In the second half of the 20^th century, domestic architecture began being used by both anthropologists and ethnologists in their study of the social relationships amongst pre-industrial communities. In this context, Claude Lévi-Strauss (1987), based on kinship studies, developed the concept of “sociétés à maison”, in which the house was at the core centre of the social organisation. This same interpretation was later adapted to archaeology through the pioneering work of Kent Flannery (Flannery and Winter 1976). To the contribution of those authors were added the studies of Amos Rapoport (1969, 1990) and those of sociologist Pierre Bourdieu (1977). In another study (1970), focusing on the architecture and distribution of objects around the traditional Berber house in northern Algeria (Cabília), developed the idea of habitus as a way to express the collective behaviour that existed between the everyday life and the trivial. Under these circumstances, the domestic routines of pre-industrial societies came to be considered an expression of social attitudes. In this same way, the study of the traditional housing unit ended up becoming a powerful analytical tool in the overall understanding of communities’ social structures.

In the 1970s, American archaeology, in close articulation with anthropology, adopted anthropological and sociological analyses that were systematically applied to the study of the domestic organisation of ancient societies in the American continent. In some cases, as for example the Mayans of Yucatán, Guatemala and Belize, there were still contemporary settlements built by the descendants of the ancient Mayans, which, in turn, allowed comparisons to be made between ancient and present-day domestic habitus. If Archaeology centred around the results of excavations, Ethnohistory was concerned with comparison analyses. The work of Richard Wilk and William Rathje in 1982, entitled “household archaeology”, summarised with great clarity the steps that followed the development of a real “prehistory of domestic life”.

At the end of the 20^th century, the first anthropological studies on houses and domestic structures emerged, and focused around the pre-industrial societies of all regions (Blanton 1994). The use of diffused diagrams to study the connectivity existing between the spaces of a house and their respective chronologies, including prehistory, was carried out (Hendon 1996). Researchers analysed the distribution of objects recovered from excavations, aimed at then interpreting the domestic units. In this same period, the study of the classical world’s domestic life started including an anthropological approach from authors as L. Nevett (1992), who carried out an analysis of the Hellenistic houses, also applied to the Roman world. In reality, the new interpretative paradigm would later be adopted to explore the rich archaeological documentation accumulated in over 200 years of excavations conducted in Pompeii and Herculaneum. This can finally be explored via alternative and complementary methodological lines expressed in, for example, the ground-breaking study of objects and artefacts developed by P. M. Allison (1992, 1997), or in the study of architecture and decoration of houses conducted by A. Wallace-Hadrill (1994) and R. Laurence (1994, 1997). Due to its remarkable state of conservation, the architecture of the houses in Pompeii and Herculaneum came to represent a fundamental interpretive context for analysing the social life of the Roman society, and consequentially inspired wider architectural studies conducted in other provincial towns.

This is the base we aim to follow when analysing a group of excavated houses included in a residential neighbourhood of the Roman city of Tongobriga, located in northern Portugal, and performing a city that was part of the Bracaraugustanus conventus, in the Tarraconensis province. To this end, a spatial and architectural assessment of the buildings was carried out, with a view to interpret it from an anthropological standpoint and better understand the new social ways that developed among the indigenous populations inhabiting the northwest Iberian Peninsula, and who came to adopt the housing models of italic origin.

Based on available archaeological data, it was possible to reconstruct the floor plan and the second floor of Tongobriga’s Roman houses. Their morphology was influenced by both external and internal access features, and therefore the articulating elements in the houses were valued. An emphasis was given towards open spaces, such as atria, peristyla and courtyards, with a morphology that shaped the distribution of the remaining areas and influenced the location of staircases connecting to the upper rooms. Equally relevant was the relationship existing between the house and the outside world, in addition to the location of streets and of other circulation elements, such as porches. Taking advantage of data concerning the Roman living space of Tongobriga, it was possible to improve our understanding on the features of Roman residential architecture in the northwest territory of the Iberian Peninsula and to assess its wide compositional versatility that questions the existence of imported and faithfully recreated typologies and models (Magalhães 2019). The conducted analysis and the comparison of the new Roman constructions with the “Castreja culture” houses from the late Iron Age, as for example the well-known indigenous settlement of Citânia de Briteiros (Magalhães 2019), shed some light on the habitus concept put forward by Bordieu (1977), and stressed its significance as an indicator of both cultural and social change occurring between the end of the 1^st century BC and the middle of the 2^nd century, more specifically between the reign of Augustus and the end of Hadrian’s era.

The Roman victory over the heterogeneous territories integrated into the Roman province of Hispania Tarraconensis took place between the end of the 3^rd century and the end of the 1^st BC. It coincided with the gradual annexation of the regions bordering the western and northern part of the Iberian Peninsula. The later integration of the north-western region in the Empire deserves some attention. We are aware that it was finalised in the Augustan era, at the end of the Cantabrian Wars, from the year 19 BC onwards. However, the area corresponding to the territory stretching between the Douro and the Minho rivers entered the orbit of the Roman world in 137 BC, following on the expedition of Decimus Iunius Brutus (Tranoy 1981; Martins et al. 2018). Still, the integration of the north-western part of the Iberian Peninsula into the Roman administration was intimately associated with the administrative reorganisation of Hispania undertaken by Augustus, and with the creation of new provinces and cities gifted with both social and political structures that underlined the consolidation of the Roman presence in the region and facilitated the integration of the indigenous populations living in this heterogeneous land (Villanueva 2016). The region was a polymorphic ethnic, cultural and socioeconomic reality; it included areas that were much closer to the Mediterranean cultural inflows, as for example the southernmost region, and others that remained marginal to these influences, located more towards the north and the inner lands (González 2006-2007).

In order to understand the nature of the cultural and social changes that took place within the indigenous groups throughout the 1^st century, as a result of their integration into the Roman administrative and economic system, it was important to analyse the specificity of the different regions in the Iberian Peninsula, which offered different realities in light of archaeological data provided by research (Fig. 1).

One of the best-known north-western Iberian sub-regions corresponded to the area between the Minho and Douro rivers, which has been the object of several studies, either in terms of settlement archaeology or in its analyses of specific cities. Among these, the works focused on the urbanism and architecture of Bracara Augusta should be mentioned. This is a city founded in the time of Augustus, around the year 15 BC, and it was the capital of the bracaraugustanus conventus (Martins and Carvalho 2017). There was also archaeological and epigraphic data related to the city of Aqua Flaviae. It acquired great prominence in the Trajan period, when the bridge over the Tâmega river was built in the year 104, under the initiative of the Aquaflavians, as it can be interpreted from the text engraved in a column recovered from the river in 1980 (Ribeiro 2010: 65-66). The city was located on the axis of the Roman road XVII connecting Bracara Augusta to Asturica Augusta, and probably functioned as the regional capital of a group of people also mentioned in this same column, alongside Padrão dos Povos that can also be found at the bridge (Martins et al. 2005; Carvalho 2008). Another Roman city located in the bracaraugustanus conventus that has been the object of extensive excavation works is Tongobriga, which evolved from an indigenous settlement (Dias 1997). Despite the scarcity of available data related to the old pre-Roman settlement, we know that the original houses had a circular plan that still exists today under part of the excavated houses assembling the small Roman housing group on site. The identified pre-Roman dwellings seem to be organised around residential sets, a common feature also present in the wide settlements from the second Iron Age period in the western part of the bracaraugustan Gallaecian region. Another common feature of the wide oppida-type settlements was their large size, which should have had about 13 hectares, an estimated dimension interpreted from aerial photography and from the layout of the walled perimeter, defined by a fortification acknowledged from the segments that displayed the typical features of the north-western castros in the Iberian Peninsula (González 2006-2007). Adding to this, the existence of an indigenous bathing structure, of the ‘pedra formosa’ type, located next to the southwestern part of the wall, was another oppida common feature in the Bracaraugustan Gallaecian region. They were mainly built after the Decimus Junius Brutus military campaign, between 138/136 BC. Adding to this, they have been associated with the reorganisation and hierarchisation of the population, and some of these settlements can be considered as true indigenous towns due to their political, economic and religious functions, alongside the centrality they had in regional terms. They normally occupied prominent positions within the landscape, and exhibited an optimal visual control over the surrounding territory. We believe that the pre-Roman settlement of Tongobriga corresponded to an oppidum, which would have had circulation axes that organised the neighbourhoods and the domestic units in them. They were bounded by walls and had several buildings unified by paved yards, which was a type of organisation in line with these types of sites. It seemed therefore acceptable that the Roman occupation gave the site’s pre-Roman nucleus a natural continuity, enabling it to inherit the singular centrality already in place. This type of settlement was also prone to house ritual and religious activities, which saw the seasonal concentration of regional populations. The “Castreja culture” bathing structures, which have only ever been identified in large oppida-type sites, as in Briteiros or Sanfins, likely provided the architectural contexts to perform some of these rituals (Silva and Machado 2007; Lemos et al. 2008).

Under these circumstances, the Roman city of Tongobriga, seems to result from the evolution of an indigenous cluster. It saw its population being gradually acculturated, whilst adopting new residential structures, alongside new daily habits that included the use of new spaces, as for example the bathing structures, in addition to being introduced to new religious and funeral rituals that the indigenous communities ended up adopting. This adaptation process took place between the Augustan period and the Flavian dynasty, at a time that was marked by the Vespasian’s attribution of Latin law to the inhabitants of Hispania, a circumstance that allowed the reinforcement of the process of cultural integration within the north-west Iberian region. It also favoured the promotion of some clusters of indigenous origin. A note should also be given to the development of several urban centres located in strategic areas from an economic and transportation point-of-view, some of which came to know relevant urban and architectural programs that documented the evolution of the territories and the economic and cultural changes operated in them (Martins et al. 2005). This seems to have been the case with Tongobriga, where the well-known Roman spaces and buildings generally dated from the end of the 1^st century, although it is admissible the existence of Roman-type buildings preceding the monumentalization of the space. This process included the construction of a public bathing structure and a large square, believed to have been the forum (Rocha et al. 2014) (Fig. 2).

Located on the southern border of the bracaraugustanus conventus, the urban settlement of Tongobriga inherited the pre-Latin name of the Iron Age indigenous site. It also seemed to have fulfilled geostrategic functions in terms of playing a role in organising the territory, being also possible that its prominent regional status arose from the centrality that the site would have already had in pre-Roman times. It was associated with the control over two important peninsular routes, one of them being the Douro river and the other the road connecting Bracara Augusta to Augusta Emerita, although this road was just a set of independent pathways covering a very rugged territory (Carvalho 2008).

Little is known about the evolution of Tongobriga until the end of the 1^st century. It is plausible to assume that some structures of Roman origin may have been gradually introduced. At the same time, the centrality of the populational nucleus, deriving from its privileged location, likely attracted both people and services. This justified its continued populational growth and was the reason behind the construction of a public baths building in the 2^nd century, at the exact site where the previous bathhouse was located. A large square was also attached to it (Dias 1997). Some of the original Roman structures may have been camouflaged by the monumental character given to the space during the Antonine era, as it happened to the known housing areas that underwent a profound reorganisation with the installation of italic type residences. These adapted in a more or less organic way to the rugged topography of the land, and therefore replaced the circular houses of indigenous tradition.

The occupation of the Roman Tongobriga was hypothetically discontinued at the end of the 7^th century. The city was referred to as Tongobria in the so-called ‘Parochiale Suevorum’, dating from the second half of the 6^th century, a manuscript also known as Divisio Theodemiri. This performs a remarkable historiographic document that accounts for the organisation of a post-Roman Galicia. All 134 parishes were identified and grouped around 13 dioceses (David 1947). A building from the end of the 4^th century/beginning of the 5^th, with floors covered in polychrome mosaics, and identified in the churchyard excavations conducted at the church of Santa Maria do Freixo, was adapted to cult practices between the 5^th/6^th centuries, and seemed to correspond to a paleo-Christian basilica, perhaps the ecclesia of the Swabian parish (Rocha et al. 2014). The Roman and late-antique settlement of Tongobriga, presumably abandoned at the end of the 7^th century, was succeeded by a Christian parish at the end of the 9^th century/beginning of the 10^th, known as Santa Maria do Freixo, which is still the name of the present-day village.

In 1882, a votive altar dedicated to Genio Tongobricensium (CIL II 5564) was discovered and that made it possible to establish a relationship between the Freixo site and the Swabian parish from the end of the 6^th century. This was the name that was then given to the place where, since the 18^th century, several ruins have gradually been uncovered. However, archaeological excavations at the Tongobriga site only really began in 1980, in a place known as ‘Capela dos Mouros’ (Moorish Chapel), where the Roman baths were to be identified. Since then, several archaeological seasons have spread over a wide area with around three dozen hectares, which has then allowed the delimitation of the archaeological site. It includes the proto-historical settlement and an area with structures with Roman chronology, known as the civic centre and consisting of public baths and a large rectangular square. To the south of this last set there is a necropolis area. All archaeological works on site have been developed as part of a permanent research project, which is today under the radar of the Northern Regional Directorate for Culture (Dias 1997; Rocha et al. 2014).

At the end of the 1^st century, the “Castreja culture” baths were replaced by a thermal complex redesigned at the end of the 2^nd century to widen the bathing area. From this period is also the construction of a natatio that occupied the central part of the previous palaestra and was surrounded by a 12-column portico. Attached to this area and to the west, there was an apodyterium, through which one could enter the baths that had, at this point in time, a bigger number of rooms. In the first half of the 2^nd century, a large square was also built. It has been classified as the forum and performed an area of about 10,000 m², 149 m long (E/W direction) by 68.5 m wide. Whatever the function this space had, we are dealing with an extraordinary example of the processes that involved the indigenous populations from the peninsular northwest, who saw their integration into the Roman cultural system.

The cultural changes associated with the inclusion of indigenous populations into the Roman world can also be visible in the residential areas uncovered in the excavations of two housing units, one on the western slope, located next to the current church and the other on the eastern slope, right next to a wall. In both cases, the contrast existing between the “Castreja culture” circular houses and the italic model houses that succeeded them was recognised. In the residential area on the western slope, a set of six houses was documented, separated by several streets of irregular layout that were part of one of the neighbourhoods formalising the Roman residential unit. This unit is believed to have not been built before the middle of the 1^st century AD. Several different construction stages were registered and they originated the gradual replacement of the previous indigenous round houses.

The latest studies carried out in Tongobriga suggested that the city had entertainment buildings, with a theatre and an amphitheatre being a possibility. On the other hand, studies on the city’s urban fabric have been carried out following on the orientation of the civic group represented by the baths/forum (Rocha et al. 2014).

The rationalisation process related with the construction layout and the reproduction of an architectural language of italic origin on behalf of the indigenous elites living in the Iberian Peninsula, in the context of the conquest of Hispania, confirmed the ability of Rome in attracting the local communities and integrating them into the new economic and social models dominating during the Roman control. In Tongobriga, these new models had an architectural expression in the remains that signalled what was old and associated with the indigenous identity, or what was new and Roman. It is also important to stress the physical proximity of the remains underlying the two cultural substrates. In terms of the dwellings, there was a reuse of materials resulting from the dismantling of the “Castreja culture” houses to build italic type residences, both in Flavian times and throughout the 2^nd century.

The areas known to have examples of Roman-style houses were located in the highest part of town, where there was a continuous evolutionary process when it came to the circular dwellings, which were destroyed and replaced with buildings with straight lines, and where the presence of the italic architectural lexicon can be appreciated. Amongst the new constructions, circulation spaces were designed with orientations that were out of the orthogonal grid of streets proposed for the foundational phase of the city. This assumes that there was a prior planning, although it was necessarily adapted to the territory, which had a well-marked slope (Rocha et al. 2014: 72-77). In reality, the organisation of the streets and houses in the western residential area corroborated the irregularity of the blocks discovered, which may have resulted from the configuration of the previous family households, a well-known model in organising the residential areas of pre-Roman settlements in the region of the bracaraugustan Gallaecia, and which were usually bounded by walls.

When analysing the housing area on the western side, it was possible to identify a series of streets sensibly oriented towards NW/SE and NW/SW. Another element that allowed us to mark the layout and the orientation of the road axes was the entrance to the houses and the identification of some pipelines that carried out the drainage of dirty water under the roads. The analysis of these data suggested the existence of an attraction axis in the western direction, and it was possible to document this orientation in four houses (Fig. 3). The so-called ‘Well House’, located in the most eastern part of the housing unit, had three entrances, with the main one allowing the access to the atrium located in the western façade. It faced a NW/SE street and delimited the housing group including ‘Houses 1, 2 and 3’ and the ‘Impluvium House’. The presence of the other two entrances in the ‘Well House’ allowed us to assume the existence of two other streets, which flanked the house in the north and south. However, when analysing the other rooms’ tangent to the NW/SE street, we failed to identify any existing entrances in that same direction, with the exception of a taberna identified in ‘House 1’. The access to the ‘Impluvium House’, ‘Houses 2 and 3’ and the ‘Stairs House’ was made from an ENE-WSW oriented street. However, and although ‘House 1’ had an entrance from an inner corridor that divided the three houses from the block where they were built, this was a secondary entry point. The main entrance faces a street located on the north, which led to an open area. This was likely the most appropriate solution the owner found to maintain an ianua that would highlight his status through the use of italic lexicon in what concerns the compartments’ layout in relation to the entrances. This way, we concluded that there was a wide variety of solutions when it came to structuring the Tongobriga’s houses, and in their adaptation to the circulation spaces that seemed to divide the lots (Fig. 4).

The group of Roman houses built between the 1^st and 2^nd centuries in Tongobriga and which we have just analysed, is a notable example of the way in which the idea of a dense and winding Mediterranean urban space developed amongst the populations of the Iberian northwest, not only due to the irregular layout of the streets, but also for the presence of, at least, one dead-end street that provided access to the inner spaces in one of the blocks. The existence of access stairs to the upper floors seemed to confirm the intensive nature of the construction process, also documented in other cities of Hispania in much earlier chronological contexts. This was the case with the Hellenistic neighbourhood of Neapolis in Ampurias (Mar and Ruiz 1993). Some indigenous settlements in northern Africa, as for example Dougga and Tiddis, amongst others, witnessed a similar evolutionary pattern, which saw the old streets of the pre-Roman settlement, of irregular layout and adapted to the topography of the slopes, being densified with the construction of two floor houses (Khanoussi 2002). In Italy as well, in the 2^nd century, the densification example of the port city of Ostia is remarkable, albeit on a very broader scale, as the houses on site ended up having five floors (Mar 2002). The case of Ostia revealed how the densification of urban centres was a consequence of the development of urban activities, not only of a residential nature, but also commercial and productive.

The residential set of Tongobriga clearly showed the same type of evolution in terms of how the urban space was thought, and how it was densified by the increased number of population and activities. Consequentially, the proliferation of the plumbing network around the analysed housing area, allowed us to perceive that more than residential needs, there was a presence of several production activities, which, alongside the commercial businesses linked to the tabernae occupying some of the façades, would have entailed an intense everyday life that certainly animated the residential areas of Tongobriga (Leão 2012; Rocha et al. 2014).

On the other hand, the analysis of known Roman houses provided us with very significant data regarding the understanding of how domestic units adopted different architectural solutions, in order to apply the italic grammar, without being submersed into a rigid urban grid that delimited blocks and divided them into lots. In reality, the variable sizes and orientations of the lots seemed to constitute the main conditioning elements of the plants, and the distribution of the houses’ inner spaces, to which would have been added the status and the economic strength of the owners. What seemed to stand out from the analysis of the preserved remains in these new typology dwellings was the repetition of architectural elements of italic matrix, although these seemed to have been almost forcibly applied. Given the physical features of the lots, it was therefore necessary to adapt several spaces, even if this resulted in the loss of their initial functions, as it happened with the atria in the ‘Stairs House’ and ‘House 1’ that failed to assume a real ordering role of the compartments, or the lack of a space separating the ianua from the ambulacrum in the ‘House 2’ atrium.

Thus, the solution found by the owners to give form and create the necessary compartments, whilst ensuring that there was enough space to carry out the activities required by the family life and those of representation, the latter being necessarily adjusted to the social level and the wealth of the owners, was to grow the house vertically. This was a phenomenon identified in all the analysed dwellings in Tongobriga. The idea of building a first floor to group all the spaces needed for the overall functioning of a house, and mainly associated with the domestic and private life, as the cubicula, enabled the ground floor to be used as a reception area and a place to house representation activities. It was therefore possible to build large triclinia and tablina.

The sizes of the lots were an important variable to consider when assessing the features of the house plans, since they influenced the organisation of the inner spaces. Sometimes it was necessary to retreat an open area, or to even sacrifice part of its ambulacrum, by narrowing it down, while trying to maintain its ordering function. There were cases however in which this function disappeared completely, and the open area functioned not as a distributor element, but rather as a point of light and ventilation. This was likely the case with ‘House 1’ and the ‘Stairs House’. Having said this, it is important to recognise that the Roman household should not be understood as a model to be faithfully reproduced; it is not a rigid typology, but should rather be viewed as a language and a lexicon represented by different architectural solutions that have been combined in very flexible ways and according to the requirements and the economic power of the lot owners.

The Roman house seen as a cultural artefact, can never be presumed to follow a closed typology, but rather a multiplicity of architectural solutions. The background was the defining element of the house and shaped its architectural project, which was, in turn, determined by the city’s specificities, as for example the topography, the urban fabric, but also the status and the financial strength of those who commissioned the works (Beltrán and Mostalac 1994).

We are therefore far away from the Vitruvian guidelines for the elite Roman house, which would preferably follow a rectangular plan and require a large area intended to accommodate atria et peristylia amplissima and laxiores ambulationes (Vitr., De arch., VI.V.2). The truth is that this is not what happened in most Hispanic cities, regardless of them having or not a Roman foundation. If we consider that the clusters of imperial foundation had orthogonal urban layouts, and generally had blocks with well-defined limits and plots of standardised sizes that were never too large, we have to admit that it would have been very difficult to accommodate the construction of wide domus, as the ones documented in cities with no orthogonal layout. A few examples of the latter are Conimbriga’s ‘Cantaber House’ and the ‘Repuxos House’, and Ampurias’ ‘Peristyle House’, or even the last stage of ‘House 2’.

There seemed to be no specific identity or model for a Roman house that has been unchangeable in time and space, as the houses have undergone several reconstructions and transformations over time. As we noted in the houses of Tongobriga, there was a great diversity in construction solutions, with details that must be assessed in both spatial and temporal terms, due to the heterogeneity of cities and their elites. This seems to shed some light in understanding certain regional traits, the change in paradigms, ideologies, ways of life, and even in the simple tastes of owners and architects. All these have the ability to influence the characteristics of the changes that have taken place in residences over time (Magalhães 2019).

The analysis of the examples of Tongobriga allows us to conclude that the Roman house cannot be understood as the result neither of the repetition of a repetitive and unique model of a scheme of organization of the house based on an axial system of organization of the compartments, present in the Pompeian house, nor of the application of a rigid typology that defined the characteristics of a house of atrium, of a house of peristyle or a house of atrium and peristyle. It would be more correct to understand the roman house as a language, necessarily variable in morphological terms, but also as a lexicon, composed of grammatical units that express several semantic possibilities, harmoniously solved in the architectonic solution that configures the syntax of the dwelling, which is necessarily result of several factors, but above all a clear reflection of the activities of the owner and of his family and of the necessities arising from the public exhibition of his status in the civic structure of the cities. In truth, the roman house was the mirror of the class and wealth of its occupants, reflecting in its architectonic characteristics the social status of its occupants, so the house worked with a powerful language that in a little literate society transmitted logics of power, privileges and values. In this way, the domus were the main scenario of the political and social life of their owners, establishing the physical display of their economic power, operating as a structure of representation and social control.