Radiografía de un castro de la Edad del Hierro: arqueología no invasiva en el asentamiento de Villasviejas del Tamuja (Botija, Cáceres) ; Radiography of an Iron Age hillfort: non-invasive archaeology in the settlement of Villasviejas del Tamuja (Botija, Cáceres)

The aim of this paper is to present the progress that has been made in the study of the hillfort of V illasviej as del Tamuj a (Botij a, in the province of Cá ceres), based on a combination of non-destructive techniq ues. The possibility of cross-referencing the results obtained w ith different geophysical methods (magnetometry, georadar and electric tomography) raises the opportunity to formulate a reliable diagnosis on the spatial organiz ation of this site. The most relevant results are described, and their contribution to a general interpretation of the urban structure and morphology of domestic spaces is analyz ed. This information is contrasted w ith previously available ex cavation data and its framing is analysed w ithin the ex isting k now ledge on this type of settlements. In addition, elements of interest about the diachronic evolution of the site are added, w hich are relevant in relation to the hypotheses formulated about the impact of the R oman conq uest. U ltimately, the potential of this type of research strategy for the study and revaluation of large and complex archaeological z ones is considered.


RESUMEN
El objetivo de este trabajo es proporcionar un primer avance del estudio del castro de Villasviejas del Tamuja (Botija, provincia de Cáceres) a partir de la combinación de técnicas no destructivas. La posibilidad de cruzar los resultados obtenidos con diversos métodos geofísicos (magnetometría, georradar y tomografía eléctrica) plantea la oportunidad de formular un diagnóstico fiable sobre la organización espacial de este enclave. Se describen los resultados más relevantes, y se valora su aportación para una interpretación general de la estructura urbana y la morfología de los espacios domésticos. Esta información es contrastada con los datos de excavación previamente disponibles, y se analiza su encuadre dentro del conocimiento actualmente existente sobre este tipo de asentamientos. Asimismo, se añaden elementos de interés acerca de la evolución diacrónica del sitio, que son relevantes en relación con las hipótesis formuladas acerca del impacto de la conquista romana. En última instancia se pondera el potencial de este tipo de estrategias de investigación para el estudio y revalorización de zonas arqueológicas grandes y complejas. fort of V illasviej as del Tamuj a, raised the opportunity to turn this site into an open laboratory to test these procedures.
The team includes researchers from the Institute of Archaeology, the U niversity of E x tremadura and the Centre for Scientific and Technical Research of Extremadura (CICYTEX). W e are putting together the efforts of specialists in geology, geophysics, remote sensing, archaeological survey and the application of geospatial technologies. Thank s to these combined efforts, w e have been able to launch a research programme w ith tw o main fronts. O n the one hand, a plan for geophysical survey that includes both magnetic and geoelectric methods and G round P enetrating R adar (G P R ). Their results are the main focus of this article. In addition, w e have implemented a battery of remote sensing methods (thermal and multispectral imaging, LID AR , and photogrammetry). W ith the combined analysis of all these results w e hope to achieve a more complete k now ledge of our case study in tw o main dimensions. F irst, w e seek the w idest possible coverage to help us understand the internal structure of the settlement. W e hope to track different typologies of built spaces, and contrast the hypothesis of the ex istence of open areas w ithin the enclosures. Secondly, w e are look ing for clues that help us to understand the diachrony of a site that w as occupied for three centuries. M ethods such as G P R and electric tomography can provide us w ith clues about the urban layout prior to the final stage, as well as the impact that the R oman presence had on the continuity of the settlement. U ltimately, w e hope that this information can provide relevant data on the degree of planning, demography, functional specializ ation of certain areas and about the variability of domestic spaces.

THE STUDY CASE
The hillfort of V illasviej as del Tamuj a is located in the province of Cá ceres, about 25 k m southeast of the capital and in the municipality of Botij a. The surrounding area is characteriz ed by a domain of meadow land w ith scrub and pasture, mainly used for livestock farming (the dehesa landscape). In this sense the archaeological site is a peculiar element in its environment, as historically it w as an area used purely for agriculture, until the farmland w as abandoned over the last few decades (F ig. 1).
The settlement comprises tw o enclosures w ith a total area of about 7 ha. The largest is located to the north (F ig. 2A) w ith a surface area of 4.7 ha. The natural defences offered by the Tamuj a river and the stream of V erraco are reinforced by a ditch and a w all flanked by towers and bastions. Enclosure B is locat-

INTRODUCTION
It is a w ell-established fact that, during the Second Iron Age (4 th to 1 st centuries BC), hillforts became a fundamental type of habitats in ex tensive parts of the central and w estern parts of the Iberian P eninsula. Today there is a consensus on assigning these settlements a fundamental role in the structuring and ex ploitation of the territory and in the control of communications and ex change routes. It has been assumed that these sites are essential in order to ex plain the development of social complex ity of these communities w ithin this geographical and temporal scope.
H ow ever, the archaeological record of these settlements is overwhelmingly limited to the definition of their w alled enclosures and their defensive architecture, w ith very few ex amples in w hich w e have a minimal idea of their intra-mural spatial organiz ation, let alone the configuration of their domestic spaces. Many conditioning factors have led us to this situation. The most important one is obviously the high cost of large scale ex cavation w ork , as w ell as the problem of preserving the ex humed structures. The maintenance needs of this type of intervention often lead to a management model that is difficult to sustain. Such large and complex archaeological sites therefore req uire action to combine their value as a cultural resource w ith their long-term conservation.
In this contex t, the grow ing introduction of non-invasive methods offers a very promising alternative. The adoption of standards in the use of these procedures has become w idespread and consolidated as a full-fledged discipline (Campana and Piro 2009;Schmidt et al. 2015 ). In recent years there has been a grow ing implementation of this type of research in the Iberian P eninsula. H ow ever, w ith regard to the environment of the case study dealt w ith here, most of the published ex periences correspond to other periods, such as R ecent P rehistory (M á rq uez and J imé nez 2013 ) and especially the R oman w orld (Corsi et al. 2012;Á lvarez M artí nez et al. 2014;M ateos et al. 2014). N evertheless, w ith regard to the settlements from the Second Iron Age there w as an almost total absence of precedents 1 . This w as a great chance that encouraged us to launch in J une 2017 the MINARQ proj ect (Minimal Invasion in Archaeology), the first results of w hich are presented in this paper. O ur main aim is to rely on the potential of non-destructive methods, as a w ay of obtaining k now ledge for the study, protection and valorisation of archaeological sites. The previous link betw een the Institute of Archaeology and the hill- Trab. P rehist.,7  ed to the south of the previous enclosure and is somew hat smaller (2.4 ha). It is also protected by a w alled perimeter reinforced by tow ers that stand out in its Southw est and N ortheast corners, as w ell as by a 10 to 12 m w ide moat that closes off its entire South and West flank until it connects with the gorge of the Tamuj a R iver. It is located in a higher position than A, the maj ority of w hose interior it actually overlook s, and is the only point in the entire settlement from w here it is possible to have medium and long-distance control over the surrounding area. Archaeological ex ploration of V illasviej as dates back to the mid-19 th century, w ith the reactivation of mining activity in the area (G onz á lez Cordero and H eras 2009). H ow ever, systematic research in the hillfort began in the 196 0s, w ith ex cavations carried out by D r. F rancisca H erná ndez . This author w as the main responsible for continuous study of the site over more than 3 0 years. This long activity included the ex cavation in several points of both enclosures (H erná ndez et al. 198 9), the systematic surface survey of the surroundings (H erná ndez et al. 2009), and above all a complete record of three necropolis associated w ith the settlement (H erná ndez and G alá n 1996 ; H erná ndez and M artí n Bravo 2017 ). The ex cavations carried out by the U niversity of E x tremadura should also be mentioned, although they w ere only partially published (O ngil 1991). The last archaeological interventions carried out in 2007 w ere oriented to prepare the site for public visiting w ithin the framew ork of the M aster P lan designed j ointly by the U niversity of E x tremadura, Complutense U niversity of M adrid, and the Institute of Archaeology. All this previous research has provided us w ith ex tensive k now ledge of the site. W e k now that human occupation in V illasviej as can be traced back to at least the 4 th century BC, although it is only over the follow ing three centuries that the ex tensive w alled settlement took shape, and w hich continued to be occupied until the early stages of R oman domination.
Based on our current k now ledge, V illasviej as is considered as a representative case on a regional scale of the forms of occupation during the Second Iron Age. F rom the point of view of a cultural ascription, F ig. 1. M ap show ing the location of the H ill F ort of V illasviej as del Tamuj a (Botij a, Cá ceres) in the Iberian peninsula. Trab. P rehist., 7 6 , N .º 2, j ulio-diciembre 2019, pp. 3 03 -3 22, ISSN : 008 2-5 6 3 8 https:/ / doi.org/ 10.3 98 9/ tp.2019.1223 9 agricultural use has k ept most of the site deforested for centuries, favouring the formation of a large flattened platform. In addition, previous ex cavations revealed the reduced stratigraphic depth of the upper area of E nclosure A and the shallow depth of the w all foundations in the rest of the site.

METHODS
Today, in order to provide a robust diagnosis of buried structures through a geophysical ex ploration, it is imperative to provide the results of multiple methods. E ach one responds to different properties of the elements of the subsoil, and therefore offers distinct and complementary images. H ow ever, their implementation entails very different timescales, w hich conditions the ability to ex plore large areas in a uniform w ay. In the case of V illasviej as, available methods allow ed us to ex plore the subsoil in tw o main dimen-the site has been traditionally framed in the southern limit of the territory of the vettoni. This w as one of the peoples that occupied the w estern area of the Meseta (plateau) of the Iberian P eninsula according to the vision of the R oman conq uerors (Á lvarez Sanchí s 1999; Salinas de F rí as 2001). N evertheless, because of its geographical location, features of other neighboring cultural areas lik e Lusitania or the Baeturia have been identified (Hernández and Galán 1996: 108).
As most of the hillforts in the area, V illasviej as is located at a secondary elevation near a main w atercourse. The aim of this lesser topographical prominence of Second Iron Age sites w as to obtain better conditions of habitability, and w as compensated by the increasing complex ity of the defensive structures. As far as the surface area is concerned, these settlements w ould have typically covered an area of betw een 1 and 2 ha. Therefore, V illasviej as w ould be w ithin a small group of 6 .5 % of the 5 4 hill forts catalogued by M artí n Bravo (1999: 204), but far from the four that are k now n w hich cover 10 ha or more, and even more than the 6 0 ha of U laca (Á lvarez Sanchí s 1999: 120).
It is important, how ever, to bear in mind that the 7 ha of our case study correspond to the total surface of the tw o w alled enclosures. W e are not certain that both were occupied simultaneously before their final stage of development. In fact, there seems to be a tendency on a regional scale for larger enclosures to be associated w ith more advanced historical contex ts, already under R oman domination (M artí n Bravo 1999: 26 5 ). The same phenomenon occurs w ith regard to the complex ity of the defensive systems, something of w hich V illasviej as w ould be a good ex ponent (Berrocal 2005 ). It seems clear that the definitive abandonment of the site occurred during the first decades of the 1 st century BC. H ow ever, a particularly controversial aspect of its historical evolution is that of the nature of its occupation in this final stage. Evidence of a military presence at this time has led to disagreement regarding the degree of continuity and the type of interaction betw een the indigenous population and the R omans. The team that has ex cavated on the site for a longer time defends the idea of a community gradually transformed by an external influence (Hernández 1993; Hernández and M artí n Bravo 2017 : 3 22), w hile other authors have emphasiz ed the sudden transformation involved by the conq uest, speak ing of a true " re-foundation" of the site by the R omans (H eras 2018 ).
In spite of these antecedents, our k now ledge about the urban structure of the hillfort w as logically very limited, since the ex cavations carried out only represent 1.2% of the total surface of both enclosures. H owever, unlik e many other settlements in its surroundings, V illasviej as offered ideal conditions for conducting a non-invasive archaeological programme. P rolonged F ig. 2. H illfort of V illasviej as del Tamuj a (Botij a, Cá ceres). Boundaries of the prospected areas, indicating the different methods used: E lectrical resistivity tomography; G round-penetrating R adar; G eomagnetic survey. Coordinate System: U TM E TR S8 9 H 3 0N . F inally, the processes follow ing the abandonment of the hillfort also had a positive influence on the visibility of the structures. The defensive w alls have acted as a large container for archaeological deposits, and their subsequent division into crop fields has slowed dow n the erosive activity that is so apparent in the w eak soils of the surrounding landscape. At the same time, centuries of tilling have led to the formation of soils w ith a matrix of sand-clay lithology. The action of light ploughs has removed the most superficial levels, so that most of the structures are only 20/ 3 0 cm from the surface, another factor that helps in identifying them clearly (F ig. 3 ).

Geomagnetic survey
The magnetic prospection is by far the method that has allow ed us to ex plore a greater surface of the hillfort. To date w e have covered a total of 3 .1 ha, w hich represents 26 % of the total prospectable surface of the site. This w ork has particularly favoured E nclosure B, w ith 44% of its prospectable area compared to 3 9% of E nclosure A.
The equipment used was a Grad601 fluxgate gradiometer from Bartington (Bartington and Chapman 2004;H immler et al. 2008 ;Schmidt 2008 ). M ost of the time it w as used in grid mode, adapting the grid to the variable dimensions of the land plots. The usual configuration of the equipment has been with an interval of 2 lines/ m and density of 8 samples/ m. W e have ex p erimented in some sectors to compare the results w ith a denser data collection (4 lines/ m). It creates a more nuanced image of magnetic variations, sions. O n the one hand, especially thank s to geomagnetic survey, w e are carrying out a " radiography" that gives us an open area understanding of the structure of the site. O n the other hand, thank s to the G P R and the tomography, w e obtain something very similar to a Computeriz ed Ax ial Tomography Scan, w hich provides us w ith a vertical reading to evaluate the diachrony of the site (F ig. 2).
In order to calibrate the relative effectiveness of each method, it is essential to characteriz e the environment in w hich w e are w ork ing, as it is decisive in the response of the sensors. F rom a geomorphological point of view , the area is w ithin the Truj illo-Cá ceres peneplain (Tena 2010), a result of the prolonged erosion of the materials of the Schist-G reyw ack e Complex . These are mainly slates, w hose resistivity values have already been studied in nearby areas w ith the same substrate (Tena 2008 ). This information has been very useful for the interpretation of the tomography. These slates are affected by a primary schistosity in a mainly N -S direction and angles close to the vertical in the northern z one of the enclosure, constituting natural cliffs that delimit the hillfort. The dip is less pronounced tow ards the south, based on the data observed in the outcrops of the Tamuj a riverbed and slope of the enclosure. The lateral continuity w ould indicate that the layout of the slates w ould have formed the original base of the settlement, forcing a terraced layout adapted to the outcrops.
As for the archaeological contex ts, w e have the information provided by the ex cavations carried out in both enclosures. The predominant building material w as provided by the local geology: slate slabs interlock ed w ith mud, forming masonry plinths betw een 5 0 and 90 cm w ide. O ccasionally, granite block s w ere used to reinforce corners and thresholds. The floors are normally made of flattened earth, although the use of the bedrock itself as a base has also been documented, levelling w ith slate, ceramics and adobes, and slate slabs (O ngil 1991: 25 0; H erná ndez and M artí n Bravo 2017 : 27 2). There are other details to tak e into account for the interpretation of the results of geophysical survey. N earby structures are often adj oined w ith very narrow gaps betw een them instead of sharing party walls. Inside the rooms is usual to find auxiliary elements, lik e benches or ovens, made w ith the same fabrics and materials. In addition, we find superimpositions that coincide in their orientation, creating the false impression of double w alls, reinforcements or subdivisions. As for the w alls, they are mostly made of rammed earth or adobe. This is a very favourable factor for the detection of the structures, as the presence of accumulations of collapsed stone inside the rooms w ould have introduced a considerable amount of noise in the reading of the data. amplify the signal containing significant data. Depending on the area, the velocity of the medium can be q uite variable, but an average of around 10.3 cm/ ns of constant velocity has been assumed for the ground w ith a dielectric coefficient of 9. With the data from the processed G P R sections a series of slices around 10 cm thick w as generated, w ithout any overlap betw een them. F inally, the interpolations generated from the sections w ere ex ported to a G IS w ork ing environment in order to compare this data w ith the other layers of information.

Electrical resistivity tomography (ERT)
E lectrical tomography provides the distribution of the electrical resistivity of subsoil materials, either along a profile (2D) or for a three-dimensional volume (3 D ) (D ahlin 2001). Its implantation in archaeological studies is fully established (Teix idó et al. 2013 ; N ow acz insk i et al. 2015 ; K vamme 2018 ). The eq uipment used in this study is the ABE M Terrameter LS resistivity meter w ith 90 electrodes. The w ork plan carried out in V illasviej as had tw o obj ectives: on the one hand, it w as proposed to create long profiles that crossed both enclosures longitudinally and transversely, w ith the intention of assessing but we have not seen significant differences in the q uality of the information provided.
The data w as processed w ith the Terrasurveyor softw are. Basic corrections w ere made to the errors derived from data collection: stripping, staggering, and occasionally filters to correct noise problems. Once ex ported to a G IS environment, w e w ork ed w ith the range of values of the tables in order to ex plore their distributions and max imiz e the detection of traces of archaeological interest. Thus, from a semi-automatic classification of the average values of magnitude for each type of identified element, we have elaborated an intermediate layer between the raw data and a final interpretative reading that w ould be visually more understandable.
The range of values associated w ith different types of buried features and anomalies has been classified as a series of discrete categories: positive and negative peak s, w all foundations (certain or probable) and thermo-altered surfaces. W ithin the latter group, w e have differentiated betw een possible pavements, burnt and possible combustion structures (kilns, fireplaces). From this reading layer, a visual interpretation has been made, combined w ith the use of contours generated at different magnitude intervals, crossing the three types of information. The result has been vectoriz ed offering an interpretative synthesis of the data obtained by magnetic prospection.

GPR
The G P R (Conyers and G oodman 1997 ;Conyers 2012; G oodman and P iro 2013 ) has been the second most used method in the survey of V illasviej as, w ith a total of 1.4 ha covering ex tensive areas of both enclosures. A total of 3 16 transects have been carried out, providing a total of 3 .47 6 georadar sections of various lengths. The w ork w as carried out at the end of J une, in ex tremely hot, dry conditions.
The eq uipment used consists of a Stream X multichannel antenna (ID S G eoradar -H ex agon G eoystems). The nominal freq uency is 6 00 M H z w ith a disposition of 12 dipoles w ith a polariz ation in H H and V V . In each pass, 11 georadar sections are obtained w ith a spacing of 8 cm for a footprint w idth of 8 8 cm. The horiz ontal distance calibration is determined using the odometer of the eq uipment and a tak ing of points of beginning and end of each transect by means of a differential G P S of sub-centimetre precision. The horiz ontal resolution is around one scan every 4 cm.
The data has been processed w ith the G P R -Slice softw are. The obj ective of the adj ustments has been to minimiz e the abundant back ground noise and to

RESULTS
Thank s to the integrated w ork w ith various geophysical methods, in a short time w e have accumulated a w ealth of information on the internal organiz ation of the settlement. M ak ing effective use of these data is a challenge that first requires evaluating each source separately, and then combining them all together.

Geomagnetic survey
The most notable feature of the geomagnetic survey results is the high contrast betw een the values corresponding to the target structures and the filling deposits betw een them. The former is almost alw ays negative, w hile the latter show a higher degree of magnetisation. As w e already pointed out in section 3 , this good result has been ex tremely favored by the building techniq ue and the absence of collapse deposits of stone and roof tiles. There is also a very limited the stratigraphic deep and the relationship betw een the archaeological deposits and the geological base.
To date, three of these profiles have been carried out, all in E nclosure A (P 1, P 2 and P 3 , see summary of their features in table 1). O n the other hand, pseudo-3 D studies w ere considered, consisting of data collection with a grid composed of 12 profiles spaced 0.5 m, w ith an electrode interval of 0.5 m. This type of register adds a horiz ontal reading to the previous one that mak es it possible to appreciate the layout of the buried structures. U sing this procedure, so far it has only been possible to ex plore an area measuring 20 x 5 .5 m in the central part of E nclosure A (Sq uare C1 in Tab. 1).
The data has been processed w ith the R E S2D IN V and R E S3 D IN V inversion programs from G eotomo Softw are. F inally, georeferencing and integration in G IS environment w as carried out. In the case of pseudo-3 D , w e also assessed the q uality of the resistivity surfaces at various depths using different interpolation methods. There are anomalies close to the surface that present w ave reverberation that is typical of a metallic element, probably related to modern livestock activity. H oriz ontal reflection anomalies have been detected that coincide w ith transects parallel to w alls. In the low er part of this enclosure, an inclined reflection has been detected below the hyperbolas of the w alls, w hich could be due to the low er contact of the soil w ith the rock at the geological level at a depth of betw een 0.6 -1.2 m. It forms a valley some meters w ide, oriented along a northeast-southw est ax is. The deepest anomalies in these sections are related to the internal structures of the metamorphic rock , diaclases and altered internal fractures. In the slices at different depths a small variation of orientation of the structures has been detected, one along a north-south ax is and w alls at about 0.18 m, and another along northw est-southeast ax es and w alls at 0.3 1 m.

Electrical resistivity tomography (ERT)
As it w as described in 3 .3 , w e carried out three long profiles (P1 to P3) in the A Enclosure in order to get an ex tensive transversal reading ot E R values across the site. R esults (F ig. 6 ) show clearly the boundary betw een the geological base and the archaeological deposits, located at a max imum depth of about 3 m in the case of profile P1 and 1 m for P2. The error in the inversion process is very low , being eq ual to 1.9 % for profile P1 and 0.95 % for P2 (Fig. 6).
Profile P1 shows three zones of anomalies, located at distances betw een 9 -16 m, 18 -28 m and 3 8 -6 0 m from the origin of the profile. The first is located betw een 1 -2 m deep from the surface and could correspond to a w all. The second, located at a depth betw een 1 -3 m, is the one w ith the largest dimensions and highest resistivity values. In the magnetic susceptibility map these tw o anomalies are not show n, probably because they are at a greater depth. The third anomaly to highlight in the profile is the most superficial, located at a depth betw een 0.2 -2 m. This feature show s a narrow correlation w ith the values recorded by the magnetometer, and may correspond to w all foundations (F ig. 7 ). Betw een 3 0 -3 8 m, areas w ith low relative resistivity are detected, w here the most notew orthy is the one located below 3 5 m. A relative maximum is also observed at 3 1 m, w hich coincides w ith a max imum of magnetic anomaly. O n the other hand, betw een 48 -6 0 m in length, structures located betw een 0.2 -0.8 m have been detected that seem to have a certain degree of collapse since they present lateral presence of dipole anomalies, w hich generally correspond to modern alterations (w ire fences, residues, pieces of agricultural machinery, etc.), thermo altered areas or the presence of buried metallic elements. In general, the range of values is q uite limited and w ith little back ground noise. This favours the clear detection of elements of less magnetic susceptibility, w hich w ould otherw ise be mask ed by the ex treme values. D istortions caused by the presence of bedrock outcrops can only be seen at some points in the higher z one. O n the contrary, as w e descend tow ards the low er points of the prospected z one, the stratigraphic pow er is greater and this type of anomalies are attenuated. The terraced arrangement of the boundaries of the landplots has favoured this process (F ig. 4).

GPR
As a w hole, the result of the G P R survey returns an image of the buried structures w ith abundant noise, which makes their interpretation difficult. The dielectric coefficient of the walls is estimated to be between 7-10 mS/ n w ith a mean propagation velocity of 9.5 -11 cm/ ns, and that of the soils is estimated at 6 -13 mS/ ns and a propagation velocity around 8 cm/ ns. Therefore, the contrast betw een the tw o media is relatively low and in some specific cases can be similar, although sufficient to obtain reflected energy. With the antenna used, a critical z one around 4 cm is estimated, meaning that w alls w ith a w ider w idth can be detected. The investigation depth w as around 1.1 m ex cept in areas of metamorphic rock , w here anomalies up to 2.3 m w ere detected.
In E nclosure A there is a maj or contrast betw een the hyperbolas 2 produced by the buried w alls and the soil surrounding them. The signal that w e identify as collapses of structures is very localiz ed, and there are few stones or soil disturbances. The w all anomalies are detected from an estimated depth of 0.26 m to 0.98 m. The low er contact of the soil w ith the geological level has not been detected ex cept in the southern part, at a low er topographical level, w here the hyperbolas could be related to the internal structures of the metamorphic rock (F ig. 5 ).
Enclosure B is where we find a greater background noise, w ith a large number of isolated hyperbolas that indicate the presence of areas w ith landslides and removals, mask ing the detection of possible structures. O n the other hand, the layout of the w all on the east side of the enclosure is clearly visible, forming a group of hyperbolas w ith large dimensions in both w idth and variations. Comparing it w ith the map of magnetic anomalies, it can be deduced that they may be the remains of w alls combined w ith a surface that w as used for some type of purpose. Betw een approx imately 3 8 -48 m, high resistivities are detected that are possibly associated w ith thick er stone masonry structures than in the previous case. Their roof is betw een 0.2 -0.5 m and increases w ith the slope, w hile the base reaches 2 m in depth, and also increases w ith the slope.
In profile P2, the layer of archaeological interest is shallower than in the previous profile. The first anomaly to be highlighted is betw een 16 -44 m from the origin of the profile, highlighting a body centered on the most ex cavated 3 2 m in the geological base, w hich could correspond to some type of infrastructure. A second anomaly is located betw een 48 -5 5 m and is the one that presents the highest resistivity values of the entire profile, which may correspond to a wall. Betw een 6 5 -8 9 m there are heterogeneous resistivities that seem to be grouped in a surface and conductor section corresponding to filler material, with two anomalies standing out that are centered on the 7 3 m and 8 5 m, some 2 m long and around 0.5 m thick .
The P3 profile was traced from south to north along almost the entire length of enclosure A (200 m long). Its purpose w as to evaluate the distribution of the archaeological level and, above all, to study the geological characteristics of the area at a greater depth. Therefore, the distance betw een electrodes is 1 m, reaching a max imum depth of about 12 m. F igure 6 show s the result of the inversion, with a final error of 0.99 %. A dotted line indicates the limit of the layer of archaeological interest, w hich is at a max imum depth of about 3 m. Between 13 -30 m from the origin of the profile, anomalies similar to those of the P1 profile (located about 40 m to the w est) betw een 18 -28 m can be observed. These could correspond to the granite w all. This coincidence in the anomaly of very high resistivity in identical sections of the two profiles, indicates that the data are reliable and valid to interpret the deepest levels ex plored. Below the archaeological level, the layer corresponding to the substrate of slate  numerous correspondences that reinforce the validity of the hypotheses proposed regarding the presence of buried elements. This can be clearly seen in the urban pattern of E nclosure A, w here the high values of electrical resistivity coincide w ith the low magnetic susceptibility bands corresponding to the structures and w ith the anomalies registered by the G P R . An illustrative ex ample of this correspondence can be seen in the front of the defensive w all at the southern side of E nclosure A. In this area, the layout of the ramparts in their W and E corners w as k now n, but the accumulation of plough soil from agricultural w ork had mask ed the presence of the w all and the location of the main entrance to the enclosure. The magnetometry (F ig. 8 A) made it possible to trace the line of the w all throughout this section, also revealing the presence of an open space flanked by a 6.7 x 6.3 m structure w ith w alls 1.5 m thick , w hich w e interpreted as a stronghold defending the entrance. O n the other side of this access, the magnetometry image w as confused, and it w as suggested that there w as a recess in the rampart to create a funnel-shaped entrance. H owever, the data provided by the GPR and profile 1 of the tomography (F ig. 6 P 1; F ig. 8 B and 8 C) recorded a strong anomaly j ust 5 m w est of the entrance, w hich w ould indicate the presence of another built mass. This w ould surely indicate the ex istence of a second tow er materials of the Schist-G reyw ack e Complex can be differentiated, and below this unit, from 7 .5 m deep, highly resistive materials appear, w ith dome morphology and resistivity values that gradually increase tow ards the nucleus. W e propose the hypothesis that this discordant presence could correspond to a granitic intrusion that w ould provok e a contact aureole, determinant of the gradual increase of resistivity values.
R egarding the results of 3 D tomography, although the ex ploration covered a very small area, a high potential for the future is evident, as it provides interesting data for the q uantitative comparison of the results provided by different sensors. As may be seen in F igure 7 , there is a high contrast that clearly differentiates structures w ith higher resistivity values. The correspondence w ith magnetic anomalies is very close. The tomography study allow s us to situate the max imum depth of the latter at 1.7 m.

General remarks
The j oint analysis of the large volume of data obtained w ith different sensors reveals the ex istence of In E nclosure A, the largest one, is evident a greater organicity and adaptation to the topography. O pen spaces are scarce. It is possible to distinguish a series of large longitudinal axes adapted to the shape defined by the w alled perimeter that organiz e the circulation in an N -S direction. The principal one, outstanding for its w idth (up to 6 m) and rectilinear layout, divides the enclosure in tw o halves. This main street runs parallel to the slope, and seems to connect directly w ith the main entrance to the enclosure, as w e saw in the previous section. We can affirm that urbanistically this formed the back bone of the entire complex . A secondary ax is runs along the w estern side of the settlement. As we will see for the B Enclosure, we never find circulation areas betw een the defensive w all and the building block s, as the ex istence of constructions attached to the inner face of the rampart has been confirmed in several points. We do not know whether the eastern side w ould have been symmetrically organiz ed in the same w ay. In this area, the geophysical readings w ere hindered by a series of terraces created by the segmentation of agricultural land, although it should be noted that these may be the fossilised remains of the original stepped design of the hill fort.
Based on these longitudinal ax es, a series of secondary streets run in a general E -W orientation, but do not transverse them at right angles. Their w idths vary betw een 3 .5 -4 m, forming block s of 11 or 12 m in w idth. This information is consistent w ith the flanking the main door, and may be offering us clues about modifications made to the southern face of the defensive w all.
At the same time, this variety of methods provides us w ith new information w here only one or more sensors are capable of recording buried features. A good ex ample of this is the information provided by the G P R in E nclosure B. O n the one hand, it has made it possible to delimit a good section of the inner face of the rampart in an area w here its layout w as more dubious. O n the other hand, thank s to magnetometry w e w ere considering the idea that at least 1/ 3 of the enclosure lack ed a dense framew ork of structures as typically found in A. As w e have seen, in this sector the GPR defined a shallow depression oriented from the SW -N E that w ould correspond to an open space, probably a passagew ay, w hich w e can therefore characterise three-dimensionally.

The internal structure of the walled enclosures
Beyond these partial valuations, the comparative w ork using the sensors mak es it possible to create an overall image of the internal organiz ation of the settlement. In this sense, the tw o enclosures have both similarities and differences. F ig. 9. P ossible Italic-type structure in E nclosure B of the hillfort of V illasviej as del Tamuj a (Botij a, Cá ceres): A O riginal data; B R eading and interpretation. D r. F rancisca H erná ndez ) or by the adj oining of w alls that ended up defining the street layout, something that is especially apparent in the central and highest area of the enclosure.
A separate issue is the diachrony that the ex cavations have revealed in the evolution of the defensive system. It has been suggested that some attachments and superimpositions to the rampart w ould denote a loss of its functional and symbolic role, to the point of proposing a " refounding" of the hillfort after the R oman conq uest (H eras 2018 : 46 5 ). This matter req uires a detailed analysis that goes beyond the limits of this w ork . W e w ill only note that geomagnetic survey has clearly revealed a main entrance to the enclosure that is w ell articulated w ith the rest of the urban layout.
As for E nclosure B, w e w ill focus on the characteristics of three isolated buildings located at its N and S sides. The first (Fig. 9) forms a SW-NE oriented q uadrangular block measuring approx imately 23 x 23 m, formed by three rectangular sections around 8 m w ide, subdivided into several elongated rooms betw een 3 .7 -5 m w ide, w hich are arranged around a central space that seems to act as a distributing corridor. This regular and symmetrical arrangement seems to have been devised in a unitary w ay. H ere w e propose its similarity w ith the Italic architectural model of the Late R epublican tripartite houses and granaries pointed out by Moret (1999: 64-65) for the "fortified houses" and castella from that period in the Iberian P eninsula. The presence of several fragments of tegulae on the surface could be indicative of the type of roof used for these buildings.
The other tw o buildings are located in the southw est end of the enclosure, in an area w here the geophysics reveals the presence of a less dense urban layout, w ith block s of structures and streets that are perfectly visible (F ig. 10).
Both groups form compact block s , w ith a total w idth of about 10 m, divided into 2 batteries of sq uare rooms w ith dimensions of 4.5 x 4.5 m in one case and betw een 5 .2 -5 .8 m each side in the other. They have a SW -N E orientation and are placed in parallel, w ith a straight street w ith a w idth of about 4 m betw een the tw o. F urther w est, a third body w ith very similar characteristics w ould have been attached to the inner face of the defensive w all. O nce again, the ex treme regularity in the layout of the floor plan and the homogeneity in the dimensions of the spaces invite us to consider them as non-domestic structures w ith a specialised function. In this case, the search for analogous ex amples leads us to review the ex cavations carried out by Schulten in the N umantine R oman camps, in particular the structures corresponding to phase III of R enieblas (D obson 2008 ). Although w ith dimensions of the block of houses ex cavated by D r. Francisca Hernández, which would have been flanked by tw o of these roads. In several points it can be seen how these block s are structured forming tw o alignments of houses facing tw o streets. In any event, despite the good contrast, w e cannot venture much more regarding the internal organiz ation of the houses. This repetitive scheme seems to break up in some areas, forming alleys that lead to small sq uares. This suggests the presence of more ex tensive and compartmentaliz ed dw elling units beyond the basic module of some 3 0 m 2 defined in the hillfort itself (Hernández et al. 198 9: 15 7 ) In the case of E nclosure B, although the entire southern half contains a dense urban layout, w e can see a large, unstructured space in the area nex t to the northern boundary of the w alled perimeter. N o maj or roads have been found, and all of the streets are narrow er (betw een 3 and 4 m). Although their design is also very straight, there is more diversity in their siz e and layout. Here, too, we repeatedly find structures attached to the entire inner face of the w all. It has also been possible to identify one of the entrances of the enclosure, corresponding to the N orthw est angle. Although the ex istence of a tow er at that point w as already known, we can now define in detail a large trapez oidal defensive structure w ith an internal compartmentation that allow s us to hypothesiz e that it had an upper room. This bastion would have flanked a gate of 7 m w ide. All these data lead us to propose that, w hile E nclosure A w as the result of the evolution of a previous urban plan, E nclosure B w as formed from the aggregation of independent structures. Some of them, due to their regularity and design, suggest an ex novo implementation, as discussed in the follow ing section (F ig. 9).

Temporal evolution of the hillfort
In the few areas that have been ex cavated, the superimposition of structures clearly indicates that the urban layout w e have documented correspond to the final stage in the life of the settlement, dated to the beginning of the 1 st century BC. In E nclosure A, the compact and regular distribution indicates a high level of urban planning. H ow ever, in its south-eastern side w e can see narrow er and more irregular streets, and more heterogeneous dw elling units. As a w hole, it could be suggested that the urban structure of the 4 th -3 rd centuries B.C w ould have been partially maintained, preserving its orientation. The evolution of the street netw ork and domestic structures came about either as a result of superimposition (documented in areas ex cavated by I. O ngil and, to a lesser ex tent, by Trab. P rehist.,7  has been in the research agenda since the late 198 0s (Almagro 1994: 13 ). H ow ever, the empirical basis from w hich this debate has developed has barely expanded in all this time.
This fact is especially strik ing w ith regard to the immediate contex t of our case study: the N orthen part of the region of E x tremadura. Although the defensive enclosures are relatively w ell k now n, w ith the ex ception of La Coraj a (Aldeacentenera, Cá ceres) no significant ex cavations have been carried out in the area, so w e still lack any information about the internal organiz ation of these hillforts (Almagro 1994: 16 ). W ithout leaving the cultural sphere of the V ettones, the situation is slightly better in the case of E l R aso de Candeleda (F erná ndez G ó mez 198 6 , 2011), w hich is today the hillfort w ith the largest ex cavated area in the region. U nlik e V illasviej as, its adaptation to a rough topography dominated by granite outcrops determines the w ay in w hich the houses are grouped together.
We seem to find a similar arrangement in the case of U laca, in Avila, w here, despite the small area that has been ex cavated, there are signs that w ould suggest a spatial differentiation of various activities and the basic articulation of the ax es of circulation. The model of spatial organiz ation in this large settlement differs considerably from that observed in V illasviej as, as the dw ellings are not grouped into compact block s, but are instead scattered w ithout a uniform orientation w ith well-defined public spaces. In the case of the Cogotas hillfort, as in V illasviej as, a series of houses attached to the interior face of the w all have been documented, forming compact clusters. H ow ever, the presence of large granite outcrops inside the enclosure w ould have made it impossible to develop a regular urban layout, as in the case of U laca (Á lvarez Sanchí s et al. 2008 ) or E l R aso. O n the other hand, there is a similarity with our case as regards the identification of spaces for rubbish dumps, spaces used for w ork ing, and other structures outside the w alls, as can be seen from the area around E nclosure A (H erná ndez et al. 198 9: 6 6 ). Another of the most outstanding cases in the area of Avila is La M esa de M iranda (G onz á lez -Tablas 2008 ; Ló pez G arcí a 2012). Thank s to recent w ork w e have a more detailed record of the structure of the houses. H ow ever, and despite the fact that here the topography w ould be more propitious for creating a w ell-organised urban scheme, as is the case in V illasviej as, w e lack any information that could help to corroborate this hypothesis (F ig. 11).
If w e move tow ards the northern limits of the V ettones region, the situation is similar to that in E x tremadura. W e have some recent ex cavations in several hillforts (Benet and Ló pez J imé nez 2008 ) but, as a w hole, there is still a total lack of k now ledge regarding th e urban layout and the organisation of domestic space. less clarity, w e can also see similarities in the structures of the much closer case of the R oman camp of Cá ceres el V iej o, w hose chronology coincides w ith that of the last occupation of V illasviej as (Abá solo et al. 2008 ). W e therefore propose the hypothesis that they could have been the barrack s of a permanent military encampment. Although w e do not yet have a direct confirmation through excavation, just 20 m from these traces w e have one of the test pits ex cavated by D r. F rancisca H erná ndez , w here the material recovered points tow ards a level of occupation dating from the beginning of the 1 st century BC (H erná ndez et al. 198 9: 26 et seq.) W e can therefore see in this E nclosure B some elements that w ould point tow ards the presence of Italic structures, possibly of a military nature, during the Late Republican period. This is not the first time that this idea has been raised (E steban and Sá nchez 198 8 ), an aspect that needs to be ex plored in detail elsew here. The presence of imported materials in the tw o parts of the settlement is overw helming. In addition, the funerary record of the necropolis of R omaz al I has provided an important collection of grave goods w ith w eapons and other elements that strongly point towards an Italic presence in the final stage of the life of this settlement. The discussion about the identification of V illasviej as as the coining center of the mint of Tamusia should not be ignored in this sense (Blá z q uez Cerrato 1995 , 2014). It' s purely Celtiberian character has given rise to the hypothesis of the presence of contingents from the eastern M eseta displaced by the R omans w ithin the framew ork of the armed conflicts of this period (Sánchez Moreno 2017: 68-69). Although all the k now n coins have a clear geographical concentration in the regional environment of V illasviej as, it is no less true that none of them comes from ex cavation contex ts in the site itself. The q uestion that w ill have to be evaluated is the type of interaction betw een the local inhabitants, Celtiberians and the R omans that is revealed by this complex palimpsest.

Villasviejas within the context of the protohistoric urban development of the Peninsular Southwest
F inally, it is w orth assessing to w hat ex tent the k now ledge gained in the case of V illasviej as w ill help to improve our k now ledge of the urbanism of Iron Age fortified settlements in the western part of Spain's Meseta. A common element throughout this w ide area from the period in q uestion is the w idespread appearance of new settlements, w hich has been interpreted as evidence of the ex pansion of a new socioeconomic model (Á lvarez Sanchí s 2003 : 106 -111). This q uestion Trab. P rehist.,7  Its location and adaptation to the topography is similar to that of the A E nclosure of V illasviej as, w ith a series of longitudinal ax es that start from an access defended by an " entrance fortress" . H ow ever, in this case the perpendicular roads are mere alleys, compared to the w ider ones in V illasviej as. O n the other hand, the excavation has provided indications that the street network was modified after the Roman conquest, blocking off sq uares and densifying the built space. Another settlement w ith occupation after the R oman conq uest is the oppidum of H ornachuelos, in R ibera del F resno (Badaj oz ). E x c avations w ithin the w alled perimeter have revealed the layout of tw o parallel streets, adapted to the topography and framing regular block s of adj oining rectangular houses (R odrí guez D í az 1991). F inally, w e have an eq ually incomplete understanding of the internal organiz ation of Iron Age settlements in P ortugal, especially in the centre of the country. Although w e have evidence of sites that are chronologically similar to V illasviej as, such as Segovia (Bargã o 2017 ) or Cabeç o de V aiamonte (F abiã o 1996 ) As far as the southern submeseta is concerned, especially in the Tagus river basin, few sites are k now n to any great ex tent. In the case of Cerro de la G avia, in M adrid, it w as possible to ex cavate a street w ith a curved outline, w hich crossed the village from north to south, starting from a possible entrance. An urban plan of compact block s and tw o main ax es adapted to the topography was identified (Morín et al. 2007 ). In turn, the ex cavations and geophysical surveys carried out in the village of E l Llano de la H orca (Santorcaz , M adrid) have revealed a more complex layout, organiz ed around a central sq uare that w as accessed from the entrance to the village and a series of radial streets that spread out from it (Contreras et al. 2014). D espite having been ex plored to a lesser ex tent, the case of Cerro de la M esa (Alcolea de Taj o, Toledo) offers the image organiz ed around several main ax es 3 .
Tow ards the South, w ithin the pre-R oman Beturia, there are eq ually few references for a comparison. In the province of Badaj oz , one of the few cases w ith open-area ex cavations is the hillfort of Los Castillej os and defensive structures. W e now also have a general view of its urban layout, with the identification of some of the main street link s. The mark ed regularity observed in both enclosures, w hich denotes a dense and orderly pattern, is strik ing in this respect. This is especially evident in E nclosure B, w here w e have been able to individualiz e complete structures that clearly correspond to foreign Italic models. The validity of these conclusions can be reinforced w ith the available evidence of previous ex cavations, although obviously they raise new q uestions that w ill req uire a feedback with the accomplishment of new field works. In any case, thank s to the k now ledge obtained, the latter can now be oriented in a much more selective w ay, w ith the conseq uent saving of resources. In this w ay w e have begun to define the guidelines for future action plans on the site, w hich also offers new resources for presentation to the public.
In a broader sense, the results from V illasviej as complete our still limited k now ledge of such a complex and debated issue as the development and evolution of Iron Age settlements, especially in their final stages. Although as far as the current territory of E x tremadura is concerned it is the only case ex plored ex tensively today, it is ultimately a model of practice ex portable to other large protohistoric settlements and in general to complex archaeological z ones of all k inds.
W ith a view to the immediate future, it w ill be necessary to complete geophysical ex ploration of the w hole complex and its surroundings. The combination of different methods w ill mak e it possible to deepen the q uantitative relationships betw een them in order to produce a more consistent and reliable diagnosis. F inally, the k now ledge w e have gained must serve to provide answ ers to q uestions about the social and economic structure of this community in its historical contex t.
In this sense, the opportunity arises to revise the previously proposed demographic estimates (H erná ndez et al. 198 9: 15 5 ). Similarly, such a clear record of the urban structure could serve as the basis for a q uantitative approach to the social logic of space through its syntactic analysis (Benech 2007 ). F inally, these considerations can add elements for the discussion about the role of V illasviej as in the organiz ation of its surrounding territory. The eq uivalent of the term " hillfort" in the Spanish bibliography is castro, that usually is opposed to oppidum by the connotations of this last one as a nucleus w ith functions of greater importance (as an administrative and/ or religious nucleus, center of services, concentration of surpluses . ..). The more intensive survey of the environment and a dialogue betw een the non-invasive ex ploration and the results of the ex cavations w ill help to resolve this issue.
there is no data on the distribution of streets or houses. In the south of the Alentej o region, the site of M esas do Castelinho offers more information about its urban layout. The record of its E nclosure B is particularly of note, revealing the ex tensive restructuring of the space, dated by its ex cavators to betw een the end of the 2 nd century and the beginning of the 1 st century BC (G uerra and F abiã o 2010: 47 2). The straight layout of the roads, organising a dense netw ork of constructions, offers a close resemblance to the urban structure that we defined in Enclosure A of Villasviejas. In conclusion, in the light of the cases that have been evaluated, w e can contex tualiz e the urban design detected in V illasviej as del Tamuj a in tw o different scenarios. In the first place, Enclosure A offers the image of a housing pattern that evolved over time, becoming denser and more complex , w ith far-reaching reforms such as the variation of the front of the ramparts in the space delimited by the defensive ditches. Following the classification proposed by Álvarez Sanchí s and R uiz Z apatero (2001), this structure w ould fit with the central street model documented especially in the Celtiberian area (E astern Meseta), w ith compact block s and a limited public space. This is the pattern w e have observed in the cases of Alentej o (M esas do Castelinho), the M iddle Tagus area or in Southern E x tremadura. This spatial organisation differs significantly from the layout exemplified by cases such as U laca or E l R aso, w here the constructions are scattered adapted to a rugged topography (Á lvarez Sanchí s 2011). W e do not k now if this model w as present in the original conception of the occupied space of V illasviej as, and it should be stressed that the image of this urban landscape that w e perceive so clearly corresponds to the final stage of the settlement, which as w e saw can be dated to the late R epublican period.
Secondly, as already noted in section 5 .4, the urban design of E nclosure B seems to respond to this late chronological contex t. Its logic of space does not seem to obey to previous cultural traditions, but rather to the need for infrastructures w ith a functionality beyond the domestic sphere. This leads us to propose that it may have been a R oman military base that w as " embedded" in the indigenous community. H ow ever, w e must evaluate more carefully this hypothesis elsew here.

CONCLUSIONS
The combination of an ex tensive battery of non-invasive methods, together w ith the ideal conditions of V illasviej as for their use, have made possible to answ er essential q uestions about the layout of the w alled perimeter, obtaining new information about its accesses