‘ ORGANIC TEMPER ’ AND THE EARLY NEOLITHIC POTTERY PRODUCTION : INTERPRETATIONAL CHALLENGES

the hallmarks of the Early Neolithic pottery production at many sites along the Eurasian Neolithisation trajectories (e.g. Todorova & Vaysov 1993; Elenski 2006; Özdoğan 2011; Çilingiroğlu 2012; Vuković 2016). The hidden potential of this archaeobotanical inclusion (see Kreiter et al. 2013; 2014; Pető & Vrydaghs 2016; Mariotti Lippi & Pallecchi 2016) lies in revealing the interplay between two fundamental aspects, which defi ne the Southeast European Early Neolithic – the agricultural (farming/crop husbandry) and the technological cycle (pottery production) – integrated locally, within a settlement-specifi c environment. To explore this, however, it is essential to establish whether the vegetal remains are those of cultivated crops (cereal plant parts), as well as whether their presence in the fabrics plausibly refl ects the intentional addition of organic materials to the clay paste as temper, rather than incidental inclusion. Thus, the article is focused on these two main questions, rather than examining all possible research avenues in detail (see Fig. 1). Based on the contrasting preliminary results obtained from three Early Neolithic Eastern Balkan key study sites, located in present-day Romania and Bulgaria (Fig. 2), this study explores a series of challenges and potential biases when interpreting such vegetal remains in the Early Neolithic context. Temper, as a pottery-making component that refl ects shared values incorporated in technological activity (e.g. Stark et al. 2000), is traditionally studied from the perspective of fabric variation, to help outline cultural group membership and classify ware types (Rice 1987, 406). Here, a range of plant inclusion variables is considered with an aim to diff erentiate between the intentional adABSTRACT Well-preserved plant remains found in clay bodies of Early Neolithic pottery of Southeastern Europe have been largely understudied. The characteristics and provenance of this ‘organic temper’ remain mostly unknown, making interpretations obscure. Based on a range of research methods, this article explores the macro and micro plant remains within the pottery clays, considering such aspects as the use of domesticated versus wild plants and actual functional temper versus organic inclusions as background noise. This innovative approach is applied to explore three diff erent Early Neolithic Balkan sites, demonstrating the importance in distinguishing between (a) deliberate addition of selected temper as a technological prerequisite; (b) sporadic occurrence of plant parts in (domestic) areas where pottery was made, (c) natural characteristics of the local clays containing organics and used as raw materials, and (d) plant use pointing towards more specifi c pottery-making techniques. Possible misinterpretations and pitfalls are discussed in using the applied integrated methodology, thus revealing crucial details on the variability of the technological approaches applied during the Early Neolithic of Southeastern Europe.


INTRODUCTION
Plant use in pottery production has been recognised in various prehistoric-and historic sites worldwide (e.g. Rice 1987;Fuller et al. 2007;Mariotti Lippi et al. 2011;Vrydaghs et al. 2014) and yet, the specifi c features of such vegetal inclusions often remain underexplored, especially in Southeastern Europe. Referred to as 'organic matter' or 'organic temper' (Spataro 2009(Spataro , 2011Kreiter et al. 2014) and usually as 'chaff ' (cf. Starnini et al. 2007;Spataro Fig. 1. Plant materials in the clay used for pottery production: some study aspects.

Acta Archaeologica
Downloaded from Brill.com12/25/2021 02:45:54AM via free access dition of plant temper (cf. Doherty et al. 2000;Mariotti Lippi & Pallecchi 2016) versus the accidental occurrence of the organics from site-specifi c context. It is necessary to study the provenience of plant material and determine the functionality of the temper components within the clay fabrics with variances such as (a) domesticated plants versus wild plants and (b) intentional temper versus background noise. Such a systematic approach allows an inference to be made as to whether the addition of organic temper was a functional prerequisite due to technological limitations of the local clays, intentional but based on other decision-making factors, or whether it merely refl ects accidental occurrences (Fig. 3). Whereas intentionally added organics reveal the complex interaction between technological choices and subsistence patterns within contemporary local environments, the background noise plants usually point towards procurement strategies and social locations where clay was collected and pottery made. There are various caveats to the possible interpretation of the true complexity of this agricultural component, embedded within mate-rial culture, especially when interpreted in the context of socially transferred ceramic traditions, technological and agricultural landscapes, and regional subsistence patterns.

ORIGIN OF PLANT MATERIALS: POSSIBLE ORGANIC INCLUSIONS
Chaff alone has been suggested as a deliberately added 'functional temper', but close examination of the vegetal inclusions is imperative to verify this claim. If chaff was used as temper, a series of individual vegetal components should be present in the clay paste, potentially refl ecting specifi c crop husbandry strategies and grain processing stages (see Hillman 1984a, 1, 39;Bogaard et al. 2017;Kreuz et al. 2005). These should be identifi able by their unique features (organics type, state, amount, fragmentation, shape and composition, et cetera).
Chaff -the non-grain parts of the cereal plant -can include the bracts, the protective part of the seeds (light chaff ), or straw fractions (heavy chaff ) (Hillman 1984b),

'Organic Temper' and the Early Neolithic Pottery Production: Interpretational Challenges
Downloaded from Brill.com12/25/2021 02:45:54AM via free access man 1984b, 128, 146) (Fig. 4). Each of these processing stages results in a clear separation between the 'heavy chaff ' (straw fractures eliminated at the early processing stages) and the 'light chaff ' separated further in a series of steps (second threshing and dehusking, secondary winnowing and grain dunking) (ibid.). Aided by the lightly blowing wind, winnowing further separates the loose cereal husks, thus aff ecting chaff composition (Hillman 1984a, 55;Jones 1984, 45). It separates the light straw and chaff from the grain -for free-threshing wheat, the chaff is removed during the fi rst round of primary winnowing, together with the light sometimes also fi nely chopped when used as temper. The thin bracts surrounding the crop seeds form a dry, inedible husk which must be removed before consumption (Hillman 1984a;1984b;van der Veen 1999). Whereas glume wheat needs multiple dehusking procedures because the grains are tightly invested in the hull, the ears of the freethreshing wheat can immediately break up and release the grain in fewer processing steps (Hillman 1984a;1984b;Jones 1984). Whereas threshing directly separates the grains from the chaff for free-threshing cereals, it breaks the hulled cereals into spikelets, which requires a second threshing procedure, usually pounding or grinding . Origin of plant materials in the clay fabrics used for pottery production. Summary of the eff ects of tempering on properties of the clay. After Rice 1987. Fig. 4. Processing stages of glume wheat (A) and free-threshing plants (B). The types and components that are remaining after (1) threshing and (2) second threshing/pounding (modifi ed after Hillman 1984;Jacomet 2006;Zochary et al. 2012;Bogaard et al. 2017). A1 -glume wheat spikelet after threshing; A2 -glume wheat spikelet components (after second threshing/pounding). 1 -individual spikelet comprising the grains; 2 -individual spikelets, fragmented: a -rachis (the stem within the ear) with glume base (where the glume is attached to the rachis) and spikelet fork (where two glumes join); b -fragmented glumes (the structures that are enclosing the spikelet); c -fragmented lemma (a membrane on the dorsal side of the grain) with part of the awn (projection); d -grain (primary product); e -palea (a membrane on the ventral side of the grain). C -imprints of cereal chaff components on studied ceramic fragments (spikelet fork with glumes -3; glumes -1, 2, 4; paleas and lemmas -5, 7; rachis -6).

'Organic Temper' and the Early Neolithic Pottery Production: Interpretational Challenges
Downloaded from Brill.com12/25/2021 02:45:54AM via free access it is not only the type but also the abundance of chaff remains that is informative of the various processing steps. For example, the higher quantity of spikelet forks and glume bases may indicate parching and third sieving, whereas their lower number is associated with the fourth sieving and hand sorting (Hillman 1984, 10, Table 1). Although these typically signal the third sieving, they still may occur sporadically in other cereal processing stages. Similarly, the higher number of intact non-basal spikelets is characteristic of a parching stage. In contrast, only a few of these are present after the fi rst and especially after the second sieving. Rachis internode segments, on the other hand, can be found in similar amounts in a range of steps (from parching to the third and fourth sieving), and basal spikelets, too, are present from the raking to the second sieving. Once fully processed, the storage of clean grains ready for consumption diff ers from that of spikelets or semi-processed grains that need further dehusking at the household level (cf. Hillman 1984a, 8-9;1984b, 126). However, when attempting to study storage practices, plant temper would not provide unidirectional data. The presence of glumes in organic-tempered pottery should not be indicative of the scale of the outdoor activities seen in contrast to daily piecemeal processing or the timing of dehusking, as glumes can be removed either before the storage or after it, immediately before the consumption of cereals. Furthermore, although glume wheat was bulk-stored as spikelets in various regions (Bogaard et al. 2017, 13, 16 for Anatolia;Jones 1981 for Greek North Macedonia;Marinova 2006 for Bulgaria), the presence of single preserved spikelets and sporadic grains in clay fabrics may be, but are not necessarily, related to specifi c storage practices.
Additionally, another variant of chaff -containing temper technically refers to the use of animal dung for pottery production (e.g. Ganetsovski 2015). However, its origin and characteristics demonstrate a completely diff erent technological approach and pottery-making traditions. Therefore, in order to interpret plant parts in ceramics, it is essential to diff erentiate between pure chaff (as resulting from cereal agriculture) and dung, which happened to contain semi-digested cereal husks. Since dung of various locally grazing and browsing domestic animals includes a variety of vegetal types, depending on the animal species and the season, it comprises not only variable proportions of chaff but also wild plants (see Charles 1998;Vergès et al. 2016). straw (Hillman 1984b, 128). The next steps, coarse and fi ne sieving, separate the components further according to size (Jones 1984, 48), the light chaff by-product usually resulting from secondary winnowing. Collected before sieving and stored separately from the straw, this light chaff is used as fi ne temper, fuel or fodder (Hillman 1984;Jones 1984).
The variable archaeobotanical sampling strategies applied during the excavations at diff erent prehistoric sites may result in a dominance of grains in the assemblage (collection of the visible plant remains) or presence of both grains and chaff (systematic sampling) (Marinova 2006;Bogaard & Halstead 2015). At understudied sites and regions, previously the focus was usually on the better preserved charred prime products (grains and spikelets) (Hillman 1981,11-12, 32); however, this record can be broadened by the analysis of the organic temper comprising well-preserved cereal processing by-products. The short-lived coarse and fi ne-sieve by-products were often used as fodder or fuel (Jones 1984, 47), but the fi nest chaff perishable parts (straw internodes and lighter nodes, leaf fragments, light chaff components, together with most of the lighter weed seeds) (Hillman 1981,11;Boardman & Jones 1990) can also be present in chafftempered pottery.
Depending on the plant types and the processing stages, the by-products may include various chaff parts: the stem of the cereal plant below (culm) and within the ear (rachis); whole or fragmented individual units of the cereal ear comprising the grains (spikelets) (Jacomet 2006;Zochary et al. 2012;Pearsall 2015) (Fig. 4). Fine chaff components present can also include the glumes surrounding each spikelet, a glume base attached to the rachis, the parts holding the grain (dorsal side membrane lemma and ventral side palea), the awn projections and a spikelet fork joining the glumes at the bottom of the spikelet (ibid.). The fi rst threshing of glume wheat results in a mix of spikelets and trodden straw; the primary winnowing is associated with spikelets, heavy straw fragments, weed seeds and heads; re-threshing and re-winnowing are refl ected by the light straw, some spikelets and straw nodes; and fi nally, coarse and fi ne sieving includes spikelets and small weed heads (Hillman 1984a;1984b). As should be apparent, each process leads to diff erent chaff products. Nevertheless, even when identifi able, the plant types and parts found in clay body do not refl ect a closed system. While each stage produces diff erent products, often, products, the organics naturally present in clay deposits, the wild plants that can be intentionally used as temper, those arable weeds that are found in combination with cereal chaff inclusions and the plants ingested by livestock, resulting from foddering and grazing and deposited onsite/off -site as dung.

IDENTIFICATION OF FUNCTIONAL TEMPER
To evaluate the technological signifi cance of the organic temper and to distinguish between the functionally sufficient amount of organics and occasional occurrences (insignifi cant amount in terms of modifying the technological properties of the clay), the organic-containing fabrics are examined as a system of interdependent constituents (clay matrix -mineral inclusions -organic inclusions).
Temper is a non-plastic material added by the potters to modify the fabrics, thus satisfying specifi c technofunctional requirements -to control plasticity, to prevent shrinkage and cracking from drying and fi ring ceramic wares, etc. (Rye 1981;Rice 1987). By mixing clay and temper, it is possible to achieve the proper texture, consistency, hardness, improving the clay properties when wet or dry, both during and after fi ring. As plant temper is expected to burn out and does not contribute to the The specifi c features of the chaff do echo certain agricultural practices and grain processing sequences, but what if the vegetal inclusions present in the fabrics were not cereal husks? Wild plants have also been intentionally used as functional temper (grasses, cattail fuzz, algae, rushes, chopped reeds (e.g. Rice 1987, 407; Hunt 2016; Doherty 2020, 54)). Importantly, wild plants still could well have been natural constituents of the raw clay sources (e.g. Ting & Humphris 2017), or present by chance in the clay paste (e.g. raw materials stored in various settlement contexts). Further complexity is introduced by the possible random occurrence of arable weeds, which cannot be considered as intentionally added temper. As arable weeds are connected with fi eld management and the immediate natural environment, they must be examined with the crops (Jones 1984;Bogaard et al. 2017;Green et al. 2019).
Plant inclusions in pottery provide data on diff erent agricultural and husbandry systems and their connection to the newly emerging technology of pottery making. For analysts, the awareness of the various possible vegetal types helps to identify potential non-chaff sources and calls for site-specifi c methodological requirements for identifi cation of both chaff and temper. As interpretation depends on the types of the plants present in pottery, it is essential to distinguish between the crop processing by-

'Organic Temper' and the Early Neolithic Pottery Production: Interpretational Challenges
Downloaded from Brill.com12/25/2021 02:45:54AM via free access adding organic temper. The variable quantity of vegetal inclusions in the clay fabrics is of crucial importance, as to whether intentionally added or occasionally present, too few plant inclusions do not aff ect the technological properties of the clays. Sporadic organic inclusions in the clay paste -meaning in very low frequency (1-5%), or as single plant fragments, do not aff ect the fabric quality due to their exceptionally low quantity, i.e. their presence most probably does not refl ect purely technological requirements. Interpreting the reasoning behind the intentional addition of too few organic inclusions thus becomes more complex.
The interrelation between the properties of clay fabrics and the added organics (or the lack of such) is also important. The silty and the fi ne-grain sandy clays (Rice 1987;Velde & Druc 1999) do not usually require the addition of temper -their technological qualities are suffi ciently high, even when the raw materials were used unmodifi ed, in their natural state. Consequently, the addition of organics to fabrics that would not require tempering was not technologically necessary sensu stricto. Given the typical amounts accepted for 'the actual temper' (above 17% for organic temper, cf. Velde & Druc 1999, 145), the added temper here is expected to range between at least 10-15%, but more often falling between 20-30%. To interpret the addition of too little/too much organic temper, and to reveal the actual sources of plant material, many factors have to be considered, including the local environment, the agricultural and husbandry practices, the site-specifi c clay deposit characteristics and the archaeobotanical evidence.

METHODOLOGICAL BACKGROUND AND MATERIALS
The ubiquitous plant-containing and low-temperature fi red Early Neolithic pottery may include: (a) domesticated and wild plants added as temper, (b) 'background crops' indicative of various activities taking place in a domestic setting and/or in areas of production, (c) 'background wild plants' that reveal the environment from where the raw materials were sourced. To understand the integration of agricultural and technological activities, together with environmental and cultural contexts at each site and regionally, a set of criteria is required.
In bonfi res, when a clay vessel is not uniformly oxidised throughout, the organics may not burn out com-stability of the ceramic body, it has been considered as providing a reducing environment and serving as a plasticity modifi er (Velde & Druc 1999). Nevertheless, other factors, such as porosity/impermeability, cooling properties, benefi cial properties for storage vessels, as well as the improved transportability of 'weightless' organictempered vessels have also been discussed (e.g. Skibo et al. 1989;van Doosselaere et al. 2014).
Before commenting on the various aspects of the study of 'organic temper', it is necessary to establish, for each site, if plant inclusions found in the clay body should be considered as temper (Fig. 5). To enhance the inherent properties of the raw materials, a certain amount of added material (temper) may be required, and if so, it must be present in suffi cient quantities, since the sporadic or inadequate quantity of inclusions would not change the characteristics of the primary raw material (cf. Rice 1987; Doherty et al. 2000). Functional temper is thus diff erent from non-functional inclusions and background noise (single organics found in the clay body). Furthermore, plant temper cannot be considered alone, without reference to the properties of the clay fabric and the mineral, inorganic inclusions. The complex integrated system of clay fabrics -consisting of various components with specifi c properties -may reveal the reasoning behind the practice of adding organic materials in the clay paste. The contrast between deliberately added plant material and 'background noise' (inclusions having no technological bearing on the clay properties) thus sheds light on why organic temper was added and how to interpret its variable quantities.
Although the intentional addition of vegetal parts as temper may well refer to the wild plants, the latter can also be naturally present as organic-containing clay deposits. For example, the presence of fi ne roots as natural inclusions in the raw material may be indicative of preferred clay deposit areas but does not refer to a potters' technological extra step of deliberately adding plant parts to the clay fabrics. The occasional presence of fossil plants (cf. Cleal & Thomas 2019), of natural plant inclusions in coal-rich soils and fi ne-organic inclusions in mountainous clay deposits is evidence of similarly unintended organic inclusion. The occurrence of such vegetal parts diff ers from the intentional use of wild plants as temper (see above). Both the type of plants and the amount of organics in the clay paste (concentration) are thus indicative of the reasons for fi ed appearance, the possible fragmentation or the 'clustering' of diff erent plant components, especially when these are placed one on top of another. However, potential identifi cation biases when examining the plant parts and species, especially within the walls of the fragments, are compensated by the possibility to study the actual sources of the plant remains (original state vs processed; dry state vs wet). High-resolution microscopy can potentially identify microremains indicative of specifi c plant taxa or vegetal parts. However, various vegetal parts may also have similar phytoliths, and diff erent plants may produce identical opal silicates (Piperno 2006). Although phytolith assemblages can be identifi ed and there are recommendations towards standardisation concerning morphometrics (Portillo et al. 2006;Ball et al. 1999;, various limitations should be considered, including the similarities between the silica skeletons of diff erent plant parts, the resemblance of cultivars to some grasses and the range of micromorphological features within parts of the same plant, as well as modifi cations caused by heat (see Lu et al. 2009;Out et al. 2016, Fig. 7;Portillo et al. 2020). Whereas the plant remains are usually poorly preserved in the thin sections of pottery fragments, the scanning electron microscopy (SEM) allows for a detailed examination at very high magnifi cation; yet, it may refer to modifi ed shapes without the fl at plane position which is necessary for making precise measurements.
In this study, both macro-and microorganic inclusions are considered within complex system of the fabric, in particular the variable ratios between clay paste, inorganic and organic inclusions. The plant remains are examined by low-power optical, high-power polarised and scanning electron microscopy in combination with selected reference collection specimens. A set of sixty samples from three Early Neolithic settlements (20 samples per site) was selected as representing a spectrum of inclusion types, concentrations and arrangements on the surface and in section, of both coarse undecorated and fine, high-quality white-onred painted vessels, of various shapes and thickness (0.5-2.5 cm). The freshly broken pottery fragments have first been examined in both sagittal (longitudinal) and coronal (frontal) plane by using a low-power Leica EZ4 Stereo Microscope (8× to 35× magnification range) according to the adapted approach for pottery studies (Whitbread 1995). The descriptions of the shapes and the concentrations (frequency) of pletely -resulting in the preservation of lightly to heavily charred plant fragments captured between the outer and the inner surfaces of the vessels. With higher fi ring temperature, it is the void plant impressions on the surface of the vessels and the microfossil remains (opal silica phytoliths) within the clay body that can possibly inform on the type of added organics. Here, the preliminary observations refer to two locations: the fi rst is the very surface of the fragments, revealing void negative impressions (a), and the second is the clay paste within the walls of the vessels, containing charred fragments (b), microfossil remains (c), and positive and negative plant impressions (d). Within the walls, the two-dimensional imprints of vegetal parts often have skewed, smashed, turned and twisted shapes resulting from the origin of the plant temper and its specifi c preparation and mixture with the clay, thus diff ering from the usual form and state of plant components.
Though otherwise crucial when reconstructing past agricultural practices, contexts and processing stages, the abundance of indicative plant parts is not the primary factor when studying organic temper in pottery. Meticulous counting of specifi c vegetal parts or species per fragment would refl ect a series of chance factors (size and thickness of the analysed fragments, clay-organics mixing properties, cross-section location, etc.). The observations on the plant temper and the ratios between a) various crops identifi ed within a pottery fragment, b) crops and wild plants, and c) specifi c plant parts are to be considered as only a partial snapshot on just some of the chaff stored in the households, used in pottery production and registered archaeometrically.
Here, the data are used opportunistically and qualitatively, rather than quantitatively and according to counting-based analysis. Not just the quantity but the distribution of the vegetal parts within the studied zones may also vary depending on the degree of homogeneity of clay-organics mixture, the location of the cross-section, the chosen surface area and the level of eff ort when smoothing the surface of the vesselsfactors that may result in varying concentrations of organic parts within a single fragment. The clay matrix characteristics and the mineral inclusions contained in the paste also have a bearing on the distribution modes of the plant parts within the fabric, whereas forming techniques and the pressure used to shape the vessels may additionally aff ect their orientation.
The identifi cation of plant morphology with the help of low-power microscopy is often hindered by the modi-

THE STUDY SITES
The pottery fragments were collected from three Southeast European open settlements with contrasting geological setting, landscape and microclimate (Fig. 2).
(1) Măgura-Buduiasca consists of pits attributed to Early Neolithic Starčevo-Criş culture, representing the initial occupation of the site and dated to the fi nal part of the 7 th mill. BC van As et al. 2004;Thissen 2005;Walker & Bogaard 2010). Located in the wide and open Danubian plain -low and alluvial homogeneous region with massive loessic accumulation, it occupies a prominence on the secondary eastern Teleorman River terrace, about 300 m from the river course Macphail et al. 2008). The site is situated atop of relatively hard bedrock, with a thick sedimentary mantle comprised of gravel and sand deposits, and is covered by up to 20 m thick loess and loess-like layers (ibid.), showing the natural range of alluvial deposits. Accentuated mineralisation of the soil organic matter, evidence of gleysation and salinisation processes are demonstrated by the local transitional steppe and silvo-steppe soils (Pârvan et al. 2011, 109-111;Blaga et al. 1996). The middle-upper Pleistocene loess deposits consist of reddish, yellowish clay, sandy silts and carbonate concretions overlying the marly alluvial-lake deposits (Coteţ 1973, 371;Macphail et al. 2008). Considering the production of pottery, the organic temper used has been reported as typical component of the fabrics at the site (van As et al. 2004).
(2) Dzhulyunitsa (6100-5700 cal BC) (Krauß et al. 2014) is located on a natural prominence -a plateaulike terrace above the Zlatarishka River (Veliko Tarnovo region), between the valley and the foothills, with several freshwater sources available immediately nearby. The river emerges onto the flat ground of a loess-filled depression and erodes sandstones and calcareous clays, combined with washed-in weathered loess from the hills (cf. Hristov et al. 2010). The Lower Cretaceous marl-limestone-sandstone facies and the Pleistocene and Holocene alluvial and colluvial deposits underlie the loessic sediments (black soil parent material) (Fotakieva et al. 1976;Hristov et al. 2010). The weathering of fine wind-blown loess in the region produced the naturally fine-grain clays around Dzhulyunitsa. Vegetal inclusions in variable quantity are registered in about two-thirds of the fragments (Elenski 2006;Dzhanfezova et al. 2014). void plant imprints on the surface and the plant remains preserved in the clay paste are based on adapted comparison charts (e.g. Rice 1987). Identification of the clearer imprints was attempted following Jacomet (2006) and Zochary et al. (2012).
Examination of the fabrics and the vegetal remains was then performed at higher magnifi cation -on 30 μm thick vertical thin sections, using polarised light optical microscope Leica DM 2500P (5× to 50×), in the sagittal plane and according to Vrydaghs & Devos (2020), allowing observation of the surface-core interaction, the oxidation transitions resulting from diff erential fi ring, and the distribution of the inclusions due to forming techniques. These interactions would not be visible if analysing the wider areas displayed by the coronal thin sections (more vegetal parts per cm). Specifi c features of the voids (mostly vesicular empty areas) left in the clay paste, estimation of the porosity and evaluation of the boundaries, were also taken into account (cf. Doosselaere et al. 2004), rather than focusing on the morphology of the plant inclusions as indicative of surviving distinctive features of various taxa (cf. Moskal del-Hoyo et al. 2017), which were usually moderately to poorly preserved in the thinsections analysed in this study.
Fragments containing plant parts were studied further as carbon-coated resin blocks and stub mounts using a Jeol 5910 scanning electron microscope with an Oxford Instruments INCA 300 energy dispersive x-ray spectrometer (SEM-EDX, <10000× magnification). As no high-precision chemical analysis was needed, and the settings were defined according to the delicate organic structures, the acceleration voltage of 15 kV was preferred with the filament current of 40 μA. The SEM observations were focused on the fabrics and the preserved phytoliths also according to (Lanning & Eleuterius 1992;Ball et al. 1999;Ball et al. 2009;Piperno 2006;Heiss et al. 2020), following the International Code for Phytolith Nomenclature (ICPN) ver. 2.0 (Neumann et al. 2019). Plant reference collections in the Archaeobotany Laboratory, School of Archaeology at the University of Oxford were used to aid identification of both vegetal macro-and microremains. The latter was prepared using sodium hypochlorite to remove the plant tissue, and mounting the phytoliths on standard glass slides (Triticum sp., Hordeum sp. and Phragmites sp. plant parts).

Preliminary low-resolution stereomicroscopic observations
The studied fragments from the three sites reveal wellpreserved and distinctive surface imprints and organic inclusions in clay body. At (1) Măgura-Buduiasca the undecorated thick (up to 15 mm) fragments contain about 20% organics, corresponding to the percentage of the mineral inclusions within the fabrics. The maximum length of the preserved organics is usually 2 mm, sometimes reaching 6 mm. The light chaff parts are present on the surface (imprints) and within the paste (imprints and charred material). Intentionally added mineral temper was not registered in the fi rst set of studied fragments. According to the morphological specifi cs of the imprints, at this stage, the registered remains are associated with    The identifi ed plant parts correspond to those registered at Măgura-Buduiasca, yet Dzhulyunitsa shows optimal preservation of sizeable vegetal temper components -such as the intact, charred spikelets with preserved glumes, paleas and lemmas found in the clay paste (Fig.  6a-c). The plant inclusions are often grouped in clusters, although their size and concentrations vary. The appearance, state and plasticity of the organic inclusions (in some fragments showing denser clustering of vegetal inclusions) suggest a possibly more liquid type of clay-temper admixture, at least in some of the cases. The variable concentrations of plant parts, usually ranging between 20 and 40%, prove no direct correspondence with the naturally variable fabrics, associated with the local clays.
Whereas the site of Măgura-Buduiasca and Dzhulyunitsa share certain similarities, the fabrics at (3) Ilindentsi-Massovets diff ered considerably. Although vegetal inclusions are usually considered as absent in the typical southwest Bulgarian Early Neolithic pottery repertoire, quite a few fragments analysed in this study contained organics, which were detected only microscopically. Often these occurred as single instances in very low quantities (1% to 5%); and when the technological properties of the clay are considered, such amounts would be indicative of non-functional temper.
The fragments showing higher quantities of plant parts (above 10%) are usually 9-20 mm thick, whereas occasional vegetal remains belong to the 5-10 mm thick vessels. The highest percentage of added organics, 30%, is registered in a 20 mm thick vessel. Still, even thicker fragments may contain only 7% organics, i.e. the relation between the thickness and the concentration of plant temper may not always be straightforward. Plant parts were also found in high-quality painted fragments, but unlike the two other northern sites, here the vegetal inclusions are hidden below an additional organics-free thin clay layer that covers the tempered body (i.e. very few visible imprints are usually present on the surface of painted wares).
The shape, sorting and size of organic particles also vary considerably -from poor (most often) to very well sorted (in rare cases). In cross-section, the sizeable inclusions are usually oriented parallel to the walls of the vessel as a result of the forming technique, and they rarely have a random distribution. The preserved length is usually between 1 and 10 mm. The association of organic temper and other mineral inclusions within the same frag-glume wheat, mostly einkorn (Triticum monococcum) and emmer (Triticum dicoccum) (Fig. 6d-f). So far, no intact charred spikelets nor grains were found preserved in the paste of clay bodies. Plant parts such as spikelet forks (Fig. 6d-f), together with the rachis and glumes, are common surface imprints, their maximum size reaching 2 mm. The glumes, paleas and lemmas dominate, and only occasional culm fragments were present both on the surface and within the walls. The infl orescence bract components are usually not found in coherent clusters or aggregates one on top of the other thus perhaps indicating continuous mixture and homogeneity of the clay paste, and possibly a not too wet state of the organic inclusions.
The sorting of inclusions is low to moderate; the longest inclusions tend to be parallel to the walls of the vessels, due to the forming techniques. Most often, the surface does not include an additional layer to cover the plant-rich paste and thus to form a more delicate fi nish. Organic inclusions are also found on the surface of smoothed/polished and decorated vessels. Although the coarser and thicker fragments tend to contain more vegetal parts, no discernible fi xed ratios between the amount of temper and the type of fabric exist.
The plant material at (2) Dzhulyunitsa shows quite variable quantities. Some of the fabrics, having higher ratios of clay to silt and sand, tend to contain more organic temper, presumably to control the shrinkage, whereas sandier fabrics include less vegetal material. However, in some cases, only tiny additions of chaff temper are present. As in Măgura-Buduiasca, no correspondence was detected between the use of temper and the quality of the surface fi nish of medium-to thick-walled vessels.
No association whatsoever between specifi c fabric and surface fi nish/decoration was registered (Fig. 7). The same amount of vegetal inclusions is observed in pottery fragments with similar thicknesses but with traces of varying and distinctive surface treatment (coarse vs decorated). Heavily tempered vessels can either have very coarse surfaces or polished and fi nely worked decorated walls. At the same time, specifi c categories such as the quality 'red-slip' pottery also demonstrate signifi cant variability, some examples showing lots of added organics, othersa small degree of plant inclusions, or none at all (Fig.  7B, 1-3). The surfaces of some of the white-painted wares contain sporadic plant imprints, too. Specifi c correlations were not established between the fabric, the thickness of the walls and the diameter of the vessel either.

High-resolution microscopic observations
The observations on the fi rst studied collection of petrographic thin-sections were focused mostly on the clay fabrics and the voids left by the organics, rather than examining in detail the surviving single phytoliths. In terms of the surrounding soil matrix, the few registered preserved phytoliths had moderate to bad visibility, following the terminology in Vrydaghs & Devos (2018). Given the function of plants in the fabrics and to the process of forming the vessels, the preservation of phytoliths derived from these plants refers mainly to technological fragmentation rather than showing traces of sedimentary transportation. The moderately preserved light-tanned to brownish phytoliths are sometimes accompanied by wellpreserved elongate dendritic phytoliths (Fig. 8, bottom right). ment reveals that fabrics having vegetal inclusions may also contain mineral grains of a bigger size. The usual 10% of other (mineral) inclusions, however, are not associated with any strict amount of added organics. In this study, the sporadic single organic inclusions in the clay paste -those found in very low frequencyusually range between 5 to 10 % and sometimes even between 1 to 5% (Fig. 5). In extreme cases, only single plant parts (reaching 2 mm in length) can be detected, and these alone do not aff ect the clay properties due to their exceptionally low amount. Interestingly, at Ilindentsi occasional non-chaff plant parts and impressions, indicative of wild plant species, possibly Avena, have also been registered (Fig. 6g).
In line with its location, Ilindentsi represents the use of diff erent raw materials and also various technological preferences. The mineral inclusions have high density and big size (< 2 mm), most often angular grains, medium to well-distributed, including polycrystalline quartz, weathered feldspars, iron-rich clay pellets, variable amounts of mica, predominantly biotite, occasional amphiboles, metamorphic and granitic inclusions. The continuous range of mineral fabrics is dominated by medium-to high-grade regional metamorphic facies and granitic material, consistent with the geology of the higher ground that immediately borders the Struma River valley. The clays belong to the fi ner grades of the Neogene sediments exposed at shallow depths nearby and eroded from the uplands bordering the Struma Graben on three sides. All mineral inclusions are natural and consistent with the locally available Neogene sediments. A very high proportion of angular mineral inclusions are indicative of the metamorphic and the granitic rocks that border the eastern area of the Struma Valley, and except for a small number of sherds that contain plant fragments, the fabrics are not tempered. The fragments refer mainly to proluvial valley side deposits.
According to the SEM analysis, well preserved and connected cells (silica microremains) are regularly observed at all three sites. The plant tissue is usually badly to moderately preserved at Ilindentsi (probably due to mechanical degradation or melting), and moderately to well preserved at the two other sites. The phytoliths were found mostly within the clay body rather than nearer the surface of the vessels. Predominantly sheet/multi-celled elements with articulated epidermal elongate cell and papillate phytoliths from infl orescence tissue, potentially also of Triticum sp. or Hordeum sp. plants, sporadic acute bulbosus (prickle) phytoliths from lamina tissue, and dendritic phytoliths from infl orescence tissue (see Ball et al. 2009) are registered at the three sites (Figs. 9, 10 & 11). Multi-cell and especially anatomically connected phytoliths (vs single or individual) (e.g. Shillito 2011, Portillo et al. 2017) are common, often consisting of more than ten parallel rows (Fig. 9) and sometimes revealing consecutive layers of various tissues (Fig. 9b). Prevailing kind is of the elongated dendritic cells, which are anatomically connected, but they are often distorted (Figs. 9 & 11d).
The cell structures and the phytolith assemblages are The Ilindentsi vegetal inclusions in thin-section demonstrate lower quantity and variety of shapes of phytoliths -these are mainly acicular, but also often wavy and bent, whereas the Dzhulyunitsa and Măgura-Buduiasca inclusions demonstrate variable shapes, often combined with channel and crack voids. The boundaries between the matrix and the plant inclusions are usually sharp at the northern sites and sharp to merging at Ilindentsi. The circular and subcircular cavities are often accompanied by residual material, which in the northern sites occurs more regularly and from fragments of larger size.
Whereas the fabrics at Măgura-Buduiasca and Dzhulyunitsa are very fine and loessic-related, those at Ilindentsi are predominantly colluvial/proluvial and naturally containing high quantities of other, mineral inclusions. At Măgura-Buduiasca, despite the homogeneity of the fine loessic raw materials, there is a slight variation in the fabrics that reflects the availability of naturally sandier local areas. Mineral temper was not established in the first set of studied samples the naturally contained mineral inclusions being too small, usually well sorted and subrounded, reflecting the local geology and dominated by single bigger (< 0.02 mm) monocrystalline quartz and sporadic feldspar grains, as well as calcareous, iron-rich inclusions and mica. The registered added inclusions refer to vegetal temper, which is spread regularly throughout the walls of the vessels in cross-section and is better visible in the middle area of the cross-sections -dark grey to black due to reduction conditions and variations of oxidation during the firing.
The Dzhulyunitsa fabrics are similar but much more diverse, presenting a broad spectrum of the local geological continuum. They are fi ne-grained, deriving from wind-blown loess weathering. The mineral inclusions in the loess-based fabrics are tiny (< 0.2 mm), refl ecting the local setting and upstream geology. They contain angular and subangular quartz grains, feldspar, muscovite mica and limestone as well as inclusions of iron oxide. In both textural and mineralogical terms, the local pottery fabrics match the natural sediments without showing signs of mineral temper. Only vegetal material is added in variable quantities to vessels of variable sizes, thickness, shapes and surface treatment. Some of the clay-rich fabrics tend to include more organic temper, but given the range of variable quantities, this is not a consistent relationship. Some of the thin-sections demonstrate that

'Organic Temper' and the Early Neolithic Pottery Production: Interpretational Challenges
Downloaded from Brill.com12/25/2021 02:45:54AM via free access Fig. 9. Secondary electron image microphotographs of phytoliths registered in fragments from the studied sites, various scale. a -an assemblage of wellpreserved phytoliths, Dzhulyuintsa; b -layers of phytoliths and a dendritic phytolith, Dzhulyunitsa; c -moderately preserved phytolith assemblages, Ilindentsi; d -phytolith assemblages with stomata, Dzhulyunitsa.   rinova 2017). As in present times, the immediate vicinity of Dzhulyunitsa was associated with riparian forests and shallow water bodies with standing or slowly running water (Marinova & Krauß 2014, 190), and wild plants and wetland vegetation, such as water chestnut (Trapa natans) and club-rush, were used at the site (ibid.). According to the present preliminary observations, their exploitation in the domestic activities did not extend to pottery production -at this stage, there is no evidence for the addition of wild plants as functional temper in the studied pottery from the three sites. Instead, undigested crop components (cereal chaff ) were used in pottery production and presently, apart from the cereals, no other crops (e.g. leguminous crops, fl ax) were found in the clay paste. Still, the presence of wild plants was identifi ed at Ilindentsi. Their sporadic occurrence and the plant type, Avena, are not indicative of the use of wild plants as a temper; they instead point to species defi ned as arable weeds, which should be considered along with the cereals. Concerning the potential confusion of fossil plants and the vegetal parts naturally contained in clay deposits, or fi ne root parts in the soil, the well-preserved and connected cells frequently observed in these samples are not likely to be purely sediment-derived plant material, as such bioclasts are usually considerably altered (see van Doosselaere et al. 2014) and do not refl ect the micromorphology of the roots.
At this stage, no traces of chaff digested by animals (see Charles 1998;Valamoti 2013) have been registered in the studied collection. Although the absence of recognised calcitic spherulite-rich assemblages (cf. Canti 1999; Portillo 2020) may not necessarily indicate the lack of dung components (as these are not always preserved), the size, preservation and intactness of the vegetal remains, as well as the type of the organics-fabrics correspondence rather point to the presence of unprocessed chaff in the clay matrix. Charcoal or ash-derived plant materials can also be excluded as the source for the organic temper, according to the presented preliminary observations.
The opportunistic qualitative observations of the plant inclusions in pottery are not suffi cient to comment on the specifi c percentage correlations between diff erent crops: einkorn, emmer and barley. According to targeted archaeobotanical studies at Măgura-Buduiasca, einkorn and (hulled) barley are well attested throughout the Neolithic (and possibly 'new type' glume wheat spikelets, at least in the later contexts) (Walker & Bogaard 2010). It indicative of the infl orescence bract components of cereal plants, probably mostly glume wheat (Figs. 9a-c & 11). Well-preserved phytoliths frequently include areas with a higher concentration of papillate (Fig. 10b-c) without epidermal appendages such as acute bulbosus/hairs (microhairs or prickles), which, in combination with surrounding cells and according to the used phytoliths reference collection of crops can be found in the central zones of glumes, paleas and lemmas in Triticum and Hordeum species. However, small depressions or pits near the edges of the papillate, typical of H. vulgare and T. aestivum (Rosen 1992;Tubb et al. 1993;Moskal del Hoyo et al. 2017;Hayward & Parry 1980), have not been registered. Plant parts containing stomata, characteristic mostly of the culm/leaves (Fig. 9d), are rare. Transverse cells, potentially associated with the aleuron layer and the endosperm (bran) (Heiss et al. 2020) were also occasionally present (Fig. 10d), whereas single trichome (hair) base phytoliths were not detected at this stage.

Origin of the plant materials in the clay
The studied regionally-specifi c fabrics are all local and expected to contain local (vs imported) vegetal inclusions. The surface of the vessels (imprints) and the clay body (imprints and charred plant parts) of fragments from the three studied sites yielded cereal chaff macro-remains (glume bases, spikelet forks, whole spikelets, rachis fragments) indicative of Triticum and possibly Hordeum species. The study of the plant remains at micro-level revealed the absolute predominance of phytoliths typically associated with the infl orescence bracts of cereals, potentially glume wheat. It should be noted that such cell structures may be very similar to those of the wild Hordeum, Aegilops and other wild grasses (Poaceae -especially genus Bromus) and crop progenitors (cf. Acedo & Llamas 2001;Moskal-del Hoyo et al. 2017). Still, even if the crop imprints on the surface of the vessels are not taken into account, the identifi able macro-remains in the fabrics, and especially the preserved crop spikelets in the clay paste, point towards the addition of wheat chaff .
Past environmental conditions at Ilindentsi and Dzhulyunitsa reveal that up to 60% of the wood charcoal assemblages consisted of light-demanding or riverside woodland taxa (Marinova & Ntinou 2017;Kreuz & Ma-

'Organic Temper' and the Early Neolithic Pottery Production: Interpretational Challenges
Downloaded from Brill.com12/25/2021 02:45:54AM via free access may also refl ect free-threshing cereal processing outside the villages, perhaps resulting in smaller chaff quantities in the settlements (see Kreuz & Marinova 2017, 647).
In pottery, such macro-remains may also be very rare (Dzhulyunitsa) or omitted simply due to their size. A detailed examination of possible free-threshing chaff parts in pottery is in process, especially concerning the north Bulgarian loess-based geological setting. Regarding the processing stages, the plant parts related to fi ne chaff most probably resulted from secondary winnowing or coarse sieving. Signifi cantly, specifi c interpretations of the chaff inclusions in pottery also depend on the storage practices. If temper consists of a mixture of plant parts that belong to diff erent species, this cannot automatically be associated with maslins (growing mixed crops). The suggestion that einkorn and emmer (and pea) could have sometimes been grown as maslins in some regions of Early Neolithic Bulgaria (Kreuz et was einkorn that was established as the dominant Early Neolithic crop in Bulgaria (Kreuz & Marinova 2017, 645, 647) and preferred as a more winter hardy plant that survives heavy rainfall (unlike emmer, Kreuz et al. 2005, 246-250). Although naked wheat and barley had a higher yield than emmer and einkorn, they needed more nitrogen than the hulled wheat species (Kreuz et al. 2005, 254). At this stage, the previously observed pattern for the Early Neolithic north Bulgarian regions, a preference to hulled barley (Hordeum vulgare var. vulgare) as a drought-resistant crop with lower requirements to growing conditions, followed by einkorn (Triticum monococcum) and emmer (Triticum dicoccum) (Marinova & Krauss 2014), does not correlate with the organic inclusions in Dzhulyunitsa pottery. However, the domination of einkorn and emmer in southern Bulgaria may seem correspondent to the plant inclusions in Ilindentsi pottery. Notably, barley (and naked wheat) rachis remains are rarely found, which bution in the fabrics used for pottery production. Still, instead of adding non-organic sources as functional temper, if and when necessary (sand, crushed rocks, etc.), only chaff was used in pottery production to temper the clay. As regards the ceramic technology, and especially the fi ring, the sandier clays would not necessitate the addition of organic temper, and strictly technologically, perhaps the reason to add organic temper was the shaping of the higher plasticity clay. Similarly, at Dzhulyunitsa the only registered temper was plant-based -mostly cereal infl orescence components. The site illustrates inconsistent use of organic temper, the added plant materials showing quite a variable abundance (0-30%). As fabrics-size-shape-surface fi nish correspondence was missing, the question becomes more complex. Given the properties of the clay body, for all organic inclusions to be categorised as clay temper, we would expect to see them in similar proportions/ suffi cient percentages. The presence of fewer inclusions would have had minimal impact on the forming, the fi ring or the use of the wares, and perhaps refl ects cultural preferences rather than practical requirements. Although there may be a tendency for the more clay-rich fabrics to contain consistent amounts of plant parts, there are still cases where local fabrics were used without any addition of organic temper. At this stage, the plant macroand microremains may be referred to specifi c crops, and especially the light chaff components of crop spikelets (predominantly einkorn and emmer). Neither cut plant parts, nor pure ash or accentuated dung features have been detected so far.
Whereas Măgura-Buduiasca and Dzhulyunitsa share some common features, the fabrics at Ilindentsi-Massovets diff er considerably. The site, located in a valley within the mountains, shows geological continuum refl ected by the microtextures of the fabrics and a range of vegetal inclusions. Although plant parts are usually reported missing in the typical southwest Bulgarian Early Neolithic pottery, some of the studied fragments contain actual vegetal temper. Furthermore, specifi c fragments include organics detectable/recognisable only microscopically. They are often single or few (1% to 5%) -amounts that do not contribute to the technological properties of the clay (i.e. non-functional temper). The range of plant inclusions are thus correspondent to a) actual functional temper; b) 'inconsistent' use of plant parts and c) a sporadic presence of organic inclusions. al. 2005, 243) cannot be assumed from evidence coming from ceramic temper. Storage practices are also a factor regarding seasonality studies. Temporality and lifecycle rituals are to be considered when examining prehistoric organic temper (Kreiter et al. 2014), but chaff present in ceramic vessels does not necessarily indicate that pottery production coincided only with the harvest season. Chaff remaining immediately after the harvest can easily be used, especially for making small clay objects (ibid.). However, chaff temper itself is not a proxy for seasonality in pottery production because of the storage componenti.e. chaff could be stored and available whenever needed. Other inclusions possibly present in the clay paste still may contain some data about seasonality.
Weeds found in pottery could give a hint on both seasonality and the harvesting methods (ear-plucking vs sickles) (Kreuz et al. 2005, 255), provided we take into account eventual biases stemming from the storage practices. Winter crop cultivation was practised, at least for some of the cereals, and generally, the winter annual weed species are the dominant group at the Bulgarian sites. Domination of perennial weeds may suggest that parts of the fi elds were not cultivated intensively (see Jones et al. 1999;Bogaard 2004;Kreuz et al. 2005, 253); however, at this stage, the intensity of fi eld management cannot be estimated based on single Avena imprints registered in pottery, as found at Ilindentsi.

Technological signifi cance: functional temper
In both textural and mineralogical terms, the local pottery fabrics perfectly match the natural sediments, i.e. no raw material processing beyond organic tempering has been registered so far at the three sites. However, specifi c features of the vegetal inclusions, and especially their abundance, show inter-and intrasite diff erences.
The preliminary observations on the studied fragments from Măgura-Buduiasca reveal consistent concentration ranges of the vegetal inclusions (> 10%-15%), indicative of plant temper intentionally added in suffi cient quantity (Fig. 12). All components of the cereal spikelets are registered in the ceramic sherds. The demonstrated consistent use of plant temper corresponds with the location of the settlement in the extensive and alluvial low Danubian plain -a homogenous region with massive loessic accumulations providing for similar raw materials. Within this uniformity, the availability of sandier local areas explains the natural variation of sand distri-

'Organic Temper' and the Early Neolithic Pottery Production: Interpretational Challenges
Downloaded from Brill.com12/25/2021 02:45:54AM via free access landscape and geology, contributing to the available raw materials (respectively the selection of suitable clays), complement the observed variation. Whereas the area of Măgura-Buduiasca in the Danube River region generally delivers the same clay opportunities (correspondent to the similar compositional amount and types of vegetal temper), the site at Dzhulyunitsa had access to both loessic low ground and high ground raw materials. The latter revealed considerable variability in pottery technology at the settlement, expressed in the addition of none, just some or lots of organic temper. The Ilindentsi potters also had a choice of raw materials -ranging from the alluvial areas to the higher mountainous terrain, and these were suitable for pottery production in their natural state.
In terms of technology, the organic temper was added in higher proportions and more frequently to the stickier, perhaps too plastic smectite clays at Măgura-Buduiasca and Dzhulyuntsa, and in lower proportions and more seldom to the drier illitic clays at Ilindentsi. The tempering practice at Ilindentsi diff ers from that at the two other settlements. The proportion of pottery with organic inclusions is considerably lower; and when present, the temper, correspondingly, is found in lower quantities. So far, no semi-charred whole spikelets have been detected, no plant parts in high concentrations or placed in clusters, one on top of another were found there. Generally, the preservation of the phytoliths is worse (compared to Dzhulyunitsa), whereas the northern settlements demonstrate better-preserved plant parts.
Although local geology is among the key technological factors, the decision making when working towards the desired technological properties of the clays also depends on important social and cultural aspects -hence the problematic interpretation of the observed variability in type and quantity of organic temper. The subsistence patterns (agricultural and husbandry practices) may also have a more direct bearing on specifi c pottery technological approaches, especially when dung is considered. Although Ilindentsi is located in the mountainous regions and dated to the developed stage of the Early Neolithic, and the two earlier sites in the Danubian valley occupy areas perfectly suited for agriculture, there is a similar predominance of sheep and goat over the cattle (see Grębska-Kulow et al. 2015). Even though dung of caprinae, at least in its natural state, is usually not very common in pottery production (in contrast to that of cattle, cf. London 1981), the question of its inclusion or absence The available local sandy clays, containing natural inclusions, are suitable for pottery production even without further processing, i.e. they would not necessarily require the addition of temper. The reasoning behind the presence of organic matter in some vessels is thus intriguing, especially given the (a) sporadic wild species and (b) the amounts of crop parts contrasting with the functional temper ranges. Based on studied pottery from one excavation unit (Square A), previous ceramic analyses refer to about 60% of organic tempered vessels at the site (Grębska-Kulova et al. 2011). The present assemblage collected from various contexts shows a considerably lower percentage of fragments containing more than 10% of plant inclusions in the clay paste (12 fragments out of 100). In this study, the percentage of plant parts considered as possible temper is above 10% (usually between 10 and 20%), reaching a maximum of 30%. Since functional temper would exceed at least 10% of the organic inclusions, the very low percentage of vegetal material contained in some of the fragments (1-5%) is probably indicative of non-functional temper but requires further explanation. The high variation in the compositional amount of organics at Ilindentsi (from 1-5 to 30%) poses the question of how much organic addition is required to change the clay properties (if added for technological reasons). At the lowest frequencies, the addition of plant parts, in their natural state, would not act as functional temper and their presence requires further examination.
Even without the addition of adequate amounts of temper, the selected clay raw materials -the silty and the fi ne-grain sandy clays -both meet the technological requirements at each stage of the operational chain; forming, drying, fi ring. Interestingly, a similar amount of organics -as percentages of plant parts in the ceramic fragments -were added both to fabrics perfectly suitable for pottery making in their natural state, and those that required additional materials to achieve the desired quality of the clay paste. It can thus be expected that the reasoning behind the presence of very few organic inclusions, along with the question of how these appeared in the clay paste, is more complex, not limited only to technological aspects.
The location of the sites within the sub-Mediterranean zone (Ilindentsi) and the Danubian plain with more continental climate (Dzhulyunitsa and Măgura-Buduiasca), results in diff erences on all key parameters of crop yield potential (see Kreuz & Marinova 2017, 642). The local higher-resolution site-specifi c contextualised analysis of the vegetal inclusions can be used to re-consider broader defi nitions, such as 'organic temper'. The plant inclusions from the three studied sites demonstrate the variability of an important Early Neolithic ceramic technology component. The potential to reveal various plant sources and crop processing steps archived by specifi c features of the plant inclusions lies in the examination of their preservation, original state when added: wet vs dry, processed vs unprocessed, origin and source, type and amount, correspondence to the mineral inclusions, combinations with wild plants.
This article is a starting point for further detailed research on the complexity of the interpretation of the organics present in pottery fabrics in Early Neolithic Balkan context. Whereas at Măgura-Buduiasca the plant parts are found in quantities correspondent to functional chaff temper, Dzhulyunitsa shows a far greater variety. At Dzhulyunitsa, not all vessels are tempered, and the presence of organic inclusions varies between functional temper and non-functional inclusions (or no temper at all). An even more complicated situation is illustrated at Ilindentsi, where plant matter can be used as functional temper, but at the same time, there are very sporadic inclusions of plant parts, as well as single wild plants. Using the vegetal component in pottery production, the set of commented features that signals this variation, reminds us once again that Neolithisation processes were much more complex and multi-layered than expected.
The three examples reveal a high degree of complexity, which is further evident when interpreting ceramic production features in a broader context. Understanding human technological choices require the study of the interrelation between organic and mineral temper and the specifi c properties of the local clays. This combination allows us to test if the presence of plant parts in the clay fabrics was a strictly technological requirement or also a cultural factor embodied (with)in social behaviour. Based on the study of ubiquitous archaeological material, the examination of the variety in plant use in pottery production thus has the potential to reveal how local subsistence patterns intertwine with various aspects of material culture technologies in the context of site-specifi c surrounding landscapes.
will be investigated in detail in the future. According to ethnographic data (Vasilieva & Salugina 1997), dungderivative temper may sometimes be used in either dry or liquid state. The presence of small and sporadic vegetal inclusions thus needs to be carefully explored, as single plant parts may have passed through the mesh when practising the latter method. Future analysis will focus on this possibility, taking into consideration the specifi cs of the dung of various animals and their eating habits (wild plants and fodder). Such examples of plant-based technological components in pottery production may be revealed by some fragments containing single, sporadic or a very low percentage of organic inclusions, e.g. Ilindentsi. Dzhulyunitsa, on the other hand, contains fragments probably showing the addition of moist plant parts -a possible indication of diff erent technological approach, a hypothesis also to be tested.
Finally, whether the newly registered intra-and intersite variation in the 'organic temper' use is meaningful, i.e. having the same importance to the ancient artisans, is another central question. Do the diff erences observed by the application of high-resolution magnifi cation techniques refl ect intentional variations or these are artifi cially created classifi cations referring to otherwise naturally fl exible components? Understanding the reasons behind the addition of organics as a technological component, and diff erentiating between the major and the secondary features of the added materials requires a spectrum of elements to be considered, rather than choosing single components. Here, the observations are based on a series of features, all pointing to the variability of the use of 'organic temper'. Local pottery-making practices are best interpreted when studied in context -the environmental conditions, subsistence patterns, land use strategies and ceramic traditions. The variation established according to several criteria may thus also refl ect the variability between contrasting practices (e.g. the colder Danubian conditions and Chernozem soils vs the hotter and drier south Bulgarian area).

CONCLUSION
The detected inter-and intrasite variability in using plants for pottery production evolves from the origin, the source of the plant parts, the types of processing and the relation to the clay fabric, all these, in turn, resulting from specifi c human decision making. A multi-layer approach and