VIB1-2). One last calcareous recipe distinguished by basaltic andesites and altered clinopyroxenes typically occurs in the Early Bronze Age 2 and 3 (Periods VIC-D) handmade painted vessels.

Forty-one out of 96 samples show recipes that are not attested in the previous phases of the site sequence (Table 1). Most of these (i.e., 24 samples) exhibit a large variety of metamorphic rocks, such as gneisses, serpentinites, metagabbros, metapyroxenites, metabasalts, and quartz- and mica schists. These lithologies are available 7 to 10 km south and southeast of the site within the Ispendere ophiolites, Maden Complex, and Malatya metamorphics. Three distinct recipes—i.e., VCMgne + metag + Ib; CMgne + metag + metaub; and Mgne + serp + metag + sc—were identified based on the incidence of each type of rock, the presence of vegetal temper, and the high- vs. low-calcareous nature of the clay matrix. Fourteen other samples are made of three recipes—i.e., VCIa, CIa + Ib, and Ia + Ib—that share granitic rocks as the main lithic inclusion, but differ in the presence or absence of microgabbros and vegetal temper as well as in the calcareous nature of the matrix. A procurement from the eastern Baskil and Yüksekova complexes, which also exhibits intermediate to sialic intrusive rocks besides gabbros and trachytic laas, is assumable. Three samples exhibit a fine-grained coral-red recipe (i.e., Q) almost totally dominated by quartz inclusions, which are not diagnostic in terms of provenance.

Discussion of the petrographic results

The distribution of samples into petro groups (Table 3) evidences elements of both continuity and change between the Late Bronze and Iron Age. Thirty-nine percent of the samples share recipes—i.e., CEb-m, VCEa-m + Ib, CEa-m + Ib, CIb, and Ia + Ib— that invariably occur in both periods. At the same time, the Iron Age marks the disappearance and appearance of old and new recipes in respectively 20% and 41% of the samples. One new recipe (petro group VCMgne + metag + Ib) now clearly predominates over both the new and older ones. In general, the Early Iron Age represents a break in the strategies of raw material procurement, with respect to both the previous and subsequent phases. Indeed, the eastern igneous deposits are mostly exploited across all the phases except during the Early Iron Age, which shows a gravitation towards the southern metamorphic and ophiolitic deposits (Fig. 7).

Table 3 Seriation table of petro groups and summarizing table (below) evidencing continuity and changes between the Late Bronze and Iron Age
Full size table
Fig. 7
figure 7

Exploitation of southern metamorphic and ophiolitic sources vs. exploitation of eastern igneous sources from the Late Bronze to the Middle Iron Age

Full size image

Interesting trends emerge when considering the Late Bronze Age and Iron Age separately in relation to the respective repertoire of wares and/or shapes. In the Late Bronze Age, the same wares are produced with different recipes, and conversely, the same recipes occur in different wares. The only exceptions are the Red Lustrous Wheel-made Ware F1 as well as the painted examples of the semi-fine D2 and storage E1/E2 wares, which are realized with specific recipes. The recipes recurring in the painted wares—i.e., CEm-a, CEb-m, and VCEm-a—date back to the prehistoric tradition, while the Red Lustrous Wheel-made Ware is made of a peculiar recipe, namely the abovementioned petro group Q. This latter fits with the fine fabric group reported for most of the Red Lustrous wares found across the eastern Mediterranean and interpreted as a Cilician production based on the latest archeological and archaeometric evidence (Kibaroğlu et al. 2019). Most of the petro-loners (Fig. 8) present the same intrusive and metamorphic local lithotypes as those mentioned above, but differently combined (Table 4). The only exception is the flat bowl 347/15 (Fig. 8g), distinguished by peculiar minero-petrographic associations not observed in other samples yet, i.e., calcareous sandstones and intrusive intermediate rocks.

Fig. 8
figure 8

Thin-section microphotographs under XP of petro-loners: a 241/15; b 325/15; c 331/15; d 306/15; e 355/15; f 342/15; g 347/15; h 350/15. Field of view = 6.85 mm (b; d; e; and h) and 4.80 mm (a; c; f; and g)

Full size image
Table 4 Inclusions, clay matrixes, and pores of the petrographic loners related to the Late Bronze to Middle Irone Age vessels from Arslantepe
Full size table

By the Early Iron Age, cooking pots started being produced with standardized paste preparation modes, as indicated by the predominance of petro group VCMgne + metag + Ib, while other shapes still exhibit variegated recipes. During the subsequent Middle Iron Age, the decorated wares of the allochthonous tradition—such as the Cypro-Phoenician and Alişar IV painted wares as well as the Syrian Red Slip and Urartian Red Burnished wares—are mostly locally produced with recipes that are well-documented along the previous site sequence (petro groups CIb, VCEm-a + Ib, Ib, CEm-a + Ib, and CEb-m). In particular, specific recipes recur, on the one hand, in the Urartian and Cypro-Phoenician wares (petro groups VCIb and CIb) and, on the other hand, in the Red Slip Ware (petro group CEm-a + Ib). However, the only Grooved Ware carinated bowl here analyzed (sample 350/15) and 1 of the bichrome Alişar IV Ware jars (sample 355/15) could not be classified into any groups, which likely reflects a foreign origin (Table 4). Both samples are non-calcareous with a medium-coarse texture; the Alişar IV Ware jar is dominated by clinopyroxenes and altered mafic minerals (Fig. 8e), while the Grooved Ware carinated bowl is rich in metapyroxenites, vegetal temper, and clinopyroxenes with exsolution and replacement textures (Fig. 8). The minero-petrographic association of both samples indicates the ophiolitic nature of the supply sources. Similar sources are widely reported in Turkey, especially along the Ankara-Erzincan and Bitlis-Zagros suture zones (Fig. 9). Considering the regions where these 2 ware classes are mostly produced and spread, a general link with the Ankara-Erzincan zone could be assumed. Locally produced and imported Alişar IV Ware vessels also differ by the surface treatments: the surfaces of the former do not exhibit any slipping layer, while those of the latter are covered by a thick regular and calcareous slip (Fig. 10).

Fig. 9
figure 9

Distribution of ophiolites in correspondence of the suture zones (Sarifakioglu et al. 2014)

Full size image
Fig. 10
figure 10

Distinct surface treatments between locally produced (a) and imported (b) Alişar IV Ware sherds. Field of view: 0.9 mm

Full size image

XRFWDS geochemistry and comparison with petrography

Upon relating the various chemical elements (Table 5) in binary plots, 3 distinct groups emerge (Fig. 11): the largest group includes both Late Bronze and Iron Age samples, while the 2 minor ones are each composed of only 3 samples mostly dating to the Late Bronze Age 1–2 phases. The first minor group includes a Late Bronze Age 1B-2 arm-shaped vessel (329/17), a Late Bronze Age 2 pedestal base (332/17), and a Late Bronze Age 2 high-necked small-sized bottle (336/17), all distinguished by higher Al2O3, Fe2O3, and Rb and minor MnO values. The sample 336/17 further differs by higher Zr and Rb contents. Unfortunately, thin sections could not be prepared from these samples due to material scarcity. The second minor group is comprised of a Late Bronze Age 2 neckless medium-sized cooking pot (327/15), a Late Bronze Age 1B short-necked medium-sized cooking pot (331/15), and an Early Iron Age flat bowl (305/17), which are richer in SiO2 and poorer in CaO, Ba, Cr, and Sr. Although these 3 samples have different minero-petrographic associations, they all share the low-calcareous nature of the clay matrix and the presence of felsic rocks (e.g., granites, gneisses, quartz-schists, trachytic to rhyolitic lavas). The sample 327/15 is further distinguished by higher Mn, V, and Ti values, which fit with the occurrence of ophiolitic rocks observed under the polarizing microscope.

Table 5 Major (wt. %) and trace element (ppm) concentrations. Elements between brackets are under the detection limit
Full size table
Fig. 11
figure 11

Binary plots showing the distinct concentrations of some elements for the samples 329/17, 322/17, 336/17, 337/15, 331/15, and 305/17

Full size image

By adding the data of locally produced samples from the previous phases, we notice that all the Late Bronze-Iron Age samples, including those of the 2 minor groups, fall into a typical local range (Fig. 12). This also concerns the 2 minor groups that show chemical affinities with Late Chalcolithic 4 to Early Bronze Age 1 cooking and Red-Black Burnished Ware vessels generally realized with metamorphic raw materials.

Fig. 12
figure 12

Elemental binary plots showing the overlap between Late Bronze-Middle Iron Age sherds and the locally produced vessels of the previous phases

Full size image

To identify potential variations and similarities among the samples, a principal component analysis (PCA) was conducted using the SPSS software package. Raw chemical values were adjusted to a constant sum of 100%, logarithmically transformed to base 10, and standardized to provide a roughly equal weight for all elements during calculation (Baxter 1994, 1995, 2003a, 2003b; Glascock 1992). CaO, Na2O, and P2O5 were excluded from statistical analyses due to their susceptibility to post-depositional alterations (Freestone et al. 1985; Schwedt et al. 2004). The same 3 groups of samples mentioned above emerge in the PCA plot that relates the first and second factors (Fig. 13). These account for 66.6% of the total variance and consider Al2O3, -Ni, Fe2O3, MgO, SiO2, -Sr, -Cr, TiO2, V and -K2O, Ba, MnO, Rb, respectively. Both the first and second minor groups differ from the rest of the samples by higher values of factor 1, which well reflect the presence in thin sections of both mafic and sialic rocks. As for factor 2, the first minor group is poorer, and the second richer. Within the largest group, which includes vessels of any wares, shapes, and periods, a subgroup of three samples now stands out due to lower values of factor 2. This subgroup only includes Late Bronze Age 1-2 necked pithos with drip marks, which do not show any distinct petrographic features compared to other samples though. This might presumably indicate the use of distinct local clay sources for this ceramic class.

Fig. 13
figure 13

Scatter plot of the principal component analysis relating factor 1 (47.2%: Al2O3, -Ni, Fe2O3, MgO, SiO2, -Sr, -Cr, TiO2, V) to factor 2 (19.4%: -K2O, Ba, MnO, Rb)

Full size image

Conclusions

The results obtained through thin-section petrography and XRFWDS geochemistry evidenced both elements of continuity and discontinuity between the Late Bronze and Iron Age. The elements of continuity concern the use of recipes made of volcanic raw materials available east of the site and already attested by the prehistoric phases. These recipes typically recur in the Late Bronze and Middle Iron Age painted productions, while unpainted vessels are distinguished by more variable and thus less specialized recipes. On a wider geographical perspective, it is interesting to note that this result contrasts with the evidence of the Late Bronze Age painted pottery from Oymaağaç Höyük/Nerik, Central Black Sea, which exhibits instead a large petrographic and geochemical variability (Mielke et al. 2021).

The most evident changes in the local production marked by the Iron Age consist in a much stronger gravitation towards the southern Anti-Taurus mountains for the procurement of raw materials, the appearance of new recipes in the common wares, the standardization of paste recipes for the cooking pots and, more generally, a trend towards a compositional homogenization. In this framework of changes, it is important to stress once again that widespread transformations and the introduction of new sets of inter-cultural connections with regions located south of the Malatya area are also clearly visible in the shape repertoire of the pottery assemblage since the very beginning of the Iron Age. At the same time, a strong standardization in terms of the decrease in shape varieties and macroscopic identification of fabrics and wares characterizes the entire Early Iron Age collection when compared to the previous Late Bronze Age. It is nevertheless worth mentioning that the morphological standardization does not unequivocally correspond to more standardized recipes, for example, as in the case of the flat bowls that were produced with at least 5 different recipes. The evidence of cooking pots recalls instead a phenomenon that took place at different times in the whole eastern Mediterranean, namely the emergence of specialized cooking pottery centers (Ben-Shlomo 2019). This is often accompanied by changes in paste preparation and vessel shaping, although the role that food-related practices had on these changes or, conversely, the impact that these changes had on food-related practices still remain poorly investigated.

Further differences concern the incidence of imports in relation to foreign wares, following a pattern that mostly reflects what has been shown by the archeological evidence. During the Late Bronze Age, the only foreign ware is the Red Lustrous Wheel-made Ware, which is always imported from Cilicia. In the Early Iron Age, both foreign wares and imports are lacking. The Middle Iron Age is in contrast distinguished by several wares of foreign tradition, which are mostly locally produced. However, some nuances are still discernible within this general trend: while there are a few imports from south-central and eastern Anatolia (e.g., Alişar IV and Grooved wares), the wares of the southern tradition (e.g., Red Slip and Cypro-Phoenician wares) are exclusively locally made. This is a noteworthy aspect that might further emphasize the fact that contacts with the regions south of Malatya were well-established since the end of the Late Bronze Age, while eastwards and westwards these became more irregular and probably did not affect the local production pattern. At any rate, our analysis also demonstrates that Alişar IV Ware was not only the product of a few centers in south-central Anatolia (d’Alfonso et al. 2022) but could also be produced far away from this region. Moreover, as already documented at an intra-regional level (d’Alfonso et al. 2022), the analysis has further shown that Alişar IV Ware local imitations and imports differ by surface treatments and decoration also at Arslantepe.

The abovementioned changes could also be briefly considered within a wider socio-economic, cultural, and historical perspective. The occurrence of imported Red Lustrous Wheel-made Ware specimens in Late Bronze Age 2 contexts confirms the gradual involvement of Arslantepe in the mechanism of expansion of the Hittite empire and further highlights interesting connections between its peripheral regions. With the collapse of the Late Bronze Age centralized system and the emergence of the Early Iron Age regional states, new trajectories of contacts and influences occur. Connections with regions located south of the Malatya area are manifested in both the procurement of raw material and the pottery repertoire. These are clearly related to the growing political and cultural role of Karkemiš during the last centuries of the second millennium BC. At the same time, the standardization of some paste recipes and shapes might imply the existence of a strong local power able to impose control over the means of pottery production. The Middle Iron Age shows instead the emergence of a new and vivid cultural environment, placing Arslantepe at the center of a renewed system of extra-regional exchange, as is shown by the presence of several wares belonging to foreign traditions. This trend reflects the multicultural nature of the Syro-Anatolian world at the beginning of the first millennium BC.

In conclusion, this study demonstrates the importance of raw material procurement and processing as indicators of socio-economic and cultural changes over the longue durée. Coupled with preliminary observations on surface treatments and decorations, our data have also better clarified the impact of exogenous traditions on the local ones. In this perspective, future analyses will focus on exhaustively characterizing the materials, tools, and gestures involved in burnishing, slipping, and painting the surfaces of these same ceramic assemblages.