1. Introduction

Our aim in this article is to present the general features of the archaeological record for a geographical area and period traditionally considered part of the Celtic region of the Iberian Peninsula. Apart from offering a summary of the material features of the record, we will explore the interpretative tendencies which, from an archaeological point of view, are normally used to try to give these features historic meaning. Obviously, it is not possible to deal with this issue without considering the current debate about the Celtic nature of the period (Armada 2002; Díaz Santana 2002; López Jiménez 1999, among others), as cultural and ethnic affiliation are undoubtedly essential factors to be considered in any historical reconstruction. However, we have preferred to remain on the sidelines on this issue for two main reasons. The first is that from the outset we knew about the general aims of this editorial project, and that other authors were taking part that would be able to deal with this issue in greater depth.¹ The second reason, which follows on from the first, is that our goal is to offer readers a representation of the archaeological remains that can be linked to other types of information (for example, from literary, epigraphic or anthropological sources) essential to the reconstruction of the cultural reality of this part of the Iberian Peninsula during the Iron Age.

2. General remarks

When describing the archaeological record from the first millennium BC in this part of the Peninsula, there are two main features that must be dealt with before any others, as they reveal to us most of the specific 'signs of identity' for the period, and also strongly condition our current knowledge of the time. The first of these features is a complete lack of information about any type of funerary rites. The second is that only one type of settlement is known: the 'castros' or hillforts, fortified settlements that were the only type in use for approximately one thousand years.

The lack of archaeological evidence of funerary rites has always been one of the features that most strongly defines the Hillfort culture, perfectly summed up in the ironic words of C. Alonso del Real who said of its people: "they lived, but did not die" (Alonso 1991). No burial site has been found to date belonging to any of the stages of the Iron Age. From the individual burials in cists or ditches of the Late Bronze Age and the appearance of the first necropoli in the Roman period in fully urban contexts until at least the second century AD (such as those in the town of Lugo or Vigo) there is a complete vacuum, void of any evidence of funerary activity. Some decades ago it was revealed that what appeared to be a necropolis with pits had been found at the hillfort of Meirás (close to the city of A Coruña), published as such by J. M. Luengo (1950). Leaving aside the question of whether all of these features were burials or not, it appears that these structures were built over the dwellings in the hillfort, indicating that their origins date from after it was abandoned. More recently, a possible burial was documented in the hillfort of Terroso, in northern Portugal (Gomes and Carneiro 1999; Silva 1986), although without doubt these are later structures from at least the second century AD, from a period which, as we will see later on in this article, marked the breakdown of the order of the hillfort culture as such (for other examples, see Soeiro 1997).

For a long time it has been suggested that Iron Age burials were superficially invisible structures, and not monumental, meaning that the burials are difficult to find. Their absence could therefore be due to the fact that they have not been found to date. However, this argument is weakened when comparing the Iron Age with other contexts in which burials are equally invisible but are known, such as the pits or cists of the Bronze Age. In any case, apparent confirmation of the non-viability of this hypothesis has been found in recent fieldwork, such as archaeological monitoring of public construction projects on the Coruña-Vigo oil pipeline (155 km), Galicia's Gas Network (more than 600 km), and several sections of new motorways and highways, carried out by the Laboratory of Heritage, Palaeoecology and Landscape of the University of Santiago de Compostela since 1992. These construction projects have made it possible to identify and document an enormous number of archaeological sites of very different types and time periods, many of which are not visible on the surface and have unusual forms (for examples, see Amado et al. 1998, 1999; Criado et al. 2000). These public construction projects have been carried out in a wide range of geographical areas, from the deepest valleys to the highest mountain ranges, frequently passing close to hillfort sites, and making it possible to identify numerous peripheral structures, such as evidence of spaces used for cultivation (Parcero 1999a). However, no evidence of funerary activity from the Iron Age has been found in the course of any of these excavations to date.

To this is added another important issue. The characteristics of the soil in most of the northwest of the Iberian Peninsula (with its underlying granite substrate and high acidity) are a serious obstacle to the preservation of organic material. Apart from a number of exceptional contexts, such as sandy areas on the coast or shell deposits, recovering organic remains from before the late Roman period is a truly difficult task. Material such as wood, antler and bone is not preserved unless previously burnt (like many of the seeds found in hillforts), or in exceptional cases preserved under conditions that make recovery difficult. So it is quite rare for perishable archaeological materials to be recovered from hillforts, except when fire was somehow involved. Even more resistant materials such as pottery or metal are found in a highly deteriorated state (Fernández Ibáñez et al.1993). This means that given the apparent lack of burial structures, monumental or not, the identification of burial sites unaccompanied by inorganic material is practically impossible.

This lack of burial structures does not imply the absence of funerary rites or of an ideology of death. It simply means that mortuary practices are not highly visible, making it extremely difficult for us to reconstruct them. It also represents a serious obstacle in the thorough description of many of the aspects of the socio-cultural dynamics of these communities. It is important to remember the essential role of necropoli in analysing the social structure of any archaeological context; a good example of this is that in the general European context, the identity of Iron Age warrior elites is primarily reflected in the necropoli from the period, whether these are burials with few monumental features (Hedeager 1992) or great tumular necropoli such as the so-called 'princely tombs' (Collis 1989). The same may be said of most of the areas in the peninsula, such as the Celtiberian region (Burillo 1998; Lorrio 1997) or the south (Berrocal 1992; Ruiz and Molinos 1993).

The second most distinctive feature is the existence of a single type of settlement, the hillforts or castros. These were fortified settlements generally small in size, less than two hectares, which only grew in size by the end of the period, covering as many as 40 hectares. Some were large oppidum-type settlements that, as we will see, were the result of a specific development at the end of the period. It is also at this time, from the first century AD onwards, that the first evidence appears of settlements different from the hillforts; previously, between the eighth century BC and the first century AD, hillforts were the only type of settlement known. For many years this was thought to be due to a lack of information about other kinds of settlements rather than their absence, an opinion that is still frequently held at least for certain regions (Camino 2002). However, the above-mentioned intensive field surveys and archaeological monitoring of public construction projects have not provided any evidence to support this theory. In fact, some examples of unfortified settlements are known with indigenous archaeological material, similar or identical to that found in hillforts, although at the moment none appear to date earlier than the end of the first century AD (see Aboal and Cobas 1999; Aboal and Parcero 1999; Ayán et al. 2002; Soeiro 1997; a summary is offered in Pérez Losada 1996, 2000).

In fact, these settlements are so specific to this time period that they have given their name to the archaeological culture we are concerned with: Cultura Castrexa. According to the traditional definition of this term, this includes the archaeological elements related to the occupation of hillforts in the northwest, a process that started at the end of the Bronze Age and ended with the Indigenous-Roman period. However, numerous changes have been identified during the period with regard to different elements: settlement patterns (Carballo 1990), architectonic structures (Carballo 1996a), types of dwellings and households (Romero 1976), material culture such as pottery or metalwork (Cobas 1999; Rey Castiñera 1999), the flow of imported goods (Naveiro 1991; Naveiro and Pérez 1992), etc; these formed the basis for a series of proposals concerning periodization (Calo 1993; Martins 1990; Peña 1992b; Silva 1986, among others). Although their content varies, all are based on three phases: 1) an initial or formative phase (broadly speaking, ninth to fifth centuries BC, usually defined as beginning with the transition between the Bronze and Iron Ages); 2) an intermediate pre-Roman period (fifth to first centuries BC, identified with the Iron Age), and 3) a final Indigenous-Roman phase, considered the 'peak' period by many authors (Calo 1993), which took place under Roman occupation and was influenced by it from the first century BC to the second century AD.

Nonetheless, this concept involves the interweaving of three different criteria for periodization for the same archaeological record: 1) based on traditional techno-typological principles (Bronze Age, Iron Age); 2) the 'cultural-local' criterion (Hillfort Culture); and 3) referring to specific historical events (the Roman conquest). In a previous publication (Méndez and Parcero 1995) we proposed a general reworking of the problem, a task that for the time being we have not continued to develop, and which is beyond the scope of this paper (see comments on this issue in Parcero 2002a). We prefer to use a scheme in which the start of the Iron Age would not be marked by techno-typological criteria, but by the appearance of the fortified settlement instead, considered not as open settlements on hill-top locations, but as occupied areas delimited by defensive structures and situated in prominent parts of the landscape. We will therefore maintain this scheme in principle, although we will refer to the first phase as the Early Iron Age, the second as the Late Iron Age, and the third as the Indigenous-Roman period, as in previous publications (Parcero 2000b, 2002a) (Table1).

Table 1 Chronological scheme used in this article (modified after Chapa and Belén 1997:14).

3. The archaeological evidence: a diachronic account

We will now go on to detail the features of the archaeological record, based on the temporal sequence outlined above. The most relevant elements for each of the three phases will be presented, beginning with the shape and structure of the settlement, followed by a discussion of material culture elements found, and finally outlining the way in which the settlements were related to their surroundings.

3.1. Early Iron Age

3.1.1. The forms, patterns and structure of settlements

The first feature that characterizes the landscape of the Early Iron Age is that the settlement is permanently occupied for the first time. The progressive decrease in the mobility of communities seen throughout the Bronze Age (Méndez 1994, 1998) had reached a critical point by this time and ultimately such mobility completely disappeared. The earliest occupied hillforts indicate that human groups opted to settle permanently in specific locations. The fact that settlements became stable is accompanied by an important novelty in the archaeological record: they were also physically delimited. For the first time it appears that there was an important emphasis on the social aspect of building community structures within settlements. The few early hillforts excavated to any great extent in Galicia (with a few more in the north of Portugal and hardly any in Asturias or the western Meseta) reveal that one of the most common features is the appearance of artificial earthworks defining the limits of the occupied area. These are most frequently ditches, such as those seen in Torroso (Peña 1992a), Alto do Castro (Parcero 2000a) or Chao Sanmartín (Villa 2002a), cases in which archaeological work made it possible to document such features directly. Apart from ditches, other "positive" structures appear, such as ramparts (normally made of earth), occasionally reinforced on one or both sides with wooden posts; this is the interpretation given to the ramparts of Alto do Castro (Parcero 2000a), Chao Sanmartín (Villa 2002a) or Punta do Tallo (Ramil 1989) (Figs. 1, 2). Examples of stone walls are also known, although these are usually very low, as at As Croas hillfort (Peña 2000) (Fig. 3). However, these 'positive' structures appear to be less frequent and large: for example, the rampart of the Alto do Castro hillfort only protects part of the settlement's perimeter (Fig. 4).

Figure 1 Stone foundation of the Chao Sanmartín rampart (Grandas de Salime, Asturias) (after Villa 2002a).	Figure 2 Torroso (Mos, Galicia), stratigraphic section of the ditch and settlement platforms (after Peña 1992a).

Figure 3 As Croas (Pontevedra, Galicia), defensive wall (after Peña 2000).	Figure 4 Alto do Castro (Cuntis, Galicia), profile of the first occupation (after Parcero 2000a).

Figure 5 Penalba (Campo Lameiro, Galicia), stone foundations of the main settlement platform (after Álvarez 1986).

Earthworks, at times on a very large scale, usually accompanied the ditches that served to prepare areas intended for settlement. The best-documented example is the Torroso hillfort, where excavation work made it possible to register and date the construction sequence of an important earthwork to the south of the hillfort, revealing the presence of perishable occupational structures (hearths, huts made out of plant material, etc.) which would have been provisional domestic areas while the settlement was being built (Peña 1992a) (Fig. 2). This case illustrates how the preparation of the site for settlement was a complex process to plan and execute, in which two activities were of paramount importance: preparing a flat surface for permanent occupation, and digging a deep ditch, used, among other purposes, to define the limits of the settlement. A similar case appears in the Penalba hillfort, where the presence of a stone wall has been suggested (Álvarez 1986, 1987) although it does not appear that there are any arguments for proposing the existence of anything other than a similar set of earthworks to those found in Torroso, reinforced on their exterior with a retaining wall of large stone blocks (Fig. 5). Despite not having been excavated, it would appear that the same is true for Penido do Medio (Ramil 1995-96).

Fig. 6. General view of Penalba (Campo Lameiro, Galicia) as seen in the mid-twentieth century (after Álvarez 1986).

The construction of Early Iron Age settlements therefore appears to have been governed by the elemental principle of defining space, using structures that involved a considerable amount of teamwork. It may be said that human communities impose their presence by modifying space, and converting it into settlements. This statement may be further refined: as we will see, these hillforts were important due to being highly prominent and visible, particularly at medium and long distances. However, a system for preparing and defining domestic space based on earthworks and digging ditches only involves a moderate alteration of the existing space: the artificial constructions adapt and delimit a space that is already prominent, and the final result is a product in which the natural elements still serve as the most outstanding features, and enhance and emphasize the perception and importance of the site. This may also be clearly seen in the current profile of the Penalba hillfort, viewed from a close distance. The terraces only slightly affect the original appearance of the hill, although the focal point is still the hill in the centre (Fig. 6). The ditches are carefully constructed defining structures, highly noticeable at a short distance, yet practically invisible at long distances.

However, these structures not only define the space occupied by the settlement but also serve to isolate it. This is how we should interpret the fact that the most frequently found items are ditches. We do not have a great deal of information about ditch sizes, although in the two cases that have been examined in detail, Alto do Castro and Torroso, the dimensions exceed those required for merely physically defining space: they are no less than two meters deep, and equally wide (Fig. 7). This issue is clarified by cases in which up to three ditches are present, such as at Torroso (Fig. 8). The defensive function of the ditches is further strengthened by the fact that they are situated preferentially - sometimes only - in the most inaccessible and difficult to defend parts of the settlements. This is once again the case in Alto do Castro, whose first inhabitants dug a ditch with a wide stockade or small rampart along the arc running from north to southwest, the only area not defined by steep natural slopes (Parcero and Cobas forthcoming).

Figure 7 Alto do Castro (Cuntis, Galicia), stratigraphic section of the defensive complex. The thick line marks the surface of Early Iron Age structures.	Figure 8 Profiles of the site of Torroso (Mos, Galicia) (after Peña 1992a).

The settlements therefore appear fortified, at least to some extent. However, the complexity of these defences is relative. There are no examples of walls as such, of lookout towers or complex access systems, of any type of architecture that was clearly intended for fortification. Evidence of a defensive function does exist, although this is based on the combination of two elements that are not 'military' as such. Firstly, the main component that defines the inaccessibility to the site is, as we shall see later on, location. The only elements that reinforce the idea of inaccessibility are the ditches, highly advantageous structures in functional terms but not visibly impressive. A ditch may only be seen at short distance, and in any case is not an architectonic element that decisively influences visual perception of a site in the same way as a wall, for example. In summary, the fortification of these sites was clearly intended, although fundamentally based on making use of existing topographic features rather than the imposition of architectonic structures by the inhabitants.

Apart from this relative modification, there is another feature involved: the extension of the settlement and the topographic form over which it is constructed coincide completely. Early Iron Age hillforts are characterized by their simplicity of form, corresponding to an adaptation to a natural topographic unit. In nearly every case it is seen that these populations were constructed on small hills that marked the base as well as the limits of the settlement. The hill fort's limits were defined, strengthened and materialized through the artificial structures without undermining the previously existing natural landforms. In other words, the populations adapted to the site chosen for settlement, and not vice-versa. The topographic profiles from Torroso and Penalba provide a good example, with an even better illustration represented by the first level of occupation at Alto do Castro, from the Early Iron Age, which was limited to the upper part of the hill, largely defined by natural slopes, and completed by a ditch with a small earthen rampart. As the site grew from the Late Iron Age onward, the appearance of the hill was substantially modified to adapt it to occupational needs. Similar developments, though on a greater scale, took place at Troña (Hidalgo 1988-89), Sâo Juliâo (Bettencourt 2000) and Chao Sanmartín (Villa 2002a).

Another important implication of the formal simplicity of the settlement is that Early Iron Age populations had a unitary internal structure. The main problem in studying this issue in detail is the small surface area typically excavated on Early Iron Age hillforts. Excavations in sites that were reoccupied in the Late Iron Age typically do not provide much information about their spatial configuration prior to the most recent occupation. In these cases, archaeological work is usually limited to revealing the existence of these previous phases, partly due to the difficulty of having satisfactory access to them (due to deep, complex stratigraphies) but also due to the lack of interest in archaeological investigation in Galicia for anything apart from establishing chronological sequences (Fernández-Posse 1998). This means that the only way of investigating this issue is to use the results of excavations of hillforts that were abandoned at the end of the Early Iron Age.

Figure 9 Torroso (Mos, Galicia), internal terracing walls (modified after Peña 1992a).

There may also be some spatial differentiation as a result of a sequence of terraced platforms, although the limited area excavated allows only a superficial examination of these types of sites and there is no evidence for any vertical, positive separation of these spaces. The different terraced levels are stepped using containing walls, which made it possible to level the ground without impeding communication between them, either physically or visibly, as was the case at Torroso (Peña 1992a) (Fig. 9).

Figure 10 Sâo Juliâo (Minho, Portugal), group of huts and foundation of defensive wall (after Bettencourt 2000).

The apparent absence of internal settlement division could be confirmed by examining different types of domestic structures. The organization of internal spaces is based on fairly large, normally circular huts, with diameters of between five to seven or eight meters (more spacious than those from the Late Iron Age, as we will see later on). Examples of this may be seen at Torroso (Peña 1992a), Penalba (Álvarez 1986, 1987), Neixón (Acuña 1976), São Julião (Bettencourt 2000; Martins 1990), Toralla (Hidalgo 1990-1) and Alto do Castro (Parcero 2000a) (Figs. 10, 11). They are fairly far apart, with wide-open spaces between them, and are always isolated; none of them have any type of neighboring structures. Only in a few cases, such as a dwelling at Torroso, is it possible to see walls that do not appear to be enclosing any type of functional space (Peña 1992a) (Fig. 12). All of them clearly demonstrate habitational use with the appearance of central hearths that are more or less architectural. No types of constructions appear to have been assigned to or earmarked for uses other than habitation in the widest sense of the term, including the possibility of their use as stables. Only one of the three buildings excavated at Torroso has a more complex structure; a tentative interpretation of this feature as a communal structure recognizes that supporting data is largely absent (Peña 1992a) (Fig. 12). The structure has a hearth like the others, although very few objects were found.

Figure 11 Toralla (Vigo, Galicia), partial plan of two Early Iron Age huts (after Hidalgo 1990-1).

Figure 12 Torroso (Mos, Galicia), general plan of the excavated area (after Peña 1992a).

In any case, the internal spatial concept of these populations appears to have been dominated by open, common spaces. One of the most noticeable features is the lack of large-scale storage structures. Compared to the large storage pits found in many settlements from the Late Bronze Age (Jorge 1988; Lima forthcoming; Méndez 1994, 1998) (Fig. 13), no spaces for storage purposes have been found to date in the earliest hillforts. As mentioned, the buildings were always isolated, without ancillary elements, either buried or constructed, that could have served for storage. The most viable option, considering the data available, is that portable containers were used, either made of pottery or of perishable material, which nevertheless would not have had the storage capacity of a structure of any type. Here it is important to remember that the capacity of pits from the Late Bronze Age site of Monte Buxel (Pazos de Borbén, Pontevedra) was between 950 and 4200 litres (Jorge 1988; Lima forthcoming). However, there is a second option: there may have been silos, although in that case they must have been located outside the areas that have been excavated thus far. It is important to emphasize that a storage area of this kind would have been for communal use, something that does not change the original interpretation: the absence of familial structures for large-scale storage.

Figure 13 Late Bronze Age site of Monte Buxel (Pazos de Borbén, Galicia), sections of storage pits (after Lima forthcoming).

The location patterns of the settlements (Carballo 1990; Parcero 2000a, 2002a, 2002b) for this period focus on prominent landforms that are highly visible in the surroundings not only directly, but also within a much wider range with a radius of at least two kilometres (Fig. 14). This dominant position gave the inhabitants visual control over a wide area, controlling large tracts of land at a long distance (Fig. 15). However, and paradoxically, visual control over areas closer to these sites tends to be quite uneven and not particularly intensive. Visibility at short and medium distances is fairly discontinuous, and is not generic and homogeneous in all directions. Instead it gives priority to the visual dominance of a specific arc at the expense of other directions. There is therefore a clear duality in quantitative visibility with regard to distance (irregular and partial close up, compared to wide range at long distances) as well as qualitatively, concentrating on the visual domain of some directions at the expense of others.

Figure 14 Coto do Castro (Cotobade, Galicia), typical placement of an Early Iron Age hillfort.

Figure 15 Visibility from the Early Iron Age hillfort of Sta. Leocadia (Arteixo, Galicia) (red dot).	Figure 16 Accessibility to and from an Early Iron Age hillfort; 30 and 40 minutes isochronic lines from Penalba (Campo Lameiro, Galicia).

Access conditions to the surroundings also feature a situation of duality, although opposite to that of visibility conditions (Fig. 16). In this case it is also possible to see a common opposition between a specific sector of the surroundings that is easily accessible and outlying areas in which mobility is difficult. However, the location of these zones is usually opposite to that of visibility: in this case the most easily accessible land is in the opposite position to the direction in which visibility is highest and covers the largest area.

Figure 17 Productive potential of the area surrounding Penalba (Campo Lameiro, Galicia).

Finally, the Early Iron Age model of site placement is characterized by locations with a predominance of light, shallow soils, generally well drained, with slight or moderate slopes. It is also normal to find areas that are difficult to make use of, or that are even non-productive (Fig. 17). A second important feature of this location model is the fact that as we move away from the settlement, the variety of environments and their possible uses increase considerably; there is normally a strong contrast between the variety of soils available at 800 meters and two kilometres from the hillfort (intervals analysed in Parcero 2000b, 2002b), as well as the percentage weight of each type of terrain in both intervals. It does not appear that the immediate and easily accessible presence of the most suitable land for agriculture was an essential factor in the selection of locations for these settlements (Fig. 18).

Figure 18 Proposed landuse in the vicinity of Penalba (Campo Lameiro, Galicia).

3.1.2. Inside the settlements: material culture

Pottery is perhaps the type of material culture in which similarities and differences throughout this period can be best observed. Early Iron Age pottery exhibits a number of characteristics that distinguish it significantly from the preceding two phases, which had much more homogeneous ceramic production traditions (Cobas and Prieto 1998). These similarities and rupture points, which summarize and represent patterns for the whole range of material culture, may be seen in different aspects of the Technical-Operative Chain (TOC) (Fig. 19). In the Early Iron Age we see the following characteristics compared to subsequent periods (Fig. 20).²

Raw material. The way the clay was collected and prepared resulted in the predominance of mica inclusions in the clay body, either as a result of selecting for a certain type of clay that was naturally high in mica inclusions, or in the preparation of the clay by adding this type of temper. It is also possible to see that the clay was not very well wedged, as indicated by the coarse angular nature of the inclusions and the abundance of pinholes noted in the fracture.

Figure 20 Basic features of Early and Late Iron Age pottery styles.

Range of forms. Modelling is characterized by the absence, or infrequent appearance, of flexed compound profiles, with edged profiles the most documented (Figs. 21, 22, 23).³ Ceramic vessels in this period were always hand-built. There is no evidence for pottery with marks indicating the use of the potter's wheel during the production process and the irregular distribution of the temper grains within the fracture does not suggest that the pottery was wheel-thrown.

Figure 21 Early Iron Age pottery; characteristic forms and decorations.	Figure 22 Early Iron Age pottery; characteristic forms and decorations.	Figure 23 Early Iron Age pottery; characteristic forms and decorations.

Finish. Smoothing, brushing and burnishing are all documented, as they continue to be in subsequent periods, although a distinctive feature of this period is the absence of spatula marks and the abundance of pieces with a granular texture.

Decoration. There are also a number of differences in the decoration as compared to the two later periods, both in the designs used and in the techniques applied. The only type of design documented to date is rectilinear geometric decoration, characterized by its simplicity and the limited variation in the decorative elements and motifs, resulting in a highly homogeneous range of decoration. In terms of decorative morphology, the differences compared to later periods are found in the treatment used for the decorative elements (always geometric and rectilinear, isolated from one another, and generally much larger than in later periods), in the treatment used for the motifs (in this phase we frequently see figures and backgrounds treated to give an effect of flexibility to the contour, and the organisation of motifs in wide bands or metopes), and even in the way in which decoration is organized with a single decorative motif located in the most visible part of the pot, with more complex versions using a repetitive symmetrical rhythm. This decoration is always simple, with the element and the motif, and sometimes even the decorative scheme, the same in some cases. These are chiefly located on the upper third of the vessel body and may be poorly delimited by fine lines that disappear in some parts of the perimeter or even make use of the morphology of the vessel (e.g. edges in the transition zone between the body and the neck of the vessel). Finally, with regard to decorative techniques, the use of stamped decoration is not found.

Firing techniques. Unlike the following two periods, firing in a reduction atmosphere dominated during this period, with firing temperatures fluctuating between 800º and 1000º.

Final product. In the final product, there are no important differences between decorated and undecorated items, as the same TOC was used for both groups.

The concept of ceramics in the first phase was therefore governed by a sense of binary opposition or duality mainly represented in the morphology of the piece and the decorative concept used. In morphological terms, apart from some minor variations, two main groups of pottery can be identified: simple profile forms and compound profile forms. The decoration used typically is located on the most visible part of the pot (revealing an opposition between the decorated and the undecorated surfaces) and on the external surface of the vessel, showing an opposition between the visible decorated surface and the invisible undecorated surface.

This binary opposition is not only seen in the simplest relationships, but can also be glimpsed within a more complex network of relationships encompassed in the TOC. In the case of the opposition between block-modelled compound profiles vs. single-modelled simple profile, we find a number of subdivisions, although these once again are the result of a binary opposition. For example, within the compound profile category we find a morphological opposition between pottery with very low rims and pottery with high rims, and in pottery with a simple profile we find an opposition between open and closed shapes. Likewise, differences are found between delimited decoration and non-delimited decoration, or between simple and composite decoration. However, a similar basic technological treatment is seen for all ceramics, despite the existence of differences that make it possible to refer to possibly independent TOCs.

The general appearance of formal similarity and simplicity may be extended to other types of material culture. Metalwork from the Early Iron Age shows a very clear formal continuity with what some authors call the bronze tradition of the Late Atlantic Bronze Age (Calo and Sierra 1983). Bronze is in fact the most widely used metal, as there are very few iron items from this period, and in general these are in the form of barely recognizable remains. Evidence of bronze metalwork is fairly common in settlements (crucibles, slag heaps, etc.), apparently indicating basically local production on what was certainly a small scale. However, this is in contrast with the presence of object types found over a wide geographical area that in general offer few novelties in shape from the characteristics of the Late Bronze Age. One example of this is the appearance of fragments of palstave and socketed axes in hillforts such as Torroso, Neixón Pequeño or A Peneda do Viso (see Teira 2003). The rest are nearly always objects used for personal adornment (brooches, necklaces, hoops) or of little daily use, and perhaps can be more easily related to other areas of social action (Fig. 24), as for example weapons (a dagger found in Torroso, a spearhead in Penalba) or the recent discovery of a large disk made of bronze and wood in Chao Sanmartín (Villa 2002a)(Fig. 25). There are also rare fragments of pieces that may be connected with community functions, such as riveted bronze cauldrons (Armada 2003).

Figure 24 Selection of bronze objects documented in Torroso (Mos, Galicia) (after Peña 1992a).	Figure 25 Chao Sanmartín (Grandas de Salime, Asturias). Wooden and bronze pieces of a big cicular object (shield?) (after Villa 2002a).

In reference to the way in which these objects were manufactured, other authors (Calo and Sierra 1983) previously suggested that the progressive increase of ternary alloys, with an increasing presence of lead, resulted in the production of objects with little or no functional value, due to their weakness as a result of large amounts of lead in the alloys used to make them. In fact, the tendency to make the bronze alloys go farther, supported by the analysis of pieces from other sites such as Torroso (Peña 1992a) or Sâo Juliâo (Bettencourt 2000, among others) represents, as we shall see later on in this article, one of the most widely-used arguments for historical reconstructions of the period. It is based on the presumed need to cope with the scarcity of raw materials, especially copper and tin, caused by large-scale raw material exchange cycles in Western Europe as a whole (Calo and Sierra 1983).

Figure 26 Early Iron Age gold torques (after Silva 1986).

Apart from ceramic and bronze items, there are relatively few materials from the Early Iron Age. Boat-shaped quern stones were still in use from the start of the period, and are fairly common in all of the hillforts. Imported materials appear as only singular objects, such as fragments of arybaloi found in O Neixón Pequeño, As Ermidas and Castroverde (Naveiro 1991), although the northwestern part of the peninsula appears to have been left out of the regular and systematic circulation routes for objects of this kind (Naveiro and Pérez Losada 1992: 67). In any case, to date only very few examples have been found. The social need for reliable and regular importation and use of exotica does not appear to have existed, at least based on the currently available record. The same is true of pieces made of gold or silver. The study of this type of material culture in the northwest has always been an eminently formal and typological question, due to the difficulty of dealing with it in any other way. Most of the pieces found are without context, and at times even from unknown locations. This not only makes it difficult to attempt to analyze patterns of distribution and deposition, it also complicates the process of chronologically dating the material. In a recent investigation (Ladra 1999a), a sample of gold torques from the Early and Late Iron Age in the northwest was examined. The criteria used were both formal and typological, with the products from the Early Iron Age objects being typically massively cast, with hardly any decoration and with square cross-sections (Fig. 26). Although the Ladra study does not indicate what percentage of the total number of currently documented pieces was studied, it is interesting to note that only nine torques are listed as belonging to this stage of the Iron Age, with a total weight of less than 700 grams. This quantity, in addition to the previously mentioned formal characteristics, would appear to contradict the idea that a fully developed gold and silver industry existed at this time. There is no widespread evidence from any contexts at this time for the production or systematic use of precious metals or for the use of other objects of value, such as imports.

3.1.3. Beyond settlements: environment, production and territoriality

Thanks to a series of studies from the 1980s and 1990s aimed at analyzing territory and landscape, it has been possible to obtain additional information about production processes and the way in which the environment was exploited. It has therefore been possible to extend the data derived from the archaeological record of hillfort excavations, which is not particularly extensive due to the preservation problems previously referred to in this article.

However, some relevant evidence, including paleoecological information, reveals that large-scale deforestation took place as the landscape was adapted to growing crops, mainly cereals. This process is clearly documented for these early stages, as shown by analyses conducted around the hillfort of Penalba (Aira et al. 1989: 80). A further example is the Alto do Castro settlement (Parcero 2000a; Parcero and Cobas forthcoming). Pollen analysis of the paleosoil found under the first phase of construction of the defences shows how the settlement was founded in a landscape characterized by the presence of woodland that gradually changed to a more open vegetation pattern in which grasslands were beginning to predominate. The analysis also showed a significant percentage of oak pollen, although indicators of human activity in the surrounding area (at a time when the settlement was being founded) begin to appear.

This process of deforestation would have continued throughout the final millennium BC, causing what has been referred to as the passage from a "Landnam" to a "cultural steppe" landscape (Ramil 1993: 53), a modification that has had global effects, and can be seen throughout the northwest at any given moment from this period of time on. This indicates the development of a pattern of intensive and constant work in clearing land for exploitation by the human communities, which must have led to changes in the production system, as compared to the more mobile and extensive subsistence strategy of the Bronze Age (Méndez 1994). It has also been demonstrated that this progressive deforestation, together with changes in the climate at the time (particularly an increase in rainfall) would have been an important agent in initiating significant erosion processes (Rodríguez López et al. 1993: 287-8).

This progressive increase in human activity is associated from the start of the period with the development of a subsistence strategy based on cereal cultivation. The seed and fruit remains reveal the presence of a consistent spectrum of cultivated species throughout the period. Millet is frequently found in sites occupied at an early stage, such as Penalba or Penarrubia (seventh through sixth centuries BC)(Álvarez 1986, 1987; Arias 1979) as well as in other later sites such as Montaz (Carballo and Fábregas 1991). Barley also appears at early hillforts such as in the early stages of Castrovite (Carballo 1998), as well as in later stages of occupation at Cortegada or Montaz, for example (Carballo and Fábregas 1991). The appearance of naked barley seeds in the last stage of occupation of the Alto do Castro site, dated from the first century BC (Parcero and Cobas forthcoming), is significant due to the primitive nature of this cereal and the generalized abandonment of its use in Europe towards the end of the Bronze Age (Buxó 1997: 193 and ff.; Cunliffe 1991: 372; Ruas and Marinval 1991: 421).

Wheat appears in both early and late occupied sites. Naked wheat forms (without husks) appear in equal amounts in early and late-occupied sites. Naked wheat forms are present in the earliest hillforts, such as Penalba, through those dating to the period of Roman occupation, such as Vigo, revealing that these wheat forms were introduced at an early date and were exploited for a long period. Spelt or husked wheat appears to have undergone the same process in Europe as naked barley: the fact that these grains were less suited for making bread and the difficulty involved in grinding them meant that they were progressively abandoned towards the end of the Bronze Age and gradually replaced by naked wheat (Barker 1985; Buxó 1997; Jones 1996).⁴ However, in the northwest there is evidence that these types of wheat continued to be used throughout the whole period in which the hillforts were occupied, as they appear in both early-occupied sites (Penalba) and sites occupied in the Roman period (Alto do Castro or Vigo). The replacement of these species with naked wheat occurs at a later stage than in other parts of the Iberian Peninsula or Europe; in any case, traditional Galician agriculture definitely moved toward the use of unhusked varieties, as husked wheat is completely absent from sites of this type (Sahuquillo and Fraga 1991). In the case of leguminous plants, only beans are represented to any great extent, without any clear chronological tendency. Species such as vetch, which were mainly used as fodder, did not appear in France, for example, until the Roman period (Ruas and Marinval 1991: 415-7).

The carpological record appears to show that a wide range of plant species and subspecies were available and exploited throughout the Iron Age. There does not appear to have been a clear pattern of evolution and replacement of one group of species by another throughout the period in which the hillforts were occupied, although the sample is too small to confirm this. What is evident is that from the first moment the hillforts were occupied, agricultural production was both diverse and complete.

Evidence of faunal exploitation is documented in studies like those by Penedo (1988) and Rodríguez et al. (1993). Three species are traditionally represented: sheep/goat, various bovine species and pig. Other species appear periodically, both wild species, such as wolves and foxes, and domestic species, including horses, chickens and dogs, although most of these appeared sporadically and at later stages. Focusing on the first group, there is a clear imbalance between the relative importance of each. Ovicaprids seem to have been the most abundant species in number, although if we consider their importance in terms of meat volume (in proportion to the weight of the others), bovine species dominate (Fernández Rodríguez et al. 1998: 180; Rodríguez López et al. 1993).

What is clear is that the main components of the herds kept were cows, goats and sheep. It is difficult to be precise about the temporal evolution of this situation, due to the lack of context for many of the known remains. However, it is possible to advance our knowledge of how each of these species was exploited by exploring the data available regarding the age of the animal when slaughtered. There is a fairly clear pattern indicating that bovine species were slaughtered when adults. This means that they were not just bred for meat, but also would have provided two other important resources: milk and muscular energy as work animals. In contrast, ovicaprids appear to have been slaughtered in the early stages of their adult life (two to four years of age), indicating that they were used primarily for meat (Penedo 1988; Rodríguez López et al. 1993). This tendency is seen even more clearly in pigs, which were slaughtered at very early ages, before reaching the age of 18 months, with bone remains showing clear evidence of having been butchered (Rodríguez López et al. 1993: 294).

Despite the fact that the types of plant and animal species exploited throughout this period do not appear to have changed greatly during the Iron Age, the specific ways in which the environment was used did undergo modifications. In the Early Iron Age the hillforts were surrounded by light, shallow, well-drained soils that were at risk from erosion if worked continually (Figs. 17, 18). The availability of areas suitable for long-term cultivation, alternating between short cycles and constant yields, was at medium or long distances from the hillforts. The immediate surroundings of the hillforts, particularly the areas that were most accessible from the settlement, were totally unsuitable for this kind of use. These types of settlements appear to conform to a productive landscape model in which the emphasis is placed on a basically extensive agrarian operation, structured around cultivating surfaces subjected to long alternating cycles. In this scenario, guaranteeing the reproduction of the group is dependent on having a wide range of resources available, as the productivity per unit of land exploited is lower, with some plots lying fallow to recover their fertility, and to control erosion.

This means that permanent fields will be either non-existent or highly reduced in number, relegated to small-scale cultivation of specific secondary species (like allotments), or will be situated at some linear distance from the settlements, and above all at some significant distance from them. Such plots not only appear at distances of more than one kilometer from the settlements (and often more than two kilometers), but also would have been difficult to access from inhabited areas, meaning that the linear distance between settlement and arable land would have taken more than half an hour to walk (Parcero 2000a). This second option is not impossible, although it is a less probable scenario. The first, and more likely, model proposes that the inhabitants of this group of settlements would have based their subsistence on a type of agriculture that was varied and probably produced constant yields (with two yearly cereal harvests), distributed among several productive plots that were large in size and were extensively exploited.

The model of occupation from the Early Iron Age on appears to represent a break with the Bronze Age system, but it does not yet exhibit the definitive formation of a new type of landscape (Méndez 1994, 1998). It is true that changes did occur, such as the appearance of millet as a crop, making it possible - whenever environmental conditions permitted - to obtain two yearly harvests, complementing the winter cereal harvest. However, this dual system appears to have been developed with Bronze Age technology with extensive, long-cycle exploitation periods punctuated by long fallow periods alongside the cultivation of other crops. Instead of immediately adopting a permanent field system, it would appear that the dual system of winter and summer cereals was adopted in the least radical way, with the separate cultivation of both species using similar technology. This was an important qualitative leap, but does not constitute a definitive break with the Bronze Age subsistence strategy.

3.2. Late Iron Age

3.2.1. The "real estate": the forms, patterns and structure of settlements

Communal labor within the social group in preparing a site for habitation was still a key feature at this time. However, unlike the Early Iron Age, this labor focused on different products and involved a different type of effort.

The first relevant change is seen in the location patterns of many of the settlements (Parcero 2000b, 2002a), which ceased to be situated in positions such as those we have described for the Early Iron Age. Although there was still a preference for prominent landforms, the further one moves away from the settlements, the less visible they become (Fig. 27). This means that the relative altitude with regard to the immediate surroundings (800 meters) is usually positive, at times markedly so, but if we consider the wider geographic radius (two kilometers) this descends to the point where the relief is neutral, and in many cases negative. Similarly, visibility is normally very intensive in the areas closest to the settlement, but only reaches medium or low values at long distances. The visual domain over surrounding areas is rarely lower than 50%, and is often far higher; however, at long distances this is much more discontinuous and quite restricted (Fig. 28). Apart from this different visual concept in terms of distance, the range of angles of visibility at these sites also differs with regard to the previous period. The characteristic view shed of these hillforts is always circular, with a very even distribution of lines of visibility from all areas of the site.

Figure 27 O Peto (Verda, Galicia) typical placement of a Late Iron Age hillfort.

The access territories of these hillforts are normally much larger than in the previous period. However, the real difference is in the positioning and location of these areas of greater accessibility, as this time their distribution matches their visibility: they are distributed in a circular fashion around the settlement, and reaching them from the hillfort is equally easy (or difficult, depending on the point of view) in any direction (Fig. 29). This means that the coincidence between visible and accessible areas is generally very high.

Figure 28 Visibility from the Late Iron Age hillfort of Lagouzos (Friol, Galicia) (red dot).	Figure 29 Accessibility from a Late Iron Age hillfort; 30-45 minutes isochronic lines from Prado (Friol, Galicia).

Figure 30 Productive protential of land surrounding Praderrei (Campo Lameiro, Galicia).

Finally, there is also an important change in the types of land surrounding these new settlements. In the Late Iron Age there is a predominance of deep, fertile soils with a potential for more intensive use, lower incidences of summer droughts, less risk from freezing, and gentle slopes (Fig. 30). This change is even more significant if we consider that by this period the hillforts were much more integrated within their surroundings, with better conditions of access and visual domain and more significantly with a special connection in terms of visibility and access to this type of land (Fig. 31).

These changes were accompanied by new construction methods. Hillforts were no longer strictly adapted to the conditions of the chosen site, but instead the natural setting was manipulated according to the needs of the group who were going to occupy it. The division and delimitation of domestic space was still an essential structural feature, although this now had a different appearance. The selection of the location, as we have seen, modified the criteria involved, and as a result the positions occupied are less visible in topographic terms. The limits of the area occupied were no longer clearly marked by the terrain, but instead the terrain was considerably modified. This is visible both in the quantity of earthworks involved in building structures to define the settlement, as well as in the formal preferences for these structures.

Compared to the predominance in the Early Iron Age of ditches and terraces, the most characteristic features of Late Iron Age hillforts are ramparts or low walls. Rather than delimitation based on negative structures, now positive architectonic structures were used that emerged from the landscape (Fig. 32). This meant that the creation of a settlement visibly and permanently altered the formal appearance of the area, introducing new structures that were visible at a great distance (Fig. 33). However, it also meant that the limits of each community's space became much more tangible. By this time both movement and perception were limited. It may be presumed that the interior of each hillfort was hidden from sight, its contents visibly defended. However, the reverse is also true: the settlements are now constrained spaces, with a limited possibility for visual escape.

Figure 32 Castro Montaz (Silleda, Galicia). Section plan of defensive wall (after Carballo 2002).	Figure 33 Castrolandín (Cuntis, Galicia). General view of ramparts and walls.

We not only observe a change in the way settlements were defined, but also in the degree to which this was done. Enclosing structures increase in number, and complexes with less than one rampart and ditch are rare. The development of multivallate sites became generalized, and in particular the access zones to the settlements became more complicated. A good example of this kind of complexity may be seen at the Fozara hillfort (Ponteareas, Pontevedra), which was occupied approximately between the fourth century BC and first century AD (Hidalgo and Rodríguez Puentes 1987)(Fig. 34). It is also important to consider the possibility of other significant structures that may no longer be visible on the surface, including groups of defensive ditches and linear trenches, as for instance in the surroundings of the site of Follente, documented in the Galicia Gas Network archaeological project (Parcero 1999)(Fig. 35).

Figure 34 Fozara (Ponteareas, Galacia). General outline (after Hidalgo and Rodríquez 1987).	Figure 35 Follente (Caldas de Reis, Galicia). The "before and after" views of the site.)

Fig. 36 Alto do Castro (Cuntis, Galicia). Reconstructive plans of the Early and Late Iron Age phases.

This increased activity in dividing up space affected the inhabited area as well, which was also internally fragmented. It is true that hillforts with a single enclosure and simple structure are still found in the Late Iron Age, but complex structures are common. The most frequently used layout was a central area normally situated on a higher area, defined by a rampart that ran around it, with additional enclosures following the slope of the hill, in turn enclosed by ramparts and different levels. Alto do Castro is a good example, especially when the site's Early and Late Iron Age occupations are compared (Fig. 36). Although only a small amount of data is available, it seems possible to extend this intentional, visible and architectural division to the productive space, as revealed by structures found in the vicinity of the Follente hillfort (Parcero 1999).

This sense of division was also extended, in a very significant fashion, to the specific structuring of domestic space in each settlement. Fragmentation is extreme, with buildings joined together with hardly any free spaces to make circulation easier (Fig. 37). The image of division is further emphasized by the fact that all domestic structures at this time were made entirely of stone, compared to huts made with perishable materials or stone flooring and wooden walls in the Early Iron Age. This means that the division of internal space was not only more extreme, but that it was monumentalized to ensure that all of the constructed elements lasted through time.

The internal layout of these settlements has traditionally been interpreted as an indication of architectonic anarchy, with a complete absence of any type of spatial organisation. However, a close examination of the situation makes it possible to detect a reiterated logic in the aggregation of structures to form what have come to be called 'domestic groups' (Peña 1989; Silva 1986). Dwellings are no longer separate and autonomous, but are instead joined together in groups of several buildings, generally surrounded by enclosing walls.

This type of spatial organisation was dealt with in some detail in a publication by A. Romero (Romero 1976), although more recently it has been the object of more specific and detailed analyses, also in pre-Roman settlements (Fernández-Posse et al. 1994) (Fig. 38). In these settlements it was observed that the groups of structures are very similar, with a main building containing the hearth, and at least one additional building, probably used for storage. These groups of structures are repeated over the whole surface area of the sites, established as autonomous groups closed in on themselves. The different ways in which these spaces were used may be clearly seen in the detailed microspatial analysis of a small group of buildings conducted at the Os Castros site in Toques, A Coruña, which made it possible to identify the presence of zones dedicated to storage, cooking, etc. (Penedo and Rodríguez Puentes 1995)(Fig. 39).

Figure 38 Cabo do Mar (Vigo, Pontevedra). Partial plan of domestic structures (after Gonzalez Ruibal 2003).	Figure 39 Os Castros (Toques, Galicia). Multi-functional spaces in a single "domestic unit" (after Penedo and Rodríguez 1995).

It is highly relevant that in some cases walls enclose the free spaces between the different structures, giving the domestic groups a very strong sense of unity and isolating them from the exterior. This was demonstrated by the recent reconstruction work carried out in the Indigenous-Roman hillfort in Sanfins, under the direction of A. Coelho Ferreira da Silva (Fig. 40). This reconstruction is a good illustration of the degree of spatial division involved in an organisation based on this type of 'family group' structure.

It may be suggested that structures of this kind only correspond to phases of occupation under Roman occupation (and influence). In fact, most traditional interpretations of this phenomenon have assumed that such structures are simply one more Roman-influenced structural change (e.g. Calo 1993). In an attempt to formalize this relationship, it has been proposed that these domestic groups be referred to as 'courtyard houses', basing their origin on an indigenous adaptation of classic architectonic styles (e.g. Acuña 1996). In fact, some of the examples we discuss here (Os Castros or Sanfins) do correspond to periods of occupation around the first and second centuries AD. However, the archaeological documentation available from other sites makes it possible to reject this theory, as revealed by evidence from the sites of Corona de Corporales (Sánchez-Palencia et al. 1985), San Juan de Paluezas (Fernández-Posse and Sánchez-Palencia 1998), and Chao Sanmartín (Villa 2002a), among others. At these sites the existence of such structural aggregations as the basic element used to organize the settlement has been documented. There is also a recurrence in the number and function of available spaces, always containing at least two basic elements: one for habitation and another for storage. A third aspect that appears at all of the sites is the tendency to isolate each unit from the others, as indicated by the direction of the entry doors into each group, which never open out into common areas, but instead face in individual directions (Fig. 41).

Figure 40 Sanfins (Paços de Ferreira, Portugal). Reconstruction of a "domestic unit" from the Indigenous-Roman period (after Silva 1996).	Figure 41 San Juan de Paluezas (León). Location and visual range available from the entrances to the other "domestic units" (modified after Fernández-Posse et al. 1998).

One of the most significant aspects of this type of spatial organisation is that in obvious opposition to the Early Iron Age, each domestic group represented a wide range of socio-economic functions. Apart from living spaces, the domestic units contain complementary spaces, the most important of which are the storerooms. These are solid, stable constructions made of stone, with doors high up in their walls to prevent the entry of damp or other elements that would have posed a risk to preserving grain. On the one hand, this implies a predictable increase in the level of agricultural production. On the other, it means that surpluses were stored and individually managed by each family group.

3.2.2. Inside settlements: material culture

In this period a series of innovations appeared that were maintained until the end of the hillfort period, throughout the developments of the Indigenous-Roman period. These innovations represent a complete break from the Early Iron Age (Fig. 42).

Raw material. Mica became less important as an element in temper visible on the surface of the pottery, with quartz generally in use as the main temper type. In some pottery it is possible to identify a more detailed clay preparation process with smaller, more rounded temper elements distributed more evenly and at times with slight direction in the fracture.

Range of forms. As regards modelling, restricted pottery forms with composite, flexed profiles appear for the first time and gradually come into general use. Occasionally the use of the potter's wheel is documented as a manufacturing technique.

Finish. With regard to the finish, the spatula becomes the more frequently applied finishing technique.

Decoration. There are thematic, technical and morphological changes in decoration type. Curved geometric elements appear for the first time in this period. It is also possible that this is when the first types of figurative decoration were developed, although the lack of context for many of the pieces on which this type of decoration is found ⁵ makes it impossible to confirm at this time that they are pre-Roman in origin. As regards morphology, the decorative elements grew smaller and were frequently combined with one another in order to create decorative motifs organized in a continuous series around the circumference of the vessel, with a diversification in the composition used for motifs (with continuous lines, strips and medallions used together in borders and metopes). The concept of spatial order in the decoration used also underwent a profound change in this period: whilst maintaining the characteristics of the previous phase for specific vessels (Fig. 43), the decoration extends now to different parts of the vessel (rim, neck etc.), instead of being limited to the most visible part of the object as in the previous period. Integral and complex decoration appeared, mainly as a repeated pattern that clearly distinguished the different parts of the pots ⁶, making use of a clear definition between the different motifs using horizontal lines (Fig. 44), which are very abundant in some cases and not only separate but also divide the motifs into 'compartments' that are isolated from each other. The successive rhythm means the decoration must be 'read' in a vertical direction; this is also the first time that vertical and horizontal directions appear on the same piece, combined with the relative visibility of the decoration on the upper part of the rim or on the inner edge. Stamped decoration is the main decorative innovation of this period (Fig. 45).

Figure 43 Late Iron Age pottery; characteristics and decorations.	Figure 44 Late Iron Age pottery; characteristics and decorations.	Figure 45 Late Iron Age pottery; characteristics and decorations.

Firing techniques. Firing in the Late Iron Age was mainly carried out in an oxidizing environment, and dark brown becomes the predominant colour of the clay body within a range of dark hues. No relevant differences are documented in firing temperature compared to the previous period, which still continues to range between 800 and 1000 degrees.

Final product. Vessels with stamp marks as the main decorative technique show a much greater degree of care and detail in their production than other types, at least visually.

We therefore see that a strong break affected nearly all of the TOC (Fig. 42). However, the main features were maintained, including those that were less susceptible to change (gathering clay locally, hand building of vessels, firing techniques and temperatures), together with some of the basic features, as the appearance of new ways of conceiving the morphology and decoration of the pieces does not imply the total disappearance of previous forms. Simple decoration was still used in wide, undefined borders, for example, but is limited to more crudely produced forms already seen in the previous phase that were possibly restricted to domestic use.

These formal modifications are included within what may be defined as a general change in the ceramic style of the hillfort culture. At this time the binary opposition between simple and compound pottery is provided by compound pottery in general, and vessels with simple profiles and reinforced edges, which incorporated important innovations in their decorative concepts. A new factor was superimposed on the previous duality resulting in a ternary series. This ternary combination may be seen in different aspects of the TOC, the most representative of which are detailed below.

• The most visible example of this new component is seen in the shape of the pottery. Here the opposition between simple and compound shapes becomes a series of oppositions between simple forms, compound edged forms, and flexed compound forms.
  
  
• In the previous phase it was also possible to observe a difference between the visible part of a decorated piece and its invisible undecorated part, while in the Late Iron Age relative visibility is characterized by placing decorations on the upper and interior part of the rim. The opposition between visibility and invisibility changes to a relationship between visibility, invisibility and relative visibility.
  
  
• Furthermore, there is a new rupture point in the opposition between the decorated and undecorated sections of the pottery. In this period decoration is not only zonal (dividing the vessel into two zones, one decorated and the other undecorated) but may also be integral (dividing the vessel into different decorated zones). This means that the opposition between different decorated areas is added to the opposition between decorated and undecorated parts.
  
  
• In contrast to the binary opposition between simple and complex decoration, at this time a ternary opposition is introduced between simple decoration, composite decoration (either vertical or horizontal "reading" directions) and mixed composite decoration (vertical and horizontal "reading" directions combined in the same object.
  
  
• It is also possible to observe a new element in the relationship between decorated and undecorated pieces. In the previous period the difference between the two was only seen in the finished product, not in the rest of the TOC, whereas in this period a new decoration (stamp marks) was introduced that implied a different technological treatment. The relationship between decorated and undecorated pottery disappeared, changing to a relationship between undecorated pottery, decorated pottery and specific decorated items using stamp marks and mixed composite decoration.

This classification system, based on ternary series, reveals itself to be a system of complex relations and combinations of choices within a wide range of varieties, within each phase of the decorative TOC. However, all of these differences are clearly brought together in an innovative morphology (pottery with flexed composite profiles) and decoration (stamping). This is a very specific type of pottery that is not incorporated as another variation, but instead as an advance over the previous scheme, in the sense that while the oppositions described for the previous phase had a consistent point of union in largely maintaining the same technological treatment, in this third category, specifically that represented by pottery with a flexed composite profile and a mixed decorative pattern, this point of convergence is lost as a different technological treatment was used.

This situation is given even more emphasis considering the ambivalent nature of stamp marks as an element both of cohesion and rupture in the hillfort culture style. Stamping was an innovative element, and represented a departure from previous practices in the northwestern peninsula, not only in terms of earlier cultural periods, but also with regard to the first phase of the Iron Age, as it does not appear until the second phase (see Rey Castiñeira 1990-91 for chronological discussion). However, it also establishes an element of cohesion and standardization in relation to a wider Iron Age context, as "this same phenomenon did not appear in isolated areas, but instead may be seen throughout the whole Peninsula..." (Rodríguez Puentes 1986: 241).

The use of stamp marks constitutes a differential factor not only in a temporal dimension, but also in a purely formal dimension, as it represents the implantation of a new decorative technique as well as a different concept of decoration, of the pottery itself, and ultimately a new conception of the TOC. Pottery with stamped decoration also introduced an exception to the homogeneity previously seen between decorated and undecorated ceramics in the hillfort style, which in turn perhaps reflects a different symbolic content. This may be derived from its connection with other types of material culture, as indicated by Rey Castiñeira (1990-91: 151), for example the similarity between vessel shapes and metal models.

The modifications in ceramic production extend to innovations observed in other types of material culture. In the case of metalwork, there was a widespread development and generalized use of iron utensils, which were also produced locally, as indicated by the presence of metalworking workshops in several settlements, such as Borneiro (Romero 1992), A Forca (Carballo 1989) or San Juan de Paluezas (Fernández-Posse 1998). In fact, these workshops appear to mark a standard of local production that may be generally applied to all of the settlements. The indirect evidence for metalworking is more numerous in all cases, with the appearance of crucibles, slag heaps and molds.

Bronze working continued to have a strong presence, mainly focused on the production of objects for personal adornment, including large quantities and varieties of fibulae (Cortegoso 2000) (Fig. 46). Iron appears to have been used more for functional elements such as picks, agricultural implements and nails (Romero 1992), whereas bronze was used for more complex and less functional objects, such as cauldrons (Armada 2003) (Fig. 47). Weapons are not particularly frequent but were certainly present, represented by the production of daggers made of iron, bronze or both (Carballo 1989) (Fig. 48). In any case, the chronology of some of the decorative elements on these pieces is still debated by some authors because most of them (particularly the weapons) do not come from closed contexts.

Figure 46 Viladonga (Castro de Rei, Galicia). Characteristic Late Iron Age fibulae (after Arias and Durán 1996).	Figure 47 A Peneda do Viso (Redondela, Galicia). Fragments of bronze cauldrons (after Armada 2003).

Figure 48 Daggers are the most characteristic Late Iron Age weapon (examples from Chao Sanmartín and Taramundi [after Villa 2002a]).

There is a similar on-going debate regarding gold and silver working, with some authors proposing very late dates for such materials, arguing that they do not appear until the first century AD (Peña 1992b). However, different readings of the material are possible based on the archaeological evidence, such as the length of time some pieces of this type were in circulation. Examples include repaired objects like the three torques from Lanhoso, the diadem from the Bedoya treasure and the pendant earring from Vilar de Santos (Pérez Outeiriño 1989). In other words, while they may not have become part of the archaeological record until the first century AD, it quite possible that they were produced and used considerably earlier. Stylistic similarities between gold and silver metalwork and some other types of materials, such as ceramics, have been well documented in stratigraphic terms (Cobas 1999) and more definitively based on datable pieces like the pendant earring from A Graña (Meijide 1990). In fact, as recently indicated (Ladra 1999b), many of the objects from Roman period contexts are found either hidden or fragmented, making it possible to consider two opposing processes: an attempt to safeguard them because of their social value, or simple loss. Both processes appear to contradict the idea that there was continuity in the manufacturing of pieces of this kind.

Gold and silverwork consists mainly of personal ornament: torques and pendant earrings are the most common gold and silver objects, although there are also examples of necklaces, bracelets and diadems (Figs. 49, 50, 51). Considering that most of these pieces come from insecure contexts, it is not practical to use them for an analysis of differential distribution or of locational criteria in domestic spaces. However, there are some ways of revealing at least part of their importance, such as quantification. Following the recent work of X.L. Ladra (1999a), even with the previously mentioned exceptions, it is possible to obtain the following data: in the case of the Late Iron Age, this author documented 51 gold torques, with a combined weight of 13.8 kilograms (without taking into account other types of objects). These amounts are not particularly impressive in absolute terms, but they are significant when compared to the small amount of gold known from the Early Iron Age (nine pieces at 700 grams). The pieces from the Late Iron Age are, unlike the Early Iron Age objects, very delicate works of craftsmanship, created with great attention to detail, in which the external appearance and form are nearly as important as the material itself. In fact, many of them are only gold plated, with their interior made of other metals. The concern for the appearance of gold objects is comparable to the case of the stamp decorated pottery, which is not accompanied by technical advances in the manufacturing process (firing temperatures, modelling, etc.) in spite of being meticulously produced with regard to external appearance.

Figure 49 Gold torque from Vilas Boas.	Figure 50 Earrings from the so-called "Bedoya treasure".	Figure 51 Fragments belonging to the diadem or belt of Mones.

Finally, there is also a gradual but considerable increase in the number and variety of imported objects. J. Naveiro (1991) has carried out the most detailed study in this area (see also Naveiro and Pérez Losada 1992), although it mainly focused on material that was Roman in origin. From the start of the Late Iron Age, the systematic appearance of foreign ceramics has been documented, mainly Punic but also Iberian, Ibero-Punic or Attic: Alto do Castro (Parcero and Cobas forthcoming), Baroña, Elviña, Castrovite, Toralla, A Forca, A Lanzada, Santa Tegra, Troña, Fozara, Castromao (Naveiro 1991; Naveiro and Pérez Losada 1992: Figs. 10, 11). With the passage of time, these materials were replaced by others of Roman manufacture with a significant number of wine amphorae from the very start of the period.⁷ The intensity of exchange was such that materials of this type are omnipresent in any occupational phase, particularly in coastal regions from the second century BC onwards. These are accompanied to a lesser degree by luxury ceramic items such as terra sigillata and objects for personal adornment, such necklace beads, with some examples of Punic origin dating from the fourth century. The situation does not appear to have changed immediately after the Roman conquest. It was only from the middle of the first century AD on that other types of imported products appeared, including functional items such as ceramics for daily use, domestic crockery, tiles, etc., at the same time that wine amphorae exhibit a spectacular decrease and almost disappear (Naveiro 1991: 63).

3.2.3. Beyond settlements: environment, production and territoriality

As mentioned previously, the limited amount of paleoecological evidence available for the Late Iron Age does not appear to show any significant modifications in subsistence patterns compared to the previous period. The same species were used for cultivation and pasture, and at present it is not possible to use this record to correctly evaluate if there was a significant increase in productivity. It does appear that there were some innovations in terms of the agricultural technology available, although once again the problem is the incompleteness of the existing record (Teira 2003). In any case, as we have already seen, it appears that it was from the Late Iron Age on that the use of iron became more generally widespread, predictable given the impact of the new technology on production.

As we have indicated, other types of evidence make it possible to suggest that there were modifications in production strategies and use of the environment, particularly those reflected in the location of the settlements and the development of significant structures for storage. The connection with specific types of terrain is a key factor in the selection of locations for these settlements. Topsoil depth and the water retaining capacity of certain soils make them unsuitable for any type of use that does not require constant work in exchange for the possibility of obtaining two yearly harvests from the same location. These are soils best suited to the application of short alternating cycle systems, which would guarantee the productivity levels required to sustain a community controlling and exploiting limited areas of arable land which, furthermore, was situated close to the hamlets, minimizing energy spent in reaching the field plots as well as maximizing their control and defense.

The other two components of this landscape model are less important, meaning that the proximity of land suitable for these uses is not a determining factor. The types of soil suitable for extensive use are always situated at a middle distance from the settlements (around two kilometers), while beyond that distance lies unproductive land.

In contrast to our proposals for the Early Iron Age, it may now be suggested that a more intensive system of land use developed in the Late Iron Age, characterized by a different type of exploitation cycle. We can assume the existence of a crop rotation strategy based on arable land subjected to intensive cultivation organized in short cycles alternating between cereals and leguminous plants. This does not mean a rejection of the idea of the maintenance of extensive cultivation strategies on light, well-drained soils, although in this case the most relevant feature is the fact that the immediate surroundings of the settlements were converted into spaces preferentially used for the development of activities of stable and permanent production. There are few direct indications of these types of spaces, although structures that were possibly used for preparing the land have been found around some sites, such as the terraces found at the Follente hillfort (Parcero 1999), which can be dated between the Late Iron Age and the indigenous-Roman period (Fig. 35).

3.3. Indigenous-Roman period

3.3.1. The forms, patterns and structure of settlements

The most visible transformation in this period is the appearance of a more complex pattern of settlement. The Roman occupation led to the implantation of new ways of occupying territory, starting with the urban settlements in the capitals of the Conventus (Lucus, Bracara and Asturica), all of which were founded in the time of Augustus and were fully developed by the middle of the first century AD.

At the same time a network of secondary settlements arose, which in principle were closely linked to the main communication roads; only at the end of the first century AD did the development of open rural settlements start to take hold, either as aristocratic homes (villae), or hamlets (for a recent discussion, see Pérez Losada 2000). In some areas these processes of transformation are manifested even earlier, as in the case of mining areas in the interior, where the restructuring of the territory started at the very beginning of the conquest, and developed according to a sudden break from the pre-Roman model of settlement, motivated by the need to have the necessary means available for the intensive exploitation of gold mines (Sánchez-Palencia and Javier 2000).

Figure 52 San Cibrán de Lás (Punxín, Galicia). Aerial view of a major oppidum.

Within this general framework, and particularly in areas without large-scale mining operations, the process of transformation was more gradual. For much of the first century AD hillforts were the only type of indigenous settlement, although morphological and other types of innovations did take place over time. The most important of these was the development of relative differences in settlement size, culminating in the appearance of large 'oppidum' type-sites that covered up to 40 hectares such as Citânia de Sanfins (Silva 1986), although this is very much an exceptional example. Other well known oppida include Monte Mozinho, San Cibrán de Lás, and Briteiros, all of which are around 20 hectares in size (Acuña 1991; Rodríguez Cao et al. 1993) (Fig. 52). This was the case in the southern part of the region, as further north the differences in size were less noticeable, and settlements rarely covered more than five hectares. Despite these differences in scale, a hierarchical pattern of settlement developed in which large sectors of land were occupied by central places around which a number of smaller hillforts were organized (Silva 1986). To date nothing comparable is known for the pre-Roman period.

Most of the smaller settlements maintain the standard forms, structures and locational models of the Late Iron Age (Parcero 2000b, 2002a, 2002b). In fact, many of them also exhibit evidence of pre-Roman occupation, and it is quite unusual to find the creation of entirely new settlements. Even the Viladonga hillfort, traditionally considered a paradigmatic example of late occupation, has recently produced indications of initial occupation in the first century BC (Arias and Fábregas 2003) (Fig. 53). However, the large oppida developed according to a locational model much closer to that of the Early Iron Age: visibility, prominence and difficult access appear to be more significant in site location decisions. This indicates a concomitant lack of concern with proximity to those soils best suited for cultivation. It is therefore highly significant that the majority of these large-scale hillforts were built on sites with previous Early Iron Age occupations, as is also the case with many central European oppida (Büchenschütz 1995). When a new location was chosen, visibility, prominence and good defensive conditions were the most important criteria (Parcero 2000b), much more than the potential productivity of the land (Fig. 54).

Figure 53 Viladonga (Castro de Rei, Lugo), aerial view (after Vilas and Durán 1996).	Figure 54 Sta. Tegra (A Guarda, Galicia). Aerial view of its commanding hilltop location.

These differences extended to the interior of the settlements as well. The organization of the internal structure of the settlements into family groups is something that was fully developed from the Late Iron Age onwards. All of the hillforts occupied from this moment on, until the definitive abandonment of the fortified model of settlement, continued to display this internal structure. However, in some of these large-scale settlements there were 'urban' developments that do not reflect this total absence of structures and elements of community organization in which the only unit of reference was the domestic group. Careful planning of space appeared, with wide streets separating regular blocks, in which the 'family groups' of structures were established (Fig. 55). These were still the elemental settlement units, although in these cases they were not superimposed over the group space (the settlement as a whole), but instead appear to have been subjected to a previous organizational scheme, according to which certain common spaces imposed their priority. However, this should not be considered a rejection of the tendency toward spatial fragmentation. On the contrary, the division is multiplied, as the 'family group' category is joined by a division brought about by the establishment of regular units that some authors have referred to as 'neighbourhoods' (Silva 1996), suggesting a shift toward a type of settlement arrangement in which population is organized according to some kind of social and/or political order beyond kinship. In these cases there are various levels of spatial division: the settlement, the 'neighbourhood', the 'family group' and the individual structure itself.

There can be no doubt that this was an important innovation, but it was restricted to the re-structuring of some specific sites and did not even include all of the large-scale settlements. At other sites such as Santa Tegra, a large hillfort mainly occupied in the first half of the first century AD (Peña 1989) (Fig. 56), or the group of small, fortified settlements that represented the predominant type for this period, structures were still grouped together in 'family' compounds that completely filled the hillfort interior. It also appears that there was a very clear chronological tendency, as the best-documented cases of this new type of organization date from between the end of the first century and the beginning of the second century AD.

Figure 55 Sanfins (Paços de Ferreira, Portugal). Orthogonal arrangement of settlement space (modified after Romero 1976).	Figure 56 Sta. Tegra (A Guarda, Galicia). Partial plan of settlement structures (after Peña 1986).

Another structural novelty was the proliferation of thermal buildings commonly called "monuments with ovens" (Fig. 57). These constructions always had a similar size of around 12 meters in length and three or four meters in width, although here the most important feature is their low profile, of around two meters. Despite this fact, they are the "most monumental type of building from the Hillfort Culture" (Almagro and Álvarez 1993: 186), both because of their capacity, and particularly because of the regular appearance of decorative elements on the wall separating the parts known as the antechamber and chamber. In the best-known cases, these are monolithic slabs with a single, small opening in the base, which are commonly known as "pedras formosas" (carved or beautiful stones) (Fig. 58). Today it would appear to be clear that these were structures used for thermal purposes. Their association with water (channels, basins etc.) and fire (ovens), as well as the lack of space, would appear to verify this interpretation. However, there is the possibility of important variations in this interpretation. Many scholars consider the ovens an indigenous version of the Roman sauna, a purely functional installation for relaxation (see for example Calo 1993: 151), although there are some conflicting points in this version. The first problem is the consistent decoration of these 'carved stones'. The second is their limited capacity: these so-called chambers never cover more than 10 square meters, and only rarely reach six square meters, and their height would have been very low, as this would have been set by the 'carved stone', which rarely reaches a height of two meters. The third, and perhaps most illustrative problem, is the type of positions they occupied. Although their location has been explained in functional terms (the need for running water, for example) the fact is that most of these structures are found in the outer limits of the settlements, set against the interior of the outermost walls, beyond the limits of the hillfort and in one case at the end of one of the ditches (Fig. 59). These are virtually the only types of structures that have so far been identified beyond the boundaries of the settlement. Furthermore, in all of the known examples it appears that locations far from the center were deliberately chosen.

Figure 57 Plan of some thermal buildings known as "monuments with ovens" (after Villa 2002a).	Figure 58 "Pedra fermosa" from Briteiros (currently in the Guimarâes museum).

Figure 59 Punta dos Prados (Ortigueira, Galicia). Location of thermal building at the edge of the internal ditch.

As with other elements of the archaeological record of this period, the idea has taken hold that these 'monuments with ovens' were exclusively related to later periods of occupation of the hillforts, during the phase of Roman dominion, specifically associated with sites that are exceptional because of their size and/or richness, i.e. 'central places', especially the large hillforts in the south of Galicia and northern Portugal. The second part of this postulate has been proven to be incorrect, thanks to the identification of structures of this type at sites in the extreme north of Galicia, and in relatively small hillforts such as Punta dos Prados (Ramil 1995-6). The recent excavation of such a structure at Chao Sanmartín represents the first stratigraphically dateable example from the pre-Roman period, from around the fourth to third centuries BC (Villa 2002a). In any case, their maximum period of development would have corresponded to the first decades after the conquest, around the birth of Christ and the first half of the first century AD (Calo 1993).

3.3.2. Inside settlements: material culture

The ceramic material culture reveals a series of standards of continuity and change that are largely parallel to those seen in the settlement record as a whole. There are a series of differences between materials from the Late Iron Age and the Indigenous-Roman period, although these are slight and do not involved many stages of the TOC, being mainly limited to the most superficial features and refining tendencies that had appeared at an earlier stage (Fig. 60).

Range of forms. Slight differences from the preceding period are seen in the modelling process (such as the generalization in the final phase of the flexed composite profile), changes in previously existing forms (such as the tendency toward convergence in the necks of pottery pieces, the development of surfaces, etc.), and the appearance of new forms (pottery with 'ear-shaped' handles). Changes are also seen in the modelling technique itself, as the potter's wheel became more widely used for manufacture, particularly in southern hillforts such as Santo Ovidio de Fafe or the Barbudo hillfort (Martins 1988). Production became more standardized and individualized, as is indicated by the potter's marks frequently seen on vessels at hillforts such as Santa Tegra (Peña 1983), Viladonga (Arias and Durán 1996), Sanfins, S. Juliao, Vila Verde and Briteiros (Silva 1986).

Finish. The spatula was more frequently used as a surface-finishing tool, and was at times the only technique used throughout the whole profile of the piece.

Decoration. With regard to differences in decoration, the theme, technique and morphology tend towards a greater simplification as compared to the previous period, with an increase in the number of decorated ceramic pieces in the record. The variety seen in the repertoire of decorative elements and motifs was restricted, with a tendency towards simpler, more linear decoration, and the appearance of new elements such as shields, rosettes, swastikas and schematic figural representations (Fig. 61). The predominant decorative techniques were burnishing and plastic decoration attached to the surface of the vessels using slip. However, the main difference is to be seen in the way in which the decoration was arranged, as this was very simple, barely visible, reiterative and not defined, despite the fact that in many cases the design is found repeated over the surface of the vessel. In some cases it is difficult to distinguish between what is actually decoration and what is simply a finishing treatment (Fig. 62). The decoration is placed mostly on the outer surface but it is also documented on the rims (Fig. 63) and very occasionally on the inner surface. The careful treatment of the surface of vessels was also emphasized, with paint and engobe seen in pottery from several sites, such as Santa Tegra (Peña 1986; Rey Castiñeira 1999: 161). This may perhaps be related to the use of interior and exterior adornments on the walls of structures, as may be seen at Briteiros, Romariz, Terroso and Âncora (Silva 1986: 52) in Portugal, or at Troña (Hidalgo 1988-89), Elviña, etc. in Galicia.

Figure 61 Indigenous-Roman pottery; characteristic forms and decorations.	Figure 62 Indigenous-Roman pottery; characteristic forms and decorations.	Figure 63 Indigenous-Roman pottery; characteristic forms and decorations.

Firing process. The firing process followed the same pattern as in the previous phase, with variable temperatures and no special link between the techniques used and the vessel types. An oxidizing environment remained the prevailing technique, while reduction firing nearly disappears.

Final product. In the final product, the most significant feature is the opposition between the greater variety of forms (Martins 1989: 92) and the increase in the simplicity of the decorations (Silva 1986: 125).

In general terms, it is possible to observe a simplification of the concept of ceramics manifested by a breakdown in the distinctions between previous vessel types combined with a tendency towards simplifying both the treatment of the surface and the decoration. There was also an improvement, at least apparently, in the manufacturing technique. The division between three formal vessel types from the previous phase was maintained. However, pottery with a flexed, composite profile is no longer an exceptional morphological type but becomes the dominant form, while vessels with edged, composite profiles become more exceptional. There are even morphological transitions between the different types, as vessels with flexed, composite profiles acquired a more globular shape, thanks to the use of necks in converging directions, while vessels with simple profiles revealed features similar to those of pieces with edged, composite profiles, as may be seen in vessels with a reinforced border that does not extend towards the interior, and pieces with an edged composite profile with relatively narrow flat surfaces. The same thing happened with the way in which decoration was conceived, generally tending towards a greater simplification:

• While most decorations are applied before the finish, some decorations are applied after the finish for the first time.
  
  
• The pattern of stamped decoration as conceived in the previous phase disappears and is replaced by a much simpler appearance, while less care is taken in the treatment of the decoration as well as the surface of the vessel.
  
  
• Stamped areas appear on pottery with composite profiles in many cases without any kind of delimitation, at times even combined with plastic decoration, in contrast to the method used in the Late Iron Age.
  
  
• Most of the vessel forms no longer display elements that delimit the decoration, and the decorative element and motif once again coincide, although in contrast to the initial phase, these do not coincide with the decorative scheme.

Changes also appear in the relationship between decorated and undecorated pieces, as there are examples in which the pottery acquires a special character as a result of sophisticated technical treatment and careful finish, without necessarily being decorated.

This was also the time when major developments took place in many other categories of material culture, although as we have already mentioned the formal and typological standards exhibit considerable continuity. We have already referred to some issues in relation to gold and silver work, which in principle would still have been in use and produced at this time. Most of the few pieces found in reliable contexts are from occupations that date to this period, although they are mainly highly fragmented pieces that were no longer used for their original purposes, possibly indicating a loss of value as objects, and the start of their re-use as raw materials (Ladra 1999b).

Figure 64 Statues of warriors presumably from Briteiros (currently in the Guimarâes museum).

One of the most noteworthy developments was in the area of sculpture and architectonic relief, with one of the most detailed studies by Calo (1994). Although it is possible that this practice started in the pre-Roman period (Carballo 1996b), it is true that it was most fully developed throughout the first century AD. These reliefs with geometric motifs decorate parts of houses, particularly the doors; the motifs used are rosettes, triskeles, loops, etc., in many cases very similar to those found on ceramics or metal objects. Furthermore, an anthropomorphic type of sculpture was developed using human heads, seated statues and particularly statues of warriors (Fig. 64). These are monumental figures, usually over two meters in height, representing armed human figures in a fairly rough and schematic way. In some cases these have been preserved more or less intact, whereas only fragments are known for others, and some have suffered important alterations due to being re-used in later periods. It appears that all of them represent male figures, generally bearded, wearing torques around their necks and shown with a small circular shield (caetra) and a dagger or sword. The type of detail varies, and is better on some sculptures than others. They are often monolithic. The figures are always standing, looking forward with a firm, hieratic appearance. Between 25 and 30 examples are known to date, depending on whether a number of doubtful fragments are accepted or not (Calo 1993: 137-8). Most of these statues are from ambiguous contexts in hillforts or their vicinity. There is little stratigraphic evidence that would make a more or less reliable chronology possible. In the few cases with a clear stratigraphic context, the statues come from occupational stages after the Roman conquest, generally towards the start of the first century AD. Eventually they are re-used as building materials, having presumably lost their original significance (Calo 1994).

Interpretations of the function of these figures vary considerably. The discovery of the base and feet of one of them on the walls of the Portuguese hillfort of Sanfíns (Silva 1986, 1996) has helped to clear the air a little, making it possible to do away with interpretations of the statues as funerary stelae. It is clear that these were literally monuments, situated so that they could be seen and made to perform a specific social function.

Finally, reference must be made to imported materials. As previously mentioned, from the middle of the first century AD on there was a dramatic decrease in the appearance of amphorae, which to date represent the most common type of foreign material found in the hillforts (Naveiro 1991: 63). From this point on other materials became more frequent, including common Roman ceramics, pieces of terra sigillata and construction materials (tegulae, imbrices). Metal objects characteristic of the Roman period, such as omega-shaped annular fibulae, came into general use. What we see here is a fairly substantial change in the patterns of exchange and acquisition of foreign objects, which were increasingly used in daily life.

3.3.3. Beyond settlements: environment, production and territoriality

In general terms, none of the records available to date (palynological, archaeobotanical or faunal) have made it possible to detect significant modifications in the patterns of environmental exploitation by indigenous communities during the Indigenous-Roman period. Similarly, the locational analysis of the settlements reveals a general continuity through time, making it possible to propose a series of criteria for environmental exploitation patterns that are comparable to the pre-Roman period.

However, some significant innovations did appear as a result of the Roman presence. Firstly, a certain level of agricultural intensification can be assumed, as the need to satisfy tributes to the Roman state would have required increased productivity at least to a certain degree. However, the most significant innovations are found in the introduction of new ways of exploiting the environment, now no longer directly linked to subsistence farming but rather to the development of industrial systems that were directly encouraged by Roman interests. Two of these are of particular importance. The first is the development of large-scale open pit mining operations in the interior of the region (Sánchez-Palencia and Javier 2000). These mining operations led to an important transformation of pre-Roman landscapes over a wide area, involving a re-structuring of previously existing types of settlement and the creation of a new network of territorial occupation. This new network, despite in many cases maintaining the hillfort as the focus of settlement for indigenous populations, modified the ways in which the hillforts were embedded in the landscape, creating a new type of settlement generally known as the 'mining hillfort' (see Sánchez-Palencia and Javier 2000).

The second new type of exploitation involved salt factories along the coast, forming an important complex that undoubtedly involved the participation of indigenous populations. However, at least to date, it has not been possible to confirm whether the appearance of these factories (whose chronological sequence is still not completely clear) was accompanied by a large-scale alteration of indigenous patterns of settlement.

Finally, it is important to note the progressive development of establishments dedicated to intensive agriculture for commercial purposes, basically manifested in the form of aristocratic villae. It has been suggested that there was a possible expansion of some crops for commercial purposes, such as the wine vine (Rodríguez López et al. 1993). In any case, here it appears possible to refer to a relatively late development, as most of the villae known to date began to operate towards the end of the first or second centuries AD (Pérez Losada 2000).

4. Summary of historical interpretations

This review has provided a considerable number of proposals for historical synthesis. At the beginning and middle of the twentieth century, interpretive tendencies were based on arguments involving mass movements of people and the arrival of central European migrants who provided the cultural template of the hillfort complex as a variant of the Hallstatt culture. Today, this interpretive debate has shifted toward a different set of arguments. The ethnogenesis of this cultural complex no longer occupies an important place in research agendas, and the discussion is now focused on sociologically characterizing the appearance and development of the Iron Age. We will go on to offer a brief summary of some of these interpretive trends.

4.1. Early Iron Age: between complexity and stagnation

In principle, the appearance of the Iron Age (here characterized by the first appearance of fortified settlements) has been related to general dynamics in Atlantic Europe as a whole from the end of the second millennium BC on. Different scholars have given special emphasis to the importance of large-scale trade networks in the exchange of metals as a dynamic element for much of the Bronze Age (Fábregas and Ruiz Gálvez 1997; Peña 1992b; Ruiz Gálvez 1988). Greater or lesser control of these trade networks and the products derived from them (metalworking in bronze) was viewed as a key element in regulating social relationships. A key event making it possible to comprehend the transition from the Late Bronze Age to the Early Iron Age was the transformation of these networks by the dynamics of change in the Mediterranean basin (Alarçâo 1992; Fábregas and Ruiz Gálvez 1997; Martins 1990; Peña 1992b) within an interpretive model directly related to the archaeological application of models of world economies (Frankenstein 1997; Frankenstein and Rowlands 1978).

This transformation leads on the one hand to types of development that were much more diversified in different parts of western Europe, and on the other hand, particularly in the case of the north-western Iberian Peninsula, to the move to a more complex socio-political structure than that of the Late Bronze Age (Calo 1993; Peña and Vá