Stratigraphy Matters: Questioning the (Re)Sacralisation of Religious Spaces from Late Antiquity to the Early Middle Ages in the Iberian Peninsula

Abstract

In recent decades, archaeological and written records have been used in combination to improve our understanding of Late Antique and Early Medieval Hispanic architectural culture (5th–10th c.). Within this renewed research context, the rebuilding of Late Antique churches and the reuse of earlier materials in both Early Medieval Christian and Islamic religious buildings; the transformation of Late Antique churches into early Islamic mosques in al-Andalus; and the rebuilding of other churches by the Christians themselves, both in al-Andalus and in the northern Iberian plateau, from the late 8th century onwards, have often been used to thread narratives concerning tradition, continuity, and re-sacralisation of earlier religious architectural spaces and, through this, religious change in the middle ages. However, the application of modern archaeological methodology reveals that these processes need to be qualified, for the stratigraphic gaps between building phases make them sometimes hard to interpret. By analysing some examples of Hispanic religious sites, this paper reflects on the many concepts related to sacralisation, explains the way archaeology is able to trace the sacralisation processes, and intends thus to highlight the complexity of these phenomena.

In recent decades, archaeological and written records have been used in combination to improve our understanding of Late Antique and Early Medieval Hispanic architectural culture (5th–10th c.; Villa del Castillo et al. 2022). Within this renewed research context, the rebuilding of Late Antique churches and the reuse of earlier materials in both Early Medieval Christian and Islamic religious buildings; the transformation of Late Antique churches into early Islamic mosques in al-Andalus; and the rebuilding of other churches by the Christians themselves, both in al-Andalus and in the northern Iberian plateau, from the late 8th century onwards, have often been used to thread narratives concerning tradition, continuity, and re-sacralisation of earlier religious architectural spaces and, through this, religious change in the middle ages.1

However, the application of modern archaeological methodology reveals that these processes need to be qualified, for the stratigraphic gaps between building phases make them sometimes hard to interpret and thus it is easy to adopt the simplest explanation. This paper reflects on the many concepts related to sacralisation and explains the way archaeology has often been able to trace it by recognising the materiality of different sacralisation processes. Some examples of Hispanic religious sites will be used to illustrate the sacralisation practices undertaken by both Christians and Muslims with the aim of highlighting the complexity of these phenomena and the risks involved in accepting established historical models to interpret material remains.2

1. Attributing and Deleting Sacredness during Late Antiquity and the Early Middle Ages

Sacralisation is the action of assigning a sacred character to something that did not have it before; re-sacralisation, recovering the sacral meaning of something that had lost it; and, conversely, de-sacralisation, removing that sacred character, including different but not incompatible actions of de-consecration (non-violent) or desecration (violent), both of which were common in the Late Antique (5th–7th centuries) and Early Medieval (8th–10th centuries) periods across the Mediterranean. In the Iberian Peninsula, this period is characterized by the convergence and divergence of different religions and their related customs, particularly Christianity and, from 711, Islam. These actions always mean the intended transformation of what existed before; they usually have a material or physical effect, and are sometimes, but not necessarily, successive to each other. Intention, materiality and diachrony are thereby fundamental concepts that interrelate beneath these phenomena, in which subtle nuance can make all the difference.

Writing about Late Antiquity, Caseau (2000, pp. 21–22) argued that sacralisation could take two ways: syncretism, which seeks the protection of all divinities and thus respects and even incorporates them; and selection, which chooses a deity and necessary rejects all others. Concerning architectural culture, these approaches entail different building actions and, therefore, a different materiality. Syncretism means the inclusion of the new divinity and it is reflected materially by maintaining the existing religious building, along with a ritual of sacralisation or dedication of a new place to the new divinity. This is illustrated, not only in the Iberian Peninsula, by the coexistence of Roman pagan temples with Late Antique Christian churches during the 5th–7th centuries, and of Early Medieval Christian churches with early Islamic mosques during the 8th–9th centuries. The selective attitude, in contrast, might involve the de-consecration and conversion of the previous architectural space or its desecration and the construction of a new religious building. Furthermore, this new construction can be founded close to the ancient religious one or be located in a new place with the aim of focussing attention on the new cult and underscoring the displacement of the former one.

When desecration implied the total destruction of the previous sacred building, this was also a costly action that required human, material, and technical resources. Likely because of this, most late Roman pagan temples were left to decay during the Late Antique period (Caseau 2000, p. 32). This is also applicable to the relationship between Late Antique churches and early Islamic mosques, as confirmed by the available material record (Bursi 2021, p. 482, on the common existence of earlier Christian buildings close to new Muslim ones).

Within this desacralization framework, the phenomenon of spolia has often been understood as a deliberate action that seeks to promote the status of the sponsors (mainly aristocratic) and to honour the new construction by linking it with the past through the presence of old elements.3 When ancient liturgical elements are reemployed in new religious buildings, they are commonly understood as bearers of sacredness. This argument is wielded particularly often with regard to the reuse of Roman altars as Christian altars, as critically re-examined by Sastre de Diego (2013, mainly pp. 134–39). However, whether this action took place as part of a de-consecration or desecration programme, or whether it affected an already ruined or a standing construction, or whether the destruction of the latter was partial or total, is of little importance: if older elements were intended to be adequately reused, they needed first to be carefully and properly removed regardless of the circumstances. Both removing and replacing entail manpower, precise skills, and tools to keep the integrity of the old elements, mainly sculptures carved from invaluable materials (usually marble), so that they can be effectively used in the new construction. Furthermore, old elements could be reused, that is, employed for their original purpose (e.g., a capital reused as a capital), or recycled, that is, employed for a different function (e.g., a capital recycled as building stone), which necessarily leads to diverse readings, mainly from a symbolic, including religious, point of view (Utrero Agudo 2020, pp. 38–40).

This brings us back to the above-noted concepts of intention, materiality, and diachrony. When considering these theoretical practices of assignation or removal of sacral character to and from Late Antique and Early Medieval Hispanic constructions, the archaeological evidence invites us to, firstly, pay special attention to recording and understanding the stratigraphy (diachrony), before interpreting materiality, and if possible, proving intentionality. Stratigraphy is essential for understanding continuity and discontinuity and can be approached following a standard methodology, regardless of period, region, and architecture.

2. From Late Roman Pagan Constructions to Early Christian Churches

Founding a Christian church always means sacralising a space, that is, dedicating a building to God. The church becomes the space where people and God meet one another. Aristocratic sponsors, both secular and ecclesiastical, were aware of that and tried to display their presence, even when they were physically absent, through foundational and consecrative stone epitaphs placed in prominent positions and even welcoming churchgoers, or by using the church as their privileged funerary space, in order to showcase their role as patrons of the building. Within this process, the role of the bishop in connecting community and God is crucial. He is responsible for sacralising the foundation by placing the relics of the saint to whom the church is dedicated and for consecrating the altar, by which action he becomes sacred himself (Castillo Maldonado 2004, p. 40).

Christian buildings standing on ground that had not been occupied previously meant the inaugural sacralisation of these places. In other instances, churches built over earlier constructions could entail the sacralisation of the latter or not, but, before any conclusion is reached, two archaeological challenges need to be seriously considered. The first is defining the function and type of the preceding building or buildings, which are especially common in previously inhabited and densely built urban centres. Existing buildings could obviously exist as well, as will be shown, in rural areas and, this fact also demands an explanation. The second is establishing the time lapse between the earlier constructions and the new ones, whose builders did not necessarily know where they were building and/or acknowledged the significance and function of existing buildings. Reoccupation of a place does not thereby necessarily mean continuity of use and, more relevantly for the case at hand, of sacrality. Reoccupation of a place might simply reflect reoccupation of a place. Sacral meaning was based on the memory of the sponsors and the community of worshippers who were able to acknowledge the signals and significance of the place, that is, the association of its past and present meanings (Markus 1994, p. 271; Dell’Elicine 2020, p. 116).

Archaeological sources show that the transformation of Roman pagan temples into Christian churches was neither homogeneous nor common across the territory of the Empire, and that it was comparatively rare before the mid-5th century AD (Arce 2006, p. 117).4 In the Iberian Peninsula, the urban pagan temples archaeologically analysed to date were neither transformed into churches nor occupied by other buildings from the 5th century onwards, and it seems that they were gradually abandoned over the next two centuries (Arce 2006, pp. 121–22; Ripoll and Arce 2015, p. 349; Martínez Jiménez et al. 2018, p. 79).5

Different factors might explain this process. First, the transformation of previous buildings always involved partial destruction and reconstruction, activities that were almost as expensive as new construction and needed official consent, at least, according to the written record, for public temples during the late Roman period (Caseau 2000, p. 31, for a general view; Buena Casa 1997 and Arce 2006, p. 116, for Hispania; López Quiroga and Martínez Tejera 2006, p. 129, for both). Second, although pagan temples were deconsecrated, they could still fulfil different political, commercial, and representative functions, which would justify their maintenance for long periods. Third, since early Christian activity in former Roman cities was initially conditioned by the cults of martyrs and their relics, the earliest martyrial churches did necessarily occupy the suburbia (as it is also exemplified by Hispanic examples; among others, Santa Eulalia in Mérida or San Fructuoso of Francolí in Tarragona), but not the earlier monumental centre, still occupied by the old pagan aristocracies and densely built up (Thomas et al. 2017, p. 316). Slightly later, episcopal complexes, understood as a set of different constructions including the main episcopal basilica, which thereby demanded large areas to be built, would often be placed at some distance from the forum for practical reasons related to the availability of space (Ripoll 2019, pp. 53–54).

Concerning rural examples, constructions at the villae of Milreu (Algarve region) and São Cucufate (Alentejo region), which had previously been regarded as Christianised late Roman pagan temples, have been recently reinterpreted as mausoleums (Graen 2005, 2008). This hypothesis completely rules out earlier ideas concerning their conversion (López Quiroga and Martínez Tejera 2006, pp. 144–46, with earlier references). This new view argues for the continuity of funerary use in São Cucufate, where the Early Medieval church has, in addition, been tentatively identified elsewhere in the villa (de Alarcão et al. 1995). At Milreu, the transformation of the former pagan mausoleum (early 4th century) into a Christian church (5th–6th centuries) has been solely based on the presence of a baptistery and a few graves on the podium, as well as on the addition of two mausoleums (Teichner 2006, p. 212). In Milreu, therefore, the funerary use seems to have been replaced by a cultic one that seems to preserve thus the site’s original sacral meaning.

Both examples can be added to the short list of late Roman funerary pagan constructions converted into early Christian churches, although the available archaeological data makes these cases far from certain. This is also the case of the mausoleum of Las Vegas de San Antonio in Pueblanueva (province of Toledo), built in the second half of the 4th century (Hauschild 1978). Three walls added to form a new rectangular space to the east of the original octagonal building could be interpreted as the sanctuary of a Late Antique church. A marble slab altar dated to the 6th–7th century, reused to cover a north-south-oriented grave, along with the presence of longitudinal cuts in the ground thought to be a trace of the chancel screens, and a belt buckle coming from another tomb (also late 6th–early 7th century), are further evidence for this conversion and date. Unlike Milreu, no baptistery pool has been attested in Pueblanueva. Something similar might have happened at the villa of La Cocosa (province of Badajoz), where a mausoleum, presumably contemporary to the nearby 4th-century villa, could have been converted into an early Christian church. The limited data available (the site was partially excavated in the 1950s) makes it impossible to confirm if the architectural spaces built to the north of the mausoleum correspond to a church, as cautiously proposed by Sastre de Diego (2010, pp. 44–47). The presence of a baptistery to the south of the mausoleum and a fragment of a marble altar table dated to the 6th century (Sastre de Diego 2010, p. 46) are the most reliable material evidence for a church here.6

The long-proposed chronologies for these buildings force caution, but the shortest possible gap, two centuries, between them, suggests that the mausoleums of Milreu, Las Vegas, and La Cocosa were not in use at the moment of their conversion into churches and draws attention to their monumental character as the most likely reason for their selection for the construction of a church there, after a long period of abandonment. In fact, based on the ceramics found upon the Late Antique graves and under its collapse level, it is clear that the Roman vault of Milreu stood until the Islamic period (Teichner 2006, p. 213). In Pueblanueva, the addition of the apse necessarily demands the rest of the building to have stood. In La Cocosa, where the alleged church could be close but not exactly upon the mausoleum, this information is, however, unknown.

Finally, within this Late Antique context, some Christian churches have been identified as the result of the conversion of former pagan temples, mainly due to the presence of reused sculptural and liturgical elements. These are the cases of São Miguel de Mota (Alandroal region) and San Juan de Postoloboso (province of Ávila), both of which were already examined by Caballero Zoreda and Sánchez Santos (1990, pp. 444–48, with earlier bibliographical references) and Sastre de Diego (2013, pp. 135–36). At the Portuguese site of Mota, however, recent archaeological works confirm that it is neither possible to date the hermitage that reused the sculptures of Envodelicus, nor to confirm the presence of a Visigothic church on the site based on the sculptural elements reused at the nearby castle of Alandroal (Schattner et al. 2005, p. 898; Villa del Castillo 2021, vol. 2, pp. 327–28, 7th-century fragments). At Postoloboso, the hermitage reuses several inscriptions dedicated to Vaelicus and sculptural and liturgical elements dated to the 7th century (Fernández Gómez 1973, pp. 209–41), but the existing construction is dated to the late medieval period, and there is no evidence for a Visigothic church on the site (Fernández Gómez 1973, pp. 249–56). Recent archaeological works (Schattner et al. 2006) argue instead for the site’s continued occupation from the pre-Roman period until today, the only hiatus corresponding to the Islamic period, although no evidence of a Visigothic church has been attested, other than the elements noted above (Villa del Castillo 2021, vol. 2, p. 375), whose origins are, however, uncertain.

But the main conclusion regarding Mota and Postoloboso, and one that could be, in my view, made extensive to other examples, is that these elements were not reused, but recycled as building material. That is, they had lost their original function. For this reason, the meaning of these elements in the past is irrelevant, because they had lost it by the time they were used for an entirely different purpose in the new church. In some cases, they were even plastered, so they were not visible and thereby not capable of bearing any meaning. At this juncture, it is worth mentioning the church of Santa Lucía del Trampal (province of Cáceres), founded in the late 8th or early 9th century (Caballero Zoreda and Sáez Lara 1999). Its walls reuse Roman construction materials (granite for the ashlar stones and marble for the decorative elements), along with a large set of 15 inscribed arae dedicated to the pagan divinity Aatecina and fragments of arae and anepigraphic stelas, all of which are dated from the late 1st to the early 3rd centuries AD (Abascal Palazón 1995). Their source is probably nearby, as different Roman settlements are known in the vicinity, but not from the site itself. These pieces were, however, already plastered in the earliest building phase (Caballero Zoreda and Sáez Lara 1999, p. 89), so they were not visible.

3. Early Medieval Churches and Mosques

Like other Mediterranean territories, on the arrival of Islam in 711 the Iberian Peninsula was peppered with Christian churches, shrines, and monasteries, in both rural and urban environments. In al-Andalus, again like in other territories of Dar al-Islam (Bursi 2021), the new Islamic rulers contested Christian power and architectural dominance by imposing their presence upon the landscape in a variety of ways, including the reconstruction, adaptation, and founding of new urban and rural settlements and civil and religious buildings. Early Medieval Christian buildings (al-Andalus and northern plateau), mainly monasteries with their corresponding churches, and Muslim foundations (only in al-Andalus) had to find their place in densely built-up urban areas, both by making use of available space and by looking for convenient locations for their respective communities. They could avoid earlier religious constructions and build new ones, either next to them or elsewhere, or they could take over and replace them. With the exception of Oviedo and León, the churches hitherto known in the northern plateau are located in rural areas, where available space was less of a problem, but the coincidence of previous and later constructions also needs to be explained.

The same two challenges noted above, regarding the function and type of existing remains and the time sequence between them and new constructions, must be considered in this instance too. Both Early Medieval churches and Early Medieval mosques are thereby analysed under a single epigraph, because their impact on previous town- and landscapes presents the same stratigraphic challenges.

The archaeological works undertaken in Early Medieval monastic sites located both in al-Andalus and the northern plateau have sometimes revealed traces of earlier constructions. This is for instance the case with the above-noted church of Santa Lucía del Trampal (Caballero Zoreda and Sáez Lara 1999, pp. 33–34, 324, remains of uncertain chronology and function), San Miguel de Escalada (province of León; Larrén Izquierdo 1990, Late Antique domestic rooms), Santa María de Wamba (province of Valladolid; Quintana Gordon and Boned Colera 1992, p. 30, structures of uncertain chronology and function), and San Salvador de Palaz del Rey (province of León; Miguel Hernández 1996, p. 134, a 4th-century rectangular structure, tentatively suggested to be of religious nature by Gutiérrez González 2015, p. 162). There is no evidence, however, to identify the earlier structures as either pagan temples or Late Antique Christian churches. The archaeological analysis of their structures (Caballero Zoreda and Sáez Lara 1999 for El Trampal; Utrero Agudo and Murillo Fragero 2022 for Escalada; work in progress in Wamba and Palaz del Rey) have found no remains beneath the currently extant walls. At Palaz del Rey and at other Christian buildings (Santa Marina, 1032, and San Isidoro, 1063, for instance) excavated in the same town of León, founded in the 1st century BC, the chances that previous constructions exist is obviously high. In all of these sites, however, if the older structures were not already ruined by the time the new building project began, they had to be destroyed to build the new church upon them, so there clearly was no intention of maintaining them.

According to some researchers (Effros 2001, for Early Medieval monasteries in Gaul, and Sánchez Pardo 2015, for Early Medieval monasteries in the north-west of the Iberian Peninsula), ancient ruins, regardless of their original function and unknown origin, could have had a major significance in the foundation of new monastic buildings. Taking this, and also the wide hiatus between occupation phases in Late Antiquity (confirmed for instance in Escalada and Palaz) and the construction of churches at least four centuries later, into account, it is legitimate to wonder if the latter, specifically in rural settings, gave more importance to the location of ruins than to the ruins themselves. In fact, other researchers (Cantino Wataghin et al. 2007, pp. 103–106; Castrorao Barba 2016, pp. 126–27, both for Late Antique Italy) have argued for other reasons behind the reoccupation of earlier sites. These usually had good connections with communication networks and access to natural resources (watercourses and quarries, for instance). Topographic reasons, along with other factors such as land ownership and the Church’s territorial and political strategy, could explain the selection and reoccupation of older sites, regardless of existing architectural structures. In addition, in my opinion, symbolic reasons needed to be perceived by both sponsor and community.

Only a few examples display evidence for the intentional preservation of the sacrality of earlier buildings, although both the original Christian church and the later building that kept its sacred character, or re-sacralised it and made it visible, are all dated to the Early Medieval period.

This is the case with the Galician church of Santiago de Compostela, founded by the Asturian king Alfonso III (899). The relics, together with the accompanying early medieval processes of inventio and peregrinatio, played here a key role in the construction of the church. The chronology of the mausoleum ascribed to St James has been dated either to the 1st or to the late 9th century, when it would have been built together with the apse and church that houses it. The similar building techniques used in mausoleum and apse seem to confirm that they are coeval, that is, that they were built by Alfonso III (Hauschild 1992, p. 96, consecrated in 899).7 Over the following centuries, this grave attracted further burials, namely two funerary rooms added to the south, one attributed to Bishop Teodomir (d. 847, the alleged discoverer of the body of St James) and another to presbyter Anastasio. Further graves, namely those of presbyters Martin (d. 1047) and Degaredo (d. 1062), and, obviously, the construction of the extant monumental cathedral, demonstrate the continued sacredness of the place, clearly driven by both the secular (Alfonso III) and ecclesiastical (bishops and presbyters) aristocracies.

A similar example of intentional preservation of memory seems to be that of the monastic church of San Millán de Suso (province of Logroño). Also subject to a long controversy about its architectural evolution and chronology, its archaeological analysis (Caballero Zoreda 2005) reveals that the original rock-cut structure was extended southwards with a two-cell standing structure (now vaulted) and with the addition of a hall with two aisles to the west. These architectural features, dated to the second half of the 10th century, were still connected with the original rock cuts. The first construction phase could be dated to the Visigothic period, based on the tradition that holds that the anchorite Aemilianus (Millán) lived there during the second half of the 6th century and first half of the 7th, and was even buried there, or to a later moment in the 8th–9th centuries according to the architectural features (Caballero Zoreda 2005, pp. 33–35). However, it is clear that San Millán kept its religious function and sacral significance from the start until the early 11th century, when it was enlarged westwards again, presumably during the reign of Sancho III, and beyond.8

Two final examples display the intentional preservation of sacral meaning of early medieval foundations, in this case suggested by epigraphic evidence. The archaeological analysis of the extant structure of the monastic church of Santiago de Peñalba (province of León; Murillo Fragero 2017), together with the recent study of its inscriptions (Guardia Pons 2017), make it clear that the original building, projected with a funerary counter-apse to the west, which housed two contemporary burials, was consecrated in 937 (the date is engraved into the original paintings of the apse) and reconsecrated, but to a different saint, probably in the early 12th century. The new consecration maintained the earlier church’s architectural structure, function, and relevance. A similar situation is attested in the already-noted late 9th-century basilica of San Miguel de Escalada, where recent archaeological analysis (Utrero Agudo and Murillo Fragero 2022) reveals that the addition of the southern portico and the reconstruction of the original partially ruined church was accompanied by the introduction of a new consecrated altar in the central apse and chancel screens in the lateral hall. It is possible that an inscription that mentions the reconstruction of a ruined church in 913, now lost, refers to this second building phase, highlighting the antiquity of the place, the humble character of the new construction, and the effort undergone by the monks with such an undertaking. Past and present are deliberately mixed in this consecration epigraph, which is clearly related to the Asturian monarchy (Utrero Agudo 2022, pp. 19–22, for a review of this and similar inscriptions, including additional bibliography).

Both of these monastic foundations demonstrate that a combination of intentionality, expressed by the inscriptions, materiality, reflected by the constructions, and diachrony and continuity, displayed by the archaeological sequence, makes it possible to establish this religious process beyond reasonable doubt in some instances.

Moving on to Islamic regions, unlike churches, mosques are not consecrated buildings (Caseau 2000, p. 46; Abumalham 2004, p. 9). The element that confers them a sacred value is the Mecca-oriented qibla, marked by the presence of the mihrab. Archaeological work undertaken in different regions of Dar al-Islam shows that the conversion of Late Antique Christian churches into mosques was a rare phenomenon (Bursi 2021, p. 490), and al-Andalus was not an exception to this (Utrero Agudo 2023). Over the last two decades, different archaeological sites have been examined and the results of this research have led to a better understanding of the twofold relationship of Late Antique churches and early Islamic mosques: the conversion of churches into mosques and the construction of mosques upon churches. Reused elements, allegedly from pre-existing churches, found in mosques have also played an important role in these arguments.

In the current state of our knowledge, no material evidence exists for the conversion of any church into a mosque, or for the partition of buildings for their simultaneous use as a church and as a mosque, in the Iberian Peninsula. This does not mean that no examples exist, as the available data is still limited, but it does at least suggest that conversion was not a common phenomenon. From a material perspective, this sort of transformation involves the reorientation of the existing building, the introduction of a mihrab, and probably other related repairs that are not always easy to trace in the archaeological record. According to the written sources, churches had to be not only reoriented but also purified to meet Islamic ritual needs, activities which were closely monitored and subject to tight rules by the Islamic state, which aimed to bring this practice to a minimum if not to avoid it altogether (Calvo Capilla 2011, p. 151). The orientation required by mosques probably made it quite difficult to reuse earlier structures and the existing urban layout. For all of these reasons, a new mosque was probably always preferred.

Although some late Roman constructions (the above-noted mausoleum of Pueblanueva, Hauschild 1978) and Late Antique churches had been put forth as possible candidates (El Gatillo, province of Cáceres, Caballero Zoreda 2003; Casa Herrera, province of Badajoz, Ulbert and Cruz Villalón 1991; Los Hitos, province of Toledo, Morín de Pablos et al. 2019; and Montinho das Laranjeiras, Algarve region, Catarino 2005–2006), the detailed examination of these examples has ruled out the possibility.9 Neither reliable remains of mihrabs (discussed by Caballero Zoreda and Sáez Lara 2009 and Caballero Zoreda 2009 for El Gatillo; Barceló 2002 and Calvo Capilla 2007 for Casa Herrera;10 De la Llave Muñoz and Escobar Requena 2017 for Pueblanueva; Calvo Capilla 2020 and Villa del Castillo et al. 2022 for Los Hitos; Coutinho 2003 for Montinho) nor further related material evidence that might point to a mosque (an ablution pool, for instance) have been attested in any of them. It is also worth emphasising that all of these sites are located in rural areas. Some of them were still occupied during the early Islamic period, but this does not mean that the Christian church was necessarily converted into a mosque.

In contrast, almost every mosque thought to have been built upon a church is located in an urban context. Only two exceptions of early Islamic rural mosques can be mentioned here, and they present similar features to those found in urban areas. Although Roman and Late Antique elements were found in the late 9th or early 10th century mosque of Almonaster (province of Huelva), the modern church of Nuestra Señora de la Concepción, no traces of a basilica were found there (Jiménez Martín 1992, p. 117). Moreover, the earlier elements are dated to the 6th century and were produced by workshops in Merida (Villa del Castillo 2021, vol. 1, p. 59 and vol. 2, pp. 277–78). The second example is that of El Cortijo de las Mezquitas (province of Malaga), where a still-standing mosque dated to the late 9th or early 10th century (Gurriarán Daza and Utrero Agudo 2018) was built upon a Roman rural settlement. Recent archaeological analysis of the architectural features and the ongoing excavation of the site reveal that the mosque was built upon an already ruined building, which was probably domestic in nature.11 Other Roman villas have been recorded in the vicinity, showing that the area was then densely occupied (Vargas Vázquez and Romero Pérez 2016, villa of Los Castillones, among others).

As all other mosques argued to have been built upon churches are located in urban centres, the above-noted challenges to define the function and type of the earlier buildings and to establish the time elapsed between constructions must be kept in mind at all times. Similarly, just as we do not know of churches turned into mosques, there are no archaeologically known examples of early Islamic mosques (8th–10th centuries) built upon earlier Christian churches.

The arguments can be presented here in summarised form (for an extended version see Utrero Agudo 2023). Excavations on the site of the cathedral of Zaragoza revealed that the 8th-century mosque was built on a previously unbuilt plot of land near the former pagan temple of the Roman forum, breaking the alignment of the latter and of the surrounding Roman street layout (Hernández Vera 2004). In the city of Toledo, the excavations at the mosque of Tornerías, dated to the 9th–10th century, revealed a 7th–8th-century structure built with reused ashlar stone blocks bound with a strong mortar and interpreted as a building with four naves, the function of which is unclear (Ruiz Taboada 2021). Similarly, at the mosque of Bāb al-Mardūm (or Cristo de la Luz), dated by its frontal inscription to 999–1000, two main features were uncovered: the north–south Roman road leading to the nearby gate (named Valmardón) in the former city wall, which was originally built in the second half of the 1st century AD; and a Roman apsidal construction of unknown function (Ruiz Taboada and Arribas Domínguez 2010). At other sites in the urban centre of Toledo, archaeology has found no traces of churches beneath later mosques (Moreno Martín 2018, pp. 158–61 for the main mosque, and pp. 166–67). And the excavations at the modern church of San Salvador in the city of Seville, which stands upon the mosque built by ‘Abd al-Raḥmān II in 829, found traces of neither late Roman buildings nor of Late Antique churches (synthesis by Amores Carredano 2009, pp. 70–71). Again, although the early 9th-century mosque reused Roman and Late Antique elements, this does not mean that a previous building was necessarily located on the same location.

The case of Córdoba deserves to be discussed separately. The city’s Great Mosque, founded in 786 by emir ‘Abd al-Raḥmān I, is probably the most paradigmatic example for the issue at hand. The debate about the identification of some earlier structures found in the courtyard by the architect Félix Hernández in the 1930s, thought to belong to the episcopal basilica of San Vicente, continues raging today. Together with the reasons wielded by several researchers (Calvo Capilla 2007, pp. 166–77; Caballero Zoreda 2009, pp. 28–30; Arce Sainz 2015)12 that rule out the interpretation of these material remains as belonging to a church, recent studies (Villa del Castillo 2021) have shown that some of the reused sculptural elements were brought here from the basilica of Santa Eulalia in Merida (province of Badajoz), after it was dismantled in the mid-9th century, to be reused at the Great Mosque of Cordoba. These reused materials were, therefore, neither local nor part of the church of San Vicente, and came from a distance of nearly 200 km. Along with this, the large group of pieces displayed today in the Visigothic Museum (located within the mosque) is actually a collection of heterogeneous pieces of different origins (Merida and Cordoba), chronology (6th to 9th centuries), and production, as well as coming from unknown contexts, so they cannot be used to argue for the presence of a church underneath the mosque (Utrero Agudo and Villa del Castillo 2023, p. 117). Finally, recent archaeological works undertaken in areas of the courtyard that were previously excavated in the 1930s have recorded a 5th-century building, thought to have been part of the episcopal complex and monumentalized during the 6th century (León Muñoz and Ortiz Urbano 2022). But no remains that can be attributed to a church have yet been discovered.

A similar timeline has been proposed for the former mosque and later medieval church of Santa Catalina and convent of Santa Clara. Excavations in the 1980s revealed lines of foundations that were thought to be part of a late 6th-century cruciform church set within a rectangular floor plan, with three semi-circular apses at the eastern end and a northern entrance (Marfil Ruiz 1996). This reconstruction has already been challenged by many researchers (compiled by Utrero Agudo and Villa del Castillo 2023, p. 116) due to the many archaeological and material obstacles for the reconstruction of this alleged church. Murillo Fragero (2023, pp. 384–85) also highlights that the foundation trenches of the 10th-century mosque cut through the collapse levels of the former construction sites, which probably date to a range of different periods (at least Roman and late Roman). The site, therefore, was already in ruins and abandoned when the mosque was built four centuries later, which clearly means that the location was not selected because of its sacral meaning, and that the previous structures were at the time probably neither visible nor identifiable.

4. Conclusions

Archaeology can help to assess the impact of new religions and, in doing so, the nature of the religion itself (Insoll 2004, p. 149). “Impact” can be read in many means, including sacralisation/re-sacralisation/de-sacralisation processes. In my view, the main issue to be clarified is not specifically the sacralisation, re-sacralisation or de-sacralisation of architectural spaces, but how widespread these practices were and whether they played a relevant role within religious change during the Late Antique and Early Medieval periods. Examples confirm what we already knew, but the real issue is to what extent this can be understood as a common and characterising phenomenon.

From the point of view of explanation, it appears that no Late Antique and Early Medieval church in the Iberian Peninsula reused earlier standing structures. I have also shown that none of the earliest mosques known to date in al-Andalus were built upon previously destroyed or abandoned churches, and that there are no clear instances of the conversion of Late Antique churches into mosques, or of their partition to be used simultaneously as a church and as a mosque. It thus seems that both early Christians (5th–6th centuries) and early medieval Christians and Muslims (8th–9th centuries) chose to build separate religious constructions rather than taking over and adapting pre-existing ones. Although there might have been some cases of intentional conversion, this does not seem to be the general rule. For churches or mosques located in densely populated Roman and/or Late Antique cities, the discovery of earlier structures underneath, not necessarily religious, is obviously not surprising. For constructions found in rural areas, existing communication networks, natural resources, and religious and geopolitical strategies, including the proximity of settlements and the need to express authority, marked the choice of location, rather than the deliberate intention of re-sacralising said locations. Therefore, constructions were more likely to respond to the location’s intrinsic advantages than to the presence of a ruin.

From a methodological perspective, stratigraphic sequences also reveal that, by and large, earlier structures seem to have decayed or to have come to ruin much before the construction of the new church or mosque, and that their traces were probably unrecognisable (and even invisible) to the new builders. In addition, the date of earlier constructions and of the new ones are often relatively widely spaced (e.g., churches dated to the 5th–6th century). On the other hand, the frequent reuse of Roman and Late Antique elements (mainly architectural sculpture) cannot be presented as evidence of a religious building on site. It has been proved many times that these elements moved from place to place, came from different regions, sometimes from a long distance, and were reused over and over again.

Only in some specific instances (Suso, Santiago de Compostela, Peñalba, Escalada) can an intended act of re-sacralisation be attested. Epigraphic and hagiographic records, material architectural evidence, and historical continuity converge in these instances; but, are we able to identify these processes of sacralisation/re-sacralisation/de-sacralisation when all these features are not in evidence?