TRACING INTERACTION ON SAMOS ISLAND: POTTERY TRADITIONS AND CONNECTIVITY AT KASTRO-TIGANI AND THE HERAION DURING THE AEGEAN LATE–FINAL NEOLITHIC/WESTERN ANATOLIAN MIDDLE–LATE CHALCOLITHIC

This paper presents the results from the pilot analytical study of the Aegean Late–Final Neolithic/Anatolian Middle–Late Chalcolithic (c. 5500–3200/3000 BC) pottery from the Kastro-Tigani settlement, located in south-east Samos (east Aegean). In addition to Crete, the region of the insular eastern Aegean has produced the earliest evidence for Neolithic habitation. The archaeological traces at Kastro-Tigani are so far the earliest known on Samos, being partly contemporary with the recently discovered Middle–Late Chalcolithic layers at the Heraion, lying in close proximity to the former site, and at the Agriomernos cave (Megalo Seitani) in the north-west part of the island. The re-evaluation of the ceramic assemblage from Kastro-Tigani has led to the laboratory analysis of 34 samples, using a combination of thin-section petrography and Wavelength Dispersive X-Ray Fluorescence Spectroscopy, in order to determine provenance and establish reference groups for the earliest local production on Samos. This first compositional characterisation of the pottery contributes new data in a relatively under-studied region and provides grounds for comparison with analytical results from the Heraion, with the aim to investigate possible relations between the sites. Hence, the identification of different strategies in pottery production, reflected in the overall distinct fabric and chemical groups, further indicates the practice of several production units and the exploitation of various raw material sources at the Pythagoreion/Chora plain. Isolated examples of possible imported ceramic vessels, as well as exotica (e.g. obsidian, acrolithic and Kilia figurines, ring-shaped features, marble vessels, kratiriskoi) are highlighted as markers of macro-scale interaction in the context of Aegean early maritime connectivity.


INTRODUCTION
Samos, located off the coast of western Anatolia, is one of the few Aegean islands that provide published evidence for its inhabitation since the mid-sixth millennium BC, while the archaeological evidence gathered so far from most of the other islands suggests that the majority were first inhabited in the late fifth and fourth millennia BC (Kouka ; Schwall , -, figs -, with further references) (Fig. ). Permanent settlements on the east Aegean islands, as well as the long-term or periodic use of caves appeared from the Aegean Late- The research focus so far and the discovery of various sites in the eastern lowlands is also due to the less dense vegetation in the area, compared to the western lowlands and uplands of Samos. According to a small-scale research conducted by Shipley in the s in the western part of the island (in the areas of Karlovassi and Marathokambos), there seems to be a diachronic geographical shift of the population density from the Roman period onwards (Shipley , -). More systematic traces appear since the Early Byzantine period (Shipley , maps -). Other stray, prehistoric traces in south-east Samos include a Late Mycenaean chamber tomb at Myli, and finds at Kavo Phanari/Cape Fonias and Mesokambos (Shipley , -, , -; Kouka , -, map ; Kouka and Menelaou , fig. ). The presence of volcanic outcrops, which could be used as a source for lithic materials, and the location of Samos in proximity to major river valleys agree with the suitability model for Pleistocene-Early Holocene exploitation of the island (Tsakanikou, Galanidou and Sakellariou ). The hitherto chronological gap between the NL and EBA at Kastro-Tigani was bridged through the recent excavations at the Heraion (Kouka ; , figs -; Kouka and Menelaou , , table ). Chalcolithic architecture at Heraion Phase  was unearthed in limited areas among and beneath house foundations of the EBA and MBA and is documented only by roof clay, wooden beams, a hearth and sea-pebble floors (Kouka , , fig. ; Kouka and Menelaou forthcoming). The stylistic study of the pottery suggests a division into two sub-phases, Heraion b and a, that correspond to Tigani III and IV and Miletus Ia-b (Kouka b, , fig The pottery of the MCh-LCh in the eastern Aegean/western Anatolia exhibits technological uniformity in macroscopic terms. With respect to fabrics and shape repertoire it is usually interpreted as homogeneous with common features being the coarse, vegetal-tempered clay pastes and the dark brown-black or red burnished surfaces with areas of discolouration, due to the fast, low-firing procedures most likely occurring in open-air constructions. The pottery assemblage from Kastro-Tigani is representative of the east Aegean region. However, despite some stylistic links with the contemporary site of Emporio on nearby Chios (overlapping roughly with phases X-VI, Hood , -, -, pls -), one of the most immediate sites for comparison with Kastro-Tigani, there are considerable differences between them, especially in Phase Tigani IV/Emporio VII-VI (Felsch , -). The morphological and stylistic features of each phase are presented in detail in the final publication of the site (Felsch ).
A number of surface treatments (wares) have been distinguished at Kastro-Tigani (Furness , -; Felsch , -). These are the black-burnished ware, with technical variations in the quality of the burnish; pattern-burnished, with narrow or wide linear stripes on a glossy background; white-painted or white-on-dark particularly attested on bowls; red slipped, black to reddish grey smoothed and plain wares corresponding to coarse vessels. Incised and punctured motifs and/or relief decorative elements appear on various vessel types but are particularly more common on black-burnished ware (Figs -).
Phase Tigani I Tigani I comprises hemispherical bowls (rounded; flaring-rimmed; S-curved; carinated; straightsided), hole-mouth jars, conical-necked jars, funnel-mouthed jars and wide-mouthed pithoi, the first so-called cheesepots, closed vessels with horned lug-handles, crescent-shaped lugs or horizontal tubular lugs, commonly with white-painted decoration (Felsch , -, pls -) (Fig. ), that in the eastern Aegean varies from site to site (Furness , , fig Sotirakopoulou (, -) previously suggested that this tradition originated and spread from south-west Anatolia towards the west Aegean, through Chios and Samos, the current distribution in western Anatolia, the Aegean islands, and Mainland Greece suggests that this decoration was a rather common feature in the Aegean during the fifth-fourth millennia BC (Schwall , -, fig. ). Mat-impressed SERGIOS MENELAOU AND OURANIA KOUKA  pottery, another class used as a marker of interaction during the fifth-fourth millennia BC (Horejs and Schwall , -), is sparsely found at Tigani I-II (Heidenreich -, , pl. :-).

Phases Tigani II-III
Tigani II comprises open vessels such as conical carinated bowls or bowls with in-turned rims, horned lugs and knobs or prong handles (Felsch , , pl. :, no. ), wide bowls with upraised knobbed or twisted handles, double-handled vessels, and other features (Felsch , -, pls -, :,, :) (Fig. ), with close parallels at Emporio X-VIII on Chios (Hood , -, fig. :-) and Vathy on Kalymnos (Benzi , -, figs -; , pl. d-e), as well as the Troad region (cf. Blum , , figs -). Dark-burnished surfaces and incised or relief decoration are common and distinguish this phase from the previous. The conical bowls and closed vessels with horned lugs continue in Tigani III, while new types include pedestaled bowls, cups, pithoi, various amphora types, globular jars with tapered or collar necks, and cheesepots (Felsch , -, pls -) (Fig. ). Variability is also observed in the surface treatments (black-burnished; red slipped and burnished; incised wares). Common decorative motifs on jars of Tigani II-III (Heidenreich -, pls :-, :-; Felsch , pls : and :, : and :, : and :) include horizontal incised triangles filled with hatchings or punctured dots (pointillé) and find close parallels at Agios Petros on Kyra Panagia in the Sporades ( ), similarly to other contemporary sites in the north-east Aegean. There are also vessel shapes which indicate strong influences from other regions. For instance, the rolled-rim bowl, sparsely found at Kastro-Tigani and the Heraion, is thought to reflect an Anatolian influence (Heidenreich -, pl. :; Sotirakopoulou , -; Kouka , ). In this phase, beakers, i.e. elongated/tapering conical vases with a flat base and two symmetrically opposed vertical perforated lugs below the rim (Felsch , pls : and :-, :-,, :V-,,), and conical bowls with vertically elongated pierced lugs (Felsch , pls :- and :V, : and :), make their appearance in marble and clay, which was thought to reflect Cycladic influences (Fig. no). However, marble conical beakers/ rhyta, with a pointed base, are known from several MCh sites in coastal and inland western Anatolia (Ş ahoglu and Sotirakopoulou , , cat. nos -; Takaoglu (Gerber , fig. A) in western Anatolia. Similar vessels with a more tapering body were recently uncovered at Yeşiltepe in inland western Anatolia (Takaoglu and Bamyacı ).
Whether the EBA I flat-based beakers, which are also imitated in clay at Tigani IVb (Felsch , pls :-, :) and Iasos (Pecorella , pls XXXVIII:, XXXIX:), reflect a long process of importation from the Aegean to western Anatolia, technological transfer, or hybridisation is unclear and requires a larger-scale systematic study. However, it is noteworthy that the late-fourth-millennium BC Aegean examples differ substantially from the mid-fifthmillennium BC Anatolian examples, which Takaoglu (, ) has described as probable heirlooms in later deposits. Such vessels were most likely of special use and symbolic value, given their rarity, size, and the labour invested in their manufacture. Furthermore, their morphological features in addition to the vertical lugs and early pointed base further strengthen such a claim. While suspension of the pointed-based beakers was a functional requirement, the later flat-based vessels suggest the continuation of their social significance in the social display of such high-valued containers, most likely for the use of liquids (Stroulia ).
Another highly distinctive ceramic type of this phase, the clay collared jar with truncated conical neck, known at Tigani IVb ( Finally, a number of non-diagnostic ceramic fragments of Tigani IVb, originally described as rhyta (Heidenreich -, -, pls :-, :-,; Felsch , -, , nos , , -, pl. :,,,) (Fig. h), are likely to represent quadrangular vessels on legs with a flat or concave surface. Decoratively (deeply incised spiral and meander motifs) and morphologically similar vessels have been invariably described as pedestaled stands, libation or cult tables, and models of altars (Furness , ; Terzijska-Ignatova  for further discussion). Although their function is uncertain (ritual vessels, liquids or pigments containers, incense-burners have been suggested), parallels from northern Greece and the Balkans, south-east and central Europe, and less commonly Anatolia during the LCh period are perhaps suggestive of the circulation of these objects or of the ideological transfer related to their use. Alternatively, the Samian examples are linked with the scoop type from Kalymnos (Benzi , , -, pls m, b).

CHRONOLOGICAL CORRELATIONS BETWEEN TIGANI III-IV AND HERAION PHASE 
The pottery from Heraion Phase  largely correlates with Tigani III-IV. Phase Heraion b dates to the Anatolian MCh (c. - BC) and includes coarse, orange, reddish brown to black coated and/or burnished hemispherical bowls with a thickened rim, conical bowls with pattern-burnished decoration, a low pedestaled bowl with slotted openings, jugs with incised handles, horned handles from jars with conical neck, and collar-necked cooking jars with globular body and vertical relief applications (Fig. ).  The LCh phase Heraion a includes deep bowls with inverted or everted rims, pyxides with vertically pierced lugs, cut-away spouted jugs with incised handles, collar-necked jars with narrow vertical or everted rims and unpierced lugs, as well as cooking pots with a pair of plastic knobs beneath the rim, or tripod ones with rectangular and trapezoidal legs, and cheesepots The pottery of Heraion b-a is comparable with that found at Tigani III-IV, Emporio VII-VI, Çukuriçi Höyük VII, Liman Tepe VII, Miletus Ιa-b, and Beysesultan LCh - and indicates a  Table ).
Re-evaluation of the published material from the old excavations in the south-west part of the settlement has allowed the identification of scattered sherds, previously falsely ascribed to the EBA period, that are diagnostic for pattern-burnished ware (Milojčić 

TRACING INTERACTION ON SAMOS ISLAND
 culture in Thessaly ( Fig. a-b). Finally, a fragment of a bone pendant or figurine from Tigani III early (Felsch , , cat. no. V, pl. :-; Fig. d ) in the form of western Anatolian marble figurines of the Kilia type  indicates the sharing of a symbolic code between the insular and coastal communities of the eastern Aegean.

ANALYTICAL METHODOLOGY
The systematic macroscopic examination of the NL-Ch ceramic assemblage of Kastro-Tigani, covering the whole range of typological and morpho-stylistic classes, defined a number of Macroscopic Fabric Groups (MFG), which further led to the selection of samples for laboratory analyses. The main technique employed in the analysed material is thin section petrography, combined with macroscopic observations, contextual/chronological information, refiring tests, and elemental analysis by Wavelength Dispersive X-Ray Fluorescence Spectroscopy (WD-XRF). Raw material samples collected from around the island, as part of a geological survey, were also taken into consideration in the determination of provenance, further supported by comparing their mineralogy with local geological maps (Theodoropoulos ), as discussed below. The selected ceramic samples are presented in Table , with information about the archaeological context and morpho-stylistic classification.
Standard petrographic thin sections of  samples were prepared at the Fitch Laboratory, British School at Athens, and examined with the polarising light microscope by Menelaou. Having been grouped into fabric classes, based upon the nature of their dominant non-plastic inclusions, clay matrix and textural features, these groups were described using the system proposed by Whitbread (, -; ). The petrographic analysis aimed at characterising the raw materials used in pottery production and, thus, determining the potential provenance of the pottery. Equally significant was the reconstruction of craft choices, manufacturing traditions and technological changes, allowing us to address issues relating to pottery production organisation. Not all stages of production (forming, finishing, firing) are reconstructed in as much detail as the exploitation, procurement and processing of the raw materials used for making the ceramic pastes.
All samples were subjected to refiring tests in controlled conditions for one hour (oxidising atmosphere) at temperatures of °C, in order to eliminate any variation in clay colour caused by the ancient firings and thus generate a basic optical distinction between different clay compositions (Whitbread , ). The colours for both the clay paste and slip were recorded with the Munsell Soil Color Charts.
Subsequently, based on the macroscopic and petrographic results, a sub-set of  samples was selected for chemical analysis with the aim of drawing preliminary conclusions on the provenance of local pottery production on Samos during the LNL-Ch period. The chemical analysis was carried out on a WD-XRF BRUKER S-TIGER wavelength dispersive spectrometer with a Rh excitation source, in order to characterise their elemental composition according to the calibration method developed at the Fitch Laboratory (Georgakopoulou et al. ). Quantitative bulk elemental analysis was undertaken on ignited powdered samples prepared as fused glass beads. Twenty-six major and trace elements were determined (Na, Mg, Al, Si, P, K, Ca, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Rb, Sr, Y, Zr, Ba, La, Ce, Nd, Pb and Th), and the data were subjected to statistical multivariate analysis using the R software.  The statistical treatment followed the approach proposed by Buxeda i Garrigos ().
A number of selected Ch samples from the neighbouring settlement of the Heraion are discussed in this paper, in relation to the fabrics identified at Kastro-Tigani. Only those samples analysed chemically are included here (   ( m) in the east of the island, the western lowlands consisting of the Karlovassi Basin, and the eastern lowlands consisting of Palaiokastro and Mytilinii basins. The latter, which constitutes our immediate area of interest, is important in terms of alluviation and hydrology. It accommodates the largest water sources on Samos (Imvrassos and Chesios rivers), which have been responsible for the formation of two large coastal plains, the Kambos-Chora plain and Mesokambos plain. As an extremely mountainous island, this part of Samos is the most easily accessible and suitable for habitation and exploitation. The island's location between the Attic-Cycladic geotectonic zone to the west and Menderes Massif to the east is responsible for its complex and heterogeneous geological background, which shares similarities with the Cyclades (especially the northern complex), a part of Attica and southern Euboea (Ring,  The Chora plain is extensively made up of Quaternary alluvial deposits with clayey-sandy material, terra rossa with grits, coarse torrential material, and recent scree, while on the coastal areas there is finer deposition of clays. The Heraion is situated within these deposits and in very close proximity to outcrops of the Pythagoreion and Hora formations to the north-west, while Kastro-Tigani is situated to the eastern limit of the Chora plain, in proximity with the confluence of the Chesios River and closer to outcrops of the Mytilinii Formation, with basaltic lavas in the lower strata (Theodoropoulos ; The geology here is mainly characterised by the presence of metamorphic rocks, intersected by other geological formations (e.g. ophiolite bodies, serpentiniteperidotite bodies, limestone deposits, igneous sills). The study of the geology through maps and geological literature is particularly useful in the assessment of provenance in the analysis of pottery. Although now generally acknowledged, no direct correlation can be established between ceramics and ancient raw material sources, due to geomorphological transformations, but an approximate analytical comparison is possible through geological prospection of locally available clayey raw materials in the determination of provenance (cf. Buxeda i Garrigos and Kilikoglou ; Montana ; Hein and Kilikoglou ). Such fieldwork, albeit preliminary,  has been carried out on Samos by Menelaou in - (Menelaou ) and allowed important insights into the identification of raw materials mineralogically similar to the ceramic fabric recipes from the Heraion (cf. Menelaou   clays in the Chora plain showed much compositional variation between the sources (Jones , -), which is also reflected in our analysis of the pottery.

Macroscopic analysis
The macroscopic features recorded were identified in  hand specimens (colour, hardness, feel, texture, lustre, porosity) through the examination of fresh breaks across the core MFG is characterised by a range of reddish yellow/greyish brown to buff (YR /, YR /) colours with varied coarse inclusions in a non-calcareous clay paste (Fig. a). Common inclusions include yellowish brown to greenish grey fragments, most likely identified as serpentinite, and frequent vegetal temper. It corresponds mainly to jars (cooking pots?) and cheesepots, with single examples of a jug and a bowl. Most vessels have a distinct grey core and heterogeneous clay body, but there is variability with respect to the surface treatments, from dark grey/greyish brown burnished to red slipped and burnished, reddish brown slipped and plain rough. All samples appear to have undergone low-temperature firings in mixed conditions. It corresponds to MFG from the Heraion (Menelaou and Kouka , table : MG).
MFG covers the majority of samples. It is characterised by a range of non-plastic inclusions, identified as volcanic-related on the basis of comparative material from the Heraion, where it corresponds to MFG (Menelaou ; Menelaou and Kouka , table : MG). The clay paste is largely heterogeneous and coarse to medium-coarse, light/reddish brown to greyish brown (YR /-/, .YR /-/) and usually has a dark core. Most characteristic are those inclusions with a light brown chalky feel and a soft/porous texture, black mafic minerals, white/grey or transparent crystalline minerals belonging most likely to quartz and feldspar grains, as well as elongate silver rock fragments and mica (Fig. b). The majority of vessels are rich in vegetal temper. MFGA varies on the basis of possible carbonate rocks (e.g. limestone). No correlation was identified between vessel type and surface treatment, the latter varying from dark grey/black burnished, greyish brown burnished, red/reddish brown slipped and burnished, red slipped, yellowish red burnished and pattern-burnished. All vessel types are represented ( Table ).
MFG constitutes a medium-coarse sandy fabric with common white transparent crystalline inclusions (most probably quartz), grey and silver angular ones related to metamorphic rock fragments, and few-rare organic matter (Fig. c). The fabric might represent an alluvial metamorphic environment. The paste is dark greyish brown coloured and homogeneously fired, whilst the surface is dark grey burnished, pattern-burnished (KT/) or decorated with white filled incisions and pointillé motifs (KT/). This group is rarely present and corresponds to MFG or MFG from the Heraion (Menelaou ).

Refiring tests
The refiring tests have confirmed observations from the macroscopic examination of the ceramic samples and have allowed the distinction between fabrics likely to be low/non-calcareous or calcareous. In Table , it becomes obvious that fabrics identified petrographically as noncalcareous refired red, corresponding mainly to MFG. The refired chips in this group exhibit a similar range of fresh break colours in reddish brown (.YR /), yellowish red (YR /) and brown (.YR /). The homogeneity in the refired colours reflects the use of similar types of clays. In contrast, those samples characterised petrographically as calcareous (Petrographic Fabric Group/PFG  and A) refired reddish yellow. This corresponds to MFG and A and implies either the use of calcareous clays or is related to the presence of calcite temper in some of the samples. Variability amongst the latter group in terms of colour, ranging from dark greyish brown (YR /-/, /) to light grey-brown (YR /) and very pale brown (YR / -/), could be due to anthropogenic clay mixing and/or the use of naturally varied clay sources, or could even reflect varied firing conditions. Significant changes in colour were recorded for MFG, represented by dark greyish brown (YR /-/) in fresh breaks, which refired to different hues of red. The slip, where preserved, turned homogeneously red (.YR / ) for all samples across all MFGs. The optical activity of the groundmass is consistently moderate to high, suggesting original firing to temperatures around -°C, mostly in oxidising/reducing atmosphere. The common presence of a grey core implies a rather short duration and so does the partly combusted vegetal temper.

Petrographic analysis
The thin sections from Kastro-Tigani were subdivided into a total of seven different petrographic fabrics, comprising small groups and isolated vessels classified as loners ( Table ). Over two-thirds of the samples are represented by a range of volcanic-related fabrics, but are further distinguished on the basis of mineralogical and technological differences, which are explained in detail below.
Petrographic Fabric Group : Ophiolite-derived with Serpentinite and Metabasites The first petrographic group (PFG; n = ) relates to naturally varied sediment sources of smallsized ophiolite outcrops and peridotite-serpentinite sills in the Selcuk nappe, occurring northwest of the Heraion in the area of Pagondas-Spatharei. It corresponds directly to PFG from the Heraion, where it constitutes one of the largest fabric groups, particularly for the manufacture of cheesepots in the LCh and amphorae in EBA II early (Menelaou and Kouka , , fig. :AB). Its non-plastics suite (Fig. a) includes mainly serpentinite fragments of various degrees of oxidation and metamorphism of basic igneous rocks, that occasionally preserve their original texture, chlorite aggregates or mafic-rich rocks showing evidence of chloritisation, serpentinised mica-rich metamorphic rocks and few metagabbro fragments ( Table ). It appears as a naturally heterogeneous group in terms of range, size, frequency, and sorting of inclusions, as well as the colour of the clay paste (predominantly yellowish brown to dark brown in PPL, dark red to reddish/greyish brown in XP). KT/ was separated as a variant on the basis of its mineralogical differences; its inclusions represent volcanic rocks of intermediate composition and a number of serpentinised rocks and metabasites (Fig. b). In his analysis of samples from Kastro-Tigani, Whitbread has identified a metamorphic fabric with zoisite schist, chlorite, and epidote-rich inclusions corresponding to cheesepots, that can be correlated with PFG to a certain degree.  Although generally compatible with the main group, KT/ (PFGA; Fig. c) varies compositionally and texturally; the vegetal temper is almost absent, the firing colour is darker,

SERGIOS MENELAOU AND OURANIA KOUKA
 the non-plastic inclusions are less packed, the serpentinites exhibit a different texture/colour than those of the main group and are outnumbered by quartz/feldspar crystals and epidote group minerals. The differences suggest the exploitation of a discrete raw material deposit that relates with the same geological formation. A few samples exhibit a darker margin possibly related to compaction of the surface due to smoothing.

Petrographic Fabric Group : Coarse Volcanic
This fabric (n = ) is dark-fired and slightly heterogeneous in terms of firing colour (majority of samples uniformly coloured in dark brown and others display some colour differentiation  between a dark core and yellowish brown-dark red colour) and density of inclusions with a bimodal grain size distribution, consisting of various volcanic rock fragments and their constituent minerals (fresh untwinned feldspars, quartz, amphibole, biotite and occasionally quartz-mica schists) (Fig. de). This fabric was first identified and described by Whitbread (Granitic Gneiss Fabric) in a previous small-scale sampling (Mavridis , -, , tables , , ) and has more recently been named as Volcanic (feldspar volcanic rock limestone schist) fabric (Whitbread and Mavridis forthcoming). Regardless of some mineralogical differences, it can be correlated with PFG from the Heraion (Menelaou ; Menelaou and Kouka , table : PG). A degree of compositional and technological variability can be observed when the samples from the two sites are compared, but the main features consist of volcanic rock fragments of intermediate to basic composition and a porphyritic or devitrified matrix, as well as vegetal temper in the form of elongate voids. Different types of Textural Concentration Features (TCFs), with one of low optical density, clear boundaries, or forming clay striations, are perhaps indicative of the natural heterogeneity in the clay source. It is defined as local to the Chora plain and relates to the small volcanic bodies intersecting within the metamorphic formations.

Petrographic Fabric Group : Calcareous Volcanic
This fabric (n = ) can be distinguished from the previous group by its high micritic content (micrite clots, rare bioclasts), substantial amount of large limestone crystals and calcareous clay concentrations in the micromass (Fig. fg), which suggests the use of a calcareous base clay or clay mixing, perhaps even tempering with volcanic sand as implied by the high bimodal distribution. Similarly to PFG, there is a range of smaller and fewer volcanic rock fragments, as well as mainly fine-grained rocks with dark devitrified matrices with acid to intermediate composition and rare volcanic glass fragments, which are probably related to tuff or tuffite deposits (Fig. f ). A considerable amount of vegetal temper occurs in all samples, as identified by elongate voids in the microstructure. The groundmass appears fairly homogeneous with respect to frequency and sorting of inclusions, textural features and colour (yellowish brown/ orange-brown uniformly coloured; slight colour differentiation between dark brown core and yellowish brown margins). It is highly calcareous and the generally moderate to high optical activity of the micromass indicates a low-firing temperature. The colour variation may reflect variable firing conditions rather than different clay sources. This fabric is chronologically and typologically varied, though mostly covering open vessels and bowls of Periods III and IV. Most samples in this group (KT/, KT/, KT/-) preserve thin layers (> mm) of red-brown iron-rich slip, being visually distinct from the calcareous groundmass.
A sub-fabric can be distinguished (PFGA, n = ), where the volcanic rock fragments are larger in size and mainly of a fresher texture. Its mineralogical composition is characterised by the constituent minerals of these volcanic rocks (Fig. h). The groundmass appears highly calcareous and the moderate to high optical activity of the micromass indicates a low-firing temperature. This subgroup is characterised by larger vessels such as jars and one cheesepot, which might explain the use of a coarser clay recipe, richer in vegetal temper.

Loner fabrics
In addition to the main groups, a number of loner fabrics have also been identified. PFG (Coarse Volcanic and Metamorphic; Fig. i) can be linked to PFGA by the presence of shared volcanic and siliceous rocks, dissociate minerals, and limestone (sparite-micrite aggregates and calcite temper), but differs on the basis of its metamorphic rock content.

TRACING INTERACTION ON SAMOS ISLAND
 loner appears more compacted and mineralogically less diverse (absence of mixed metamorphic rocks). A possible local provenance in deposits of the Chora plain on the south-east coast cannot be excluded. PFG (Well-sorted Quartz and Muscovite Mica; Fig. k) is also texturally different and shares some common features with PFG, but it stands out by the quantity and sorting of muscovite mica in the groundmass, related to mica-rich low-grade metamorphic rocks. Similar fabrics, found in the EBA at the Heraion in small quantity, are believed to be imported from the western Anatolian coast (Menelaou ).
PFG (Medium-coarse Sandy/Alluvial Metamorphic; Fig. l ) has a well-packed texture and is dominated by silt-and sand-sized monocrystalline quartz grains and a few metamorphic rocks, as well as rare micritic bioclasts. The weakly bimodal to unimodal grain size distribution of the fabric and the uniformity of its constituents suggest that a residual, unrefined sandy sediment rich in silicate minerals was used for the manufacture of the ceramic vessel. A possible match was identified in Whitbread's Felsic Fabric Group (Mavridis , -, , tables , , ; Whitbread and Mavridis forthcoming). Its mineralogy reflects an alluvial metamorphic environment and is taken as broadly local.

Petrographic results from the Middle-Late Chalcolithic Heraion
There was undoubtedly a strong ceramic manufacturing tradition at the Heraion since the foundation of the site. This concerns a rather homogeneous manufacture in terms of fabric, finish, forming, and firing techniques employed for the production of the full range of domestic vessels, large bowls, jars of various types, jugs, and cooking pots.
The petrographic results distinguished at least three main fabrics at the Ch Heraion and a number of subgroups and loners ( Table ). PFG (Ophiolite-derived with Serpentinite) is directly linked with PFG at Kastro-Tigani, but their mineralogical and textural differences could be a symptom of exploiting discrete raw materials deposits that are related to the same geological formation (ophiolite outcrops and peridotite-serpentinite sills in the Selçuk nappe) and the natural heterogeneity of these exploited deposits (Menelaou and Kouka , table , fig. A-C ) (Fig. ab). This fabric is used at the Heraion exclusively for the manufacture of vessels related to cooking or the preparation of food (cheesepots, cooking jars), similarly to Kastro-Tigani (KT/, KT/), but at the latter site it is also used for other functional categories.
PFG (Coarse Mixed Metamorphic) and PFGA (Red Coarse Metamorphic) were characterised as broadly local to the Chora plain (alluvial deposits of the Ambelos nappe) and are linked on the basis of a heterogeneous, naturally mixed lithology, with more common coarse high-grade metamorphic inclusions and related minerals (Fig. cd ). This series of fabrics or sub-groups comprises the most common clay recipe used throughout prehistory at the Heraion, and its compositional variability is also supported by comparable clay samples collected in the vicinity of the site (Menelaou and Kouka , fig. D-I ). On the basis of its compositional and technological variability, PFG was further divided into more sub-groups (Menelaou ), C (Coarse Metamorphic -Common Quartz-Mica Schist and Vegetal Temper; Fig. e) and D (Sand-tempered Metamorphic). The latter fabric consists of Ch/EBA I dark red-slipped collar-necked jars and is characterised by oxidised quartz-muscovite schists and greywacke fragments, as well as limestones that are occasionally fossiliferous (Fig. f ). The presence of such coarse inclusions in a generally very fine groundmass strongly suggests the intentional addition of sand temper in the clay paste. This sub-group is texturally distinctive and finds very close parallels in the Sand-tempered fabric group from Çukuriçi Höyük in western Anatolia, which is considered as local and covers EBA I samples belonging to closed vessels and tripod cooking pots (Peloschek , -, fig. , Phases IV-III). This could potentially represent imports at the Heraion, but the provenance will become more confident with more analysed comparative material. No further parallels could be identified petrographically for metamorphic fabrics at Kastro-Tigani.
PFG spans from the Ch to the EBA III and is defined as broadly local at the Heraion. It relates to the small volcanic bodies intersecting as sills within the schist formations in the eastern side of SERGIOS MENELAOU AND OURANIA KOUKA  the Ambelos Massif. Although exhibiting substantial mineralogical and textural similarities with the volcanic fabrics from Kastro-Tigani, on the basis of an intermediate-basic composition and a porphyritic or devitrified matrix, these are, however, distinct. The volcanic class from the Heraion shows a greater variability and differs from that at Kastro-Tigani in the sorting and distribution of minerals, the calcareous content, and the occasional presence of metamorphic inclusions (Fig. gh). It is probably more similar to PFG from Kastro-Tigani. Perhaps it should be distinguished in sub-groups: A (Fresh and Altered Volcanic) and B (Felsic Volcanic and Metamorphic).
Two secure imports were further identified petrographically at the Ch Heraion. The first belongs to a cheesepot that is made in the Muscovite-rich Medium-grade Metamorphic fabric (Fig. i) and can be correlated with the red-brown metamorphic fabric series with a provenance in the north-west Cyclades, possibly the island of Kea (see Menelaou and Kouka ,   . Perhaps in support also of a special use of these vessels, it is noteworthy that an almost complete cheesepot at Kastro-Tigani was found containing a large triton (Felsch , -, no. ; Fig. i).
The second import, a sherd from a closed vessel in pattern-burnished ware, corresponds to the Porphyritic Intermediate Volcanic fabric (Fig. j), which appears in substantial frequency in EBA I-II (Menelaou ). Based on morphostylistic and fabric parallels, this vessel has a provenance in coastal western Anatolia (cf. volcanic fabrics from Ulucak Höyük III: Müller, Kiriatzi and Çevik ), but a closer resolution is currently infeasible in the absence of published comparanda.
Chemical analysis WD-XRF analysis confirmed the macroscopic and petrographic observations and distinction of the main fabric classes. The compositional variation matrix for  elements was calculated (Table ) in order to assess the total variation and variability of each element (Buxeda i Garrigos ). As suggested by the evenness variability graph (Fig. ), the total variation value is very high, both for all the  chemical elements (vt = .) and after excluding Cu, Na, P, and Pb from the multivariate statistical treatment (vt = .). This is reflective of a polygenic population (Buxeda i Garrigos and Kilikoglou ), in agreement with the results of the petrographic analysis. The variability of the excluded elements may be related to post-depositional alteration/contamination in the pottery. The element that introduced the highest variability in the dataset was Cr, followed by Ni, MgO, and Th. After showing the lowest variance in the log-ratio transformation of  element concentrations, Al was chosen as common divisor. The results of testing the similarity of all samples towards the groups' average compositions, through cluster analysis, allowed for establishing three main clusters, further divided to more sub-clusters, and two loners in the dendrogram (Fig. ). These differences were also confirmed through a principal component analysis (PCA). According to Table  relatively homogeneous compositions are observed among the clusters, with variations in a series of elements as described below, related to some major and minor elements.
As observed through the cluster analysis, the chemical data clustering seems to broadly overlap with the petrographic groups (Fig. ). Cluster A (n = ) consists entirely of samples petrographically grouped to PFG, with samples from both Kastro-Tigani and the Heraion. It stands out from the rest of the groups, both mineralogically and chemically, due to its high concentrations of Cr, Ni, Co, and MgO (as well as Fe  O  ). This is associated with the  (Menelaou and Kouka , -). This chemical sub-cluster is distinguished by higher and lower values in K  O and Cr concentrations respectively, compared to the other clusters. Although clustered together (Fig. ), perhaps due to the frequency of metamorphic rocks, HR/ has comparatively higher values in K  O (. per cent), Rb ( ppm) and Sr ( ppm), and lower values in V ( ppm), Cr ( ppm), Fe  O  (. per cent), and Ni ( ppm), and is also petrographically different. B (n = ) is petrographically similar to B, on the basis of metamorphic-related minerals, as indicated in the relatively elevated iron content (- per cent)   most likely as a result of iron-rich secondary minerals in the schist rocks, but differs in the concentration values of Cr, Co, Ni, and Sr. KT/, petrographically singled-out as a loner fabric (PFG), appears in B together with two samples from the Heraion on the dendrogram (Fig. ); this is also supported in the PCA plot, where it appears closer to Cluster C (Fig. ). However, it differs chemically, as seen in the higher values of MgO and K  O and the lower values in Mn and Zn. Interestingly, sub-clusters B and B are broadly grouped together in the dendrogram (Fig. ), suggesting that the metamorphic-rich raw material sources were preferably exploited at the Heraion and not at the nearby settlement of Kastro-Tigani. With regard to trace element composition and internal homogeneity, Cluster C and its subvarieties are linked between one another, confirming the petrographic observations, possibly related to the presence of igneous rocks. However, their chemical differentiation was consistent with the petrographic results, with chemical clusters and fabric groups largely matching, further confirmed by the low total variation values obtained for the individual sub-clusters. In particular, C (n = ) corresponds to PFG, C (n = ) to PFG also including sample KT/ (PFG) which is petrographically different but clustered together with PFG samples on the basis of its high CaO percentage (. per cent), and C (n = ) corresponds to PFGA. The high frequency of volcanic and less commonly of metamorphic minerals in the aforementioned clusters could account for the relatively high values in Zn, in comparison with the values obtained for Clusters A and B. Nevertheless, different values in certain trace elements exist between the chemical groups of Cluster C, for instance in the values of V, Co and others. CaO concentrations range between . per cent (Cluster C, low calcareous) to . per cent (Cluster C, intermediate calcareous), and . per cent (Cluster C, high calcareous), in association with clay pastes rich in carbonates. The differentiation was consistent with the petrographic results, which showed an assumed non-calcareous composition for C. In terms of CaO presence, Cluster A appears to be intermediate calcareous (. per cent), while Clusters B (. per cent) and B (. per cent) are low calcareous.
The presence of two samples behaving as chemical loners, namely KT/ and KT/, is observed in both the dendrogram and the PCA plot (Figs -). Although distinguished petrographically as a variant of PFG, KT/ is chemically different in the concentration values of certain elements (lower Cr and Ni; higher V and Mn). Nevertheless, this sample plots in association with Cluster A in the PCA, where the samples form a loose and scattered group on the right side of the plot. Loner KT/ plots apart from all clusters in the PCA and is also distinguished in the dendrogram; its provenance remains undetermined, although on a fabric basis it could represent a distinct raw material source of the repetitive metamorphic lithology on Samos.


The chemical results reflect on the one hand the complex geology of Samos and, thus, of the exploited raw material sources for pottery manufacture at the south-east part of the island, which is firmly represented in the mineralogically very different fabric groups. On the other hand, it is noteworthy that, despite the nature of the dataset and its very high total variation, perhaps also due to the small number of the samples analysed, as well as the loose clustering and fairly high relative standard deviation values for most elements, the results indicate clear differences between the three main clusters. These differences show a good correlation with the petrographic results, suggesting the exploitation of different clay sources or even the concurrent operation of distinct production units during the LN-Ch. Intra-cluster variability and the formation of chemical sub-groups may in fact also be the outcome of technological variability (Buxeda i Garrigos and Kilikoglou , -). A comparison between the results from Kastro-Tigani and the Heraion suggests the existence of compositional differences. While the volcanic fabrics identified at Kastro-Tigani are mineralogically similar to those used in later periods at the Heraion, they are chemically distinct. Similarly, the chemical clustering of metamorphic fabrics from the Heraion have no parallel in the Kastro-Tigani assemblage. These variations, also evident in the PCA plot, may indicate the selective exploitation and use of slightly different raw materials and/or clay paste recipes, although the chronological difference between the sites could also account for technological and compositional differences between them. The natural heterogeneity of the geological deposits exploited and repetitive geology of Samos, obvious in the analysis of raw material samples (Menelaou and Kouka , figs -), could also explain much of the chemical and petrographic variability observed in the samples.

Determining local provenance and relations between Kastro-Tigani and the Heraion
The combination of several methods and stages of analysis allowed us to produce some useful patterns in the determination of provenance of the analysed ceramics. In the absence of kiln materials  that could be used as reference groups in such an effort, our analytical work has developed a pilot dataset and has defined the mineralogical and chemical signatures of local and possible imported pottery. A general drawback in the assessment of provenance between raw material sources and pottery relates to the fact that the resources exploited by the ancient potters are not immediately represented in today's landscapes. However, the mineralogical comparison between the two can provide useful information through petrography, and hints towards determining the geological provenance of a ceramic fabric (e.g. Montana ).
Based on a thorough bibliographical study of the geological background in south-east Samos, we may infer Samos' prehistoric communities' choices of resources. More particularly, PFG at Kastro-Tigani corresponds petrographically and chemically to PFG at the Heraion (Menelaou and Kouka , table , fig. A-C). Despite some mineralogical differences between the two, these seem to relate to naturally varied melange sediment sources of small-sized ophiolite outcrops and peridotite-serpentinite sills (Selçuk nappe), occurring north-west of the Heraion in the localities of Myli, Pagondas, Spatharei, and Mavratzei ( Ophiolite-related fabrics are generally compositionally varied, associated with the inherent lithological variability of the raw material sources. This is further exemplified in the recognition of variable samples (KT/) and sub-group A, which suggest the exploitation of discrete outcrops situated within the same geological formation used for the main  Evidence for on-site pottery production, i.e. preservation of kilns, has been so far rarely preserved and /or documented in Aegean NL sites, with the exception of two pottery kilns and a clay kiln model at MNL Imvrou Pigadi (Kyparissi-Apostolika ) and one at LNL II Dimini in Thessaly (Chourmouziadis ). Intramural pottery kilns are preserved in the east Aegean during the MBA-LBA at Ialysos on Rhodes, Miletus, and Liman Tepe (Marketou ; Niemeier ; Mangaloglu-Votruba , ).

SERGIOS MENELAOU AND OURANIA KOUKA
 class. The limited occurrence of such geological deposits at the aforementioned localities allows a more confident suggestion of local provenance. The possible relation of PFG with primary sources implies the direct exploitation of these deposits and perhaps even the existence of a production centre beyond the Heraion itself. At the Heraion this chronologically and typologically very consistent fabric is well represented, especially during the Ch and EBA I, for the manufacture of cooking pots and cheesepots but less frequently during EBA II early for the manufacture of amphorae. At Kastro-Tigani the majority of samples are typologically varied and date to the LCh (Phase IV), with the exception of two samples that date to Phases II-III. Only a detailed geological study of the area and more extensive analysis of clay samples may help to establish whether the differences are due to (a) the exploitation of distinct deposits (perhaps localities of Pagondas-Spatharei for the Heraion and the locality of Mavratzei for Kastro-Tigani), (b) the manufacture of pottery in this clay recipe at separate production centres for Kastro-Tigani and the Heraion or (c) its importation and consumption at Kastro-Tigani from the western part of the Mytilinii Basin (Heraion). A previous small-scale resource prospection and clay sampling in the area of Pagondas-Spatharei has identified raw materials that are petrographically identical with this fabric (Menelaou and Kouka , table ).
The volcanic fabric series at Kastro-Tigani comprises the majority of samples. The compositional differences between PFG and PFG (and sub-fabric A) can correlate with the chemical sub-clusters, reflecting the use of disparate raw materials. Samos has a limited presence of volcanic rocks (trachydacites, minor basalts, rhyolites, weathered tuffs and tuffites) restricted at the lower series of the Karlovassi and Mytilinii Neogene basins. Small volcanic bodies, formed in the early stages of the continental sedimentation of the basins, appear as sills along the eastern margin of the Karlovassi Basin and the western margin of the Mytilinii Basin during the Middle-Late Miocene, intersecting within the metamorphic formations (Theodoropoulos ; Pe-piper and Piper , fig. ; Owen, Renaut and Stamatakis ). Such occurrences appear mostly near Pagondas, Pyrgos, and Mavratzei, together with red loams and clays, while mostly minor rhyolites and sodic basalts crop out at the small-sized Palaeokastro Basin, situated at the eastern part of Samos and the margins of the Vathy Basin to the north-east. Common volcanic rocks are identified also in PFG, but differences occur in the presence of more carbonate minerals and arrangement of inclusions. Regardless of the presence of intermediate to basic volcanic rocks and their constituent minerals in a similar clay paste identified at the Heraion (PFG), it is noteworthy that these are, to some degree, mineralogically and chemically recognisable. This is seen in the occasional presence of metamorphic rocks (see HR/), which are linked to the schist formations of the Mytilinii Basin, the almost total lack of clay pastes equivalent to PFGs-A from Kastro-Tigani, and the predominance of volcanic rocks with a weathered top. The latter are predominantly spherical, perhaps having been reworked through water action. These are related to volcaniclastic rocks and basaltic deposits with a weathered top occurring at the western margin of the Vathy Basin and near Pagondas and Pyrgos, overlain by felsic pyroclastic deposits that correlate with the Mytilinii Formation in the centre of the basin (Pe-piper and Piper , ; Menelaou and Kouka , table , fig. JK ). The mineralogical and chronological (predominantly EBA II-III at the Heraion) differences in the use of these volcanic-related fabrics at both sites suggest the exploitation of distinct raw material outcrops. The clay samples from the Chora plain analysed so far suggest the local provenance of this fabric series (Menelaou and Kouka , fig. L).
A third group distinguished in our analytical work (Figs c-f and ) corresponds to the metamorphic fabric series that occurs diachronically only at the Heraion from the Ch to at least the MBA for the manufacture of the whole range of vessel types and functional classes (Menelaou and Kouka , table , fig. DE,G-I ). Corresponding to PFG and its subgroups, it represents the main local fabric series, although the compositional variability in terms of coarseness, frequency, and quantity of metamorphic-related inclusions implies the exploitation of several raw material sources in the Chora plain. This fabric series was most likely produced from red alluvial deposits (schist bodies) that cover the immediate area of the Heraion, extending from the Chora plain to the foothills of the Ambelos Massif, and is broadly related to the metamorphic substrate of Samos (Fig. ). However, the inherent variability of metamorphic TRACING INTERACTION ON SAMOS ISLAND  geologies and the repetition of different geological formations over a wide area can prevent discrimination and a more precise provenance determination between the various subgroups. The compositional consistency with Samos' geology was also supported by comparable clay sources sampled in the vicinity of the Heraion (Menelaou and Kouka , fig. F ), but even the most detailed geological prospection may not be able to provide direct petrological links due to intra-variation of the same outcrops and erosion/transportation of alluvial sediments through the action of rivers.
Aside from the aforementioned main clay recipes used synchronically and diachronically at the sites, a number of loners have also been identified at Kastro-Tigani, dated to Phases III and IV and representing amphorae/jars. Although a local provenance could be suggested for the majority of loners, it remains undetermined whether the mineralogical and textural differences reflect connections with other areas on the island that have been considered 'empty' so far (PFG; Fig. i) or represent off-island products. The inconsistency of these loners is reflected to a certain degree also in the chemical analysis, where PFG and PFG differ in their trace element concentrations from their main clusters, while PFG behaves as a loner in the dendrogram and PCA plot (Figs -).
Reconstructing the Middle-Late Chalcolithic Samian ceramic traditions: a diachronic technological view Shifting away from a solely typological and morpho-stylistic approach, in this paper we adopt a theoretical framework that concentrates on the social dimensions of technological practice, according to which technology is a socially embedded dynamic process of combined social and material engagement (Sillar and Tite ). An attempt was made to reconstruct the operational sequence of pottery production, i.e. the various stages in the manufacturing process and related past technical systems, and social, cultural, and economic acts in the ways of making and transforming the raw materials to finished products (cf. Roux ). This section focuses on the micro-scale processes of pottery production, defined here as the interactions involving potters, their raw materials, and the social context in which the craft practice took place. By highlighting where in the production sequence variability and change occur, we can draw inferences concerning the location and organisation of ceramic production and the make-up of the earliest potting groups operating on south-east Samos, as well as the shared repertoire of technical know-how or learning environment that characterises different communities of practice.
More particularly, raw material exploitation and procurement strategies at Kastro-Tigani were focused on the use of two main lithological categories, with further varieties, distinct in the petrographic fabric variability discussed above. Interestingly, the analytical results can be generally correlated with our macroscopic observations. A picture of continuity in the choice of raw materials emerges. Regarding raw materials preparation and processing, the Ch potters chose either to use clays in their natural state or to process and manipulate the raw materials by refinement and mixing. The heterogeneity and freshness seen in PFG is indicative of the use of largely unprocessed raw materials, but the occasional presence of dark brown clay pellets might also suggest clay mixing. According to the freshness of inclusions in PFG, the clay derives from a primary source close to volcanic parent rocks, but the angularity of the constituent minerals could even suggest crushing and tempering of the raw materials. Although mineralogically linked with the previous group, PFG exhibits the exploitation of rather different carbonate-rich raw material sources. The presence of fresh calcite perhaps suggests tempering, while the identification of texturally distinct red striations (e.g. KT/, ) and micrite in the base clay might be indicative of some sort of mixing. It is worth noting that in Tigani Phase III a technological change occurs in the use of higher-calcareous clays, as illustrated in sub-fabric A. This differs from PFG because of a finer groundmass, perhaps a result of clay refinement, and the presence of well-distributed sand-sized limestone and well-rounded mineral inclusions. This fabric reflects the technological knowledge and decisions of potters to use a clay mix with coarser and abundant limestone, which would thus increase the toughness and strength of the manufactured jars (cf. Tite, Kilikoglou and Vekinis ).


As already discussed in the previous section, procurement and exploitation strategies at the Heraion exhibit some differences from Kastro-Tigani Phases III-IVconsidering also the chronological gap at Tigani III-IVthat could suggest the use of separate but lithologically compatible sources for the manufacture of pottery during the MCh-LCh period, but also the preferred exploitation of others. It is striking that, while the metamorphic fabric series (PFG and subgroups) at the Heraion constitutes the longest ceramic mix used from the Ch till at least the MBA, owing also to the widespread availability of related raw materials in the Chora plain, it is absent from Kastro-Tigani, even from its Period IV.
A common practice in the preparation of the various clay mixes can be seen in the use of vegetal temper, although varying in frequency. It is not as much present in PFG as in PFG, where it mostly occurs in cooking pots of Phases II-III (e.g. KT/, , ). Chaff-tempering is better embedded in the operational sequence of the Heraion's potting communities, which provides a good link between local fabrics and implies a long-lasting continuity of practice until EBA II. Vegetal-tempering might have been related to more than just techno-functional criteria (improvement of clay workability, thermal stress resistance, etc.). It is a widespread technological practice among various regions with a noticeable decrease from EBA II onwards (e.g. Kos: Vitale and Morrison ; Chios: Hood , ; Miletus: Kouka , ; Beycesultan: Lloyd and Mellaart , ), but it is noteworthy that it did not form part of the potting traditions at the neighbouring site of Çukuriçi Höyük, despite sharing lithologically comparable raw materials and relevant clay recipes (Peloschek : red metamorphic-rich fabrics; buff calcareous ophiolite-related fabrics).
In terms of forming techniques, our macroscopic observations agree with the general tradition of handbuilt ceramics during the MCh-LCh period. Macroscopic evidence of relic coils (observed also petrographically in the form of concentrically arranged inclusions), pinching marks and finger grooves, secondary formation of handle attachments with no pushed-through examples and jar rims/necks joined separately the main body, and indentations on the exterior surfaces suggest the combination of different techniques. These may include coiling, slab-building, and pinching, and perhaps represent different potting traditions.
Several surface treatments and decorative elements have also been identified macroscopically and complemented with petrographic observations. No particular relationship seems to exist between fabric and surface treatment, but some degree of chronological patterning could be observed. For instance, dark burnishing is by far the most popular during the earlier phases at Kastro-Tigani, with a minor presence of white-painting and smoothing. The latter is found on the exterior surface of cheesepots, occasionally preserving traces of a red slip on the interior, most likely for reasons linked to their function in food preparation activities (e.g. reducing permeability: cf. Menelaou and Kouka , -). Dark burnished and red/reddish brown slipped surfaces of Tigani I-II are preferred for vessels made in PFG. From Phase III onwards there seems to be a higher variability in surface finishes (black polished; red slipped and burnished; light brown slipped; pattern-burnished; incised and pointillé), even in typologically similar vessels, which should be examined in the framework of established stylistic regionalism and transfer of technological skills during mid-fifth and fourth millennia BC, as well as the exercise of different craft decisions by the local potting communities. Burnishing marks usually occur vertically but can also be multi-directional. The most obvious and better-preserved slips, which likely had an iron-based composition (red to reddish brown, in relation to burnished surfaces; e.g. KT/, -) appear consistently in the thin sections of PFG. Similarly, at the Heraion the slip layers vary in colour and thickness, especially those preserved on vessels of the volcanic fabric series (PFG). Others exhibit areas near the surface edges with a strong, parallel orientation and slightly different optical activity, probably related to compaction due to smoothing, and the creation of a self-slipped surface (PFG, non-calcareous and iron-rich clay). Surface finishes at the Heraion are less varied (absence of white-painted, very limited patternburnished and black-burnished, etc.).
Finally, the assessment of the firing conditions was undertaken through a combination of macroscopic and microscopic information. More particularly, the pottery was generally fired to low temperatures and varied atmospheric conditions with short oxidation, as suggested by the TRACING INTERACTION ON SAMOS ISLAND  pronounced grey/reduced cores or core-margins colour differentiation, discolouration/mottling of surfaces, medium to high optical activity of the groundmass, and freshness of limestone inclusions in PFG. The presence of partially combusted vegetal temper in some cases attests further to the low-firing temperature employed and related little control over the procedure. Previous targeted SEM-EDS analysis on pottery from the Heraion (Menelaou and Kouka , , fig. ) supports our evidence from Kastro-Tigani and an equivalent temperature in the range of -°C. The prevalence of burnished finishes correlates also with employing lowtemperature firing procedures. Firing at low temperatures also decreased the risk of breakage of such coarse-grained clay recipes.
Despite the small number of analysed samples at Kastro-Tigani, it is important to highlight some general correlations between preferred ceramic clay pastes and vessel types, with a marked identification of period-specific choices. To be more exact and following a chronological narrative, PFG is predominantly in use during Phase I for the manufacture of typologically and functionally different vessels (jars, bowls, cooking pots). PFG is mostly used in Phase III predominantly for the manufacture of bowls, with single examples also in Phases I and IV respectively, while sub-fabric A is used concurrently in Phases III-IV for the preferred manufacture of jars. PFG, although known since Phases II-III, becomes the main ceramic recipe in Phase IV for the manufacture of different vessel types. The latter fabric, similarly to the equivalent at the Ch Heraion (Phase ), is used for the manufacture of cheesepots and cookingrelated vessels, but appears typologically more diverse at Kastro-Tigani. During the Ch at the Heraion, PFG and PFG are used for the manufacture of cooking pots and cheesepots, while PFG appears to be multi-functional.

Craft organisation and modes of pottery production in south-east Samos
The overall picture emerging from the technological assessment of MCh-LCh ceramic assemblages from the two neighbouring settlements is one of continuity and change. The organisation of production shows by no means a linear evolution and appears rather complex, with choices and decisions affecting discrete stages of the manufacturing procedures. The analytical work revealed that while potters at Kastro-Tigani had access to visually and lithologically distinct raw material sources in their immediate landscape, these were used largely concurrently from the earliest phase for functionally, typologically, and stylistically varied ceramic products. Nevertheless, different requirements led to divergent technological choices in manipulation practices and the preparation of distinct paste recipes. For instance, cheesepots are manufactured in different clay pastes, obviously with different technological properties, but potters adapted their methods of surface finishing even if other, available clays would provide the finished product with a selfslipped interior. The intra-fabric variations might represent the coexistence of different manufacturing traditions, particularly when considering the correlation with certain shape categories. Thus, continuity and shared practices are evident in the likely exploitation of clay sources, forming and firing, but marked discontinuities are observed in morphological features and surface finishing and the visual appearance of pottery.
At the Heraion we identify a different potting community of practice, where lithologically similar geological raw materials resources were exploited in proximity to the site and beyond, at least during the MCh-LCh period, but with evidence for rather different decisions in the manufacturing stages. This implies to a certain degree that direct transmission of knowledge and interaction existed between the potting communities from both sites, in the sense of verbal guidance about clay recipes and location of resources. Based on a combination of analytical data, stylistic, typological, and technological evidence, we can assume that some sort of specialised production existed, occasionally with clay recipes correlating with certain shapes or chronological periods. However, the two sites exhibit differences in the production of shapes (i.e. certain vessel types and wares are missing from the Heraion), which could also be a symptom of consumption preferences, technical expertise and tradition, divergent regional interactions, or visibility in the archaeological record and chronological inconsistencies between the sites in absolute terms.


Overall, a significant overlap in the craft organisation of local pottery production is observed, marked with the coexistence of multiple potting groups. Beyond physical attributes in the selection of raw materials, the longevity of certain clay recipes implies an in-depth knowledge of material properties, while shared practice suggests the inter-generational transmission of technical knowledge and technological traditions.
Aegean Late-Final Neolithic/Anatolian Middle-Late Chalcolithic cultural interaction and maritime connectivity The concepts of connectivity and mobility in the prehistoric Aegean, as well as the mechanisms of communication, have taken up a significant amount of research in the last decades, particularly discerned through similarities in ceramic assemblages. In attempts to explain regional diversities, narratives of typological and stylistic influences alone are not sufficient, as we usually miss evidence lying beneath the surface. Nevertheless, judging from such morphostylistic similarities and shared techniques across such a wide geographical scale, we may assume that different sites were participating in networks, with common forms and decorative styles conveying social and symbolic ideologies. That is then why the identification of ceramic markers (e.g. cheesepots, pattern-burnished decoration, mat-impressed, etc.) is important not simply as narrating a homogeneous cultural view of a given region, but because, together with exotica, it can materialise the movement of people, cultural transmissions, consumption preferences, skills and status, and sharing of values (e.g. Stark, Bowser and Horne ).
The archaeological evidence from the MCh-LCh settlements of Kastro-Tigani and the Heraion illustrates two sites located respectively in the eastern and western part of the once deep Gulf of Pythagoreion that shared the same micro-landscape in the southern part of Samos Island. These agricultural sites commonly exploited the regional maritime environment through their safe anchorages just opposite the Maeander River Delta, which constitutes the most direct gate to the inland of western Anatolia, and is located on the natural sea-stream that connects the southern with the northern part of the east Aegean seascapes. Moreover, the nodal location of Samos in the central part of the east Aegean facilitated the connection between the south-east and the north-east Aegean islands, as well as between the Cyclades and western Anatolia (Fig. ).
The aforementioned natural advantages elucidate resemblances at a micro-regional level in settlement longevity, architecture, ceramic, and lithic technology and production and importation of raw materials (obsidian) and various finished products (marble figurines, ceramic and marble vessels), from the mid-fifth through the late fourth millennia BC. The material culture of the discussed Samian sites is comparable with that documented in other east Aegean islands (from Lemnos to Rhodes) and in the western Anatolian littoral (from Kumtepe to Miletus), and is best exemplifieddespite the underlying local ceramic traditionsin the homogenising recognition of a cultural koine in this part of the Aegean during these millennia (Kouka ; Tuncel and Ş ahoglu , -). This becomes clear predominantly from the previous discussion on ceramic styles and technologies, but also from the presence of similar imported raw materials and exotic artefacts with a special symbolic value, either from the Cyclades or from western Anatolia, that reflect shared cognitive and social affinities, transgression of social borders, and the diffusion of ideas. Aside a handful of ceramic products (cheesepot, jars) identified at the Heraion as off-island (Menelaou and Kouka , , fig. A), deriving from the Cyclades and western Anatolia, a small number of undetermined loner fabrics at Kastro-Tigani imply different contacts during the Ch period. This is indirectly also reflected in the preferred consumption of certain shapes and wares at each site. The patterning in the distribution of the various wares characterising the spatial framework of this paper could shed light into the mechanisms of technical transmission, being either indirect by observation or direct by verbal guidance and experiential learning, which may allude to the establishment of visually uniform regional traditions. Such a study, in fact, addressing the impact of long-term population fluctuations that influence the sharing of ceramic similarities, was carried out for the MNL and LNL (de Groot ).


Of particular interest in our discussion of maritime connectivity is the acquisition of obsidian from Melos as a result of a long-distance maritime trade network between the Cyclades and the eastern Aegean island-and coastscapes. Such a network was established since the seventh millennium BC, as indicated by Melian obsidian tools found in the earliest known NL sites of western Anatolia at Ulucak VI and Çukuriçi Höyük XIII in the Izmir region (Horejs , -), and two LNL axes of jadeite from Syros at Çukuriçi Höyük XI and IX (Horejs , fig Further witnesses of Samos' maritime communication with the western Aegean are the ringshaped attachment on the lug handle of a bowl from MCh Tigani II or III (Fig. c) imitating another prestige marker of the Aegean LNL-FNL used particularly in western Anatolian funerary contexts, such as at Bakla Tepe, until the EBA II (Keskin , )the marble acrolithic figurines from LCh Tigani IV and Heraion  of Thessalian inspiration (Fig. ab), the Early Cycladic I marble and clay flat-based conical beakers with vertical pierced lugs from Tigani IVb (Fig. no), as well as the kratiriskoi (Fig. m) also from the same phase. Conical beakers with flat base and kratiriskoi, either imported from the Cyclades or locally produced occur also as grave goods in the EBA I cist-graves of the cemetery at Iasos (Pecorella , fig. :-, pls XXXVIII:, XXXIX:), along with ingots of possibly Cycladic copper in the form of arm-rings or perforated axes (Pecorella , pls XXXII:,, XXIII:, XLII:, XLVI:): these appear to signify another strong link between western Anatolia and the Cyclades towards the end of the fourth and the beginning of the third millenia BC.
Such long-distance maritime routes seem to have been intersected by local, short-distance maritime and land trade routes among the east Aegean islands, the littoral and the inland of western Anatolia, established through the obsidian trade. Clear markers of these routes are the MCh marble conical beakers and marble anthropomorphic figurines of the Kilia type with triangular, backwards bending head with relief elongated nose, long neck and wing-like arms (cf. Fig. d ). Pointed conical beakers were found, as mentioned earlier, in MCh sites along the coast and in the inland of western Anatolia (see references above), as well as at Koukonisi on Lemnos and Tigani IIb-III on Samos (see references above). The discovery of conical beakers as final or unfinished products and of Kilia-type figurines at Kulaksızlar in the inland of the Izmir region, and the identification at this site of the only so far known marble workshop in the region during the mid-fifth millennium BC, suggest a possible provenance of the above Anatolian examples from this regional workshop (Takaoglu ; , -, figs -). Moreover, the workshop at Kulaksızlar seems to be contemporary with typologically comparable examples (Takaoglu ) of the late fifth millennium BC (FNL) from Kephala on Keos (Coleman , , pls , ; Coleman and Facorellis , ). Given the highly specialised manufacture, and their low frequency primarily in domestic contexts of the MCh, the pointed conical beakers could be considered as high-valued containers, most likely for the use of liquids  fig. :), being usually discovered in fragments (heads, torso), in combination with their unsecure provenance mostly from mixed LCh/EBA contexts raises the question of intentional fragmentation of such figurines during special social events (ancestral ritual feasts) and their inheritance to the next generations, as was proposed for a Dokathismata type Early Cycladic II marble figurine in an early MBA context at Miletus (Kouka b, ). The high symbolic value of Kilia figurines becomes more evident by its clay equivalent, a triangular head fragment uncovered at EBA I Çine Tepecik (Tuncel and Ş ahoglu , , fig. :). Thus, Kilia figurines seem to have represented artefacts with a high symbolic value, which were circulated along with conical beakers and Melian obsidian (Kouka , ) from Kulaksızlar to the north-western and western Anatolian coast and, based on seven fragmented examples from Çine Tepecik IV, possibly also in the Maeander Valley towards Aphrodisias, Can Hasan and Karain to the east and south-east and Tigani on Samos to the west. If we acknowledge that such exotic items signify exchange and mobility, and by extension social boundaries, then we may assume that elites, seafarers or others from Samos, being the last anchorage for Cycladic longboats with obsidian in their cargos off the Milesian coast, have gained such prestige objects from interaction with their equivalents travelling from both the Cyclades and western Anatolia.
Maritime interregional trade networks should have been managed by wealthy FNL Cycladic settlements, such as Strofilas on Andros (Televantou ), for the transfer of Melian obsidian, metals (not in the case of Samos), artefacts with symbolic value, ceramic containers, technologies and ideas, alongside the mobility of people, towards the eastern Aegean seaboard, particularly at gateway communities occupying such nodal geographical positions. Such incentives and motives must have gradually opened new ways of communication at both sites of the Aegean, through the establishment of supra-regional routes, while the presence of the symbolic artefacts and adoption of exotica mentioned above signifies the emergence of societies with inequalities also in the eastern part of the Aegean well before the dawn of the Bronze Age.

CONCLUSIONS
This study provides new insights and contributes to the currently enriched picture of MCh-LCh pottery production and circulation of artefacts on Samos, with implications in understading connections and cultural trajectories in the eastern Aegean/western Anatolian region. The incorporation of Samos into this wider perspective of chronological associations and sociocultural affinities was further allowed due to the increase in recently excavated data from a number of sites and evaluation of existing evidence. A more integrated analytical approach has been employed in our selective re-examination of the old ceramic assemblage from Kastro-


Tigani, providing a fresh perspective beyond mere description of styles and shapes and the characterisation of ceramic chronological phasing. Through reconstructing the diachronic technological and provenance profile of the local ceramic sequence at this site, complemented through the analysis of samples from the partly contemporary neighbouring site of the Heraion, we were able to develop a nuanced understanding of pottery craft organisation and traditions in south-east Samos. Moreover, the discussion of ceramic and other exotica imports at both sites suggests that the local communities were engaged in maritime communication Aegean networks, with an apparent increase in the material visibility of connections in the Aegean LNL-FNL period. Building on these findings, our next step is to expand further our understanding of Samos' regional and interregional connections in later prehistoric periods. Finally, the presentation of analytical results from two MCh-LCh sites on Samos will allow comparisons with currently ongoing and future relevant studies.