1. INTRODUCTION

This work is a morphological and metrological study of the spaces identified as triclinia in four cities of Lusitania, Conimbriga (Condeixa-a-Velha), Capera (Cáparra), Augusta Emerita (Mérida), and Mirobriga (Santiago do Cacém), these being the urban centres that best represent the dining rooms of this provincial space, both due to the number of examples available for study and to their geographical distribution (Fig. 1).

Holding large social meals, in familiar, clientelist, or associative contexts, was an important element of socialization of daily life in Roman times, reflected, at least in the case of middle- and high-ranking families, in the way in which their dwellings were designed (Dunbabin 2003: 2 and 39-41Dunbabin, K. 2003: The Roman Banquet - Images of Conviviality. Cambridge University Press, Cambridge.).

In the Iberian Peninsula, rooms for holding banquets have been known since the 2^nd c. BCE, mainly in residential buildings at the Neapolis of Emporion. These spaces, identifiable as androi, show markedly Hellenic morphological characteristics and proportions (Olmos 1995: 52-55; Mar & Ruiz de Arbulo 1993: 364-376Mar, R.; and Ruiz de Arbulo, J. 1993: Ampurias romana: historia, arquitectura y arqueología. Ausa Editora, Sabadell.). This reflects the specificities of the furniture of Hellenic tradition, its arrangement being distinct from their Roman counterparts (Dunbabin 1998Dunbabin, K. 1998: “Ut Graeco More Biberetur: Greeks and Romans on the Dining Couch,” in I. Nielsen, and H. S. Nielsen (Eds.), Meals in a Social Context, pp. 81-101. Aarhus Studies in Mediterranean Antiquity, 1, Aarhus University Press, Aarhus.; Morvillez 2005Morvillez, E. 2005: “Sed nudo latere et parvis frons aerea lectis… Sur les dimensions des tapis en T+U et les types de lits employés,” in H. Morlier (dir.), La mosaïque gréco-romaine, IX, pp. 1325-1334. Collection de l’École française de Rome 352, École française de Rome, Roma.; Uribe Agudo 2013: 23-27Uribe Agudo, P. 2013: “Espacios convivales en las viviendas urbanas del Valle medio del Ebro desde la etapa postnumantina hasta el conflicto sertoriano,” Aquitania: revue interrégionale d’archéologie, 29, pp. 19-41. https://doi.org/10.3406/aquit.2013.1402 ).

However, these spaces, which were part of a Hellenized cultural matrix that, in Emporion, persisted until the Julio-Claudian period (Olmos 1985: 52Olmos, R. 1985: “Hedykoitos y agathos daimon. Inscripciones en dos mosaicos tardohelenísticos de Ampurias,” in Mesa redonda sobre el mosaico romano en España, 43-59. Ministerio de Cultura, Madrid.), would not constitute a direct antecedent of the Roman dining rooms found in most of the Peninsular space (Uribe Agudo 2013: 23-27Uribe Agudo, P. 2013: “Espacios convivales en las viviendas urbanas del Valle medio del Ebro desde la etapa postnumantina hasta el conflicto sertoriano,” Aquitania: revue interrégionale d’archéologie, 29, pp. 19-41. https://doi.org/10.3406/aquit.2013.1402 ). These are thought to be direct adoptions of the Roman model and, as such, are bound to reflect different social conducts and specificities of design. Spaces of Roman tradition designed for holding the convivium would accommodate the characteristic eating furniture, the lecti triclinares, typically arranged in a “U” shape. These would inittially reflect a strict moral conduct regarding the number of attendants and their designated seats, according to social status. However both the rooms and their associated norms seem to have evolved under the Empire, as dinning rooms grew larger, with consequent changes in the spacing and layout of the couches (Dunbabin 1998: 92Dunbabin, K. 1998: “Ut Graeco More Biberetur: Greeks and Romans on the Dining Couch,” in I. Nielsen, and H. S. Nielsen (Eds.), Meals in a Social Context, pp. 81-101. Aarhus Studies in Mediterranean Antiquity, 1, Aarhus University Press, Aarhus.)

Besides these purposely designed dining rooms, meals could also take place in various indoor or outdoor spaces prepared for the occasion, giving rise to the distinction between ‘dining areas’ and ‘dining rooms’ (Foss 1994: 105Foss, P. W. 1994: Kitchens and Dining Rooms at Pompeii: The Spatial and Social Relationship of Cooking to Eating in the Roman Household, A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Classical Art and Archaeology) in The University of Michigan, I.; Uribe Agudo 2009: 154Uribe Agudo, P. 2009: “«Triclinia» y salones triclinares en las viviendas romanas urbanas del cuadrante nordeste de la Península Ibérica (I a.C.-III d.C.),” Archivo español de arqueología, 82, pp. 153-189. https://doi.org/10.3989/aespa.082.009.007 ).

Textual sources use different names for these spaces, namely triclinium, cenatio, oecus, cenaculum, or even tablinum, but often do not provide a description that might clarify whether these terms were synonymous or what the differentiating characteristics might be (Uribe Agudo 2013: 20Uribe Agudo, P. 2013: “Espacios convivales en las viviendas urbanas del Valle medio del Ebro desde la etapa postnumantina hasta el conflicto sertoriano,” Aquitania: revue interrégionale d’archéologie, 29, pp. 19-41. https://doi.org/10.3406/aquit.2013.1402 ). On the other hand, some designations seem to relate to the dimensions of the rooms - cenatio; cenatiuncula - or to have a qualitative value - cenatio modica (Plin. Ep. 2.17.10); cenatio cotidiana (Plin. Ep. 5.6.21); cenatio mica (Mart. 2. 59) (Foss 1994: 85-88Foss, P. W. 1994: Kitchens and Dining Rooms at Pompeii: The Spatial and Social Relationship of Cooking to Eating in the Roman Household, A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Classical Art and Archaeology) in The University of Michigan, I.).

Vitruvius sustains the ideal proportion of a length that doubled the width, while also providing prescriptions regarding their solar orientation (Vit. VI, III, 8), concerning use according to the seasons - hibernum triclinium (Vit. VI, IV, 1), triclinium vernum, triclinium autumnalium and triclinium aestivum (Vit. VI, IV, 2).

According to the testimony of Marcius Servius HonoratusServius Honoratus, M.: Servii Grammatici qui feruntur in Vergilii carmina commentarii, recensuerunt G. Thilo et H. Hagen (1881), Georgius Thilo, B. G. Teubner, Leipzig. Available at Maurus Servius Honoratus, Commentary on the Aeneid of Vergil, SERVII GRAMMATICI IN VERGILII AENEIDOS LIBRVM PRIMVM COMMENTARIVS., line 698 (tufts.edu) (Serv. A. 1.698), in the 4^th c. CE the term triclinium lost its literal value since it continued to be used even in contexts where it supposedly was not applicable, as in the case of dining rooms equipped with a stibadium, which had only one piece of furniture of semi-circular shape. In these cases, according to the same source, the most correct terminology for this type of room would be cenatio or basilica (Foss 1994: 86Foss, P. W. 1994: Kitchens and Dining Rooms at Pompeii: The Spatial and Social Relationship of Cooking to Eating in the Roman Household, A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Classical Art and Archaeology) in The University of Michigan, I.).

Roman iconographic sources start representing banqueting scenes from 1^st the century BC to 5^th c. CE, attesting to the enduring significance of these social practices. Apart from the occasional drapery or clearly idealised scenery, little is shown of the rooms where these banquets were held. These representations come mainly from funerary or domestic contexts, and in the form of stone reliefs, mosaics, and pictorial depictions; and while the first two come from diverse geographical and chronological backgrounds, the latter come essentially from Vesuvian sites affected by the eruption of 79 CE (Dunbabin 2003: 4-5Dunbabin, K. 2003: The Roman Banquet - Images of Conviviality. Cambridge University Press, Cambridge.).

Commonly, crockery for drinks and foodstuffs is seen in small tables placed in front of the reclined diners; in secondary tables, or placed over heating devices both of which are located in the anteroom of the banqueting scene (Dunbabin 2003: 55, fig. 26; 71, plate XIIDunbabin, K. 2003: The Roman Banquet - Images of Conviviality. Cambridge University Press, Cambridge.).

As is observed in these representations, holding meals in these spaces would require specific furniture, implying that such rooms must have appropriate dimensions for placing the couches where people would lie down/recline, and also to fit other types of furniture like boilers for heating food and drinks.

Taking this into consideration, this study aims at characterising and comparing the dimensions of triclinia documented in the urban dwellings of Lusitania, in an attempt to identify which dimensions would be characteristic or appropriate for these spaces in this provincial space.

2. IDENTIFYING THE DINING ROOMS

The spaces interpreted as having this function are generally located in the central axis of the house, with a large opening onto a peristyle or gardened space (Farrar 2013: 15-19Farrar, L. 2013: Ancient Roman Gardens. The History Press, Gloucestershire.), and feature dimensions that distinguish them from the remaining spaces. This distinction is seen as a reflection of their importance, consequently identifying them as spaces of representation, usually due to their architectural features and decorative quality.

In some cases, parietal or floor decoration was used to distinguish two areas: an entrance or reception area and the space in which the banqueting itself would take place (Guiral Pelegrín & Mostalac Carrillo 1993Guiral Pelegrín, C.; and Mostalac Carrillo, A. 1993: “Influencias itálicas en los programas decorativos de cubicula y triclinia de época republicana y altoimperial en España, Algunos ejemplos representativos,” Espacio, Tiempo y Forma, Serie I, 6, pp. 365-392. https://doi.org/10.5944/etfi.6.1993.4593 ; Dunbabin 2003: 41-42Dunbabin, K. 2003: The Roman Banquet - Images of Conviviality. Cambridge University Press, Cambridge.; Uribe Agudo 2009: 155Uribe Agudo, P. 2009: “«Triclinia» y salones triclinares en las viviendas romanas urbanas del cuadrante nordeste de la Península Ibérica (I a.C.-III d.C.),” Archivo español de arqueología, 82, pp. 153-189. https://doi.org/10.3989/aespa.082.009.007 ; Correia 2013: 187-188Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.). In these cases, approximately 2/3 of the space was dedicated to the placement of the lecti triclinares, while the remaining 1/3 was allocated to the entrance or reception area (Uribe Agudo 2009: 155Uribe Agudo, P. 2009: “«Triclinia» y salones triclinares en las viviendas romanas urbanas del cuadrante nordeste de la Península Ibérica (I a.C.-III d.C.),” Archivo español de arqueología, 82, pp. 153-189. https://doi.org/10.3989/aespa.082.009.007 ; Guiral Pelegrín 2018: 621Guiral Pelegrín, C. 2018: “Cubicula y triclinia pintados en Hispania: articulación del espacio, sistemas decorativos e iconografía2,” in Y. Dubois, M. E. Fuchs, and A. Spühler (Org.), Pictores per Provincias II - Status Quaestionis, Actes du 13e Colloque de l’Association Internationale pour la Peinture Murale Antique (AIPMA), Université de Lausanne, Anthropole, 12 au 16 septembre 2016, pp. 621-638.). This distribution is noticeable in terms of the mosaic floors designed for these spaces, particularly in the “T + U” configuration, the “U” being reserved for the placement of lecti around a central carpet (Dunbabin 2003: 41-42Dunbabin, K. 2003: The Roman Banquet - Images of Conviviality. Cambridge University Press, Cambridge.; Uribe Agudo 2009: 156-158Uribe Agudo, P. 2009: “«Triclinia» y salones triclinares en las viviendas romanas urbanas del cuadrante nordeste de la Península Ibérica (I a.C.-III d.C.),” Archivo español de arqueología, 82, pp. 153-189. https://doi.org/10.3989/aespa.082.009.007 ; Correia 2013: 261-262Cardullo, S. 2016: Le Antiche Misure Romane. Casa Editrice Kimerik, Messina.). The remaining area, near the entrance, would be reserved for the circulation of the servants attending the banquet, but also for the placement of support furniture, namely boilers or braziers for heating food and drinks, as evidenced by several representations, such as the fresco of the triclinium of the Casa dei Casti Amanti in Pompeii or the mosaic of the triclinium of the House of Orpheus in Sepphoris (Dunbabin 2003: 55, 166 and 167Dunbabin, K. 2003: The Roman Banquet - Images of Conviviality. Cambridge University Press, Cambridge.).

In the Vesuvian area, for example, some rooms present niches dug in the walls¹This happens, for example, in the triclinium of Casa del Gran Portale, in Herculaneum, and Casa del Cinghiale, in Pompeii (Mols 2008: fig. 17)., made to fit the furniture, namely beds (Mols 2020: 38Mols, S. 2020: Wooden furniture in Herculeneum - form, technique and function. Brill, Leiden/Boston.) and triclinium lecti (Mols 2008: 155Mols, S. 2008: “Ancient Roman household furniture and its use: From Herculaneum to the Rhine,” Anales de Prehistoria y Arqueología, 23-24, pp. 145-160.). This architectural feature immediately tells us what the function of the room was at a given time.

The presence of these niches, however, is not common and suggests that the rooms had to be adapted to the furniture. This further suggests that the furniture size was somewhat standardized (Mols 2020: 38Mols, S. 2020: Wooden furniture in Herculeneum - form, technique and function. Brill, Leiden/Boston.), or that a specific size was commonly desired²This idea is backed by the study of the preserved furniture from Herculaneum (De Carolis 2007; Mols 2020)..

Simultaneously, this could also imply that these rooms were not originally meant to be dining rooms, hence their adaptation to fit the lecti.³This is the case of Casa del Gran Portale, in Herculaneum, which was restructured after the earthquake of 62 AD, where the triclinium was fashioned from a previous representation room (Dardenay 2021: 110-112).

On the whole, this suggests that the dining rooms had specific architectonic features and a layout that required a minimum size for functioning properly.

The existence of a relationship between the dimensions and function of dining rooms has already been proposed by Éric Morvillez (1996Morvillez, E. 1996: “Sur les installations de lits de table en sigma dans l’architecture domestique du Haut e du Bas-Empire,” Pallas, 44, pp. 119-158. https://doi.org/10.3406/palla.1996.1386 and 2005)Morvillez, E. 2005: “Sed nudo latere et parvis frons aerea lectis… Sur les dimensions des tapis en T+U et les types de lits employés,” in H. Morlier (dir.), La mosaïque gréco-romaine, IX, pp. 1325-1334. Collection de l’École française de Rome 352, École française de Rome, Roma. and Katherine Dunbabin (1998)Dunbabin, K. 1998: “Ut Graeco More Biberetur: Greeks and Romans on the Dining Couch,” in I. Nielsen, and H. S. Nielsen (Eds.), Meals in a Social Context, pp. 81-101. Aarhus Studies in Mediterranean Antiquity, 1, Aarhus University Press, Aarhus., in their works on room dimensions and the respective mosaic carpets, suggesting their suitability for certain types of lecti and, consequently, for different convivial contexts. Regarding the Peninsular area, María Pilar Galve Izquierdo has also observed similarities in the dimensions of some triclinia of Celsa (Casa de la Tortuga and Casa del Emblema) and the opus signinum pavement of the Pamplona Archdeaconry (Pompaelo) (Galve Izquierdo 1996: fig. 103 apudGalve Izquierdo, M. P. 1996: Los antecedentes de Caesaraugusta, Estructuras domésticas de Salduie. Institución Fernando el Católico, Zaragoza. Uribe Agudo 2009: 158Uribe Agudo, P. 2009: “«Triclinia» y salones triclinares en las viviendas romanas urbanas del cuadrante nordeste de la Península Ibérica (I a.C.-III d.C.),” Archivo español de arqueología, 82, pp. 153-189. https://doi.org/10.3989/aespa.082.009.007 ). Likewise, Virgílio H. Correia has associated some of the dining rooms in Conimbriga, which feature a distinct mosaic layout, with a different social context of use (Correia 2013: 265Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.).

We have focused only on the sizing of the rooms themselves as their arrangement constitutes a separate line of inquiry that largely exceeds the scope of this paper.

3. THE METROLOGICAL ANALYSIS OF DINING ROOMS

4. CONIMBRIGA

Twelve dining rooms have been documented in this city (Correia 2013Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.). The identification of this type of room is relatively reliable in many cases due to the opus tesselatum or opus signinum markings for the placement of lecti on the pavements. In other cases, it is suggested by the dimensions and layout of the rooms.

The premise that the representational rooms are situated on an axial arrangement in relation to the centre of the house, presiding over gardened spaces or refreshed through water displays and reflecting pools, has a clear correspondence in Conimbriga, including the existing secondary dining rooms, either at the Casa de Cantaber or the Casa dos Repuxos⁶Also published under the designation Maison aux jets d’eau (Morand 1996).. Although they are also situated in smaller axialised areas, the architectural treatment of these peristyles reveals a similar care in the scenography, intended to be visualised from the dining room (Fig. 2).

Two other dining rooms were also documented in Conimbriga, which are not located in residential buildings. One of them is situated in the Edifício das Latrinas do Forum and is part of a set consisting of a succession of three compartments, the last of which features a pavement in opus signinum displaying the demarcation of an area intended for the placement of lecti. The second one is situated in Edifício Oeste da Zona C, a building only partially known. It also displays a configuration with a succession of three compartments, one of which features a marking on the opus signinum pavement defining a “U”- shaped area for the placement of lecti. Buildings with this type of structuring are not uncommon in Conimbriga. It has been suggested that these buildings could be scholae or the sedes of collegia based on their layout (Correia 2016Correia, V. H. 2016: “A presença dos espaços de reunião no tecido urbano: o exemplo da cidade de Conimbriga (prov. Lusitania, Portugal)”, in O. Rodríguez Gutiérrez, N. Tran, and B. Soler Huertas (coords.), Los espacios de reunión de las asociaciones romanas, Diálogos de la Arqueologia y la Historia en homenaje a Bertrand Goffaux, pp. 345-357. Editorial Universidad de Sevilla, Sevilla.).

Regarding the chronology of the buildings where dining rooms have been identified, the oldest are the Insula do Aqueduto, from the early 1^st c. CE (Correia 2013: 131Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.) and the Casa do Medianum Absidado, pre-dating the mid-1^st c. CE (Correia 2013: 92Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.), while the majority date from the Flavian period and early 2^nd c. CE, namely Casa de Cantaber (Correia 2001: 123Correia, V. H. 2001: “Conimbriga, Casa atribuída a Cantaber, Trabalhos arqueológicos 1995-1998,” Conimbriga, 40, pp. 83-140. https://doi.org/10.14195/1647-8657_40_4 ; 2013: 101Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.), Casa dos Repuxos, which has a previous phase that will not be discussed here (Correia 2013: 154-156Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.), Casa dos Esqueletos (Correia 2013: 171Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.) and the Edifício da Latrinas do Forum (Correia 2013: 118Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.). The most recent building is Casa da Cruz Suástica which, although also built in the 1^st c. CE, was redesigned in the 3^rd c. CE, resulting in its final appearance (Alarcão 2010: 28-45Alarcão, J. 2010: As Casas da Zona B de Conimbriga. Centro de Estudos Arqueológicos das Cidades de Coimbra e Porto, Coimbra.; Correia 2013: 166Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.). No chronological data are currently available for the Edifício Oeste da Zona C (Correia 2013: 124-125Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.).

The identified dining rooms feature considerably different scales, and three different sizes can immediately be observed in houses with several dining rooms, such as Casa de Cantaber and Casa dos Repuxos (Tab. 1).

The larger rooms, identified as oeci triclinia (Correia 2013: 153, 260-262Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.), have been documented in Casas de Cantaber, Repuxos, Suásticas, and Esqueletos and measure between 716 and 780 cm in width and between 940 and 1335 cm in length.

The rooms with intermediate dimensions have been identified as either triclinia or cenationes. The latter designation was attributed to Room 6 of Casa de Cantaber and Room 29 of Casa dos Repuxos (Correia 2013: 140 and 265Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.), measuring 630 cm in width by 910 cm in length and 610 cm in width by 810 cm in length, respectively. The latter also displays an interesting feature in terms of its construction technique, which consists in the construction of two brick walls flush against the south and east walls of the room, with an apparent cladding function, a situation we shall address in the section dealing with the discussion of the data. In addition, the dining room identified at the Edifício Oeste da Zona C and considered a triclinium (Correia 2013: 125Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.) also has similar width dimensions, at 620 cm, as do the triclinium of the Edifício das Latrinas do Forum, Room 16 of the Insula do Aqueduto and Room 8 of Casa do Medianum Absidado, although these are slightly smaller than the previously mentioned ones.

Two smaller rooms are also considered possible cenationes (Correia 2013: 139 and 154Correia, V. H. 2013: A Arquitectura Doméstica de Conimbriga e as Estruturas Económicas e Sociais da Cidade Romana, Anexos de Conimbriga 5, Coimbra.). These feature measurements of 403 cm in width and 466 cm in length in the case of Room 24 of Casa dos Repuxos, and, in the case of Room 65 of the Casa de Cantaber, a width of 420 cm, with a restored length of 620 cm⁷According to the plans published in Correia (2001 and 2013). The room is not completely exposed nowadays..

5. CAPERA

In Capera, the several excavations that took place since 1967 have led to the identification of three residential buildings: Capera 1, Capera 2, and Capera 3 (Bejarano 2020: 110-114Bejarano Osorio, A. 2020: “Capera (Cáparra, España)”. in A. Pizzo (ed.), La arquitectura doméstica urbana de la Lusitana romana, pp. 109-116. Serie MYTRA, 6, Instituto de Arqueología de Mérida, Mérida.).

Of these, only Capera 3 features a compartment identifiable as a dining room. The building was identified during the works carried out between 2009 and 2010 as part of the Proyecto Via de la Plata II, which focused on the insula to the south of the forum (Gijón 2013: 1461 and 1464Gijón Gabriel, M. E. 2013: “El Proyecto Vía de la Plata II - Extremadura: Actuaciones en la ciudad romana de Cáparra,” in J. Jiménez Ávila, M. Bustamante Álvarez and M. García Cabezas (Eds.), Actas del VI Encuentro de Arqueología del Suroeste Peninsular, pp. 1459-1470. Gráfica Rejas, Mérida.).

Its construction took place between the end of the 2^nd and the beginning of the 3^rd c. CE; it was abandoned around 270 CE and reoccupied and transformed from the end of the 4^th c. CE onwards. One of the most evident transformations was the longitudinal subdivision of the triclinium through the construction of a wall, which also spanned the corridor of the peristyle (Gijón 2013: 1466 and fig. 2Gijón Gabriel, M. E. 2013: “El Proyecto Vía de la Plata II - Extremadura: Actuaciones en la ciudad romana de Cáparra,” in J. Jiménez Ávila, M. Bustamante Álvarez and M. García Cabezas (Eds.), Actas del VI Encuentro de Arqueología del Suroeste Peninsular, pp. 1459-1470. Gráfica Rejas, Mérida.).

This house features two peristyles, a bigger one and a smaller one. The room identified as a possible triclinium is situated in the northeast corridor of the bigger peristyle (Fig. 3). This room, besides displaying an elaborated parietal decoration in relief, features distinctive dimensions (Tab. 2) that justify its interpretation as triclinium (Gijón 2013: 1466Gijón Gabriel, M. E. 2013: “El Proyecto Vía de la Plata II - Extremadura: Actuaciones en la ciudad romana de Cáparra,” in J. Jiménez Ávila, M. Bustamante Álvarez and M. García Cabezas (Eds.), Actas del VI Encuentro de Arqueología del Suroeste Peninsular, pp. 1459-1470. Gráfica Rejas, Mérida.).

6. AUGUSTA EMERITA

The dining rooms that have been recognised in Augusta Emerita are relatively few compared to the number of identified domestic buildings (Corrales Álvarez 2016a: 217Corrales Álvarez, A. 2016a: La Arquitectura Doméstica de Augusta Emerita. Anejos de AEspA LXXVI, CSIC, Madrid.).

In fact, out of the 178 residential buildings inventoried (Corrales Álvarez 2016bCorrales Álvarez, A. 2016b: La Arquitectura Doméstica de Augusta Emerita: Catálogo de la edilicia doméstica emeritense. Anejos de AEspa LXXVI, CSIC.), only sixteen rooms can be somewhat reliably interpreted as being linked to this function and allow this type of analysis. Amongst these, there are seven representational rooms or triclinia featuring apses. These were intended for a later style of triclinium furniture, the stibadium (Corrales Álvarez 2016a: 218-220Corrales Álvarez, A. 2016a: La Arquitectura Doméstica de Augusta Emerita. Anejos de AEspA LXXVI, CSIC, Madrid.), and for this reason not included in this work.

Nearly all of the nine studied examples are situated in an axial or semi-axial position with a scenic framing provided by a peristyle, the only exceptions being Room U of Casa de Mitreo, which is an inner room, the room found in Casa de la Calle Pareja 32, whose excavation area did not allow its position to be identified in the plan, and Room 29 / Room A’ of Casa del Anfiteatro which, although it apparently faces an exterior space, is an area whose architectural structure is unknown because it has not been fully excavated (Fig. 4).

The oldest residential buildings are dated to the 1^st and 2^nd c. CE, namely Casa del Mitreo, from the 1^st c. CE (Lequement 1977: 50Lequement, R 1977: “Rapport preliminaire sur deux sondages effectués a Mérida, septembre-octubre 1973“, in Noticiario Arqueológico Hispano, 5, pp. 147-166. and Gijón 2002Gijón Gabriel, M. E. 2002: Informe sobre la intervención arqueológica realizada en la Casa del Mitreo (Mérida). Departamento de Documentación del Consorcio de la ciudad monumental de Mérida, nº de intervención 151. apud Corrales Álvarez 2016b: 881Corrales Álvarez, A. 2016b: La Arquitectura Doméstica de Augusta Emerita: Catálogo de la edilicia doméstica emeritense. Anejos de AEspa LXXVI, CSIC.), the Casa de la Calle Pareja 32 (Barrientos 2000Barrientos Vera, T. 2000: “Intervención arqueológica en el solar nº32 de la C/ Pareja. Un ejemplo de reutilización de estructuras desde época altoimperial hasta la tardoantiguedad,” Memoria, 4, pp. 221-277. apud Corrales Álvarez 2016b: 139Corrales Álvarez, A. 2016b: La Arquitectura Doméstica de Augusta Emerita: Catálogo de la edilicia doméstica emeritense. Anejos de AEspa LXXVI, CSIC.) and Casa Alcazaba 2, both from the 2^nd c. CE (Corrales Álvarez 2016b: 553Corrales Álvarez, A. 2016b: La Arquitectura Doméstica de Augusta Emerita: Catálogo de la edilicia doméstica emeritense. Anejos de AEspa LXXVI, CSIC.). The grounds of the Casa del Anfiteatro probably featured a previous 1^st c. CE construction of some kind, but it was demolished and replaced by another one, built in the 3^rd c. CE (García 1966: 43García, E. 1966: Informe sobre las casas romanas de Mérida y excavaciones en la Casa del Anfiteatro. EAE, 49. Madrid: Ministerio de Educación Nacional, Dirección General de Bellas Artes, Servicio Nacional de Excavaciones Arqueológicas. ; Hernández 1993: 778Hernández, J. 1993: La pintura mural romana de Mérida (inserta en la estructura urbana y doméstica de la ciudad). Unpublished Ph.D. dissertation submitted to the Departamento de Prehistoria y Arqueología, Universidad Nacional de Educación a Distancia, Madrid.; Mateos 1995: 200Mateos, P. 1995: “Proyecto de arqueología urbana en Mérida: desarrollo y primeros resultados,” Extremadura Arqueológica, IV, pp. 191-215.apudCorrales Álvarez 2016b: 836Corrales Álvarez, A. 2016b: La Arquitectura Doméstica de Augusta Emerita: Catálogo de la edilicia doméstica emeritense. Anejos de AEspa LXXVI, CSIC.).

Dining rooms of different scales have also been identified in Augusta Emerita (Tab. 3). The largest, Room 29 / Room A’ of Casa del Anfiteatro (Pizzo 2004: fig. 2Pizzo, A. 2004: “La Casa de Anfiteatro de Augusta Emerita”, Memoria 7, Mérida - Excavaciones Arqueológicas 2001, Consorcio Ciudad Monumental Histórico-Artística y Arqueológica de Mérida, Mérida, pp. 335-349.; Corrales Álvarez 2016b: fig. 1005Corrales Álvarez, A. 2016b: La Arquitectura Doméstica de Augusta Emerita: Catálogo de la edilicia doméstica emeritense. Anejos de AEspa LXXVI, CSIC.), corresponds to a large room (Tab. 3) surrounded by a corridor (Pizzo 2004: 344Pizzo, A. 2004: “La Casa de Anfiteatro de Augusta Emerita”, Memoria 7, Mérida - Excavaciones Arqueológicas 2001, Consorcio Ciudad Monumental Histórico-Artística y Arqueológica de Mérida, Mérida, pp. 335-349.). This room has been the subject of various interpretations and has been referred to as a possible tablinum (García 1966: 26García, E. 1966: Informe sobre las casas romanas de Mérida y excavaciones en la Casa del Anfiteatro. EAE, 49. Madrid: Ministerio de Educación Nacional, Dirección General de Bellas Artes, Servicio Nacional de Excavaciones Arqueológicas. ), triclinium (Balil 1976: 90Balil, A. 1976: “Sobre la arquitectura doméstica en Emerita,” in A. Blanco (ed.): Augusta Emerita. Actas del Simposio Internacional Conmemorativo del Bimilenario de Mérida, pp. 75-91. Dirección General del Patrimonio Artístico y Cultural del Ministerio de Educación y Ciencia y Patronato de la Ciudad de Mérida, Madrid.), or oecus (Sánchez & Nodar 1999: 371Sánchez, G.; and Nodar, R. 1999: “Reflexiones sobre las casas suburbanas en Augusta Emerita: estudio preliminar,” Memoria, 3, pp. 367-386.), although the “T + U” shaped structure of its mosaic pavement indicates that it is a dining room, commonly identifiable as triclinium (Corrales Álvarez 2016a: 217Corrales Álvarez, A. 2016a: La Arquitectura Doméstica de Augusta Emerita. Anejos de AEspA LXXVI, CSIC, Madrid.). It is the only exemplar with these dimensions ever found in the city and present in this sample.

The rooms with intermediate dimensions feature widths varying between ca. 715 cm and 770 cm, and even more variable lengths, between ca. 420 cm and 1355 cm. They are represented by Room R of the Casa del Mitreo, Room E of the Casa Alcazaba 2, and the Salón of the Casa Morería 6, corresponding to a room of large dimensions, similar to those commonly interpreted as triclinium or oecus (Alba 2011: fig. 3Alba Calzado, M. 2011: “Los espacios domésticos en la ciudad visigoda de Emerita (ss. V-VIII),” in J. M. Álvarez and P. Mateos (eds.): Congreso Internacional 1910-2010, el yacimiento emeritense, pp. 521-546. Ayuntamiento de Mérida, Mérida. apud Corrales Álvarez 2016b: 359Corrales Álvarez, A. 2016b: La Arquitectura Doméstica de Augusta Emerita: Catálogo de la edilicia doméstica emeritense. Anejos de AEspa LXXVI, CSIC.).

Two other spaces may be classified as intermediate in size, but are slightly smaller than those listed above. These correspond to the room identified at the Casa da Calle Pareja 32, interpreted as a possible oecus or triclinium on account of both its dimensions and decorative features (Barrientos Vera 2000: 229-232Barrientos Vera, T. 2000: “Intervención arqueológica en el solar nº32 de la C/ Pareja. Un ejemplo de reutilización de estructuras desde época altoimperial hasta la tardoantiguedad,” Memoria, 4, pp. 221-277. apud Corrales Álvarez 2016b: 139-141Corrales Álvarez, A. 2016b: La Arquitectura Doméstica de Augusta Emerita: Catálogo de la edilicia doméstica emeritense. Anejos de AEspa LXXVI, CSIC.), with a width of ca. 645 cm wide, but with an indeterminate length due to the state of preservation of the structure; and also, to Room 22/Room C from Casa del Anfiteatro (Pizzo 2004: fig. 2Pizzo, A. 2004: “La Casa de Anfiteatro de Augusta Emerita”, Memoria 7, Mérida - Excavaciones Arqueológicas 2001, Consorcio Ciudad Monumental Histórico-Artística y Arqueológica de Mérida, Mérida, pp. 335-349.; Corrales Álvarez 2016b: fig. 1005Corrales Álvarez, A. 2016b: La Arquitectura Doméstica de Augusta Emerita: Catálogo de la edilicia doméstica emeritense. Anejos de AEspa LXXVI, CSIC.) with a width of ca. 610 cm and a length of ca. 910 cm. This room is situated on the central axis of the house, aligned with the entrance vestibulum and at the centre of the peristyle, and has been considered a representational room (García 1966: 23García, E. 1966: Informe sobre las casas romanas de Mérida y excavaciones en la Casa del Anfiteatro. EAE, 49. Madrid: Ministerio de Educación Nacional, Dirección General de Bellas Artes, Servicio Nacional de Excavaciones Arqueológicas. apud Corrales Álvarez 2016b: 836Corrales Álvarez, A. 2016b: La Arquitectura Doméstica de Augusta Emerita: Catálogo de la edilicia doméstica emeritense. Anejos de AEspa LXXVI, CSIC.). A second opening was identified at the far end of this room. This opening was eventually walled up, but it originally extended the visual axis towards another part of the house, possibly an outside area. Although no mosaic pavement was found during the excavations, this set of characteristics supports the interpretation of the room as a triclinium aestivum (Pizzo 2004: 343Pizzo, A. 2004: “La Casa de Anfiteatro de Augusta Emerita”, Memoria 7, Mérida - Excavaciones Arqueológicas 2001, Consorcio Ciudad Monumental Histórico-Artística y Arqueológica de Mérida, Mérida, pp. 335-349.).

The smaller rooms correspond to the compartments identified as triclinium at Casa Moreria 6 (Alba 2011: fig. 3Alba Calzado, M. 2011: “Los espacios domésticos en la ciudad visigoda de Emerita (ss. V-VIII),” in J. M. Álvarez and P. Mateos (eds.): Congreso Internacional 1910-2010, el yacimiento emeritense, pp. 521-546. Ayuntamiento de Mérida, Mérida. apud Corrales Álvarez 2016b: 359Corrales Álvarez, A. 2016b: La Arquitectura Doméstica de Augusta Emerita: Catálogo de la edilicia doméstica emeritense. Anejos de AEspa LXXVI, CSIC.), to Room U of Casa del Mitreo, interpreted as a triclinium aestivum (Corrales Álvarez 2016b: fig. 1063Corrales Álvarez, A. 2016b: La Arquitectura Doméstica de Augusta Emerita: Catálogo de la edilicia doméstica emeritense. Anejos de AEspa LXXVI, CSIC.), and Room 23/Room D of Casa del Anfiteatro, also considered a triclinium on account of being paved by a high-quality mosaic known as “mosaico del otoño” [lit. Autumn mosaic] (Balil 1976: 88Balil, A. 1976: “Sobre la arquitectura doméstica en Emerita,” in A. Blanco (ed.): Augusta Emerita. Actas del Simposio Internacional Conmemorativo del Bimilenario de Mérida, pp. 75-91. Dirección General del Patrimonio Artístico y Cultural del Ministerio de Educación y Ciencia y Patronato de la Ciudad de Mérida, Madrid. apud Pizzo 2004: 343Pizzo, A. 2004: “La Casa de Anfiteatro de Augusta Emerita”, Memoria 7, Mérida - Excavaciones Arqueológicas 2001, Consorcio Ciudad Monumental Histórico-Artística y Arqueológica de Mérida, Mérida, pp. 335-349.; Corrales Álvarez 2016b: fig. 1005Corrales Álvarez, A. 2016b: La Arquitectura Doméstica de Augusta Emerita: Catálogo de la edilicia doméstica emeritense. Anejos de AEspa LXXVI, CSIC.).

7. MIROBRIGA

Dining rooms have already been identified in four houses in Mirobriga, namely Room 7 of Casa da Calçada⁸Also referred to as Construção/Domus 3 in Quaresma 2012. (Sousa & Felício 2020: 1373Sousa, F.; and Felício, C. 2020: “A Arquitectura Residencial em Mirobriga (Santiago do Cacém): Contributo a partir de um estudo de caso,” in III Congresso da Associação dos Arqueólogos Portugueses 2020 - O Estado da Questão, pp. 1371-1384.), Room 5.1.2 of Casa Periquito/Casa 5⁹Formerly referred to as Casa 1 in Teichner et al. 2014. (Kopf 2018: 59Kopf, J. 2018: “Archäologische und stratigraphische üntersuchungen zu Haus 5,” in F. Teichner (Ed.), Mirobriga, Eine Stadt im fernen Westen des Imperium Romanum, pp. 58-91. Kleine Schriften aus dem Vorgeschichtlichen Seminar Marburg, Heft 62, Philipps - Universitat Marburg.), Room 8.3.5 of Casa 8¹⁰Formerly referred to as Casa 4 in Teichner et al. 2014. (Oberhofer 2018b: 137Oberhofer, K. 2018b: “Archäologische und stratigraphische üntersuchungen zu Haus 8,” in F. Teichner (Ed.), Mirobriga, Eine Stadt im fernen Westen des Imperium Romanum, pp. 128-147. Kleine Schriften aus dem Vorgeschichtlichen Seminar Marburg, Heft 62, Philipps - Universitat Marburg.) and representational space, Room 7.1.2 of Casa 7¹¹Formerly referred to as Casa 3 in Teichner et al. 2014. (Oberhofer 2018a: 110Oberhofer, K. 2018a: “Archäologische und stratigraphische üntersuchungen zu Haus 7,” in F. Teichner (Ed.), Mirobriga, Eine Stadt im fernen Westen des Imperium Romanum, pp. 109-127. Kleine Schriften aus dem Vorgeschichtlichen Seminar Marburg, Heft 62, Philipps - Universitat Marburg.). There are other probable dining rooms, namely Room 2 of Casa dos Frescos¹²Also referred to as «Casa Oeste» in Rosário et al. 2002; however, this designation corresponds to the area of the city where the «West Houses» are located (Biers 1988). and Room 13 of Casa da Hospedaria (Casa 22 in Cortés 2018, p. 153-154Cortés, A. 2018: “The Roman Houses of Flavian Mirobriga and Their Evolution - A first approximation to the archaeological remains”, in F. Teichner (Ed.), Mirobriga, Eine Stadt im fernen Westen des Imperium Romanum, pp. 148-155. Kleine Schriften aus dem Vorgeschichtlichen Seminar Marburg, Heft 62, Philipps - Universitat Marburg.), two partially preserved rooms whose length and width dimensions support their interpretation as dining rooms. However, both the Casa dos Frescos and the Casa da Hospedaria do not feature all the necessary measurements for a complete analysis due to their state of preservation. In the first case, only the width was preserved, and in the second, only the length. Regarding Casa 8 and in the scope of this study, only the room interpreted as triclinium in the 3^rd phase was considered¹³The proposal presented by the author for the floorplans of the two previous phases isn’t, in our view, supported by the data. (Fig. 5).

Although the plan of the buildings is fairly preserved, most of these spaces have no preserved floor, with the only exception being the so-called Casa dos Frescos, which presents an opus signinum floor that covers the full extent of the preserved room. This floor, however, doesn’t show any indicators regarding the placement of lecti. Casa dos Frescos, as given away by its name, yielded a mural painting identified in the space that can be interpreted as a dining room. The layout of this decorative element also does not provide any additional information regarding the organisation of the space. Room 7 in Casa da Calçada hasn’t, to this date, been fully excavated and may still have a preserved floor.

As such, the identification of the dining rooms was essentially based on an empirical comparison with more complete examples from other sites, based on their dimensions and position in plan (Cortés 2018: 148-152Cortés, A. 2018: “The Roman Houses of Flavian Mirobriga and Their Evolution - A first approximation to the archaeological remains”, in F. Teichner (Ed.), Mirobriga, Eine Stadt im fernen Westen des Imperium Romanum, pp. 148-155. Kleine Schriften aus dem Vorgeschichtlichen Seminar Marburg, Heft 62, Philipps - Universitat Marburg.).

The chronology of Mirobriga’s domestic buildings has been ascribed to the time between the Flavian period and the beginning of the 2^nd c. CE, namely Casa 7 (Oberhofer 2018a: 109Oberhofer, K. 2018a: “Archäologische und stratigraphische üntersuchungen zu Haus 7,” in F. Teichner (Ed.), Mirobriga, Eine Stadt im fernen Westen des Imperium Romanum, pp. 109-127. Kleine Schriften aus dem Vorgeschichtlichen Seminar Marburg, Heft 62, Philipps - Universitat Marburg.), Casa Periquito (Kopf 2018: 58Kopf, J. 2018: “Archäologische und stratigraphische üntersuchungen zu Haus 5,” in F. Teichner (Ed.), Mirobriga, Eine Stadt im fernen Westen des Imperium Romanum, pp. 58-91. Kleine Schriften aus dem Vorgeschichtlichen Seminar Marburg, Heft 62, Philipps - Universitat Marburg.) and Casa da Calçada (Sousa & Felício 2020: 1374Sousa, F.; and Felício, C. 2020: “A Arquitectura Residencial em Mirobriga (Santiago do Cacém): Contributo a partir de um estudo de caso,” in III Congresso da Associação dos Arqueólogos Portugueses 2020 - O Estado da Questão, pp. 1371-1384.).

Casa da Hospedaria is possibly the oldest example with a proposed chronology for its construction in the middle of the 1^st c. CE (Biers et al. 1982, p. 37-39Biers, J.; Biers, W.; and Soren, D. 1982: “Excavations at Mirobriga: the 1982 season,” Muse, 16, pp. 29-43.; Slane et al. 1984: 61-62Slane, K. W.; Biers, J; Darling, J. K.; Miksicek, C.; and Soren, D. 1984: “Mirobriga: a Portuguese-American project in southern Portugal,” Muse, 18, pp. 35-53.).

Casa 8 was originally built during the Flavian period but was redesigned over time; the phase studied herein dates from the mid- to late 2^nd c. CE (Oberhofer 2018bOberhofer, K. 2018b: “Archäologische und stratigraphische üntersuchungen zu Haus 8,” in F. Teichner (Ed.), Mirobriga, Eine Stadt im fernen Westen des Imperium Romanum, pp. 128-147. Kleine Schriften aus dem Vorgeschichtlichen Seminar Marburg, Heft 62, Philipps - Universitat Marburg.) and is the city’s most recent example known so far.

Casa dos Frescos was largely excavated in 1972 but no stratigraphic records are known, which makes it impossible to assign a precise chronology to its construction. However, the possible dining room identified therein dates from a later phase, and the most direct chronological element is the preserved mural painting covering an opening that was eventually walled up. This painting was dated to the 2^nd half of the 2^nd c. CE, based on stylistic criteria (Rosário et al 2002: 167-168Rosário, M. A.; Duarte, F. L.; and Maciel, M. J. 2002: “Levantamento e leitura iconográfica dos frescos romanos de Miróbriga visíveis in situ no ano 2000,” Trabalhos de Antropologia e Etnologia, XLII (1-2), pp. 165-191.) and should, for this reason, be considered with due caution.

The dimensions and proportions of the dining rooms documented in Mirobriga are identical to one another (Tab. 4). The most coherent set is composed of the triclinia of Casa da Calçada (651 cm in width by 830 cm in length), Casa Periquito (669 cm in width and 844 cm in length), Casa da Hospedaria (with an estimated width of ca. 650 cm by 879 cm in length, according to the planimetric reconstruction of the house) and Casa dos Frescos (measuring 661 cm in width but with a still unknown length). Regarding these last two examples, and though they are not fully preserved, the dimensions observed, both in terms of length and width, are consistent with those observed in the other dining rooms of the city, thus strengthening their interpretation as such.

The proportions of the other two examples, Casa 7 and Casa 8, do not match the previously mentioned ones. The triclinium from Casa 7 measures 730 cm by 660 cm, displaying an almost square morphology and its layout is not immediately perceptible, while the triclinium from Casa 8 features the largest dimensions, with a width of about 760 cm and a length of about 910 cm.

8. DISCUSSION

8.1. Metrical analysis

 ⌅

The measurements recorded in the sample of dining rooms studied herein allow us to recognise not only the existence of various scales for these spaces but also what seems to be some standardisation of the dimensions, namely in terms of widths, which supports the definition of four size clusters. The length measurements show some variations, not enabling the definition of clusters, although some of the length measurements are recurrent. This suggests the existence of some flexibility in the definition of the lengths, as opposed to the widths (Tab. 5).

Cluster	City	Building	Room	Width		Mean		Median		Length		Module	Intended measure		Proportions
				cm	ft (29,63)	cm	ft (29,63)	cm	ft (29,63)	cm	ft (29,63)	ft	Width (ft)	Length (ft)
Cluster 1	Augusta Emerita	Casa del Anfiteatro	Room 29/Room A’	1020	34.42	_	_	_	_	1350	45.56	5	35	45	7x9
Cluster 2 - A	Conimbriga	Casa de Cantaber	Room 20	770	25.98	751.1	25.35	752	25.37	1335	45.05	5	25	45	5x9
		Casa dos Repuxos	Room 33 *	715	24.13					1200	40.49	5	25	40	5x8
		Casa da Cruz Suástica	Room 18	744	25.11					1078	36.38	5	25	35	5x7
		Casa dos Esqueletos	Room 9	780	26.32					940	31.72	5	25	30	5x6
	Augusta Emerita	Casa Morería 6	Salón	770	25.98					1025	34.59	5	25	35	5x7
		Casa Alcazaba 2	Room E	740	24.97					1355	45.73	5	25	45	5x9
	Mirobriga	Casa 7	R 7.1.2	730	24.63					660	22.27	5	25	25	5x5
		Casa 8	R 8.3.5	760	25.64					910	30.71	5	25	30	5x6
Cluster 2 - B	Augusta Emerita	Casa del Mitreo	Room R	715	24.13	715	24.13	715	24.13	1100	37.12	6	24	36	4x6
			Room U	715	24.13					545	18.39	6	24	18	4x3
Cluster 3 - A	Conimbriga	Casa dos Repuxos	Room 29 *	610	20.58	646	21.8	655	22.1	810	27.33	7	21	28	3x4
	Augusta Emerita	Casa Calle Pareja 32	undesignated	645	21.76					no data	no data	7	21	no data	3x?
	Mirobriga	Casa da Calçada	A7	655	22.11					835	28.18	7	21	28	3x4
		Casa dos Frescos	Room 2 *	655	22.11					no data	no data	7	21	no data	3x?
		Casa Periquito	R 5.1.2	665	22.44					845	28.51	7	21	28	3x4
Cluster 3 - A or B	Conimbriga	Ed. Oeste da Zona C	undesignated	620	20.92	_	_	_	_	no data	no data	5 / 7 ft	20 / 21	no data	4x? / 3x?
Cluster 3 - B	Conimbriga	Casa de Cantaber	Room 6	630	21.26	591.5	19.96	586	19.77	910	30.71	5	20	30	4x6
		Ed. Latrinas do Forum	undesignated	580	19.57					465	15.69	5	20	15	4x3
		Insula do Aqueduto	Room 16	575	19.41					875	29.63	5	20	30	4x6
		Casa Medianum Absidado	Room 8	590	19.91					730	24.63	5	20	25	4x5
	Capera	Capera 3	undesignated	582	19.57					744	29.87	5	20	30	4x6
	Augusta Emerita	Casa Morería 6	Triclinium	575	19.41					430	14.51	5	20	15	4x3
		Casa del Anfiteatro	Room 22/Room C	610	20.58					910	30.71	5	20	30	4x6
			Room 23/Room D	590	19.91					910	30.71	5	20	30	4x6
	Mirobriga	Casa da Hospedaria	Room 13	no data	no data					890	30.03	5	no data	30	?x6
Cluster 4	Conimbriga	Casa de Cantaber	Room 65	420	14.17	411.5	13.88	_	_	620	20.92	5 / 7 ft	15 / 14	20 / 21	3x4?/ 2x3?
	Conimbriga	Casa dos Repuxos	Room 34 *	403	13.61					466	15.72	2 / 7 ft ?	14	16 / 21 ?	2x3 ?

Table 5.  Cluster comparison table (own work).

In this sense, the relative consistency of the width values supports the interpretation that this measure would be the most relevant, and should not be compromised, possibly due to the dimensions of the furniture and the required arrangement, the same not applying to the length.

Clusters 2 and 3 are divided into two subclusters, even though their sizes are almost identical. This distinction is due to the apparent use of a different module in the design of the rooms, resulting in slightly different measurements among them, but not different enough for their classification in a new cluster (Fig. 6).

Figure 6.  Dining rooms size dispersion graph. Rooms are labelled according to their location (AE Augusta Emerita; CA Capera; CN Conimbriga; M Mirobriga). Coloured lines represent the different proposed modules (Red: 5 ft; Blue: 6 ft; Yellow: 7 ft) (own work).

8.1.1. Cluster 1

 ⌅

The first cluster consists of only one example, corresponding to a room 35 ft wide and 45 ft long. The greatest common divisor (GCD) is 5, both for width and length, so this may have been the module used to define the room. As there is only one example in the sample, it is not possible to characterize the cluster. However, we would not rule out that there may be more examples of this size.

8.1.2. Cluster 2

 ⌅

This cluster includes ten examples and has been divided into two categories: A and B, according to the module used. Cluster 2-A, with eight examples, has a probably intended width of 25 ft and a variable length, between 30 and 45 ft. The GCD is 5, so this may have been the module used. This cluster has one noteworthy exception, Room R 7.1.2 of Casa 7 from Mirobriga. This room measures 24.6 ft by 22.3 ft, and it is not clear, at first sight, what the intended measurements were, namely, how to distinguish length from the width and what module was used for its sizing.

The explanation for this lack of proportion may lie in the way the spaces within the house were laid out. The house has an internal length of about 50 ft, of which 25 ft have been set aside for the width of the peristyle and, apparently, another 25 ft for the triclinium. The dining room would thus have a square proportion with 25 ft to a side. However, the wall separating the triclinium from the peristyle was built within the modules reserved for the triclinium itself, leaving the correct proportion for the layout of the peristyle, but ultimately taking away space from the dining room, thereby limited to a length of only 22.3 ft (Sousa & Felício 2022: 504-505Sousa, F.; and Felício, C. 2022: “Desenhar e construir em Mirobriga (Santiago do Cacém, Portugal): Contributo para o conhecimento dos modelos arquitectónicos no séc. I d.C. a partir de um estudo metrológico,” in Small Towns, una realidad urbana en la Hispania romana, pp. 495-508. Serie Mytra 10, Instituto de Arqueología de Mérida, Mérida.).

Hence, if we consider 24.6 ft (25 ft intended) as the width, only the length saw its proportions compromised, while the width remained well proportioned, a situation that suggests, once again, that the width would be the most important measurement, possibly because it was meant to accommodate the furniture. This interpretation is further strengthened by the arrangement of the “T+U” markings found in the Edifício das Latrinas do Forum and the Edifício Oeste da Zona C, in Conimbriga, placed on the longer side of the room; and by the dining room of Casa Moreria 6 that, given the location of the door, would not allow the placement of lecti in the narrower side of the room, the width corresponding, therefore, to the longer side.

Cluster 2-B consists of only two examples found in the same house in Augusta Emerita, Casa del Mitreo. Both feature a width of ca. 24 ft and probable intended lengths of 18 ft, in one case, and possibly 36 ft in the other (real 37.12 ft). The GCD for these values is 6 so this may have been the module used to define these rooms.

The proximity of the intended measurements of the 24 × 36 ft proposed for Room R and the 25 × 35 ft found in the other examples from Cluster 2-A accounts for an added difficulty in distinguishing these sub-clusters. However, the values obtained from Room U, in the same house, have a less arguable correspondence to a 6 ft module, giving weight to an intended measure of 36 ft, with a surplus of 1.12 ft, for the length of Room R.

8.1.3. Cluster 3

 ⌅

This cluster includes a total of 15 examples. It was also divided into two subclusters differentiated by the apparent variation in the module used.

Cluster 3-A consists of five examples and is characterised by a width of 21 ft and a length of 28 ft in the complete examples, having a GCD of 7.

Two of these examples are in poor condition or were not fully excavated resulting in only one reference measurement available; we are referring to the Casa de la Calle Pareja 32 (Augusta Emerita) and the Casa dos Frescos (Mirobriga). In these cases, the width dimensions are very close to 21 ft, so this may have been the intended width, which indicates a module of 7 ft.

To properly ascertain if these should fit into Cluster 3-A or 3-B, it would be necessary to have their length preserved or excavated. However, in two of these cases, the measurements obtained are well over 21 ft. If in Casa de la Calle Pareja 32 we may have to account for an excess for the installation of the finishing, in Casa dos Frescos, the plastered surface is preserved and accounts for a width of 22.11 ft, making it less likely that these examples could belong to Cluster 3-B.

Cluster 3-B includes nine examples and is characterized by a width of 20 ft and a length between 15 and 30 ft, its GCD being 5 and therefore this may have been the module used. Casa da Hospedaria (Mirobriga) is not totally preserved and only its length (30 ft) can be assessed. The proportions usually documented for this cluster support an estimate of 20 ft for its width, consistent with the restored dimension proposed for this building by Ada Cortés (2018: 154)Cortés, A. 2018: “The Roman Houses of Flavian Mirobriga and Their Evolution - A first approximation to the archaeological remains”, in F. Teichner (Ed.), Mirobriga, Eine Stadt im fernen Westen des Imperium Romanum, pp. 148-155. Kleine Schriften aus dem Vorgeschichtlichen Seminar Marburg, Heft 62, Philipps - Universitat Marburg..

Room 29 of the Casa dos Repuxos (Conimbriga) also fits into this cluster. As previously mentioned, this room has two internal brick walls built flush against its South and East walls. Being aware of this situation, we have chosen to record two sets of measurements for this room, i.e. the length and width both with and without the internal walls¹⁴The table only includes the measurements of the internal walls..

The measurements obtained without the walls were 598 cm wide by 772 cm long, corresponding to 20.2 × 26 ft. The walls are 16 cm (South) and 19 cm (East) thick, plus 11 cm and 8 cm of plaster and pictorial finishing on the North and West walls respectively, resulting in a rectangular space 582 cm wide by 753 cm long, corresponding to 19.64 × 25.41 ft.

The function of these inner walls is not clear. Their construction, fragile and disorganized in terms of the placement of the constructive elements, does not seem to be related to a reinforcement of the main walls, being more like a regularization filler.

Both the construction technique and the recorded measurements, with and without these walls, seemingly support its interpretation as an attempt to correct the proportions of the room, bringing it closer to the 20 × 25 ft values.

As for the case of Edifício Oeste da Zona C (Conimbriga), due to the block being, to this date, largely unexcavated, only the measurement of its width is available. However, unlike the rooms from Casa de la Calle Pareja 32 and Casa dos Frescos, its width (20.92 ft) could fit in either subcluster A or B. As such, in the absence of a length that could help us verify what module may have been used, we ascribed it simply to Cluster 3, without specifying a subcluster.

8.1.4. Cluster 4

 ⌅

The last defined cluster includes only two examples. Room 65 of Casa de Cantaber has a width of 420 cm and a restored length of ca. 620 cm. These measurements could correspond to a room of ca. 15 × 20 ft, with a module of 5 ft, or a room with 14 × 21 ft, having thus a module of 7 ft.

The other example, Room 29 of Casa dos Repuxos, is 403 cm wide by 466 cm long, corresponding to ca. 14 × 16 ft. In this last case, the interpretation of the module presumably used depends on the room’s intended proportions at the time of its design: on one hand, a module of 2 ft can be admitted since it is the GCD of 14 and 16, in which case the room would feature the originally intended proportions; on the other hand, one might also consider that the room would have been designed using 7 ft modules and a 14 × 21 ft proportion, the length having been shortened by 5 ft due to the irregular configuration of the plot, not unlike the situation observed at Casa 7 of Mirobriga.

If that was the case, this could reinforce the idea that Room 65 was also designed with a 7 ft module and had intended measures of 14 × 21 ft.

However, considering these possibilities and the scarcity of examples of any of the dimensions documented in these two cases, their characterisation as a cluster becomes difficult due to the lack of comparative elements, still not supporting the establishment of a trend concerning their proportions.

8.2. Global overview

 ⌅

From an accounting point of view, cluster 3 is the most numerous, with 15 examples, nine of which feature a five-foot module while six feature a seven-foot module. Arguably, the intended dimensions were 20 ft in width in the first case and 21 ft in the second. This difference would not be sufficient to differentiate the rooms in terms of a specific function in the dining activities spectrum.

Cluster 2 counts ten examples, eight of which were designed according to a five-foot module and two with a six-foot module, resulting in a width of 25 ft in the first case and 24 ft in the second. In this cluster, there is also a minimal difference that should not have had any influence on the intended function of the rooms.

With the exception of clusters 1 and 4, whose samples are not large enough to establish criteria, most of the dining rooms included in this sample tend to measure between 20 and 25 ft in width, which is probably the optimal dimension for this type of room.

Since width variation observed in these clusters and subclusters is not significant (20 and 21 ft for cluster 2; 24 and 25 ft for cluster 3) the differences should not result in functional differences but may be related either to space restrictions related to the laying out of the building on the plot, or to the mathematical or theoretical-methodological principles used to calculate the proportions of the rooms. In this sense, the use of different modules (two, five, six and seven ft) may be related to different theories of architectural thought for the sizing of spaces, such as the existence of architectural traditions or schools that used different modules for the calculation of architectural elements and spaces (Fig. 6).

Figure 7.  Global overview of the floorplans of the rooms under analysis and the respective proportions and modules (own work).

However, the corroboration of any of the hypotheses for the use of one or another module still lacks more comprehensive metric studies of each of these buildings, in order to characterise their design processes.

The remaining examples, corresponding to clusters 1 and 4, stand at both ends of the studied sample. In the case of the only example from Cluster 1, its large dimensions allow us to infer that it may have had a significant representational function and may have been a space ordered with such architectural characteristics, possibly in accordance with the status of its owner. Conversely, the examples from Cluster 4, both due to their small dimensions and their location in secondary peristyles, would appear to be dining rooms of less formal use, or a more familiar nature. Once again, the lack of examples with such dimensions in this sample inhibits a reliable interpretation.

The rules of proportion prescribed by Vitruvius for the sizing of the dining rooms determine that the length should be twice the width (VI, III, 8). However, this rule of proportion was not observed in any of the examples addressed herein. This suggests, once again, the existence of several schools of architecture that presumably used different methodologies and principles for the proportioning of spaces. In this sense, the Vitruvian ‘school’ does not seem to be reflected in the sizing of the dining rooms in Lusitania.

9. FINAL REMARKS

The sizing observed in the dining rooms under analysis here, seems to be the result of a design conceived to serve a specific function, the holding of a banquet, which required furniture of specific configuration and dimensions, boilers for heating foods, as well as space for servants and guests to circulate.

In this sense, the repetition of certain measurement patterns, represented by the four clusters, denotes the existence of architectural maturity in the conception of these spaces and, above all, of a functional purpose involving criteria that were apparently not very flexible. This is further suggested by the inner walls in Room 29 of Casa dos Repuxos, which were, apparently, built to correct the proportions of the room. This also suggests little influence of the ordering party, with the criteria for the sizing of the rooms being left to the architects. The latter would act according to their own school of thought, as suggested by the various modules and proportions documented, which in turn differ from the thought advocated in the only textual source that has reached us - the prescriptions given by Vitruvius were not documented in any of the rooms we’ve analysed.

Another conclusion that can be drawn from this analysis is that the design of these rooms seems to have been based on linear measurements and not on area measurements. This can be deduced from the relative homogeneity of the widths, which allows the definition of clusters, as opposed to the more heterogeneous set of lengths. This results in rooms from the same cluster having different areas.

This results in rooms with different areas, possibly a less important criterion, as long as the necessary width for the placement of lecti was ensured, as well as some circulation space in the anteroom, where more variability was admissible depending on the available space. This observation seems to go along with Mark Wilson Jones’ observations that Roman’s attitude to design doesn’t follow rigid rules or recipes but instead relies on principles that can be adapted to certain contexts (Jones 2000: 9Jones, M. W. 2000: Principles of roman architecture. Yale University Press, New Haven and London.).

In this regard, room orientation also doesn’t seem to have been a rigid prescription, with no particular pattern being detected. The location of the rooms themselves, either in inner areas or in peristyles, seems to be a more critical factor for their seasonal use. In some cases, the placement of windows may also have played a significant role in allowing a greater solar exposure.

As for the different dining room scales identified, sometimes coexisting within the same house, this may reflect different party numbers and/or social contexts of use, which, in this case, would reflect the commissioner’s wishes, although the sizing criteria, which still needs to be better understood, were probably left to the architects.