Absence of evidence or evidence of absence? The microarchaeology of an ‘empty’ square enclosure of the Late Iron Age La Tène Culture

Archaeological features without macro-finds (i.e. finds visible with the naked eye) represent a common phenomenon that poses considerable methodological problems in interpretation. To tackle these problems, we present an approach based on integrating micro-archaeological methods and macro-observations to help elucidate ‘invisible’ activities or at least to exclude certain others. Specifically, we combine micromorphological and micro-debris analyses with the scarce macro-finds and their fragmentation degree to clarify whether the lack of evidence for certain activities really reflects their absence or merely bad preservation conditions. This study focuses on a Late Iron Age ‘empty’ enclosure excavated in the settlement site of Haselbach, Lower Austria. ‘Empty’ square enclosures of the La Tène Culture have been controversially interpreted as either sanctuaries, domestic buildings or burial enclosures, although these interpretations were rarely supported by any material evidence. Using a multi-proxy approach enables us to reconstruct the formation processes of the Haselbach feature as well as to rule out a range of activities, although a clear definition of its primary function was not reached. Based on our case study, we propose a research protocol that can be applied to research on similar ‘empty’ features in order to better validate their functional interpretations.


Introduction
'Empty' features, i.e. archaeological features without macroscopically visible finds (e.g. pottery, bones, metal artefacts), are often neglected for subjective or practical reasons: features with no finds do not seem worthy of study, features with finds are more attractive and hence preferred. Methodological reasons also play a role: they may be more difficult to date, their function is not evidenced by finds. Importantly, macro-finds themselves do not necessarily relate either to the dating or to the function of the studied features, depending on their formation processes (Schiffer 1996;LaMotta and Schiffer 1999;Kuna 2015). Their importance for studying past activity patterns is thus largely over-estimated, especially when compared to micro-finds and other kinds of 'invisible' evidence, also known as the 'microarchaeological record' (Weiner 2010).
This paper focuses precisely on features that present no or only a few macro-finds, a common phenomenon in archaeological excavations. We explore the potential of 'empty' features by implementing a combination of microarchaeological techniques and assess the reasons behind the absence of evidence of past activities, be it macro-or microscopic. We chose ditched enclosures, a typical category of Late Iron Age settlement features in Europe, as our case study. The methodological approach we propose here is, however, applicable to all kinds of 'empty' structures, without regional and chronological boundaries.
Ditched enclosures are widespread features of the Late Iron Age La Tène Culture in Central and Western Europe. Based on the size and layout of the ditches, they obviously served different purposes (Buchsenschutz 2000;Büchsenschütz 2019;Sievers 2006;Venclová 2006). The largest enclosures, with an area spanning from one to several hectares, are commonly interpreted as the delimitation of elite farmsteads ('Viereckschanzen' in German, 'fermes indigènes' in French literature;von Nicolai 2006;Wieland 2019;John 2019;Krause 2019;Fichtl and von Nicolai 2020;Fichtl 2021). Another category of quadrangular ditches encompassing around one thousand square metres are regarded as sanctuaries or feasting places (e.g. Ribemont-sur-Ancre: Fercoq du Leslay 2017; Fercoq du Leslay et al. 2019; Frauenberg near Leibnitz in Styria, Austria: Tiefengraber andGrill 1997, 2008;Grill 2009;Tiefengraber 2015; for an overview: Table Ronde 2000; Arcelin and Brunaux 2003;Barral and Thivet 2019). The smallest size category (up to a few hundred square metres, with side lengths of 4 to ca. 20 m) is the most numerous and shows the greatest variability in terms of use. Some of them delimited sanctuaries, banquet or feasting places (Arcelin and Brunaux 2003;Barral and Thivet 2019), others contained burials and thus delimited an ancestral space (Villes 1999;Baray 2003;Ramsl 2020; on the relation of these two categories cf. Trebsche 2021). Only recently, another category of 'enigmatic' enclosures devoid of burials, depositions, sacrifices, offerings and food remains was discovered. This type of enclosure was rather negatively defined as belonging to none of the aforementioned categories and vaguely labelled 'special places' or domestic enclosures (e.g. Herbin et al. 2013;Maguer et al. 2019;Le Quellec 2019).
In this paper, we focus on the latter category of small quadrangular enclosures, i.e. those that do not provide evidence of their function in the form of macro-finds, i.e. artefacts detectable with the naked eye during excavation and interpreted as primary depositions or primary refuse. Our case study is represented by a square ditched enclosure excavated in Haselbach, a Middle La Tène period settlement located in Lower Austria (on the settlement: Trebsche and Fichtl 2016b;a;Fichtl and Trebsche 2018;Trebsche and Fichtl 2020). Apart from rare macro-finds (isolated ceramic sherds, burnt daub and animal bones), this enclosure appeared completely 'empty', thus prompting questions about its intended function, actual use and infilling processes.
The next section presents a short overview of small 'empty' La Tène enclosures, followed by a detailed investigation of the square enclosure from Haselbach (feature 63) and its central post hole (feature 62). The analysis further includes a rectangular pit (feature 50) that cuts ditch 63. Our main questions are as follows: is there evidence for 'invisible' activities (i.e. activities not producing traces or refuse detectable with the naked eye)? What kind of activities can be ruled out based on absent evidence? We attempt to answer these questions by combining multiple lines of evidence: macro-finds (pottery, burnt daub, animal bones and one coin) and their fragmentation patterns, radiocarbon dating, micro-debris and micromorphological analyses. In the discussion, we combine both 'positive' and 'negative' evidence to evaluate the existing hypotheses (see below) on the function and use of these small square enclosures.

Enclosures without macro-finds or with very few macro-finds
While the interpretation of enclosures that yielded burials, primary depositions of offerings or primary refuse seems clear (in cases where the primary character of refuse can be demonstrated), the intended function and actual use of enclosures without finds or with only a few macro-finds is highly uncertain. Several explanations for the lack of macrofinds and several hypotheses for the interpretation of 'empty' enclosures without or with very few macro-finds have been published (cf. Bollacher 2009;Kellner-Depner 2016, 117): -Heavy erosion or deep ploughing destroyed previously existing burials, offerings or other finds; hence, the 'emptiness' is caused by formation processes independent from the original use of the enclosure. -Ritual use-either with no offerings, or only organic, decayed offerings, or offerings deposited elsewhere than in the ditch. -Domestic use.
-Workshop or working place.
The diverse interpretations are illustrated based on a few selected examples spanning (geographically) from East to West and (functionally) from ritual to profane ( Fig. 1 and Table 1). Rather than providing a complete typology of small square enclosures, we restrict ourselves highlighting the diversity of the phenomenon and its different interpretations.
The first site is located in Roseldorf (district of Hollabrunn, Lower Austria, Austria). It represents a large Middle La Tène (LT C) settlement with an area of approximately 40 ha according to magnetic surveys, characterised by a ditch surrounding the southern part of the settlement and three cult areas with seven square enclosures. These are interpreted as sanctuaries based on hundreds of weapon depositions, thousands of animal bones and human remains. While this interpretation is beyond doubt for the large enclosures (Holzer 2010;Holzer 2014b;Holzer 2019;cf. Trebsche 2020a), two of the smaller enclosures look different ( Fig. 2/1-2 Table 1/4). Inside, four post holes forming a square of 4.8 m side length reminded Christian Bollacher of the enclosures of the Saint-Benoît type, which were commonly part of funeral sites during the Early La Tène period. Bollacher interprets the Riedlingen ensemble as a monumental funerary building, where the institutionalised remembering of the dead was expressed in architectural form (i.e. the four-post-construction with ditch) and a ritual vessel deposition (Bollacher 2009, 123).
For these three sites from Austria and southern Germany, a ritual function was deduced rather from the settlement context and analogies (mainly the larger sanctuaries from Roseldorf) than based on finds. In contrast, small enclosures in northern France yielded settlement waste and/or special depositions. Their interpretation often remains ambiguous, for example in the case of Soupir 'Le Parc' (Département Aisne, Région Hauts-de-France, France) in the Aisne valley (Haselgrove et al. 2005). The quadrangular enclosure had an entrance (interruption) at the eastern side and two corresponding post-holes (Fig. 2/5 and Table 1/5). Due to the lack of human remains, Colin Haselgrove proposed a domestic rather than a funerary function, although he did not exclude a link with the nearby funerary structures. According to him, the existence of an enclosure with partially ritual function in a settlement would be no surprise (Haselgrove et al. 2005, 368).
Another very well studied square enclosure is that from Borre (Département Nord, Région Hauts-de-France, France) (Herbin et al. 2013) (Fig. 2/6 and Table 1/6). The few finds from the ditch infill were interpreted as domestic rubbish that had been secondarily deposited in the ditch, indicating a domestic function of the building (Herbin et al. 2013, 87).
Another enclosure with very sparse finds was discovered next to a rural settlement at Fontenay-le-Comte 'Genâts' (Département Vendée, Région Pays de la Loire, France) (Bouvet et al. 2003, 87 Table 1/8). This enclosure was interpreted as a cult place for the owners of the nearby farm. Notably, in the vicinity of this enclosure, another square ditch (9.30 × 8.7 m) and-at a distance of ca. 100 mthree circular enclosures without any finds were uncovered.
At the site Château-Thierry (Département Aisne, Région Hauts-de-France, France), an elongate enclosure surrounded a small square enclosure with very little find evidence (Le Quellec 2019) (Fig. 2/9 and Table 1/9). Somewhat north of it, two pits with depositions of 17 vessels and one potin coin were discovered. Vincent Le Quellec emphasised the long maintenance of this enclosed space for more than two centuries, which he interpreted as a place for modest familial or local rituals.
At Bessines 'Grand-Champ-Est' (Département Deux-Sèvres, Région Nouvelle-Aquitaine, France), a large sanctuary with a trapezoidal ditched enclosure (ca. 41 × 33 m) with an inside palisade and three wells in all but one corner was Finally, to illustrate the blurred distinction between burial and ritual enclosures, the site of Fontaine-la-Gaillarde (Département Yonne, Région Bourgogne-Franche-Comté, France) deserves mention (P. Nouvel in: Barral et al. 2003, 154 f.  Table 1/11). One hundred fifty metres to the west of an important Early La Tène (LT A) necropolis, two square ditches closely resemble the grave enclosures. However, they provided no evidence of burials or human remains, only some ceramics and a series of 68 coins and three Nauheim type fibulae, dating to the second half of the first century BC.
Based on these eastern Austrian, southern German and French sites, the occurrence of small square enclosures with sparse evidence of macro-finds is apparently not uncommon throughout the Eastern and Western La Tène Cultures, especially during the Middle and Late La Tène period. Unfortunately, this category of features has attracted much less attention than the 'rich' enclosures with abundant valuable offerings, i.e. the sanctuaries with remains of sacrificed weapons, animals, food and drink, or with primary or secondary burials of humans. From a methodological point of view, the interpretations of small enclosures mostly rely on 10 Bessines the analogy with the mentioned sanctuaries, sometimes also considering the context and, of course, the macro-finds, if present. Only rarely have archaeozoological and archaeobotanical analyses been conducted. To our knowledge, other scientific methods such as geoarchaeological analyses (micromorphology, sedimentology, geochemistry etc. ) have not yet been used in such contexts, despite the increasing application of these methods when investigating La Tène period ritual and settlement features especially in France, Switzerland and Czech Republic (e.g. Brönnimann et al. 2020aBrönnimann et al. , b, 2021Fechner 2000;Gillet et al. 2018;Goláňová et al. 2020;Langohr 2000;Parma et al. 2011).
It is our intention to fill this research gap by applying methods suitable for the investigation of the microarchaeological record in the sense of evidence that is 'invisible' for the naked eye (Weiner 2010). In the following, we present our case study conducted at the Haselbach site in Lower Austria, starting with the presentation of the excavated features, the evaluation of the macro-finds (mainly pottery, burnt daub and animal bones) and microarchaeological investigations (micro-debris and micromorphological analyses).

Site location
The site of Haselbach is located in the municipality of Niederhollabrunn, district of Korneuburg, in the federal state of Lower Austria, some 28 km north-northwest of Vienna. It extends over a flat plain at an altitude of ca. 210 m above Adriatic Sea level, between two small brooks (Hatzenbach and Senningbach) which flow into the Danube River, ca. 11 km to the south.
Geologically, Haselbach lies in the 'Outer Zone' of the Waschberg area, which forms the outermost strip of the Alpine-Carpathian ceiling stack north of the Danube (Wessely 2006). Gravel blankets as well as Pleistocene loess/loamy loess are present within a radius of about 3 km around the site. Small areas of Tertiary iron-bearing clays and sands occasionally interrupt these loess deposits, which are adjoined to the east by Tertiary schistose clays and clay marls (Geological Map in Lower Austria Atlas 4.0; http:// atlas. noe. gv. at).
Today, the area is farmland, but before the fields were drained in the twentieth century, it was surrounded by wetlands with riparian forests and wet meadows (Sauberer et al. 1999). The soils covering the site and its immediate surroundings are formed by Chernozem, interrupted only by small areas of Kastanozems, Gleysols and colluvia (www. boden karte. at). The La Tène period settlement lies on the mentioned loess/loamy loess substrate (AC to C horizons), covered by a ca. 0.30-m-thick (on average), entirely ploughed humus layer (Ap horizon).

Research history
After the identification of a La Tène period settlement measuring ca. 6.6 ha during a 2010 survey and geophysical campaign (Trebsche 2016; Trebsche 2020b), a French-Austrian archaeological excavation project was started in 2015. Its aim was to investigate the structure and organization of this Iron Age site (Trebsche and Fichtl 2016b;a;Fichtl and Trebsche 2018;Trebsche and Fichtl 2020). Specific focus was laid on clarifying the function and use of the characteristic pit houses (semi-sunken dwellings) as well as that of other constructions in the settlement. Among the numerous built structures identified during five field campaigns (2015-2019)-which include pit houses, pits, post-holes and, possibly, a well-one in particular posed questions related to its nature and function. This feature is a square enclosure (Feature 63) located in the west half of the settlement at the edge of an otherwise unbuilt area (Fig. 3).

Description of investigated features
The quadrangular enclosure Feature 63 measures 7.42 × 7.66 m on its outer edges (Figs. 4 and 5). The eastern side is somewhat longer (6.95 m) than the other three sides (6.43-6.53 m, measured on the bottom from corner point to corner point). The ditch shows a V-to U-shaped cross-section and is 0.45-0.65 m wide and 0.20-0.32 m deep (in its preserved state). The upper part of the ditch was clearly destroyed by intensive ploughing, so the original dimension must have been wider and deeper (see reconstruction below). The edges had been dug a bit wider (0.70 m) and not as deep (0.20 m) as the long sides, which probably reflects the use of a spade for digging. The fill layer consisted of dark brown, clayey-sandy sediment 1 , which was disturbed by animal burrows but otherwise homogeneous in composition-at least from a macroscopic perspective. Neither traces of re-cutting nor separate infill strata were observed. Based on the documentation, it can also be excluded that posts or piles once stood in the ditch; hence, it represents a simple enclosure and no palisade or building with walls.
The centre of the enclosure featured a large oval posthole (Feature 62) with vertical, straight walls (length 1.10 m, width 0.88 m, depth 0.90 m). Its bottom showed a clear circular impression of a thick round post with a diameter of 0.40 m and a flattened underside. While the overall post-hole is slightly shifted from the centre of the square enclosure, the post is placed precisely in its centre; i.e. it leaned on the southwestern wall of the post-hole.
In the fill layer of the post-hole, the traces of a round post wedge (interface 62-04, diameter 0.12×0.15 m, depth 0.08 m) with a black, clayey fill layer were recorded. Both the main post and the wedge were intentionally pulled out, as the ghosts of the post and the wedge did not appear in the upper fill layers of the post-hole, the dark brown stratigraphic unit (SU) 62-02 and the greyish-brown SU 62-01. A 0.22-m-deep extension at the western edge of the post-hole also indicates an extraction of the main post with a lever (Fig. 5B).
The precise alignment of the square enclosure and the central post leave no doubt that these two features belong to the same construction. The central post must have been extraordinarily high because it is approximately five times deeper than the other post-holes documented in the Haselbach settlement.
The northern side of the square ditch 63 was cut exactly in the middle by a rectangular pit (Feature 50), which was 2.34 m long, 0.94 m wide and max. 0.20 m deep preserved. It was dug only when ditch 63 had already been filled and thus represents a second use phase. The fill layer of pit 50 consisted of homogeneous greyish-brown clayey-sandy sediment with traces of charcoal.

Reconstruction of Feature 63
The reconstruction of Feature 63 must consider the partial destruction of the ditch by modern ploughing (Fig. 6). The 0.30-m-thick humus layer above the ditched enclosure was completely ploughed. We should therefore consider that the original fill of the ditch probably extended for at least further 0.30 m, that is the thickness of the current humus layer above these features. As a result, the original ditch must have been at least 0.55 m deep. Assuming that the inclination of the destroyed upper parts of the wall was the same as for the preserved walls, the original width of the ditch was almost 1.0 m (0.97 m) instead of the preserved average width of 0.55 m. Thus, the original minimum volume of the ditch can be calculated as approximately 8.48 m 3 (preserved volume: 1.89 m 3 ).
As argued above, the precise alignment of the square enclosure and the central post demonstrates that they belong to the same construction. Based on practical considerations, we assume that first the mighty post (diameter 0.4 m) was erected and then the ditch was created around the post (rather than vice versa). The excavated sediments (volume: 8.48 m 3 ) were probably deposited inside the square enclosure (inner area: 5.5 m × 5.5 m), resulting in a pyramidal elevation of max. ca. 0.5 m around the central post.

Radiocarbon dating
Three radiocarbon samples were taken in order to determine the absolute chronology of the investigated features. Samples were measured by Accelerated Mass Spectrometry at the Poznan Radiocarbon Laboratory in Poland. Two horse bones (samples HAS-03 and HAS-04) from SU 63-01 produced almost identical radiocarbon ages, with calibrated results pointing to the third century BC (with greatest probability 307-210 BC, and 301-201 BC, respectively; for the exact calibration results, see Table 2). One charred grain of barley from stratigraphic unit [SU] 50-01 (sample HAS-14) confirmed that feature 50 is significantly younger than the square enclosure 63. The calibrated date (150-46 BC, 1 sigma) is the latest of all radiocarbon dates (n=36) measured from Haselbach so far, indicating that the filling of this feature belongs to the final abandonment phase of the settlement.

The macro-finds
The macro-finds comprise all finds that were recovered manually with the naked eye during the excavation ( Fig. 7 and Tables 3 and 4). In total, the infill (SU 63-01) of the square ditch 63 yielded nine pottery fragments, 54 animal bone fragments, nine pieces of burnt daub, four complete mollusc shells and fragments of two more molluscs (probably natural intrusions). The ceramic fabrics correspond with the typical Middle La Tène period pottery found elsewhere in the Haselbach settlement, although the sherds lack characteristic features allowing for a more precise dating. Among the infill strata of post-hole 62, only the two uppermost layers 62-01 and 62-02 contained finds: three animal bone fragments and one piece of burnt daub. The richest feature was pit 50, with 22 pottery sherds, 39 animal bone fragments, 48 pieces of daub, one iron fragment with a circular cross-section (length 3 cm, Staedelin 2021, 123, 125) and one gold coin found in its fill layer SU 50-01. The find density (calculated as the total number of finds divided by the volume, i.e. finds per m 3 ) varies considerably among the three features: the fill layer of the ditched enclosure 63 contained 38.1 finds/m 3 , the filling of post-hole 62 only 6.5 finds/m 3 and the fill layer of pit 50 252.3 finds/m 3 . Counting only pottery, Feature 63 from Haselbach contained 4.8 fragments/m 3 (46.6 g/m 3 ), Feature 50 contained 50 fragments/m 3 (1113.6 g/m 3 ) and Feature 62 yielded no pottery at all. Comparable figures from other Iron Age settlements are available only for pottery. In the La Tène period and Roman period settlement of Březno near Chomutov (Czech Republic), the average number of pottery fragments in sunken features is 118.6 fragments/m 3 (2033 g/m 3 ), varying from 10.0 to 242.1 fragments/m 3 (278-4096 g/m 3 ) (Salač et al. 2007, 267 Table 1). In the oppidum of Manching (excavation area Altenfeld), five sunken features yielded an average of 122 pottery fragments/ m 3 (1349 g/m 3 ) (Salač et al. 2007, 268 Table 2). Although the find densities for all features from Haselbach have not yet been evaluated, the comparison with other published settlement data shows that Feature 50 ranges among typical Late Iron Age features, whereas Features 63 and 62 show lower pottery find densities than recorded elsewhere.
Regarding the internal find distribution within the infill of Feature 63, the distribution of pottery fragments seems random, whereas the daub fragments concentrate at the northern side and animal bones (especially from domestic horse, see below) were mostly found at the southern side and the southwestern corner of the enclosure (Fig. 8).
The fragmentation of macro-finds shows marked differences in the two most abundant Features 63 and 50 (Table 3, Fig. 9). The average length of pottery fragments is significantly lower in the ditch infill (SU 63-01) than in the pit infill (SU 50-01). A detailed examination of the histograms of fragment lengths (Fig. 9A) confirms that fragmentation of pottery is much higher in SU 63-01 than in SU 50-01, i.e. the assemblage from SU 50-01 contains much larger fragments. Correspondingly, the average pottery sherd (9.8 g/sherd) from SU 63-01 weighs less than half of the average pottery fragment from SU 50-01 (22.3 g/sherd). This is much lower than the average of the whole Haselbach pottery assemblage (26.4 g/sherd, n=16766). Another indicator is the degree of sherd erosion, classified semi-quantitatively in three classes based on the sherd edges (Table 4). A high The burnt daub fragments show the same tendency in fragmentation: They weigh only 2.2 g on average in SU 63-01, but 10.0 g in SU 50-01. In the fragmentation histogram, this marked difference is less clearly visible (Fig. 9C). Nonetheless, all 'daub' finds are so heavily fragmented that it is not possible to distinguish daub from wall constructions and burnt clay from hearths or ovens.
In contrast with pottery and daub, the animal bone fragments from both stratigraphic units show almost identical average weights and lengths (Table 4). The fragmentation histogram (Fig. 9B) shows similar patterns for SU 50-01 and SU 63-01 with two exceptions: First, some very large (> 15 cm) bone fragments from the latter unit, and second, the high number of fragments in the classes 0-1 cm and 1-2 cm in SU 63-01, which reflects the especially careful find retrieval during the excavation of Feature 63.

Archaeozoological analysis
The faunal remains were identified at the Museum of Natural History Vienna (1st Zoological Department, Archaeological-Zoological Collection). To quantify the material, NISP (number of identified specimens) and weight analysis were used. Aging was determined by examining the epiphyseal fusion (Habermehl 1975), since no relevant teeth were found (deciduous teeth: premolars Pd 4 /Pd 4 and permanent teeth: molars M 3 /M 3 ). Sexing was possible only for one mandibula of a horse based on teeth and alveoli (Hillson 2005 (1976). The height at withers was calculated for one horse radius following the factors set by May (1985) and for one metacarpus of a goat based on the factors set by Schramm (1967). Finally, butchery marks were interpreted based on Binford (1981).
In total, 53 animal bones and teeth weighing 1019.6 g were identified and analysed from the enclosure in Haselbach ( Fig. 10 and Table 5); 28 fragments remained unidentified (34.1%). The vast majority of the archaeozoological material derives from Features 50 (n=25) and 63 (n=21); only one bone was identified from Feature 62. The identified material mainly belongs to domesticated taxa: bones of cattle, sheep/goat, pig and horse were documented. Remains of an intrusive specimen of European ground squirrel (n=6) were retrieved from Feature 63 ( Table 5).
The prevalence of the domesticated species and the general composition of the faunal remains bear significant similarities to the faunal material from other contexts within the settlement of Haselbach, which comprises more than 6,000 identified animal remains (Saliari and Trebsche 2023). However, the relatively high number of horses (NISP=10) documented in the enclosure (Feature 63) is remarkable ( Fig. 10(a-c 1 ); Table 5), especially when considering the  (Pucher 1998;Pucher 2006;Galik 2008;Grill 2009). The Great Sanctuary (Feature 1) from Roseldorf constitutes a notable exception: there, horses were a crucial part of the ritual activities, represented by 10.8% based on the number of identified specimens (Abd El Karem 2013).
The most common modification documented on the animal remains from the enclosure is butchery marks. Such marks were also found on bones of horses, confirming that hippophagy was not unusual (Pucher 1998;Bruckner-Höbling 2009;Abd El Karem 2013;Saliari et al. 2016). All boneswith the exception of one (horse) radius and one (goat) metacarpus-exhibited chop marks. Chops on the cranium were documented for cattle (zygomaticum), sheep/goat (corpus mandibulae and caput mandibulae of proc. condylaris), pig (maxilla) and horse (zygomaticum, corpus mandibulae), suggesting dismemberment. Diagonal chops on long bones of all species at different parts on the joints and the midshaft indicate dismemberment and portioning, longitudinal chops on long bones of cattle and horse reveal marrow extraction. A comparison with material from other contexts in the settlement of Haselbach shows the existence of similar chop marks and a very low number of cut marks (Saliari and Trebsche 2023).
Furthermore, gnawing marks caused by Rodentia (Fig. 10(c 2 )) were recorded at the distal epiphysis of a horse tibia (Feature 50, find number: 2464). Slight weathering was noted at the distal epiphysis of a pig humerus (Feature 63, find number: 3269). These finds suggest that at least some of the faunal remains were left exposed on the ground or were only partially buried. The relatively well-preserved pig humerus suggests that the bone was not exposed for an extended period of time.     Finally, the height at withers was reconstructed based on one horse radius at 120.7 cm (Fig. 10a)

Micro-debris analysis
A total of 21 samples of loose sediment were systematically collected from Features 50, 62 and 63 for micro-debris analysis and wet-sieved to recover the heavy fraction of each sample. Remains between 10 and 1 mm were collected, separated into different categories and counted at the Microarchaeological Laboratory of the University of Innsbruck (Table 6)  -Graphite = mineral particles with platy structure that leave clear marks on paper. -Fish bones = identifiable scales, otoliths and thin cranial bone particles. -Iron oxides = highly magnetic reddish (rusty) mineral particles, such as fragments of metal and pigments.
The density (ρN) of each residue category was then calculated by dividing the number of finds by the volume of the samples (Table 7) enabling a comparison between samples with different sediment volumes.
Given the considerable variability in density between debris types, these were grouped in three categories (frequent, few, trace) based on their total density value, and then plotted in ArcGIS® 10.8 by means of spline interpolation (Fig. 11). This way of grouping the remains enabled better visualising the data of the less abundant categories, which would have been otherwise masked by the most abundant residues.
The most evident results are that Feature 50 presents the highest density of micro-residues (which corresponds to the macro-finds there) as well as the largest variety of debris types (Fig. 12). Burnt daub is the most abundant type of remain here, whereas lithic remains of the very coarse sand and gravel fraction as well as molluscs prevail in Features 62 and 63.
The interpolation maps (Fig. 11) do not reveal particular distribution patterns, and the micro-remains seem to be equally scattered in the fill of the three features.

Micromorphological analysis
A total of 20 samples were collected for micromorphological analyses from the sediment fill of the three contexts (Features 50, 62 and 63), and 12 of them were selected for this study (Table 8). Blocks of sediment were cut from the exposed profile(s) of the investigated features, targeting the lowermost portion of the fill directly in contact with the underlying loess substrate (Fig. 13a, b).
The 30-μm uncovered thin sections (Fig. 13c, d) were studied with a Leica DM2700 P polarising microscope, at magnifications ranging from 12.5× to 400×, under plane polarised (PPL), cross polarised (XPL) and oblique incident light. The standard micromorphological analytical guidelines and manuals were used for the description (see Tables 9, 10 and 11) and interpretation of the nature, characteristics and spatial arrangement of the sediment components (Verrecchia and   Bullock et al. 1985). Pedologically, the three features are located in an area covered by Chernozem soils that developed on loess substrate. Such soils are relatively dry and present moderate water storage capacities as well as moderate water permeability. The horizons composing the soil profile are A1p, A2, AC and C. The top A-horizons are characterised by a moderate humic content consisting of litter. The calcium carbonate content, which is already high in the upper humic horizons, increases downwards in the transitional AC-horizon and in the unweathered C-horizon, leading to neutral to alkaline soil reactions. These Chernozem soils are good for agricultural work and represent high-quality arable land. Based on soil erosion risk-assessments, they are considered at moderate risk from wind and runoff (https:// boden karte. at).
All three samples from Feature 50 present a poorly to well-developed crumb structure, with a well-expressed porosity that, in the case of sample 2372, reaches 50%. Only in one case (sample 2370) can two deposits be clearly distinguished. The lowermost corresponds to the loess substrate in which this feature was dug. The loess is characterised by a calcitic crystallitic b-fabric and an almost completely minerogenic composition, represented by quartz, feldspars, calcite and, to a lesser degree, mica and glauconite (Fig. 14a). Despite the rare presence of charred plant material and burnt clay aggregates, moved downwards by soil fauna and/ or flora, this deposit represents the hardly to unweathered substrate of the C-horizon.
The other samples show a single deposit composed of a sandy clay loam with abundant interspersed amorphous organic matter, conferring the deposit with a brown/dark brownish grey coloration in which loess aggregates of different sizes are evenly mixed. The groundmass is characterised by abundant Fe/Mg ox(hydr) impregnations and small nodules as well as by indications of biologic activity (passage features, mesofauna pellets, calcite precipitation in root channels) (queras; see Fig. 15) produced by the activity and decay of plant roots (Durand et al. 2018) and occasional fresh roots. More residues of anthropic origin were found in this mixed organic-rich deposit: charcoals, charred seeds and bone fragments, as well as aggregates of dung pellets containing what appear to be rare calcitic spherulites (Canti and Brochier 2017;Canti 1998). The groundmass contains possible fungal spores, ashes, burnt clay aggregates, (rare) pottery fragments and few silica slags ( Fig. 14b-g). The groundmass of sample 2372 is interspersed with few small rounded phosphatic aggregates (also containing rare spherulites-possible ashes from burnt dung?) (Fig. 14h). The orientation and distribution of all particles is random but overall homogeneous throughout the deposit.
The two samples from Feature 62 present a moderately massive structure, less crumb than in Feature 50. Sample 2778 is characterised by a slightly sandier groundmass (Fig. 16b) Table 7 Haselbach. Samples analysed for micro-debris, with their volume (before wet-sieving) and the numeric density of each class of residue in the three features although the mineral composition of these samples is the same (Fig. 16a). Both samples are constituted by a single microlayer, mainly composed of the carbonate-rich loam mixed with unsorted (from pebble to sand-sized) rounded organic and iron-rich aggregates. The latter can be interpreted as infiltrations from the topsoil due to biologic activity (Fig. 16c).
Beside the minerogenic component, no refuse connected to human activity was documented. Both samples show indications of biologic activity (Fig. 16d), although less intense than in the Feature 50 (especially root channels).
Two out of six samples from Feature 63 (no. 2679 and 3254) present two different microlayers, as in Feature 50. One  is the loessy light grey, micritic and rather compact substrate underlying the features fills; the second is the organic-rich fill itself, with a crumb groundmass but an overall similar composition as the substrate (Fig. 17a). The boundary between the two deposits is very diffuse and digitated. The remaining samples of Feature 63 feature a single layer composed of a mixture of more or less coarse aggregates of the two types of deposit (Fig. 17b). Impregnative pedofeatures are very common and Fe/Mg ox(hydr) nodules are slightly more abundant than in Feature 50 (Fig. 17c-d). Rare phosphatic nodules are present in sample 2759 along with occasional rounded cases of possible fungal spores (Fig. 17f) in all samples. Porosity is well-expressed in the mixed organic loamy deposit of all samples, showing clear traces of biologic mixing, mostly animal but also due to root growth. As in Feature 62, no refuse or particles related to human activity were recorded, apart from a rare bone fragment (Fig. 17e) in sample 3095.

Formation processes
Overall, the fill deposits of Features 63 and 62 closely resemble the loessy soil material in which they were dug. Micromorphological analyses also reveal that these fill

Table 10
Haselbach deposits are fairly homogeneous and that bioturbation contributed to the mixing of the sediments. In both features, the same homogeneity is reflected by the micro-debris distribution patterns. Nonetheless, modern bioturbation and pedogenetic processes are not responsible alone for the structure and composition of the fills, in which case also the surrounding AC transitional horizon would have been affected by the same processes and the archaeological features would have been hardly recognisable at all.
Considering that there is no indication of a previous use of the area before Feature 63 was created, we can assume that the enclosure was dug directly in the original soil capping. The different horizons (A, AC and C) of the prehistoric Chernozem had thus been mixed together in the soil heaps that resulted from the construction of the features. Indeed, the dark coloration of the fill reflects the presence of amorphous organic matter, which in turn indicates that humic material had been incorporated in the deposits. Some of the more organic aggregates interspersed in the fills (see Fig. 16b) may instead represent 'intrusive' modern Chernozem topsoil material. This same (mixed) soil material appears to have then infilled the structures when their use came to an end.
As for how this happened, the lack of stratification or of repeated accumulations suggests rapid sedimentation processes. The absence of layered and repeated depositions of natural fill could also be due to either regular maintenance or to their use for a short time span. Cleaning activities-consisting of the removal of sediment-would still have been micromorphologically visible in the form of truncations and/ or remnants of the original stratigraphy, despite the intense bioturbation (as indeed observed in other structures that are not part of this study). The lack of such features therefore points to events of short duration and/or not too prolonged use of the structures. After they were sealed, the area of the enclosure was not used anymore, or at least no construction was built in their place. The only exception is represented by the shallow pit, Feature 50. Despite being constructed in a later moment in time-and partly cutting the deposits of Feature 63-there appears to be no difference in the sedimentary nature of this fill. Once the structures were backfilled, with the onset of a period of stability, pedogenesis resumed and promoted the further mixing of the deposits.

Types of refuse
In general, no indication of primary refuse-in the form of either trampled debris or layered accumulation-could be identified in any of the studied features. The low quantities of anthropogenic particles additionally point to the lack of specific activities in the area. Despite intense reworking and the lack of clear-cut trampled activity-areas, the diverse occurrence of macro-finds and micro-refuse among the three features can be used to highlight the different nature and possible use of these structures.
Among them, the shallow rectangular pit Feature 50 presents the highest number and variety of both macro-finds and micro-refuse. Based on the find density, the state of fragmentation and erosion, the macro-finds (ceramics, burnt clay, bones and one gold coin) from SU 50-01 represent secondary refuse (Schiffer 1996;LaMotta and Schiffer 1999), i.e. refuse that was redeposited from the place of its origin and finally deposited in the abandoned pit 50. The different types of micro-debris, which include bones, burnt bones, burnt  and what appears to be an (empty) fungal spore [f?]-or another non-pollen palynomorph-are randomly distributed. PPL; e section through a hollow bone with osteocytes (the fine black dots covering the surface of the particle). The structure of the bone is well-preserved, although perhaps exposed to heat (based on its grey-brown coloration). Left part of the image in XPL; right part in PPL; f aggregate of finely trimmed charcoals and calcitic ashes (yellow arrow). PPL; g detail of a large particle characterised by a reddish-orange matrix withdense row of elongated fibres. Micro-XRF analyses revealed its elemental composition and identify it as bone (perhaps burnt, based on its colour (e.g. food preparation, roasting and sweeping of fire rake out together with fragments of the fireplace itself), rather than from the waste of specialised workshops (e.g. metal working or pottery production). In contrast to Feature 50, the square enclosure ditch (Feature 63) and its central post-hole (Feature 62) provided a very low density of macro-finds and almost no micro-debris. The more fragmented state of the ceramic and daub macrofinds from the fill layer 63-01 in the square ditch indicate that they probably belong to the category of tertiary refuse, i.e. redeposited secondary refuse or 'residual' refuse (in the sense of Kuna 2015). Only the animal bones found in SU 63-01, which show a similar fragmentation pattern and are partially even less fragmented as those from SU 50-01, could also belong to secondary refuse. Almost no (perhaps none at all) particle related to human activity was found in the micromorphological samples of either the square enclosure (Feature 63) or the central post-hole (Feature 62). The higher occurrence of stones and shells (recorded also in Feature 50) in the micro-debris seems merely to reflect the natural composition of the mixed soil material.

Inferences from absence
Based on the macroscopic absence of intentional depositions, domestic rubbish or production refuse disposal in the Fig. 17 Haselbach. Photomicrographs of thin sections from Feature 63 illustrating the main features of its deposits: a Overview of the two different sediment types, recurring in all studied samples. Although they present overall the same texture, coarse composition and fabric, their groundmass varies notably: the deposit on the left side of the image is reddish-dark brown, partly organic and oxidised. On the right side, a grey micritic groundmass is characterised by a calciticcrystallitic groundmass. The boundary between the two deposits is relatively evident and quite convoluted. XPL; b same as the previous image but in this case the reddish/oxidised deposit is visible in the form of rounded and sub-rounded aggregates [a] interspersed in the micritic layer. PPL; c-d examples of Fe/Mg ox(hydr) [Fe] frequently observed in the groundmass of the deposits from Feature 63. Both photomicrographs are in PPL; e detail of the only bone fragment found in the ditch of this enclosure. The bone structure appears isotropic in XPL (on the left), suggesting bad collagen preservation; see also a series of fine dark dots-osteocytes-in PPL (on the right); f detail of the circular, small-sized (~250 μm in diameter) particles, hollow and with an organic orangey and smooth coat, possibly belonging to a fungal spore [f?]. PPL. Graphics: S. Cereda square enclosure (Feature 63), we rule out such activities inside or next to it. In the Introduction, we emphasised that positive evidence in the form of macro-finds is rarely indicative of the actual use of archaeological features. Does this make the reverse conclusion true? When can the lack of evidence be considered as evidence for a lack of activity (cf. Wallach 2019)? The argument of lacking macroscopic evidence in Feature 63 is strengthened by the paucity or complete absence of material evidence in micromorphological soil samples and among the micro-debris sieved from bulk sediments. The lack of such evidence cannot be related to the conditions of preservation, as demonstrated by the occurrence of micro-remains in the overlapping Feature 50. Offerings, sacrifices, domestic or artisanal activities usually leave a visible and significant footprint in the form of depositions or refuse, which is not the case with Features 63 and 62. In contrast with almost all other La Tène period settlement features at Haselbach, Feature 63 (square enclosure) was kept clean from refuse and was not used for depositions of any kind. The macroscopic observations are substantially validated by the combined microarchaeological analyses providing only negative evidence. This treatment of the enclosure underlines its special function in the middle of the settlement. This calls for considering activities that do not necessarily leave visible or invisible material traces inside or around the enclosure (Feature 63). Examples might include assemblies, trading, jurisdiction, making music, dancing, etc. The lack of macro-and microscopic find evidence from the Haselbach enclosure provides a valuable argument that such non-debris-producing activities may have exclusively been taking place in this unbuilt space.
From an epistemological perspective, 'one must be aware that the justification for inference from absence, as outlined above, is essentially local since it rests on the failure to find sought-after evidence in an intensive localised search' (Wallach 2019, 7). The conclusions we draw are therefore valid only for the Haselbach enclosure (Feature 63). Nevertheless, the Haselbach case certainly represents a useful analogy for similar 'empty' enclosures that were not investigated using this microarchaeological approach.

Conclusion
Micro-debris and micromorphological analyses yielded a better, more in-depth understanding of the 'empty' square ditched enclosure (Feature 63) and its central post-hole (Feature 62) as well as of the shallow pit (Feature 50) cutting the ditch, compared with the evaluation of the few macro-finds alone. Notably, in Feature 50, the spectrum of micro-debris is broader than that of the macro-finds, including charred plant remains, fish bones, graphite and dung (?), which without appropriate sampling would have escaped our attention.
Although bioturbation and sediment mixing play a major role in each of the studied features, they cannot alone explain the absence of activity-related remains. Indeed, despite presenting similar post-depositional characteristics, the shallow cavity 50 still contains refuse material that indicates its infilling with domestic refuse-or maybe even deliberate disposal. While these results do not provide a direct answer to the original question (what was the square enclosure Feature 63 used for?), they do indicate that the lack of traces related to specific activities in its fill reflects an actual 'evidence of absence' rather than merely poor preservation of the archaeological deposits. Currently, we cannot say what type of practices took place in Feature 63. Our research showed that the use-life of the enclosure was probably rather short. Moreover, the only potentially purposeful activity attested-the secondary deposition of animal bones (mainly from horse)-reflects the only distinguishable episode of human activity near that structure. Importantly, even though an attribution of the structure to cultic and/ or ceremonial purposes goes beyond the interpretive potential of this study, we can conclude that the lack of normal domestic/craft-related refuse, together with its peculiar form and location, place the structure beyond the sphere of mundane architecture. This aspect is underlined by the fact that the enclosure ditch was kept clean from (macro and micro) rubbish, which is otherwise omnipresent in the filling of practically all La Tène period features at the site. We expect the final interpretation of the Haselbach and other similar enclosures to ultimately rely on analogy-and context-based interpretations (for the methodology see Trebsche 2009). This makes the evaluation of the other settlement features (mainly sunken dwellings and storage pits) from Haselbach as well as from La Tène period enclosures from other sites crucial for comparison with the data presented in this paper.
Although the microarchaeological investigation in our case study has not clarified the original function and use of the enclosure, it has helped to exclude many hypotheses. In this process of elimination, further analyses-such as geochemical investigations, fungal spore and phytolith studies or environmental DNA analyses-should be pursued in the future.
Our approach could be applied to other excavations with 'empty' features. We are, of course, aware that such cost-intensive analyses might not be feasible in most rescue excavations and even in some planned projects. We do recommend, however, to take as many samples for different purposes as possible in a systematic manner during the fieldwork. This would enable starting with a small sub-sample of sediment samples to retrieve micro-debris and a small number of micromorphological samples for the post-processing phase of the project. Our experience shows that even a limited effort can yield reasonable results, i.e. identify invisible activities or exclude certain activities. As a standard procedure for the analysis of quadrangular enclosures, we suggest to first investigate at least eight wet-sieved samples from the four sides and the four corners of the feature. In a second step, at least four micromorphological samples can be analysed. Depending on the results, further micromorphological and other analyses should be undertaken.
Finally, the study of an 'empty' enclosure also contributes to challenging the interpretation of those enclosure ditches filled with abundant finds such as iron weapons, animal bones or human remains which were clearly recovered in a secondary or tertiary position but nevertheless commonly interpreted as the remains of the primary use of the corresponding feature. Consequently, intensive sampling and scientific analyses should also be regularly integrated into the research on find-rich enclosures. Special emphasis should be put on the lowermost fill layers of the ditches, which in some cases have been proven to be deposited during a phase of natural erosion (e.g. Roseldorf, Object 1; Roetzel et al. 2009). The same also applies to the upper find layers in order to reconstruct the fill process and the provenance of the macro-finds.
Author contribution SC and PT conceptualised the study and wrote the main manuscript text. PT directed fieldwork and sampling and analysed the macro-finds. SC performed the micro-debris and the micromorphological analyses. MM recorded the macro-finds. KS performed the archaeozoological analysis and wrote the corresponding section of the text. BZ sorted and determined the micro-debris. All authors reviewed the manuscript.
Funding Open access funding provided by University of Innsbruck and Medical University of Innsbruck. The excavation project at Haselbach was funded by the State of Lower Austria (Department of Culture).

Data availability
The authors declare that the data supporting the findings of this study are available within the article.

Consent for publication Not applicable
Competing interests The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.