Neither “Celtic” nor “Dacian”: the site of Židovar at the edges of La Tène, Carpathian and Roman worlds

The paper reviews long-standing interpretation of the late Iron Age site of Židovar as “Celtic”, “Dacian” and “Celto-Dacian”. Arguing that this standpoint is derived from biased culture-historical ethno-determinism, the evidence is reconsidered from excavation journals of Branko Gavela and published research on Židovar. Evidential basis is discussed, such as sratigraphic difficulties and chronology, as well as some common misconceptions of the site’s characteristics. Deadlocks are emphasized regarding the conclusions on its ethnic belonging. The paper calls for a new approach that goes beyond ethno-cultural determinism and urges the employment of “relational locality”. This perspective considers the site and its immediate surroundings as the first order community, i.e. the spatio-social focal point entangled in diverse, multidirectional and supra-regional relational networks. This would mean that the community of Židovar actively mediated different templates coming from the “globalized” koines of La Tène Pannonian, DanubianCarpathian and Roman worlds, and bricolaged them in distinctive local ways.


Introduction
The purpose of this paper is to reconsider traditional interpretative frameworks that are persistently employed in explanations of the late Iron Age in the areas of southern Pannonia, Carpathian Basin and central Balkans. Particularly, we address the case of one specific archaeological site -Židovar -in order to demonstrate limits of the culture-historical paradigm that has been used to interpret its character, and then propose another possible approach. The reevaluation we offer, as well as the alternative standpoint, is aimed at opening the discussion about how we perceive and interpret the late Iron Age, and whether some deeply embedded preconceptions aid better understanding of the period's dynamics or, to the contrary, results in simplified and problematic judgments.
Let us begin by a brief sketch of the general outlines of the late Iron Age studies in the region of SE Europe, i.e. the states of former Yugoslavia, Romania and Bulgaria. As it was pointed out on many occasions, the traditions of (prehistoric) archaeology in these countries stems from the Central European version of the culture-historical archeology (see e.g. Biehl, Gramsch and Marciniak eds. 2002). In the sub-discipline of protohistoric (i.e. the late Iron Age) studies this means that archaeological remains were linked to the data from ancient historiography, geography and epigraphic texts in a particular manner: ancient written information about "barbarian" populations on the fringes of the Mediterranean world served as the ultimate framework into which archaeological evidence was squeezed, with the intention to perfectly fit historiographical narratives (see students of the Department of Archaeology at the University of Belgrade and were led by Branko Gavela (1914-1994. At that time, Gavela was the doctoral student and teaching assistant to Professor Miloje Vasić , and the first lecturer holding the official position of a teacher in prehistoric archaeology. Since Vasić stubbornly held his eccentric interpretation of Neolithic Vinča as an Archaic Greek colony, and extremely lowered the chronology of the whole Balkan-Pannonian prehistory (see e.g. Palavestra 2020), Gavela's excavations, among other, aimed to show that what was considered as early prehistoric cultural features actually stretched all the way to the late Iron Age, and that Neolithic and Bronze ages in the region were much younger than generally accepted in European archaeological tradition. In such a constellation, and possibly aware of erroneous views of his Professor's, Gavela chose to deal with the late Figure 2. The Židovar plateau with excavated areas indicated with grey (Jevtić, Lazić and Sladić 2006, Drawing 1) Iron Age layer of Židovar. 1 This work became his doctoral thesis defended in 1951 and published as soon as 1952, being the first monograph entirely devoted to a late Iron Age site in Serbia (Gavela 1952). Already here Gavela gave decisive interpretation of the nature of the site and aptly named the book Celtic oppidum Židovar (Keltski opidum Židovar). By identifying the excavated material as typically La Tène (in the first place, gray and painted wheel-thrown pottery) and associating it with the historiographical narrative on the Celts and Scordisci (established in local academic tradition during the first half of XX c.), Gavela saw Židovar as an exemplary Celtic settlement in southern Pannonia. Citing ancient sources that give information about the Celts living together with the Thracians and Illyrians, he comprehended the Scordisci as the Celts intermixed with various autochthonous tribes and, to a certain extent but not decisively, assimilated with the natives. Furthermore, Gavela referred to Strabo's information about the Scordisci as Dacian allies and postulated that the Celtic/Scordiscan presence at Židovar can be explained by this neighborly cooperation. On the other hand, speaking about curious case of the Celts in what was (according to ancient geography) considered as a purely Dacian area, Gavela denied any evidence of the Dacian presence or influence: [Židovar's] archaeological inventory does not show the Dacian provenance whatsoever, nor does it show the Dacian cultural influences. Neither in the younger nor in older stratum of this site there are any signs of closer cultural and economic contacts between the Dacians and the inhabitants of Židovar. Both archaeological material and the type of enclosed settlement at Židovar demonstrate complete cultural independence from the Dacians. (Gavela 1952, 55, translation by the authors) However, soon after the publication of his conclusions, Gavela was faced with criticism, as Milutin Garašanin (1920Garašanin ( -2002 disputed his ethnic interpretation. Contrary to Gavela, Garašanin postulated that there is no sufficient evidence of historical and archeological nature to support the thesis about the Celtic presence in the Banat area. Most importantly, he understood Židovar as a Dacian settlement (dava) and generally regarded the excavated material as having all the features of the "Daco-Getaean culture", especially the coarse ware (Garašanin 1953, 7 fn. 23). Garašanin saw the finds of the Celtic fine pottery as generally characteristic for Daco-Getaean sites, i.e. as not suggestive for Židovar's Celtic provenance. He repeated this opinion in another article, this time using the mixture of pieces of written evidence and conjectures typical of ethno-deterministic conceptualization of the past: As the stronghold of the Scordisci, Židovar would not have justification. In the period of the last layer at Židovar, i.e. I c. BCE, we know that the Scordisci were in the close contact with the Dacians: already in 109 BCE they jointly attacked the Romans. During the reign of Burebista, in the first half of I c. BCE, they were under the Dacian rule. Under such circumstances, it is hardly probable that the Scordisci could build any fortification on their northern border, towards the Dacians. To the contrary, Židovar makes full sense as a Dacian stronghold. At the verge of mountainous area of the Dacians, dominating the Banat plain, Židovar represents one prominent position within the system of Dacian fortresses which are well known in Transylvania and Oltenia. (Garašanin 1957, 91, translation by the authors) Garašanin was well familiarized with the Romanian literature where he had found analogies for the material he referred to. Additionally, some of the most respected and influential Romanian authors also identified Židovar as a Dacian settlement (Daicoviciu 1960, 275;Crişan 1969, 244 fn. 458; see also Rustoiu, Ferencz and Drăgan 2017, fn. 2). Their interpretation followed the same logic as Garašanin: Židovar was in the (historically attributed) Dacian territory, and archaeological material, first and foremost pottery, proves its Dacian character.
Not surprisingly, Gavela did not comply with this critique and held his grounds with some minor amendments. In fact, he subsequently set the resolving of the question of ethnic character of Židovar as the main aim of further excavations (1964)(1965)(1966)(1967) and, interestingly, tried to reinforce the original argument even in his field journals and excavation reports. Therefore, instead of stratigraphic and contextual observations and descriptions of finds that might be expected in this type of accounts, we read the thoughts on why cultural layers at Židovar are Celtic and not Dacian, when the Scordisci could have settled the site, and under what historical circumstances (Gavela 1964(Gavela , 1966. The final interpretation, published in several papers, Gavela (1970Gavela ( , 1976 supposed that the Scordisci stayed at Židovar only several decades as a sort of foreign (Celtic) enclave in what was a homogenous (Dacian) ethnic environment. According to his understanding, the arrival of the Scordisci to Židovar begun with their major defeats inflicted by the Romans at the end of the II and beginning of I c. BCE, ended at the turn of eras, and was only a stage in the course of their retreat. Additionally, he pointed out that mutual cultural influences are not surprising and comes as a completely natural occurrence: "The Dacian cultural material could reach the areas of foreign, even far away ethnic environments, equally as the Celtic culture could enter the Dacian ethnic area" (Gavela 1970, 124, translation by the authors). Finally, Gavela asserted that percentage of the Dacian pottery is notably smaller in comparison to the overwhelming share of the Celtic fine ware N "C " "D " 801 Етноантрополошки проблеми, н. с. год. 16 св. 3 (2021) (although he never actually published any kind of quantification and statistical analysis, but gave subjective approximation). In summary, Gavela changed his previous radical denial of any Dacian involvement at the site, but maintained that Židovar was essentially a Scordiscan settlement with some inflow of the Dacian cultural elements. The other late Iron Age specialists in Yugoslavia basically followed the general principle of canonical understanding of the Scordisci when dealing with Židovar. This meant that the Scordisci were seen as an ethnic group of heterogeneous structure, derived from the Celtic immigrants who merged with the local ethnicities (see further in Mihajlović 2019, 93-116, 277). With such a perspective Židovar has been interpreted as the easternmost Scordiscan settlement, but with very strong presence of the "Dacian element". The latter is asserted thanks to the coarse pottery of the "Dacian" type, found also at other sites further west, in the regions considered to be the core of the Scordiscan territories. Thus, the notion of ceramics that "prove the Dacian inflow" to "the Celtic newcomers' culture" is combined with the ancient accounts about Burebista's expansion and creation of the Dacian kingdom, giving the grounds to propose a significant "Dacian" ethnic component within the population of the "Scordisci", additionally intensified during the rule of the famous Dacian king (Todorović 1974, 41, 138-139;Jovanović B. 1974, 299, 306, 310;Sladić 1986, 31;Jovanović M. 1992). Simply, the debate of Gavela with Garašanin and Romanian colleagues is resolved by referring to one of the favorite interpretational principles of cultural-historical paradigm: ethnic blending of different elements, manifested through "mixed" material culture.
Most recent line of thought of what was going on at Židovar appeared independently in the Romanian and Serbian archaeologies, and to some extent changed the logic of interpretation by employing chronological differentiation of the Celtic, Dacian and Celto-Dacian habitations. This was first proposed by Glodariu (1983, 54) who suggested that the "Dacian material" reflects the taking of previously Scordiscan stronghold by Burebista, and that the Dacians continued to live there afterwards. Similar opinion is expressed by the directors of the latest excavations (1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014) at Židovar who differentiated three separate stratigraphic layers from I c. BCE to I c. CE. According to their interpretation the first settlement belonged to the Scordisci because of the fine gray La Tène ceramics and Celtic character of many object from the Židovar hoard; the second phase of habitation begun around the middle of I c. BCE when Židovar was taken by the Dacians, which is mainly judged by the alleged increase of the "Dacian pottery"; while the last layer is characterized by impoverished material culture of both "Celtic" and "Dacian" nature, as well as sporadic finds of the early Roman products (Jevtić, Lazić and Sladić 2006, 5-6;Lazić 2006, 26-28;Jevtić and Ljuština 2008;Ljuština 2014). In other words, there was an ethnic shift at the site ("Scordiscan to Dacian"), followed by the relatively short-lived "Dacian" domination, and then the ethnic fusion that saw the end of the settlement in the course of the Roman conquest during I c. CE. This perspective has gained support among the Romanian researches, who largely agree with the proposed scenario (with important exception of Egri 2019). However, they stress that Židovar was more closely inclined to the Dacian cultural templates and kingdom, and can be related to sites of the so-called limes Dacicus (Rustoiu, Ferencz and Drăgan 2017;Rustoiu and Ferencz 2018, 131-132), which means that it is understood as more "Dacian" than "Celtic/Scordiscan" from the middle of I c. BCE onwards.
To summarize, the history of interpretation of Židovar has moved from extremely Celtic/Scordiscan (Gavela) or Dacian (Garašanin, Daicoviciu, Crişan), to fused Scordiscan or Celto-Dacian (Todorović, Jovanović B., Sladić, Jovanović M.), to ethnically differentiated by chronology (Glodariu, Jevtić-Lazić-Sladić, Ljuština, Rustoiu-Ferencz). Nonetheless, all of these perspectives are embedded in the ethno-deterministic discourse, with the essentialist idea that "determining" ethnicity answers all the other unknowns about some site or period (especially if there are written accounts that "explain away" particular or general dynamics). Regardless of their cursory divergences, they imply unquestionable existence of large and en bloc group identities that allegedly correspond to archaeological classification and geographical distribution of certain kinds of materiality. This kind of interpretations revolves around generalized (and fundamentally arbitrary) assessments (e.g. "typical Dacian cultural elements" or "common Celtic material") and stem from the comprehension of culture as a normative manifestation of ethnicity. Such perspectives simplify the realities of the late Iron Age and marginalize all the other social aspects that communities of this period certainly lived. This is why another approach should be tried out, one that does not a priori give an absolute advantage to some general academic construct such as "Celts", "Dacians", "Thracians", etc., but instead minds the local context and relationalities. In short, rather than explaining away particular phenomenon with (ancient and academic) general categories of contested meanings, it is more productive to consider specificities of distinct local and micro-regional cases. We will tackle alternative perspective further in the text, but in order to do so it is first necessary to get some facts straight.

Stratigraphic difficulties and uncertainties
As already mentioned, Židovar was one of the first prehistoric sites to be researched in post-WWII Serbia. This brings along some important methodological baggage, as the excavation methods of this period were not yet as stand-ardized and developed as they became several decades later. In other words, Gavela's excavations and documentation procedures have some limitations that are intrinsic for the formative phase of proper archaeological methodology in Serbia. This meant that the selection and collection of finds were in accord with the purpose of archaeology as it is seen in the 50's-70's in the Yugoslavian context: to ascertain relative chronology (as 14C dating was still largely unavailable and accepted with caution) and culturally determine sites and areas, which in the case of protohistoric studies actually meant attribution of ethnic labels known from ancient literary sources. This is why Gavela's (unpublished) field journals (Židovar Excavation Journals 1964, 1965, 1966, 1967, 1971, 1977; ŽEJ further in text) and published reports (Gavela 1964(Gavela , 1965(Gavela , 1966 largely differ from what we are used to read in that sort of archaeological narratives. In his field writings Gavela did not offer any detailed descriptions of stratigraphy or particular contexts, he was not systematically defining separate stratigraphic units, features, pits, cuts and fills, nor did he record precise positionings (by X, Y and Z axes) of finds and archaeological situations (with several exceptions). While he did write about houses, floors, ovens, parts of earth-walls of dwellings, fireplaces, zones of charcoal and characteristics of the soil, this was done in a very general and loose manner, without mentioning defined shapes, sizes, or locations inside the trenches. Likewise, while he used photography, sketches and drawings as the means of recording, and treated some contexts with more attention, the main focus was on selective collecting of the "diagnostic" material. This included ceramics and metal objects that could serve for relative dating and definition of their wider cultural belonging (through typo-stylistic features), as well as exceptionally nice and preserved pieces suitable for museum display. Also, it is more than probable that Gavela was not keeping all of the found potsherds, but saved only the most illustrative and characteristic parts of vessels. The gathering of artefacts was done by a system of "excavation layers", i.e. the digs of soil of more or less regular thickness of circa 0.20 cm (i.e. one shovel deep) by which the relative depth of finds was determined. Thus, the ceramic material was collected en masse and packed separately according to excavation layers from each of the opened trenches. Although this enables us to comprehend general cultural layers of Židovar even today, the use of such a system of recording has irreversibly deprived us of studying distinct stratigraphic units and contexts. Furthermore, the amassed pottery was never systematically analyzed or statistically processed, 2 so we lack these pieces of information as well.
In conclusion, the excavation methods and manner of documentation used by Gavela, which were directly derived from the theoretical framework of cultural-historical archaeology, prevents us to utilize the research results and data for anything more than general stratigraphic picture.
However, Gavela is not to be judged harshly for documentation that is not up to our standards, as this way of archaeological recording was not uncommon prior the shift towards more careful and rigorous evidence-collecting brought by the processual and similar tendencies. Furthermore, and crucially, Židovar's stratigraphy is per se a difficult one, because of the poor state of the layers' preservation, confirmed by the most recent excavations (1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014). The late Iron Age stratigraphic section (dated to I c. BCE-I c. CE) is circa 1.1 m thick and it consists of three layers differentiated by the quality of their contents (soil and archaeological material -see Lazić 2006;Jevtić and Ljuština 2008;Ljuština 2014). Already Gavela has pointed out that both late and early Iron Age layers have very scarce architectural remains, contrary to the abundance of ceramic material, and that they imply an improvised settlement (ŽEJ 1966, 30 th August, 14 th October). He has also remarked that the uppermost layer (layer nr. III according to Jevtić, Lazić and Sladić 2006) suffered severely by the process of erosion caused by the rain and the proverbially strong Banat winds (ŽEJ 1964, 13 th October). In other words, the most recent phase of habitation, dated to I c. CE by the Roman period objects (Jovanović 1997), did not reveal any remains that could be clearly defined as floors, walls, or postholes of dwellings and accompanying architectonic features. True, the detailed reports of the latest excavations are not published in entirety, but none of the existent publications mention clear-cut stratigraphic units that could aid us to grasp the layout and dynamics of the youngest settlement at Židovar. Instead, this sub-layer is composed of the debris mixed with soil and archaeological material (with few remains of ovens in situ) that does not offer intelligible picture of its character (Jovanović 1997;Jevtić and Sladić 1999, 96;Lazić 2006, 23).
Situation is similar with the oldest layer, i.e. the remains of the first late Iron Age settlement, whose construction is dated to the end of II or beginning of I c. BCE. According to the information from the most recent research, the oldest habitation phase could be discerned only by the residues of earthen house walls, zones of ash and charcoal, late La Tène type potsherds and few metal objects. Judging by this situation and some areas of the completely destroyed layer, it has been presumed that the debris of the oldest settlement were severely disturbed and devastated by the construction and usage of the settlement from the subsequent layer (nr. II according to Jevtić, Lazić and Sladić 2006). This scenario is allegedly supported by the stratigraphic position of the Židovar treasure, supposedly hidden at the end of the first phase of habitation: the hoard is found below the oldest late Iron Age sub-layer, while some of its parts were scattered within it. Thus, it has been suggested that the dispersion took place during the N "C " "D " 805 leveling of terrain that occurred as a preparation for the construction of buildings of the second late Iron Age layer, i.e. the settlement's second phase (Lazić 2006, 20-21). These stratigraphic data are combined with historical account on Burebista's conquest and the conclusion is reached that the hoard is hidden by the Scordisci who, threatened by the Dacian king, left Židovar in a hurry (Lazić 2006, 26-28). The second late Iron Age layer contains the best-preserved remains. Both Gavela's (ŽEJ 1965, 2 nd and 6 th October; 1977) and recent excavations uncovered the house floors of specific construction, even with household items still on their surfaces (Sladić 1991, 139;Lazić 2006, 21-23). It is supposed that these buildings were erected on the surface leveled with loess and bright-colored clay, the materials also used for the construction of houses. Because of the leveling and unawareness of residents of the second phase about the hoard in the soil below their dwellings, it is proposed that this settlement was constructed as completely new, after the radical change that involved the ethnic shift from the Scordisci to the Dacians (Lazić 2006, 26-28). The directors of the latest research stress that houses of the second sub-layer were very large buildings of rectangular plan and dimension up to 10-12 m in length and 6-8 m in width (Sladić 1997, 23;Lazić 2006, 21, 23), which makes them quite a remarkable phenomenon, unobserved so far in any of the late Iron Age sites in the region. Nonetheless, although the architectural remains of this sub-layer are of much better quality than the previous and subsequent ones, the situation is far from clear. The house floors and walls of wattle and daub are not preserved in their entirety, but have survived as the intermittently found remains that were then used to suggest possible plans and sizes of the houses. In the words of one of the researchers: "Floors and walls of these structures are not completely preserved so their actual size could be assumed only on the basis of large quantities of loess and clay of light-yellow color used as building material and many separate piles of burnt wall daub. In the corners and along some of the edges of these structures were encountered evenly distributed circular holes for timbers used as supports for the roof and walls made of wattle and daub" (Lazić 2006, 21; also repeated in Jevtić and Ljuština 2008, 29). It is then unsurprising that there are literally no ground plans and perimeter outlines of these houses in any of the published works dealing with the second layer of the late Iron Age Židovar. 3 In other words, the second late Iron Age habitation phase is indeed more comprehensible than the other two, but what was presented as relatively clear evidence of the settlement's nature (i.e. the type of houses) is actually a very loose and idealized reconstruction (that is of course possible, but for now far from certain). Additionally, there are uncertainties regarding the renovations of the houses from this phase as Jevtić  (2006,88) mentions "later building phase of second stratum", i.e. "the level of restoration of larger rectangular house". Finally, the circumstances of cessation of this phase are equally difficult to discern. It is noted that some of the houses had burned in a fierce fire and none of them were ever again repaired, which is taken as the evidence of complete abandonment of the site soon after Burebista's death (Lazić 2006, 27). However, it has never been reported how many houses were found in the first place (it is possible to count only four or five of them in the published works and Gavela's excavation journals), how many of them were actually destroyed, and if the traces of fire (and/or some other supporting evidence) are of such nature to convincingly suggest violent destruction of the second phase of habitation. Therefore, even though the second late Iron Age layer contains far richer archaeological situations and clearer stratigraphic units, it is almost as difficult and confusing as the first and the third ones.

Chronological ambiguities and interpretative obstacles
Complicated stratigraphic situation poses serious challenges, especially bearing in mind that all the late Iron Age phases of Židovar are defined and explained by referring to the events described in ancient narratives. Researchers have been trying to mitigate the inconclusiveness of archaeological evidence by its stretching and linking to allegedly pivotal historical events, but this created more problems than it solved. The chronology is a case in point. The first phase of the late Iron Age Židovar is dated to the end of II or beginning of I c. BCE, but the majority of authors cite the defeat of the Scordisci by Scipio Asiaticus in 85 BCE as a blow that definitely set them on the course of retreat and refuge across the Danube (Gavela 1976, 34;Sladić 1997, 55;Lazić 2006, 27;Jevtić 2006, 167;Ljuština 2014, 220). Nevertheless, this interpretation is problematic for several reasons, not least because Asiaticus most probably led a very limited campaign, both in terms of time and geographic span, and it is highly unlikely that his operations had such drastic impact on populations of the central Balkans (see Mihajlović 2019, 198-199). This means that the explanation of the causes for the establishment of the "Scordiscan" settlement at Židovar, as well as chronology based on it, are dubious and do not help us with better comprehension of the site. Whatever the reasons for the formation of first habitation phase were, it is more reasonable to relate them to the tendency of enclosing the late Iron Age sites at the end of II c. BCE, generally taking place in the region, and most probably not directly generated by the Roman military activities farther south (Mihajlović 2019, 210-219, 299-302;. Consequently, the interpretation of the first late Iron Age settlement of Židovar will have to take alternative direction, not only in terms of explanatory methodology, but also concerning the methods of dating.
The current understanding of the beginning of the second phase and its character is equally dubitable. As it is mentioned earlier, the second phase is interpreted in the light of the Dacian conquest and colonization, which caused the Scordiscan deserting of Židovar. It is also supposed that this habitation layer arose as completely new, and terminated violently soon after the dissolution of the Dacian hegemony. Whereas there is no doubt that such a powerful political and military force impacted the communities in vast areas (directly or otherwise), the definition of archeological layers and phases by straightforwardly referring to Burebista's reign as a pivotal moment is an extremely problematic standpoint. First, it has to be considered that his domination lasted for a limited period of 30 years at best (but more plausibly circa 60-44 BCE, see Rustoiu and Ferencz 2018), which would mean that the entire second late Iron Age layer at Židovar (with the most abundant archaeological traces) had a very limited lifespan of only several decades. Second, even if Židovar and surrounding areas indeed were parts of the king's political structure, there is no ground to automatically claim violent and turbulent power turns, either in the course of the alleged "Dacian conquest and Scordiscan retreat", or in the post-Burebista time. Finally, the era of Burebista's domination did not necessarily have to leave archaeologically recognizable traces of the kind we would expect, especially given the general state and quality of layers at Židovar. To summarize, the initiation of second construction phase at Židovar could have been somewhat older or younger than the period of the presumed "Dacian control"; it is hard to expect that Burebista's hegemony would be archaeologically evident if e.g. the local community submitted without armed conflicts; and, similarly, it is highly uncertain if the events after the breakdown of Burebista's supremacy would leave any consequence of archaeologically visible kind. Simply, the beginning of the second habitation phase could have been either somewhat older or later than Burebista's direct involvement in the area of Židovar, just as the ending of this settlement could have happened years or decades after the king's era and for the reasons unrelated to the dynamics in the realms of the kingdom's successors. In conclusion, the forcing of archaeological evidence to fit historical dates and events could contribute to extremely misguided conclusions, not least because of the elusive stratigraphy and difficulties in defining more nuanced chronology and archaeological units/features. In this light, it is indicative that Rustoiu, Ferencz and Drăgan (2017, 203) rightly date some of the material found in the second habitation layer (such as brooches of spoon-bow, Jezerine and Alesia types - Fig. 4/1-4) to the Augustan period, and attribute its general chronology to the second half of I c. BCE and the beginning of I c. CE. The information given by Jevtić (2006, 88) is also instructive in this regard: on the floor of the later building phase of one of the houses from the second habitation layer there was a bronze spoon-bow brooch ( Fig. 4/2), which is generally dated from the last quarter of I c. BCE to the mid I c. CE (16c type according to Rustoiu 1997, 48-50, 129-130 have lasted considerably longer than Burebista's epoch and terminated closer to the turn of eras. The interpretation of the third and most poorly preserved habitation layer is burdened with the same kind of difficulties. As we have seen, its onset is very vaguely dated sometime after Burebista and the alleged fierce ending of the second phase, but it was never explicated by the researchers if some hiatus is hypothesized and if so, how long it took (Lazić 2006;Jevtić 2006). On the other hand, Rustoiu, Ferencz and Drăgan (2017, 203) place the third phase entirely in I c. CE, citing the chronologically sensitive types of brooches as an argument. As for the cessation of the last habitation layer, it is presumed that it had something to do with the Roman involvement in the region, i.e. conflicts with Decebalus, either during Domitian (Jevtić 2006, 88;Rustoiu, Ferencz and Drăgan 2017, 204) or Trajan (Jovanović 1997). There is no doubt that this phase at Židovar saw some contacts with the Roman Empire that had spread to the Danube and rearranged previous power relations and social structures in the whole region. As for chronology, however, the end of this phase is not at all unambiguous as the associations with Domitian's or Trajan's wars would instruct us to believe.
Namely, of chronologically indicative Roman objects, so far none conclusively belong to the end of I or beginning of II c. CE. The only two Roman imperial coins found at the site is denarius issued for Octavian in 42 BCE (Bakić 2005, 14;Fig. 3/5), and Augustan as minted in 7 BCE (Bakić 2005, 15) with the countermark AVC ( Fig. 3/6), struck in 9-14 CE (countermark type 54.39 according to Werz 2009, 168-169). 4 Similarly, the only datable piece of Roman pottery is found underneath the floor of the house found in NW part of the site, belonging to the second habitation phase (ŽEJ 1977, 10-15 th October;Sladić 1991). However, the fragment most probably belonged to the fill of a pit from the last settlement phase, since many were found in the trench and reportedly seriously devastated the floor of the mentioned house (Sladić 1991, 139). The shard ( Fig. 5/2) is a piece of foot and bottom of a terra sigillata bowl (or plate) and bears the stamp SYM[PH]/ORV[S] (Sladić 1991, 148, fig. 9/7). This workshop was most probably located in the Po valley and its production is dated to 1-20 CE (Mongardi 2014, Tab. 2 nr. 265). 5 As for the unearthed brooches, five 4 Another bronze as is found in 2002 at the adjacent site of Tobolocia (presumed to be the "suburban" area of Židovar). It is not yet conserved and thus it is unidentifiable, but the large SC at the centre of one of its sides is clearly discernible and indicates an issue from Julio-Claudian epoch. 5  There was a Gaulish maker of the same name and of much later period (125-150), but his stamps run as Symphori M (Hartley and Dickinson 2011, 386) and thus do not correspond to our example.
of them are suggestive: the example of Alesia type brooch ( Fig. 4/4) is dated to the period of Caesar and Augustus (Feugère 1985, 306;Istenič 2005); the unique specimen of a hybrid brooch ( Fig. 4/5) which combines elements of the Gorica and Almgren 18 types 6 can also be dated to the second half of I c. BCE or the beginning of I c. CE; the fibula ( Fig. 4/8) of Langton-Down type (Feugère type 14b1b) can be placed from the Augustan to Claudian times (Plantos 2019 with older bibliography); and two strongly profiled brooches with an open catch-plate ( Fig. 4/6, 7) are widely dated from the Augustan to early Flavian period or even later (Riha 1979, 73-74;Rustoiu 1997, 130-131). Finally, at the Archaeological Collection of Faculty of Philosophy in Belgrade, where the material and documentation of Gavela's and 1996 excavations are kept, there is a photo of two pieces of terra sigillata from 1996 excavations (Fig. 5/3). While we could not locate the actual shards, judging by the photo, these could be identified as two pieces of a bowl produced in Rheinzabern in the style of B. F. Attoni and dated in the late Antonine and Severan period (Brukner 1981, 21-22, T. 32;Rusu-Bolindet 2005, 203-204;Gabler 2016, 123;Hajdu 2018, 452-453). 7 These pieces correspond well with Jovanović's (1997, 73, 77) assertion that according to archaeological evidence settlement continued to exist after Trajan's wars, albeit he does not explicate to what objects he refers. A later date is also plausible for the fragment of circular bronze-plate mirror (Jevtić 2006, 158, fig. 90) of a kind which entered the wider usage in the province of Moesia in II c. CE. Although the mentioned chronologically sensitive items are not enough to reach a conclusion regarding the end of the third settlement phase at Židovar, two observations can be made. First, judging by the dated finds, and even taking into account that some of the objects could have been in use longer prior to deposition, it is indicative that there is (so far) no conspicuous evidence for the duration of the settlement up to the Domitian's war (86-88) or Trajan's conquest of Dacia (106). Of course, this cannot be excluded, since the material (especially from the recent excavations) is not yet fully analyzed and published, but it seems that the termination of habitation at Židovar is currently defined exclusively by giving the priority to general historical events in the region. Second, the identified pottery fragments from the late II or the beginning of III c. CE open the possibility of either continual or temporary habitation at Židovar in this period (the latter option being more probable as the former would leave more tangible remains). This in turn implies potentially on-andoff usage of the Židovar plateau: it was perhaps settled continually in the first half of I c. CE (according to more precisely dated artifacts), abandoned in the course of the second half of the same century, and then settled again (once or several times) in a downsized extent during II c. Whatever the case, it remains dubious how accurate is the current understanding of the last settlement phase of the site, i.e. if the last habitation layer and all of its features could be treated as an outcome of one continual stage of occupation, or it was indeed more perplexing.
Stratigraphic and chronological unknowns pose serious obstacles for understanding the character of the site as well. To begin with, it is not certain if the plateau accommodated settlement of a kind usual in the areas of Syrmia and Bačka (see Mihajlović 2020), or it was inhabited by some sort of privileged portion of the society, according to the model presumed for the Iron Gates and "Dacian" settlements (Egri 2014, 177;2019, 126-128). The latter has its merits as the Židovar plateau was of small size (0.5 ha), possibly enclosed with rampart and palisade (see further in text), perhaps had large houses (if the current reconstruction turns out to be accurate), yielded fragments of imported Roman bronze vessels, and sheltered the hoard of precious objects. Furthermore, the site of Tobolica in immediate vicinity (below the plateau), two and a half times larger than the Židovar plateau, was used in the late Iron Age, possibly as a fortified area with the role of a "suburban zone" (Ljuština 2014, 223). In light of this evidence, Židovar could be considered as the secluded residence of a group of people with greater social and economic capacities (the martial elite, as it is usually postulated -Jevtić 2006, 167-168). On the other hand, although there are finds of horse gear (spurs and pieces of bids -Sladić 1997, 63), arrowheads, spearheads and fragments of swords, there is also direct evidence of crop production, processing and preparation of food. This is indicated by the finds of grinding stones and rotary quern (Gavela 1952, 35;ŽEJ 1966, 18 th June;1977, 10 th October; Ljuština 2011), cereal seeds (Medović 2003, 182-183), possible storage area within the remains of one of the houses (Lazić 2006, 23) 8 and bread ovens (Sladić 1991, 139-144). Furthermore, the presence of coarse cooking and storage ware on the house-floors (Lazić 2006, 21-23), combined with the absence of any finds of obviously exclusive ceramic tableware (the common painted pottery has been found), point to the "usual" household activities at the site. Likewise, the presence of osteological remains of domestic and wild species (with one and two thirds share respectively) demonstrates husbandry and, N "C " "D " 811 Етноантрополошки проблеми, н. с. год. 16 св. 3 (2021) especially, hunting as important subsistence activities (Radišić and Ljuština, in print). 9 Hence, even if some sort of a group with greater socio-economic capacities indeed resided at Židovar, the available evidence does not suggest the lifestyle divergent from the patterns commonly found in other late Iron Age settlements in the middle Danube region. The specificity of Židovar that can be taken into consideration is its seemingly advantageous position allowing control and manipulation of passage, and the role of a mediating node within supra-regional relational networks of various kinds. In other words, the site should be examined with regard to its local spatial and social setting instead of lumping it in ethno-cultural generalizations such as "Celtic" or "Dacian" (see the last section). Some widespread misconceptions and confusions of evidence Unsurprisingly, all of the above problems (from issues of documentation, to complicated and hard to interpret stratigraphy, to uncritical correlation of archaeological evidence and historical accounts) contribute to some widespread and persistent confusion concerning evidence, which additionally clouds the un- Figure 5. Some of the indicative ceramic material found at Židovar: (1) "Dacian" cup/mug from the first late Iron Age layer at Židovar; (2) terra sigillata shard with stamp Symphorus; (3) Rheinzabern terra sigillata shards; (4) fragment of a locally produced pitcher following Roman provincial shape. derstandings of the late Iron Age Židovar. One of the common misconceptions is the alleged destruction of different phases by a strong fire. This conclusion is reached already by Gavela who, due to the presence of "a large amounts of charcoal, soot, ash and carbonized bones", as well as demolished houses and burnt potsherds, postulated that "the settlements was set on fire purposely every time it had to be abandoned" (1952,14,42). He repeated this later as well, claiming that the burnt and destroyed dwellings were the deed of the Celts themselves, or of the unknown enemy of theirs (Gavela 1976, 33, 35). However, it is interesting that mentions of purposeful fire and settlement burning appeared only in Gavela's published works, whereas such information was not explicated in his excavation journals. These writings indeed contain information on the burnt materials and ashes, but within them it was never reported that such residues characterized wide zones or whole layers, and it seems that they refer to the localized cases of ovens, hearths and dumping pits of ash, or building activities such as fired floors, walls and beams (ŽEJ 1965, 2 nd October;1966, 16 th , 20 th June, 31 st August; 1977, 15 th September). 10 This is even more telling in light of Gavela's accounts of the scorched stratums from the older (early Iron and Bronze Age) layers at the site (ŽEJ 1965, 11 th , 13 th October; 1966, 24 th June, 3 rd September; 1977, 22 nd , 27-28 th September). Consequently, it can be deduced that Gavela gives ambiguous and confusing notes about destructions by fire, which could be deemed inconclusive at best, or useless at worst, when it comes to the scenario of violent ending of the Židovar settlements. Other researchers have also been mentioning traces of burning, but unfortunately this has not removed the ambiguities. Sladić, who partook in Gavela's excavations, explicated that the remains of a house from the second layer of habitation found in 1977 was burnt, without any hints about large-scale fire (1991, 139). As we have already stressed, Lazić (2006, 23, 28) stated that some of the houses from the second phase were burnt, but this assertion is too vague for any relevant judgment. As for the last phase of Židovar settlement (I c. CE), authors of the most recent excavations decidedly dismiss its destruction by fire and imply that the houses were abandoned peacefully (Jevtić and Sladić 1999, 96; see also Jovanović 1997, 77-78).
It is then not curious to read further interpretative misusage of these accounts, although they most probably stem from the initial evidence confusion. For example, Ljuština (2014, 220) implies that houses of the earliest horizon (the phase dated to the first half of I c. BCE) were destroyed by fire. Albeit she might have had the insight into the documentation of the most recent research, it is strange that no earlier works of the excavators ever mentioned the destruction by burning for the first settlement at Židovar. Similarly, the violent destruction of the first phase is cited by the Romanian colleagues, but with more amplified 10 It is noteworthy that the cited pieces of information about traces of fire in the Late Iron Age layers are actually all there is in Gavela's excavation journals. N "C " "D " 815 Етноантрополошки проблеми, н. с. год. 16 св. 3 (2021) statement that it "ended in a powerful fire" (Rustoiu, Ferencz and Drăgan 2017, 199;Rustoiu and Ferencz 2018, 129 fn. 32). Although they have quoted earlier publications for this view, in none of the referred works such scenario is ever mentioned. The misconceptions about the demolitions of Židovar continued up to date, so Egri (2019, 80, 84, 128) accepts that both first and third phases of settle ment ended with destructions, although the last habitation layer was never before interpreted as being destroyed by burning or otherwise. The confusion is similar when it comes to fortification of Židovar. It was Gavela who already in his book (1952,15) postulated the existence of a palisade, thanks to the discovery of post-holes in the southernmost section of the plateau. He encountered 14 holes of 0.1 m in diameter and in mutual distance of circa 0.18 m, allegedly in two rows that formed zigzag outline. In 1977 field campaign the structure made of beaten earth was discovered in the NW part of Židovar plateau, and it was presumed that this is a part of earthen rampart (ŽEJ 1977, 23-28 September). Additionally, Gavela reported the discovery of drywall stone structures he found in trenches, both at the periphery and in the central area of the plateau. These structures were "unearthed next to the outer parts of dwellings, especially at their corner sides", and "according to the places where they are preserved and visible, it can be concluded that these drywalls were erected for the protection and greater stability of houses exposed to north and east windblasts" (Gavela 1952, 15, translation by the authors). In other words, Gavela discovered stone structures that had some function in housing (most probably of the second habitation phase), registered in later excavations as well, together with parts of grindstones (ŽEJ 1977, 15 th September).
Ever since Gavela's first reports, it has been widely accepted that Židovar had some sort of earthen rampart enclosing with palisade, but further research did not provide conclusive evidence. For the presumed northern portion of the earthen wall, Jevtić (1997, 45-46) postulated the early Iron Age provenance, according to the 1977 excavation accounts and stratigraphic position of the unearthed earthwork. Additionally, in her reviews of archaeological results and recent attempts to find the remains of fortification at the plateau, Ljuština (2014, 222, 224) rightly pointed out that there is no conclusive evidence of the site's earthen fortification. Having in mind the destructive effects of erosion caused by rains and strong winds, the lack of reliable traces of enclosure is not an unexpected situation. However, Gavela's reports on drywall structures were taken at a face-value and misinterpreted as the proof of a stonewall stronghold of the Dacian type. This was done already by Crişan (according to Rustoiu 2005, 68 fn. 245) and Glodariu (1983, 54), and it persists in the most recent summaries (Rustoiu 2005 from overenthusiastic efforts to fit various particular phenomena inside the general interpretational framework of chronology and regional dynamics. In any case, there is no substantial data of any kind to support the supposition about Židovar's stone drywall. Lastly, there is the issue of the previously discussed alleged large houses of the second settlement phase. Even though their layout, size and architectural details are not certain due to heavy devastations of the second habitation layer, the reports of Jevtić, Sladić and Lazić are not only acknowledged, but also misread or even overstated in some reinterpretations. Namely, these large houses are believed to have had apsidal ending to the NW side and thus resembled the "Dacian" houses or sanctuaries (Rustoiu 2005, 68-69;Rustoiu, Ferencz and Drăgan 2017, 199;Rustoiu and Ferencz 2018, 129 fn. 32). This is however never asserted in any of the field journals, reports and publications dealing with Židovar, and could be regarded as the case similar to the alleged drywall fortification: the daring attempt to shed some more light on the unknown, but without solid evidential basis.
In conclusion, whereas it is most welcomed to hypothesize and widen the span of interpretations that could enrich our perspectives, it is crucial to be careful not to stretch the evidence too far. This is especially the case when the data and available documentation are difficult in character and of dubious quality. In such instances, their cursory usage can lead to various overestimations and it will certainly not help to form interpretations of greater quality but, instead, contributes to the creation of biased and wrong understandings. "Celtic", "Dacian", "Celto-Dacian" -based on what?
Having in mind all of the deadlocks and shortcomings of the evidence from Židovar, both in terms of preservation and documentation, the question must be addressed of the actual basis of the interpretation of site's (ethnic) character. As hinted in the opening section, the current comprehensions of Židovar are derived from ethno-determinism that is an innate feature of culture-historical archaeology. More precisely, the interpretations of the site rest on theoretical thinking that could be considered problematic and anachronistic. Contrary to the widespread beliefs of practitioners of this traditional approach about its atheoretical nature, it is a viewpoint charged with theoretical preconceptions that are deeply biased and preset interpretative outcomes. In the case of Židovar this meant to go to great lengths in order to demonstrate and decidedly define the site's ethnic attribution as the professedly most important research problem one could ever tackle. In what follows we review on what grounds the "Celtic" or "Dacian" arguments lie, and pay special attention to the problem of the supposedly distinct "Dacian" ceramic material. The "Celtic" interpretation of Židovar springs from a long-lasting conviction that the La Tène material culture is the direct proof of the Celtic migrations (described by ancient authors), and that its distribution shows the territories reached and dominated by the Celtic tribes, especially from c. 300 BCE onwards. However, the so called "Celtic debate" contributed to the realization of much more complex scenarios than the epic-historical comprehension of the ethnically and culturally unified Celts. These involve regular population mobility (spurred by the Hellenistic world's need for armed manpower and hence recruitment of continental mercenaries); technology and style transfers through the global La Tène networks that pervaded the late Iron Age Europe; and internalization and merging of the new techno-stylistic trends and practices with local ones (summarized with bibliography in Mihajlović 2019, 181-186, 278-289). Over the last couple of decades, the general caution has been formed regarding equitation of the La Tène material culture and Celtic ethno-cultural belonging, as the term is too general and vague to signify any real group identification in the past, or meaningful analytical category in modern research. Therefore, the term "Celts/ Celtic" is still in use, but more and more as the nominal shorthand for perplexing and compound world of the continental Europe in last three centuries BCE.
The same is valid for the "Scordisci", which are seen as the southern Pannonian and central Balkan Celts, that fused with local tribal groups and created a mixture of the autochthonous and La Tène cultural elements (summarized in Tasić 1992). This interpretation has been challenged on several grounds since the "Scordisci" were ancient literary construction from the second half of II c. BCE, that covers much more than it reveals the realities of the late Iron Age world in the area (Mihajlović 2019). Although the term still dominates the research, the increasing number of scholars raises caution towards the traditional ethno-deterministic axiom of "Celts + natives (Illyrians, Pannonians, Thracians) = Scordisci" and avoid its simplifying connotation (e.g. Radišić 2017; Egri 2019; Drnić 2020). In other words, the material culture of La Tène or "latènized" style in the region cannot be utilized as an indication of "Celticity", nor can the "Scordisci" be treated as the stable ethnic unit with intrinsic cultural features and clearly defined territory. Needless to say, all of this applies in the case of Židovar as well: its "Celtic" and "Scordiscan" ethnic character cannot be asserted by the presence of La Tène material, as this interpretative approach is too flawed and built on scientifically questionable postulates. Similarly, the utilization of ancient narratives as a direct frame of reference for Židovar (such as the topos of its establishment by the Scordisci who retreated from Scipio Asiaticus) is not only highly unreliable, but misunderstands the logic of the Greek and Roman geographic and ethnographic worldviews and writings derived from them.
The comprehension of the "Dacians", usually imagined as a singular ethno-cultural collective, also underwent revisions in recent years. Based on gen- Issues in Ethnology and Anthropology, n. s. Vol. 16 Is. 3 (2021) eral doubts in previous historiographical and archaeological perspectives on the European protohistory, the opinions have been formed about the "Dacians" as a composite category, much more complex than surviving ancient literature and XIX-XX c. scholarship pictured it (Lockyear 2004;Egri 2019, 28-31, 119-121). For example, by utilizing statistical analysis of funerary data from Romania, northern Bulgaria and northern Serbia, Popa (2018) demonstrated that it is possible to define at least two supra-groups, containing seven different (smaller) groups of material similarities, which suggests regional and supra-regional sharing of some practices and objects. What this study indubitably proves is that archaeological evidence could be defined as separate clusters in various ways according to different starting positions (e.g. as ethno-cultural areas or as regional and supra-regional groups of similar material traits), which in turn can be variously interpreted in terms of what kind of identity they represented. Consequently, to claim the existence of strictly defined Dacian ethno-cultural characteristics is to build on anachronistic theoretical outlook that artificially ethnicizes the past and makes it resemble modern nationally defined identities and dynamics.
One of the key means in proving the presence and spread of the "Dacians" is by referring to finds of the so-called "Dacian pottery". This category includes coarse ceramic material of pot and cup/mug shapes and more delicately made pedestal "fruit-bowl" (fructiera). As we have mentioned, Židovar was declared "Dacian" exactly thanks to these types of vessels, and the allegedly increased share of this kind of pottery (in comparison to the "Celtic" grey fine ceramics) gave grounds to assert the "Dacian" capture and colonization of the site. This was first postulated by Milutin Garašanin who, together with Draga Garašanin, started to further promote the thesis that such pottery is an indubitable sign of the "Dacian" ethnic presence (M. Garašanin 1953;1957;D. Garašanin 1964). The Garašanins formed their opinion under the influence of the contemporary Romanian literature, where the principle was firmly established by the authors who were defining the "Dacian ethno-cultural area" (e.g. Crişan 1969). More interestingly, thanks to their authority within Serbian/Yugoslavian archaeology, the formula was afterwards accepted and the whole series of works about time and nature of the "Dacian" presence emerged over the years and even up-todate, all citing the finds of crudely designed cups/mugs and pots as the crucial indication (Todorović 1974, 41, 138-139;Jovanović B. 1974, 299, 306, 310;Sladić 1986, 31;B. Jovanović and M. Jovanović 1988, 95-96, 195;Popović 1992, 38;2000, 85, 87-88;Jevtić and Ljuština 2008). According to such viewpoint, wherever the "Dacian cups/mugs" are found, from Syrmia in the west to the Morava valley and the site of Krševica to the south, there must have been some Dacian involvement through migration or otherwise. The claim is not limited to the late Iron Age, since the "Dacian cups/mugs" are also found N "C " "D " 819 Етноантрополошки проблеми, н. с. год. 16 св. 3 (2021) at the sites of the early Roman period, giving grounds to understand them as the proof of ancient mentions of resettling of the Transdanubian (i.e. Daco-Gethaean) population from their original territory to the right banks of the Danube, somewhere in the province of Moesia (on these accounts see Boatwright 2015). In general, this perspective is led by a simple rule: the mentioned type of pottery is the evidence of the Dacian ethnicity -full stop. According to such a viewpoint, no subtleties and nuances are used or required, and the possibility of shared material culture without ethnic pretext is an incomprehensible abstraction and a heretic idea. This outlook disregards that no sort/type of object is a straightforward index of ethnicity, and at the same time neglects the fact that the "Dacian" cup/mug and pot are very simple shapes, that could have evolved from centuries old traditions of handmade pottery and easily spread with no ethnic meanings whatsoever. While diverse sorts of mobility could have and most probably did routinely occur all around the late Iron Age continental Europe and the Mediterranean, borrowings, adoptions and adaptations of objects and modes of production did not inevitably match ethnic fluctuations. As it has been argued, the "Dacian" pottery, and especially cups/mugs and pots, first appeared in II c. BCE and became widespread in the assumed core of the Dacian area by the beginning of I c. BCE (Crişan 1969, 153-163;Pupeză 2010, 133-134). While they are regularly considered as the emblem of the Dacian identity, it is telling they were not present in the inner Carpathian basin until the end of II c. BCE (Pupeză 2010, 146-147). Even more indicative is the fact that they are found in the same period at the "Scordiscan" sites. Namely, the oldest late Iron Age phase at Gomolava, dated to the second half of II c. BCE, as well as the first habitation layer at Židovar (end of II-beginning of I c. BCE), already contained the "Dacian" pottery, notably the cups/mugs and coarse pots (Sladić 1986, 29-33;Fig. 5/1). Additionally, for Gomolava is beyond any doubt that such pieces were produced within the settlement, and most probably were in everyday usage, but for different purposes than the fine gray and painted La Tène ("Celtic") ware (Jovanović B. 1979, 12-13;1991, 151-155). If indeed the cups/mugs were utilized as lamps (Jovanović B. 1991, 155;Popović 2000, 87;Pupeză 2010, 147) and/or had some other simple purpose (such as a lid -Egri 2019, 80), and the cylindrical pots were cooking vessels, their extensive and continual diffusion comes as no surprise. Although all of this was well known to the late Iron Age researches, they have failed to go beyond the embedded ethno-deterministic approach, which created completely artificial categorization of pottery types.
Furthermore, there is a commonly held, completely sound and in many cases confirmed archaeological position that various objects such as jewelry or brooches were the products of local workshops/artisans, made under the influence of foreign craft and style traditions. For example, and not to go further, the majority of items from the Židovar hoard are explained exactly in this manner  (Jevtić, Lazić and Sladić 2006). Having this in mind, why should be the ceramic material (such as gray wheel-thrown or coarse pottery) treated as an index of ethnic belonging? If there were influences in the form of technology and knowhow transfers, 11 through local emulations, travelling craftspeople or chain-effect apprenticeships, there is indeed no good reason to assume, let alone axiomatically hold, that some simple and widespread kinds of pottery implied ethnicity. Indeed, there is a range of objects with distributions that traverse traditionally supposed ethnic boundaries and suggest a rather dynamic and "ethnicity-free" transfer and exchange of technology, knowledge, practices and objects. In such examples, to name but a few and leave aside the obvious cases of widely shared fibulae designs of the late La Tène scheme, one could include Laminci type belt buckles (Plantos and Ciută 2016) and (more or less) continually used kantharos type of vessels (Rustoiu and Egri 2010;Dizdar 2010;Popović 2014).
Consequently, the dichotomy between the raw "Dacian" and fine "Celtic" ware could have easily been related to consumption practices i.e. the habits of producing wares for different purposes in two basically distinct ways (see Egri 2019, 81). None of these variations were necessarily linked to ethnic symbolism and expression of belonging via ceramic vessels, but instead suggest local appropriations of what has been archaeologically classified as the "Celtic" and "Dacian" pottery. Therefore, to avoid biased views and misleading conclusions, it is more objective to regard these not with ethnic labels but as much larger ("global" if you wish) techno-stylistic traditions that could be loosely and conditionally defined as Pannonian La Tène and Danubian-Carpathian respectively. Plausibly, the Pannonian La Tène features (wheel-thrown, gray, fine and painted pottery) were introduced and fused with local templates from the end of the IV c. BCE onwards (cf. e.g. Sladić 1986, 48-52, 64-66;Jovanović B. 2018, 160-174), while the Danubian-Carpathian ones (mainly the coarse pots and famous "Dacian" cups, probably stemming from the older pottery matrices) were adopted and widespread from the second half of the II c. BCE (Crişan 1969;Pupeză 2010). Contrary to traditional and theoretically disputable perspective, neither were the La Tène techniques the simple outcome of "Celtic mass invasion" (see Thér, Mangel and Gregor 2017), nor were the Danubian-Carpathian features the result of "the expansion of the Dacians". Most probably, the distribution of practices of production and consumption of both techno-stylistic features went by heterogeneous means through multi-leveled and diverse relational networks: e.g. exchange, copying, chain-effect artistry transfers via apprenticeships, travelling manufacturers, small-scale mobility, long distance marriages, etc. In this scenario, both Pannonian La Tène and Danubian-Carpathian ceramic fashions N "C " "D " 821 Етноантрополошки проблеми, н. с. год. 16 св. 3 (2021) could be regarded as commonly accepted preferences of production and usage (or, simply, trends), rather than indexes of ethnicity and domination of particular ethno-cultural groups. Hence, the case of Židovar cannot be longer explained by ethnic generalizations and instead has to be approached with different set of conceptual apparatus.
The bottom-up approach: Židovar and its "relational locality" The alternative way to engage in a more productive interpretation of Židovar has been already hinted by previous scholarship, although it was not matured, due to the inherent ethno-deterministic distortions of theoretical perspective. Several researchers have pointed out the specificities of material articulation of communities living at the crossroads of different ethno-cultural areas and communication routes. The site of Židovar and the area of Iron Gates (Đerdapska klisura; Porțile de Fier) are such instances: they have been regarded as the zones of permeation of cultural influences coming from both the "Scordiscan" and "Dacian" environments. Gavela asserted that foreign cultural elements, even far away from their original environment are the matter of course, and that "phenomena of cultural assimilation, syncretism and, in some cases, symbiosis lie in the nature of life itself and its endless dynamics, in laws that determine the link and dependence of economic and social processes and their influences on development and life-course of a culture" (1970,124, translated by the authors). Popović proposed that the Danube was neither a cultural nor ethnic boundary, but actually the linking factor and communication route around which similar populations lived: Therefore, in all those shapes that are roughly assigned to the Dacian material culture, we should perhaps seek local specificities that are hard to recognize, due to insufficiently researched and scarcely known regional circumstances, and our tendency to define [archaeological material] by using analogous and already determined [large] cultural complexes. In our case this would mean that certain forms, above all the coarse -house ware, must have had local origin. Also, the pronounced concentration of curved knives shows the connections with the Balkan area and the long tradition that, with the help of the Celtic technology in the course of late La Tène, made this tool (and weapon) so popular. (Popović 1991, 173, translation by the authors) In a similar manner, Sladić postulated that economic interests and vivid exchange contributed to cultural unification of the Yugoslavian part of the Danube region (1997, 66). Of course, all of the cited occasional views of this sort stemmed from the traditional matrix of ethnic intermixture and spread of culture via migrations and diffusion, especially in the border zones of two conceived "ethno-cultural complexes". However, the role of shared objects/templates across large areas, as well as the trends of techno-stylistic and cultural unification was accurately noticed, even though overshadowed by the search of ethnic attribution.
By far, all of the previous heralds of different perspective are surpassed by work of Mariana Egri, summarized in her recent book (2019). Not only she emphasizes the importance of supra-regional connectivities and cultural exchange in general, but she has directly focused on Židovar and the area traditionally seen as a boundary region between the "Dacians" and "Scordisci". Employing the concept of contact zones and connectivity, she casts doubt on interpretations of material culture as a proxy for ethnic dynamics and reaches the following conclusion: Neither is the increased presence of "Dacian" pottery related to this kind of demographic influx, or that of particular types of jewellery. These ceramic forms were already identified on site during the first phase, even if in reduced quantities and always mixed with "Scordiscian" forms, pointing to a heterogeneous cultural environment in which many daily practices or styles of consumption coexisted, sometimes also leading to the appearance of hybrid ones, a phenomenon which is commonly encountered in contact zones. Židovar was actually part of contact zone between the Scordisci and the Dacians; the social and cultural models of the former prevailed until the middle of the 1st century BC, when the latter got the upper hand. (Egri 2019, 81) Albeit Egri loosely conforms to the previous problematic interpretations about the Scordisci and the Dacians as clear-cut ethnicities, that interacted in a manner of well-defined socio-cultural groups, the insistence on the thesis of contact zone opens the room for further elaborations. One such approach, the closest to the perspective we elaborate here, is proposed by Dragăn (2012) in her study of the late Iron Age burials in the Iron Gates region. She calls for avoiding ethnic attribution of archaeological evidence (except as conditional nominal generalization) and focusing on functional analysis instead, that can reveal articulations of other and much more important social identifications and connectivities with no ethnic significance (Dragăn 2012, 436-437).
Expanding these standpoints further, Židovar can be comprehended as the first order community (sensu Varien and Potter 2008), i.e. the basic socio-spatial focal point for relations that operated both within the settlement and in supra-local networks (Mihajlović 2019, 307-309;. This scenario entails that the most intense and vibrant social relationalities and dynamics went on locally and, accordingly, the strongest sense of belonging and collectivity was situated at the habitation area and its immediate vicinity, and only then at some wider levels, such as regional and supra-regional affiliations of any kind. In other words, instead of explaining the particular (i.e. Židovar) by superficially imposing general N "C " "D " 823 Етноантрополошки проблеми, н. с. год. 16 св. 3 (2021) and misleading categories (i.e. "Scordiscan/Celtic" and/or "Dacian") it is more instructive to focus on the importance of the local and micro-regional settings. Of course, this absolutely does not imply that wider associations never existed, or had no importance for the entire populations or at least some group(s) of the local community. Instead, the proposed view means that we should reframe the order of importance used to define the phenomena of the past and acknowledge that "globality" of any sort exists and is articulated only through and at the local level. In simple terms, it is plausible to assume that whatever type of larger associations the (parts of) Židovar community had, they could operate solely if accepted and (re)negotiated by local social forces (probably of diverse sorts). Thus, if we conceptualize Židovar with the help of Massey's (2005) definition of place as the specific intersection of diverse ongoing relationalities, and remember the site's favorable geographic position at the crossroads between the Danube, Pannonia, Carpathians and Balkans, its distinctive locality could spring as the most important feature. Not only it could be defined as a contact zone, but the community of Židovar should be comprehended as a group which actively mediated and localized diverse social and cultural matrices and put them in use for their own purposes (compare Janković 2014 for similar processes in the Moesian Danube region). This could be explained as a series of internalizations of technology, design, utilization patterns and meanings, all of which in different periods and extents became parts of local traditions and ways of life. While some of these could be linked to the feelings of wider belonging and expression of supra-local identities, it is impossible to claim a priori which phenomena had such a role and to what kind of affiliation they were associated. Similarly, the local community, or at least its dominating parts, could enter and built larger social and identity structures, but it is currently impossible to speculate what was their extent and nature. 12 As local communities such as the Židovarian one participated in different relational networks (see e.g. Egri 2014, 176), one would expect the dynamic and perhaps relatively frequent shifts of allegiances. Correspondingly, the switches in shares of one over the other "global" flows, or introduction of new materialities and practices, could have meant fluctuations of intensity and quality of relations (in some cases maybe ongoing political and economic tendencies), but it is much less plausible that they signified the changes in all-encompassing ethnic affiliations. The idea of localization of certain aspects of wider cultural koines (e.g. La Tène, Danubian-Carpathian and Roman imperial) in bricolaged ways also brings the notion of creation of relational constellations of people and associated objects (sensu Van Oyen 2016). This means that local communities, or some of their parts, created specifically fused environments by means of internalizing various (material and immaterial) phenomena mediated through diverse kinds of relations. Indeed, the entanglement of (parts of) Židovar community with varied-scaled relationalities and networks through objects is manifested in all phases of its existence. The simultaneous utilization of both the Pannonian La Tène and Danubian-Carpathian ceramic traditions (and possibly consumption practices) illustrates this very well at the level of daily routine in all three habitation phases. Similar is the case of the Židovar treasure from the first settlement phase: since the hoard of luxurious items from diverse production provenances and areas of origin (see Jevtić 2006;Živković et al. 2014) demonstrates amassing and associating with different kinds of objects, it could be understood as the capacity to participate in multidirectional strings of long-distance exchange and relationalities. Furthermore, at the conditionally termed middle-range level (i.e. not as common as pottery and not so uncommon as the content of a hoard), coins found at Židovar testify to the same point. The chance find of Aninoasa-Dobreşti type (Group B -Dobreşti; Fig. 3/1), dated to the last decades of II c. or the beginning of I c. BCE, implies links with the area of Oltenia (Preda 1973, 274-288). The quinarius of P. Sabinus minted in 99 BCE (Crawford 1974, 331, no. 331;Fig. 3/2) and denarius of C. Naevius Balbus from 79 BCE (Bakić 2005, 14;Fig. 3/3), are most probably original issues and hint (at least indirect) connectivities that led to the Roman world. The "imitation" of Roman denarius with Dioscuri reverse (Fig. 3/4) most probably originated from the "Dacian" emissions (Jovanović 1997, 77;Bakić 2005, 15) and testifies to contacts in that direction. The finds of brooches 13 of the late La Tène scheme (published in Jovanović 1997, figs. 57, 59, 60;Jevtić 2006, 82-87), hybrid and Jezerine types ( Fig. 4/1, 5), as well as the belt buckles of Laminci type (Sladić 1997, fig. 51) suggest the connections to westwards production centers during the second habitation phase. On the other hand, the appliqué of the Piatra Cravii type (Jevtić and Ljuština 2008, 30), the silver ring with palm stamps (Rustoiu, Ferencz and Drăgan 2017, 204), the part of a hoarse bit (Božič 1984, 137), and spoon-bow fibulae ( Fig. 4/2, 3) are all interpreted as items from the "Dacian" production area. After the Roman conquest of the right bank of the Danube and the creation of the provinces of Illyricum/Pannonia and Moesia (9 BCE-14 CE), Židovar stayed outside the imperial political system within (what was in the Roman eyes) the Barbaricum (Mihajlović 2018). However, it continued to play a role of a settlement that retained contacts with nearby "globalized" cultural areas, albeit in the changed general setting. The relations with the populations from the Roman world apparently were less intensive than those with the Dacian, although the previously discussed imported material from the provinces (Fig.  4/4, 6-8, 5/2), as well as local ceramic forms modeled after the Roman imperial types (Fig. 5/4), indicate further localization of trends that came via supra-regional networks. Of course, it is one thing to identify the imports originating from different production sources and relational networks, but completely another matter to discern their roles within the local community. For the latter, we would need much clearer stratigraphic and archaeological evidence and extensive cross-referencing of a wide range of different contextual cases, but (unfortunately) the data from Židovar currently does not allow such a study. In any case, however, to pose the question whether Židovar was a "Celtic" or "Dacian" settlement is a false dilemma: both categories are academic constructs, made by linking ancient ethnonyms with selectively defined assemblages of material culture, neither was as stable and pervasive as is usually taken for granted, and sticking to them generates misguided conceptualization of the past.