1. Introduction

Unlike other cultural aspects of the Iron Age in the Iberian Peninsula, an updated summary of the domestic architecture of the so-called Cultura Castreña, or Hillfort Culture, in the northwest has not yet been produced (Figure 1). The last monograph on this topic is that of Romero Masiá (1976). This deficiency affects archaeological research in the Iberian Peninsula and represents a major limitation for foreign researchers focused on the issues related to domestic space during this period as well as those interested in studying the typical circular houses found on hillforts and their similarity to the Iron Age dwellings of the Atlantic area. In fact, to date very few papers on this subject have been published in English², and most of those deal only tangentially with domestic space (Blanco et al. 2003; Criado and Parcero 1997; González Ruibal 2006; Parcero 2003; Queiroga 2003; Sastre 2002).

Figure 1. Location of study area in Europe.

Studies of the Iron Age house in the northwestern Iberian Peninsula, from the excavations of Martins Sarmento in Briteiros and Sabroso in the 1870s to the present, are an accurate reflection of the historiographic evolution of the archaeology of the northwest (Ayán 2002). Until the 1980s the dominant socio-cultural German School represented the traditional perspective regarding the Hillfort Culture, a perspective that still has some defenders in the scientific community. According to these parameters, the objective of "hillfort archaeology", conceived as a scientific practice and a political tool, is the definition of a cultural group distinct from the rest, conceived as a living organization and defined by a number of "objective" formal characteristics: a scattered environmentally specific territory, the predominance of the circular shape in settlements and structures, the absence of town planning, a specific type of traditional craftsmanship in precious metals and a unique ceramic tradition. In particular, this archaeological perspective developed out of the Asturian and minhoto regionalisms of the last third of the nineteenth century, especially the Galician cultural and political nationalism of the 1920s (Ayán 2005a, 2006; Díaz Santana 2002; González Ruibal 2006-2007; Marín 2005).

According to this organic-historicist approach, the presence of a culture in the northwest that is different from other cultures of the Iberian Peninsula can already be attested in proto-historic periods and represents the origin of Galician ethnicity. According to Galician archaeology, there is total continuity between Iron Age Celtic civilization and the traditional rural world in this region. Influenced by the Hamburg School, most of the research into hillfort houses during the central decade of the twentieth century was focused on the study of the historical continuity between Iron Age dwellings and domestic vernacular architecture, which supposedly retains ancient building techniques in remote mountainous areas on the borders of Asturias, Galicia and León (García y Bellido 1967).

In the 1960s there was a reaction against these Celtic theories, reflected in the appearance of the more neutral term Cultura Castrexa or Hillfort Culture, and in the acceptance of an endogenous origin of the Galician Iron Age communities through Mediterranean and central European influences. In spite of these changes, however, archaeological research has continued to encourage a typological and formalist approach to the interpretation of hillfort houses. The houses in the fortified settlements have been conceived as study subjects in themselves, like fibulae, torcs or ceramic vessels, rather than as a social construction; they are viewed as an architectonically functional form within the context of a cultural landscape. During the 1970s hillfort archaeology was particularly obsessed with temporal parameters; dwellings were sectioned to expose internal stratigraphic profiles and the first radiocarbon dates were obtained. All of these methodological approaches were combined with the formal classification of the houses, in the form of ground plans and territorial distribution studies. It was not until the 1980s that the foundations of a social archaeology of hillforts were laid under the delayed influence of British functionalist spatial archaeology (Clarke 1977, 1984; Hodder and Orton 1990).

This historiographic trend has placed a number of significant conditions and limitations on the archaeological record available for the study of domestic space in Iron Age northwestern Iberia, resulting in a situation that typically surprises European colleagues: even though northwestern Iberia has produced the largest number of excavated settlements, almost nothing is known about hillfort social organization. The archaeological documentation of proto-historical archaeological sites in the northwest is so deficient in significant ways that to a large extent it is impossible to analyze the built environment at these sites. The resulting paradoxical limitations on the investigation of the Iron Age in the northwest have been exacerbated by an approach that pays little attention to the technological, functional and symbolic aspects inherent in any domestic architecture. As M.D. Fernandez Posse has correctly pointed out:

What is really sad is that such attitudes have been maintained in spite of the most meaningful and abundant building records in peninsular proto-history: we only have to compare the small areas that have been excavated in the Iron Age settlements on the plateau at those times with the many hillforts that have already produced evidence of large open-air areas within their built nuclei. So far the analysis of such structures, especially studies regarding the inner structure of the hillforts, has not surpassed the morphological level of Cuevilla's³ work (1998: 78) [author's translation].

Even the information from recently excavated hillforts reflects this phenomenon, replicating the limitations and gaps that are a direct consequence of traditional field archaeology, which has been concerned mainly with exhuming architectonic structures. The old archaeological excavation work focused on exposing wall courses in order to get a clear sense of the ground plan of the structures within archaeological sites. Possible spatial relationships between buildings were not considered, nor were the positions of features found within structures a concern.

This practice—still applicable today in projects that developed within the framework of classical archaeology—results in the loss of important information, especially a lack of proper contextualization of finds. This deficiency makes it impossible to speculate on the internal organization of housing areas or to identify different activity areas. In addition, no data are available regarding structures or smaller building elements (such as internal room dividers or postholes) that could be used as a foundation for hypothetical reconstructions of the domestic sphere.

These deficiencies combined are also an obstacle to functional interpretations of hillfort structures. As a result, ignorance regarding the rules governing the arrangement of the domestic trousseau leads archaeologists to assume that every excavated structure is an individual house, without paying attention to the possible spatial interrelationships within a housing group. Also, and despite the perennial obsession with stratigraphic profiles, no complete stratigraphic sequences exist on which to posit the possible evolutionary development of the built environment within the hillforts. The combination of all these factors prevents an in-depth analysis of the domestic area and seriously complicates any attempt to supersede the descriptive and typological direction in which traditional archaeology has continued to develop.

However, since the early 1990s a number of factors have led to an improvement in the archaeological record available for the hillforts. First of all, it is worth mentioning the application of modern excavation technologies and updated recording conventions (in the form of the Harris matrix, for example); secondly, at a theoretical level the contextual and/or post-processual analyses that resulted from the development of research directions in landscape and architectural archaeology have had a positive impact; and finally, the involvement of a new generation of archaeologists in cultural resource management has resulted in regionally-based projects that have allowed the documentation of previously unknown archaeological sites. In particular, several major archaeological lacunae in our knowledge of the Iron Age cultural landscape have been filled in this way, including evidence for farming terraces and hillfort mines as well as the earliest settlements in the plain that were not fortified during the Roman occupation.

Table 1. Periodization of the northwestern Iberian Peninsula: Recent Prehistory and Protohistory (calibrated dates; LAr, IEGPS, CSIC-XuGa proposal). Key: EN (Initial Neolithic); MN (Middle Neolithic/Megaliths); LN (Late Neolithic); EBA/MBA (Early/Middle Bronze Age); LBA (Late Bronze Age); EIA (Early Iron Age); LIA (Late Iron Age).

This historical context and these new discoveries are the starting point for this article, which provides a very specific perspective for those specialists who do not know at first hand the range of material available in relation to Iron Age hillfort spatial analysis in this region. A chrono-cultural diagram outlining the phases that are commonly accepted for the northwest by most researchers is presented above (Table 1). Included in this updated diachronic synthesis of Iron Age architecture and domestic space in the northwestern Iberian Peninsula are maps showing hillfort distribution patterns and the locations of the archaeological sites mentioned in the text (Figure 2).

Figure 2. Map showing the main prehistoric sites mentioned in the text.

2. Before the hillforts: domestic space in recent prehistory

The archaeological record of the northwestern Iberian Peninsula is substantially limited by the highly acidic soil, which is inimical to even minimal preservation of most ancient archaeological remains, whether osseous, metallic, or organic. Another important limitation is the almost total absence of information about domestic space in the recent prehistory of this area. The fact that most of these prehistoric settlements are invisible as well as poorly preserved, together with their location on slopes with light soils without much stratigraphic depth, means that archaeological research has traditionally been focused on mortuary ritual, which is monumental and visible in the form of Neolithic megalithic tumulus and cist burials of Bronze Age date.

2.1. The relative invisibility of domestic space in recent prehistory

During the past fifteen years, excavations undertaken in the context of lineal subsurface projects (turnpike roads, modern two-lane highways, gas pipelines) have helped to identify a great number and variety of previously unknown domestic structures and have thrown some light on this archaeological gap (Lima Oliveira 2000).

To summarize, the nature of domestic space during the Early to Middle Neolithic (the beginning of the fourth millennium BC) is practically unknown. There are a few documented remains, like the archaeological sites of O Regueiriño (Moaña, Pontevedra) (Lima Oliveira 2005) that refer to small temporary camps with ephemeral housing that was only occupied for short periods of time (Bello and Peña 1995: 112; Lima Oliveira 2000: 46), based on the paucity of material remains and structures. There was minimal investment in building resources and the building materials used were generally perishable. These temporary camps, which left few traces in the landscape, have contributed to the relative invisibility of domestic space during the megalithic phase of the Neolithic. This situation changed during the Late Neolithic in the third millennium BC (Suárez Otero 2002: 12-3). Therefore, the historical evolution of these Neolithic farming societies is characterized by a great variability of settlement forms on a regional scale. Primitive hamlets appear (Vicent 1991) in the form of more permanent settlements with more evidence for internal complexity in the late Neolithic. Examples include the archaeological site of Zarra de Xoacín (Lalín, Pontevedra) (Aboal et al. 2004-2005) with its burned stone structures, foundation trenches and postholes, and the monumental archaeological site of Montenegro (Moaña, Pontevedra). Here the extent of occupation is reflected in the archaeological structures as well as in communal activity areas within a permanent hamlet with roofless houses, ditches used to store agricultural products, a great number of postholes, small demarcation trenches and compartmentalization of space—all the material hallmarks of a typical settlement of that time. In this case, however, we would like to emphasize the fact that the most obvious remains of houses are in the form of circular depressions three to four meters in diameter, with foundation trenches and postholes.

Figure 3. Hut floor, Os Remedios settlement (Bonilla and César 2005: 56).

The same pattern can be seen nearby in the contemporary archaeological site of Os Remedios (Moaña, Pontevedra). Here, on an artificial platform delimited by a surrounding palisade, an extensive domestic space was laid out in the form of a permanent area with ditches, burned structures and paving in which the foundations of a circular house roughly 4.5 meters in diameter are clearly documented based on imprints of postholes located on the edge of the foundation trench (Figure 3). Similarly, most of the postholes assigned to the first occupation of the settlement are circular or oval (Bonilla and César 2005: 56).

This is a construction type that ultimately establishes itself as the standard template in the recent prehistory of the northwestern Iberian Peninsula. The stability of this architectural pattern suggests the existence of communities with a cultural tradition that was symbolically based and reproduced a spatial logic marked by circular visibility patterns. In this sense, it seems significant that these Neolithic societies built their commemorative stone tumulus structures on the same circular plan as their settlements and houses.

Figure 4. Inhabited area of Montenegro settlement (Photo LAr, IEGPS, CSIC-XuGa).

The Montenegro archaeological site exhibits this socio-spatial logic as well. In the upper part of the settlement, coinciding with a terrace along the foothill down to the Ría de Vigo a monumental circular area about 20 meters in diameter was documented with its main entrance facing southeast (Figure 4). The edge of the southwestern quadrant is defined by a line of stones thrust into the ground while the granite rock outcrop naturally defends the remainder of the site. Inside is a circular posthole structure (Gianotti and Cancela 2005: 51) that echoes the organizational model of Neolithic tumular space, with the following characteristics: circular shape, dual and asymmetric organization of space, entrance facing the rising sun, incorporation of natural elements in monumental architecture and concentration of ceramic and lithic material in the southern part of the structure (Criado and Villoch 1998, 2000; Mañana et al. 2002). It is suggested based on this that there was a formal, regular pattern in these agro-pastoral communities in which the circle was used in funerary architecture as well as in monuments erected for social gatherings⁴.

2.2. Hamlets and dwellings during the Bronze Age: the weak imprint of a primitive rural society

Domestic space during the Bronze Age is still a mystery in most of the northern Peninsula (Blas Cortina 1984: 71). In Galicia, the excavation record is so poor that it was almost impossible to outline even a minimal characterization of the settlements of this period up to a few years ago (Fábregas and Meijide 2000: 71-2). In fact, on the eastern Cantabrian coast where cave occupations have survived, remains of circular dwellings have been documented on some of these archaeological sites, for example at the Cantabrian cave of Mirón or the Basque cave of Iruaxpe III (Arias and Armendáriz 1998: 61-2). During the second millennium BC in the northwestern Iberian Peninsula regional variability was extensive, with different occupational strategies coexisting while the social environment was gradually developing from the first primitive rural societies to the birth of fortified settlements during the Iron Age (Méndez and Rey 2005: 95).

Broadly speaking, the settlements documented for the Late Neolithic and Bronze Ages tend to exhibit an occupation-abandonment-reoccupation cycle in the same areas. While this recurrent process does not indicate permanent settlement it also is not consistent with an entirely mobile way of life. Rather, settlement during this period follows a regular pattern, with a gradual but significant impact on the environment that created territories and gave a meaning to predefined points within these territories over time, ultimately developing into permanent settlements (Criado and Martínez 2005: 94).

This period of transition coincides with the adoption of a much more diversified settlement pattern than in previous periods, and it suggests the consolidation of a dual model of territorial occupation that survived until the introduction of hillforts in littoral areas, as in the case of the O Morrazo Peninsula (Méndez and Rey 2005: 104). In that area, the settlements followed two clear siting patterns: on the one hand, locations near catchment areas or near the coast, and on the other hand locations on hillsides at higher elevations, on small peneplains and terraces, but always at a threshold of 400 meters above sea level.

During the first half of the 1980s the first archaeological sites of this type were documented in O Morrazo, including the settlements of Lavapés and Illa de Barxés (Peña Santos 1985: 74) on the Mesa de Montes hillsides. The best known of these sites is Lavapés, which clearly dates to the Early Bronze Age⁵. Except for several houses that are oval in plan—one of which reused the bases of several flat millstones in its construction—and certain groups of stones that were arranged intentionally, no elements related to posthole bases were documented (Peña Santos 1985: 76).

Inland in the northwest from the Early Bronze Age on settlement locations are directly related to humid catchment areas that were periodically and intensively reoccupied. That is why they can be defined as areas of accumulation rather than as individual settlements (Méndez 1994: 87). These areas of accumulation were connected with basins, usually had a braña, or permanent summer camp/pasture in the lower part, and were located in mountainous areas suitable for clearing for farming. They also tended to be in an outlying position with respect to the valley and were linked to important migration corridors. The room structures suggest a type of settlement that was not permanent, with houses made of perishable materials. Cattle pens were also made of perishable post fences (Méndez 1995a: 70-1).

Figure 5. Hut floor, Os Laguiños settlement (Photo LAr, IEGPS, CSIC-XuGa).

The archaeological sites of A Lagoa (O Bocelo), O Fixón (García Lastra 1984; Suárez Otero 1995, 2002: 18) and Os Laguiños (Morrazo) (Mañana and Chao 2005: 61) are good examples, based on non-permanent occupation of a vast basin, with circular dwellings made of perishable materials and conjoined enclosures for cattle (Figure 5). For its part, the settlement of Os Torradoiros (Vázquez Liz 2005: 67), with an associated cultural area in the form of a necropolis), shares these characteristics, with an elliptical domestic structure 5.74 x 4.89 meters in size and a stone foundation.

These communities ensured their survival by practicing hoe or primitive plow agriculture and by exploiting nearby pastures for their bovine livestock, which determined settlement location (Méndez 1994, 1998: 176-8). There are a variety of structures in these settlements (pits, lineal ditches, dwelling foundations), evidence of longer stays and greater settlement stability in this area. The areas given over to cleared land agriculture are quite close to one another and there are demarcation ditches, presumably to separate the housing areas from the farming and pasture lands (Lima Oliveira 2000: 47). All these signs of segmentation of domestic space suggest the social division of the landscape into plots (Criado 1993: 38-41).

2.3. Progressing toward fortification

In the course of the Bronze Age domestic space became more complex. The Early and Middle Bronze Age communities lived in small hamlets that have left very little trace in the landscape and reflect great mobility resulting from a self-sufficient economic system. Power was marked in places of worship, in caves and metal ore deposits. This partly explains the absence of a mortuary record and the monumental nature of domestic space during this period (Jorge 1988b: 92-94, 1990: 225-31).

It was during the period known as the Late Bronze Age that more economically and socially complex communities appear in close proximity to these earlier sites, apparently in response to bronze metallurgy and exotic prestige goods exchange (Martins 1990: 199-201; Méndez 1995b: 77). In southern Galicia and northern Portugal the development of the different socio-political units was asymmetrical; open settlements coexisted with new settlements located on hilltops. These were fortified settlements engaged in exchanges along the Atlantic coast that controlled the exploitation, manufacturing and distribution of metallic goods (Jorge 1990; Martins 1987, 1990).

The archaeological record reflects this dual pattern in the organization of domestic space (Martins 1985: 218-9). The conditions were favorable for increased sedentism related to a complex agro-pastoral economy based on a system of planting crops and leguminous plants on a rotation basis. This was also when new plant domesticates, such as broad beans, were introduced. As part of this process, many open settlements spread throughout the present provinces of Pontevedra, Ourense, Minho and the Douro Littoral, documented by pits dug out of the rock creating a new spatial conception: a landscape divided into plots (Gonçalves et al. 1978, 1980).

A paradigmatic example of this trend is the archaeological site of Bouça do Frade (Baiâo, Porto) (Figure 6). It is situated on the southeast slope of a hillock, controlling the bottom of a valley. It can be considered an open settlement because it is not fortified but is located near a river without prevailing defensive worries. At this settlement the occupational surface was characterized by a number of ditches dug out of the rock (reused as garbage dumps), postholes and some domestic structures of a transitory nature containing artifacts such as boat-shaped millstones, ceramic receptacles for storage, transportation and cooking as well as flint tools (Jorge 1988c: 53-7, 60). The final period of intensive occupation took place during the eighth century BC (Jorge 1988a: 134).

Figure 6. Hut plan, Bouça do Frade (Jorge 1988a: Fig. 15).

The same pattern can be seen in the neighboring archaeological sites of Monte Calvo and Lavra (Baião, Porto) (Figure 7). At the former, which lacked any kind of defensive structure, a number of pits had been dug into the bedrock and several postholes correspond to an open, probably temporary settlement Gonçalves 1981: 86-7, Fig. 4). At Lavra, during the second period of occupation, some pits dug into the substratum were recorded, along with the remains of burned seeds, burned structures and postholes (Sanches 1982, 1988: 126-7, Fig. 1, 1995).

Figure 7. Plan of excavated area at Lavra settlement (Sanches 1988: 126-7, Fig. 1).

Figure 8. Hypothetical reconstruction of a Final Bronze Age settlement by A. González Ruibal (2006-7: 100, Fig. 2.14).

These settlements belong to a cultural context that was doubly influenced, on the one hand, as part of a Late Bronze Age cultural interaction sphere in the northern plateau, with an economy based primarily on agro-pastoral practices that is characterized by storage ditches and decorated pottery of the Cogotas I type (Jorge 1988c: 68); on the other hand, exhibiting some features of the Late Bronze Age Atlantic coast area, such as the decorated pottery of Baiões type and local products such as wide horizontal rimmed cups. The archaeological site of Monte Buxel in southern Galicia (Lima and Prieto 2002: 78-86) is the best example of an open-air settlement from the Late Bronze Age. It featured nine globular pits used for plant food storage, especially cereal grains, with a combined capacity of 1000 - 2000 liters. These archaeological sites represent the climax of productivity and production of agricultural surplus by the communities settled in the northwest in recent prehistory (Figure 8).

Between the tenth and ninth centuries BC in northern Portugal hilltop settlements with natural as well as artificial defensive systems began to appear, such as S. Julião, Barbudo, Coto da Pena, Faria, Roriz, S. Jõao do Rei, Monte da Saia, Santa Marta de Falperra, Monte do Padrâo, Facha, Castro do Peso, Castelo de Aguiar, Santiago de Lagarelhos, Alto da Caldeira or Castelo de Matos (Martins 1987: 745-51) (Figure 9). These proto-castros were related to the increase in trade and exchange in the Atlantic coastal area due to the exchange of metallurgical products. Communities were differentially affected by these varied economic interests, which gave rise to a system of interdependence that created a more heterogeneous society, with multiple settlement strategies and internal configurations linked by interregional cooperation (Jorge 1988b: 97).

Figure 9. Hut plan, San Julião (Bettencourt 2000d: Plate IX).

The best-known settlement of this type is the hillfort of S. Juliâo, located in the central basin of the Cávado River (Figure 10). Its monumental defensive system composed of earth and stone ramparts and an outer ditch dug out of hard coarse-grained bedrock demarcates an upper platform that is laid out as a residential area. The test excavations documented traces of a possible dwelling from the first phase of the occupation. This house is defined by a number of stones arranged between two rock outcrops. Section 3 is within the settlement area adjoining the slope and consists of remains of two rooms. These rooms are demarcated by stone circles with perishable superstructures (Martins 1985: 202-203, Plate IV 12; 1988: Plate XLVII). Inside the rooms there are several levels of clay and gravel paving, as well as traces of some hearths. House 1 seems to have been no more than five meters in diameter (Martins 1990: 123-5). The area between the houses and the slope was used as a garbage dump. During the second stage of occupation the housing area was reorganized and some houses with circular paving were constructed, one of which contained a hearth and postholes immediately on top of the previous structure. Radiocarbon dating assigns the construction of the slope⁶ to between the tenth and ninth centuries BC (970-810 BC), while the second stage of occupation of one of the dwellings⁷ was dated between the ninth and eighth centuries BC (860-740 BC) (Martins 1985: 216-7, 1986).

Figure 10. Hut plans, San Juliâo (Bettencourt 2000d: Plate V).

The excavations that took place at this site between 1989 and 1996 (Bettencourt 2000d) clarified the occupational sequence further. During the first phase of the Bronze Age (S. Julião Ia) the housing area extended across the upper platform, and the house foundations were about four to five meters in diameter. These houses were built using local granite, oak posts and leguminous plants for roofing material as well as walls (Bettencourt 2000d: 105). In the second occupational phase in the ninth century BC the settlement was provided with a monumental fortification system. Excavation on the acropolis yielded evidence for structures with gravel paving demarcated by stones, one of which was 5 to 5.50 meters in diameter. In later structures these stones were smaller, 3.50-3.75 meters in diameter, but the floors continued to be covered with gravel and the entrances faced south-southeast (Bettencourt 2000d: 106).

The oldest occupational level at Castro do Barbudo (Vila Verde) can be assigned to the same horizon, as the radiocarbon dating⁸ of a pile of burned wood from a probably domestic structure proves. However, this could also have been part of a palisade documented in Level 2 Section 4 (Martins 1989: 66, Plate XL, 2). In general we can say that the defensive aspect of the site was not the predominant concern in its location, since the settlement is on the hillside rather than the summit. This is also the case at the site of Alto da Caldeira (Baião, Porto) (Jorge 1981: 72-3).

Another settlement located in an elevated position that has been studied in detail is Castelo de Matos (Baião, Porto). This settlement was an important metal production center with long distance trade connections, as evidenced by the discovery of a piece of amber (Queiroga 1984). A gravel pavement was uncovered under the walls of the medieval castle that was part of a circular house on a platform laid out as a residential area (Figueiral and Queiroga 1988: 141-2, Fig. 4).

The settlement of A Santinha (Amares), located on a low peak in the Cávado River basin belongs in this settlement category as well. The occupation documented on the acropolis and on its south slope can be assigned to the tenth century BC (Bettencourt 2001: 43-5), and according to the excavator, the types of structures, their arrangement as well as the pottery types and other macro remains support the hypothesis that the upper plateau was a community area. Here there was a big dwelling with stone foundations containing pits used to store agricultural products (cereal grains, cruciferous plants, leguminous plants and fruits). The same area was used as a metallurgical workshop (Bettencourt 2001: 44). Dwellings built of perishable materials such as oak beams, saprolite and unfired clay occupied the southern slope. The second phase of occupation dates to the beginning of the Early Iron Age and involved a reorganization of the settlement, with larger storage structures and greater agricultural diversification.

In the settlement of Sola (Braga) the second Bronze Age phase of occupation is documented in the form of the foundations of a circular house that had reused part of the Late Neolithic wall. This house foundation consisted of overlapping stones, marked by a number of postholes of different sizes (Bettencourt 1991-1992: 103, 2000e).

All of these proto-castros were rich in metal items in the form of weapons, ornaments, and tools used in community rituals; they controlled the tin ore resources, agricultural land and the extensive road systems that facilitated supraregional trade (Fábregas and Ruíz-Gálvez 1994). This is also seen in areas⁹ like León and Zamora, western Asturias and the provinces in Galicia, where archaeological fieldwork has increased in the last few years. In the northwestern part of the northern plateau there is a bipartite settlement pattern, with settlements located in the plain or in the mountains and occupied during the formative phase of the Cogotas I culture following the traditional chronology (Celis Sánchez 2002a: 98); these sites are known only from their posthole remains. The period from the end of the tenth century to the end of the ninth century BC, the formative phase of Soto de Medinilla, marks the longest period of occupation of housing groups. Post structures, silos, granaries, circular dwellings, the first palisades and the development of a building technology using clay have all been documented (Celis Sánchez 2002a: 103). Examples include the hillfort of Mazada in Zamora (Gallegos de Campos) (Esparza and Larrazabal 2000), Los Cuestos de la Estación in Benavente (Celis Sánchez 1993) and the hillfort of Sacaojos (Misiego et al. 1999) (Figure 11).

Figure 11. Hut plans at the hillfort of Sacaojos (Misiego et al. 1999: 53).

In western Asturias, the development of the Navia River Basin Archaeological Project has resulted in a complete rewriting of the occupational history of the area as it was known before the early 1990s. Until then, some interpretations even suggested that there had not been a Hillfort Culture here before the Romans. The most recent research has documented the existence of fortified settlements by the end of the Bronze Age at sites like the hillfort of Chao Sanmartín (Villa 2002a: 162-4) and the fortified settlement of Os Castros (Taramundi) (Villa 2002a: 178-9, Plate 7). A pseudo-rectangular house with rounded corners was built on the plateau of Chao Sanmartín that was eight meters long and five meters wide. This level of occupation dates back to¹⁰ the second half of the eighth century BC. The structure contained plain pottery vessels as well as a round wooden disc decorated with plates and bronze strips (Villa 2002a: 161, Plate 1).

Figure 12. Hut floor, Carballeira do Espírito Santo settlement (Photo LAr, IEGPS, CSIC-XuGa).

The Galician provinces, an area that has not been investigated much from an archaeological perspective, also gradually made the transition toward fortified settlements during this period. The archaeological site of A Carballeira do Espírito Santo (Chapa, Silleda, and Pontevedra) is a good example of the general pattern seen in domestic space during the Late Bronze Age in those areas, which were far away from the dynamic developments that characterized the Rías Baixas and northern Portugal. An open housing area, roughly 106 meters (east-west) x 50 meters (north-south) in size, was situated on a shelf sloping slightly to the east toward the Escuadro and Toxa River valleys at this site, with a wide vista from SSE to NW. This open, unfortified Late Bronze Age settlement¹¹, built from perishable materials with pits, postholes and demarcation ditches, is just 50 meters away from the Castro of Chapa, an Iron Age fortified settlement. The most significant feature documented at A Carballeira do Espírito Santo was a circular foundation trench roughly 40-60 centimeters wide surrounding a structure about 4.40-4.30 meters in diameter with a possible entrance in the center of the southwest wall and a small posthole in the center of the western side (Aboal et al. 2003: 46) (Figure 12).

By the end of the Bronze Age the nomadic settlements in the Galician provinces had moved into the fertile bottomlands of the valley. While the addition of fortifications monumentalized these previously open settlements defensively, the internal organization of domestic space and the small circular dwellings constructed of perishable materials were retained from previous periods, as can be seen in the earliest levels of the excavated Galician hillforts from the eighth to fifth centuries BC. We are now beginning to find some Bronze Age settlements that are the immediate predecessors of the hillfort domestic space plan. Up to now, these sites have been presumed to be located on the plateaus of the fortified settlements themselves and some stray bronze finds suggest a certain continuity with traditional Atlantic Bronze Age metallurgy. The archaeological record described here leads us to uphold the existence of a structured cultural tradition throughout recent prehistory. Based on the available archaeological evidence, there is marked continuity between the Bronze Age communities and the first fortified settlements that retained the basic organizational characteristics of domestic space first seen in earlier periods (Martins 1985: 220; Parcero et al. 2007: 167-8).

The problem of the origin of the circular house in the northwestern Iberian Peninsula is in line with this pattern as well. This is part of a formal typologizing debate regarding the reconstruction of the evolution of the hillfort that has continued throughout the twentieth century. It reflects the Celtic invasion theories of culture history (García y Bellido 1941) and is based on an overemphasis on the effects of the East and the Mediterranean on cultural developments in the Iberian Peninsula (Almeida 1965: 203-4; Calo and Sierra 1983: 34-5; García y Bellido 1971: 25-35) as well as the concept of integration via a proposed Atlantic koiné.

The existence of a roundhouse tradition in the Atlantic area from Galicia in the south to Scotland in the north has been expressed by a number of authors, especially Cunliffe (1997: Chapters 7 and 8). Henderson argues that "the occurrence of circular settlements along the western fringes of Europe and their widespread and exclusive use in the British Isles have led a number of authors to suggest that the circular form may represent one visible aspect of a cultural tradition, distinct from the rest of the European continent" (2007: 309). The most recent contributions to this debate question the existence of an ancient tradition dating back to the Neolithic period. They argue that the appearance of the circular house is coterminous with the emergence of settlements located on hilltops during the Late Bronze Age (González Ruibal 2006-2007: 99-100; Silva 1986: 35-6). We suggest that the new archaeological excavations, especially in Galicia, have clarified the evolution of domestic space in recent prehistory, clearly demonstrating the continued existence of circular huts in open settlements and the existence of settlements on elevated landforms in different regions¹².

3. House, family and community in the Early Iron Age

During the Bronze Age human settlement in the northwestern Iberian Peninsula consisted of small open villages of huts and wells made of perishable materials, at the most demarcated by relatively shallow perimeter ditches. There are a few exceptions in some southern areas where settlements are occasionally found on prominent landforms. These new fortified settlements not only represent a novel form of habitation, but they eventually became widespread spatially and remained the dominant settlement type for a long time, as is proved by the fact that the hillfort is still the only type of settlement known from the first millennium BC (Carballo and González 2003: 40; Parcero et al. 2007: 144).

This process of fortification began during the tenth and ninth centuries BC in the littoral and southern regions of the northwest and spread progressively throughout the entire peninsula at different rates in different areas (Alarcão 1992: 44-5). Although the information we have regarding the development of fortified domestic architecture from its origins up to its final abandonment as a cultural tradition is incomplete, we can outline the developmental sequence at those sites whose occupational stages have been studied in detail (Figure 13).

Figure 13. Map of main hillfort sites mentioned in the text.

One hypothesis is that the Phase 1 domestic architecture followed the existing culturalpattern of the Atlantic Late Bronze Age, with most circular houses constructed of perishable materials. This would support the idea of continuity between the Bronze Age and early Iron Age traditions (Bettencourt 2000a: 84.5), which is also exhibited in the ceramic assemblages (Cobas and Parcero 2006: 79-85; Cobas and Prieto 1998: 161) and in the proximity of the new settlements to the old open villages (Parcero 2003: 276). The main break with this earlier tradition occurs when fortifications appear, a phenomenon that influenced the settlement as well as the group, whereas the domestic space inhabited by individual families still reproduced previous spatial and architectural patterns.

Figure 14. Hut floor, As Croas hillfort (Peña Santos 2000: 158).

During this initial stage of the hillfort period there was a clear predominance of small, fortified settlements of less than one hectare, with a single enclosure that usually took advantage of the natural conditions of the surrounding terrain. Evidence for architectural planning can be seen clearly at sites like the hillfort of As Croas in Pontevedra (Peña Santos 2000: 154-5), which was abandoned during the building phase. The site is about one hectare in size, with an upper area and a lower terrace curving around the summit that is protected by a parapet enclosing most of the domestic structures. The masonry wall is about 180 centimeters thick and about one meter and a half high; it was constructed of stone from the settlement itself (Figure 14).

Sites like As Croas demonstrate how the spatial demarcation of these settlement followed a defensive strategy: the building of a defensive ditch and the material used to raise parapets and walls came from the material extracted by the construction of the ditch. The domestic architecture then used the inside wall course as a spatial reference without recourse to individual foundations. The buildings were clearly adapted to the morphology of the terrain, although at some sites terracing modified the natural landforms. Within the walls the distribution of the houses was apparently arbitrary depending on the settlement topography, suggesting that the structures either exploited the land individually or formed groups that followed natural shelves or slopes.

Within the domestic area small circular houses between three and five meters in diameter were built from perishable materials that followed a curve established by an irregular stone base on which was placed a straw-clay surface or pallabarro. The hearth is usually in the center of the house, so rooms for various domestic activities¹³ were arranged around the lareira or hearth.

The first dwellings of this type were documented in the 1970s on hillforts located in Rías Baixas, for example the archaeological site of O Castro Pequeno de Neixón (Boiro, A Coruña), which underwent several decades of excavation but whose field records have, unfortunately, not been well published (Acuña 1976). However, the data we have outline the occupational sequence of one of the first fortified settlements, as well as evidence for important metallurgical activity during the Late Bronze Age (Álvarez et al. 2005; Ayán 2005a: 67; Rey 1992: 167). At O Castro Pequeno a series of house areas could be documented, from buildings that used perishable materials to circular houses with schist stone bases, adobe and straw floors and thatched roofs of straw. These can be dated back to the seventh and sixth centuries BC, in line with the architectural remains exhumed at another littoral site, A Lanzada (Acuña 1976, 1977; Calo and Sierra 1983: 35; Fariña 1983: 34). The people who dwelled in O Castro Pequeno used a masonry technique that exploited the schist and quartzite blocks available from the Neixón deposits as building materials. The houses, each with a central hearth, were all the same size, although there are some indications for the existence of a larger house nine meters in diameter (Balil 1971: 13) (Figure 15).

Figure 15. Hypothetical reconstruction of residential area of Castro Pequeno de Neixón (Álvarez et al. 2005: 112).

From 1980 on, new excavation techniques were applied with greater methodological rigor, resulting in a more reliable archaeological record for understanding the nature of domestic space as conceived by the first inhabitants of the hillforts. For example, the occupants of the hillfort of Torroso (Mos, Pontevedra), which has produced evidence of outstanding metallurgical activity during the seventh century BC, made use of masonry architecture and erected individual houses within a large area otherwise free of buildings during the sixth occupation phase at the site, the first time this is seen in the Galician hillfort world (Peña Santos 1992: 93, Fig. 21) (Figure 16). One of these houses has a circular plan, another a spiral plan and a third a combination of the two plans, replacing earlier circular structures that were all built using perishable materials during the fourth and fifth stages of occupation, and which are attested only by a few postholes, paving and some burned structures in a central location (Peña Santos 1992: 18). The circular house, which was 7.20 centimeters in diameter, was built using a double masonry wall course, with more elaborately worked stones and a better finish in the external face. Clay and saprolite were used as mortar. The walls were 40 centimeters wide and lacked any type of foundation, being built directly on the occupational surface (Peña Santos 1992: 18, Fig. 22 and Plate 22). The other stone structures used the same technique, although their sizes differ; the spiral house is five meters in diameter while the mixed plan house is more than 15 meters along its longest axis¹⁴.

Figure 16. Plan of excavated domestic area, Torroso hillfort (Peña Santos 1992).

This domestic architecture in the Rías Baixas appears to have developed due to the monumentalization of the family unit, a strategy that exploited a new technical resource, the use of masonry as a tool to build houses and defensive systems combined with a traditional way of building houses that was culturally determined, the circular house shape. This domestic architectural model was also adopted in the southwestern area of present-day Galicia based on excavations carried out at hillforts like Penalba (Álvarez Núñez 1986), As Croas (Peña Santos 2000), Pena Redonda (González Ruibal 2004a, 2005a) and Alto do Castro (Parcero 2000a: 64-5), which share some characteristics: isolated houses with an average diameter of five meters and a housing area characterized by a significant amount of open space between the structures.

Figure 17. Hut floor, Penalba hillfort (Álvarez 1991: 39).

The Castro of Penalba (Campo Lameiro, Pontevedra) was a stable settlement with an agricultural and harvesting economy in the Late Bronze Age until it was destroyed by fire (Álvarez Núñez 1986: 61, 1991: 41). It was then rebuilt using new building methods typical of the Iron Age, with a housing area made up of circular house bases between five and five and a half meters in diameter with a central hearth. Subsequent rebuilding episodes maintained the original arrangement, with successive fireplaces placed on top of earlier ones (Álvarez Núñez 1989: 38) (Figure 17).

At the aforementioned Castro of As Croas (Salcedo, Pontevedra), two test excavations carried out in the upper part of the central enclosure documented a circular wall that was 40 centimeters thick and excellently made, with a house plan similar to that at Torroso. The internal paving was made of saprolite while the outer paving was made of small stones. This level of occupation dates back to the seventh and sixth centuries BC, corresponding to the construction of the settlement (Peña Santos 2000: 155-60).

Figure 18. Early Iron Age levels of occupation (Phases IV-VI) at Pena Redonda hillfort (González Ruibal 2004a: Figs. 15, 20).

For its part, the site of Pena Redonda (Pontecaldelas, Pontevedra) presents certain aspects of interest due to its unusual location on a terrace in a foothill region with associated caves as well as the documentation of a long occupational sequence. The oldest housing area consists of the remains of a circular dwelling built with stones and perishable material that can be assigned to the last years of the early Iron Age¹⁵ with two stone-built houses dating to the following period (González Ruibal 2004a: 20-32). The oldest hut is demarcated by a not very deep trench with a few postholes that seem to have originally been built with branches instead of posts, considering the reduced depth of the postholes. The excavator has argued (González Ruibal 2004a: 30-1, Fig. 20) that it is formally comparable to the oldest level of occupation of the hillforts of A Graña (Acuña and Meijide 1991: Fig. 1) and Falperra (Bettencourt 2000b: Table IX) (Figure 18). Another level of occupation from this period was documented at the fortified settlement of Alto do Castro (Cuntis, Pontevedra). Here a circular stone house that had been devastated by fire was excavated. The radiocarbon dates for the burned roof beams (CSIC-1934, cal BC 410-370, two sigma) indicate that the construction and occupation of this domestic structure occurred before the end of the fifth century BC (Parcero 2000a, 2002a: 64-5).

On the other hand, the oldest level of occupation is relatively poorly documented throughout the southwest during the sixth and fifth centuries BC on the rock outcrops of some hillforts that were densely occupied, such as Montealegre (González and Rodríguez 2006; Losada Diéguez 1927) or Troña (Hidalgo 1985: 27). Some sites in the Galician provinces and in the northern coastal region assigned to this phase were also explored using test cores. These documented the existence of perishable architecture, establishing a construction pattern that would continue into the following period. Examples include the sites of Castro de Lobos and Castro Pedro (Penedo and Rodríguez 1991: 202-5), Penarrubia (Arias Vilas 1979), Penedos do Castro (García Valdeiras 2004: 120), Punta do Tallo (Ramil 1989: 60-1), San Trocado (Carballo and González 2003: 39), As Laias (Álvarez and López 2000), Saceda (Carreño 1991: 60; González Ruibal 2005b) and Castromao (Orero Grandal 2000: 181). In the latter hillfort, evidence of pottery was uncovered on the plateau and in the wall foundations, some wares of Alpiarça type, others with a coarse temper and some that may have come from Greece; bronze tools, including the pin of a pivot fibula and the bow of another long pin together with the remains of wooden posts (Calo and Sierra 1983: 34) indicate a level of occupation that can be assigned to the seventh and sixth centuries BC.

Figure 19. Hut floor, A Graña hillfort (Acuña and Meijide 1991: 51).

The site of A Graña stands out within this group of Phase I fortified settlements (Acuña and Meijide 1991: 54; Meijide 1990). The earliest level of occupation that could be documented was a house floor with walls of perishable materials that were demarcated by a trench. Some remains of posts and stakes were found at this site as well. A clay hearth was placed in the approximate center of each house, demarcated by three big stones lying on their sides and a central posthole. Radiocarbon analysis of a charcoal sample from one of the postholes (Gd-6074) resulted in a 2,650±100 BP date for the posthole and another sample from the fireplace (Gd-6068) is within the same range at 2,610±70 BP (Meijide 1990: 114-5) (Figure 19).

Archaeological remains dating to the early Iron Age are also known from Asturias hillforts (Villa 2003: 115-9) like Olivar and Camoca (Villaviciosa) (Camino 2003: 165-6) as well as several along the Navia River basin, including Os Castros de Taramundi, San Chuís in Allande and especially the aforementioned Chao Sanmartín. At this last site, on the hilltop plateau a fortified settlement with a defensive ditch and palisade was established at the end of the Bronze Age between 801 and 778 BC. Three unique features were documented here (Villa 2002b: 151-2; Villa 2003: 116-7; Villa and Cabo 2003: 144-6), including a large house with rectangular plan and rounded corners (12.50 x 4.40 m). This contained some fragments of a large bronze disc 110 centimeters in diameter consisting of a piece of wood decorated on both sides with copper alloy plates and fixed with rivets and nails. Similar bronze pieces were found outside the upper enclosure at the site. It was possible to document the existence of older structures made of perishable materials (Figure 20). Secondly, a deep recess in a wall was used as a tomb. This was contemporary with the monumental enclosure around the acropolis. Finally, a pyre was discovered that was also used during the eighth century BC sheltered by a rock. Chao Sanmartín is a good example of these early fortified settlements and opens up the possibility of new approaches that stress the symbolic nature of cultural space in this type of settlement during this initial phase.

Figure 20. Hypothetical reconstruction of the upper enclosure of Chao Sanmartín (Villa 2003: 140; Fig. 2).

A similar process could be documented in some areas of León and Zamora during the early Iron Age. Circular houses with painted benches, hearths, central postholes and conjoined granaries became quite widespread during this period. The characteristics of the houses were standardized in terms of size and technology, as can be seen at the hillforts of San Juan de Torres, Villacelama, Camarzana, Manganesos de la Polvorosa, and La Mazada (Celis Sánchez 2002a: 101-8). It is important to highlight the extensive sequence documented at Los Cuestos de la Estación (Benavente, Zamora) during the early Iron Age (Celis Sánchez 1993) (Figure 21).

Figure 21. Plan of Phase VII at the settlement of Los Cuestos de la Estación (Celis 1993: 106; Fig. 7).

The first level of occupation in the area that was tested consisted of a circular house made of perishable materials, on top of which there was another housing area with trenches, hearths and stamped clay floors. Successive phases were marked by the superimposition of circular adobe houses, with Phase 6 the most important because of the appearance of a rectangular house (Celis Sánchez 1993: 102-4, Plate 6). During Phase 7 there was a major reorganization of domestic space in the settlement, and this was also when a large circular adobe house with plastered walls 8.40 meters in diameter was constructed. Fourteen layers of paint were documented in this house, and there was an oval hearth in the center of the structure. The postholes outside the house indicate that there used to be outer reinforcements to support the roofing system. During Phase 8, another house was placed on top of this last large structure. This was a smaller house (7.50 m diameter), with a foundation base made of a course of adobe with a bench in the northeastern part of the wall and a central hearth decorated with spatula-applied paintings (Celis Sánchez 1993: 107-8, Plate 8, Fig. 9). During the last period of occupation (Phase 9), the existing domestic architecture was replaced by a large rectangular structure.

To finish this virtual tour of the northwest we have to return to the question of the origins of the fortification phenomenon, and we have to focus on the paradigmatic Castro of Coto da Pena (Caminha, Viana do Castelo), situated at the confluence of the Coura and Miño Rivers. In the oldest occupation level at this site, which dates back to the middle of the ninth century BC, a circular house was excavated (House I) (Silva 1986: 34-5, Fig. 16, E-F). This structure¹⁶ was built directly on bedrock, and in some parts of the perimeter it is possible to see some shallow cuts in the rock where the walls were located. The house plan is an elongated shape oriented north-south, with a curvilinear profile, about six meters (north-south axis) by 4.5 meters (east-west axis). The walls are about 60 centimeters thick and are formed by two courses of small stones with many stabilizing wedges cemented with saprolite mortar and quartzite pebbles. Some ceramic fragments of the Baiões type were found together with some metallic objects, including a sickle and a sheet bronze container. This is why the excavator assigned it to the same occupational level as the Castro da Senhora da Guia (Baiões), which also had a smelting facility in the upper portion of the site. Paving and circular clay hearths belonging to circular houses built of perishable materials were found here as well (Silva 1986: 35).

Generally speaking, despite the paucity of evidence, we can outline a number of formal regularities in the arrangement of the domestic architecture during Phase 1. First, we see a characteristic in this early phase that is typical of hillfort settlements, with isolated houses that do not even share walls. This isolationism, in the case of these circular structures, was partly an artifact of the use of the conical roof, which made shared walls a problem for the normal evacuation of water around the whole perimeter of the house. The topographic and climatic conditions combined with the circular shape made a more or less orthogonal arrangement of the buildings impossible. What we have is an apparent lack of spatial arrangement of the structures, with many unused gaps between the houses and very narrow alleyways that made unimpeded movement quite difficult. The spaces that were used as passages are characterized more by the absence of structures rather than by any qualities expected of spaces that joined one or more buildings. From a structural point of view, the architectural methods used were relatively homogenous, and no significant size differences are seen. The housing areas of these settlements were characterized by their internal permeability, a low rate of built space, freedom of movement within the settlement, the existence of open space between the houses and the absence of a hierarchical system of arranging the settlement space.

The sociological interpretation of this spatial pattern has tended to stress the fact that all of the architectural approaches appear to have adopted this spatial isolation pattern, with its suggestion of independent family domestic units. The lack of dividing walls and the consistent presence of open spaces between the houses implies that domestic space was not maximized, since semi-detached houses would have saved space as well as making construction more convenient. In this regard, the hillfort model contrasts with the spatial arrangement found in other geographical areas during this period of the Iron Age, such as the Iberian Culture.

There are some considerations that may help us understand the logic of this spatial arrangement. The hillfort, as a monumental marker and visual reference point in the landscape, stresses residential development and reinforces the community identity. The internal area of the enclosure can be considered the realm of communal family acitivity, since they are not likely to have segregated their households spatially if there had not been a need to define the different domestic areas, given the absence of evidence within the enclosure for a hierarchical organization. In addition, the inner spatial arrangement suggests that demographic pressure was absent, coupled with a low level of social competitiveness and the apparent predominance of the community over the family. The fact that there are no dividing walls inside the houses suggests an open internal domestic space without subdivisions; this uniform use of the interior space may reflect a high level of internal cohesion within the parental unit as well as an absence of privacy. At the same time, the fact that there is only a single area in which all everyday activities take place seems to argue for relatively little socioeconomic complexity.

The dwelling is a shelter as well as a production and consumption unit; most daily work is carried out inside or around it. The possible activity areas, including cooking, manufacturing of tools, storage, and sleeping, follow a radial arrangement around the hearth. The lareira (hearth) is the central focus of the house and its inhabitants. It is a reference point around which all other objects and people are organized, with the fire supplying the vital light and heat required for domestic life. All this indicates the existence of communities, like those found on the hillforts, which shared beliefs and developed rituals to maintain the beneficent power of the hearth. There are no data confirming this hypothesis, but the important role that the hearth obviously played as a basic element for all the houses of this period is clear. At some sites the archaeological evidence indicates that in the houses that had undergone some rebuilding, the hearth kept the same position, which could also point to its symbolic nature.

Anthropological research emphasizes the fact that the house in pre-modern and pre-industrial societies had great symbolic significance for the people living in it; it was not just a roof over their heads, but it also expressed and represented, as if in microcosm, their concepts of society and the world around them. In this regard, archaeological research has worked out some of these symbolic and cosmological aspects in different prehistoric contexts, for example in the ritual orientation of houses and hearths, or in the discovery of a spatial pattern defined by a binary system that reproduces cosmological concepts.

However, the cosmological influences in this respect have perhaps been expanded out of proportion, similar to the drastic reversal that took place in Britain in Iron Age studies, which decided the binary opposition functionalism vs. ritualism in favor of the latter. These trends culminated in the formulation and spread of the cosmological model, described in its most developed form by Parker Pearson (1999). The foundations of this model were laid by Wait (1985) and Boast and Evans (1986) with respect to the potential ritual associations of roundhouse orientation, and were continued by Hill (1988, 1989, 1996) in his studies of orientation in southern England. Ideas about roundhouse entrance orientation were tested by Alastair Oswald (1991, 1997), who concluded that in the British Iron Age orientation was dominated by ritual concerns, specifically sun worship, as indicated by the predominant east or east-south orientations. This trend culminated with the suggestion that life inside the Iron Age roundhouse was ritually structured. The cosmological model was welcomed with a certain amount of caution by several authors but a process of deconstruction has already begun. One example is Brück (1999a: 325), who has indicated how such work has inadvertently strengthened the functional/ritual dualism in archaeology, while Hunter (1999) has contested claims regarding a left/right distribution of finds in wheel-houses of Scotland. Both Brück (1999b) and Webley (2003) stress the diversity actually expressed in the archaeological record and warn against synthetic and essentially static models that deny human agency. Furthermore, recent testing of the model at site level showed the archaeological evidence to be considerably more complex than the model might suggest (Hodgson et al. 2001). Woodward and Hughes (2007) and Pope (2003) found the model wanting when it was tested against large datasets. Finally, Pope (2007) has shown that the use of domestic space in prehistory was not generally dictated by sun-based belief systems reflected in roundhouse entrance orientation, but that the people designing roundhouses had both structural and spatial concerns in mind. According to this view, the ordering of space within the roundhouse tends to reproduce that created by the architecture itself: the roundhouse form allows the possibility of organizing space along center/periphery lines, which could be explained as hearth-inspired, but also as a conscious organizing principle, while front/back may be a more subconscious mechanism, associated with the desire for light and the potential for contact/ privacy. Pope also welcomes the recent move in prehistoric studies away from structuralism, replacing the functional/ritual, structure/agency and male/female dualisms with concepts of practice and identity. The deconstruction of structuralism and the acceptance that there is no one reading of the archaeological record—that meaning is wholly dependent on context—signals that we have finally entered the post-processual phase in prehistoric studies (cf. Hodder 1989: 70).

However, with respect to hillfort domestic architecture there is an important empirical problem when trying to reach cosmological conclusions of this type. The architectural shape of the circular house, apparently part of a strong cultural tradition that predates the Iron Age, may be the reflection, on a micro-spatial scale, of the cultural landscape. Thus, there could be a correspondence between the different spatial areas of the Castro Culture landscape, as follows:

• Housing space
• Spatial arrangement of structures within the settlement
• Economic territory that shapes and organizes the settlement
• Cultural, symbolic space¹⁷ (Figure 22)

Figure 22. Ritual structure of Guidoiro Areoso (Rey 1995: 12).

The round house could, as suggested by Parker Pearson (1996:19), act as "a microcosm of the Universe". It could thus be viewed as a metaphorical representation of the spatial arrangement of community territory, organized during Phase II following the model of a concave landscape. This is suggested by research in landscape archaeology in Galicia as well (Criado 1993; Parcero 1995), which the author considers compelling.

The round plan shares the round shape of the walled military enclosures, which in their turn coincide with the round shape of the land exploited directly by the community, where all the economic activities necessary for survival are carried out: cultivated lands, pasturage for animals, extraction of raw materials. The hillfort, a monumental milestone built to be visible and to make visible, occupies a central position and the other economic spheres that make up the land directly controlled by the settlement are arranged around it. This landscape model consists of two primary spatial areas: a core zone consisting of the settlement, arable land and pasturage, and land in the peripheral waste ground that serves as a boundary with other communities.

Thus, we suggest that in this initial stage the hillfort societies underwent a process of territorialization and domestication of the landscape. From that starting point a circular spatial model was created to organize material reality, which probably initiated a similar restructuring of the society itself. The social space of the hillforts then was organized by means of strategies intended to stress the identity and cohesion of the community with respect to the outer world: a visible and monumental settlement with restricted access thanks to the wall built around the settlement that simultaneously concealed and isolated the housing area.

This new materiality is consistent with the decline of the hierarchical society that characterized the Late Bronze Age, in which elites who had contact with the Atlantic world based their prestige on the control of bronze metallurgy. During Phase I there was a caesura in the process leading to the emergence of a stratified society; the forms of acquisition and display of power were going through a crisis during this period, and the community was seen as more important than the leaders (Carballo and González 2003: 40; Parcero 2002b: 231-2). Within this context of growing productivity and surpluses, the system used by these communities to keep these resources from being monopolized by particular individuals or groups increased the sense of exclusion and rivalry with respect to other communities while redirecting the internal clashes toward the outside world. Potential clashes that might have resulted from conflicts over the disposition of surpluses were resolved by investing those surpluses in communal construction projects represented by the building of the settlement and its walls (Parcero et al. 2007: 180-2).

To summarize, it is important to highlight four phenomena that influenced the construction of domestic space during the early Iron Age, which also, in our opinion, define it:

1. On the one hand, the emergence of these fortifications marks the appearance of a stable settlement pattern, the beginning of a sedentary way of life in which the connections of the community were symbolically monumentalized within the housing area, and where a significant investment was made in the built environment. However, the houses of this phase still followed a building tradition seen in earlier nomadic settlements during the Bronze Age. This clear continuity with the Late Bronze Age can be seen in the siting pattern of the first hillforts, which are located in areas with great visibility and very close to the previous period's housing areas (Parcero 2000b), as well as in the pottery traditions (Cobas and Prieto 1999).
2. On the other hand, this period was the beginning of a gradual process of monumentalization of housing, which followed to a large extent the formal pattern of the architectural imprint of the Bronze Age.
3. The early Iron Age was characterized by architectural polymorphism and the presence of a great number of regional variations on the construction of domestic space (Jorge 1988b: 97-98; Peña Santos 1992: 21-2, 2000: 159-60). The dynamic area of the Rías Baixas seems to have been an exception, since the proto-monumentalization process that took place there did not occur in the provinces or on the northern coast, where architecture made of straw and adobe continued to be used long into this period.

Within this context, we suggest that the hillfort architectural style in the northwest was already in place by the first phase of the Iron Age, based on a pattern of formal regularity that consisted of different codes of material culture that were consistent with each other and with a specific logical pattern. This is what defines and differentiates these fortified settlements from Iberian fortresses or from hillforts in the British Isles.

4. Domestic space during the late Iron Age: the consolidation of the castros house pattern

During the second phase of the castreño or hillfort period, which has conventionally been dated back to the fifth and fourth centuries BC, the settlement organization model became consolidated due to the widespread shift to stone construction for domestic buildings (Almeida 1984). It is clear that sites like Montealegre, Troña, and As Laias o Castromao had a long sequence of occupation, and the rooms were accordingly extended. The most elevated portion of the settlement was typically the place chosen for early Iron Age settlement, but in the late Iron Age there is a tendency to construct terraces upon which new houses were built that adjoined the original croa, or central part of the castro.

Most of the houses excavated from this period, regardless of the building material used, follow the same basic pattern: a circular or oval plan, about four to five meters in diameter with roofs made of rye straw or plant branches from scrublands such as broom or gorse. Inside the houses there was usually a hearth, a place to store the grain and the domestic trousseau. Divisions within the houses are not common. In order to maximize the available space (about 20 square meters) some houses appear to have had a kind of attic built with light materials. This has many ethnographic parallels in later northwestern vernacular architecture, for example houses in La Corona de Corporales (Sánchez-Palencia and Fernández Posse 1985: 95), Santa Trega (Patiño 1989: 59), and in the later mining hillforts of Chano and Cervantes (López and Álvarez 2000). This architectural model reflects family units with an average of four to five people per house¹⁸ (Camino 2002: 144-6).

It is quite clear that during this period there was a gradual process of artificialization and monumentalization of domestic space. These processes can be seen in the new settlements that emerged during this period (A Forca, Fozara, Cortegada, Coto do Mosteiro, Recarea, Cameixa, As Orelas, Castro Grande de Neixón) and were a response to the socioeconomic changes that resulted from the occupational model mentioned previously.

Once again, the areas of Rías Baixas and northern Portugal are a good example of this architecturalization process of domestic space. The settlements were located on the old Phoenician and Punic trade circuits and as a result they experienced major economic development due to the gradual development of trade relationships coupled with a major population increase. We are only now beginning to identify a process of proto-urbanization in these hillforts. They became central places at which goods moving through the territory along these trade routes were distributed. The convergence of these roles may explain the complex morphology exhibited by most of these sites during this period. This included larger built areas as well as a reduction in the communal, public space. This process can be seen in the archaeological evidence uncovered at Forca (Carballo 1987), Toralla (Hidalgo 1995), Fozara (Hidalgo and Rodríguez 1987), A Lanzada (Suárez and Fariña 1990) and at the recently excavated Castro da Punta do Muíño do Vento (Acuña Piñeiro 2006), Facho de Donón (Schattner and Suárez 2004; Schattner et al. 2005; Schattner et al. 2006), Pena Redonda (González Ruibal 2004a, 2005), Montealegre (Aboal and Castro 2006), O Achadizo (Rubinos et al. 1999) and the Castro Grande de Neixón (Ayán et al. 2005) (vid infra).

There is some contextualized evidence in northern Portugal at the castros of Coto da Pena and Romariz (Silva 1986: 37-43), Cividade de Terroso (Póvoa de Varzim) (Flores and Carneiro 2005) and Cossourado (Paredes de Coura) (Matos da Silva 1995-97). Cividade de Terroso is one of the best-preserved fortified settlements in the Douro-Minho Rivers area, and the ruins found there date to the late pre-Roman period. However, some circular stone buildings that appear to map¹⁹ family units can be dated to the fourth and second centuries BC (Silva 1986: 39-40, Plate XVIII) (Figure 23). The walls consist of two parallel courses with gravel fill, while the paving is made of saprolite; occasionally they are decorated with circle, braid or brush designs. Rectangular or cylindrical stone bases were placed in a central position on these pavements to support posts (Flores and Carneiro 2005: 73, 110-2). This occupational phase ended in a conflagration, probably related to the 138 BC raids of Decimus Junius Brutus (Flores and Carneiro 2005: 73, 113) (Figure 24).

Figure 23. Plan of Early Iron occupation levels at Cividade de Terroso (Silva 1986: 39-40; Plate XVIII).	Figure 24. Excavation of a late Iron Age hut at Cividade de Terroso (Flores and Carneiro 2005).

The settlement of A Cividade de Cossourado (Paredes de Coura, Viana do Castelo district), looking out over the river Coura valley, has a surface area of 10 ha. It gives the impression of great monumentality, enhanced by three defensive walled enclosures. The domestic area excavated during the 1990 dates to the fifth and second centuries BC and features circular and oval structures whose dimensions contrast with the average size of other houses in this region at this time (Gouveia and Matos 1998/2000: 25, Fig. 1) (Figure 25). The circular houses exhibit careful treatment of the walls, which have both outer and inner faces, whereas the oval structures are rougher and not as meticulously built (Matos and Gouveia 2004: 5). In both cases, based on the archaeological evidence, the roofing system consisted of a superstructure resting on wooden posts laid across the walls with an external covering of broom or gorse. According to the excavators, the circular structures were the location of housing and domestic activities, while the oval structures were associated with craft working, storage and the shelter of animals. In the southwestern area of the site, next to a circular tower, an oval building with an internal bench could be documented that has been interpreted as a meeting place for the community²⁰ (Matos and Gouveia 2004: 6) (Figure 26).

Figure 25. Cossourado hillfort (Photo LAr, IEGPS, CSIC-XuGa).	Figure 26. Oval building with a built-in bench documented at Cossourado hillfort (Photo LAr, IEGPS, CSIC-XuGa).

In Galicia, the site of A Forca has yielded interesting data regarding domestic space in this area between the fourth and second centuries BC. Despite the limited extent of the area investigated at this site, the information presented is quite significant, consisting of open tiled areas, water channels, a high building density per square meter, as well as different types of building plans and granite masonry walls (Carballo 1987: 107-10). The site reflects a clear architectural polymorphism, with a number of circular huts four and a half meters in diameter as well as some square or rectangular structures with rounded corners (Figure 27). Excavations at this site, which was constructed in the first half of the fourth century BC and abandoned during the second half of the second century BC, has contextualized material remains of controversial date, such as circular mortars, the use of square house plans, braided decoration, fibulae with a long bow and without a spiral (longo travesaño sem espira), and vitreous paste eyed beads, all of which are related to a great number of imported Punic materials (Carballo 1987: 140-1).

Figure 27. Plan of domestic area at A Forca hillfort (González Ruibal 2006-2007: 351; Fig. 4.46, from an original drawing in Carballo [1987]).

Figure 28. Plan of domestic area at Castro de Punta do Muíño do Vento, Cabo do Mar hillfort (Gonzalez Ruibal 2006-7: 354; Fig. 4.48).

Following the coastline, we arrive at the Vigo River, where we find the Castro de Punta do Muíño do Vento, excavated in 2001 and 2002 at the time of the construction of the Maritime Museum in Vigo. The site was occupied permanently from the seventh century BC until the first century AD; it was located on a sandy substrate in which a hoard of bronze axes and two rings, dating between the sixth and the fifth centuries BC was found. Two circular huts were documented here, built from perishable materials with thick tamped clay floors that retain the postholes of the uprights that supported the roof. Both structures were associated with large quantities of Iberian-Punic²¹ ceramic material and can be dated to the sixth and fifth centuries BC. In the following centuries, the settlement's domestic space became monumentalized by the construction of circular stone houses with clay floors and central hearths (Acuña Piñeiro 2006: 44) (Figure 28).

This type of domestic architecture became quite widespread in this area around this time, as can be seen at a neighboring hillfort located on the island of Toralla (Vigo), where two circular huts were partially excavated (Figure 29). The huts were built using granite and quartzite masonry, and are between six and six and a half meters in diameter, with central hearths and doors facing south (Hidalgo 1995: 195, Plate 1); the oldest level of occupation dates to the fourth and third centuries BC. We see the same phenomenon on the inland hillforts, in pre-littoral areas linked to the Vigo River and the Meridian Depression, such as in the northern and eastern areas of the earliest enclosure of the hillfort of Troña (Hidalgo and Rodríguez 1988: 135), the hillfort of Fozara (Ponteareas, Pontevedra) (Hidalgo and Rodríguez 1987), the second stage of occupation in the fourth and third centuries BC at Alto do Castro (Cuntis, Pontevedra) (Cobas and Parcero 2006: 104-5, Figs. 18-21), and the second stage of occupation at Pena Redonda, where a major reorganization of this densely populated settlement was documented in the form of five circular huts. These huts are organized in a monumentally planned and crowded space that is very characteristic of this period in the Rías Baixas region (González Ruibal 2004a: 58).

Figure 29. Hut foundation and floor, Toralla hillfort (Hidalgo 1995: Plate 1).	Figure 30. Aerial view of the excavated area at Facho hillfort (www.castrenor.com).

The same dynamics are seen to the north at the littoral hillforts of O Facho de Donón (Cangas do Morrazo, Pontevedra) and A Lanzada (O Grove, Pontevedra). The first of these exhibits a long-term occupation from the Late Bronze Age to the Roman Period (Figure 30). The sanctuary devoted to the god Berobreo was built at this time (Fariña and Suárez 2002; Schattner et al. 2004: 26). The trenches of the 2003 excavation campaign uncovered the first oval structure at this site, with granite masonry walls about four and a half meters in diameter and an east-facing door. The expansion of the excavated area uncovered a series of circular or oval domestic structures as well as a D-shaped house plan that was rectangular with rounded corners (Schattner et al. 2006: 173, Fig. 3).

A Lanzada is still an exceptional case in the context of other northwestern domestic architecture. A housing area was found in the so-called Level II suburbium, unprotected by any visible defensive system (Suárez and Fariña 1990: 316); most of the structures have irregular trapezoidal plans with curved wall faces that combine curved and straight lines and are associated with hillfort pottery of fourth and third century BC date as well as material of Punic origin. Recent studies have suggested a Mediterranean influence for this suburbium at A Lanzada, which would explain these architectural peculiarities (Suárez and Fariña 1990: 332), but another possibility is that the area was a temporary habitation for Punic merchants in this area (González Ruibal 2004c, 2006b).

To the north but still in Rías Baixas we find the Arousa River and the hillfort of O Achadizo (Boiro, A Coruña). The suburbium excavated here included an interesting housing area located up against the settlement wall consisting of a circular hut with successive levels of occupation and related deposits of debris. The stratigraphy of this structure allowed the reconstruction of most of the occupational sequence of the site from the early Iron Age to the Galician-Roman period (Carballo et al. 2003: 100-2; Rubinos et al. 1999). The settlement reached the peak period of occupation during Phase II between the fifth and the second centuries BC (calibrated dates) (Rubinos et al. 1999: 154).

The sociocultural processes that we have been describing, which are characteristic of Rías Baixas and northwestern Portugal, seem to have had no parallel in the rest of what was to become northern Gallaecia, where a different sociospatial model appears to have been established, at least based on the limited number of sites excavated in this area. The most important of these is the hillfort of A Cidá de Borneiro (Cabana de Bergantiños, A Coruña) (Figure 31 and 32). Seventy-five percent of its housing area has been excavated, uncovering about 32 buildings, most of which have round plans, a central hearth that partly supports the roof and a central post. Most have central hearths that are part of the structure supporting the roof with a post embedded in the floor (Romero Masiá 1992: 132). Based on the arrangement of these houses, above all the ones located in the central part of the settlement, we can say there was a tendency to form small groups of houses, representing different family groups (Lestón 2006: 19). Until recently, only one level of occupation dating to the late Iron Age had been documented. The hillfort was abandoned by the end of that period.

Figure 31. Plan of Borneiro hillfort (Romero Masiá 1992: 132).	Figure 32. Aerial view of Borneiro hillfort (Lestón 2006: 19).

Apart from the littoral or prelittoral areas, there are two other northwestern regions further inland that have been studied in depth, thanks to systematic CRM projects as well as excavations of the fortified settlements. The work carried out by X. Carballo at Terra de Trasdeza, the first of these sites, clearly demonstrates that the hillforts in this area are part of a different local tradition, with more domestic architecture built using perishable materials. Other examples include the initial stage of occupation at Castrovite to the second century BC (Carballo 1998: 12-3; Carballo and González 2001) and the third century BC occupation documented in the lower part of the hillfort of Cortegada, where the first stone houses coexisted with an architecture built with perishable materials (Carballo 1996: 354).

Figure 33. Excavated floors of huts constructed of perishable materials at Vilela hillfort (Álvarez et al. 2006).

Territorially speaking, there is less monumentalization of domestic architecture as we move from west to east. The Meridian Depression served as a boundary for the distinctive local tradition of Rías Baixas that could have been cultural (Carballo 2003), separating different lands, but with a domestic architecture still directly connected to the previous tradition, as can be seen in the occupational levels during this period at the hillforts of Graña (Meijide 1990), Cameixa (López Cuevillas and Lorenzo Fernández 1986), Coto do Mosteiro (Orero Grandal 1988), A Moura (Samos, Lugo) (Rodríguez García 1999) or Vilela (Taboada, Lugo) (Álvarez et al. 2006) (Figure 33). The occupation at the last of these sites can be dated to the fourth and third centuries BC, with rectangular, round-cornered houses built entirely from perishable materials (López et al. in press).

It is within this context that we must consider the complex hillfort of As Laias (Barbantes, Ourense), which overlooks the Miño River near Ourense. The occupation of this settlement during the late Iron Age produced the material remains of buildings scattered across the rocky hillside where it is located, including some remarkable terracing work. Generally speaking, the remains of houses with hearths were built into the rock face of the mountain, which in some cases resulted in alignments of houses on different levels, and sometimes a house floor was cut directly into the mountain face itself. In some cases the house walls served to control the erosion of vegetation and clay facing, while in other cases the houses were protected by stone foundations. Masonry constructions were also documented, in one case with a central hearth. On one of the terraces a complete occupational family unit was excavated on a horizontal surface thanks to a strong retaining wall. This consisted of a circular hut with a high-quality light clay floor, a central hearth and some wall bases made of clay. This hut was linked to a courtyard or auxiliary area that was quadrangular in plan with a stone foundation (Álvarez and López 2000: 528-9).

The research project in Las Médulas, southwestern León province, developed during the last few years by one of the CSIC research teams under the supervision of F.J. Sánchez- Palencia and M. Dolores Fernández Posse, has also produced quite interesting evidence for internal spatial distribution and the social interpretation of the architectural record of hillforts in this area during this period. The main aim of the project in question is the study of Roman Imperial development of a mining landscape in this area, especially the changes this would have produced in the native socio-economic structure and its influence on the coordination of new settlement strategies, including the occupation of the area and its exploitation (Fernández-Posse et al. 1994; Sánchez-Palencia et al. 1996; Sastre 1998). Within this larger context, most of the research has focused on the excavation of two pre-Roman hillfort sites: Corona de Corporales (Sánchez-Palencia and Fernández-Posse 1985, 1987) and San Juan de Paluezas, which were occupied beginning in the third century BC, and on the analysis of the archaeological evidence for these two settlements (Fernández-Posse 2001). The interpretative framework is based on the comparison of the anthropological model of rural societies to the archaeologically documented evidence from the site (Fernández-Posse and Sánchez-Palencia 1998) (Figure 34).

Figure 34. Reconstruction of a domestic unit at San Juan de Paluezas hillfort (Fernández-Posse and Sánchez-Palencia 1998: 132; Fig. 3).

This architectural study has tackled the structural analysis of the remains based on the quality and characteristics of the materials used, as well as effort and resource investment, coupled with the functional analysis of the different buildings found, which included shops, storerooms, houses, stockyards, yards, etc. and their spatial relationship with other settlement features such as roads, accessibility routes, attachments, spatial demarcations, etc. This research has contributed to our understanding of these sites and has made it possible to identify a local building tradition that made use of schist as a raw material and had square and/or rectangular plans as the fundamental cellular unit and a distinctive feature of domestic space (Figure 35). Also documented is the individualization of nuclear family units, self-sufficient entities in terms of production and consumption that were separated from other units and lived in a domestic area made up of different attached large rooms, of which the storerooms were the most important.

Figure 35. Hypothetical reconstruction of a hut at Corona de Corporales hillfort (Sánchez-Palencia and Fernández-Posse 1987).

In western Asturias the archaeological record is quite detailed thanks to several systematic excavations carried out on sites in the Navia River basin where domestic architecture is characterized by the use of circular, elliptic or rectangular house plans with rounded corners as well as the use of slate and schist joins combined with clay. Dividing walls were not used, resulting in the tangential juxtaposition of independent wall courses in cases where buildings were in close proximity (Villa 2003: 120). This architectural model was used in the fourth century BC at Chao Sanmartín, Monte Castrelo de Pelou, which was occupied from the fourth to third centuries BC, at Os Castros (Taramundi), at the hillfort of San Chuís, where huts with stratigraphic evidence dated back to the fourth and first centuries BC were found, and at the littoral hillfort of Cabo Blanco (El Franco), where some dwellings were documented dated between the fourth century BC and the first century AD (Villa 2003: 120-1) (Figure 36).

Figure 36. Plan of San Chuís hillfort (Jordá et al. 2002: 19; Fig. 2).

We must keep in mind all this regional architectural variety, as well as the existence of local building traditions and different rates of socio-economic evolution, if we hope to outline a number of more or less widespread tendencies characteristic of domestic space at this time. The interpretative framework that follows is aimed at highlighting the social and cultural guidelines that have a bearing on the constructive process of the architectural record as the material expression of culture. We are far from defining a model for domestic space in the northwestern Iron Age, but some hypotheses can be generated.

Figure 37. View of Baroña hillfort (Photo LAr, IEGPS, CSIC-XuGa).

First of all, we must highlight the significant stability of the internal organization plan of the hillfort settlements. This plan survived into the late Iron Age and its origins probably date back to the first millennium BC. We can verify the continuous repetition of a dwelling plan that is characterized by a well-defined type of house, as can be seen at hillforts like Borneiro (Romero Masiá 1987) or Baroña (Calo and Soeiro 1986) (Figure 37).

For their part, the fortified settlements of the late Iron Age have some relatively homogenous features, such as the retention of the ditch as the first element demarcating the living area and the increasing complexity of the defensive system over time, with the appearance of other types of walls, including the compartmentalization of structures in hillfort units in western Asturias as early as the fourth century BC (Villa 2000, 2003).

The restructuring that took place in these settlements during this period included the extension of the living area and the rebuilding of houses in stone, but according to the same construction plan as before. It seems that a particular type of settlement model was consolidated and stabilized during this time—the hillfort or permanent settlement—in response to the socio-economic needs of a growing rural population that had become permanently established in the region they were exploiting.

There seems to have been a marked demographic increase during this period that could have resulted in increased competition for land, initiating a gradual social segmentation process, a dynamic that has been characterized as defining every rural society (Chayanov 1986; Shanin 1976; Wolf 1982): the ideal of autarchy that had supported the family unit of production and consumption was confronted by the imperative of social cohesion within the settlement imposed by defensive needs and the collective labor requirements that characterize a survival economy.

Seen in this context, the stability of the model may be explained by the need to legitimize that continuity in the settlement through the repetition of architectural elements like dwelling types at a time of increasing social competitiveness. Paradoxically, the individualization of the family unit contrasted with the role played by the settlement itself as a reference point for the creation of social identity. This would explain the considerable regularity found in the built areas of these sites, and from this we can deduce that there was great cultural integration and harmony when it came to observing the rules and values of the community's everyday life.

So, in the intermediate stage of development of the hillfort world there was a clear architectural model accepted by the group, a style that was the result of a cultural tradition respected by the community. Related to this, the domestic hillfort architecture has some of the defining features of what anthropologists have called primitive architecture (Guidoni 1989; Rapoport 1972). In this type of tradition-oriented socio-cultural formation the development of architectural design is based on the technical knowledge available to the community. Hence any member of the group would be able to build their own house. However, there was always a pre-determined model pre- or proscribing certain aspects of the design that also met most of the cultural, physical and maintenance requirements. This is typically a uniform model allowing little innovation and with a predominantly standardized shape.

According to A. Rapoport (1972: 15-6) this construction pattern is totally lacking in theoretical or aesthetic pretentions; it is a non-specialized and open-ended adaptation to the conditions imposed by the settlement location and the microclimate; it is also characterized by a respect for any existing structures and by the presence of an established common heritage and a hierarchy of values to which the construction activity adapts itself point by point. Therefore, in such societies, including traditional rural ones, cultural tradition imposes a rigid system of control over the construction of a house, establishing a model everyone recognizes, which explains the absence of designers or specialists. Also, in these communities there is no difference between magic and work, religion and the secular realm, ritual and secular use of space; all the material culture manifestations reflect the cosmovision accepted and shared by the group. The domestic architecture, the building shape, is a physical incarnation of that logic, of that tradition; the house and the everyday activities carried out inside it symbolically exhibit the metaphorical and ontological foundations of the particular worldview of its inhabitants (Parker Pearson and Richards 1994).

We therefore suggest that hillfort house architecture was mainly influenced by socio-cultural factors. Its long survival was due to a strong cultural tradition that perpetuated that architectural style because it met the social and symbolic needs of the hillfort societies perfectly. This is the only explanation we can propose for the long survival of this type of building; Romanization and the gradual abandonment of the indigenous tradition that supported this model ultimately resulted in the replacement (although not definitively) of this type of construction with a different one. Domestic hillfort architecture appears to reflect a style of construction and a technology belonging to a different social reality than that of communities in the previous period.

There also was a clear spatial division between individual domestic units in the hillfort communities, which were conceived as isolated housing areas with more control over movement and access. In some cases they were internally divided into different zones with different functions, indicating the appearance of a new type of courtyard house that would become quite popular in southern Galicia. Control over access to the buildings was increased with the use of hallways and decoration appears on thresholds, lintels and jambs in special cases; the visibility of the house was emphasized more and the external walls were more carefully finished. Both of these phenomena seem to have taken place during the last stage of the late Iron Age.

In our opinion, this period marks the final social sanctioning of a construction tradition in the form of an architectural style that would not undergo important technical modifications until the Roman period. This architectural tradition was used as a tool to shape a social structure that followed a socio-economic logic defined by a specific process: the increase of agricultural exploitation during this period, cause and consequence of the consolidation process of a rural society and economy. The structures for storing crops became more common as did structures built to store surpluses. This can be seen at sites like Castrovite (Carballo 1998: 13-6), in the upper area of the hillfort of As Laias, where many storage structures with large amounts of carbonized grain were found dating to the late Iron Age (López and Álvarez 2000) and in the upper area of the Castro Grande of Neixón (Boiro, A Coruña) (Ayán in press).

The success of these communities resulted in increased demographic pressure and competition for resources; the population increase transformed the configuration of the hillfort enclosures into a chaotic conglomeration of clustered dwellings. There was a need to consolidate group identity that directly competed with the self-assertion process of the family units that were the production and consumption units within this economy. Access control, house monumentalization and the need for attached areas such as hallways were understood as spaces in which people could carry out their individual domestic activities.

This model is reflected in domestic units, especially the circular dwelling that was like a microcosm (as the hillfort is like an imago mundi) of social space, although the dichotomy displayed in this case was the opposition between the family unit and the community. Hence the main aim of the architectural solutions, regardless of the environmental conditions, was to achieve the total isolation of the house. The house in the hillfort is a physical reincarnation of an ideal environment: the rural ideal of survival, of self-sufficiency, that governed the life of a family unit conceived as a production and consumption body. For its part, the hillfort was the physical expression of that same ideal: the settlement as the identity framework of a close-knit rural community. Both are devices that perpetuate and enable that ideal, exhibiting that worldview and the way of life supporting it. This is the reason that we can talk about a distinct hillfort architecture today.

Figure 38. Communal structure at the Briteiros oppidum (www.castrenor.com).

Some communal architectural elements that appeared inside the settlement at this time are spatially differentiated, such as the communal hut dating to the late Iron Age at Chao Sanmartín (Villa and Cabo 2003: 152), identical to those at Coaña, Mohías and Pendía that were replaced during the late Roman empire with a monumental area like a square paved with slate and decorated with continuous benches along the north and east walls. Huts with oval plans, continuous benches and no evidence of domestic activity were documented in northern Portugal as well, for example a structure at the well-known Briteiros hillfort that is about 11 meters in diameter (Figure 38), and at the aforementioned and somewhat smaller hillfort of Cossourado in Paredes de Coura, which was occupied between the fifth and second centuries BC (Matos da Silva 1995-1997).

On the other hand, the hypothetical evolution of a hierarchical social organization and the eventual emergence of the Germanic society that created a divided landscape (Parcero 2002b: 175-82, 229; Parcero et al. 2007: 216-7) contrasts with the standardized and homogeneous houses, with their self-sufficient family units, of this earlier period (Camino 2002: 150) (Figure 39). However, these structures do not reflect an egalitarian society; evidence of a system of rank and class can be seen in the housing culture, as in other parts of the Celtic world (Brañas 2005; García Quintela 2007), one of a number of similarities between the Atlantic zone and central Europe, where a wood-based architecture, built with perishable materials, also housed a hierarchical society. This can be seen in the archaeological assemblages found in some tombs and in the Greek and Roman sources²².

Figure 39. Hypothetical reconstruction of a Moriyón hillfort hut structure (Camino 2002: 145; Fig. 2).

5. Oppida and courtyard houses in the late Iron Age

Certain sites in littoral and southern areas, including Santa Trega (Peña Santos 1984-1985, 1988, 1989, 1990), Troña (Hidalgo 1987), Elviña (Ayán 2001; Bello 2006), Castromao (Orero Grandal 2000), Saceda (González Ruibal 2005a) and S. Cibrán de Lás (Álvarez et al. 2004) were located in areas that were part of the trade route system and began to undergo significant economic development due to the development of trade relationships and an increase in population beginning in the second century BC. This proto-hillfort development process spread across a wide area (Almagro 1994, 2002), and is clearly documented in northern Portugal at sites like Briteiros, Sanfins, Cividade de Terroso and Santa Luzia. The convergence of these factors may explain the complex morphology of most of the sites located in the bracarense area.

As mentioned previously, the current discussion focuses on the origin and chronology of these oppida, since most researchers believe that this development process is typical of a change of era, encouraged and promoted by the stability supposedly brought to the northwest by Rome. In this sense, the Roman influence is thought to have shaped the hillfort domestic landscape, especially in southern Galicia (Conventus bracarense), characterized by the development of the orthogonal city plan, the introduction of square and rectangular house plans, the introduction of decorative motifs in houses and complex defensive systems, the construction of infrastructure, including areas for public use, and the improvement of stone building techniques (Acuña 1977; 1996; Almeida 1984; Balil 1971; Blanco 1960; Maluquer 1975; Romero 1976). The chronology was interpreted as the result of a change of era and these settlements were defined as Galician-Roman settlements (Peña Santos 1988).

Today, thanks to the significant progress made in research and the recovery of new material in the course of the past 15 years of excavations, we have enough data to revise this historiographic model and to incorporate the concept of oppida dynamics in the northwest into the discussion of the general evolution of the Iron Age along the Atlantic coast and in central Europe (González Ruibal 2005: 268-9; Parcero et al. 2007: 224-5). These major settlements should not be automatically interpreted as the product of Romanization, but as the result of an endogenous evolution or a late progression of the city concept introduced from the Mediterranean regions to the northern and western parts of the Iberian Peninsula (Parcero et al. 2007: 238-41).

We propose that during this late stage (from the end of the second century BC until the Roman period) there emerged, in specific areas, a series of large central places that reached a significant degree of monumentality and exhibited a complex spatial organization in line with a very characteristic model of domestic architecture. There also were significant technological improvements in hillfort architecture that demonstrate an increased use of iron tools to work the stone. This process explains the explosion of stonework during this period, which is reflected in a diversified final product that included different types of tools and a different treatment of the stone itself. All of this resulted in an increasingly complex defensive system in which settlement walls were made entirely of stone whose inner and outer faces exhibit irregular masonry decoration and/or the marks of polygonal or pseudo-reticular tools.

Figure 40. Excavated area of the oppidum of San Cibrán de Lás (Álvarez et al. 2004).

The domestic spaces also increased in size in response to larger numbers of people. About one thousand people are calculated to have inhabited the oppidum of S. Cibrán de Lás, for example (Yolanda Álvarez, pers. comm.), resulting in the reorganization of the interior of the settlement and of domestic space (Figure 40).

The interior organization of settlements during this period was based on an east-west axis, possibly due to southern influences, but more likely the product of a monumental materialization of the architectural pattern that had already developed in the late Iron Age at sites like the Castro Grande de Neixón (Ayán et al. 2005) or Castrolandín (Ayán et al. in press), where a central wall encircles a number of rock outcrops situated in the northern part of the settlement, with two entrances, one facing east, the other facing west.

The excavation of the defensive ditches continued to be the main source of stone for the erection of the stone architecture and the walls remained the primary referent for the organization and coordination of the settlement. On the other hand, the walls achieved an unprecedented monumentality, with the external facing stones worked with picks and helicoidal tools. Apart from its undeniable defensive and coercive functions, directed both outward and inward, the architecture clearly represents a polysemous and multidimensional sign that occasionally becomes true monumental scenery, a symbol of prestige and identity.

Figure 41. Plan of excavated area at Cividade de Ãncora (Silva 1986: 48-51; Plate XXVII).

A proto-urban settlement type appears to exist in northern Portugal with an internal spatial organization based on divided groups of buildings identified as family areas or courtyard houses, the habitations of individual families living in the settlement (Almeida 1984; Silva 1986). One of the best-studied examples is a group of housing units excavated between 1978 and 1980 in the hillfort of Cividade de Ãncora (Caminha, Viana do Castelo) (Silva 1986: 48-51, Pl. XXVII) (Figure 41). The central architectural element consists of three circular buildings, one of which had a hallway, as well as a building with a sub-rectangular plan and a furnace, which were arranged around a paved courtyard containing a tank or fountain. According to the excavator, the main domestic activities (eating and sleeping) took place in the circular unit, which had carefully built walls and paving, as well as plaster decoration; hearths and individual benches were arranged around the walls, which surround an open-plan area without supporting posts, like the other circular structures. The rectangular structure had a similar function, also with a hearth, two individual benches and a bread oven. The house with its atrium and continuous benches attached to the walls was clearly a meeting place for the three resident nuclear families (Silva 1986: 49). The residential complex included an attached funerary area (quite rare in the northwest) with cremation cist burials (Silva 1986: Plate XXVII).

Figure 42. Reconstruction of a domestic unit in the oppidum of Sanfins (Sande 2006: 76).

Figure 43. Hypothetical reconstruction of a domestic unit at Cividade de Terroso (Flores and Carneiro 2005: 125; Fig. 7).

The architectural structure of the courtyard-house is well-known thanks to the experimental archaeology carried out by A. Coelho in the city of Sanfins. The occupied space in this oppidum was orthogonal, and about forty housing areas were documented within it that the excavators interpret as domestic family units (Almeida 1984; Silva 1995) (Figure 42). The functional interpretation of the rebuilt area, which consists of a confined space containing three circular houses and a large rectangular structure facing a central courtyard, was based on a formal analysis that examined three variables: the quality of the construction, the presence or absence of decorative motifs, and the domestic assemblage in each building. According to these criteria, the circular shape of the enclosure, which has a hall like an atrium, a central hearth and a triskelion, or great triple spiral decoration, on one of its walls, can be understood as a prestigious building that may have been used as a meeting place and/or a domestic sanctuary. This type of housing unit, composed of páteo, composta e lacha according to the Mediterranean pattern, also incorporated several buildings with different functions, such as a kitchen/bedroom, silo/hórreo or raised granary, and threshing yard floor (Figure 43). It would have housed a large family, a self-sufficient production and consumption unit along the lines of the process observed in the intermediate stage, as at the site of San Juan de