Arqueología de la construcción V : man-made materials, engineering and infrastructure: Man-made materials, engineering and infrastructure [1 ed.] 8400101421, 9788400101428

Citation preview

ANEJOS

AEspA

LXXVII

ISBN 978-84-00-10142-8

Man-made materials, engineering and infrastructure

(eds.)

anejos de

aespA LXXVII

ARQUEOLOGíA DE LA CONSTRUCCIÓN V 5th International Workshop on the Archaeology of Roman Construction Man-made materials, engineering and infrastructure

th

ARQUEOLOGíA DE LA CONSTRUCCIÓN V

5 International Workshop on the Archaeology of Roman Construction

2016

Janet DeLaine Stefano Camporeale Antonio Pizzo

ARCHIVO ESPAÑOL DE

9 788400 101428

Cubierta _ARQUE_V_vol_77.indd 1

CSIC

ARQVEOLOGÍA

21/12/16 11:25

ANEJOS DE ARCHIVO ESPAÑOL DE ARQUEOLOGÍA LXXVII

ARQUEOLOGÍA DE LA CONSTRUCCIÓN V

Anejos de AEspA

Director: Ángel Morillo Cerdán, Universidad Complutense, Madrid, España. ­ spaña. Secretario: Carlos Jesús Morán Sánchez, Instituto de Arqueología, CSIC-Junta de Extremadura, Mérida, E Comité Editorial: Pedro Mateos Cruz, Instituto de Arqueología, CSIC-Junta de Extremadura, Mérida, España; Adolfo Domínguez Monedero, Universidad Autónoma, Madrid, España; Inés Sastre Prats, Instituto de Historia, CCHS, CSIC, Madrid, España; Miguel Cisneros Cunchillos, Universidad de Cantabria; José Miguel Noguera Celdrán, Universidad de Murcia, España; Victorino Mayoral Herrera, Instituto de Arqueología, CSIC-Junta de Extremadura, Mérida, España; Susana González Reyero, Instituto de Historia, CCHS, CSIC, Madrid, España; M.ª Ángeles Utrero Agudo, Instituto de Historia, CCHS, CSIC, Madrid, España. Consejo Asesor: Francisco Pina Polo, Universidad de Zaragoza, España; Luis Caballero Zoreda, Instituto de Historia, CCHS, CSIC, España; María Paz García-Bellido, Instituto de Historia, CCHS, CSIC, Madrid, España; Juan Manuel Abascal Palazón, Universidad de Alicante, España; Filippo Coarelli, Universitá degli Studi di Perugia, Italia; Trinidad Tortosa Rocamora, Instituto de Arqueología, CSIC-Junta de Extremadura, Mérida, España; María Ruiz del Árbol Moro, Instituto de Historia, CCHS, CSIC, Madrid, España; Pilar León-Castro Alonso, Universidad de Sevilla, España; Almudena Orejas Saco del Valle, Instituto de Historia, CCHS, CSIC, Madrid, España; Carmen García Merino, Universidad de Valladolid, España; Javier Arce, Université Lille, Francia; Bárbara Böck, Instituto de Lenguas y Culturas del Mediterráneo y Oriente Próximo, CCHS, CSIC, Madrid, España; Domingo Plácido, Universidad Complutense de Madrid, España; Pietro Brogiolo, Università di Padova, Italia; Teresa Chapa Brunet, Universidad Complutense de Madrid, España; Monique Clavel-Lévêque, Université Franche-Comté, Besançon, Francia.

Janet DeLaine Stefano Camporeale Antonio Pizzo (eds.)

ARQUEOLOGÍA DE LA CONSTRUCCIÓN V Man-made materials, engineering and infrastructure Proceedings of the 5th International Workshop on the Archaeology of Roman Construction Oxford, April 11-12, 2015

CONSEJO SUPERIOR DE INVESTIGACIONES CIENTÍFICAS MADRID, 2016

Reservados todos los derechos por la legislación en materia de Propiedad Intelectual. Ni la totalidad ni parte de este libro, incluido el diseño de la cubierta, puede reproducirse, almacenarse o transmitirse en manera alguna por medio ya sea electrónico, químico, óptico, informático, de grabación o de fotocopia, sin permiso previo por escrito de la editorial. Las noticias, los asertos y las opiniones contenidos en esta obra son de la exclusiva responsabilidad del autor o autores. La editorial, por su parte, solo se hace responsable del interés científico de sus publicaciones.

Este volumen se ha beneficiado de los fondos del proyecto de investigación del VI Plan de Investigación «Análisis de soluciones técnico-constructivas, modelos arquitectónicos y urbanísticos de la arquitectura romana de Lusitania: orígenes y transformación de una cultura arquitectónica» (ref. HAR2012-36963-C05-05). Investigador principal: Dr. Antonio Pizzo.

Imagen de cubierta: Giovanni Biagio Amico, L’architetto prattico, vol. I, Palermo 1726 Imagen de contracubierta: detalle de la imagen anterior

Catálogo general de publicaciones oficiales: http://publicacionesoficiales.boe.es Editorial CSIC: http://editorial.csic.es (correo: [email protected])

© CSIC ©  Janet DeLaine, Stefano Camporeale, Antonio Pizzo (eds.) y de cada texto, su autor ©  De las ilustraciones, los autores y las instituciones mencionados a pie de figura NIPO: 723-16-272-8 e-NIPO: 723-16-273-3 ISBN: 978-84-00-10142-8 e-ISBN: 978-84-00-10143-5 Depósito Legal: M-39482-2016 Maquetación, impresión y encuadernación: Gráficas Blanco, S. L. Impreso en España. Printed in Spain En esta edición se ha utilizado papel ecológico sometido a un proceso de blanqueado ECF, cuya fibra procede de bosques gestionados de forma sostenible.

SUMARIO

Introduction......................................................................................................9 J. DeLaine, S. Camporeale, A. Pizzo I.  CONCRETE TECHNOLOGY AND USE Technological confidence in Late republican and imperial era Roman architectural and maritime concrete construction...........................................................15 M. D. Jackson L’uso delle polveri pozzolaniche nei grandi cantieri della Gallia Cisalpina durante l’età romana repubblicana: i casi di Aquileia e Ravenna.................................29 J. Bonetto, G. Artioli, M. Secco, A. Addis Lime mortar production in Ostia: material analysis of mortar from the Hadrianic period........................................................................................................45 J. Wehby Murgatroyd II.  STRUCTURAL AND CONSTRUCTIONAL USES OF METAL The bronze truss of the portico of the Pantheon in Rome...................................59 D. Heinzelmann, M. Heinzelmann “Armored” columns in the Roman imperial period..............................................75 C. M. Amici III.  PRODUCTION, SUPPLY AND USE OF BRICK AND TILE The social life of tile in the Roman world...........................................................87 P. Mills Ceramic building material: production, supply and use in Roman London..........99 I. M. Betts Tegulae mammatae or lateres? On the presence and use of tegulae mammatae in the delta of the Rhine....................................................................................111 A. G. Gazenbeek The end of the tegulae mammatae? A review on their name, function(ality) and presence in the Roman North ............................................................................121 T. Clerbaut IV.  PRODUCTION, SUPPLY AND USE OF MUD-BRICK AND PISÉ DE TERRE Mud brick and pisé de terre between Punic and Roman.....................................131 B. Russell, E. Fentress

L’industrie de la brique crue dans la colonia Lugdunum. Typologie, approvisionnement et organisation de la production.......................................................145 B. Clément V.  LIFTING MACHINES IN THEORY AND PRACTICE Taking a bearing on Hero’s anti-crane and its un-windlass: the relationship ­between Hero of Alexandria’s mobile automaton and Greco-Roman construction machinery, artillery and water-lifting machines............................................167 D. Keenan-Jones, I. Ruffell, E. Mcgookin Lifting blocks, 1st-5th century AD: the inclined plane...........................................185 G. Martines, M. Bruno, C. Conti The restoration of the columns of the templum Castoris during Verres’ praetorship: the machina and organisation of the building site......................................201 P. Ducret VI.  PRACTICAL SOLUTIONS TO ENGINEERING PROBLEMS Costruire in terreni paludosi: sistemi di fondazione e bonifica in uso in età romana in Italia settentrionale fra tradizione e innovazione........................................209 C. Previato Facing structural problems in ancient times. A structural assessment during the construction of the Mausoleum of Hadrian.......................................................231 P. Vitti VII.  INFRASTRUCTURE AND ORGANISATION OF CONSTRUCTION Les techniques et les étapes de la construction des salles de soutènement des thermes de Longeas (Chassenon, France)..........................................................251 A. Coutelas, D. Hourcade Tracce di cantiere dall’area del Capitolium di Brescia: evidenze archeologiche e materiali dai recenti scavi..................................................................................275 A. Dell’Acqua The ‘Lower Agora’ of Pergamon. The organisation of a major building site in Roman Asia Minor............................................................................................299 B. Emme, A. Öztürk (†) VIII.  THE ORGANISATION OF CONSTRUCTION IN REPUBLICAN AND PRE-REPUBLICAN ITALY Quantifying monumentality in a time of crisis. Building materials, labour force and building costs in Late Republican central Italy............................................317 D. Maschek Networks and workshops. Construction of temples at the dawn of the Roman ­Republic............................................................................................................331 P. S. Lulof Conclusions....................................................................................................... 343 L. C. Lancaster

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

INTRODUCTION JANET DELAINE, STEFANO CAMPOREALE, ANTONIO PIZZO

The archaeology of construction is a relatively new discipline within the broader field of Roman architecture. It was given a major impetus by the efforts of an active international group of researchers to create a series of workshops on the subject. Thanks to Antonio Pizzo of the Archaeological Institute of Mérida, Stefano Camporeale of the University of Siena, and Hélène Dessales of the Ecole normale supérieure in Paris, the initial series of workshops began in 2007 at Mérida, quickly followed in 2008 at Siena, and in 2010 at Paris. Given their enormous success, and with the invaluable contribution of Jacopo Bonetto of the University of Padua, it was decided to extend the series with a further workshop in 2012.1 The great success of these previous four workshops demonstrated the very strong interest which exists among scholars of Roman history and arch­ aeology in the constructional aspects of ­Roman architecture. This discipline can be briefly defined as the study of the processes involved in Roman construction projects, covering the production and supply of materials, the techniques used in construction, and the problems of logistics, manpower, and the economics. The archaeology of Roman construction is seen as complementary to the more traditional areas of the study of Roman architecture, such as building typologies and use, design and decoration, or the func-

1   For the archaeology of construction see DeLaine 1997; Cantieri antichi 2002; and the chapter by Hélène Dessales in Bukowiecki et al. 2008: 19-24. For the proceedings of the workshops see Camporeale et al. 2008; Camporeale et al. 2010; Camporeale et al. 2012; Bonetto et al. 2014; syntheses of the results of the first workshops are in Pizzo 2009 and Camporeale 2010.

tion of the finished buildings in society, but also uniquely has the potential to contribute much to ongoing debates about the size and shape of ­Roman urban economies. The previous workshops looked at the question of ‘cantieri’ – building sites and processes – first in Italy and west (Mérida) and then Italy and the eastern Roman provinces (Siena), while the Paris meeting focused on the economic aspects, and the Padua workshop concentrated on the quarrying and supply of building stone. The broad focus of this present workshop is on the development of specific technologies for solving constructional or structural problems, including the production and supply of man-made materials, the contribution of engineering and machines to the functioning of building sites, and the nature and division of labour these require. It has also tried – with some success – to bring into the discussion the north-western provinces, and lessfrequently discussed materials as diverse as metals, mud-brick and pisé de terre. Alongside the naturally occurring materials of stone and timber, Roman builders used a wide variety of materials that had to go through one or more sequences of processing, usually at a separate production site or more rarely, for example in the case of mortar, on site. The supply of bulk materials such as brick and mortar are key in the logistics of site construction, while speciality products such as metal elements usua­lly represent special and sometimes unique so­lutions to specific structural and constructional problems. Man-made materials also often requi­red high levels of skilled manpower and spe­cialised equipment or plant in their processing, thus contributing substantially to the cost of construction.

10

Janet DeLaine, Stefano Camporeale, Antonio Pizzo

The volume is divided into eight sections, corresponding to the four pairs of sessions which constituted the workshop. The first session, on concrete technology and use, began with the most common and most typically Roman of all manmade materials – pozzolanic mortar. Here advances in scientific analyses are proving fundamental in advancing our understanding not only of what materials were used and how, but also where, geographically, and in what contexts they were used, and the chronology of use. In contrast, metal (session II) is much less well-known as a building material in the Roman world; the papers on the structural and constructional uses of metal give new insights into both bronze and iron technology and its applications. The third and fourth sessions were devoted to the production, supply and use of building materials made of fired clay and of earth. Building materials of fired clay (which the British call CBM or ‘ceramic building materials’) and including brick, tile, wall tubes, and vaulting tubes, are, like mortar, one of the hallmarks of Roman construction, and one of the most ubiquitous and certainly the most widespread materials in the archaeological record. Yet they have tended not to be treated as a main focus of study, although the ­recent conference in Rome has gone a considerable way to redressing the situation.2 Even more neglected in standard accounts of Roman construction, however, are materials made of unfired earth, mud brick and pisé de terre, often very difficult to detect in the archaeological record; furthermore they are certainly rarely brought into the same discourse as fired brick and pozzolanic mortar. Yet the papers presented here argue that they were at least as common, and not only in earlier periods or rural vernacular constructions. In the fifth session attention was focused on engineering and machines. The Romans have long been admired for their engineering skills, but this is an area which has so far attracted relatively little interest within the archaeology of construction. Large-scale construction is not possible without the use of complex lifting machines, or devices such as the truss, used in supporting large formwork, yet all of these require s­ pecialised equipment and skills, and have an impact on the ­scheduling of work and the division of labour. ­Machines, especially the lifting machines essential for large-scale construction, are particularly  Bukowieki et al. 2015.

2

Anejos de AEspA LXXVII

­ nder-researched in relation to construction prou cesses, something that the papers in the section on lifting machines in theory and practice go part way to addressing. The papers in the sixth session looked specifically at practical solutions to engineering problems, and in particular the problems of found­ations – another often invisible element in construction – in relation to actual site conditions, helping to throw light on construction processes, the choices made by builders and architects, and their responses to building on unstable ground. The final pair of sessions session returned to themes more familiar from earlier workshops, espec­ially the infrastructure and organisation of construction. These provided some rare surviving evidence of building work in progress, and discuss strategies used for construction where resources were limited. The final pair of papers continue the theme but with a chronological emphasis, examining seasonal demands for resources and manpower in the Republican period, and identifying networks of production in pre-Republican Italy. These papers all bring us as close as possible to the individual human actors – the builders – who are, ultimately, the real focus of our common inter­est, the archaeology of Roman construction. The workshop in Oxford attracted a great deal of interest, with almost 60 proposals for papers, and over 80 delegates and speakers, from 10 different countries. One of the guiding principles for the workshop was to maintain the wide range of approaches and the multidisciplinary nature which characterised the earlier four. As befits a workshop, the emphasis was on discussion and the exchange of ideas, elements which made the previous ones so fruitful and enjoyable. Given the very large number of high quality proposals for papers, it was therefore decided for the first time to include posters as an integral part of the workshop, divided into their appropriate sessions and included in the discussion at the end of each session. Several of these are being published in the journal Arqueología de la arquitectura, forming an important complement to the papers presented in this volume. Finally the editors would like to thank the Classics Faculty Board, the Craven Committee, the Meyerstein Research Fund of the School of Archaeology, and the Lorne Thyssen Research Fund for Ancient World Topics at Wolfson College, all of the University of Oxford, and the RSK Group for their generous support of this workshop. Particular thanks go to the Consejo Supe-

Anejos de AEspA LXXVII

rior de Investigaciones Científicas, who, since the first workshop, has continued to finance the publication of the series Arqueología de la construc­ ción in the Anejos de Archivo Espanol de Arqueolo­ gía. Finally, we would like to thank the Workshop secretary, Alejandra Albuerne, for her outstanding contribution to the whole organisation of the workshop, Kate del Nevo of the Faculty of Classics for her logistical support, and the graduate volunteers from the School of Archaeology. REFERENCES Bonetto, J., Camporeale, S. and Pizzo, A. (a cura di) 2014: Arqueología de la Construcción IV. Las cante­ ras en el mundo antiguo: sistemas de explotación y procesos productivos (Padova, 22-24 de noviembre de 2007), Anejos de Archivo Español de Arqueología 69. CSIC, Mérida. Bukowiecki, E., Dessales, H. and Dubouloz, J. 2008: Ostie, l’eau dans la ville. Châteaux d’eau et réseau d’adduction, Collection de l’Ecole française de Rome 402. Ecole française de Rome, Rome. Bukowiecki, E., Volpe, R., Wulf-Rheidt, U. 2015: “Il laterizio nei cantieri imperiali. Roma ed il Mediterraneo”, Archeologia dell’Architettura, 20, in press. Camporeale, S. 2010: “Archeologia dei cantieri di età  romana”, Archeologia dell’architettura, 15, pp. 171-179.

Introduction

11

Camporeale, S., Dessales, H. and Pizzo, A. (a cura di) 2008: Arqueología de la Construcción I. Los pro­ cesos constructivos en Italia y en las provincias roma­ nas: Italia y provincias occidentales (Mérida, 26-27 de octubre de 2007), Anejos de Archivo Español de Arqueología 50. CSIC, Mérida. Camporeale, S., Dessales, H. and Pizzo, A. (a cura di) 2010: Arqueología de la Construcción II. Los pro­ cesos constructivos en Italia y en las provincias roma­ nas: Italia y provincias orientales (Siena, 13-15 de noviembre de 2008), Anejos de Archivo Español de Arqueología 57. CSIC, Madrid-Mérida. Camporeale, S., Dessales, H. and Pizzo, A. (a cura di) 2012: Arqueología de la Construcción III. Los procesos constructivos en el mundo romano: la econo­ mía de las obras (París, 10-11 de diciembre de 2009), Anejos de Archivo Español de Arqueología 69. CSIC, Madrid-Mérida. Cantieri antichi 2002: “Cantieri antichi. Giornata di studio tenuta il 25 ottobre 2001”, Römische Mit­ teilungen, 109, pp. 340-429. DeLaine, J. 1997: The Baths of Caracalla. A Study in the Design, Construction, and Economics of Largescale Building Projects in Imperial Rome, Journal of Roman Archaeology Suppl. 25. Journal of Roman Archaeology, Portsmouth, R. I. Pizzo, A. 2009: “La Arqueología de la Construcción. Un laboratorio para el análisis de la arquitectura de época romana”, Arqueología de la arquitectura, 6, pp. 31-45.

I CONCRETE TECHNOLOGY AND USE

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

TECHNOLOGICAL CONFIDENCE IN LATE REPUBLICAN AND IMPERIAL ERA ROMAN ARCHITECTURAL AND MARITIME CONCRETE CONSTRUCTION MARIE D. JACKSON Department of Geology and Geophysics, University of Utah, Salt Lake City, Utah, USA

ABSTRACT: Imperial Roman builders’ confidence in more-or-less standardized lime-volcanic ash mortars is recorded by daring structural designs in architectural concrete in Rome and the rock-like durability of maritime concrete in Mediterranean harbor structures. Vitruvius emphasized know­ ledge and expertise, or scientia, in late Republican construction technologies (de Architectura 1.1.1-2). These principles are recorded by the first century BCE archaeological record. The Tomb of Caecilia Metella (30 BCE) in Rome, for example, has extraordinarily resilient wall concretes with mortars reinforced with strätlingite crystals. The harbor construction at Caesarea Palaestinae (23 to 15 BCE), to which builders shipped enormous quantities of pumiceous ash from the Gulf of Naples, crystallized zeolite and Al-tobermorite mineral cements. The physical processes that underlie late Republican builders’ advances in concrete technologies and which form the foundation of Imperial builders’ trust and reliance on pyroclastic rock concretes are now being explored with new volcanological and mineralogical research. KEYWORDS: Roman concrete, Microscopy, Volcanic pozzolans, Technical advances, Vitruvius. RESUMEN: La seguridad de los constructores imperiales romanos en morteros de cal con cenizas volcánicas, más o menos estandarizados, se ha documentado a partir del atrevimiento en diseños estructurales de hormigón en Roma y la durabilidad parecida a una roca del hormigón marítimo en estructuras portuarias del Mediterráneo. Vitruvio hizo hincapié en el conocimiento y la experiencia, o scientia, en las tecnologías de construcción de época tardo-republicana (De Architectura 1.1.1-2). Estos principios se han documentado en el registro arqueológico del primer siglo antes de Cristo. La tumba de Caecilia Metella (30 a.C.), en Roma, por ejemplo, presenta hormigones de estructuras extraordinariamente resistentes con morteros reforzados con cristales de estratlingita. Otro ejemplo es la construcción del puerto en Caesarea Palaestinae (del 23 al 15 a.C.), a la que los constructores enviaron enormes cantidades de cenizas de pómez desde el Golfo de Nápoles, zeolita cristalizado y cementos de mineral de Al-tobermorita. Los procesos físicos que representan los avances de los constructores republicanos en las tecnologías específicas y que forman la base de la confianza y dependencia de los constructores imperiales en los morteros de rocas piroclásticas están siendo analizados, en la actualidad, con nuevos procesos de investigaciones vulcanológicos y mineraló­gicos. PALABRAS CLAVE: hormigón romano, microscopía, puzolanas volcánicas, avances técnicos, Vitruvio.

INTRODUCTION In the early second century CE, builders integrated an expansive and intricate complex of ­concrete structural elements to create the Markets

of Trajan (~110 CE) at the foot of the Quirinale hill in Rome. Nearly two thousand years later, this remains a highly functional monument that embodies the qualities of fine workmanship (subtilitas), magnificence (magnificentia), and design

16

Marie D. Jackson

(dispositio) that Vitruvius described in the late first century BCE as the tests of good architectural practice (de Architectura 6.8.9). The 32 m long vaulted span of the Great Hall (fig. 1) that anchors the market has a complex concrete mix design, which utilizes varying proportions of scoriaceous and pumiceous volcanic ash, or harenae fossiciae, and conglomeratic brick and vitric tuff coarse aggregate, or caementa. This sophisticated structure records the technical knowledge of engineers and builders, as they expanded their expertise with vaulted concrete construction (Lancaster 1998; Vitti 2007; Jackson et al. 2009; Jackson et al. 2014; Brune and Perucchio 2012). We do not have a written record of the Trajanic builders’ perceptions of their concrete work, either in the monuments of Rome or in the massive harbor con­struction at Portus, 15 km east of Rome. Therefore, we cannot make a direct assessment of the confidence that they placed in these mix designs, which were founded on different types of volcanic rock and brick aggregate (fig. 2). Even so, fine-scale examinations of both the architectural and maritime concretes reveal the close attention that builders paid to the selection of volcanic ash pozzolans for these exceptionally durable mortars, and the resilient nature of the cementitious binding components that developed over time in these materials (Jackson et al. 2013; ­Jackson et al. 2014).1 It was with these highly refined mortar mixes that the architects and engineers who built Trajan’s markets assured the very long term structural integrity and mechanical stability of a serviceable, working monument of great beauty. Their workmanship and expertise seem to reflect an audacious intent to enhance the human perception of interior and exterior architectural spaces at a temporal scale that has persisted far beyond the dur­ation of the Roman Empire. At Trajan’s port, builders created a concrete that continues to grow crystalline cementitious hydrates as it ages, similar to a volcaniclastic rock in the earth’s crust (Jackson 2014). This pyroclastic rock concrete can provide a template for modern engineers as they search for innovative solutions for improving the durability of large-scale maritime structures that must remain stable in an aggressive seawater environment for very long periods of time. 1   A pozzolan is a material that reacts with lime (CaO) in the presence of moisture to form stable, binding, cementitious hydrates (Massazza 1988).

Anejos de AEspA LXXVII

Fig. 1. Photograph of the Great Hall in the Markets of Trajan, Rome (Archives, Museo Fori Imperiali).

It now appears that Republican era builders first mastered mortar technologies for terrestrial concretes in Rome during the mid- and late-second century BCE (Mogetta 2015). These are preserved mainly in the rectangular podia underlying public monuments that were faced with opus quadratum blocks of volcanic tuff ­dimension stone, as at the Temple of Magna Mater (150 to 100 BCE) (Mogetta 2015: 7-19). Builders experimented with pyroclastic and reworked epiclastic ashes as pozzolans in mortar formulations and with the preparation of lime for many decades; the decimetersized coarse aggregate (caementa) of the concrete is usually volcanic tuff. Many mortars formulated prior to the Augustan period (27 BCE to 14 CE) contain a component of weathered, dusky-brown scoriaceous ash (harena fossicia) from an ancient soil, or paleosol, that formed in the uppermost horizon of the mid-Pleistocene Pozzolane Rosse pyroclastic flow (Jackson et al. 2010). These deposits are now known to exist within the center of Rome, and were used in many Republican era mortars (Marra et al. 2015). There are also particles of volcanic tuff and palagonitic ash (iron-rich glass fragments) in many early mortars, as well as reworked and weathered ash from a mid-Pleistocene, epiclastic, palagonitic ash horizon that overlies the paleosol at the top of the Pozzolane Rosse pyroclastic flow (fig. 3). Parallel experimentation in the development of the mortars of maritime concretes also occurred during the later Repu­blican era. Porous mortars of Egnazia ­harbor concretes (first to second century BCE) drilled by the  ROMACONS program contain pumiceous pozzolan (pulvis) with Campi Flegrei and Vesuvian trace element ratios, yet the coarse aggregate (­caementa) is local calcareous, fossiliferous sandstone (Jackson 2014: 145-

Anejos de AEspA LXXVII

Technological Confidence in Late Republican...

17

Fig. 2. Photographs of drill cores of Trajanic era conglomeratic concretes with brick and volcanic tuff coarse aggregate (caementa), a) core from the eastern wall of the Great Hall, 20 cm diameter, Sovrintendenza Capitolina Beni Culturali di Roma Capitale, b) core PTR.2001.01, from the western mole of Portus Traianus, 9 cm diameter, ROMACONS drilling program.

160; Oleson et al. 2014: 265-267). The early piers constructed at Portus Cosanus (~50 BCE), for example, are the only concretes drilled by ROMACONS that incorporate a substantial proportion of beach sand in their volcanic ash-lime mortars (Oleson et al. 2014: 248-252). These Republican era mortar mixes exhibit good resilience to chemical decay, however, and have persisted in the archaeological record for more than 2000 years. Based on many years of empirical observations of architectural concretes in Rome, E. B. Van Deman hypothesized in 1912 that by late first century BCE, when Vitruvius wrote de Architectura, builders in Rome had begun to develop improved technologies for the production of a concrete, in which one of the distinctive characteristics of the construction, the dusky red color of the mortar, arises from the use in it of a special variety of pozzolana [the arena rubra of Vitruvius (2.4.1)], the introduction of which by Augustus, or by his predecessor Julius Caesar, marks an epoch in the history of concrete construction (Van Deman 1912a).

Conglomeratic concrete with this very robust mortar was employed, for example, in the foundation and the opus reticulatum and brick-faced walls of the Theater of Marcellus (46 to 44 BCE) and the substructure and the brick-faced walls of the Tomb of Caecilia Metella (30 BCE) (Gerding 2002; Jackson et al. 2007; Jackson et al. 2011). These mortars have reddened and dark gray scoriaceous volcanic ash pozzolan, the rubra and nigra harenae fossiciae described in de Architectura

(2.4.1). This is scoriaceous ash excavated from the middle and lower horizons of the mid-Pleistocene Pozzolane Rosse pyroclastic flow that erupted from Alban Hills volcano about 457,000 years ago (Marra et al. 2008). The scoriae have relict glassy groundmass, fresh leucite crystals and occa­ sional replacement by analcime, and thin clay mineral (halloysite) and zeolite (mainly phillipsite and chabazite) surface coatings (Jackson et al. 2010). These clay and zeolite mineral components are known to have excellent pozzolanic properties (Massazza 1998). The selection and calcination of limestone, and lime preparation and hydration methods had likely improved, as well (de Architectura 2.5.1-3). By the early Augustan age, builders and their private and governmental sponsors had apparently begun to develop a prudent trust in the cement­itious resilience and mechanical durability that resulted from these prescribed architectural and maritime mortar mix designs and installation procedures. Recent mineralogical research demonstrates that early Imperial age builders were thoroughly justified in placing their confidence in these mortar formulations. The Tomb of Caecilia Metella (fig. 4a) and Theater of Marcellus mortars have durable calcium-aluminum-silicate-­ hydrate (C-A-S-H) binder and platy strätlingite (Ca2Al(AlSi)O2(OH)10·2.25H2O) crystals that reinforce the interfacial zones of scoriae with the cementitious matrix, similar to the highest quality Trajanic era mortars (Jackson et al. 2011; Jackson et al. 2014). Builders’ firm trust in the long term performance of the well-bonded Pozzolane Rosse

18

Marie D. Jackson

Anejos de AEspA LXXVII

Fig. 3. Petrographic micrographs showing aggregate components of later Republican era mortars in Rome, plane polarized light, a) the Porticus Aemilia, or Testaccio shipshed (174 BCE or 110-100 BCE (?)) (Cozza and Tucci 2006; Mogetta 2015), sub-rounded clasts of vesicular palagonitic glass from the Pozzolanelle pyroclastic flow, b) the Temple of Concord (121 BCE), analcime, diopside, and sanidine sand grains, a rounded lava pebble, and fine palagonitic glass particles from the epiclastic San Paolo Formation, c) the Temple of Castor and Pollux (117 BCE), a clast of Tufo del Palatino, and palagonitic glass particles likely from the upper horizon of the Pozzolane Rosse pyroclastic flow, d) the Tabularium (~70 BCE), a fragment of Tufo Giallo di Prima Porta (TGdPP), clasts of weathered Pozzolane Rosse (PR) with analcime crystal fragments and the upper paleosol horizon of reworked palagonitic glass (pa), e) the Caesaeran Rostra (~50 BCE), and f) the Temple of Saturn (~48 BCE) with clasts of weathered Pozzolane Rosse (PR), some with thick, yellow coatings of halloysite clay mineral, and particles of the upper paleosol horizon of reworked palagonitic glass (pa), g) another Temple of Saturn (~48 CE) mortar with Pozzolane Rosse scoriae, analcime and sanidine grains, and clasts of Tufo Rosso a Scorie Nere and Tufo Giallo di Prima Porta, perhaps excavated near the building site, and h) the Tomb of Caecilia Metella (30 BCE) with Pozzolane Rosse from the reddened intermediate alteration facies, with intact leucite (lc), glassy ground mass, and a fine ash accretion on a larger scoria perimeter. This is the typical appearance of well-consolidated mortars Imperial age mortars with and well-calcined and hydrated lime and dusky-red and dark-gray Pozzolane Rosse pozzolan from the middle and lower alteration facies of the pyroclastic flow deposit (Jackson et al. 2010).

Anejos de AEspA LXXVII

Technological Confidence in Late Republican...

19

Fig. 4. Crystalline cementitious hydrates in very cohesive, well-consolidated Republican era mortars, back-scattered images from scanning electron microscopy, a) the Tomb of Caecilia Metella showing abundant strätlingite plates in the cementitious matrix, and b) the Baianus Sinus breakwater in the Bay of Pozzuoli, showing hydrocalumite crystals associated with acicular Al-tobermorite crystals.

harena fossicia mortar that used ash excavated from the middle and lower horizons of the pyroclastic flow – not the uppermost weathered paleosol horizon – is implied by the sophisticated structural designs of Imperial age concrete in the monuments in Rome, which required long term mechanical resilience to respond to static and ­dynamic loads without collapse (Brune and Perucchio 2011; Brune and Perucchio 2012). These include the Colosseum (70 to 79 CE) (Lancaster 2005b), the Forum of Trajan (107 to 112 CE) (Bianchi et al. 2011), the Markets of Trajan (Jackson et al. 2014) and the Baths of Caracalla (211 to 215 CE) (DeLaine 1997). The addition of pumiceous volcanic ash to the mortars of the vaulted concrete coverings of these monuments, however, shows far more variation (Lancaster et al. 2010; Marra et al. 2013). Examinations of the latest Republican era harbor structures along the central Italian coast and certain passages of de Architectura (2.6.1-6, 5.12.2-6) indicate that builders also deliberately focused their technological expertise on improving the volcanic ash-hydrated lime mortars that bind seawater based conglomeratic concretes. The mortars of  maritime concrete structures in the Bay of Pozzuoli, including the Baianus Sinus breakwater (~35  BCE), for example, contain pumi­ceous volcanic ash pozzolan (pulvis) from Campi Flegrei volcano, based on the trace element ratios and mineral assemblages of relict pumice clasts (Brandon et al. 2008; Jackson et al. 2013b). Pliny the Elder would describe the resilience of these concretes after about 100 years of service life in Naturalis Historia, written before the Vesuvius eruption in 79 CE. Pulvis volcanic

ash is also the predominant pozzolan of maritime concrete structures along the central Italian coast: as at Portus Cosanus and Santa Liberata (~50 BCE) in Roman Etruria, Portus Claudius (~50 CE) and Portus Traianus (~115 CE) at Portus, Portus Neronis (~65 CE) at Anzio, (Jackson 2014: 148-159) as well as at Anxur harbor in Terracina (second century  CE) (D’Ambrosio et al. 2015). The tremendous volume of pumiceous ash pozzolan that was shipped from the Gulf of Naples­to create the immense harbor installations in the eastern Mediterranean region (Hohlfelder and Oleson 2014: 232) suggests a bold and assured confidence in the ­seawater mortar mix, and also in formwork and installation procedures (Brandon 2014: 212-222). Remarkably, all of the maritime mortars drilled by ROMACONS crystallize Al-tobermorite ([Ca4(Si5.5Al0.5O17H2)] Ca0.2Na0.14H2O), a rare, layered, calcium-silicatehydrate mineral, in relict lime clasts and the cementitious matrix (Vola et al. 2011; Jackson et al. 2013b; Jackson 2014: 286-291) (fig. 4b). Strätlingite and Al-tobermorite play an important role in the durability of experimental pozzolanic concretes but are difficult to produce with conventional Portland cement mixes. The material design and installation of the latest Republican and early Imperial architectural and maritime structures were produced by engin­ eers and builders who had been able to observe the performance of previous concrete constructions over several decades of service life through diverse deleterious impacts: fires, floods, earthquakes, substructure instabilities, and chemical attack in the intertidal marine environment. The virtues required of the builders of these ­enormous

20

Marie D. Jackson

accomplishments in construction ­engineering were described by a second century CE Roman designer of a aqueduct tunnel in Algeria as patientia, virtus and spes, “patience, valor, and hope” (Oleson 2008). The purpose of this article is to give some examples of how archaeometric investigations can potentially validate aspects of builders’ confidence, or lack of it, which arose from the implementation of these virtues towards improving varied concrete technologies. These seem to be expressed in certain passages of de Architectura and Naturalis Historia and the many advancements in pozzolanic mortar production that are preserved by the archaeological record. REPUBLICAN AGE TECHNICAL COMPETENCIES WITH CONCRETE MASONRY Roman builders’ geotechnical expertise using opus quadratum masonry improved steadily through the Republican era. This expertise, and the concretes that builders developed during the second century BCE (Mogetta 2015) culminated during mid-first century BCE in innovations that produced the integrated dimension stone and conglomeratic concrete masonry that forms the basis of Imperial age monumental architecture (Jackson and Marra 2006; Jackson et al. 2011). Certain monuments in the westernmost Roman Forum (fig. 3) provide important landmarks for builders’ evolving selections of harenae fossiciae for the mortars of these concretes. The Temple of Concord and the Temple of Castor and Pollux, for example, were constructed in 121 BCE and 117 BCE, and rebuilt in 10 CE and 6 CE, respectively. The mortars of the older concretes contain volcanic ashes that have been interpreted as coming from epiclastic sediments (Jackson et al. 2007). These are earthy deposits of reworked volcanic and sedimentary sands and gravels. They were eroded from the tuff and granular ash deposits of the Roman landscape prior to each aggradational phase associated with episodes of mid-Pleistocene climate change and sea level rise, and deposited by tributaries to the Tiber River into the center of Rome (Marra et al. 2008). The predominant mortar aggregate of the 121 BCE concrete of the Temple of Concord concrete is indeed the epiclastic mid-Pleistocene San Paolo Formation (Marra and Rosa 1995), and the foundation of the temple was excavated from

Anejos de AEspA LXXVII

these deposits.2 Here, the San Paolo Formation is composed of clasts eroded from glassy Monti Saba­tini tuffs, mainly Tufo Giallo di Prima Porta and Tufo Rosso a Scorie Nere, feldspar and quartz sands, and rounded lava pebbles (fig. 3b). The concrete has soft, porous Tufo Giallo di ­Prima Porta as caementa, likely excavated from outcrops of the pyroclastic flow downslope. By contrast, the concrete of the 117 BCE concrete of the Temple of Castor and Pollux contains caementa of soft, poorly consolidated Tufo del Palatino, traditionally called “cappellaccio”, and particles of the tuff also occur in the mortar (fig. 3c). The mortar also contains abundant dull, dusky-brown Pozzolane Rosse scoriae as well as other pumiceous and palagonite particles. This was thought to be reworked ash of the epiclastic Aurelia Formation (Jackson et al. 2007). The current interpretation, however, is that the ash was excavated from the Pozzolane Rosse pyroclastic flow, since new examination of the stratigraphy of pyroclastic flow deposits in the center of Rome reveals that the Pozzolane Rosse ignimbrite did flow from Alban Hills volcano into the area now occupied by Stazione Termini (Marra et al. 2015). The early Imperial era concrete at both temples, however, has far more cohesive mortar with the reddened intermediate alteration facies of Pozzolane Rosse as pozzolan, similar to the well-bonded mortars of Tomb of Caecilia Metella concretes (fig. 3h) (Jackson et al. 2007; Jackson et al. 2010). Other late Republican age concretes from the western Forum, as from the walls of the Tabular­ ium (~70 BCE), the Caesarean age Rostra (~50 BCE), and the podium of the Temple of Saturn (~48 BCE), also contain Pozzolane Rosse pozzolan, as well as tuff clasts that seem to be small particles of the caementa (fig. 3d-g).3 The Pozzolane Rosse of these mortars is mainly the upper alteration facies of the deposit, and contains common pedogenic features such as root traces and thick, dense, yellow clay mineral (halloysite) coatings associated with the intensely hydrolytic midPleistocene paleosol that formed on the surface of the pyroclastic flow. These mortars also incorporate common particles of the reworked and weath2  See Jackson and Marra 2006; Jackson et al. 2010 for geological nomenclature. 3   The Navalia (~174 BCE or 110-100 BCE (?)) or Testaccio shipshed (Cozza and Tuzzi 2006; Jackson et al. 2007; Mogetta 2015; Marra et al. 2015), contains variable mortar mixes, here mainly sub-rounded, weathered palagonite clasts of the Pozzolanelle pyroclastic flow (fig. 3a).

Anejos de AEspA LXXVII

Technological Confidence in Late Republican...

ered palagonitic glass deposit that overlies the Pozzolane Rosse pyroclastic flow south of Rome (see fig. 3d, f); these were mistakenly identified as Tufo Lionato tuff in previous descriptions of the mortars (Jackson et al. 2007; Jackson et al. 2011). The concretes of several structures in the Largo Argentina Sacred Area also have mortars with Pozzolane Rosse pozzolan (Marra et al. 2015), mainly from the weathered upper horizons of the pyroclastic flow. Overall these are poorly cohesive, friable mortars. This is due in part to the weathered, opaque nature of the scoria groundmass and the thick yellow clay mineral coatings on the surfaces of these particles, which seem to have less pozzolanic reactivity with lime and, therefore, less bonding with the cementitious matrix. No strätlingite has been detected through microscopy or Xray diffraction studies. For the mortars of the Tomb of Caecilia ­Metella (~30 BCE) with Pozzolane Rosse volcanic ash pozzolan excavated from the middle and lower horizons of the pyroclastic flow (fig. 3h), builders evidently greatly improved methods for lime calcination, hydration, and mixing of the mortars. This is reflected in the well-bonded mortar fabric, and the bundles of strätlingite plates that reinforce the cementitious matrix and scoria interfacial zones (fig. 4a). The importance of Pozzolane Rosse scoriaceous ash excavated from the dusky red middle horizon and dark gray lower horizons of the pyroclastic flow, first identified qualitatively by Van Deman (1912a), has to do with the nature of the mineral textures on the surfaces of the scoria particles. These are thin, delicate coatings of opal, halloysite clay mineral, and zeolite mineral on the surfaces of scoriae, and especially in accretions of fine ash particles (Jackson et al. 2010).4 These minerals and the glassy groundmass of the scoriae reacted with hydrated lime, or portlandite (Ca(OH)2), to produce highly durable, poorly crystalline calcium-aluminatesilica­ -hydrate (C-A-S-H) binder and crystalline  cementitious microstructures, mainly platy strätlingite, especially in accretions of microscoria on scoria perimeters (fig. 3h) (Jackson et al. 2011). These cementitious components provide internal cohesion and mechanical resilience, measured as tensile strength or fracture toughness (Brune et al. 4  Finely ground zeolite minerals and poorly crystalline metakaolinite, a clay mineral derivative that is similar to halloysite, are important reactive alumino-silicate materials in Portland cement concrete systems (Snellings et al. 2012).

21

2013; Jackson et al. 2014), to support the complex static loads required of concrete structural elements in the Imperial age monuments (Brune 2010) and the dynamic loads induced by intermittent ground shaking from moderate magnitude earthquakes (Galli and Molin 2013). It should be noted that not all Imperial age concretes have this highly resilient Pozzolane Rosse mortar. The wall concretes of the Domus Aurea (68 CE) contain reworked epiclastic ash (Jackson et al. 2010), for example, and second and third century CE concrete structures at the Villa dei Quintilli on the via Appia Antica, for example, employ Pozzolane Rosse, Pozzolane Nere and Pozzolanelle ash quarried near the building site (Belfiore et al. 2015). Furthermore, construction disasters and building collapse were common occurrences in late Republican and Imperial Rome. It seems that insufficient lime in the pozzolanic sand mix was the principal cause of these construction failures, “the result of which the aggregate is laid without a proper mortar”, as described by Pliny the Elder (Naturalis Historia 36.176) (Oleson 2011). The advances in concrete technologies that are exemplified by the Tomb of Caecilia Metella and the Theater of Marcellus occurred at about the same time that Vitruvius was writing de Architectura. He described how over many centuries of practice (usus meditatio) Republican era builders developed a good empirical knowledge (scientia) of the material characteristics of the geologic materials of the Roman landscape (de Architectura 2.4.1-3, 2.5.1-3, 2.7.12.4.1-3, 2.5.1-5), and from this technical competency and confidence began to integrate new theoretical concepts (ratiocinationes) and skillful effort (fabrica) with new dimension stone and concrete materials and designs (Jackson and Kosso 2013). Recent studies of these monuments (Gerding 2002; Jackson et al. 2011) support an older point of view – that of E. B. Van Deman writing 100 years ago – that it was the late Republican era builders who developed the soph­ isticated pozzolanic mortars and complex concrete construction methods that underlie Imperial age monumental architecture. This contrasts with the “architectural revolution” hypothesis that ­focuses on the Fire of 64 CE as the driving force for concrete material innovation (Lechtmann and Hobbs 1998). Rather, private investment in both architectural and seawater concrete development in first century BCE, especially in domestic construction in Rome and in piscinae, or fish raising pools, associated with coastal villas (Mogetta

22

Marie D. Jackson

2015: 19-31; Hohlfelder and Oleson 2014: 227230) seems to have been a critical factor. The contribution of first century CE builders was to use this late Republican era material in increasingly daring structural designs. ASPECTS OF TECHNOLOGICAL CONFIDENCE In de Architectura, which is the most comprehensive written record of late Republican builders’ perceptions of the material properties of their concrete structures; Vitruvius uses the participle, confidens, only two times. In the first (de Architectura 6, preface 2.5), he describes the remarks of the Greek philosopher Theophrastus (371-287 BCE), “urging men to acquire learning rather than to put their trust (confidentes) in money” (Morgan 2005). In the second (de Architectura 7, preface 10), he expresses gratitude to ancestral writers from whose ocean of intellectual services... we have gained the means of writing with more eloquence and readiness, and trusting (confidentes) in such authorities we venture to put together a new manual of architecture (translated Granger 1934).

Republican era scholars may have trusted in an ancient corpus of texts as a dynamic source for new writing, but Vitruvius does not explicitly include confidens as a quality of builders’ perceptions of their expertise in construction technologies. Instead, he describes an ethos of building that incorporated a long tradition of observation, display and exhibition, imitation, thoughtful reflection, skill, and the ability to learn, which created new advances in construction durability and architectural design (Jackson and Kosso 2013). He does seem, however, to express indirectly several aspects of confidence, trust, and reliance in Republican builders’ technical expertise as they developed sustainable concretes for monumental and maritime construction. One aspect of confidence can be considered as the feeling of assurance that arises from the appreciation of one’s or another’s abilities.5 In the case of the very late Republican era builders, this could have developed from a burgeoning rec5  Definitions of confidence come from the Merriam Webster Dictionary and Thesaurus online and the Charlton T. Lewis, Charles Short, A Latin Dictionary online, http://www. perseus.tufts.edu/hopper/text?doc=Perseus:text:1999.04.0059.

Anejos de AEspA LXXVII

ognition that they had truly created new technological competencies with concrete construction. This quality of assurance seems to underlie Vitruvius’ exuberant expression of satisfaction with the aesthetics and functionality of monumental archi­tecture that was achieved through collaboration among architect, building owner, contractor or overseer, and workmen (de Architectura 6.8.9-10), When the work has achieved a magnificent appearance, the praise for the expense goes to the owner; when the [workmanship is] fine, the management by the overseer is pronounced good; and when it has true beauty due to proportion and symmetry the glory goes to the architect...[who], having already imagined [the work] before it is begun, has a definite [idea] of what it will be with respect to [its] elegance, beauty, and quality of design (translated Jackson and Kosso 2013).

Vitruvius placed this statement at the end of a detailed narrative describing the means to stabilize buildings, and emphasizing the importance of arched structures with voussoirs in reinforcing concrete foundations and substructures (de Architectura 6.8.1-4). The concrete foundations of the Tomb of Caecilia Metella and the Theater of Mar­ cellus, which were constructed when he was writing de Architectura, are some of the first brickfa­ced arched opus testaceum concrete structures in Rome (Gerding 2002; Ciancio Rossetto 1995). The builders who created these structures had no knowledge of the remarkable cementitious microstructures – a highly durable C-A-S-H binder and fine strätlingite reinforcements at the micron scale (Fig. 4a) – that they had produced. Even so, their empirical explorations and innovations with lime hydration, selection of specific Pozzolane Rosse alteration facies as pozzolan, and development of brick-faced conglomeratic concrete m ­ asonry form the foundations of the confidencia expressed by Vitruvius regarding the principles of fine work­ manship (or elegance), magnificence and design of architectural practice, which would so influence the construction of Imperial era monuments in Rome. Another aspect of confidence is the feeling of believing in the trustworthiness or reliability of a person or thing. It is this sense of trust and reliance, of feeling assured or protected against discontinuance or change, that seems to be reflected in Vitruvius’ discussion of “stability and the means by which buildings may be planned so as

Anejos de AEspA LXXVII

Technological Confidence in Late Republican...

to endure without defect” (de Architectura 6.7.7). Indeed, he dedicates chapter 6.8 to explaining methods through which the foundations of buildings “will be assuredly sound and durable” and “how buildings can be carried out without failure” (de Architectura 6.8.1, 6.8.8). On the other hand, he also expresses a lack of trust or fiducia soft (molli caemento) conglomeratic concrete walls (de Architectura 2.8.9), Those structures in soft rubble... are not the ones that will resist ruin as time passes. And thus, when assessors are appointed to evaluate party walls they never assess soft rubble walls according to their initial cost, but rather, when they look at the price recorded in the original contracts, they deduct 1/80th of that sum for each subsequent year and the remaining amount is fixed as the current value of the walls. They have rendered judgment, in effect, that such walls cannot last more than 80 years... (translated Rowland and Howe 1999).

The concrete walls and foundations of structures constructed 80 years prior to de Architectura, that is, prior to about 100 BCE, would likely have been considered as made of flawed, unsound, and poorly-cohesive cementitious materials to some degree by Vitruvius’ contemporaries. Such walls, with porous, friable tuffs such as Tufo ­Giallo della Via Tiberina or Tufo del Palatino as coarse aggregate and weakly cohesive mortars, persist at the Temple of Concord (121 BCE) and the Temple of Castor and Pollux (fig. 5b, c), but many such walls apparently did not survive well in Republican era Rome. This is reflected in Vitruvius’ explanation of “depreciation” in the Roman legal code (Rihll 2013),6 which implies that concrete walls in buildings under private ownership were expected to withstand deterioration for no more than 80 years of service life. First century BCE houses also frequently collapsed, burned, or were demolished during uncontrolled urban growth, as recorded by Strabo (64/64 BCE to 24 CE) (Geografica 5.3.7), because of this concourse of blessings [in] the city... the building of houses, which goes on unceasingly in consequence of the collapses and fires and repeated sales (these last, too, going on unceasingly); and indeed the sales are intentional collapses, as it 6  The discussion of ‘depreciation’ or ‘amortization’ by Rihll (2013) in late Republican Rome construction also explores the role of these concepts in the economic and moral values of ancient Roman thought.

23

were, since the purchasers keep on tearing down the houses and build new ones, one after another, to suit their wishes (translated Jones 1999).7

Owners and potential buyers of tenement or apartment buildings, lacking a sense of fiducia in the longevity of these unstable concrete structures, would have wanted a fair assessment of their amortized value over time (Rihll 2013). Strabo­(5.3.7) described interventions by Emperor Augustus to prevent further building collapse and establish a firm building code, Now Augustus Caesar concerned himself about such impairments of the city, organizing for protection against fires a militia composed of freedmen, whose duty it was to render assistance, and also to provide against collapses, reducing the heights of the new buildings and forbidding that any structure on the public streets should rise as high as seventy feet; but still his constructive measures would have failed by now were it not that the quarries and the timber and the easy means of transportation by water still hold out (translated Jones 1999).

These interventions evidently included the cohesive Pozzolane Rosse mortar mix that is ­preserved in many of the monumental buildings of the Augustan construction program (Van ­Deman 1912a). The mechanical and cementitious resilience of these concrete structures would eventually extend far beyond the expected 80-yearlifetime of Republican age urban construction, until the present millennium. By the late first century CE, builders in Rome could observe that the more-or-less standardized Augustan era cohesive mortar mix had performed well in conglomeratic concrete architectural structures for 50 to 100 years of service life. They seem to have developed an increasingly firm trust in these materials, which gave rise to the expression of another aspect of confidence, recorded by increasingly audacious and innovative architectural designs in vaulted construction in Imperial age Rome (Lancaster 2005). The successful completion of these monumental constructions required a sophisticated empirical understanding of the mechanical demands of strength, ductility and toughness on concrete structural elements (Brune 2010).

7  For a more complete description of these building disasters and construction fraud see Oleson 2011.

24

Marie D. Jackson

Anejos de AEspA LXXVII

Fig. 5. Photographs comparing Republican era concrete walls at a) the Temple of Concord (121 BCE) with soft, porous Tufo Giallo della Via Tiberina opus incertum facing and conglomeratic aggregate (molli caementa, de Architectura 6.8.9) and b) the Tomb of Caecilia Metella (30 BCE) with highly refined brick opus testaceum facing and Tufo Lionato conglomeratic aggregate.

Late first century BCE builders could observe that maritime concrete harbor structures constructed with pulvis pumiceous volcanic ash at Portus Cosanus and in the Bay of Pozzuoli, for example, had remained intact and immobile on the seafloor for a few decades. Vitruvius wrote a rather cautious statement regarding the importance of pumiceous volcanic ash from “the vicinity of Baiae and the territory of the municipalities around Mount Vesuvius” in the cementing processes of maritime concretes, stating that: When three substances [lime (calx), pozzolan (pulvis) and rubble (caementum, tuff coarse aggregate)] formed in like manner through the violence of fire come into one mixture, they suddenly take up water and cohere together. They are quickly hardened by the moisture and made solid, and can be dissolved neither by the waves nor by the power of water (de Architectura 2.6.1) (translated Granger 1931).8

Whether Imperial era builders did indeed under­take the daunting task of transporting tens of thousands of kilos of pulvis volcanic ash from the Campi Flegrei and Somma-Vesuvius volcanic ­districts to harbor construction sites along the 8  In this translation of de Architectura 2.6.1, Granger (1931) correctly interprets “recepto liquore una cohaerescunt et celeriter umore duratae solidantur” as Vitruvius’ empirical description of the process of hydration that drives pozzolanic reaction between calcium hydroxide and pumiceous volcanic ash constituents. This process rapidly produces cohesive C-AS-H binder (Jackson et al. 2013). Literally, this would be “when these three substances had come into one mixture, suddenly they cohered into one [substance] by accepting liquid [and] they quickly solidified due to the moisture and were made hard.”

Italian peninsula and far distant Mediterranean harbors can be addressed qualitatively through measurement of trace element ratios in pumice clasts from the mortars of the harbor concretes. Although much dissolution and alteration of glass has occurred in these particles, which also contain cementitious hydrates (Jackson et al. 2013), trace element ratios and mineral assemblages of bulk pumice samples from the harbors studied by the ROMACONS drilling teams do all fall in the range of Gulf of Naples pumice compositions (fig. 6). Therefore, it seems quite possible that tens of thousands of tons of pulvis ash were shipped to Imperial harbor sites at Caesarea Palaestinae (23 to 15 BCE), Portus Claudius, Portus Neronis, Alexandria (50 CE), and others (Hohlfelder and Oleson 2014: 222-230). The most astonishing of these is the Caesarea Palaestinae harbor that was built in the open sea on the central coast of Israel, and whose mortars employed about 24,000 m3 of pumiceous ash pozzolan. The harbor was constructed contemporaneously with the later phase of the Theater of Marcellus in Rome. The engineers and builders who conceived this massive building program, commissioned by King Herod in 25 BCE, seem to have an audacious confidence in both their technological competencies and the seawater concrete they formulated with pulvis pozzolan. Pliny the Elder, writing about 80 years later, would describe the rock-like durability and resistance to wave forces of this maritime concrete (Naturalis Historia 25.166), For who could marvel enough that on the hills of Puteoli there exists a dust (pulvis)... that, as soon as it comes into contact with the waves of the sea

Anejos de AEspA LXXVII

Technological Confidence in Late Republican...

25

Fig. 6. Trace element studies, Zr/Y and Nb/Y, of powdered pumice clasts from the volcanic ash pozzolan of the ancient maritime mortars compared with Mediterranean pumice deposits beyond the Gulf of Naples. See Jackson (2014, fig. 7.11) for descriptions of pyroclastic deposits and reference chemical analyses.

and is submerged, becomes a single stone mass, impregnable to the waves and every day stronger... (translated Eichholz 1962).

Modern concrete engineers have a great deal to learn from Roman technologies in this regard.

CONCLUSIONS By late first century BCE Romans had perfected technologies for the production of concre­ tes to resolve constructional and structural problems of durability in the monuments of Rome and in central Italian coast and eastern Mediterranean maritime harbor structures. Examinations of de Architectura (2.4.1-3, 2.6.1-6, 5.12.2-6, 2.5.1-3) and Republican era monumental and harbor structures indicate that builders deliberately focused their technological expertise on improving the material and physical properties of the volcanic ash-lime mortars that bind these conglomeratic concretes. In Rome, very high quality mortars using specific alteration horizons of Pozzolane Rosse scoriaceous volcanic ash (harena

fossicia) from Alban Hills volcano were employed, for example, in the foundations and walls of the Tomb of Caecilia Metella (~30 BCE) and Theater of Marcellus (46-44 BCE, 23-17 BCE). This would become the principal formulation of Imperial era foundation and wall concrete. By contrast, mortars of the harbor concretes at Portus Cosanus and Santa Liberata (~50 BCE) and Bai­ anus Sinus (~35 BCE) all contain pumiceous ash (pulvis) from Campi Flegrei volcano, identified through pumice trace element and mineralogical signatures. This would become the formulation of Imperial maritime concretes, as at Casearea Palestinae (23 to 15 BCE), whose mortars employed about 25,000 m3 of Gulf of Naples ash. Mortars from Tomb of Caecilia Metella and Theater of Marcellus concretes have a sophisticated calciumaluminum-silicate-hydrate binder and platy strät­ lingite crystals that reinforce interfacial zones, similar to the highest quality Trajanic era mortars. All maritime mortars drilled by the ROMACONS drilling program grow Al-tobermorite, a rare calcium-silicate-hydrate mineral, in relict lime clasts and the cementitious matrix. These are highly desirable components of modern concretes

26

Marie D. Jackson

but very difficult to produce with conventional Portland cement mixes. Romans were therefore thoroughly justified in placing their confidence in these sophisticated mortar formulations. ACKNOWLEDGEMENTS All samples were collected with the permission of the Sovrintendenza Capitolina Beni Culturali di Roma Capitale and Soprintendenza ­Archaeologica di Roma. I thank Massimo Vitti and Lucrezia Ungaro at the Markets of Trajan for their support and collaboration over many years. Elisabetta Segala and Lisa Gianmichele assisted with sample collection in the monuments of the western Forum and the Tomb of Caecilia Metella. I am e­ specially grateful for discussions with Tracey Rihll, whose examination of Vitruvius’ remarks regarding depreciation in Republican era walls forms the foundation of the exploration of technical confidence presented in this article. John P. Oleson contributed ­critical insights into the interpretation of literary sources. REFERENCES Belfiore, C. M., Fichera, G. V., La Russa, M. F., Pezzino, A., Ruffolo, S. A., Galli, G. and Barca­, D. 2015: “A multidisciplinary approach for the archaeometric study of pozzolanic aggregate in Roman mortars: the case of the Villa dei Quintilli (Rome, Italy)”, Archaeometry, 57.2, pp. 269-296. Bianchi, E., Meneghini, R., Jackson, M., Brune, P. and Marra, F. 2011: “Archaeological, structural, and compositional observations of the concrete arch­itecture of the Basilica Ulpia and Trajan’s Forum”, in Ringbom, A. and Hohlfelder, R. L. (eds.), Building Roma Aeterna. Current Research on Roman Mortar and Concrete. Proceedings of the Conference March 27-29, 2008, pp. 72-93, Commentationes Humanarum Litterarum Series 128. Societas Scienti­ arum Fennica, Helsinki. Brandon, C. J. 2014: “Roman formwork used for under­water concrete construction”, in Oleson, J. P. (ed.), Building for Eternity. The History and Techno­ logy of Roman Concrete Engineering in the Sea, pp. 189-222. Oxbow Books, Oxford. Brandon, C., Hohlfelder, R. L. and Oleson, J. P. 2008: “The concrete construction of the Roman harbours of Baiae and Portus Iulius, Italy: The ROMACONS 2006 field season”, International Journal of Nautical Archaeology, 37, pp. 374-392.

Anejos de AEspA LXXVII

Brune, P. F. 2010: The Mechanics of Imperial Roman Concrete and the Structural Design of Vaulted Monuments, Ph.D. Thesis, University of Rochester, USA. Brune, P., Ingraffea, A. R., Jackson, M. D. and Perucc ­ hio, R. 2013: “The fracture toughness of an imperial Roman mortar”, Engineering Fracture Mechanics, 102, pp. 65-76. Brune, P. F. and Perucchio, R. 2011: “Numerical simulation of the mechanical behavior of opus caementic­ ium: opportunities and challenges”, in Ringbom, A. and Hohlfelder, R. L. (eds.), Building Roma Aeterna. Current Research on Roman Mortar and Concrete. Proceedings of the Conference March 27-29, 2008, pp. 96-108, Commentationes Humanarum Litterarum Series­128. Societas Scientiarum Fennica, Helsinki. Brune, P. F. and Perucchio, R. 2012: “Roman concrete vaulting in the Great Hall of Trajan’s Markets: structural evaluation”, Journal of Architectural Engin­eering, 18, pp. 332-340. Cozza, L. and Tucci, P. L. 2006: “Navalia”, Archeologia Classica, 57, pp. 175-202. D’Ambrosio, E., Marra, F., Cavallo, A., Gaeta, M. and Ventura, G. 2015: “Provenance materials for Vitruvius’ harenae fossiciae and pulvis puteolanis: geochemical signature and historical-archaeological implications”, Journal of Archaeological Science, Reports, 2, pp. 186-203. DeLaine, J. 1997: The Baths of Caracalla. A study in the design, construction, and economics of large-scale building projects in imperial Rome, Journal of Roman­ Archaeology Suppl. 25. Journal of Roman Archaeology, Portsmouth, R. I. Eichholz, D. E. (transl.) 1962 (reprinted 2001): Pliny, Natural History, Books 36-37. Harvard University Press, Cambridge, Mass. Galli, P. A. C. and Molin, D. 2013: “Beyond the damage threshold: the historic earthquakes of Rome”, Bulletin of Earthquake Engineering, 10.6, pp. 1-32. Gerding, H. 2002: The tomb of Caecilia Metella: tumu­lus, tropaeum, and thymele, Doctoral thesis, Lund University, Sweden. Granger, F. (transl.) 1931 (reprinted 2002): Vitruvius, On architecture, Books 1-5, Volume I, Books 1-5. Harvard University Press, Cambridge, Mass. Granger, F. (transl.) 1934 (reprinted 2004): Vitruvius, On architecture, Books 6-10, Volume II. Harvard University Press, Cambridge, Mass. Hohlfelder, R. L. and Oleson, J. P. 2014: “Roman maritime concrete technology in its Mediterranean context”, in Oleson, J. P. (ed.), Building for Eternity. The History and Technology of Roman Concrete Engin­eering in the Sea, pp. 223-235. Oxbow Books, Oxford.

Anejos de AEspA LXXVII

Technological Confidence in Late Republican...

Jackson, M. D. 2014: “Seawater concretes and their material characteristics”, in Oleson, J. P. (ed.), Building for Eternity. The History and Technology of Roman Concrete Engineering in the Sea, pp. 141187. Oxbow Books, Oxford. Jackson, M. D., Chae, S. R., Mulcahy, S. R., Meral, C., Taylor, R., Li, P., Moon, J., Yoon, S., Emwas, A.-H., Vola, G., Wenk, H.-R. and Monteiro, P. J. M. 2013: “Unlocking the secrets of al-tobermorite in Roman seawater concrete”, American Mineralogist, 98, pp. 1669-1687. Jackson, M. D., Ciancio Rossetto, P., Kosso, C. K., Buonfiglio, M. and Marra, F. 2011: “Building materials of the Theater of Marcellus, Rome”, Arch­aeometry, 4.4, pp. 728-742. Jackson, M. D., Deocampo, D., Marra, F. and Scheetz, B. E. 2010: “Mid-Pleistocene volcanic ash in ancient Roman concretes”, Geoarchaeology, 25, pp. 36-74. Jackson, M. D. and Kosso, C. K. 2013: “Scientia in Republican era. Stone and concrete masonry”, in Evans, J. R. (ed.), A Companion to the Archaeology of the Roman Republic, pp. 268-284. WileyBlackwell­, Malden, Mass. Jackson, M. D., Landis, E. N., Brune, P. F., Vitti, M., Chen, H., Li, Q., Kunz, M., Wenk, H.-R., Monteiro, P. J. M. and Ingraffea, A. R. 2014: “Mech­ anical resilience and cementitious processes in imperial Roman architectural mortar”, Proceedings of the National Academy of Sciences, 111.52, pp. 1848418489. Jackson, M. D., Logan, J. M., Scheetz, B. E., Deocampo, D. M., Cawood, C. G., Marra, F., Vitti, M. and Ungaro, L. 2009: “Assessment of material characteristics of ancient concretes, Grande Aula, Markets of Trajan, Rome”, Journal of Archaeological Science, 36, pp. 2481-2492. Jackson, M. D. and Marra, F. 2006: “Roman stone masonry: volcanic foundations of the ancient city”, American Journal of Archaeology, 110, pp. 403-436. Jackson, M., Marra, F., Deocampo, D., Vella, A., Kosso, C. and Hay, R. 2007: “Geological observations of excavated sand (harenae fossiciae) used as fine aggregate in ancient Roman pozzolanic mortars”, Journal of Roman Archaeology, 20, pp. 1-30. Jones, H. L. (transl.) 1923 (reprinted 1999): “Strabo, Geography”, Books 3-5. Harvard University Press, Cambridge, Mass. Lancaster, L. C. 2005a: Concrete Vaulted Construction in Imperial Rome. Innovations in Context. Cambridge University Press, Cambridge. Lancaster, L. C. 2005b: “The process of building the Colosseum: the site, materials, and construction

27

techniques”, Journal of Roman Archaeology, 18, pp. 57-82. Lancaster, L. C. 1998: “Building Trajan’s Markets”, American Journal of Archaeology, 102, pp. 283-308. Lancaster, L. C., Sottili, G., Marra, F. and Ventura, G. 2010: “Provenancing of light weight volcanic stones used in ancient Roman concrete vaulting: evid­ ence from Rome”, Archaeometry, 53.4, pp. 707‑727. Lechtman, H. N. and Hobbs, L. W. 1987: “Roman concrete and the Roman architectural revolution”, Ceramics and Civilization, 3, pp. 81-128. Marra, F., D’Ambrosio, E., Gaeta, M. and Mattei, M. 2015: “Petrochemical identification and insights on chronological employment of the volcanic aggregates used in ancient mortars”, Archaeometry, online 20 JAN 2015, DOI: 10.1111/arcm.12154 Marra, F., Karner, D. B., Freda, C., Gaeta, M. and Renne, P. R. 2008: “Large mafic eruptions at the Alban Hills volcanic district (Central Italy): chronostratigraphy, petrography and eruptive behavior”, Journal of Volcanologic and Geothermal Research, 179, pp. 217-232. Marra, F., D’Ambrosio, E., Sottili, G. and Ventura, G. 2013: “Geochemical fingerprints of volcanic materials: identification of a pumice trade route from Pompeii to Rome”, Geological Society of America Bulletin, 125, pp. 556-577. Marra, F. and Rosa, C. 1995: “Stratigrafia e assetto geologico dell’area Romana”, in Funiciello, R. (ed.), Memorie descrittive della carta geologica d’Italia, la geologia di Roma, pp. 50-118. Istituto Poligrafico e Zecca dello Stato, Rome. Massazza, F. 1988: “Pozzolana and pozzolanic cem­ ents”, in Hewlett, P. C. (ed.), Lea’s chemistry of cement and concrete, pp. 471-632. Arnold Publishers, London (fourth edition). Mogetta, M. 2015: “A new date for concrete in Rome”, Journal of Roman Studies, 105, pp. 1-40. Morgan, M. H. (transl.) 1914 (reprinted 2005): Vitruv­ ius. The ten books on architecture. Harvard University Press, Cambridge, Mass. Oleson, J. P. (ed.) 2008: The Oxford Handbook of Engin­eering and Technology in the Classical World. Oxford University Press, New York-Oxford. Oleson, J. P. 2011: “Harena sine calce: Building disasters, incompetent architects, and construction fraud in ancient Rome”, in Ringbom, A. and Hohlfelder, R. L. (eds.), Building Roma Aeterna. Current Research on Roman Mortar and Concrete. Proceedings of the Conference March 27-29, 2008, pp. 9-26, Commentationes Humanarum Litterarum Series 128. Societas Scientiarum Fennica, Helsinki.

28

Marie D. Jackson

Oleson, J. P., Jackson, M. D. and Vola, G. 2014: “Append­ix 3: catalogue and descriptions of concretes drilled from harbor structures by ROMACONS”, in Oleson, J. P. (ed.), Building for Eternity. The History and Technology of Roman Concrete Engineering in the Sea, pp. 243-283. Oxbow Books, Oxford. Rihll, T. 2013: “Depreciation in Vitruvius”, The Classical Quarterly, 63.2, pp. 893-897. Rowland, I. D. and Howe, T. N. 1999: Vitruvius Ten Books on Architecture. Cambridge University Press, Cambridge. Snellings, R., Mertens, G. and Elsen, J. 2012: “Supplementary cementitious materials”, Reviews in Mineralogy and Geochemistry, 74, pp. 211-278.

Anejos de AEspA LXXVII

Van Deman, E. B. 1912a, 1912b: “Methods of determining the date of Roman concrete monuments (first and second papers)”, American Journal of Arch­aeology, 16, pp. 230-251, 387-432. Vitti, M. 2007: “Mercati di Traiano”, in Ungaro, L. (ed.), Il Museo dei Fori Imperiali nei Mercati di Traia­no, pp. 5-19. Electa, Milano. Vola, G., Gotti, E., Brandon, C., Oleson, J. P.  and  Hohlfelder, R. L. 2011a: “Chemical, mine­ ­ ralogical and petrographic characterization  of ­Roman ancient­hydraulic concretes cores from Santa Liberata, Italy, and Caesarea Palestinae, ­Israel”, Periodico di Mineralogia, 80, pp. 317338.

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

L’USO DELLE POLVERI POZZOLANICHE NEI GRANDI CANTIERI DELLA GALLIA CISALPINA DURANTE L’ETÀ ROMANA REPUBBLICANA: I CASI DI AQUILEIA E RAVENNA JACOPO BONETTO*, GILBERTO ARTIOLI**, MICHELE SECCO**, ANNA ADDIS* * Università degli Studi di Padova - Dipartimento dei Beni Culturali ** Università degli Studi di Padova - Dipartimento di Geoscienze

RIASSUNTO: Nuove ricerche sulle fortificazioni repubblicane di Aquileia sono state avviate nel 2011 nell’area dei Fondi Cossar con l’obiettivo di raccogliere dati cronologici e architettonici sul sistema difensivo della colonia. Lo scavo ha interessato i livelli di fondazione della cortina e ha rivelato l’uso di laterizi di modulo greco e di grandi masse di calcestruzzo di straordinaria resistenza meccanica. Le analisi archeometriche condotte su questi blocchi hanno mostrato l’uso nella malta di una polvere vulcanica di probabile provenienza centro-italica o flegrea. L’uso di polveri vulcaniche è stato riscontrato anche nelle malte usate per la costruzione del muro di difesa dell’insediamento di Ravenna (fine del iii sec. a.C.). I dati così raccolti hanno messo in luce ancora sconosciute pratiche di diffusione e uso delle polveri vulcaniche per la realizzazione di leganti idraulici in regioni poste a grande distanza dai luoghi di origine del materiale e in una fase (iii-ii sec. a.C.) quando simili procedure conoscevano le prime sperimentazioni in Italia centrale. PAROLE CHIAVE: Fortificazioni romane, Analisi archeometriche, Malte antiche, Polveri pozzolaniche. ABSTRACT: New research on the Roman republican walls of Aquileia was started in 2011 by the University of Padua in “Fondi Cossar” area, aiming at providing chronological and building information about the defensive system of the colony. The excavation has revealed the use in the foundations layers of Greek bricks and concrete masses which are quite exceptional for their mechanical strength. The archaeometric analysis performed on these blocks show the use in the lime-mortar of a volcanic ash probably imported from central Italy or the Campi Flegrei. The use of volcanic ash has also been revealed from analysis of the mortars used in the brick defensive walls of Ravenna (III century BC). The data collected throws light on so far unknown practices of trade and the use of volcanic ash in regions far away from their districts of origin and at a very early stage when similar technical procedures had just been carried out in central Italy. KEYWORDS: Roman urban defences, Archaeometric analysis, Ancient mortars, Pozzolanic ashes. RESUMEN: En el año 2011, con el objetivo de recopilar datos cronológicos y arquitectónicos en relación con el sistema defensivo de la colonia, se han efectuado nuevas investigaciones sobre las fortifi­ caciones de Aquileia, en el área de los Fondi Cossar. La excavación ha interesado los niveles de cimentación de la muralla y ha revelado el empleo de ladrillos de módulo griego y grandes cantidades de hormigón de extraordinaria resistencia mecánica. Los análisis arqueométricos practicados en estos elementos de mortero han demostrado el uso de un polvo volcánico de probable proveniencia centroitálica o de los Campos Flegreos. El empleo de polvos volcánicos se ha evidenciado, además, en los morteros usados en la construcción del muro de defensa del asentamiento de Rávena (finales del iii a.C.). Los datos recopilados han evidenciado diferentes prácticas todavía desconocidas sobre la difusión y el uso de los polvos volcánicos para la realización de morteros hidráulicos en regiones situadas a gran distancia de los lugares de origen del material y en una fase (iii-ii a.C.) en la que los mismos procedimientos se encontraban en fase de experimentación en Italia central. PALABRAS CLAVE: fortificaciones romanas, análisis arqueométricos, morteros antiguos, polvos volcánicos.

30

JACOPO BONETTO, GILBERTO ARTIOLI, MICHELE SECCO, ANNA ADDIS

LE RICERCHE AD AQUILEIA E LE MURA REPUBBLICANE A partire dal 2007 l’Università degli Studi di Padova ha avviato un ampio progetto di ricerca incentrato sulla città romana di Aquileia, posta nel settore orientale della pianura padana a breve distanza dalle rive settentrionali del golfo di Venezia (fig. 1). Il centro venne fondato come colonia latina nel 181 a.C. con l’invio da parte del Senato di Roma di 3000 coloni e fu rinforzato nel 169 a.C. con l’aggiunta di 1500 famiglie (Liv. 39, 22, 6-7; 39, 45, 6-7; 39, 54, 2-13; 39, 55, 1-6; 40, 34, 2-3; sul rinforzo: Liv. 43, 1, 5-6; Chiabà 2009). Favorita dalla funzione di snodo commerciale tra il Mediterraneo e l’Europa continentale (Strab. 5, 1, 8), che già era propria dell’emporio preromano esistente nell’area almeno dalla prima età del Ferro, Aquileia conobbe un grandioso sviluppo urbanistico e architettonico ben noto per la tarda età repubblicana e l’epoca imperiale e assunse rilievo di primo piano nel quadro dell’impero fino ad essere considerata da Ausonio la nona città più importante del mondo antico (Aus. 11, 9, 4). In questo scenario storico-archeologico di primo livello per la ricerca, favorita anche dalla mancata so-

Anejos de AEspA LXXVII

vrapposizione di grandi centri urbani sul sito antico, gli interessi dell’Università di Padova hanno riguardato vari aspetti dell’insediamento romano, tra cui la cultura artistica (Clementi et al. 2009), con lo studio della ricchissima produzione musiva, gli aspetti architettonici (Ghedini e Novello 2009; Bonetto e Salvadori 2012) e tecnico-costruttivi dell’edilizia pubblica e privata del centro (Previato 2015) così come l’assetto urbano, ad oggi ancora poco conosciuto (Tiussi 2009; Ghiotto 2013). Da diversi anni un interesse particolare è stato anche rivolto allo studio delle fortificazioni urbane di Aquileia, che gli scavi del passato e le testimonianze letterarie hanno permesso di conoscere in forme molto chiare nella loro realtà strutturale e nella loro successione (Brusin 1966; Bonetto 2004; Villa 2004 per le fasi tardoantiche e altomedievali; Bonetto 2009). Questa importanza e abbondanza di testimonianze sulle difese urbane repubblicane di Aqui­ leia sono dovute al ruolo che il centro ebbe sin dalla fondazione come piazzaforte strategica destinata a contrastare le pressioni delle popolazioni celtiche, che già pochi anni prima della deduzione avevano invaso la pianura padana orientale fondando un oppidum nell’area della futura città.

Fig. 1. Immagine satellitare dell’Italia settentrionale con la posizione dei due centri di Ravenna e Aquileia.

Anejos de AEspA LXXVII

L’uso delle polveri pozzolaniche nei grandi cantieri...

31

Il lungo racconto liviano della fondazione sottolinea questa grande valenza difensiva, militare e di esposizione bellica del centro, che viene conservata inalterata per tutta la sua storia fino ad epoca tardo imperiale, quando Aquileia torna ad assumere il ruolo di baluardo per la penisola contro le devastanti invasioni delle popolazioni balca­ niche. Negli anni più recenti l’interesse per le fortificazioni più antiche della colonia si è rinnovato ed ha ricevuto nuovo impulso grazie all’avvio di un ampio programma di indagini presso il settore sud-orientale del nucleo urbano più antico corrispondente, nella topografia moderna, all’area dei Fondi Cossar (fig. 2). LE INDAGINI SULLE FORTIFICAZIONI DEL 2011 E 2012 Lo scavo sistematico di quest’area1 ha preso avvio nel 2009 ed ha riguardato principalmente una delle residenze private già note da indagini del secolo scorso (Bonetto et al. c.s.); una di queste (domus di Tito Macro) è stata oggetto tra il 2009 e il 2013 di uno scavo sistematico, che si è anche ampliato a comprendere gli assi stradali ad essa connessi (Bonetto e Ghedini 2013; Bonetto e Ghiotto 2014; Ghiotto 2014; Centola et al. c.s.). Tuttavia durante questo periodo le ricerche si sono anche indirizzate verso una porzione dei Fondi Cossar, prossima a quella delle domus, dove era nota da indagini del passato la presenza di un significativo tratto delle più antiche mura della città. Queste porzioni delle difese più antiche (cortina e torri), come altri in zone diverse della città, vennero infatti indagati da G. Brusin negli anni Trenta del secolo scorso per determinare il perimetro delle fortificazioni della colonia repubbli­ cana (fig. 3). Le indagini fornirono importanti informazioni sull’imponente sistema di fortificazioni e sui suoi caratteri costruttivi, ma i reso­conti dell’epoca risultano molto limitati sul piano delle informazioni edite (Brusin 1932).2 Per questo è stata avviata una nuova campagna di ricognizione dei tratti delle mura (2011) e sono state identificate le trincee di scavo operate 1   Il Progetto è sostenuto dalla Fondazione Aquileia, dalla società Arcus di Roma e dalla Soprintendenza Archeologia del Friuli Venezia Giulia. 2  Ulteriori brevi verifiche vennero condotte nello stesso settore da L. Bertacchi nel 1988, ma di queste indagini restano solo alcune fotografie conservato in Archivio.

Fig. 2. Pianta di Aquileia con indicazione delle mura repubblicane (A-B-C-D-E-F-G) e della posizione dei Fondi Cossar presso i quali è stato eseguito lo scavo di un tratto di fortificazione (elaborazione dell’autore).

da G. Brusin. Dopo lo svuotamento delle terre di riempimento è stato quindi possibile rimettere in luce la superficie del muro di difesa della città, conservato per alcuni corsi dell’alzato e apparso ben leggibile nel suo assetto strutturale e planimetrico (Bonetto e Pajaro 2012). Da subito sono però emersi vari problemi che hanno reso complesse e non del tutto esaurienti le veri­fiche programmate. Da un lato è stato notato che gli scavi compiuti da G. Brusin avevano rimosso, senza lasciare documentazione, una parte consistente della stratigrafia connessa alle fasi di costruzione, utilizzo e ripristino della fortificazione; dall’altro il progressivo approfondimento dello scavo si è dovuto confrontare con la presenza di vene d’acqua sotterranee che hanno prodotto costante allagamento dell’intera superficie del saggio di scavo ad una quota superiore di circa 0,5 m a quella della cresta del muro di fortificazione. Lo scavo è proceduto quindi in grandi difficoltà tecniche e logistiche, come la ciclica ri-sommersione delle strutture dopo i pompaggi meccanici e soprattutto l’impossibilità di approfondire l’indagine oltre

32

JACOPO BONETTO, GILBERTO ARTIOLI, MICHELE SECCO, ANNA ADDIS

Fig. 3. Aquileia. Tratto delle mura repubblicane rimesso in luce da G. Brusin negli anni Trenta del secolo scorso. È ben visibile la struttura omogenea in laterizi cotti (Archivio del Museo Archeologico nazionale di Aquileia).

il limite presso il quale il flusso e il ruscellamento dell’acqua sono risultati incontro­abili. Nonostante questi vincoli operativi, già nel 2011 erano stati raggiunti, in alcuni limitati punti, contesti stratigrafici non toccati dalle precedenti indagini e connessi alla costruzione delle fortificazioni. Con la campagna 2012 lo scavo ha permesso in primo luogo di raccogliere significativi indicatori per definire la cronologia di costruzione delle mura. Fino ad oggi infatti le fortificazioni erano state assegnate da quasi tutti gli studiosi al momento fondazione della città nel 181 a.C., al rinforzo del 169 a.C. o ai decenni immediatamente successivi (Bonetto 2004: 167-170), pur senza il supporto di indicatori derivati da scavi stratigrafici. Lo scavo condotto tra il 2011 e il 2012 a ridosso della fronte interna delle mura ha recuperato all’interno dei livelli di fondazione alcuni indicatori radiometrici tra loro coerenti che sembrano confermare pienamente una datazione nella prima metà del ii sec. a.C. per l’impianto del muro.3 3   Due campioni di carbone prelevati dagli strati addossati alla fondazione del muro forniscono date calibrate di morte del materiale vegetale comprese rispettivamente tra il 380 e il

Anejos de AEspA LXXVII

Lo scavo ha però anche consentito di giungere ad una piena conoscenza degli aspetti architettonico-strutturali delle fortificazioni più antiche della colonia e di verificare dettagli prima di adesso del tutto sconosciuti. La porzione di muro messa interamente in luce (circa 4 m) appartiene al lato orientale delle mura che fronteggiano la linea del porto fluviale e si trova presso l’angolo sud-orientale del perimetro (fig. 4). La vicinanza di questo tratto alla torre angolare determina un allargamento improvviso dello spessore della cortina da 2,38 m (8  piedi) a 4,76 m (16 piedi). Solo alcune assise dell’alzato del muro risultano conservate, mentre sono risultate visibili quelle appartenenti alla fascia di passaggio tra l’alzato e le fondazioni, distinti da una risega di pochi centimetri. In questa parte il muro, contrariamente a quanto in precedenza creduto (Bonetto 2004: 160), è realizzato con struttura omogenea e l’intero spessore è costruito con vari filari di laterizi cotti, rilevati per 7 corsi appartenenti alla fondazione e all’alzato, che sono stati oggetto di misurazione sistematica sia sulla fronte del muro sia sulla superficie rasata. Gli elementi fittili impiegati mostrano con ricorrente regolarità lati di 0,36-0,38 m e spessore di 0,075 m.4 Questo modulo, non noto nell’edilizia romana, fornisce un’informazione essenziale nello studio delle fortificazioni più antiche di Aquileia poiché rimanda con precisione al mattone ricordato da Vitruvio (Vitr. 2, 3, 3) con il nome di pentadoron, derivato dalla misura di cinque palmi e usato, secondo l’architetto augusteo, dai Graeci per le opere pubbliche. Questo dato, assieme alla tipologia greco-ellenistica delle torri (a pianta pentagonale) e delle porte (a cavedio circolare) delle mura (Bonetto 1998: 62-63, 78-79), lascia supporre la presenza di maestranze ellenistiche 170 a.C. e tra il 400 e il 200 a.C. (campione 12875A, US 10031B. Età radiometrica 2240 ± 45. Età calibrata: 68,2%: 390 BC (19,6%) 350 BC e 300 BC (48,6%) 200 BC; 95,4%: 400 BC (95,4%) 200 BC. Campione 12874A, US 10031A. Età radiometrica 2199 ± 35. Età calibrata: 68,2%: 360 BC (43,3%) 280 BC e 260 BC (24,9%) 200 BC; 95,4%: 380 BC (95,4%) 170 BC. Per cautela va però precisato che esiste una pur remota possibilità che alcuni livelli di fondazione siano collegati ad un intervento di restauro successivo alla costruzione delle mura e riferibile a fasi tardo repubblicane, cui rimanderebbero alcuni pur incerti indicatori ceramici. Su tale dettaglio sarà necessario in futuro effettuare riscontri e ulteriori approfondimenti. 4   I laterizi risultavano spesso spezzati e quindi non è facile determinare la forma originaria (quadrata o rettangolare). Tuttavia la netta prevalenza delle misure indicate su tutte le fronti dei laterizi posti al di sotto della risega fa pensare che la forma quadrata fosse prevalente nelle porzioni più basse del muro.

Anejos de AEspA LXXVII

L’uso delle polveri pozzolaniche nei grandi cantieri...

Fig. 4. Aquileia, fondi Cossar. I resti del muro repubblicano emerso nel corso dello scavo del 2012 (foto autore).

impegnate nel cantiere delle mura di Aquileia. Questo tema dei laterizi e delle maestranze di matrice greca attive nei cantieri di Aquileia sarà peraltro trattato in altra sede (Bonetto c.s.). Qui interessa notare che l’approfondimento dello scavo lungo il fronte interno delle mura e al di sotto del livello della risega di fondazione ha messo in evidenza alcune masse di calcestruzzo5 (fig. 5), apparse da subito dotate di un’eccezionale durezza e compattezza, costituite da calce di colore molto chiaro e puro (pressoché bianco) e aggregati costituiti da ciottoli, schegge lapidee, ghiaia e frammenti fittili di dimensioni tendenzialmente inferiori ai 10 cm. Questi grandi blocchi di forma irregolare presentano dimensioni eccezionali con lunghezze superiori ad 1,5 m, ma la loro reale estensione in relazione al muro non è stata verificata a causa delle difficoltà logistiche incontrate nello scavo. È certo che questi blocchi sono presenti a ridosso dell’angolo che caratterizza la cortina nel punto del suo allargamento presso la torre, ma non è certo se tali blocchi siano presenti anche nei settori limitrofi del muro o addirittura per tutta l’estensione del perimetro difensivo; certamente però essi si estendono anche oltre il limite dello scavo condotto nel 2011 e 2012. Incerto è rimasto anche il loro spessore in profondità, come incerto, d’altronde, è rimasto lo sviluppo in profondità delle fondazioni stesse del muro, immerse nei fanghi acquitrinosi. L’unico dato certo e sicuro è che questi grandi ammassi 5  Si usa in questa sede la terminologia di Ginouvés e Martin 1985: 51, ove con calcestruzzo si intende una miscela preparata prima del suo impiego in opera (a differenza dell’opera cementizia) con acqua, legante e aggregati diversi di dimensioni superiori ai 5-8 mm.

33

Fig. 5. Aquileia, fondi Cossar. In primo piano è visibile uno dei blocchi di calcestruzzo idraulico individuato nel corso dello scavo del 2012 a ridosso della fronte del muro e della sua fondazione. La presenza dell’acqua di falda in continua risalita ha reso estremamente complessa la pulitura e l’esatta perimetrazione di questi elementi di rinforzo della struttura (foto autore).

cementizi risultavano addossati alla fronte del muro per una parte della struttura sottoposta alla risega di fondazione. Le difficoltà incontrate nello scavo e nel rilievo hanno quindi reso in parte difficile la ricostruzione dell’assetto stratigrafico e del procedimento costruttivo di questo tratto delle difese (fig. 6); sulla base di quanto notato nello scavo a­ ppare probabile che la realizzazione dell’imponente opera difensiva sia proceduta con l’iniziale scavo di una grande fossa di fondazione, forse in qualche modo protetta dalle infiltrazioni d’acqua; entro tale fossa dovette quindi essere costituita la fondazione del muro in laterizi e quindi dovettero essere realizzate le grandi masse di calcestruzzo tra le pa­reti delle fosse di fondazione e la base del muro. Dopo il processo di tiraggio e presa, queste masse di diverse tonnellate di peso poterono svolgere una funzione di rinforzo costruttivo per evitare cedimenti laterali del muro nei punti di maggiore sensibilità statica e rilievo defensionale come gli angoli e le torri del perimetro. L’impossibilità di raggiungere in profondità la base delle fondazioni e in estensione i limiti delle fosse di fondazione ha lasciato però aperta anche la teorica possibilità, seppur decisamente meno probabile anche per ragioni di statica, che i blocchi di cementizio siano stati addossati al muro non in fase di costruzione ma in un momento imprecisato per consolidare o rinforzare il muro. Sia che i blocchi rientrino nei piani della costruzione del muro sia che facciano parte di un intervento di consolidamento successivo, essi tro-

34

JACOPO BONETTO, GILBERTO ARTIOLI, MICHELE SECCO, ANNA ADDIS

Anejos de AEspA LXXVII

Fig. 6. Aquileia, fondi Cossar. Schema ricostruttivo della struttura delle mura repubblicane nel tratto identificato presso lo scavo del 2012. Permangono dubbi sulla relazione tra i blocchi di calcestruzzo e il muro nei livelli profondi della fondazione (elaborazione dell’autore).

vano ragione nel particolarissimo contesto ambientale in cui venne fondata la città ed eretto il muro. Questo venne infatti realizzato all’interno di un suolo umido e instabile, caratterizzato quasi certamente anche in antico6 dalla penetrazione di livelli di acqua di risorgiva, dotato di bassa portanza e quindi instabile e certamente poco adatto a sostenere costruzioni imponenti come le mura. Il difficile quadro ambientale era ben noto e sottolineato già dagli autori antichi, come Vitruvio, che ricorda espressamente le moenia di Aquileia e 6  Uno dei problemi più difficili da affrontare nel caso urbano di Aquileia è quello dell’oscillazione del livello della falda tra l’età antica e l’epoca moderna. Appare addirittura assai diverso il livello della falda quale si evince dalle foto degli inizi del secolo scorso ad adesso ed è quindi decisamente complesso provare a determinare la quota delle acque sotterranee per l’età antica. Non è da mettere in dubbio però che l’area della città rimase sempre e comunque caratterizzata da assetto umido e instabile sul piano statico per la presenza (accertata da recentissimi carotaggi) di livelli di limi, argille e sabbie permeati d’acqua fino a notevole profondità dal piano d’uso antico.

di altre citta del litorale adriatico in paludibus constituta (Vitr. 1, 4, 11).7 Questi caratteri umidi e insicuri dei suoli d’Aquileia spiegano quasi certamente anche la soluzione adottata lungo il lato occidentale delle mura in un’area particolarmente depressa nota come le Marignane; qui infatti il muro in laterizio venne costruito su una base in pietra d’Istria che evitava la risalita dell’acqua (Bonetto 2004: 158-160). Lungo il lato orientale, interessato dai recenti scavi, fu scelta la tecnica di consolidamento del muro tramite le masse di calcestruzzo, ma la necessità di costruire in terreni anche qui invasi continuamente dalle acque suggerì ai costruttori l’adozione di un tipo di miscela legante del tutto particolare che apre nuovi scenari sul panorama tecnico-costruttivo e commerciale della colonia latina posta all’estremo nord del golfo adriatico. 7   Item si in paludibus moenia constituta erunt... Exemplar autem huius rei Gallicae paludes possunt esse, quae circum Altinum, Ravennam, Aquileiam, aliaque quae in eiusmodi locis municipia sunt proxima paludibus...

Anejos de AEspA LXXVII

L’uso delle polveri pozzolaniche nei grandi cantieri...

35

LE ANALISI ARCHEOMETRICHE DEL CALCESTRUZZO DELLE FONDAZIONI Il calcestruzzo è stato sottoposto ad indagini archeometriche comprendenti analisi petrografica attraverso microscopia ottica in luce polarizzata (OM), analisi della composizione mineralogica attraverso la tecnica della diffrattometria ai raggi X delle polveri (XRPD) ed analisi micro tessiturali e chimiche mediante microscopia elettronica a scansione (SEM) con associata microanalisi chimica a dispersione di energia (EDS). Tutte le analisi sono state eseguite presso i l­ aboratori del dipar­ timento di Geoscienze dell’Università di Padova. L’analisi petrografica eseguita su una sezione rappresentativa (fig. 7) del calcestruzzo ha evidenziato un rapporto standard legante/aggregato della miscela cementizia pari a 1/3. L’aggregato risulta scarsamente selezionato ed è caratterizzato da una distribuzione bimodale con dimensioni medie relative alle classi granulometriche delle ghiaie fini e delle sabbie medie. La sua morfologia è sub-arrotondata e la sfericità osservata è medio-bassa. Mineralogicamente l’aggregato è composto principalmente da rocce carbonatiche, per lo più grainstone bioclastici, e secondariamente da clasti di calcare micritico e quarzite. Nella matrice legante si osservano sia aree ad elevata birifrangenza dovute ad accumuli millimetrici di calcite che testimonia l’impiego di un legante calcico (Pecchioni et al. 2008), sia aree a più bassa birifrangenza che indicano la presenza di ceneri vulcaniche di dimensioni sub-micrometriche e costituite totalmente da vetri silicatici, associate ad allumino e silicati di calcio idrati riconducibili alla reazione pozzolanica tra la calce e i succitati materiali piroclastici. A supporto di

Fig. 7. Sezione sottile (30 μm) di una porzione rappresentativa del calcestruzzo. In evidenza due frammenti di aggregato piroclastico con comportamento pozzolanico (elaborazione degli autori).

questa ipotesi si sottolinea la presenza nella matrice legante di agglomerati sub-millimetrici giallastri di materiale piroclastico incoerente. L’analisi mineralogica delle fasi è stata eseguita su una porzione significativa del campione cementizio micronizzata fino all’ottenimento di una polvere sub-micrometrica. I diffrattogrammi sono stati ottenuti tramite diffrattometro PANalytical X’Pert PRO a geometria Bragg Brentano, dotato di rilevatore RTMS X’Celerator. Il calcestruzzo è principalmente composto da calcite ascrivibile sia alla componente legante interessata da reazione di carbonatazione, sia all’aggregato. Le analisi incentrate sulla frazione legante (fig. 8) hanno evidenziato, oltre all’occorrenza di calcite, la presenza di un’abbondante frazione amorfa, identificabile tramite l’articolazione del fondo del pattern di diffrazione, associata a obermorite (Jackson et al. 2013), un silicato di calcio idrato del gruppo delle fasi C-S-H, ed emicarboalluminato di calcio, un alluminato idrato di calcio del gruppo delle fasi AFm (Matschei et al. 2007). Questi minerali,

Fig. 8. Diffrattogramma della frazione legante del calcestruzzo, in evidenza le fasi C-S-H e AFm (elaborazione degli autori).

36

JACOPO BONETTO, GILBERTO ARTIOLI, MICHELE SECCO, ANNA ADDIS

Anejos de AEspA LXXVII

Fig. 9. Immagine in elettroni retrodiffusi al microscopio elettronico a scansione di un aggregato piroclastico con comportamento pozzolanico. Il clasto è composto principalmente da un vetro ricco in silicio (punto analisi in rosso e relativa microanalisi EDS), è caratterizzato da aree a bordo grano ricche in fasi C-S-H e Afm (punto analisi in verde e relativa microanalisi EDS) ed è circondato da una matrice legante composta da carbonato di calcio (punto analisi blu e relativa microanalisi EDS) (elaborazione degli autori).

che fanno parte dei costituenti dei cementi moderni, si formano a seguito di reazioni idrauliche dovute all’interazione tra materiali pozzolanici ricchi in silice amorfa e calci idrate. La natura pozzolanica della miscela utilizzata per la realizzazione del legante messa in luce dalla analisi petrografica è stata confermata dalle analisi mineralogiche. Le analisi SEM-EDS, effettuate mediante microscopio elettronico a scansione CamScan ­ MX3000 con sorgente LaB6, confermano l’ete­ rogeneità delle matrici leganti, costituite da aree ricche in calcio identificabili come accumuli di carbonato di calcio, associate ad aree ad alto contenuto di calcio, silicio ed alluminio le quali confermano l’occorrenza di polveri vulcaniche finemente disperse nel materiale ed interessate da reazione pozzolanica con la calce idrata. Tali reazioni si sviluppano talora ai bordi degli agglomerati piroclastici (fig. 9), composti prevalentemente da vetro ricco in silice, i quali risultano ricchi di fasi C-S-H e AFm nelle porzioni interfacciali. Ulteriori analisi chimiche, soprattutto di elementi in

traccia, sono attualmente in corso al fine di identificare l’area di provenienza dei materiali idraulicizzanti utilizzati negli impasti; si può comunque affermare sulla base dei dati disponibili che non ci sono rocce paragonabili a quelle individuate in Italia settentrionale a nord dell’Appennino e che quindi la provenienza del materiale individuato nelle malte aquileiesi è con buona probabilità centro italica o sud-italica. DATI E PROBLEMI SULL’USO DELLE POLVERI VULCANICHE I risultati delle analisi ora esposti hanno un’eccezionale importanza per la storia della tecnologia costruttiva romana per diverse ragioni. L’uso documentato ad Aquileia di polveri vulcaniche costituisce infatti un possibile interessante esempio dell’impiego di materiali8 dai distretti 8   In questa sede non può essere preso in considerazione un importante problema che riguarda la precisa provenienza delle polveri vulcaniche usate in antico a fini edilizi. Infatti in quasi

Anejos de AEspA LXXVII

L’uso delle polveri pozzolaniche nei grandi cantieri...

vulcanici di Campania e Lazio9 che per le loro proprietà chimico-fisiche, risultavano adatti per essere combinati alla calce aerea e produrre leganti e malte o calcestruzzi idraulici. Le riconosciute caratteristiche fisico-meccaniche di queste miscele (Jackson 2014) sono descritte in un ben noto testo di Vitruvio (Vitr. 2, 6, 1-6), che tratta delle “mirabolanti” proprietà delle polveri raccolte attorno a Baia e al Vesuvio in grado di fare presa anche in assenza di ossigeno (e quindi in acqua), e sono più tardi menzionate ancora da altri autori (Brandon et al. 2014: 11-36), che talvolta menzionano esplicitamente l’esportazione e l’uso della pozzolana (come Plin. nat. 16, 201-202). Le possibilità di un uso tanto particolare le rendevano indispensabili in alcuni particolari contesti e ne garantirono una diffusione in specifiche aree o cantieri del mondo antico dove erano realizzate opere edilizie in suoli umidi o a diretto contatto con l’acqua. Tale diffusione è stata l’oggetto di numerosi studi (Brandon et al. 2014: 4-5) e di un ampio progetto di ricerca, denominato ROMACONS, i cui risultati sono stati recentemente pubblicati in forma estensiva (Brandon et al. 2014; Gazda 2011). In questo studio sono esatutti gli studi condotti il rinvenimento di tale materiale allo stato grezzo o all’interno di malte/calcestruzzi con reazioni idrauliche ha fatto ritenere troppo spesso certa (o scontata) la loro provenienza dall’area flegrea attorno al porto di Pozzuoli, anche in assenza di analisi archeometriche o sulla base di risultati di analisi non decisive. Le ricerche archeologiche e archeometriche hanno effettivamente dimostrato che le polveri campane ebbero indubbiamente una diffusione e notorietà massime, ma hanno anche chiarito che non sempre è possibile indicare con certezza l’afferenza dei materiali a questa zona geografica ed esistono concrete possibilità che anche altre zone del mondo antico abbiano fornito materiali dalle proprietà simili e abbiano costituito importanti bacini di approvvigionamento per usi locali o anche per l’esportazione. Tra queste è noto con certezza il distretto dei Colli Albani (su cui vedi la nota seguente), mentre altre aree di produzione (per usi locali) sono state identificate in Germania (Lamprecht 1989: 84 e Lamprecht 1996: 61, 75, 87 per edifici di Colonia dove sembra usata una “pozzolana” di Renania, definita Trass) e in Turchia (Brandon et al. 2014: 224-225); altre ancora sono suggerite e discusse (isole egee come Melos e soprattutto Santorini: Siddal 2000: 342-343 e Brandon et al. 2014: 3 con bibliografia prec.; Volsinii: Marra e D’Ambrosio 2013). Lo stesso Vitruvio parla della produzione in area etrusca di un materiale (Vitr. 2, 6, 6) eccellente in structuris... in terrenis aedificiis come il carbunculus, la cui origine è strettamente associata a quella delle polveri vulcaniche e che viene associata negli ultimi studi alla regione dei Monti Sabatini e del distretto vulcanico di Vico (D’Ambrosio et al. 2015: 198). 9   L’area laziale è connotata dalla presenza di due grandi distretti vulcanici rappresentati dai Monti Sabatini a nord della città e dei Monti Albani a sud della città. Su questi due comparti e sul materiale da essi prelevato per le opere architettoniche romane sono fondamentali gli studi di: Jackson e Marra 2006; Jackson et al. 2007; Jackson et al. 2010; Marra et al. 2011; Jackson et al. 2014; Belfiore et al. 2014.

37

minati con grande attenzione i casi di opere marittime (porti, piscine, peschiere, etc.) realizzate con leganti idraulici in varie parti del Mediterraneo, e ne è stata documentata la grande diffusione dal Portogallo all’Asia Minore con una specifica concentrazione (65% dei casi) lungo le coste dell’Italia tirrenica, tra la Toscana e la Campania (Brandon et al. 2014: 123-140). Le grandi realizzazioni di opere marittime hanno attirato in forma prevalente l’attenzione degli studiosi che si sono occupati dell’uso delle polveri vulcaniche nell’edilizia antica, sia per l’effettiva straordinarietà delle opere compiute con esse in ambito marino, sia per l’enfasi che il noto testo vitruviano poneva nella particolarissima e miracolosa proprietà del materiale in grado di fare presa sott’acqua (Vitr. 2, 6, 1: ... sed etiam moles cum struuntur in mari, sub aqua solidescunt). Minore interesse catalogico o analitico ha invece suscitato il pur documentato impiego delle polveri pozzolaniche in costruzioni realizzate in ambienti terrestri, la cui diffusione, densità e cronologia restano sostanzialmente mal noti. Fanno eccezione le casistiche ben note di ambito urbano o laziale, dove ricerche e studi condotti da diversi gruppi hanno dimostrato il frequente impiego in età imperiale romana di leganti addizionati con polveri vulcaniche provenienti prevalentemente dai distretti vulcanici dei Colli Albani e Sabatini. Esemplari in tal senso sono gli studi sui Mercati di Traiano a Roma (Jackson et al. 2010; Jackson et al. 2014), sulla villa dei Quintili (Belfiore et al. 2014)10 o su alcuni altri complessi del centro cittadino (Marra et al. 2015), dove la presenza delle malte pozzolaniche appare strettamente legata, da un lato, alla prossimità dei centri di approvvigionamento delle polveri e, dall’altro, alla se­conda delle caratteristiche tipiche delle malte idrauliche costituita dalla particolare resistenza meccanica. Dal punto di vista geografico l’uso delle polveri vulcaniche per la realizzazione di leganti idraulici è noto in un numero di casi decisamente maggiore per le aree mediterranee e centro-italiche, mentre sembrano molto poco numerosi gli esempi di impiego al di fuori di queste aree e nelle regioni centro-settentrionali d’Europa (Gazda 2011). Isolate attestazioni sono documentate nei centri di Marsiglia (porto) e di Frejus (porto e ter10  Lo studio ha messo a confronto con attenta logica interdisciplinare lo studio di 38 campioni di malta prelevati dalle strutture della villa dei Quintili con campioni di sabbie di cava provenienti dai monti Albani e afferenti ai generi della Pozzolana Rossa, Pozzolana Nera e Pozzolanella.

38

JACOPO BONETTO, GILBERTO ARTIOLI, MICHELE SECCO, ANNA ADDIS

me) nella Francia meridionale (Excoffon e Dubar 2011: 178 e passim), dove sono usati tufi vulcanici sia in forma di polvere sia in forma di aggregati dell’opera cementizia. Non si può tuttavia escludere che questa ridotta documentazione di altri esempi in aree dell’Italia settentrionale ed europee sia dovuta ad una modesta abitudine alla verifica archeometrica dei leganti in aree così distanti dai bacini laziali e campani. L’USO DELLE POLVERI VULCANICHE E DEI LEGANTI IDRAULICI NEI CANTIERI DELLA CISALPINA REPUBBLICANA In questo quadro di almeno apparente (se non reale) modesta diffusione delle polveri vulcaniche al di fuori della penisola italiana o delle aree mediterranee per costruzioni marittime e addirittura al di fuori dell’ambito laziale per costruzioni su terra, il caso delle masse cementizie con leganti pozzolanici riscontrati presso le mura di Aquileia assume dunque particolare importanza documentaria e costituisce ad oggi l’attestazione più settentrionale nel quadro geografico mediterraneo ed europeo. I dati di Aquileia acquistano ancor maggiore rilievo storico in quanto possono essere affiancati ad indicazioni parallele provenienti da un altro centro della Cisalpina romana come Ravenna (fig. 1), così da apparire indizio di un fenomeno forse più diffuso di quanto pensato e di particolare significanza. In questo secondo centro costiero dell’Adriatico le mura realizzate dai Romani alla fine del iii sec. a.C., indagate negli anni Ottanta del secolo scorso (fig. 10), presentano infatti caratteri molto simili a quelle di Aquileia per diversi aspetti (Manzelli 2000; Manzelli 2010; Manzelli 2015): anche in questo caso la cortina (spessore 2,3 m) è realizzata con tecnica omogenea attraverso laterizio cotto per la sua intera estensione e le misure dei mattoni (0,49 × 0,49 × 0,05 e 0,52 × × 0,52 × 0,05 m: triemiplinthos dorico o, in alternativa, triemiplinthos samio) rimandano anche in questo caso inequivocabilmente alla presenza di maestranze greche (Manzelli 2010 e Manzelli 2015).11 A differenza di quanto avviene ad Aquileia, dove la presenza del legante tra mattoni è 11  Indiziate anche dalla presenza di alcune lettere probabilmente greche, di un numerale certamente greco (Phi = 500) e di alcune abbreviazioni tachigrafiche, ricorrenti in ambito ellenico, incise a crudo sui mattoni.

Anejos de AEspA LXXVII

pressoché nulla, a Ravenna gli elementi fittili sono legati da malta tenacissima su cui è stata eseguita un’analisi tramite diffrattometria a raggi X e microscopia elettronica. I primi risultati, pur editi in forma molto sintetica, hanno portato ad affermare che le malte vennero “prodotte con un legante idraulico costituito da una miscela di calce aerea e di pozzolana”. Secondo le osservazioni eseguite, nei campioni “compare anche l’augite, minerale presente in rocce vulcaniche e tipicamente nei materiali pozzolanici soprattutto dell’area laziale” (Costa et al. 2000: 28; Manzelli 2000: 74; Manzelli 2010: 158 e nota 4).12 Nonostante non esistano al momento maggiori informazioni e report analitici su questo fondamentale caso di studio, sembra comunque evidente che anche nel caso delle fortificazioni ravennati si prestò particolare attenzione alla qualità dei leganti attraverso l’aggiunta di materiale idraulicizzante non locale. In sintesi sembra quindi che in entrambi i centri romani dell’arco adriatico (Aquileia e Ravenna) la costruzione delle fortificazioni avvenga con importazione di materiale vulcanico “pozzolanico” dall’Italia centro-meridionale per la costituzione dei complessi monumentali che rendevano sicuri i caposaldi militari dell’avanzata dell’esercito e della potenza di Roma verso l’Europa continentale. Questo quadro ricostruttivo offre molti dettagli da sottolineare. Una prima annotazione riguarda la natura del materiale trasferito in Italia settentrionale a fini edilizi. Gli studi fino ad ora condotti (e forse il caso qui presentato di Aquileia) potevano suggerire che l’esportazione su lunga distanza di polveri pozzolaniche interessasse solo le piroclastiti prodotte attorno all’area flegrea. Ma le analisi di Ravenna, se confermate, dimostrano invece che anche i distretti estrattivi laziali (Jackson et al. 2005; Jackson e Marra 2006; Conticelli et al. 2010; Jackson et al. 2010; Belfiore et al. 2014), che erano fino ad ora noti come bacini di approvvigionamento per i soli cantieri di area romana (Brandon et al. 2014: 226; Jackson et al. 2014; Belfiore et al. 2014), costituiva invece un’area geologica da cui il materiale era trasferito anche verso cantieri a grande distanza. 12  Le analisi sono state condotte presso i laboratori dell’Italcementi di Bergamo. La genericità del dato fornito in seguito al percorso analitico rende non del tutto provata la provenienza laziale del materiale vulcanico presente in queste malte e ha spinto ad avviare un nuovo piano di analisi sui leganti di questo complesso architettonico.

Anejos de AEspA LXXVII

L’uso delle polveri pozzolaniche nei grandi cantieri...

Inoltre appare dimostrato e assai significativo che il trasporto del materiale vulcanico era ben organizzato e ricorrente anche per destinazioni, come quelle dell’Italia settentrionale, che distavano centinaia di chilometri dai luoghi di approvvigionamento. Per raggiungere Aquileia e Ravenna dalle aree vulcaniche laziali e/o campane potevano essere scelte due vie, entrambe estremamente onerose: la prima poteva prevedere una difficile circumnavigazione dell’Italia e la risalita per l’Adriatico con un tragitto estremamente lungo (circa 1.000 miglia marine) e complesso; non si può escludere che il trasporto navale fosse ridotto tramite il trasferimento del materiale via terra lungo la via Appia e dall’area flegrea a Brindisi per risalire quindi l’Adriatico e raggiungere Ravenna e Aquileia. Un’altra possibilità ancora poteva comportare un trasporto via terra del materiale dal versante occidentale a quello orientale della penisola, fino a Ravenna, e un successivo trasporto via acque interne lagunari da Ravenna ad Aquileia. Il percorso copriva una distanza di oltre 550 km via terra e di 110 miglia marine, risultando comunque particolarmente impegnativo per costi e manodopera impiegata. È ragionevole tuttavia pensare ad una preferenza per il trasporto marittimo secondo una prassi che sembra usuale, come sembra dimostrare l’impiego delle polveri pozzolaniche in varie parti del Mediterraneo.13 Inoltre il trasferimento marittimo in ambito italico è ben documentato dal ritrovamento di alcune imbarcazioni con carico di pozzolana nel porto di Pisa (Slayman 1999a; Slayman 1999b; Giachi e Pallecchi 2000: 350)14 e nel Tirreno settentrionale (Liou e Pomey 1985: 562-563).15 Va anche detto che, seppure non siano note le quantità di materiale impiegato nei due siti, tuttavia la notevole estensione delle mura in entrambi 13  Il progetto ROMACONS (Brandon et al. 2014) ha mostrato come il materiale di Pozzuoli fosse esportato in tutto il Mediterraneo per esigenze di carattere pubblico in ambito portuale. Materiale di Pozzuoli è attestato a Caesarea (Oleson e Brandon 2010), a Soloi-Pompeiopolis in Turchia (Stanislao et al. 2010), al porto di Chersonesos a Creta (lo studio del porto di Chersonesos e le connessioni con l’Italia è in Brandon et al. 2005). 14  Si veda però anche più recentemente Marra e D’Ambrosio 2013 e D’Ambrosio et al. 2015: 201 che hanno ripreso in considerazione il materiale ritenuto di provenienza flegrea e ne hanno fissato una probabile origine dall’area di Volsinii Novi. 15  Il relitto di Madrague de Giens al largo delle coste meridionali della Francia, conteneva diverse migliaia di anfore con polveri vulcaniche interpretate come pozzolana dell’area flegrea. Un altro relitto presso Marsiglia (“Chrétienne M”) trasportava polveri “pozzolaniche” secondo Joncheray e Joncheray 2002.

39

i casi e la presenza diffusa in tutto il perimetro di Aquileia di suoli umidi, tali da richiedere almeno in via teorica specifici leganti, rendono probabile un ampio utilizzo di materiale pozzolanico, che non avrebbe senso pensare localizzato od occa­ sionale; è quindi da ritenere probabile una massiccia importazione di varie migliaia di metri cubi di additivi per i due cantieri16 e, di conseguenza un’articolata organizzazione economico-produttiva per la fornitura del materiale, forse stimolata dal potere militare e statale così fortemente impegnato sul piano dell’avanzata verso le terre del nord. Questa complessità della gestione dei grandi progetti costruttivi coloniali si nota anche sul piano locale dell’organizzazione dei singoli cantieri. I dati sopra esposti suggeriscono infatti che le ­attività di costruzione delle mura urbane repubblicane dell’Italia settentrionale vedessero la compartecipazione di protagonisti diversi e ben integrati. Infatti ad Aquileia come a Ravenna è da ritenere cruciale il ruolo svolto dai quadri militari dell’esercito, che costituivano i soli in grado di sostenere sul piano logistico grandi operazioni di urbanistica; ma è evidente che all’interno o a fianco della compagine militare i cantieri dovettero vedere attivi in forma del tutto sinergica progettisti e maestranze greche o magnogreche (Bonetto 2015; Bonetto e Manzelli 2015) per gli aspetti progettuali e di produzione dei materiali fittili; una terza sinergica componente doveva però essere rappresentata da maestranze centro-italiche specializzate nell’impiego delle polveri pozzola­ niche,17 certamente collegate o addirittura coincidenti con i fornitori del materiale provenienti dalle regioni laziali o campane. Nell’uso delle polveri pozzolaniche a Ravenna e Aquileia l’aspetto però forse più ricco di spunti appare quello relativo alla cronologia di realizzazione dei complessi architettonici. L’età di cos­ truzione delle mura di Ravenna e l’impiego in esse dei leganti è infatti fissata con buona sicurezza tra la fine del iii sec. e l’inizio del ii sec. a.C. (Man16   Sul trasporto dei materiali vulcanici nel Mediterraneo vedi alcune indicazioni di Excoffon e Dubar 2011: 177-178. Nel caso del porto di Cesarea in Israele il calcolo delle quantità di materiale pozzolanico impiegato produce una stima di circa 17.600 mc (Votruba 2007). 17   L’esistenza di un forte legame con la penisola e l’ambito laziale è denotato dalla presenza nei livelli di cantiere più antichi delle mura di Ravenna di fittili di produzione volterrana e laziale giunti evidentemente in città con le avanguardie operative e tecniche romane: sul materiale recuperato nello scavo vedi Manzelli 2000: 10-11 e fig. 6.

40

JACOPO BONETTO, GILBERTO ARTIOLI, MICHELE SECCO, ANNA ADDIS

zelli 2000 e Manzelli 2010). Pur con qualche mo­ desto dubbio ancora da verificare, la data di impianto delle fortificazioni di Aquileia e la posa a ridosso delle stesse dei blocchi con pozzolana sembra da porsi pochi decenni dopo quelle della città romagnola e entro la metà del ii sec. a.C. Questo panorama stimola considerazioni storiche rilevanti: da un lato si può notare che l’impiego su larga scala di miscele essicanti (malte e calcestruzzi) avviene in Cisalpina nella stessa epoca in cui esse sono introdotte nei grandi cantieri dell’Urbe e della Campania (Adam 1988: 82-84; Rakob 1983; Lamprecht 1996: 138-150) o addirittura in leggero anticipo rispetto ad essi.18 Ma la presenza in un’età così antica delle polveri pozzolaniche all’interno delle miscele essicanti di Ravenna e Aquileia appare ancora più importante sul piano storico; fino ad ora gli studi hanno infatti pur documentato, con vari gradi di affidabilità, l’impiego di materiale vulcanico nelle malte di edifici dell’Urbe già dalla prima metà del ii sec. (e continuativamente nel corso del i sec. a.C.), ma è stato pure dimostrato che tali additivi sono derivati dagli stessi suoli vulcanici in cui i cantieri erano aperti19 senza ricorso ad importazione di materiale da altri distretti d’origine; lo stesso fenomeno si riscontra in vari altri contesti (Vulci, Volsinii Novi, Pompei e altri), dove pure il materiale vulcanico usato nei leganti tra iii e i sec. a.C. proviene da bacini di approvvigionamento locali  secondo il “naturale” principio dell’identità tra luogo di provenienza e luogo di impiego del materiale costruttivo (D’Ambrosio et al. 2015: 201 e passim; Miriello et al. 2015).20 18   Si veda l’importante contributo di Moggetta 2015 in cui la cronologia di diffusione del calcestruzzo e dell’opera cementizia in Italia centrale viene spostata ai decenni centrali del ii sec. a.C., in contrasto con le precedenti opinioni che ne vedevano un avvio di impiego già in epoche più antiche. 19   Fondamentale appare in questo senso il contributo di Marra et al. 2015, che prende in esame 14 campioni provenienti dai complessi della Porticus Aemilia, del Tempio della Concordia, del Tempio dei Dioscuri, del Tempio B e di altre strutture di Largo Argentina. In questo studio (vedi in particolare le sintesi di pp. 15-19) viene notato come negli edifici costruiti fino al grande incendio del 111 a.C. (Tempio di Saturno, Porticus Aemilia, Tempio della Concordia, Tempio dei Dioscuri) sono utilizzati sedimenti vulcanici locali ricavati dalla lavorazione del materiale usato per la muratura (Tufo del Palatino e Tufo Lionato) o raccolti nelle aree finitime ai cantieri, mentre successivamente a tale data, in corrispondenza alle grandi ristrutturazioni urbanistiche, si inizia ad impiegare la Pozzolana Rossa cavata pur sempre nell’ambito dell’area urbana e suburbana di Roma. 20  Secondo Brandon et al. 2014: 2-3, “the large-scale production of pozzolanic mortars for applications in watersaturated environments, however, began at some point in the third or second century BC, most likely in the landscape of

Anejos de AEspA LXXVII

Invece l’esportazione su medie o lunghe distanze di tali polveri e degli aggregati lapidei vulcanici appare fenomeno ben più raro o del tutto sconosciuto per questi orizzonti cronologici: sebbene non si disponga oggi di uno studio dettagliato in merito, tale pratica del commercio delle pozzolane potrebbe essersi diffusa dall’inizio del i  sec. a.C.21 e in forma massiccia a partire dall’età augustea.22 I dati di Ravenna e Aquileia potrebbero dimostrare quindi sia l’inizio di un largo impiego delle polveri pozzolaniche in un’epoca decisamente più antica rispetto a quanto fino ad ora creduto sia il loro trasferimento in contesti geografici esterni a quelli di origine già in un’epoca tanto precoce. Naturalmente tali considerazioni vanno ritenute ancora preliminari e ipotetiche, perché è possibile che la mancanza di estese ­campagne di indagini archeometriche sui leganti adoperati in edifici terrestri in Italia centrale (o in altre regioni) non abbia ancora rivelato un uso  ugualmente precoce ed esteso delle miscele idrauliche. Comunque sia, allo stato attuale della ricerca  questi dati costituiscono un punto di rife­ rimento decisamente nuovo nelle valutazioni sull’epoca di diffusione delle miscele essicanti con speciali componenti idraulicizzanti nel mondo romano. Oltre all’aspetto cronologico i dati qui discussi offrono un ulteriore motivo di riflessione in rapporto al contesto di frontiera (Ravenna), colo­ niale (Aquileia) e per certi aspetti “periferico” (rispetto al quadro dell’Italia centrale) in cui sono documentati i precoci usi di malte e calcestruzzi pozzolanici. In base a quanto si è detto sembra infatti che nel quadro delle grandi imprese di militarizzazione e urbanizzazione delle regioni a nord degli Appennini l’apparato tecnico dello Campi Flegrei volcanic district...”, ma tale affermazione non è supportata da alcun riferimento archeologico datato. M. Blake (Blake 1947: 346) data l’inizio della costruzione del porto di Baia al 199 a.C., ma anche in questo caso senza supporti di cronologia sicuri. 21  Un documento preciso è rappresentato dal carico del relitto di Madrague de Giens, rinvenuto al largo delle coste francesi, che comprendeva anche 6-7000 anfore contenenti pozzolana vulcanica, datato tra il 70 e il 65 a.C. (Liou e Pomey 1985: 562-563). 22   Un panorama molto ricco sui tempi e sui modi della diffusione delle polveri pozzolaniche si trova nel recente Brandon et al. 2014: 223-226 (R. L. Hohlfelder). Alcune annotazioni cronologiche sono anche in Excoffon e Dubar 2011: 177-178 e in Jackson et al. 2014: 1; così brevi indicazioni sono date da Belfiore et al. 2014: 1; tutti questi studi indicano in varie forme il i sec. a.C. come momento di grande diffusione delle polveri vulcaniche a fini edilizi.

Anejos de AEspA LXXVII

L’uso delle polveri pozzolaniche nei grandi cantieri...

Stato e dell’esercito abbiano avviato una serie variegata e assai innovativa di sperimentazioni che forse non avevano ancora trovato campo d’applicazione e di prova in contesti comprensibilmente più conservatori quali quelli dei consolidati contesti edilizi centro-italici. Diversi fattori possono aver influito su questa evidente, marcata spinta “in avanti” verso nuove forme del costruire che sembrano trasformare la Cisalpina del ii sec. a.C. in un vero laboratorio dell’edilizia romana. In primo luogo l’intervento massiccio di militari e coloni è attuato in siti e aree frequentemente libere da ogni preesistenza o connotate da presenze architettonico-insediative “deboli” sul piano tecnicostrutturale, dove venne dispiegata senza condi­ zionamenti particolari ogni possibile azione costruttiva. Le opere di urbanizzazione e di infrastrutturazione territoriale si dovettero inoltre confrontare con contesti ambientali sconosciuti ad operatori italici per materie prime disponibili, natura dei suoli e condizioni ambientali che imposero lo sviluppo di nuove tecnologie e la loro rapida adozione. In questo campo è sufficiente pensare alle infinite disponibilità di argilla per lo sviluppo dell’industria laterizia o alla presenza ubiquitaria di acque subaeree e sotterranee quali stimoli all’impiego di malte adeguate ad ambienti umidi come sono le pozzolane di cui si è discusso. Ma l’orizzonte regionale del periodo in cui queste novità tecnologiche si inseriscono è anche più ampio e spiega nel suo insieme il fenomeno osservato. Non va infatti trascurata l’esistenza di una cultura edilizia locale (celtica e veneta) che già aveva messo a frutto sistemi e tecniche adeguate allo specifico contesto e da cui i costruttori romani utilmente assunsero nozioni e capacità; su questo punto è sufficiente fare riferimento alle costruzioni in legno per contrastare i suoli umidi e acquitrinosi.23 E più importante ancora appare il fecondo incontro in Cisalpina tra i nuovi progettisti romani e gruppi di architetti, artigiani e maestranze di origine magnogreca e greca che per antica consuetudine e nuove speranze risalivano il canale adriatico e portavano in pianura padana una tradizione costruttiva (oltre che artistica e artigianale) ellenistica di eccelsa qualità, che introduce novità straordinarie come il laterizio cotto e altre pratiche costruttive sconosciute allora in ambito romano. È proprio l’incontro di queste tre tradizioni tecnico-architettoniche diverse ed im23  Vedi su questo il contributo di C. Previato in questo volume.

41

portanti, come quella celto-veneta, quella greca e quella romana, fuse in un contesto ambientale diversificato e ricchissimo di materia prime, che fanno della Cisalpina del ii sec. a.C. un teatro multiculturale di grandi innovazioni per l’architettura romana e, probabilmente, per nulla subalterno o dipendente sul piano delle innovazioni costruttive dal contesto italico. Come sembrano dimostrare gli emblematici casi dell’uso delle polveri vulcaniche e dell’introduzione del laterizio cotto, la Cisalpina non conosce così, almeno nel campo dell’architettura, un passivo processo di romanizzazione inteso secondo superati canoni concettuali, ma è invece sede d’avanguardia della sperimentazione di alcune tecniche che contemporaneamente o più tardi troviamo impiegate con sicurezza nel cuore dello Stato romano. Su questo quadro geografico, cronologico ed operativo è forse possibile aggiungere qualche riflessione ancora grazie ai risultati delle analisi archeometriche condotte sul calcestruzzo usato nelle fondazioni delle mura aquileiesi. L’osservazione microscopica ha potuto infatti evidenziare due fatti legati a dettagli importanti delle pratiche di cantiere: la miscela di calce aerea e di polveri pozzolaniche è costituita da un amalgama omogeneo ottenuto con accurata e voluta combinazione degli elementi prima dell’aggiunta dell’acqua. Inoltre le procedure sono ancora meglio precisate grazie all’analisi in microscopia del rapporto percentuale tra la quantità di calce aerea e di polveri  pozzolaniche che fornisce valori pari a circa 30:70; questa relazione volumetrica appare molto vicina al rapporto di 1:2 prescritto da Vitruvio oltre cent’anni dopo proprio per la miscela delle calci con il pulvis puteolanus nelle malte idrauliche finalizzate alle opere marittime (Vitr. 5, 12, 2).24 Questi dettagli fanno intuire che lo sperimentalismo e la spinta all’innovazione costruttiva dei grandi progetti urbani della Cisalpina repubblicana dovettero maturare precocemente già nel corso della piena età repubblicana, generando prassi cantieristiche che si diffonderanno fino ad essere codificate nei manuali di architettura dell’ormai consolidata prassi costruttiva augustea.

24   Eae autem structurae, quae in aqua sunt futurae, videntur sic esse faciendae, uti portetur pulvis a regionibus, quae sunt a Cumis continuatae ad promunturium Minervae, isque misceatur, uti in mortario duo ad unum respondeant. Sui rapporti tra polveri pozzolaniche e calci e sui rapporti tra leganti e aggregati nelle malte antiche vedi le note di Jackson 2014: 160163. Inoltre vedi il contributo di J. Wehby Murgatroyd in questo volume.

42

JACOPO BONETTO, GILBERTO ARTIOLI, MICHELE SECCO, ANNA ADDIS

In sintesi, seppur è vero che non si possiedono ancora dati sufficienti per affermare che le applicazioni di certe tecniche e materiali ben noti in Italia centrale per la tarda repubblica siano state introdotte a partire dalle sperimentazioni attuate nel nord della penisola e nel quadro della colonizzazione cisalpina, è forse almeno il tempo di osservare il grande mondo dei sistemi edilizi di epoca romana come un universo policentrico (e non romanocentrico) nel quale varie componenti culturali e geografiche hanno contribuito in forme diverse, sinergiche e sincroniche alla progressiva costituzione ed evoluzione del linguaggio tecnico dell’architettura, ognuna generando e distribuendo propri fondamentali apporti. BIBLIOGRAFIA Adam, J.-P. 1988: L’arte di costruire presso i Romani. Materiali e tecniche. Longanesi, Milano. Belfiore, C. M., Fichera, G. V., La Russa, M. F., Pezzino, A., Ruffolo, S. A., Galli, G. e Barca, D. 2014: “Multidisciplinary Approach for the Archaeometric Study of Pozzolanic Aggregate in Roman Mortars. The Case of Villa dei Quintili (Rome, Italy)”, Archaeometry, c.s. Blake, M. E. 1947: Ancient Roman Construction in Italy from the Prehistoric Period to Augustus. A Chronological Study Based in Part upon the Material Accumulated by the Late Dr. Esther Boise van Deman. Carnegie Institution, Washington D. C. Bonetto, J. 1998: Mura e città nella Transpadana romana. Fondazione Antonio Colluto, Portogruaro. Bonetto, J. 2004: “Difendere Aquileia, città di frontiera”, in Cuscito, G. e Verzár-Bass, M. (a cura di), Aquileia dalle origini alla costituzione del ducato longobardo. Topografia – urbanistica – edilizia pubblica. Atti della XXXIV Settimana di studi aquileiesi (Aquileia, 8-10 maggio 2003), pp. 151-196, Antichità Altoadriatiche 59. Editreg, Trieste. Bonetto, J. 2009: “Le mura della città”, in Ghedini, F., Novello, M. e Bueno, M. (a cura di), Moenibus et portu celeberrima. Aquileia: storia di una città, pp. 83-92. Libreria dello Stato, Roma. Bonetto, J. 2015: “Diffusione ed uso del mattone ­cotto nella Cisalpina romana tra ellenizzazione e romanizzazione”, in Bukowiecki, E., Volpe, R. e Wulf-Rheidt, U. (a cura di), Il laterizio nei cantieri  imperiali. Roma e il Mediterraneo, Atti del I Workshop “Laterizio” (Roma, 27-28 novembre ­ 2014), Archeologia dell’architettura, XX, pp. 105113.

Anejos de AEspA LXXVII

Bonetto, J. e Salvadori, M. 2012 (a cura di): L’architettura privata ad Aquileia in età romana. Atti del Convegno nazionale (Padova, 21-22 febbraio 2011), Quaderni di Antenor 24. Padova University Press, Padova. Bonetto, J. e Pajaro, G. 2012: “Le fortificazioni repubblicane. Area I, saggio 1”, in Bonetto, J. e Ghiotto, A. R. (a cura di), Aquileia. Fondi ex Cossar. Rapporto 2012, pp. 12-19, Antenor quaderni 24. Padova University Press, Padova. Bonetto, J. e Ghiotto, A. R. (a cura di) 2014: Aquileia  – Fondi ex Cossar. Missione archeologica 2013. Università degli Studi di Padova, Padova. Bonetto, J. e Ghedini, F. 2014: “Vitruvio ad Aquileia. La casa ad atrio dei fondi ex Cossar”, in Clini, P. (a cura di), Vitruvio e l’archeologia. Tra norma e prassi. Atti del III Symposium di studi vitruviani (Fano, 8-11 novembre 2012), pp. 48-73. Marsilio, Venezia. Bonetto, J. e Manzelli, V. 2015: “Le mura repubblicane”, in Malnati, L. e Manzelli, V. (a cura di), Brixia. Roma e le Genti del Po. Un incontro di culture. iii-i sec. a.C. Catalogo della Mostra (Brescia, maggio 2015-gennaio 2016), pp. 153-154. Giunti, Firenze. Bonetto, J., Dobreva, D., Madrigali, E. e Centola V. c.s.: “Luisa Bertacchi ai fondi Cossar: innovazione e modernità”, in Salvadori, M. e Ventura, P. (a cura di), Luisa Bertacchi. Una vita per l’archeologia. Atti del Convegno di studio (Aquileia, 23-24 settembre 2011), Aquileia Nostra. Brandon, C. J., Hohlfelder, R. L., Jackson, M. D. e Oleson, J. P. 2014: Building for Eternity. The History and Technology of Roman Concrete Engineering in the Sea. Oxbow Books, Oxford. Brandon, Chr., Hohlfelder, R. L., Oleson, J. P. e Stern, Ch. 2005: “The Roman Maritime Concrete Study (ROMACONS). The harbor of Chersonios in Crete and its Italian connection. Étude du ciment hydraulique romain: le port de Chersonios (Crète)”, Méditerranée, 1.2, pp. 25-29. Brusin, G. 1932: “Scavi e loro assetto”, Aquileia Nostra, 3.2, pp. 135-142. Brusin, G. 1966: “Le difese della romana Aquileia e la loro cronologia”, in Corolla Memoriae Erich Swoboda dedicata, Römische Forschungen in Niederösterreich, 5, pp. 84-94. Centola, V., Furlan, G., Madrigali, E. e Previato, C. c.s., “La domus dei fondi ex-Cossar ad Aquileia: tradizione architettonica e innovazione tecnica”, Atti del XVIII Congreso Internacional de Arqueologia Clasica – Centro y periferia en el mundo clásico (Mérida, 13-17 maggio 2013), Mérida.

Anejos de AEspA LXXVII

L’uso delle polveri pozzolaniche nei grandi cantieri...

Chiabà, M. 2009: “Dalla fondazione all’età tetrarchica”, in Ghedini, F., Novello, M. e Bueno, M. (a cura di), Moenibus et portu celeberrima. Aquileia: storia di una città, pp. 7-22. Libreria dello Stato, Roma. Clementi, T., Rinaldi, F., Novello, M. e Bueno, M. 2009: “La produzione musiva”, in Ghedini, F., Novello, M. e Bueno, M. (a cura di), Moenibus et portu celeberrima. Aquileia: storia di una città, pp. 231252. Libreria dello Stato, Roma. Conticelli, S., Boari, E., Avanzinelli, R., De Benedetti, A. A., Giordano, G., Mattei, M., Melluso, L. e Morra, V. 2010: “Geochemistry, isotopes and mineral chemistry of the Colli Albani volcanic rocks: constraints on magma genesis and evolution”, in Funiciello, R., Giordano, G. (a cura di), The Colli Albani volcano, pp. 107-139, Special Publications of IAVCEI 3. The Geological Society, London. Costa, U., Gotti, E. e Tognon, G. 2000: “Nota tecnica: malte prelevate da mura antiche dallo scavo della Banca Popolare di Ravenna”, in Fortificazioni antiche in Italia. Età repubblicana, pp. 25-28, Atlante tematico di topografia antica 9. “L’Erma” di Bretschneider, Roma. D’Ambrosio, E., Marra, F., Cavallo, A., Gaeta, M. e Ventura, G. 2015: “Provenance materials for Vi­ truvius’ harenae fossiciae and pulvis puteolanis: geochemical signature and historical-archaeological implications”, Journal of Archaeological Science: Reports, 2, pp. 186-203. Excoffon, P. e Dubar, M. 2011: “L’emploi de tuf volcanique et de la pouzzolane dans quelques constructions de Forum Iulii (Fréjus, Var)”, Revue du Centre Archéologique du Var, pp. 171-181. Gazda, E. K. 2001: “Cosa’s contribution to the study of Roman hydraulic concrete. A historiographical commentary”, in Goldman, N. W. (a cura di), New Light from Ancient Cosa. Classical Mediterranean Studies in Honour of Cleo Rickman Fitch, pp. 145177. P. Lang, New York. Ghiotto, A. R. 2013: “Nuovi dati e nuove ipotesi sulla pianificazione urbana di Aquileia”, Rivista di Archeologia, 37, pp. 99-114. Giachi, G. e Pallecchi, P. 2000: “Analisi preliminari sui materiali”, in Bruni, S. (a cura di), Le navi antiche di Pisa ad un anno dall’inizio delle ricerche, pp. 348-351. Polistampa, Firenze. Ghedini, F. e Novello, M. 2009: “L’edilizia residenziale”, in Ghedini, F., Novello, M. e Bueno, M. (a cura di), Moenibus et portu celeberrima. Aquileia: storia di una città, pp. 111-125. Libreria dello Stato, Roma. Ginouvès, R. e Martin, R. 1985: Dictionnaire méthodique de l’architecture grecque et romaine, 1. Maté-

43

riaux, techniques de construction, techniques et formes du décor, CEFR 84.1. École française de Rome, Roma. Jackson, M. D. 2014: “Sea-water concretes and their material characteristics”, in Brandon, C. J., Hohlfelder, R. L., Jackson, M. D. e Oleson J. P., Building for Eternity. The History and Technology of Roman Concrete Engineering in the Sea, pp. 141187. Oxbow Books, Oxford. Jackson, M. D., Marra, F., Hay, R. L., Cawood, C. e Winkler, E. M. 2005: “The judicious selection and preservation of tuff and travertine building stone in ancient Rome”, Archaeometry, 47, pp. 485510. Jackson, M. D. e Marra, F. 2006; “Roman stone masonry: volcanic foundations of the ancient city”, American Journal of Archaeology, 110.3, pp. 403436. Jackson, M., Marra, F., Deocampo, D., Vella, A., Kosso, C. e Hay, R. 2007: “Geological observations of excavated sand (harenae fossiciae) used as fine aggregate in ancient Roman pozzolanic mortars”, Journal of Roman Archaeology, 20, pp. 1-30. Jackson, M., Deocampo, D., Marra, F. e Scheetz, B. 2010: “Mid-pleistocene pozzolanic volcanic ash in ancient Roman concretes”, Geoarchaeology. An International Journal, 25.1, pp. 36-74. Jackson, M. D., Chae, S. R., Mulcahy, S. R., Mera, C., Taylor, R., Li, P., Emwas, A., Moon, J., Yoon, S., Vola, G., Wenk, H. e Monteiro, P. J. M. 2013, “Unlocking the secrets of al-tobermorite in Roman seawater concrete”, American Mineralogist, 98, pp. 1669-1687. Jackson, M. D., Landis, E. N., Brune, P. F., Vitti, M., Chen, H., Li, Q., Kunz, M., Wenk, H.-R., Monteiro, P. J. M. e Ingraffea, A. R. 2014: “mechanical resilience and cementitious processes in imperial Roman architectural mortar”, in Proceedings of the National Academy of Sciences of the United States of America, published online before print December 15 (doi: 10.1073/pnas.1417456111). Joncheray, A. e Joncheray, J. P. 2002: “Chrétienne M, trois épaves distinctes, entre le cinquième siècle avant e le premier siècle après Jésus-Christ”, Cahiers d’Archéologie Subaquatique, 14, pp. 57-130. Lamprecht, H.-O. 1989: “Verwendung von Beton bei Wasserbauten in der Antike”, Mitteilungsblatt der Bundesanstalt für Wasserbau, 65, pp. 79-96. Lamprecht, H.-O. 1996: Opus caementicium. Bautechnik der Römer. Beton Verlag, Düsseldorf (5.a ed.). Liou, B. e Pomey, P. 1985: “Directions des recherches archéologiques sous-marines”, Gallia, 43, pp. 547576.

44

JACOPO BONETTO, GILBERTO ARTIOLI, MICHELE SECCO, ANNA ADDIS

Lugli, G. 1957: La tecnica edilizia romana con particolare riguardo a Roma e Lazio. G. Bardi, Roma. Manzelli, V. 2000: “Le mura di Ravenna repubblicana”, in Fortificazioni antiche in Italia. Età repubblicana, pp. 7-24, Atlante tematico di topografia antica 9. “L’Erma” di Bretschneider, Roma. Manzelli, V. 2010: “Le mura repubblicane”, in Bolzani, P. (a cura di), La Banca popolare di Ravenna. Storia, architettura, arte e archeologia (1885-2010), pp. 158-167. Ravenna. Manzelli, V. 2015: “Mattoni delle mura repubblicane”, in Malnati, L. e Manzelli, V. (a cura di), Brixia. Roma e le Genti del Po. Un incontro di culture. iii-i sec. a.C. Catalogo della Mostra (Brescia, maggio 2015-gennaio 2016), pp. 109-110. Giunti, Firenze. Marra, F., Deocampo, D., Jackson, M. D. e Ventura, G. 2011: “The Alban Hills and Monti Sabatini volcanic products used in ancient Roman masonry (Italy). An integrated stratigraphic, archaeological, environmental and geochemical approach”, Earth Science Reviews, 108, pp. 115-136. Marra, F. e D’Ambrosio, E. 2013: “Trace elements classification diagrams of pyroclastic rocks from the volcanic districts of central Italy: the Case Study of Ancient Roman Ships of Pisa”, Archaeometry, 55.6, pp. 993-1019. Marra, F., D’Ambrosio, E., Gaeta, M. e Mattei, M. 2015: “Petrochemical identification and insights on chronological employment of the volcanic aggregates used in ancient Roman mortars”, Archaeo­ metry (doi: 10.1111/arcm.12154). Matschei, T., Lothenbach, B. e Glasser, F. P. 2007: “The AFm phase in Portland Cement”, Cementicious Concrete Research, 37, pp. 118-130. Miriello, D., Barca, D., Bloise, A., Ciarallo, A., Crisci, G. M., De Rose, T., Gattuso, C., Gazineo, F. e La Russa, M. F. 2010: “Characterisation of archaeological mortars from Pompeii (Campania, Italy) and identification of construction phases by compositional data analysis”, Journal of Archaeological Science, 37.9, pp. 2207-2223. Moggetta, M. 2015: “A new date for concrete in Rome”, Journal of Roman Studies (doi:10.1017/ S007543581500043X). Oleson, J. P. e Brandon, G. 1992: “The technology of King Herod’s harbour”, in Vann, R. L. (a cura di), Caesarea Papers. Straton’s Tower, Herod’s Harbour,

Anejos de AEspA LXXVII

and Roman and Byzantine Caesarea, pp. 49-67, Journal of Roman Archaeology suppl. 5. Journal of Roman Archaeology, Ann Arbor, Mich. Orlandos, A. 1966: Les matériaux de construction et la technique architecturale des anciens grecs. Première partie. E. de Boccard, Paris. Pecchioni, E., Fratini. F. e Cantisani, E. 2008: Le malte antiche e moderne tra tradizione ed innovazione. Pàtron, Bologna. Previato, C. 2015: Aquileia. Materiali, forme e sistemi costruttivi dall’età repubblicana alla tarda età imperiale. Padova University Press, Padova. Rakob, F. 1983: “Opus caementicium – und die Folgen”, Römischen Mitteilungen, 90.2, pp. 359-372. Siddal, R. 2000: “The use of volcaniclastic material in Roman hydraulic concretes. A brief review”, in McGuire, W. J., Griffiths, D. R., Hancock, P. L. e Stewart, I. D. (a cura di), The Archaeology of Geological Catastrophes, pp. 339-344. The Geological Society, London. Slayman, A. L. 1999a: “Sunken ships of Pisa”, Archaeology, 52.3, p. 14. Slayman, A. L. 1999b: “A cache of vintage ships”, Archaeology, 52.4, pp. 36-39. Stanislao, C., Rispoli, C., Vola, G., Cappelletti, P., Morra, V. e De Genarro, M. 2011: “Contribution to the knowledge of ancient Roman seawater concretes. Phlegrean pozzolan adopted in the construction of the harbour at Soli-Pompeipolis (Mersin, Turkey)”, Periodico di Mineralogia, 80.3, pp. 471488. Tiussi, C. 2009: “L’impianto urbano”, in Ghedini, F., Novello, M. e Bueno, M. (a cura di), Moenibus et portu celeberrima. Aquileia: storia di una città, pp. 61-81. Libreria dello Stato, Roma. Villa, L. 2004: “Aquileia tra Goti, Bizantini e Longobardi: spunti per un’analisi delle trasformazioni urbane nella transizione fra tarda antichità e alto medioevo”, in Cuscito, G. e Verzár-Bass, M. (a cura di), Aquileia dalle origini alla costituzione del ducato longobardo. Topografia – urbanistica – edilizia pubblica. Atti della XXXIV settimana di Studi aquileiesi (Aquileia 2003), pp. 561-632, Antichità Altoadriatiche 59. Editreg, Trieste. Votruba, G. F. 2007: “Imported building materials of Sebastos Harbour, Israel”, The International Journal of Nautical Archaeology, 36.2, pp. 325-335.

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

LIME MORTAR PRODUCTION IN OSTIA: MATERIAL ANALYSIS OF MORTAR FROM THE HADRIANIC PERIOD JENNIFER WEHBY MURGATROYD RSK Environment Limited, Hemel Hempstead, United Kingdom

ABSTRACT: Choices made during mortar production, especially material selection, affect the performance and durability of the final product, a fact apparently known to ancient builders. Analyses of mortar from seven different Hadrianic-era structures in Ostia were undertaken to investigate such choices made by builders regarding aggregate selection and mortar mix designs. Samples were evaluated with petrographic analysis, including optical microscopy. Mortar components were quantified with modal analysis of thin sections (point counting), allowing an approximation of the original mix design for each sample. All samples comprised lime binder and volcanic tuff aggregate exhibiting varying degrees of natural alteration. The aggregate types present were ideal for fostering the pozzolanic reactions that have resulted in their durability. Patterns of aggregate lithology revealed three distinct mortar types within the group of seven structures, based on their unique aggregate profiles, suggesting specific mix designs were in use across the city at the same time. KEYWORDS: Aggregate lithology, Petrography, Optical microscopy, Point counting, Pozzolans, Tuff, Ancient builders. RESUMEN: Las elecciones efectuadas durante la producción de mortero, especialmente la selección de los materiales, condicionan el resultado y la durabilidad del producto final, un hecho aparentemente conocido por los antiguos constructores. En Ostia, se han llevado a cabo análisis de mortero sobre siete diferentes estructuras de época de Adriano para investigar este tipo de elecciones por parte de los constructores respecto a la selección de los agregados y la composición de la mezcla del mortero. Las muestras se evaluaron con análisis petrográfico, incluyendo microscopía óptica. Los componentes del mortero se cuantificaron con el análisis de láminas delgadas, permitiendo una aproximación a la composición original de la mezcla para cada muestra. Todas las muestras con aglomerante de cal y agregados de toba volcánica presentan diferentes grados de alteración natural. Los tipos de agregados presentes resultaban idóneos para mejorar las reacciones puzolánicas que han manifestado su durabilidad. Los patrones de litología agregada han revelado tres tipos distintos de mortero dentro del grupo de siete estructuras, en función de sus perfiles únicos agregados, lo que sugiere la presencia de mezclas específicas que estaban en uso en toda la ciudad al mismo tiempo. PALABRAS CLAVE: litología, petrografía, microscopia óptica, puzolanas, toba, constructores ­antiguos.

INTRODUCTION The ruins of the ancient city of Ostia lie near the western coast of central Italy, roughly 30 km from Rome in the region of Lazio. This part of Italy­represents a complicated geography with an

even more complex geological profile. The region lies between two volcanic districts, Monti Sabatini to the north and Colli Albani to the south, with a sedimentary basin along the Mediterranean coast to the west. The area is rich with volcanic material from numerous eruptive events that outcrop

46

Jennifer Wehby Murgatroyd

throughout the region, often overlapping into ­intermixed deposits (Karner et al. 2001: 195-215). The Tiber River snakes its way through the region, cutting through the volcanic deposits to empty into the Mediterranean Sea, carrying with it a mixture of volcanic and sedimentary material that had been eroded, transported, and redeposited along the way. Erosional and depositional processes have extended the basin, pushing the coast further out to sea by approximately 3 km in the last centuries, meaning that while once an ancient coastal city, Ostia­is now situated inland (Giraudi et al. 2009: 380). Ostia Antica, as the site is known today, is an impressively complete archaeological site, with many of the structures preserved to a considerable height and even upper floors still intact in some cases. The ruins extend for more than a square kilometre, though geophysical surveys completed by Heinzelmann revealed that the standing remains represent only half of the original site (Martin et al. 2002). During the Hadrianic period, the primary construction technique comprised masonry walls exhibiting a typical thickness of 44 cm for internal walls or 60 cm for external weight bearing walls, with facings keyed into a conglomeratic concrete core (DeLaine 2003: 723-724). The facings comprised either brick and reticulate stone (fig. 1a) or solely brick (fig. 1c). Ancient builders in Ostia apparently understood what modern builders know today: lime mixed with water and sand made a good binder for the brick facings and produced a solid concrete core when mixed with coarse aggregate (fig. 1b). The structural engineering success of these buildings cannot be overstated given their centuries of use, constant exposure the elements, local earthquakes and shifting geography, and later (sometimes unsympathetic) attempts at preservation and repair. Add to this the specific stresses from spending decades as an active archaeological site and tourist attraction and the technical skill of the ancient builders become all the more apparent. The material properties and performance of cementitious materials directly depend on the type and mix proportions of binder and aggregate, a fact that was apparently known to ancient builders. There is no shortage of advice from ancient Roman authors who gave detailed instructions regarding the required materials and procedures for mixing mortar under various conditions. Cato, in De Agri Cultura XV, suggests an optimal sand to lime ratio of two to one for enclosure walls, for example. However, in De Architectura II.5.1, Vitruvius stipulates that mortar for standing structures

Anejos de AEspA LXXVII

Fig. 1. Detail of facing and core sections of a damaged wall from the Case a Giardino showing a) areas of reticulate; b) areas of exposed core; and c) an area of brick facing (author).

should be made with three parts pit sand to one part lime. The fact that various mix specifications were preserved in ancient texts suggests an empirical understanding of the interaction between the binder and aggregate materials, as well as a familiarity with practical building techniques. By detailing specific mix proportions, ancient authors displayed a clear understanding of the performance properties of different materials and how to mani­ pulate them in order to best achieve a desired result. What is not clear, however, is the degree to which the academic prescriptions for best practice – as described in the ancient texts – was put into practise on site by the builders themselves. Visual assessments of mortar in situ can provide basic information about its compo­ nents and condition, for example the presence of lime and its current friability, and basic descriptions and identification of aggregate particles. Modern concrete scientists employ various an­ alytical techniques to further describe the components, condition, and mix quality of mortar and concrete samples on a microscopical level, and by extension, evaluate the material production process. Due to the similarities between modern and

Anejos de AEspA LXXVII

Lime Mortar Production in Ostia: Material analysis...

ancient mortar, material analysis can be applied to ancient samples in order to likewise evaluate the decisions made by individual builders. Where forensic evaluations of modern mortar samples are typically employed to investigate failure and degradation mechanisms, this project sought to explain the success of these ancient mortars. Date1

Address2

The study investigated mortar samples from seven public, private, and domestic buildings from across the ancient city: the Case a Giardino, Caseggiato del Serapide, Caseggiato del Larario, Casa Basilicale, Piccolo Mercato, Terme del Foro, and Insula dell’Ercole Bambino (figs. 2-3). This group represented both small- and large-scale Building function

Access

Case a Giadino

125CE

III, IX, 1-23 Apartments Mixed-use Brick and shops and reticulate

Caseggiato del Serapide

125CE

III, X, 3

Residence

Caseggiato del Larario

117CE

I, IX, 3

Residence and shop Mixed use

Casa Basilicale

118CE

I, IX, 1 Private residence

Private

Brick and reticulate

Piccolo Mercato

118CE

I, VIII, 1

Storage, warehouse

Public Brick and reticulate

Terme del Foro

160CE

I, XII, 6

Public bath hose

Public

II, VI, 3-4

Residence Private

Brick

Insula dell’Ercole Bambino 130CE

47

  Dates reported by DeLaine from brick stamps (DeLaine 2002: 77-99).   Addresses follow the standard nomenclature of Regio, Insula, Building.

1 2

Fig. 2. Structures included in this study.

Fig. 3. Images of each structure (author).

Private

Fracing materials

Brick Brick

Brick

48

Anejos de AEspA LXXVII

Jennifer Wehby Murgatroyd

construction projects undertaken by separate collectives of builders, groups which were initially identified by DeLaine’s architectural research that included these particular structures (DeLaine 2002: 41-76). The samples from Case a Giardino represented a special case, as samples were collected from three unique locations, including the facade of the main entrance, the outer shops of the courtyard, and the central apartments. An additional sample was unfortunately received without precise provenance information. The mortars were treated as four unique samples and assessed separately. All structures were built during the reigns of  Hadrian and Antoninus Pius, dating from ­117-160 CE, as determined from studies of their recorded brick stamps (DeLaine 2002: 77-99). This fairly limited time frame reasonably could be considered to represent a single generation of builders working in Ostia’s construction trade. Hypothetically, local techniques and procedural standards could have been widespread throughout Ostia, perhaps resulting in mortars that were similar, if not entirely consistent, across the city. The aim of this study was to test the hypothesis that the builders in Ostia employed locally consistent materials and techniques to produce mortars of similar character and quality. The test protocol incorporated analysis of mortars from each of these seven structures to determine their material properties and current condition state as a proxy for their durability. Of primary concern was mortar composition, particularly the builders’ choice of binder and aggregate, how they processed the materials, and the mix proportions they employed. MATERIALS AND METHODS Samples were collected by hand, using a hammer and chisel or trowel to remove mortar from wall facings and exposed areas of the core from two separate walls in each structure. Where possible, samples were taken from pre-damaged areas­, such as holes and spalled areas of the facing (fig. 1), in order to minimise the new damage caused by sampling. Laboratory investigations comprised visual assessment and petrographic analysis to evaluate the physical properties of the samples, including the type, particle size, shape, and specific lithology of fine aggregates, representational proportions of the different aggregate

lithologies, and the characteristics and condition of the lime binder. Aggregate particle lithology was identified petrographically, comparing key features of each particle (i.e. texture, mineralogy, degree of alteration, specific alteration products) to known samples of fresh material from both Monti Sabatini and Colli Albani. Modal analysis of mortar components comprised a point count of each component encountered during measured transects across a thin section of mortar, using a Leica polarising microscope at 100X magnification with a step distance of 0.5 mm for a minimum of 500 counts per slide. Point counts were completed on six slides from each structure, and every slide was counted twice. Quantitative results have been reported below as mean percentage of the total count for each structure. RESULTS Mortar description and components All mortar samples exhibited a moderately soft to moderately hard exterior, appearing white to light grey in colour, with an air lime binder and volcanic tuff aggregate. The mortars were generally well-adhered to the bricks and stones of the facings, making them difficult to remove in most cases. High-power microscope analysis revealed that the binder was fully carbonated and commonly contained sub-rounded to rounded voids, typically 300-500 µm in diameter (fig. 4a). Spor­ adic fine cracks (10-100 µm wide) and micro­ cracking (1-10 µm wide) were observed in the binder of all samples (fig. 4b). The fine aggregate component comprised angular to sub-rounded particles of unwelded tuff with a nominal 2 mm maximum size, exhibiting sporadic to common interfacial partings from the binder (fig. 4c). All samples also contained minor or trace proportions of other inclusions, including loose mineral crystals, leucitic lava fragments, clay particles, chert, quartzite, ceramic fragments, and rare root traces and unidentifiable degraded lithic fragments. The observed loose minerals included primary grains from aggregate particles (leucite, clinopyroxene, quartz, feldspar, micas, hauyne), secondary lithic alteration minerals (analcime ­after leucite), and secondary mortar binder alteration products (calcite, stratlingite). Particles of unincorporated lime were present in varying quantities and were visible in hand specimen

Anejos de AEspA LXXVII

Lime Mortar Production in Ostia: Material analysis...

Fig. 4. Photomicrograph set of mortar samples showing a) light and dark grey areas of variable binder carbonation and void space marked “V,” – aggregate particles appear black in cross polarised light; b) microcracking in the matrix, cracks and voids appear yellow in plane-polarised light; c) typical variability of fine aggregate particles of volcanic tuff, which appear black, brown, or tan in plane polarised light; d) unincorporated lime particles, outlined in black, appear pinkish grey in cross polarised light.

(fig.  4d). The lime particles were commonly degraded and exhibited shrinkage cracks, interfacial parting from the binder, high carbonation, spor­ adic alteration to calcite, and material loss during thin section preparation. Mix proportions The first approximation of mix proportion for each sample was determined from the ratio of binder, aggregates, and void space recorded with modal analysis (fig. 5). Results indicated that binder content varied from 27% to 38%, with an overall average of approximately 33%. The aggregate accounted for 57% to 68% of the constituents, with an average of approximately 61%. The average void content across all samples was approximately 6%, including the anomalously high value of 9% in the mortar from Insula dell’Ercole Bambino and a low value of 4% in the mortar from the courtyard of the Case a Giardino. These results suggested an average binder:aggregate mix proportion of approximately 1:2. Aggregate lithology The mortars from Ostia contained aggregates from both the Monti Sabatini volcanic district to the north and the Colli Albani to the south. Ag-

49

gregates from the two locations exhibited distinct mineralogical and morphological characteristics indicative of their natural alteration processes. Figure 5 illustrates the averaged aggregate profile for each structure. The table includes data from some slides that exhibited poor repeatability between counts due to occasional misidentification of aggregate lithology; however, removing those data from the analysis did not change the interpretation of results. The observed aggregate particles from the Colli Albani volcanic district comprised approximately 88% of the overall total aggregate across all samples. These materials included Pozzolane Rosse (fig. 6a), Tufo Lionato (fig. 6b), Pozzolanelle (fig. 6c), and Pozzolane Nere (fig. 6d) in descending order of abundance. Colli Albani part­icles exhibited a sub-angular to sub-rounded shape and typically varied in size from sub-10 µm to 2 mm, with rare particles over 1 cm in size. In situ, Pozzolane Rosse naturally occurs with three distinct alteration facies displaying increasing degrees of hydrothermal alteration. Alteration features include accumulation of halloysite clay on the outer grain rims, groundmass alteration, and needles of the zeolite phillipsite growing in vesicles that are sometimes lined with opaline silica (Jackson et al. 2010: 57-62). All three facies were present within the mortars, but the intermediate alteration facies accounted for nearly 70% of all Pozzolane Rosse particles identified across all samples. Similarly, the Tufo Lionato and Pozzolanelle lithologies represent two strata of a single eruptive event known as the Villa Senni eruption of Colli Albani, and these can distinguished from each other by the texture and degree of alteration within the groundmass. In outcrop, Tufo Lionato typically exhibits heavy hydrothermal alteration and considerable growths of the rhombohedralshaped zeolite chabazite. Although both types of Villa Senni material were present in the mortars, the more highly altered Tufo Lionato material comprised over 65% of the Villa Senni aggregate observed in the sample set. Aggregates from the Monti Sabatini district included Monti Sabatini ­Volcanic ­District (MSVD) reworked air fall ash (fig. 7a) and Tufo Giallo di Prima Porta (fig. 7b). These materials commonly exhibited a sub-rounded shape, rather than the more angular shape of the Colli Albani material, and ranged in size from 100 µm to 10 mm. Particles often retained thick clay coatings on the

0.0

1.7

1.4

10.5

1.7

4.7

27.1

9.6

3.7

2.5

4.8

24.4

0.0

1.3

0

0

1.2

1.2

2.2

33.9

0.2

26.2

66.6

 4

29.4

16.8

17.1

3.4

22.7

0.5

3.7

0.3

5.6

23.8

Case a Giardino Courtyard

0.0

0.9

6.8

0

0.5

1.4

0.5

15.6

0.6

37.5

64.2

 6.2

29.6

1

0.0

1.6

2.1

0

0.6

1.2

0.5

34.6

1.6

17.7

60.0

 6.1

33.9

20.5

14

1.1

8.5

8.1

4.7

1.4

6.1

27.9

Cas. del Serapide

0.0

2.8

0.7

0

0.8

1.8

0.7

46.1

0.1

3.5

56.6

 5.7

37.7

27.1

19

1

2.4

0.1

5

0.8

5

32.7

Cas. del Larario

0.05

2.4

0.9

0

0.9

1.4

1.7

6.1

1.3

47.2

62.0

 6.7

31.3

4.7

1.4

6.3

28.3

12.7

5.9

0.8

7

24.3

Casa Basilicale

0.02

1.2

1.4

0

0.4

3.7

1.4

4.6

3.3

46.1

63.1

 5.7

31.2

4.4

0.3

1.7

32

12.4

4.6

1.1

5.9

25.3

Piccolo Mercato

Fig. 5. Results of point counting modal analysis reported as percentage of the total count (mean values per structure). * = not counted with modal analysis but trace amounts observed in thin section.

14.9

0.7

2.3

23.4

11.8

6.-

0.2

6.2

23.4

Case a Giardino Façade

0.0

2.1

1.1

1.1

0.4

8.8

0.7

35.8

0.4

10

60.3

 5.2

34.5

24.4

11.5

1.5

7

1.5

4.1

1

4.4

30.1

Terme del Foro

0.1

2.8

4.4

0

0.3

3.7

1.2

12.8

3.3

31

59.6

 9.1

31.3

8.3

4.6

1.3

27.2

2.5

8.3

0.8

6.1

25.2

Insula dell’Ercole Bambino

Jennifer Wehby Murgatroyd

  Unfortunately this particular sample was received without precise provenance information.

2.2

0.0

Free Crystals

Root Traces

0

0.0*

0.9

Ceramic

Line Lumps

0.3

0.7

Other Lithics

3.9

0.5

0.6

0.0

Tufo Giallo di Prima Porta

49.1

Tufo Lionato

12.2

2.8

41.5

64.5

 6.2

29.2

Case a Giardino Apartment

MSVD reworked air fall ash

7

56.1

Pozzolanelle

Villa Senni Aggregate

0.2

Least Alteration Facies

0.3

5.9

Intermediate Alteration Facies

Pozzolane Nere Aggregate

1.1

7.2

67.9

Greatest Alteration Facies

Pozzolane Rosse Aggregate

Aggregate Lithology Breakdown

Aggregate Total

1.6

3.1

Linear crack

Irregular woids

4.6

Unidentifiable relict clasts

 4.7

22.9

27.4

Matrix

Void space

Binder

Case a Giardino Sample OS-C151

50 Anejos de AEspA LXXVII

Anejos de AEspA LXXVII

Lime Mortar Production in Ostia: Material analysis...

Fig. 6. Photomicrograph set of aggregate particles from Colli Albani: a) Pozzolane Rosse, with a particle of unincorporated lime; b) Tufo Lionato; c) Pozzolanelle; d) Pozzolane Nere. Images collected by author in plane polarised light.

e­ dges, which cemented multiple particles of lithologies (occasionally including small fragments of Pozzolane Rosse) within single grains. These erosional and depositional features suggested this material was more likely part of a redeposited epiclastic deposit than a natural outcrop. The Monti Sabatini aggregates combined represented less approximately 7% of all aggregates identified in the sample set, suggesting these were not a ­major component in any of the mortar mixes. Additional observations Petrographic observations provided good evidence for the interaction between the binder and

51

aggregate particles within the matrix. In many cases, the lime binder had penetrated deeply into the vesicles of larger Pozzolane Rosse and Tufo Lionato fine aggregate particles (fig. 8a). The ­edges of larger Pozzolane Rosse fragments were commonly lined with microscoria, which previously had been cemented along the edges with clay (fig.  8b). These clay coatings have been replaced by the mortar binder or left as empty void space, most likely consumed via thermal dissolution in the pozzolanic reaction (He et al. 1995: 1701). Rare instances of larger Tufo ­Lionato parti­ cles exhibited intact grains of the zeolite chabazite in the central vesicles, which was not consumed by the reaction (fig. 8c). The majority of Tufo Lionato particles, however, did not exhibit the zeolites, suggesting the chabazite was consumed by pozzolanic reaction. The mortars exhibited no evidence of alkalisilica reaction, delayed ettringite formation, or sulphate attack, all of which are common degradation mechanisms of cementitious materials (Bendetti et al. 2004: 344-345). Although all samples exhibited some cracking, as reported in fig.  2, the cracking generally was not excessive. In all cases the linear cracks represented less than half of the observed void space, indicating the mortars have maintained a good level of durability. Fine shrinkage cracks and tension gashes, which indicate a mortar has dried too quickly, tend to form in the matrix fairly soon after installation; these were primarily limited to those within lime lumps and were only rarely observed in the binder matrix.

Fig. 7. Photomicrograph set of aggregate particles from Monti Sabatini: a) Monti Sabatini Volcanic District reworked airfall ash, with accreted Pozzolana Rosse and Tufo Lionato particles; and b) Tufo Giallo di Prima Porta. Images collected by author in plane polarised light.

52

Jennifer Wehby Murgatroyd

Anejos de AEspA LXXVII

Fig. 8. Photomicrograph set of key petrographic features in aggregate particles: A) mobilised cement redeposited in vesicles of Pozzolane Rosse (cross polarised light); B) microscoria in Pozzolane Rosse (plane polarised light); and C) relict chabazite (grey, marked Ch) in Tufo Lionato (cross polarised light). Images collected by author.

DISCUSSION Petrographic analysis of these mortars indicated they were generally well-made, with only limited evidence of degradation. As noted above, cracking observed in the matrix was never excessive, and the type of cracks observed were generally related to structural age or sample collection rather than to the initial mortar production or installation. The facing samples were all fully carbonated, which means they would have reached their highest strength potential. The carbonation mechanism in air lime mortar is driven by the presence of water, and a well-made mortar will have been kept fully hydrated throughout the initial set. Mortar that has been allowed to dry too quickly will exhibit characteristic shrinkage cracks, which were absent in all of the mortar samples in this investigation. Analysis of the point count data revealed that the sample included three groups of mortar based

on relative proportion of aggregate type, specifically the ratio of Pozzolane Rosse aggregates to material from the Villa Senni eruption (fig. 9). The identified mortar groups could not be correlated to specific archaeological or architectural features of the structures themselves, including location within the city, age of the building, function of or access to the structure, or the type of facing used in construction. The samples could only be categorized by the type and proportion of mortar constituents. The first mortar type was dominated by Pozzolane Rosse aggregates and included the samples from the Casa Basilicale, the Piccolo Mercato, the  Insula dell’Ercole Bambino, and the central apartments and facade from the Case a Giardino. In all cases, the mortar included greater than 30% Pozzolane Rosse and 15% or less Villa Senni Mat­ erial. The second mortar type was strongly dominated the Villa Senni eruptive materials (Tufo Lionato and Pozzolanelle), which comprised less

Anejos de AEspA LXXVII

Lime Mortar Production in Ostia: Material analysis...

53

Fig. 9. Aggregate lithology profiles as determined by modal analysis, averaged per structure, and reported as percentage of the total count.

than 10% Pozzolane Rosse and greater than 40% Villa Senni aggregates. These aggregates were clearly predominant in the samples from the Caseggiato del Larario and in one sample from the Case a Giardino. The samples from the remaining structures (the Caseggiato del Serapide, the Terme del Foro, and the courtyard of the Case a Giardino) contained a mixture of aggregates without a strongly dominant component. These mortars typically contained more fine aggregate particles with intact authigenic clays and erosional textures, suggesting the aggregate was possibly epiclastic in nature. The binder:void:aggregate ratios were variable across the sample set, but exhibited some explainable patterns when viewed in the context of mortar types (fig. 10). Three of the Pozzolane Rosse group of mortars (the Casa Basilicale, the Piccolo Mercato, and the Insula dell’Ercole Bambino) exhibited within-group consistency, with all structures exhibiting a binder content of approximately 31% and aggregate content of 60-63%, although the void space varied rather considerably. This result was consistent with a set of mortars that have been produced following a single standard, preferentially selecting a specific aggregate type and tightly controlling the quantity of both lime and aggregate. The three mortars from the known loc­ ations in the Case a Giardino likewise exhibited a

consistent binder content of approximately 29%, with aggregate content between 64 and 66%. This suggests tightly controlled production conditions, specifically controlling the proportions of lime to aggregate throughout the site as a whole, despite variations in aggregate profiles. The Villa Sennitype mortar from the Caseggiato del Larario was a unique case, exhibiting a high binder content of approximately 38% and a low aggregate content of almost 57%. The variable mix proportions observed in the potentially epiclastic mortars from the Caseggiato del Serapide and the Terme del Foro distinguished these mortars from those in the Pozzolane Rosse and Villa Senni groups. These samples were consistent with mortar that has been less tightly controlled during production. The Pozzolane Rosse mortars and, to a lesser extent the Villa Senni mortars from Caseggiato del Larario and the unprovenanced sample from the Case a Giardino, exhibited a nearly homogeneous aggregate composition with a s­trongly ­do­minant aggregate lithology, unlike the epiclastic mortars which were not dominated by any single aggregate type. The homogeneity of aggregates seemed to suggest a preferential selection of a specific type of aggregate. The builders would have been required to maintain control over the material supply at every stage – from collection at the quarry to storage, transport to the building

54

Anejos de AEspA LXXVII

Jennifer Wehby Murgatroyd

Fig. 10. Binder:Void:Aggregate ratios, as determined with modal analysis, reported as percentage of the total count (mean values per structure) and grouped in ascending order of binder percentage.

site, and eventual mortar production – in order to prevent inadvertent mixing of various materials. The level of organisation and quality control required for this suggests that the material would have been handled with care and considered an important component in its own right. While it is not possible to know why the ancient builders believed these materials were worth special hand­ ling, the materials themselves exhibited key characteristics that can explain their good performance as aggregates. Both Pozzolane Rosse and the Villa Senni aggregate can be considered naturally pozzolanic materials. Pozzolanic aggregates contribute to the strength and durability of the mortar by combining with lime, in the presence of water, to form calcium-alumina-silicate-hydrates within the cem­ entitious matrix (Moropoulou et al. 2004: 1). Modern builders foster this reaction with the addition of synthetic pozzolans, such as pulverised fuel ash, ground glass blast-furnace slag, or even brick dust (Mertens et al. 2009: 233). In the case of the Ostian mortars, the natural alteration products of the aggregates, the clay and zeolites, contributed the silica and alumina required to generate and complete the pozzolanic reaction (Snellings et al. 2009: 48). This means that the ancient­builders were able to achieve the best pos-

sible result without mineral additives, as required for modern concretes. The further implication of preferential selection of aggregates with naturally beneficial components is that the builders well under­stood the physical properties and their benefits in the mortar. CONCLUSIONS Modal analysis from point counting thin sections allowed for direct quantification of main constituents of mortar samples from Hadrianicera structures in Ostia. Distinct types of mortar were identified among the sample set, all of which comprised lime binder and unique aggregate compositions. At this stage of the research, the clearest evidence for specific mortar mix design is that of the Pozzolane Rosse mortars. Multiple samples exhibited consistent composition, aggregate profiles, and petrographic features. The possibility of a Villa Senni mortar is highly likely, given the similarity of quality across the samples from the Caseggiato del Larario, the relative consistency of aggregate types observed within the samples, and the unique binder and aggregate proportions. However, more structures should be studied to increase our understanding of this

Anejos de AEspA LXXVII

Lime Mortar Production in Ostia: Material analysis...

mortar type, and by extension, the groups of builders who used it. The ancient builders, specifically the mortar producers, were obviously skilled technicians who followed systematic procedures, as seen in the consistency of material selection and distinct mix designs within the Pozzolane Rosse and Villa Senni type mortars. Furthermore, the material chosen was naturally ideal for the use as aggregate, with features that ultimately imparted good durability to the final product. The data suggest that groups of builders can indeed be identified by the material properties and components of the mortar they installed during construction. The information gleaned from this research was determinable only with detailed material analysis, which was able to identify specific aggregate selection and apparent quality control during production. These analyses can be used to supplement on-site structural observations to better understand key aspects of ancient structures, and by extension, the ancient building trade. ACKNOWLEDGEMENTS The author would like to thank, with much gratitude, Dr Marie Jackson for invaluable support and assistance with the initial research design­, as well as for the loan of geological source materials and reference samples of mortar from Ostia. Thanks also must go to Dr Janet DeLaine and Prof Mark Pollard for serving as thesis advisers throughout the D.Phil research, during which the bulk of this work was conducted. PETROG point counting equipment and software were very generously provided for use by Conwy Valley Systems Limited. Finally, thanks must also be extended to Dr Ian Sims and RSK Environment, Limited for financial support and use of laboratory facilities, which have allowed for continued research on this project. REFERENCES Bendetti, D., Valetti, S., Bontepi, E., Picciolo, C. and Depero, L. E. 2004: “Study of ancient mortars from the Roman Villa of Pollio Felice in Sorrento (Naples)”, Applied Physics A, Materials Science and Processing, 79, pp. 341-345. DeLaine, J. 2002: “Building activity in Ostia in the second century AD”, in Bruun, C. and Zevi, A. G. (eds.), Ostia e Portus nelle loro relazioni con Roma.

55

Atti del Convegno dell’Institutum Romanum Finlandiae (Roma, 3-4 dicembre 1999), pp. 41-101, Acta Instituti Romani Finlandiae 27. Institutum Romanum Finlandiae, Roma. DeLaine, J. 2003: “The builders of Roman Ostia: Organisation, status and society”, in Huerta, S. (ed.), Proceedings of the First International Congress on Construction History, Madrid, 20th-24th January 2003, pp. 723-732. Instituto Juan de Herrera, Madrid. Giraudi, C., Tata, C. and Paroli, L. 2009: “Late Holo­cene evolution of Tiber river delta and geoarchaeology of Claudius and Trajan Harbor, Rome”, Geoarchaeology, 24, pp. 371-382. He, C., Osbaeck, B. and Makovicky, E. 1995: “Pozzolanic reactions of six principal clay minerals: Activ­ation, reactivity assessments and technological effects”, Cement and Concrete Research, 25, pp. 1691-1702. Jackson, M., Deocampo, D., Marra, F. and Scheetz, B. 2010: “Mid-Pleistocene pozzolanic volcanic ash in ancient Roman concretes”, Geoarchaeology, 25, pp. 36-74. Jackson, M., Marra, F., Hay, R., Cawood, C. G. and Winkler, E. M. 2005: “The judicious selection and preservation of tuff and travertine building stone in ancient Rome”, Archaeometry, 47, pp. 485-510. Karner, D. B., Marra, F. and Renne, P. R. 2001: “The history of the Monti Sabatini and Alban Hills volcanoes: Groundwork for assessing volcanic-­ tectonic hazards for Rome”, Journal of Volcanology and Geothermal Research, 107, pp. 185-215. Martin, A., Heinzelmann, M., De Sena, E. C. and Granino Cerere, M. G. 2002: “The urbanistic project on the previously unexcavated areas of Ostia (DAI-AAR 1996-2001)”, Memoirs of the American Academy in Rome, 47, pp. 259-304. Mertens, G., Snellings, R., Van Balen, K., ­Bicer-Simsir, B., Verlooy, P. and Elsen, J. 2009: “Pozzolanic reactions of common natural zeolites with lime and parameters affecting their reactivity”, Cement and Concrete Research, 39, pp. 233-240. Moropoulou, A., Cakmak, A., Labropoulos, K. C., Van Grieken, R. and Torfs, K. 2004: “Accelerated microstructural evolution of a calcium-­silicatehydrate (C-S-H) phase in pozzolanic pastes using fine siliceous sources: Comparison with historic pozzolanic mortars”, Cement and Concrete Research, 34, pp. 1-6. Snellings, R., Mertens, G., Hertsens, S. and Elsen, J. 2009: “The zeolite-lime pozzolanic reaction: Reaction kinetics and products by in situ synchrotron X-ray powder diffraction”, Microporous and Mesoporous Materials, 126, pp. 40-49.

II STRUCTURAL AND CONSTRUCTIONAL USES OF METAL

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

THE BRONZE TRUSS OF THE PORTICO OF THE PANTHEON IN ROME DOROTHEE HEINZELMANN*, MICHAEL HEINZELMANN** * LVR-Amt für Denkmalpflege im Rheinland / Universität zu Köln, Kunsthistorisches Institut, Abteilung Architekturgeschichte ** Universität zu Köln, Archäologisches Institut

ABSTRACT: The portico of the Pantheon in Rome offered an extraordinary example of Roman engineering: a truss completely constructed of bronze. Although it was dismantled in 1625 on the orders of Pope Urban VIII a large number of architectural drawings, including two particularly de­ tailed sketches by Francesco Borromini, give evidence of its original structure. A current research project at the University of Cologne is dedicated to the investigation of the bronze truss, focussing on its history, design, construction technique, material properties and statics. The investigation is based mainly on an analysis of the traces left by the roof structure on the portico itself, using among other things a complete documentation by a high resolution laser scan; secondly it uses Borromini’s drawings for a reconstruction, and finally one rivet, the only remnant of the ancient structure (today in the Antikensammlung Berlin), has been analysed to determine its chemical composition and re­ produced in order to examine its physical properties. KEYWORDS: Rome, Pantheon, Roman construction technique, Bronze roof truss, Reconstruction. RESUMEN: El pórtico del Panteón de Roma ofrece una extraordinaria obra de ingeniería romana: una armadura construida completamente en bronce. A pesar de que fue desmantelado en 1625 por orden del papa Urbano VIII, un gran número de dibujos arquitectónicos, entre ellos dos bocetos particularmente detallados de Francesco Borromini, evidencian su estructura original. Un proyecto de investigación de la Universidad de Colonia se está ocupando de la investigación de la armadura de bronce, centrándose en su historia, diseño, técnica de construcción, propiedades del material y estática. La investigación se basa principalmente en un análisis de las huellas existentes en la estruc­ tura del techo del propio pórtico, utilizando, entre otras cosas, una documentación completa de alta resolución realizada con láser scan. En segundo lugar, se utilizan los dibujos de Borromini para una reconstrucción y, finalmente, se ha analizado un remache, el único resto de la antigua estructura (hoy en el Antikensammlung Berlín), para determinar su composición química, efectuando una reproduc­ ción con el objetivo de examinar sus propiedades físicas. PALABRAS CLAVE: Roma, Panteón, técnica de construcción romana, armadura de cubierta en bronce, reconstrucción.

The Pantheon in Rome is among the most outstanding buildings of Roman ­construction and has captivated generations of architects and engin­ eers since the Renaissance (fig. 1). While today it is primarily famous for the construction of its cu­ pola, until 1625 it offered another showpiece of antique building technique: a roof truss, origi­ nally consisting completely of bronze, above the

pronaos. Although Emperor Constans II had probably already purloined the gilded roof panels in 663, the actual supporting structure remained intact into the early 17th century and was recorded in sketches and drawings by numerous, particu­ larly 16th-century architects.1 In 1625 Pope Urban   For a general overview see Fine-Licht 1968: 46-58 and 184.

1

60

DOROTHEE HEINZELMANN, MICHAEL HEINZELMANN

Anejos de AEspA LXXVII

Fig. 1. Rome, Pantheon. View from North-East (photo: D. and M. Heinzelmann).

VIII gave the order to remove the bronze support­ ing structure in order to cast cannons, and to re­ place it with a timber roof truss.2 This timber con­ struction seems to be at least partially the one that is retained today.3 During the disassembly Francesco Borro­moni, who was also occupied with repairs to the  col­ umns of the pronaos, was able to examine the con­ struction of the roof truss more closely and take measurements. He left behind highly detailed drawings, which are much more accurate than the previous ones and are particularly important for the reconstruction.4 In addition, it was established   Rice 2008: 337-352.   An analysis of the existing roof structure seems still to be a desideratum. Between 1900 and 1902 restoration works were carried out on the pronaos roof structure during which parts of the trusses were replaced by new beams. See Lapenna 2006: 153-164 and the correspondence in Archivio Centrale dello Stato (Rome), Ministero Pubblica Istruzione, Divisione Generale AA.BB.AA., Divisione Monumenti, III° versamento, IIa parte, Busta 717, fascicolo 1169. 4   Thelen 1967: vol. I, 32-33, vol. II, C 25 and C 26. 2 3

at the time how much bronze was obtained to melt down. The total amount was approximately 180 tonnes, from both the supporting elements and the rivets joining them. At the time the rivets were regarded as coveted collectors’ items and were handed to high-ranking guests as gifts. Of all pieces formerly mentioned in collections, how­ ever, only one single rivet is known today in the Berlin antiquities collection. Astonishingly, so far there seem to have been no close examination of the Pantheon’s antique bronze roof truss.5 The current state of research is still constituted by the chapter on the pronaos in Kjeld de Fine-Licht’s fundamental work on the Pantheon from 1968.6 The latter had already ana­ lysed numerous Renaissance drawings, but evi­ dently he had no knowledge of Borromini’s draw­ ings in the Albertina in Vienna. During a current 5   Discussions among others in Thelen 1967: vol. I, 32-33; Haselberger 1996: 182-189; Valeriani 2006: 111-112. 6   De Fine-Licht 1968: 35-84.

Anejos de AEspA LXXVII

The Bronze Truss of the Portico of the Pantheon in Rome

61

Fig. 2. Rome, Pantheon. Portico, upper part of the central nave seen from east (authors).

research project at the Archaeological Institute of the University of Cologne, an attempt is now ­being made in co-operation with various partners to draw from all available components as accurate a profile as possible of the original roof truss. The following elements serve as a basis: 1. Structural analysis: On one hand the struc­ tural traces of the roof truss in the Pantheon itself are being examined. The documenta­ tion available for this includes a complete 3D laser scan of the Pantheon that was made as part of the Bern Digital Pantheon Project in 2006/07 and records the current status at a precision of a few millimetres (figs. 2-4).7 2. Structural records and drawings from the 15th17th century: Based on the drawings created up to 1625, the aim was to make as detailed a reconstruction of the roof truss as possible that can be linked with the traces on the  Graßhoff et al. 2009: 7-12.

7

s­ tructure. Borromini’s drawings play a central role in the process. 3. Material analyses: With the aid of an analysis of the chemical composition of the rivet in the Berlin antiquities collection, re-castings from the same alloy were prepared by the Institute of Metallurgy at RWTH Aachen University and the material was tested for its physical properties. The results form the basis for a static calculation of the supporting structure, which will be processed by the Brandenburg­ ische Technische Universität Cottbus (Prof. Dr.-Ing. Werner Lorenz). 4. Finally, also in co-operation with the BTU Cottbus we intend to have parts of the sup­ porting structure re-cast to their original size, in order to be able to recreate the actual con­ struction process, particularly the riveting technique, in tests. Only an interim status of the project can be presented here.

62

DOROTHEE HEINZELMANN, MICHAEL HEINZELMANN

Anejos de AEspA LXXVII

Fig. 3. Rome, Pantheon. 3D-Laserscan, East-West-section of the point cloud of the portico (Bern Digital Pantheon Project).

Fig. 4. Rome, Pantheon. 3D-Laserscan, longitudinal sections of the point cloud of the portico (Bern Digital Pantheon Project).

Anejos de AEspA LXXVII

The Bronze Truss of the Portico of the Pantheon in Rome

Briefly on the construction history of the Pan­ theon and pronaos. The present Pantheon is ­already the third building on this site, after its pre­ decessors, the buildings of Agrippa and ­Domitian, were destroyed by lightning strikes.8 Based on a mention in Cassius Dio (69, 7, 1), the present ­Pantheon is traditionally ascribed to the Emperor Hadrian. However, the latest examination of the brick stamps by Lise Hetland has supplied strong arguments for placing the commencement of con­ struction – and thereby the building’s actual con­ ception – already under the reign of Trajan, while it was only completed under Hadrian. Indeed, the rotunda up to the start of the cupola already seems to have existed when Hadrian came to ­power.9 Commencement of construction was probably be­ tween 110 and 114 AD; the building should have been completed in 125 AD at the latest, as a first Senate session is attested for this year.10 Within the mere 10-year or so construction period, various planning changes can be obser­ ved.11 At least one of these presumably also con­ cerns the height of the pronaos, as the so-called intermediate block is integrated into the rotunda’s masonry only up to mid-height. Mark WilsonJones assumes that the original plan for the pro­ naos probably provided for the erection of 50-foot granite columns, instead of the only 40-foot col­ umns actually used.12 After this re-planning, how­ ever, all components of the intermediate block and the pronaos are well harmonized and seem to be based on a uniform plan. The roof truss alone shows certain discrepancies in connection with the intermediate block, which perhaps point to renewed re-planning or unsuccessful planning. However, it cannot be ruled out that a later alter­ ation may also be involved here. Restoration measures are proved by an inscription on the archi­trave of the pronaos under Septimius Severus in 202 a.D. (CIL VI, 896).13 The pronaos exhibits a width of 34.2 m and a depth of 15.6 m (fig. 5). Its rear wall forms the fa­ çade of the intermediate block, which opens up to the entrance portal into the rotunda. The portico is  8   For a brief summary of the building history see i.a.: Fine-­ Licht 1968: 172-179, 185-190; Heinzelmann 2009: 142-144; Ziolkowski 1999: 54-61; La Rocca and Virgili 1999: 280-285; La Rocca 2015: 49-78.  9   Hetland 2007: 95-112. 10   Cassius Dio 69, 7. 11   For a summary of the construction history see recently: DeLaine 2015: 160-192; Hetland 2015: 79-98; Wilson Jones 2015: 193-230. 12   Wilson Jones 2009: 81-85; Wilson Jones 2015: 213-224. 13   See also: Fine-Licht 1968: 41, 180, 190.

63

Fig. 5. Rome, Pantheon. Plan and section of the portico and the intermediate block (after: De Fine-Licht 1968, figs. 98 and 105).

structured by 8 columns on the front and 3 col­ umns on the sides, the interior being subdivided into three naves by means of two pairs of addi­ tional columns. The inner width of the middle nave is approx. 12.60 m between the colums, that of the side naves approx. 7.80 m each. Above the columns runs a classic entablature, which for its part bears the roof truss and the front triangular pediment. Crucial for the reconstruction of the roof truss are the masonry zones above the two inner

64

DOROTHEE HEINZELMANN, MICHAEL HEINZELMANN

Anejos de AEspA LXXVII

Fig. 6. Rome, Pantheon. Portico, upper part of the central nave seen from east (photo: D. and M. Buchholz).

column placements and the inner sides of the pediment (fig. 6). Above the columns are pillar structures consisting of seven layers of partly re­ used marble and travertine blocks with rough, un­ smoothed surfaces. They are connected by arches made of opus caementicium or plastered masonry. They serve to reinforce and stabilize the upper structure, while the arch shape helps to relieve the marble blocks underneath. This zone of the pillar structures and arches is clearly not meant to be visible. On these stone pillars rests the roof truss, which now consists of timber (fig. 7-8). The tim­ ber beams largely reuse the inlets of the former metal roof truss. The present roof truss consists of three triangular rafter units, which form the main supporting structure. The trusses in the cen­ tral nave form a palladiana (Queen post truss), as is characteristic of Roman roof trusses from the Middle Ages onwards.14 In this way it was possi­ ble to bridge the span of more than 12 m. As the general setting of the building remain the same, the Baroque supporting structure cor­ responds in many elements to the ancient roof truss. Nevertheless, it differs from it in essential points, as will be shown. Numerous traces, such as now unused oblique inlets in the pillars above the  columns or beam holes on the gable walls, provide clear clues to the original roof ­construction.  For the types and evolution of roof constructions in Rome cf. Valeriani 2006: 32 with fig. 6 (C1). 14

However, a distinction must be made between ­ancient and more recent traces as some of them are evidently post-antique. From the various drawings and sketches of the Renaissance period it follows that the ancient roof truss, like the present one, consisted of three truss units above all three naves, which were con­ nected to one another longitudinally by means of purlins.15 Several drawings, of which the Metro­ politan sketch of an anonymous French artist is the most detailled, reflect a blend of view and di­ agonal perspective, in order to ­emphasize the cha­ racteristics of the roof truss (fig. 9).16 For exam­ ple, the girders were not massive beams, but thin slice-like metal profiles, that ran parallel to one another and were connected to one another by means of rivets. Above the central nave there was a triangular construction made up of a horizontal tie beam, two rafters and a central king post. Fur­ ther support for the horizontal tie beam was pro­ vided laterally by two steep oblique struts, which correspond to the inlets in the masonry. The lat­ eral naves had simple triangular constructions made up of a horizontal tie beam, the external rafter and an oblique strut running counter to the rafter. Above the rafters, girders running longitu­ dinally in the manner of purlins are represented, 15   For a compilation of the most important drawings see i.a.: Fine-Licht 1968: 51-60. 16   Yerkes 2013: 87-89 with fig. 1.

Anejos de AEspA LXXVII

The Bronze Truss of the Portico of the Pantheon in Rome

Fig. 7. Rome, Pantheon. Portico, central nave looking South (authors).

Fig. 8. Rome, Pantheon. Portico, eastern aisle looking South (authors).

65

66

DOROTHEE HEINZELMANN, MICHAEL HEINZELMANN

Anejos de AEspA LXXVII

Fig. 9. Anonymous French draftsman, mid 16th century, elevations of the Pantheon portico roof structure and bronze truss (The Metropolitan Museum of Art New York, Goldschmidt Scrapbook, 68.769.1, verso).

Anejos de AEspA LXXVII

The Bronze Truss of the Portico of the Pantheon in Rome

67

Fig. 10. Francesco Borromini 1625, ancient truss of the Pantheon portico (Vienna, Albertina, Thelen 1967 C 25).

obviously U-shaped. However, none of the drafts­ men seems to have seen the construction from close up, and quite a few details remain unclear. Only Borromini found it possible to examine the construction closely during the dismantling of the roof truss in 1625 and to document it. In particular, two drawings of his are essential for the reconstruction of the ancient s­ upporting struc­ ture (figs. 10-11).17 Both drawings are to scale. A laser scan superimposition has shown that they are accurate down to a few centimetres. The first sheet (C. 25) seems to be a former sketch, while the second one (C. 26) is an elaborate final draw­ ing. The two sheets show different details and prove that not even Borromini was able to record the exact shape of the components and their con­ nections completely until the dismantling was ­under way. The first of the two drawings shows a schematic full section of the roof truss of the pronaos, in which only the right-hand half is rep­ resented in detail (fig. 10). Both the full extent and  virtually every component is provided with measurements in palmi.18 Individual construction 17   Vienna, Albertina; cf. Thelen 1967: vol. I, 32-33; vol. II, C 25 and C 26. 18  1 palmo romano corresponding to 0,2234 m; cf. Thelen 1967: vol. I, 105.

e­ lements are additionally emphasized at an enlar­ ged scale, so as to be able to represent them more accurately, among them being one of the rivets with part of a girder in the top right of the page. In this detail it is also clear that the supporting elements consisted only of thin plates, which were probably each about 2.7 cm thick.19 The second sheet shows a complete cross-­ section of the roof truss above all three naves with additional measurements indicated (fig. 11). A special emphasis is given to the tie beam in the central nave, being marked purple. Unlike what is represented in the older drawings, Borromini was able to recognize that this girder was obviously 19   For a discussion whether the construction consisted only of bronze or of timber beams encased in bronze plates cf. i.a. Valeriani 2006: 111-112 with further bibliography. Inigo Jones comments that the construction of the Pantheon roof truss depicted in a drawing by Palladio was clearly of bronze plates pinned together (“y lame di bronso dobble and pinned together”, Jones 1970: vol. I, 57). See also: Alberti 1485: book VI (o iiii verso): “Extant in hanc usque diem ad porticum Agrippae contignationes aeneis trabibus pedum quadraginta” (in edition Bartoli 1784: VI, 11); Palladio 1570: book 4, XX: “Le Traui del portico sono fatte tutte di tauole di bronzo”; Serlio 1584: book 3, 52v: “Questo ornamento si trova in essere al presente sopra il portico del Pantheon, & è tutto di tauole di bronzo”; Fanucci 1601: 401 (book 4, 36): “Innanzi a essa porta sta un superbo portico sostenuto da grosse, & grandi colonne di marmo coperto tutto con traui, trauicelli, & teuole di metallo senza alcun legname, o altra materia.”

68

DOROTHEE HEINZELMANN, MICHAEL HEINZELMANN

Anejos de AEspA LXXVII

Fig. 11. Francesco Borromini 1625, ancient truss of the Pantheon portico (Vienna, Albertina, Thelen 1967 C 26).

trapezoidal with a middle part located higher up. Due to the oblique bend of the lateral sections of the tie beam, it was possible to bring these some­ what parallel to the rafters and thus produce par­ ticularly force-locked connections. In these areas it was thereby possible to produce overlaps with multiple construction units, which ensured con­ siderably higher stability. On the top side of the wall pillars, slabs sup­ port the main girder and short vertical struts, which stabilize this statically important area. While this construction was so unusual that Bor­ romini picked it out in colour, more precise de­ tails seem to have become clear to him only later, presumably during the dismantling. Marked on the left side with a powerful subsequent stroke and picked out as a detail top left, it is illustrated how the elements are connected at this central node: The rafter above the central nave termi­ nates beyond the pillars between central and lat­ eral nave and is connected there with the lower rafter. The trapezoidal tie beam ends above the pillars without touching them, being stabilized at their pivot point by steep oblique struts. In the centre a king post is connecting the trapezoidal tie beam up to the ridge. Borromini’s drawings also tell the size and number of the purlins as well as the respecti­ ve rivet connections. It is

therefore clear that all nodes were secured by ­series of two to seven ­rivets. In a preliminary modelling by the University of Applied Sciences of Dresden, an attempt was made to reconstruct all components of the an­ cient roof truss (fig. 12). Each rafter unit con­ sisted of 18 individual parts, which were joined together by a total of 60 rivets. Arrangement of the individual elements within a rafter unit must have been performed in probably five construc­ tion layers; as a result one rafter unit had a thick­ ness of approx. 13.5 cm. The biggest individual element was the trapezoidal tie beam with a length of approx. 12.6 m and a weight of approx. 1.8 tons. Two such rafter units were joined together by means of the rivets to form one double truss with a distance of approx. 35 cm between them. The endings of each double truss were inserted into the inlets in the masonry provided for this pur­ pose. The total thickness of one such double truss must have been about 60 cm. A total of three of these double trusses, each at a distance of about 4 m from the other, formed the main supporting structure. Laid longitudinally on top of these were 17 U-shaped purlins, at intervals of 1.5 m, which bore the roof panels of the covering pre­ sumably also made of bronze. Similar long bronze

Anejos de AEspA LXXVII

The Bronze Truss of the Portico of the Pantheon in Rome

69

Fig. 12. Rome, Pantheon. Reconstruction of one rafter unit of the ancient bronze truss after Borromini (Oliver Frost, HTW Dresden).

roof panels of a thickness of 1 cm are still intact today on the upper ring of the Pantheon’s cupo­ la.20 All supporting elements must have been cre­ ated as individual cast elements. As the upper wall sections were not at all worked or meant to be visible, a ceiling construc­ tion beneath the roof truss must be assumed. Semi-circular supports on the façade of the inter­ mediate block point to barrel-vaults having been planned (figs. 7-8). As the available space to sup­ port the barrel vaults on the lateral entablature is only a few centimetres wide, it seems to be most plausible to assume the use of bronze plates for the vaulting, suspended from the roof truss struc­ ture. In the central nave the vaulting would have formed the continuation of the upper side of the coffered barrel vaulting made of opus caementi­ cium above the entrance portal. This also explains the complicated solution of the trapezoidal tie beam, because only in this way was semi-circular vaulting possible (fig. 13). However, semi-circular vaulting was not pos­ sible in the lateral naves. Although the round arches occur in the brickwork here as well, the 20  See photographs of the bronze-covered ring in FineLicht 1968: 142-143.

horizontal tie beams of the lateral roof construc­ tion are lower and cut off these arches at the top, so that there would have been no room for semicircular barrel vaulting. Only a slightly segmen­ tal arch-shaped vaulting, such as Fine-Licht also assumed, would be conceivable in the lateral naves. This is also indicated by segmental archshaped traces on the rear of the pediment wall (fig. 14). As these do not match the semi-circular arches arranged on the intermediate block, this points to a re-planning during the construction process.21 This point throws up questions concerning the planning process and the collaboration be­ tween architect and carpenter or metalworker. Though the shape of the ancient roof truss seems to be largely traceable, numerous questions re­ main concerning the construction process and technical aspects. 21   As Mark Wilson Jones proposed in the context of an original design with a taller pronaos and a proportionally higher entablature (see above note 12), it would be plausible that in the original design the roof trusses had enough space to be constructed without the trapezoidal form of the tie beam and with semicircular vaulting in the lateral naves as well. A change of plan is also possibly indicated by preparatory drawings analysed by L. Haselberger and connected to the Pantheon portico; cf. Haselberger 1996: 182-189.

70

DOROTHEE HEINZELMANN, MICHAEL HEINZELMANN

Anejos de AEspA LXXVII

Fig. 13. Rome, Pantheon. Prelimenary reconstruction of the ancient bronze truss and ceiling of the portico (authors).

Fig. 14. Rome, Pantheon. Portico, western aisle looking North (authors).

Anejos de AEspA LXXVII

The Bronze Truss of the Portico of the Pantheon in Rome

71

Fig. 15. Bronze rivet of the Pantheon portico, Antikensammlung der Staatlichen Museen zu Berlin, inventory no. Fr 1765 p (authors).

The only known intact remnant of the origi­ nal structure is one rivet, which is located in the Antikensammlung Berlin (fig. 15).22 Its dimen­ sions are astonishing: it is nearly 53 cm long, the shaft is 6.6 cm thick and it weighs 15.5 kg. At one end it exhibits a round head, the other end shows slight warping and chisel traces, which probably arose during the dismantling of the roof truss (figs. 16-17). In addition, the rivet shows a coat of paint, which is omitted at all zones at which gir­der elements were fixed. This coat of paint confirms the representation on the Renaissance drawings, which show the double girders with an interval in between. Based on the chemical composition of the rivet’s alloy executed by the Antikensammlung Berlin,23 cast bodies were prepared by the Insti­ tute of Physical Metallurgy of the RWTH Aachen University, and their physical properties were ana­ lysed. The results show that the bronze used on the Pantheon had very similar properties to mod­ ern bronze, for example in relation to compressive and tensile strength. However, it has not yet been possible to determine one factor, porosity, which could have had a considerable effect on loadbearing capacity. To this end we are planning an experimental recreation of the production pro­ cess, in co-operation with the Institute of Casting 22   Inventory no. Fr 1765 p; see Heres 1982: 196-197 with fig. 3 and Pl. 30.14. 23   Peltz 2011: 29-31.

Technology of the RWTH Aachen University. In order to pursue the question of how the construc­ tion process may have been performed, we are ad­ ditionally planning to re-cast individual parts of the supporting structure and to test during practi­ cal experiments how the riveting may have been carried out. Clarifications would include, for ex­ ample, whether riveting was done cold or hot. Based on this information, a concluding eval­ uation of the static behaviour of the roof truss is to take place, which will be realized by the Bran­ denburgische Technische Universität Cottbus (Prof. Dr.-Ing. Werner Lorenz). Overall, we have a construction that remains without comparison, both in design and in mate­ riality. The bronze truss as it can be traced now differs in many respects from known ancient tim­ ber roof structures.24 But as a metal construction as well it seems so far to be without parallel. Al­ though there are written sources indicating the use of metal in roof constructions in antiquity, as for example Janet DeLaine was able to note for the Baths of Caracalla, the remnants are too few to be able to reconstruct their appearance with certainty.25 Therefore, the roof truss of the Pan­ theon remains to this day a unique item in the construction history of ancient roof supporting structures.

24 25

  Cf. Valeriani 2006; von Kienlin 2011.   DeLaine 1987: 147-156 with further bibliography.

72

DOROTHEE HEINZELMANN, MICHAEL HEINZELMANN

Anejos de AEspA LXXVII

Fig. 16. End of bronze rivet of the Pantheon portico with rests of paint coating and chisel traces from dismanteling, Antikensammlung der Staatlichen Museen zu Berlin, inventory no. Fr 1765 p (authors).

Fig. 17. Head of bronze rivet with traces of paint coating and negative of girders, Antikensammlung der Staatlichen Museen zu Berlin, inventory no. Fr 1765 p (authors).

Anejos de AEspA LXXVII

The Bronze Truss of the Portico of the Pantheon in Rome

REFERENCES Alberti, L. B. 1485: De re aedificatoria. Accuratissime impressum opera magistri Nicolai Laurentii Alama­ ni, Florentiae. Bartoli, C. 1784: I dieci libri di architettura di Leon Battista Alberti. Giovanni Zempel, Roma. De Fine-Licht, K. 1968: The Rotunda in Rome. A study of Hadrian’s Pantheon. Gyldendal, Kopenhagen. DeLaine, J. 1987: “The ‘cella solearis’ of the Baths of Caracalla. A reappraisal”, Papers of the British School at Rome, 55, pp. 147-156. DeLaine, J. 2015: “The Pantheon builders: estimating manpower for construction”, in Marder, T. A. and Wilson Jones, M. (eds.), The Pantheon from Antiqui­ ty to the Present, pp. 160-192. Cambridge Universi­ ty Press, Cambridge. Fanucci, C. 1601: Trattato di tutte l’opere pie dell’alma citta di Roma. Per Lepido Facij & Stefano Paolini ad instanza di Bastiano de’ Franceschi, Roma. Grasshoff, G., Heinzelmann, M. and Wäfler, M. (eds.) 2009: The Pantheon in Rome. The Bern Digital Pantheon Project, Pantheon 2. Lit Verlag, Wien. Haselberger, L. 1996: “Die Fronthalle des Pantheon: Ein Werkriss des Dachstuhls?”, in Schwandner, E.L. (ed.): Säule und Gebälk. Zu Struktur und Wand­ lungsprozess griechisch-römischer Architektur, pp. 182-189, Diskussionen zur antiken Bauforschung 6. Von Zabern, Mainz. Heinzelmann, M. 2009: “Il Pantheon”, in von Hes­ berg, H. and Zanker, P. (eds.), Storia dell’architettu­ ra italiana. Architettura romana. I grandi monumenti di Roma, pp. 142-151. Electa, Milano. Heres, G. 1982: “Beiträge zur antiken Bronzekunst”, Staatliche Museen zu Berlin. Forschungen und Berichte, 22, pp. 196-197. Hetland, L. 2007: “Dating the Pantheon”, Journal of Roman Archaeology, 20, pp. 95-112. Hetland, L. 2015: “New perspectives on the dating of the Pantheon”, in Marder, T. A. and Wilson Jones, M. (eds.), The Pantheon from Antiquity to the Present, pp. 79-98. Cambridge University Press, Cambridge. Jones, I. 1970: Inigo Jones on Palladio: being the notes by Inigo Jones in the copy of I quattro libri dell’ar­ chitettura di Andrea Palladio, 1601, in the library of Worcester College, Oxford, vol. I and II. Repro­ duced by Courtesy of the Provost and Fellows. O ­ riel Press, Newcastle upon Tyne.

73

Kienlin, A. von (ed.) 2011: Holztragwerke der Antike. Internationale Konferenz 30. März.-01. April 2007, Byzas 11. Ege Yayınları, Istanbul. La Rocca, E. 2015: “Agrippa’s Pantheon and its ori­ gin”, in Marder, T. A. and Wilson Jones, M. (eds.), The Pantheon from Antiquity to the Present, pp. 4978. Cambridge University Press, Cambridge. La Rocca, E. and Virgili, P. 1999: “Pantheon”, in Steinby, E. M. (ed.), Lexikon Topographicum Urbis Romae V, pp. 280-285. Quasar, Roma. Lapenna, M. C. 2006: “Indagini archivistiche. Ar­ chivio Centrale dello Stato: Regesto 1875-1927”, in Belardi, G. (ed.), Il Pantheon. Storia, tecnica e restauro, pp. 153-164. Soprintendenza per i beni ar­ chitettonici e per il paesagio, Roma. Palladio, A. 1570: I quattro libri dell’architettura. Do­ menico de’ Franceschi, Venezia. Peltz, U. 2011: Nägel, Stifte, Niete, BAR International Series 20266. Archaeopress, Oxford. Rice, L. 2008: “Bernini and the Pantheon Bronce”, in Satzinger, G. and Schütze, S. (eds.), Sankt Peter in Rom 1506-2006. Beiträge der internationalen Tagung vom 22.-25. Februar 2006 in Rom, pp. 337-352. Satz­ inger, München. Serlio, S. 1584: Tutte l’opere d’architettura. Francesco de’ Franceschi, Venezia. Thelen, H. 1967: Francesco Borromini – Die Hand­ zeichnungen, Veröffentlichungen der Albertina 2. Akademische Druck und Verlagsanstalt, Graz. Valeriani, S. 2006: Kirchendächer in Rom. Imhof, Pe­ tersberg. Wilson Jones, M. 2009: “The Pantheon and the phas­ ing of its construction”, in Graßhoff, G., Heinzel­ mann, M. and Wäfler, M. (eds.), The Pantheon in Rome. Contributions to the Conference Bern, Novem­ ber 9-12, 2006, pp. 69-87, Pantheon 1. Bern Studies, Bern. Wilson Jones, M. 2015: “Building on adversity: the Pantheon and problems with its construction”, in Marder, T. A. and Wilson Jones, M. (eds.), The Pan­ theon from Antiquity to the Present, pp. 193-230. Cambridge University Press, Cambridge. Yerkes, C. Y. 2013: “Drawings of the Pantheon in the Metropolitan Museum’s Goldschmidt Scrapbook”, Metropolitan Museum Journal, 48, pp. 87-120. Ziolkowski, A. 1999: “Pantheon”, in Steinby, E. M. (ed.), Lexikon Topographicum Urbis Romae IV, pp. 54-61. Quasar, Roma.

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

“ARMORED” COLUMNS IN THE ROMAN IMPERIAL PERIOD CARLA MARIA AMICI Dipartimento di Beni Culturali, Università del Salento

ABSTRACT: In the Roman imperial period several high tech applications of metallurgy were ex­ ploited in construction. To reinforce columns iron rods were used, inserted in holes made by a tubu­ lar drill the length of the column shaft. The rods were anchored at the top or at the bottom, creating a sort of armored column as a means to reinforce faulty supporting elements, demonstrating a high degree of specialization and workmanship. The cost/benefit ratio of this specialized work was advan­ tageous only in particular contexts; it was actually used only in periods of great demand for valuable marbles, when there was a concomitant development of workshops and expertise at the quarries of origin as well at the places of destination, of storage and of distribution of the material; is not like­ ly that such a specific and difficult process was performed at the construction site, involving a very specialized workmanship and high technical expertise. KEYWORDS: Building archaeology, Iron technology, High tech devices. RESUMEN: En la construcción de época romana imperial se explotaron diferentes aplicaciones metalúrgicas de alta tecnología. Para reforzar columnas se utilizaron barras de hierro, insertadas en orificios realizados con un taladro tubular de la longitud del eje de la columna. Las barras se anclaban en la parte superior o en la parte inferior, creando una especie de columna armada para reforzar elementos de apoyo defectuosos y demostrando un alto grado de especialización de la mano de obra. La relación costo/beneficio de este trabajo especializado resultó ventajosa solo en contextos particulares. Se utilizó exclusivamente en períodos de gran demanda de mármoles valiosos, cuando se produjo un desarrollo simultáneo de talleres y expertos en las canteras de origen, así como en los lugares de destino, de almacenamiento y de distribución del material. No es probable que un proceso tan específico y difícil se realizara en el lugar de la construcción, implicando una mano de obra muy especializada y de alta experiencia técnica. PALABRAS CLAVE: arqueología de la construcción, tecnología del hierro, componentes de alta tecnología.

In the Roman imperial period metallurgy reached a high degree of complexity and speciali­ zation. In spite of the difficulty of finding remains of metal still in place, it is possible to demonstrate that sometimes iron was used in construction in a very peculiar way, taking advantage of its strength without revealing its presence.1 1   This subject, presented here in preliminary form, will be developed in a forthcoming publication. All illustrations and

There is evidence for several columns, mono­ lithic or composed of a limited number of drums of different heights that have two or three holes made by a tubular drill along the major axis of the shaft.2 Long rods, square in section, were drawings are by author. For general references on metallurgy see Forbes 1964-1972; Tylecote 1976 (1992); Healey 1978; Loiseau 2005: 117-129. 2   The use of a tubular drill is assumed also by Leon Battista Alberti, De Re Aed., lib. X, VII, II, 930: “Vidimus marmora

76

Carla Maria AMICI

i­nserted inside the holes, to obtain a resistant frame in order to avoid flaking and cracking in sectors where the column, of expensive colored marble, was not uniform in quality and conse­ quently had varying degrees of resistance to com­ pressive load. The rods are always inserted at the same distance along the diameter of the shaft when two holes are present, and at a homogene­ ous distance from the center in case of three holes, in order to provide evenly distributed reinforce­ ment. From the analysis of each case, it seems that this device was used both during the original construction and during phases of restoration. Not considering cases of single or sporadic col­ umns, not surely related to specific buildings, and therefore of little value for a broad understanding of the device, there are four cases that provide ­sufficient details for a correct interpretation. In the Roman colonia of Privernum, near ­Latina, Italy,3 the theater was not anticipated in the original planning of the town, but was under construction in the Augustan period; the building has subsequently underwent several changes. In the columns of the frons scaenae dating from the end of the 1st and the beginning of the 2nd century AD, two or three iron rods were in­ serted in holes made by tubular drill, evenly dis­ tributed along the major axis of the shaft. In the best preserved example, two drums of Africano of different lengths, with maximum pre­ served length of 0.43 m and 1.20 m, are connected by two rods, at least 60 cm long and with a cross section of 2 × 2 cm, inserted in holes of 3.8 cm in  diameter (fig. 1). To achieve the connection ­between the two drums, two rectangular pegs (3.5 × 1.5 cm; depth 2.2 cm) have been used, and two light external iron cramps (8.5 + 11 cm; width 1.2 cm), and not leaded dowels as usual, allowing slight movements and adjustments during d ­ rilling. The forged iron rods were inserted in holes up to 1.60 m long, in any case far enough to reach an area providing a satisfactory resistance, and were anchored in the lower part of the column. Before insertion in this part, the head of the rod was bent at 90° for about 9 cm, hot hammered to flatten it suitably, and then lodged into the corresponding recess that had been chiseled on the lower face of the column (fig. 2). pedes longa plus duodecim a summo ad imum traiecto foramine pervia lato palmum: quod effecisse ænea fistula tornatili et  harena aptissimis coniecturis indiciisque ex ipso lapide intelleximus.” 3   Cancellieri 1998; Amici 2012: 565-581.

Anejos de AEspA LXXVII

Fig. 1. Privernum, (Latina, Italy). Drums in Africano marble of a column of the frons scaenae of the theatre. f = holes for iron rods; t = cutting for pegs; g = light external cramp.

All the other columns were partly or com­ pletely reduced to pieces, probably to recover the metal rods, and used to fill the pit of the auleum of the theater in the early Middle Ages, as clearly indicated by the connected stratigraphy. Of the remaining 350 pieces of minute size, 89 have drill holes for the insertion of metal rods, closely comparable with those present in the two drums of Africano. Ten pieces, five of Africano and five of Giallo antico retain two holes, equidis­ tant on the diameter, and two, one of Africano and one of Giallo antico, retain three, evenly spaced with respect to the center of the column, providing a uniformly distributed support. The diameter of the drill holes ranges from about 3 to about 4 cm, and presumably in relation to the size of the rod and to the size of the column. Two ­series of columns in Giallo antico and two in Afri­ cano can be reassembled (with diameter of 0.45 m and 0.35 m); in connection with the remains of the theater still in place they allow for a well founded reconstruction of the frons scaenae. The same device was applied in the frigidar­ ium of the Antonine Baths in Carthage, where two series of monolithic columns in red granite, 0.9 m and 0.6 m of diameter, have two holes made by tubular drill, running along the major axis of the shaft (figs. 3-4). The Antonine Baths are in a very poor condition, and only the underground services are preserved, but it is possible to relocate the columns at the entrance of the frigidarium and in front of the norther cold water tub. In both cases, the columns supported only an entablature, topped by a relieving arch. The red granite marble seems of very good quality; in this instance the iron rods were probably used during a phase of

Anejos de AEspA LXXVII

“Armored” columns in the Roman imperial period

77

Fig. 2. Lower part of drum A; cutting for the head of rods 1 and 2, bent and flattened. Inv. 3220: example of cuttings for the head of the rods (f) in the upper part of a column. A residue of iron is still in place. X = cutting for dowel, with channel for lead.

restoration, to reconnect two or more sections of broken column, without shortening the broken sections or leveling the contact surfaces. In the Villa of Trajan in Arcinazzo,4 a large and sophisticated building abandoned before being completed, and therefore never used, two columns found in the area of the peristyle display the same technical solution.5 One of the columns, in Giallo antico of poor quality, with a maximum length of 1.85 m, and a diameter of 0.50 m at the bottom, has two parallel holes drilled along the major axis of the shaft, equidistant on the diameter (fig. 5). The holes, preserved for a remaining length of 0.8 m and 1.30 m, have a diameter of 5 cm, suitable for metal rods with 3/3.5 cm of section. The second column, in Portasanta, with a maximum length of 1.90 m, and a diameter of 0.45 m at the bottom, has three 4 cm diameter drill holes, two of which partially overlap (fig. 6). Probably only two holes were planned, equally spaced along the diameter of the column; during the drilling one of  them veered off line, making it necessary to repeat the operation. If, as it seems likely, the columns be­ longed to the west side of the peristyle of the ­villa, and to the main entrance of the great hall,   Fiore and Mari 2003; Fiore and Mari 2008: 81-90.   The local archaeological museum houses a fragment of the lower part of another column in Portasanta, with a single hole for metal rod, diameter cm. 5. On the basis of the previous examples, the presence of another hole in a symmetrical position with respect to the center can be safely assumed. 4 5

they supported only wooden beams and the roof terrace. Finally, in the eastern palaestra of the Baths of Trajan in Rome a fragment of a column in Porta­santa, preserved for a length of 1.5 m, with a diameter of 0.88 m, has three longitudinal holes, approximately parallel to the major axis of the shaft, with a maximum length of 1.17 m and about 6 cm in diameter (fig. 7). They are roughly equidistant, at about 30 cm from each other; in a virtual reconstruction of the horizontal section of the shaft they are arranged with a very similar disposition of the three dowels usually used to connect two drums of a column, optimizing the distribution of the links. The shaft fracture allows to appreciate the thin circular inci­ sions left by the tubular drill along the internal surfaces of the holes, and a very clear groove left by its end at the bottom of one of the holes. The column belongs to the portico of the palaestra, where columns supported flat arches reinforced by metal bars; in this case it is not possible to estab­lish if the iron rods have been used during the original construction or during a later ­restoration. In all these cases the dimension of the rod seems to vary according to the dimension of the column; the rods were inserted, not embedded, to prevent cracks in the marble shaft in contact with iron, the empty space probably filled with wet sand or other material with similar characteristics of adaptability.

78

Carla Maria AMICI

Anejos de AEspA LXXVII

Fig. 3. Carthage (Tunisia), Antonine Baths, columns of the frigidarium. All the columns have two tubular holes for the insertion of metal rods (A-G). A further fragment (H) can be found a few meters away.

Fig. 4. D = lower part of a column, diameter 0.59 m; 0.66 m max length preserved; with a central hole for a dowel. The distance between the two longitudinal holes, almost parallel, is approximately 28 cm; the distance from the circumference of the column is about 16 cm. Not surprisingly, the surface of the lower part of the column is flaked in relation to the cutting for the heads of the rods bent at 90°. G = upper part of a column, diameter 0.51 m; 1.88 m max length preserved. The terminal part of two holes for metal rods are preserved, max length preserved 7.4 cm. The distance between the two holes is about 15 cm; the distance of each hole from the outer circumference of the column is about 18 cm.

Anejos de AEspA LXXVII

“Armored” columns in the Roman imperial period

Fig. 5. Arcinazzo (Rome, Italy), Villa of Trajan. Column in Giallo antico, of poor quality, diameter at the lower part 45 cm, with two parallel holes drilled along the major axis of the shaft, equidistant on the diameter. Max length preserved: 0.8 and 1.3 m, diameter 5 cm.

Fig. 6. Arcinazzo (Rome, Italy), Villa of Trajan. Fluted column in Portasanta, diameter at the lower part 45 cm, with three parallel holes drilled along the major axis of the shaft. The holes are almost completely obstructed by earth, but a length of 0.40 m at least can be measured; diameter 4 cm.

79

80

Anejos de AEspA LXXVII

Carla Maria AMICI

Fig. 7. Trajan’s Bath (Rome, Italy), eastern Palestra. Column in Portasanta with three holes for longitudinal metal rods. Hole B retains the imprint left by the head of the tubular drill.

The rods were housed far enough in the verti­ cal holes to reach an area which could ensure a satisfactory degree of resistance. In this way the transmission of loads on the columns to the found­ ations bypassed the segment of weak resistance, and ensured the stability of the shaft. Resorting to such a complex mechanism was very demanding. The overall analysis of the re­ cords currently available, reveals that the ­cost/bene­ fit ratio of this practice was advantageous only in particular contexts. This kind of specialized work was actually used only in periods of great demand for valuable marbles, apparently throughout the 2nd century AD, when there was a concomitant de­ velopment of workshops and expertise at the quarries of origin as well at the places of destina­ tion, of storage and of distribution of the mate­ rial. From this point of view, a similar problem is offered by columns with faulty surface revetted with marble slabs fixed with iron dowels,6 or 6   Fant 2001: 175-176; De Nuccio and Ungaro 2003: 528, n. 288 (M. Bruno).

c­ olumns supplemented or repaired with segments with sinusoidal linking:7 it is not likely that so spe­ cific and difficult process was performed on the construction site, involving very specialized work­ manship and high technical expertise (fig. 8). Equally surprising and demanding is this spe­ cific use of the tubular drill, for holes of more than 1.3 m in length, and up to 9 cm in diameter, using tubes of increasing length, in relation to the  depth and the size of the hole.8 Generally drilling was carried out by placing the object to be drilled vertically, making the most of the force ap­ plied by  the operator; usually, abrasive was in­ serted from the upper end of the tube. In this case,   Hoffmann 2007-2008: 105-110.   Bessac 1986: 226-228, 241-244. At present, no tubular drill suitable for large holes is recorded, but a small copper tubular drill, 1.8 cm in diameter, used for surgery on the skull, is recorded in a tomb in Bingen (Germany), dated between the late 1st and early 2nd century AD. It corresponds perfectly to the type of drill called “modiolus” and whose structure and functioning are described with precision by Celsius, De med. VIII, 3, 1-3. cfr. Künzl 1996, 2451; abb. IV, page 2587; Künzl 2002: 5. 7 8

Anejos de AEspA LXXVII

“Armored” columns in the Roman imperial period

81

Fig. 8. A = Basilica Ulpia, Trajan’s Forum (Rome, Italy). Column in gray granite with sinusoidal linking to the next part. B = Ostia antica (Rome, Italy). Column in Africano revetted with marble slabs fixed with dowels. C = Privernum (Latina, Italy). Circular grooves left by the tubular drill on the internal surface of a hole for iron rod. D = Privernum (Latina, Italy). Circular cutting left by the tubular drill head.

it should be considered that the drill must have been operated in a vertical position, with an inclin­ ation not exceeding 30°, otherwise the sand would not act uniformly. It is possible, but as yet not documented, that tubular drills existed in which abrasive grains, for example corundum, were in­ corporated in the head, as in some modern saws. Considering both the time and effort required to drill marble or granite columns, raises the possi­ bility that hydropower was employed, as is now taken for granted with marble saws;9 in this case placing horizontally the column to be drilled seems most appropriate. This could better explain the cases with errors, with converging or overlap­ ping perforations, and support the suggestion of the preparation of reinforced columns in work­ shop rather than in the construction site. Additionally, it should be noted that the com­ bination of tubular drill holes and metal vertical elements has other, sporadic, applications; for in­ stance, to ensure the anchoring of metal decora­  Ritti et al. 2007: 138-163.

9

tive elements on marble blocks, as in the Basilica Ulpia in Trajan’s Forum (fig. 9).10 From a comparative analysis of all the exam­ ples available at present, it is clear that this solu­ tion has been used, regardless of the type and function of the building, during the original con­ struction, and sometimes also during restora­ tions, to reinforce columns of valuable material, but it was never used for columns subject to heavy load or in a critical position. It created a sort of “armored” column as a means to reinforce faulty supporting elements, bypassing the segment of weak resistance, assur­ ing the stability of the shaft without resorting to  external clamps that would have been visible, 10  Amici 1982: 24. Careful examination is required of tubular drill holes sporadically found in entablatures or in columns, often not related to iron rod. For instance, I think that the drill holes, approximately 2.4 cm in diameter, present in some of the architectonical elements of the front of the Nympheum of the lower triclinium of the Domus Augustana, are to be connected with the passage of water and not with the insertion of iron elements. Von Hesberg 2004: 63-65.

82

Carla Maria AMICI

Anejos de AEspA LXXVII

Fig. 9. Basilica Ulpia, Trajan’s Forum (Roma, Italy). Entablature of the attic with four vertical holes made by tubular drill, for anchoring metal decorative elements. Length of the holes: 30 cm; diameter 8/10 cm.

Fig. 10. Haghia Sophia, Istanbul (Turkey). Columns with external clamps, in iron and bronze, put in place during different periods to ensure stability and consistency to columns. Highlighted two bronze clamps of the original setting of the columns of the basilica.

Anejos de AEspA LXXVII

“Armored” columns in the Roman imperial period

a­ ltering the perception of the integrity of the mate­ rial, the smoothness of the surface, and therefore the elegance of the architectural element; this lat­ ter technique became standard procedure in later periods, due to its speed and ease of use (fig. 10). In the Roman imperial period, on the con­ trary, according to a very characteristic Roman building rule, the complex mechanism allowing an edifice to stand was to be hidden, not revealed. REFERENCES Amici, C. M. 1982: Foro di Traiano: Basilica Ulpia e Biblioteche. Roma. Amici, C. M. 2012: “A cloaca maxima in the Roman town of Privernum, Italy: the project, the plan, the construction”, in Carvais, R., André, G., Nègre, V. and Sakarovitch, J. (eds.), Nuts & Bolts of Construc­ tion History. Culture, Technology and Society, pp. 565-581. Picard, Paris. Bessac, J.-Cl. 1986: L’outillage traditionnel du tailleur de pierre de l’Antiquité à nos jours, Revue Archéologique de Narbonnaise suppl. 14. CNRS, Paris. Cancellieri, M. 1998: Privernum. L’area archeologica. “L’Erma” di Bretschneider, Roma. De Nuccio, M. and Ungaro, L. 2003: I marmi colorati della Roma imperiale. Marsilio, Venezia. Fant, J. C. 2001: “Rome’s marble yards”, Journal of Roman Archaeology, 14, pp. 167-198. Fiore, M. G. and Mari, Z. 2003: Villa di Traiano ad Arcinazzo Romano. Il recupero di un grande monu­ mento. Grafica Ripoli, Tivoli. Fiore, M. G. and Mari, Z. 2008: “La villa di Traiano ad Arcinazzo Romano”, in Valenti, M. (ed.), Resi­ denze imperiali nel Lazio. Atti della Giornata di studio (Monte Porzio Catone, 3 aprile 2004), pp. 81-90. Cit­ tà di Monte Porzio Catone, Monte Porzio Catone.

83

Forbes, R. J. 1964-1972: Studies in Ancient Technology. Brill, Leiden. Healey, J. F. 1978: Mining and Metallurgy in the Greek and Roman world. Thames and Hudson, London. Hesberg, H. von. 2004: “Die Domus Imperatoris der neronischen Zeit auf dem Palatin”, in Hoffmann, A. and Wulf, U. (eds.), Die Kaiserpaläste auf dem Palatin in Rom. Das Zentrum der römischen Welt und seine Bauten, pp. 63-65. Von Zabern, Mainz. Hoffmann, V. 2007-2008: “Die geflickten Säulen des Pantheons und ihre Schwestern“, Boreas, 30-31, pp. 105-110. Künzl, E. 1996: “Forschungsbericht zu den antiken medizinische Instrumente”, in Aufstieg und Nieder­ gang der römischen Welt, II, 37, 3, pp. 2433-2639. De Gruyter, Berlin. Künzl, E. 2002: Medizinische Instrumente der Römischen Kaiserzeit im römisch-germanischen Zentralmuseum. Römisch-germanisches Zentralmuseum, Mainz. Loiseau, C. 2012: “Les métaux dans les constructions publiques romaines. Application architecturales et structures de production (ier-iiie siècle ap. J.-Ch.)”, in Camporeale, S., Dessales, H. and Pizzo, A. (eds.), Arqueología de la construcción III. La economía de las obras (Paris, 10-11 de Diciembre de 2009), pp. 117-129, Anejos de Archivo Español de Arqueología 69. CSIC, Madrid-Mérida. Ritti, T., Grewe, K. and Kessener, P. 2007: “A relief of a water-powered stone saw mill on a sarcophagus at Hierapolis and its implications”, Journal of Ro­ man Archaeology, 20, pp. 138-163. Thébert, Y. 2003: Thermes romains d’Afrique du Nord et leur contexte méditerranéen. Études d’Histoire et d’archéologie, Bibliothèque des Écoles françaises d’Athènes et de Rome 315. Écoles françaises de Rome, Roma. Tylecote, R. F. 1976 (1992): History of Metallurgy. Institute of Materials, London.

III PRODUCTION, SUPPLY AND USE OF BRICK AND TILE

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

THE SOCIAL LIFE OF TILE IN THE ROMAN WORLD PHIL MILLS Freelance/Hon. Research Fellow University of Leicester

ABSTRACT: This is a brief survey around the Greco-Roman world of evidence for the social status of tile makers and the social importance of ceramic building material (CBM), especially roof tile. A review of the economic evidence of CBM suggest it had a higher cost than is sometimes assumed, and that it could be a marker of social status. Evidence suggests that tile makers had different statuses in different regions. In Britain, tile makers are free, but production moves from itinerant tile makers to more centralised production. In the East, Italy and North Africa they appear to be mainly slaves working on large scale specialist ceramic sites. In Sagalassos and Bulgaria they are farmers, paying rents in brick and tile, but with higher literacy. In all cases tile makers had to exist as part of a community of specialist Roman builders and architects, and required patrons. KEYWORDS: Ceramic Building Material (CBM), Roof tile, Brick, Economics, Quantified assemblages, Pottery, Urbanisation, Lebanon, Syria, Bulgaria, Turkey, Italy, Tunisia, Britain. RESUMEN: Se trata de un breve estudio sobre la evidencia de la condición social de los productores de latericio y la importancia social del material cerámico de construcción en el mundo grecorromano, especialmente la teja. Una revisión de los datos económicos del material cerámico de construcción sugiere un coste más alto del que a veces se supone que, además, podría ser un indicador de posición social. Las pruebas revelan que los fabricantes de latericio tenían diferentes estatus sociales en diferentes regiones. En Britania, los fabricantes son libres, y la producción varía de productores de material latericio itinerantes a una manufactura más centralizada. En el este, en Italia y en el norte de África parecen ser principalmente los esclavos que trabajan en sitios a gran escala, especializados en cerámica. En Sagalassos y Bulgaria son los agricultores que pagan rentas en ladrillos y productos latericios, con una más alta alfabetización. En todos los casos, los productores tuvieron que existir como parte de una comunidad de constructores romanos especialistas y arquitectos, siendo necesaria la presencia de promotores. PALABRAS CLAVE: material cerámico de construcción, teja, ladrillo, economía, cuantificación de  la producción, cerámica, urbanización, Líbano, Siria, Bulgaria, Turquía, Italia, Túnez, Gran ­Bretaña.

INTRODUCTION In recent years there has been a flourishing of interest in ceramic building material (CBM), representing as it often does a very large part of the Roman archaeological record. This recent research shows a number of problems with the tra-

ditional idea that CBM is a high bulk, low value commodity that is essentially locally made (e.g. Peacock 1984; Tomber 1987), echoing now long overthrown ideas about coarse utilitarian pottery in the Roman world (Orton et al. 1993: Table 1.1). In particular it has become apparent that large quantities of CBM, especially roof tile, are trans-

88

Anejos de AEspA LXXVII

Phil Mills

Fig. 1. Map showing principle locations mentioned in the text (by the author).

ported in bulk around a number of regions. Brick generally has a more restricted, local distribution although some forms would also appear to have wider distributions (e.g. Towcester pink grog tempered flue tile, Mills 2013b). This paper will start with a summary of the economics of the ceramic building material and their social frame work; this is followed by a brief survey of how this is reflected in material recovered from around the Roman world, largely based on the author’s own work. The principle sites referred to are located on the map in figure 1. THE ECONOMICS OF CBM AND ITS SOCIAL CONTEXT There are some 40 ship wrecks from Parker’s (1992) survey which have tile as a primary cargo – some 4% of the whole gazetteer, or 3% of the sample Parker (1992) examines. This compares with c. 2% for stone cargoes. Study of the geographical location of these ship wrecks (Mills 2013a: fig. 1.2) suggests a number of clusters, including Frejus in France and the central Adriatic Sea. Parker (1992) presents the evidence, based on fabric and stamp analysis, for material being exported from Fréjus, and it would seem plausible that the cluster in the central Adriatic is indicative of a regional trading network as well. To these

frameworks we can add the major industry centred on Cilicia which fed the Levant, including Ras el Bassit, Homs (Newson et al. 2012), Beirut (Mills 2013a) and Tyre (Mills 2013a); and an apparent industry based in western Bulgaria supplying bricks along the Danube. In Italy, Campania was an important supplier of CBM to central ­Italy (Mills and Rajala 2011a), but was also supply­ing Tunisia and elsewhere in North Africa (Graham 2011; Mills 2013b). In Britain, the amount of work on developer-funded projects since the early 1990s has allowed the identification of a number of regional supply zones (Mills 2013b). Early attempts to explain this pattern suggested that such examples of the long distance transportation of CBM could be explained by the distorting effect of the annona. In this model state subsidised cargoes of grain were transported from North Africa, and on the return trip empty ships could be loaded with roof tile as ‘ballast cargo’ (Tomber 1987). A number of objections can be raised to this model, and are detailed in Mills (2013a). These include that this model does not really apply to the situation in Beirut,1 with CBM travelling the wrong way (along with oil and wine from Cilicia). Interestingly in the case of Carthage, relatively few buildings in the city actu1  Report submitted N. Baudray Rimourski, University Quebec 2008.

Anejos de AEspA LXXVII

The Social Life of Tile in the Roman World

ally benefitted from tiled roofs, as opposed to flat or domed roofs, primarily of opus signinium (c.f. Hurst and Gibson 1994). This contrasts with ­urban centres in Italy, the Levant, the Balkans and Britain where the majority of roofs appear to have been tiled. A number of functional explanations can be entertained for why CBM may be produced at specialist centres and then transported over long distances: namely the need for skilled tile makers, the correct raw materials, and the space in which to form dry, fire and stack large quantities of brick and tile (Mills 2013a). The transportation and construction with CBM require skilled workers, as the loading and unloading of a cargo of several thousands of roof tiles would not be a trivial affair. In addition the production of tile required a large infrastructure of building designers who knew how to use ceramic building ma­ terial for a desired aesthetic effect, and skilled builders who knew how to use it, how to build with it, and had patrons who were willing to pay for them to do it and wanted to have buildings in the ‘Roman style’. Given this, it is worth revisiting the assumption that roof tile was inexpensive. There is a surviving section of the edict of Diocletian (Section 16, 20-24 in Erim and Reynolds 1973: 103 and 108;  cf. Giacchero 1974: Section 15, 88-92) that has some prices for CBM remaining. Unfortunately the price for roof tile does not survive, although reference is made to ‘best quality tegula with im­ brex’. Other ceramic building materials (e.g. bricks) survive with a cost, and Warry (2006: 122) has suggested a cost of c. 4 denarii for a tegula based on the relative amounts of clay used in the forms for which prices exist. Mills (2013a) used a similar figure, for his estimate. This cost of 4 dena­ rii is equivalent to a single day’s unskilled labour. This would make even a modest roof (300 tiles for a c. 9 × 12 m house) 1200 denarii – a substantial outlay. Given that the most sensible way to transport the material is by the roof-load, with some excess to cover breakages, this would make a cargo of tile potentially quite profitable as a commodity in its own right. Mills (2013a: fig. 5.5) shows that by volume it can be compared to the cost of wine and other relatively high cost commodities. Whilst the cost of roof tile in Roman pe­r­ iod  does not survive, the cost in the Hellenistic world does. Wikander (1988) has a useful summary of evidence, both directly from inscriptions on the tiles themselves and from literature. In-

89

scriptions dating from 350-180 BC suggest prices ranging from 2.5 obols to 1 drachma, 3 obols each, to 2.5 obol – 3 drachma a pair,2 and confirmed from sources such as e.g. Herondas III The School Master: “I have to pay three half-oboli for each tile” (after Wikander 1988), which comes to the equivalent of the cost of 1 day’s unskilled labour (given as ½ Dr. or 3 obols). This is interestingly consistent with the cost of tile in the Roman period suggested above. There is more evidence surviving about the expected wages of the tile maker, with the brick maker to be paid 2 denarii for 4 fired bricks and preparation of the clay (Giacchero 1974: Section 7, 15-16). Given an estimated output of c. 200 bricks a day (c.f. Warry 2006: 119) this compares well with the expected wages of skilled labourers e.g. ‘stone mason with maintenance 50 denarii’, i.e. equivalent to 100 bricks (Giacchero 1974: Section 7, 2). The evidence then is that roof tile is very expensive and tile makers are well rewarded skilled labourers. Why then should there be such an explosion of the use of roof tile in urban and rural settings around the Hellenistic-Roman world from around the 2nd century BC? Wikander (1988) suggests that the main reason would have been to stop fires being so devastating in densely built up urban cores. Brodbribb (1989: 7-8) references the Charter of Tarentum, dating from around 89 BC (Crawford 1996: No. 15, I, 26-28): Every person who is or shall be a decurio of the  municipium of Tarentum, or who shall have declared his vote in the senate of the said municipium, shall in all good faith possess a house of his own in the town of Tarentum or within the territory of the said municipium, roofed with not less than 1500 tiles.

and Section 14 of the Lex Ursonensis (HispEp 4.825): Whoever would be a decurion in the colony Genetiva Julia, that decurion shall own a building within the area marked round by the plough that has not less than 600 roof-tiles. Whoever would be a citizen of the colony, but not a decurion, within two years of the colony being founded he shall own a building that has not less than 300 roof-tiles.

2

  Based on the material collected by Martin 1965: 81-84.

90

Phil Mills

He also points out the use of the use of tiles to raise war funds, (after Dio Cassius, Hist. XLVI, 31,3) with an assessment of 4 asses per roof tile. An interesting restriction on tile production in the city is also suggested in the Lex Ursonensis (Crawford 1996: No. 25, Section 76, 23-27): 76 “No person shall possess within the town of the colony Julia pottery works or a tile factory of larger size than to produce 300 tiles per day. If any person possesses such works or factory, the said building and ground shall become the public property of the colony Julia, and any person at will shall have the right to claim such building and any magistrate charged with jurisdiction in the colony Genetiva Julia shall pay without malicious deception into the public funds a sum of money equivalent to the value of the same” (translation from Johnson et al. 1961: no. 114, pp. 97-114).

This would suggest that there is some pressure to restrict tile production, maintaining a high price and thus that it is used to mark status. THE SOCIAL CONTEXT OF CBM AROUND THE ROMAN WORLD This section is largely based on the author’s own work in Britain, Syria, Lebanon, Turkey, Bulgaria, Italy and Tunisia, but with reference to other work as appropriate. The methodology employed on the study of CBM is outlined in Mills (2013a) and (Mills 2013c). Where possible social status of the site has been evaluated using the pottery, based on fine ware supply and functional analysis (Evans 2001). Britain A recent overview of the state of knowledge of the civilian usage of CBM in Roman Britain can be found in Mills (2013b) whilst aspects of military usage are explored by Warry (2010). The earliest use of CBM is with the arrival of  legionaries with the Claudian invasion. Early civili­an adoption was confined to retired l­ egionary veterans settling in the province, and the elites of Roman client kingdoms (e.g. Fishbourne palace). There is a big expansion in the use of Roman ­architecture in the mid 2nd century with urban centres and villas being constructed within the limits of the Hadrianic civil boundary (Jones and

Anejos de AEspA LXXVII

Mattingly 1990: Map. 5.11). The nature of the structure of the tile industries appears to depend on the particular civitas capital. For instance the industries around Cirencester (McWhirr and Viner 1978) are very different from those at York (McCormish 2012) and London (Betts this volume). It would be interesting to see if there is a link to where the legions situated at these locations originated from, and the early phases of the local CBM industries. In the region of Cirencester (McWhirr and Viner 1978) there is evidence of municipal tile works with wide distribution from quite early on. Elsewhere the initial production of civilian roof tile appears to be very local. One interesting aspect is in the evidence of roller relief stamped flue tiles (Betts et al. 1994). These are stamps applied by a roller onto a face of the flue tile. In quantified assemblages they occur at c. 2.5%,3 suggesting that only one out of a batch of c. 40 were marked in this way. The most likely explanation is of flue tile makers sharing facilities such as a kiln. It is not uncommon to find more than one stamp from any location (Betts et al. 1994), implying that different tile makers were working side by side on the same projects. There is variation of which stamps are found together, which implies that these itinerant tile makers are free, that is not tied with a specific group of workers, as would be expected for slaves. One interesting example, Die 18 (Betts et al. 1994), shows a herringbone type pattern. There are variants of it noted which clearly show a split in the roller used to mark the wet clay. This split occurred whilst the roller was in use in Leicester (or possibly London). Figure 2 shows a plot of the occurrence of Die 18, in its split and un-split incarnations, alongside Die 9 which is found in many of the same sites in Leicester. This shows the movements of tile makers in the mid to late second century AD. It is interesting that travel is apparently restricted to the cultural boundaries suggested by the Late Iron Age ‘tribal’ boundaries, reflecting a pattern seen in pottery usage (Evans 2013; Evans in press b). These boundaries seem to have affected the development of emerging market economies in the different regions of Britain. It is important to note that a specialist tile maker would not be working on their own: they would have needed to work with architects who knew how to build hypocaust structures, 3  Alchester, report submitted E. Saur, University of Edinburgh 2005.

Anejos de AEspA LXXVII

The Social Life of Tile in the Roman World

Fig. 2. Map showing the find spots for flue tile with die patterns 9, 13, and 13 split, laid over the Roman road system and location of civitas and colonia (by the author).

and perhaps with other specialist manufacturers, ­masons, and carpenters, and those who could build in these materials as well as those who wanted to commission and pay for these structures. After this initial surge in building activity it is interesting to note that tile industries became much more sedentary in the later Roman period, and outside of London the tiled structures appear to have been more commonly built in the countryside. Outside of the south east, tile industries associated with pottery industries can be noted at: Harrold, Bedfordshire;4 Holme on Spalding Moor (Mills 2015); Crambeck, North Yorkshire (Mills in press); Horningsea, Cambridgeshire (Evans et al. in press); they also occur in the Severn Valley ware tradition,5 and with Towcester pink grog tempered wares (Mills 2013). These represent a few of the more easily recognised industries, based on macroscopic fabric analysis, and it is likely that more will be recognised as work continues. It should also be noted that more 4   Leicester Shires development. Analysis report submitted ULAS 2005. 5  Worcester Magistrates Court. Report submitted to Hereford and Worchester Archaeology 2004.

91

local modes of production, such as kilns at villa estates (e.g. Ward 1999) continue through this per­iod as well. Figure 3 shows the location of these tile kilns, the presence of other tile kilns, as well as the distribution of CBM (roof tile and flue tile) identified as coming from Towcester. The distribution of the pottery from the kilns at Towcester has been studied by Taylor (2004) who identified three distribution zones Taylor (2004: fig. 3): a core, a heartland and an outer for the supply of this pottery. The sites where CBM in the same fabric are recorded are: Croughton (Mills 2008); Alchester;6 Alcester (Cheer and Booth 2001); the villa site TR99 H from the Transco Gas pipeline from Churchover to Newbold Pacey (Mills 2010); Salford Priors (Evans 1999); and Worcester Magistrates Court (WMC).7 There is also a possible, presence at Piddington Villa (Ward 1999). The proportion of CBM in the pink grog tempered fabric by number of fragments is shown on the map. This clearly shows that a very similar distribution pattern is emerging for the CBM as for the pottery. It would be interesting to see how this pattern develops with further work, as it would be expected that the marketing of tile, in large batches for single commissions, would have a much more stepped pattern than that of pottery (Hodder 1974). Even though tile makers were now stationary in this period, there was still a requirement for the other specialists to oversee transportation and the building of the structures at the final location, and it would perhaps be organisationally more useful for such specialists to have been based around the tile manufactures so that building works could be commissioned in one piece and carried out relatively efficiently. Further work would hopefully establish if this correlates with other specialist building services, such as the mosaic schools (Jones and Mattingly 1990: Map 6:41). Syria and Lebanon The supply to the Northern Levant was dominated from the 2nd century BC until the early 6th century AD by specialist ceramic production centred in Cilicia. Both roof tile and bricks were 6  Alchester, report submitted E. Saur, University of Edinburgh 2005. 7  Worcester Magistrates Court. Report submitted to Hereford and Worchester Archaeology 2004.

92

Phil Mills

Anejos de AEspA LXXVII

Fig. 3. Location of main tile fabrics discussed in the paper, as well as the distribution of PNK GT tile and pottery (by the author, after Taylor 2004: fig. 3).

made here, although bricks only have a restricted distribution, while roof tile is found in Ras el Bassit,8 Beirut (some 25% of the overall assemblage, Mills 2013a), Tyre (Mills 2013a), and in the Homs hinterland (Newson et al. 2012). There is  very little in the way of writing or stamping on material from this source in the Hellenistic or Early Roman period. There are a few stamps on 5th century items, usually with Christian icono­ graphy (Mills 2013a), paralleling the importation of stamped material from Cyprus (Mills 2013a, Manning et al. 2002). It seems reasonable to ­suggest that specialised production was mainly 8  Report Submitted N. Baudray Rimourski, University Quebec 2008.

­ ndertaken by relatively low status or servile lau bourers, with not much evidence for the epigraphic habit. Hellenistic roof tile was of the Corinthian form (Wikander 1988) with faceted imbrices and tegulae which hooked over each other, rather than with the cutaways common in the west (c.f. Warry 2006). Interestingly in this period there seems to be a differentiation between yellow roof tile used for public buildings and red roof tile used for private buildings. When Beirut became a Roman colony, there was an increase in the use of the curved imbrex common in the west, and no apparent differentiation by colour. By the fifth century AD the distinctive use of colour for private and public buildings is again observable, and

Anejos de AEspA LXXVII

The Social Life of Tile in the Roman World

there is an increase in the Corinthian style im­ brex  in public architecture (although this latter pattern does not seem so clear cut elsewhere in the Levant). Cilician tiles are also an important component of the rural settlements surveyed in the Homs Hinterland (Newson et al. 2012). The traditional flat roofs of basalt or limestone appear to have been the dominant style in these settlements of the Hellenistic through to Byzantine periods, and fired clay roofed tiles corresponded to at best a single structure per settlement. In the Byzantine period these may have been churches, but it is not clear what they would have been in the Hellenistic period. Of note, further north in the limestone massif central the standing ruins of the olive manufactory at Serjilla suggest that it originally had a pitched roof of tiles, so these tile structures could be related to agricultural structures of this sort. Turkey In Antioch9 and its immediate core hinterland Cilician tiles are still found. However, in the northern hinterlands there are a number of 5th-6th century settlements, the pottery of which suggests were of very low status, probably focusing on grain agriculture; functional analysis (Evans 2001) shows a limited range of utilitarian vessels for cooking and eating and no fine ware supply. There is however evidence of a tiled structure in the settlements. These tiles are not from the Cilician source but from elsewhere, possibly north of the Antioch hinterland. As with the Cilician supply, there is no evidence of stamps or graffiti, although further work is required to determine the nature of the industrial set up in this hinterland region. The presence of tiled roofs here is somewhat at odds with their high status in other contexts around the ­Roman world. A number of explanations need to be tested further. For instance, these structures may be related to grain processing or storage, or they could be accommodation for a steward or land owner, or communal accommodation for the (presumably servile or indentured) work force, similar to the specialist settlements which had private contractors supplying the military along the

  Report Submitted L. Swartz, ULC, Chicago 2012.

9

93

King’s Road in north-west Britain (Swann and Philpott 2000; Evans in press a). In Western Turkey, at Sagalassos,10 there is evidence of a completely different set up for the manufacture of CBM and variations in its use. In the city itself brick is much more commonly used for walls and flooring than at the previously mentioned site. Roof styles are mainly Corinthian tegulae with curved or faceted imbrices, with a few Laconian-style pan tiles noted. There are a  number of slipped tiles which may have been originally used on churches or other high status buildings. In the smaller settlements in the territory of Sagalassos the main roof style appears to be Laconian, which may be indicative of a political or social identity being referenced by the arch­itectural differences between the city and its subordinate settlements. The roof tile and brick were not made within the city itself, but in the surrounding valleys (Degryse and Poblome 2008). There is a large number (over 156) of animal prints, a range of complex signatures, stamps and graffiti. Altogether this evidence suggests that the tiles were produced on working farms, mainly for use within the city itself. The range of signatures suggests that it was important that tallies were kept of how much was produced by a particular tile maker. The presence of stamps would indicate that there was another tier of economic monitoring above that of the individual tile maker for some projects, perhaps marking the ownership of the output from a particular farm, or perhaps monitoring the contributions of ­landowners hol­ding several properties. The graffiti show that the epigraphic habit is more in evidence in the tile ma­kers here than in the northern Levant. This would suggest a somewhat higher status for tile makers who are also farmers (or employed by farmers), and that perhaps roof tiles could be a form of rent for tenant farmers to pay land owners. Bulgaria In Bulgaria, It would appear that CBM was transported along the Danube, with production in the west of the country, but with tile from this region reaching Varna. The majority of roof tile would appear to be in the Laconian style,

10

  Report Submitted J. Poblome, Leuven 2014.

94

Anejos de AEspA LXXVII

Phil Mills

a­ lthough there is at least one structure in the ‘Sicilian’ style in the city of Nicopolis Ad Istrum (Poulter 2007) in the early Roman period. The site at Dichin11 has high levels of animal prints (0.4% of fragments) similar to that seen at Sagalassos (above) which would also suggest that production is diffused amongst working farms in a similar way. Also of note is the large number of bricks used in constructions other than bathhouses in Roman Bulgaria. This may reflect the lack of suitable building stone, but may also be part of a ­regional cultural style. Italy In Italy, much attention, rightly, has been paid to the wealth of data from the stamped bricks of the Tiber valley used in the city of Rome and the economics of the brick industry (Helen 1975; DeLaine 1995; DeLaine 1997; DeLaine 2001; Graham 2006). Analysis shows the development of complex ties between land ownership, slave production and where the bricks were used; land owners building urban structures from brick and tile produced on their own land, and also drew on  their wider networks of patronage. Graham (2006) in particular demonstrates that control of the brick market was an important node in power relationships, and its importance led it to becoming an imperial monopoly by the 3rd century. Outside of the rather exceptional case of Rome, recent analysis of pottery and tile from field survey around Nepi (Mills and Rajala 2011a) gives some evidence of the social status of tile outside of Rome. The fabric analysis of the tile suggest it was from Campania, which is a dominant supplier of specialist ceramics, especially mortaria as well as tile, in the early Roman Period (the link between brick and mortaria manufacture was made in Hartley 1972). Interestingly the only evidence for ceramic roofing was from the sites of the larger villas, and also a very unusual site which was a focus of the extra mural cemetery of the urban core of Nepi. This image of the rural roofscape (Mills 2013a) is at odds with the Renaissance image of a tiled landscape and it would be interesting to see if this pattern is replicated elsewhere in Italy, and how it varies around the Roman world. 11   Report Submitted A. Poulter, University of Nottingham 2004.

North Africa The importation of Campanian roof tile and brick into North Africa has been well established (Tomber 1987; Wilson 2001 Graham 2011). As noted above this gave rise to the ‘ballast cargo’ model. Study of stratified sequences suggest that the earliest use of Campanian roof tile is in the Roman colony at Carthage (Mills 2013a). In Carthage local industries have developed by the Vandal period This includes one poor quality batch with a number of footprints which may have been made during the fifth century Vandal occupation. Apart from this, graffiti and stamps on locally produced material are scarce, and it seems plausible to suggest that a lot of local production was linked with the prodigious amounts of ceramics produced in North Africa. Whilst the form of roof tiles is of the ‘Sicilian’ type – that is flanged tegulae with cutaways and curving imbrices – their method of manufacturing tegulae is unusual, with elaborate keying on the underside of the tiles, as opposed to the usual sand impressions from the drying platform. As noted above, structures roofed with fired clay tile are rare in the urban centre, although common in villas in the countryside. In the city most roofs appear to be flat or domed, using mortar and vaulting tubes (Wilson 1992). The most plausible explanation is the lack of suitable timber for such roofs as well as the remnants of any Punic cultural traditions. The relative scarcity of roof tiles and the early importation of them would suggest that they are a significant status marker. DISCUSSION This is a very preliminary attempt at an outline of the social contexts of the manufacturing and use of ceramic building materials, largely based on the author’s work around the Roman world. It is clear that this context is complex and changes across the empire, and over time. However we can suggest that the earliest use of roof tile is socially constrained: by expense. In addition to its usefulness as a fire break in densely populated urban centres (Wikander 1988) this would also have been a useful tool in the developing and entrenching of the polis (and indeed civitas) model of the city. The social status of tile makers and the structure of the industry varied around the empire. In

Anejos de AEspA LXXVII

The Social Life of Tile in the Roman World

the east there appear to be specialised servile workforces mass producing tiles for wide distribution. In the Balkans there is more farm-produced material with literate tile makers perhaps producing tile as a way of paying rent to landowner elites in the cities. It Italy the industry is so important it is taken over by the state by the third century (Graham 2006), having been an obvious source of  wealth and power for landowners. In Britain Roman architecture is introduced by the military but adopted by the local elites, at least in the more romanised south and east of the province. Tile makers appear to be free and itinerant, working as part of a larger body of Roman building specialists, in the 2nd century. By the later third and fourth century, tile production was more centralised, although specialists would still have to be mobile. There are also a number of other more local small scale tileries working at the same time. In any case tile makers and users cannot live in a vacuum – they need imperial culture making them to want to build houses in such style, as well as the financial resources to do so. The tile maker must work directly or indirectly with other skilled labourers to actually build the desired structure. This discussion has mainly been looking at the civilian aspects of tile production. Military production would appear to be very much part of the internal logistics of a legion, one of whose main impacts on the wider Roman world was to introduce tile into new areas. REFERENCES Betts, I., Black, E. W. and Gower, J. 1994: “A corpus of relief-patterned tiles in Roman Britain”, Journal of Roman Pottery Studies, 7, pp. 3-167. Brodbribb, G. 1989: Roman Brick and Tile. Sutton, Wolfboro. Cheer, P. and Booth, P. 2001: “Roman tile (AES767)”, in Booth, P. and Evans, J. (eds.), Roman Alcest­ er. Northern Extramural Area 1969-1988 Excava­ tions, pp. 62-64, Roman Alcester Series 3. Council for British Archaeology, York. Cooley, A. E. and Salway, B. 2012: “Roman Inscriptions 2006-2010”, Journal of Roman Studies, 102, pp. 172-286. Crawford, M. 1996: Roman Statutes, BICS supplement 64. Institute of classical studies, London. Degryse, P. and Poblome, J. 2008: “Clays for mass production of table and common wares, amphorae  and architectural ceramics at Sagalassos”, in

95

­ egryse, P. and Waelkens, M. (eds.), Sagalassos VI. D Geo and Bio-Archaeology at Sagalassos and its ­Territory, pp. 231-254. Leuven University press, Leuven. DeLaine, J. 2000: “Bricks and mortar. Exploring the economics of building techniques at Rome and ­Ostia”, in Mattingly, D. J. and Salmon, J. (eds.), Economies Beyond Agriculture in the Classical World, pp.  271-296, Leicester-Nottingham Studies in Ancient Society. Routledge, London-New York. DeLaine, J. 1997: The Baths of Caracalla. A Study in the Design, Construction, and Economics of Largescale Building Projects in Imperial Rome, Journal of Roman Archaeology Suppl. 25. Journal of Roman Archaeology, Portsmouth, R. I. Delaine, J. 1995: “The supply of building materials to the city of Rome. Some economic implications”, in Christie, N. (ed.), Settlement and Economy in Italy 1500BC to AD 1500. Papers of the 5th Conference of Italian Archaeology, pp. 555-562. Oxbow Books, Oxford. Erim, K. T. and Reynolds, J. 1973: “The Aphrodisias copy of Diocletian’s Edict on Maximum Prices”, Journal of Roman Studies, 63, pp. 99-110. Evans, J. 1999: “The Salford Priors Roman pottery”, in Palmer, S. (ed.), “Excavations in the Arrow Valley, Warwickshire”, Transactions of the Birmingham and Warwickshire Archaeological Society, 103, pp. 101126. Evans, J. 2001: “Material approaches to the identifi­ cation of different Romano-British site types”, in  James, S. and Millett, M. (eds.), Britons and ­Romans: Advancing an Archaeological Agenda, pp. 26-35, CBA Research report 125. Holywell Press, Oxford. Evans, J. 2013: “Balancing the scales Romano-British pottery in early Late Antiquity”, Late Antique Ar­ chaeology, 10, pp. 425-450. Evans, J. in press a: “King Street, The Roman frontier in the North west and Roman military supply”, Journal of Roman Pottery Studies, 16. Evans, J. in press b: “Forms of knowledge; changing technologies of Roman-British pottery”, in Millett, M., Moore, A. and Revell, L. (eds.), Oxford Univer­ sity Press Handbook of Roman Britain. Oxford University Press, Oxford. Evans, J., McCauley, S. and Mills, P. J. E. in press: The Horningsea Roman Pottery Industry in Context: An Area Study of Ceramic Supply in the Cambridgeshire Region. East Anglian Archaeology, Norfolk. Evans J. and Mills, P. J. E. 2013: “Excavation, processing, and studying the pottery from Ras el Bassit, Syria”, Late Antique Archaeology, 9, pp. 553-572.

96

Phil Mills

Fulford, M. G. and Peacock, D. P. S. 1984: Excava­ tions at Carthage, The British Mission, Volume 1, 2, The Avenue du Président Habib Bourguiba, Sal­ ammbo. The Pottery and Other Ceramic Artefacts from the Site. The British Academy, University of Sheffield, Sheffield. Giacchero, M. 1974: Edictum Diocletiani de pretiis: Edictum Diocletiani et collegarum de pretiis rerum venalium. In integrum fere restitutum e Latinis Grae­ cisque fragmentis editit Marta Giacchero. Istituto di storia antica e scienze ausiliarie, Genova. Graham, S. 2006: Ex Figlinis: the Network Dynamics of the Tiber Valley Brick Industry in the Hinterland of Rome, British Archaeological Reports International Series 1486. Archaeopress, Oxford. Graham, S. 2011: “A brick stamp of Agathobulus”, in Stone, D. L., Mattingly, D. J. and Ben Lazreg, N. (eds.), Leptimus (Lamta), Report no 3, The Field survey, pp. 481-485, Journal of Roman Archaeology Suppl. 87. Journal of Roman Archaeology, Portsmouth, R. I. Hartley. K. F. 1972: “The marketing and distribution of mortaria”, in Detsicas, A. (ed.), Current Research in Romano-British Coarse Pottery, pp. 39-51, Council for British Archaeology Research Paper 10. Council for British Archaeology, York. Helen, T. 1975: Organization of Roman Brick Produc­ tion in the First and Second Centuries AD: an Inter­ pretation of Roman Brick Stamps, Acta Instituti Romani Finlandiae 9.1. Academia Scientiarum Fennica, Helsinki. Hodder, I. 1974: “Some Marketing Models for Romano-British Coarse Pottery”, Britannia, 5, pp. 340359. Hurst, H. R. and Gibson, S. 1994: “Roman and later building materials and reconstructions”, in Hurst, H. R. (ed.), Excavations at Carthage. The British Mission, Vol. II, I. The Circular harbour, North side. The site and Finds other than Pottery, pp. 53-63, British Academy Monographs in Archaeology 4. Oxford University Press, Oxford. Johnson, A. C., Coleman-Norton, P. R. and Bourne, F. C. 1961: Ancient Roman Statutes. University of Texas Press, Austin. Jones, B. and Mattingly, D. 1990: An Atlas of Roman Britain. Guild Publishing, London. McCormish, J. M. 2012: An Analysis of Roman Ceram­ ic Building Material from York and its Immediate Environs, M.A. Dissertation, University of York. McWhirr, A. and Viner, D. 1978: “The production and distribution of tiles in Roman Britain with particular reference to the Cirencester region”, Britan­ nia, 9, pp. 355-377.

Anejos de AEspA LXXVII

Manning, S., Tomber, R., Sewell, D. A., Monks, S. J., Ponting, M. J. and Ribeiro, E. C. 2002: The Late Roman Church at Maroni Petrera, Cyprus. The A. G. Leventis Foundation, Cyprus. Martin, R. 1965: Manuel d’architecture grecque V. 1, Matériaux et techniques, Paris. Mills, P. J. E. 2008: “The ceramic building materials”, in Dawson, M. (ed.), “Excavation of the Roman Villa and Mosaic at Rowler Manor, Croughton, Northamptonshire”, Northamptonshire Archaeolo­ gy, 35, pp. 45-93. Mills, P. J. E. 2010: “Ceramic building materials”, in Palmer, S. C. (ed.), “Neolithic, Bronze Age, Iron Age, Romano-British and Anglo Saxon sites excavated on the Transco Churchover to Newbold Pacey gas pipeline 1999”, Birmingham and Warwickshire Archaeological Society Transactions for 2009, 113, pp. 136-138. Mills, P. J. E. 2013a: The Ancient Mediterranean Trade in Ceramic Building Material: A Case Study in Car­ thage and Beirut, Roman and Late Antiquity pottery 2. Archaeopress, Oxford. Mills, P. J. E. 2013b: “The supply and distribution of ceramic building material in Roman Britain”, Field Methods and Techniques in Late Antique Archaeolo­ gy, 10, pp. 451-470. Mills, P. J. E. 2013c: “The potential for ceramic building material”, Late Antique Archaeology, 9, pp. 573594. Mills, P. J. E. 2015: “The Iron Age and Roman pottery from Hayton, Yorkshire”, in Millett M. (ed.), Exca­ vations at Hayton. York Archaeological Monographs, York. Mills, P. J. E. in press a: “The ceramic building material”, in Millett, M. (ed.), Excavations at Thwing, East Yorkshire. York Archaeological Monographs, York. Mills, P. J. E. and Rajala, U. 2011a: “The Roman ceramic material from the field walking in the environs of Nepi”, Papers of the British School at Rome, 79, pp. 147-240. Mills, P. J. E. and Rajala, U. 2011b: “Interpreting a ceramiscene landscape. The Roman pottery from the Nepi Survey Project”, in Mladenovic´, D. and Russell, B. (eds.), Proceedings of the Twentieth An­ nual Theoretical Roman Archaeological Conference, 2010, Oxford, pp. 1-17. Oxbow Books, Oxford. Mills P. J. E. and Rajala, U. 2014: “Supply and distribution of Late Roman coarseware from the Nepi Survey Project”, in Poulou-Papadimitriou, N., Eleni Nodarou, E. and Kilikoglou, V. (eds.), LRCW 4. Late Roman Coarse Wares, Cooking Wares and Am­ phorae in the Mediterranean. Comparison Between Western and Eastern Mediterranean, pp. 29-37,

Anejos de AEspA LXXVII

The Social Life of Tile in the Roman World

­ ritish Archaeological Reports International Series B 2616. Archaeopress, Oxford. Newson, P., Abdulkarim, M., McPhillips, S., Mills, P., Reynolds, P. and Philip, G. 2012: “Landscape study of Dar es-Salaam and the basalt region north west of Homs, Syria”, Berytus, 51-52, pp. 1-35. Orton, C., Tyers, P. and Vince, A. 1993: Pottery in Archaeology, Cambridge Manuals in Archaeology. Cambridge University Press, Cambridge. Peacock, D. P. S. 1984: “The ceramic building materials”, in Fulford, M. G. and Peacock, D. P. S. (eds.), Excavations at Carthage, The British Mission, Vol­ ume 1, 2, The Avenue du Président Habib Bourguiba, Salammbo. The Pottery and Other Ceramic Artefacts from the Site, pp. 242-246. The British Academy, University of Sheffield, Sheffield Poulter, A. G. 2007: Nicopolis Ad Istrum. A Late Ro­ man and Early Byzantine City. The Finds and Biolog­ ical remains. Oxbow Books, Oxford. Swan, V. G. and Philpott, R. A. 2000: “Legio XX VV and tile production at Tarbock, Merseyside”, Bri­ tannia, 31, pp. 55-68. Taylor, J. 2004: “The distribution and exchange of pink, grog tempered pottery in the East Midlands: an update”, Journal of Roman Pottery Studies, 11, pp. 60-66.

97

Tomber, R. S. 1987: “Evidence for long-distance commerce: imported bricks and tiles at Carthage”, Rei Cretariae Romanae Favtorum Acta, 25-26, pp. 161174. Ward, C. 1999: Iron Age and Roman Piddington: The Roman Ceramic and Stone Building Materials 19791998, Fascicule 4. The Upper Nene Archaeological Society, Northamptonshire. Warry, P. 2006: Tegulae. Manufacture, Typology and use in Roman Britain. British archaeological Reports British Series 417. Archaeopress, Oxford. Warry, P. 2010: “Legionary tile production in Britain”, Britannia, 41, pp. 127-147. Wikander, O. 1988: “Ancient roof tile – use and function”, Opuscula Atheniensia, 17 (15), pp. 203-216. Wilson, A. I. 2001: “Ti. Cl. Felix and the date of the second phase of the East baths”, in Stirling, L. M., Mattingly, D. J. and Ben Lazreg, N. (eds.), Leptimi­ nus (Lamta) Report No 2. The East Baths, Cemeter­ ies, Kilns, Venus Mosaic, Site Museum and Other Studies, pp. 25-28, Journal of Roman Archaeology Suppl. 41. Journal of Roman Archaeology, Portsmouth, R. I. Wilson, J. R. A. 1992: “Terracotta vaulting tubes (tubi fittili): on their origin and distribution”, Journal of Roman Archaeology, 5, pp. 97-129.

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

CERAMIC BUILDING MATERIAL: PRODUCTION, SUPPLY AND USE IN ROMAN LONDON IAN M. BETTS Museum of London Archaeology, London

ABSTRACT: During the 1st-early 2nd century AD a wide variety of different ceramic building material types were used in London. Most were made at tileries known to have been situated in or near London, but additional supplies were also brought in from other areas, such as north Kent and Sussex. The Sussex tilemakers appear to have been the first in Roman Britain to have produced hollow voussoir tiles, and were among the first to use wooden rollers to key tiles. The later tileries supplying London and other parts of south-east England produced a more limited range of tile types. Some of these tileries transported their products over considerable distances, sometimes by road, other times by utilising water transport. KEYWORDS: Ashtead, Bath buildings, Box-flue, Brick, Calcareous tile, Ceramic ‘spacer’, Cohort of Gauls, Columns, Forum-basilica, Harrold, Hypocaust heating, North Kent, Paving brick, Pipes, Procuratorial tile, Reigate, Roof tile, Thames, Tilemakers, Voussoir, Wall tile. RESUMEN: Durante el siglo i y comienzos del ii, en Londres, se empleó una amplia variedad de tipos de materiales de construcción cerámicos. La mayor parte se fabricaron, posiblemente, en los talleres situados en la misma Londres o cerca, aunque suministros adicionales se importaron de otras áreas, como Kent y Sussex. Los talleres de Sussex parecen haber sido los primeros en Britannia romana en la producción de latericios adovelados huecos y fueron entre los primeros en utilizar rodillos de madera para marcar los latericios. Los talleres posteriores que suministraban Londres y otras zonas del sureste de Inglaterra producían una variedad más limitada de tipos de ladrillo. Algunos de estos talleres transportaban sus productos a distancias considerables, a veces por carretera, otras veces utilizando el transporte en agua. PALABRAS CLAVE: Ashtead, termas, Tubulus, ladrillo, latericio calcáreo, distanciadores cerámicos, cohorte de galos, columnas, foro-basílica, Harrold, hipocausto, norte de Kent, latericio para pavimentación, tuberías, latericios con sellos del procurador provincial, Reigate, teja, Thames, talleres, dovela, ladrillo.

EARLY TILE PRODUCTION Large quantities of Roman tile (‘tile’ in this context meaning both brick and tile) have been ­recovered from excavations in London over the last 40 years. Tile production would appear to have commenced very soon after the establishment of Roman London in c AD 48. Tile was certainly being used for building construction before the Boudican revolt. The revolt against Roman rule, led by

Queen Boudica widow of King Prasutagus of the Iceni of East Anglia, destroyed much of the city in AD 60/61 (Marsden 1980: 31). Roof tile (tegula and imbrex), brick, box-flue (tubulus), half box-flue and opus spicatum paving bricks are present in preBoudican contexts on London sites (Dunwoodie 2004: 22; Pringle 2007: 205; Wroe-Brown 2014: 11). Much of the hypocaust material may derive from an early bath house situated somewhere in the Fenchurch Street area east of the forum/basilica.

100

Anejos de AEspA LXXVII

Ian M. Betts

Londinium

fort

city wall amphitheatre

Cheapside forum/ basilica

Bread Street

Paternoster Square

Huggin Hill public baths Cannon Street

River 0

New Fresh Wharf

Tham

es

500 m Fig. 1. Roman London.

The use of ceramic building material expanded considerably during the late 1st-early 2nd century. This was a period of rapid urban expansion, with the construction of numerous private dwellings alongside an ambitious public building programme (Rowsome 2008: 27). Amongst the ­public buildings dating to this period which incorporated ceramic building material are Huggin Hill baths, the forum-basilica and the early 2nd century rebuild of the amphitheatre (fig. 1). The majority of ceramic building material used in London in the 1st-early 2nd century was roofing tile (tegulae and imbrices) and various sizes of brick. Other tile types were generally used in much smaller amounts. In the latter category are half box-flue tiles (fig. 2), normal full box-flue tiles (fig. 2), solid voussoir tiles, hollow voussoir (tubulus cuneatus) tiles, notched wall tiles (parieta­ lis) (fig. 2), opus spicatum paving bricks, round pila bricks, semi-circular bricks (fig. 3) and socketed water pipes (tubuli). Very few ridge tiles are known from London, or elsewhere in south-east England, suggesting imbri­ces were utilised to cover the crest of the roof. Many tile types were required in relatively small amounts because the majority of London domestic buildings were of clay and timber construction, with the use of tiles restricted to roofs and tiled hearths.

Early Roman tile production in south-east England was marked by a period of innovation and experimentation. One innovative form was the socketed box-flue, where the top and base of each tile was modified to allow the tiles to slot into one another (fig. 4). This would have allowed the tiles to lock more securely into position but they probably still had to be secured to the wall by the use of iron clamps. The disadvantage of an interlocking system is that the tiles would have had to have been carefully made and fired to ensure they interlocked, which may explain why relatively few tiles of this type are known from Roman Britain. The example illustrated is  from Huggin Hill baths which was constructed during the Flavian period (AD 70-100). The scored keying is typical of box-flues made during the 1stearly 2nd century. Brodribb (1987: 74) illustrates a socketed box-flue from Gelligaer in South Wales, again with scored keying. Box-flue tiles with a far more unusual method of securing them to the wall were made at a tilery adjacent to Ashtead villa in Surrey. These have a protruding ‘fish-tail’ like extension to the top of the tile which could be set into the wall behind, so removing the need for iron clamps (fig. 5). Their big disadvantage was that these ‘fish-tails’ were very prone to breakage and it may have been

Anejos de AEspA LXXVII

Ceramic building material: production, supply...

101

b a

c

e

Fig. 3. Nibbed segmented quadrant and semi-circular bricks from Ashtead villa, Surrey (David Bird, Surrey Archaeological Society).

­ ifficult to align them to the masonry wall behind. d It was certainly not an idea which appears to have been repeated elsewhere. The box flue tiles with fish-tail extensions were not the only innovative tile produced by the Ashtead tilemakers. They also made a series of unusual segmented quadrant bricks with a nib extending from one corner edge (fig. 3).1 These were used with semi-circular bricks to form c. 336 mm wide projecting semi-circular columns. The projecting nibs were used to anchor the curved bricks

to the wall behind. Semi-circular bricks could also be used back to back to form circular columns. Semi-circular bricks were used in London, and there are also a number of fragmentary bricks with projecting nibs but these all seem to be a type of solid vaulting brick known as the ‘armchair’ voussoir (Brodribb 1987: 46) (fig. 2). Semi-­circular bricks were one of the products of a procuratorial tileworks operating somewhere in London, as some are marked with procuratorial tile stamps (Betts 1995: 213) (fig. 13). The same tilery was responsible for the manufacture of roofing tile, brick and notched wall tile. It may also have made box-flues, such as the socketed flue-tiles installed in the public baths at Huggin Hill. The origin of box-flues is often more difficult to identify as a much smaller proportion of such tiles were stamped (Brodribb 1987: 124). The notched wall tiles and semi-circular bricks used in London may have been made exclus­ ively at the procuratorial tileworks. Notched wall tiles, used with cylindrical bobbin-like ‘spacers’, formed a continuous jacket around the walls of bath houses, or similar heated structures. Wall tiles are known from a number of sites in London so it is surprising that only one definite ceramic ‘spacer’ has been found (fig. 2).2 It would be expected that spacer bobbins would be present on the majority of sites with wall tiles, so their ab-

1   The illustration of two nibbed brick in Brodribb (1987: 22) is incorrect, they should be quadrant shaped.

2   Bloomburg, London (Museum of London Archaeology, site code BZY10, context [3166]).

0

250 mm

d

0

100 mm

Fig. 2. Half-box flue (a), box-flue (b), notched wall tile (c), spacer bobbin (d) ‘armchair’ voussoir (e).

102

Ian M. Betts

Anejos de AEspA LXXVII

Fig. 4. End of a socketed box-flue tile with scored keying from the Huggin Hill Roman Baths, London (author).

Fig. 5. Box-flue tile with unusual ‘fish-tail’ protrusion from Ashtead villa tileworks, Surrey (David Bird, Surrey Archaeological Society).

sence is puzzling. It is possible that the majority of bobbins were made of wood rather than cer­ amic and have rotted away leaving no trace in the archaeological record. A number of more unusual tile types have been found on single London sites, suggesting they were supplied for a specific building project. Among these oddities is a single hexagonal shaped brick from Bread Street, London (Betts 2013: 34) (fig. 6). Similar shaped bricks found at Caersws fort in Powys, Wales, and Fingringhoe, Essex were used as flooring (Brodribb 1987: 54). The Bread Street example would almost certainly have been used, or intended to have been used, in a similar decorative brick floor. From the initial construction phase of the 2nd-century forum/basilica in the Leadenhall Court/Gracechurch Street area were a  number of trapezoidal bricks (Brigham and Crowley 1992: 100) (fig. 7). None have been found elsewhere suggesting they may have been specially made for the building. One tilery, believed to be have been situated somewhere along the south coast of Sussex, supplied London with a variety of innovative tile types. Recent work by Lynne Lancaster (2012: 421) suggests the Sussex tilemakers were the first in Ro-

Anejos de AEspA LXXVII

Ceramic building material: production, supply...

103

Fig. 6. Hexagonal brick from Bread Street, London (Andy Chopping, MOLA).

Fig. 8. Hollow voussoir tile with relief-patterned keying made by the Sussex tilemakers (Lynne Lancaster, Ohio University).

Fig. 7. Trapezoidal brick from the Roman basilica/forum, London (author).

Fig. 9. Large curved brick relief-patterned keying from Bloomberg, London (Andy Chopping, MOLA).

man Britain to produce hollow voussoir tiles (fig. 8). These would have been located at the top of vertical rows of box-flue tiles where they would have formed the vaulted roofs of bath buildings (Lancaster 2012: 428). The Sussex tilemakers also produced other tiles needed in the construction of bath buildings including single and double box-flue tiles and some remarkable curved bricks. These bricks may have been set horizontally on the top of walls above vertical lines of box-flues. One interesting feature of the box-flue and voussoir tiles is that many are twice as thick as later examples produced at other tileries. It is possible that the Sussex tiles were ‘over-engineered’ as the tilemakers were still unsure of their structural stretch when installed in bath buildings.

The large curved bricks are a type unique to the Sussex tilemakers. An almost compete example from London measures 571mm in length (breadth 213-c 250 mm, thickness 31-50 mm) (fig. 9). Why none were made elsewhere is uncertain, perhaps the much more readily available curved imbrices were employed in a similar position instead. All the curved bricks are keyed with an exceptionally long wooden roller carved with a diamond and lattice pattern (relief-patterned die type 37) (Betts et al. 1994: 110). Most wooden rollers used to impress keying patterns into the clay prior to firing were much smaller, typically 50-100 mm in breadth (fig. 10). The length of these keying patterns, adjusted for slight shrinkage during firing, suggests the use of rollers c 40 and c 85 mm in diameter.

104

Ian M. Betts

Fig. 10. Tiler using a wooden roller to key a box-flue tile.

The overwhelming majority of tileries operating in Roman Britain produced roofing tile (tegu­ la and imbrex) and brick, with generally only a small proportion of other types. The proportion of different tile types produced by Reigate and a north Kent tilery listed in fig. 11 is typical of most British tileries. Roofing tile and brick were the types of tile most needed for building construction. By the late 1st century it is probable that cer­ amic tiled roofs were ubiquitous in urban areas such as London, not least because they represented less of a fire risk than inflammable roof coverings such as thatch and wooden shingles. There is also evidence for the use of tiled roofs, along with brick, in smaller roadside settlements such as Brentford to the west of London (Betts 2013: 72), as well as more rural locations, such the Harmondsworth and Harlington area further west.3 Roofing tiles have also been found associated with various rural farmsteads between Rainham and Upminster to the east of London (Howell et al. 2011: 85).4 In these areas roofing tile was not only used on domestic dwellings but covered what appear to be agricultural buildings. Bricks were a major component of most ­tilemakers output. Not only were they used to 3   Betts, I. M. 2015: West London landscapes (sites CFL94/ CLH89, HL80, MFH87, NHS97, SPD85, WGF79): Building materials, unpublished archive report, Museum of London Archaeology. 4  Betts, I. M. 2006: East London gravel sites: building material, unpublished archive report, Museum of London Archaeology.

Anejos de AEspA LXXVII

c­ onstruct walls or as levelling courses in walls of masonry construction, they could be employed as paving or used in the base of hearths, ovens or similar structures. A remarkable feature of the output of the Sussex tilemakers is the unusual proportion of tile types manufactured. They do not seem to have made roofing tile, instead over half their output comprised box-flue and voussoir tiles and curved bricks for installation in hypocaust heating systems (figs. 8, 9, 11). They also produced a smaller proportion of flat brick and tegula mammata tiles. The bricks generally measure between 59mm and 100mm in thickness suggesting they are probably of larger sesquipedalis or bipedalis type (Brodribb 1987: 3). The principal use of these large square bricks was to cap stacks of pilae in the floor of hypocaust heating systems. The tegula mammata tiles are different from those supplied to London from other tileries. The clay nibs are attached to a surface completely covered by combed keying. Each tile probably had two nibs set in diagonally opposite corners (Pringle 2009: 193, fig. 147). How these tiles were used is still uncertain. It is clear that the tiles made by the Sussex tilemakers were primarily intended for use in the construction of bath buildings. The products of the Sussex tilemakers shown in fig. 12 may not just show the location of their tiles but the work of a group of tilemakers/builders who specialised in the construction of bath buildings and hypocaust heating systems. The roofing tile and other building material derive from separate tileries so it seems reasonable to assume that the other buildings, which required more simple construction skills, are the work of separate building contractors, the more skilled work of constructing bath buildings being entrusted to the Sussex tilemakers. Castleford Roman fort in West Yorkshire provides an interesting parallel. The more simple fort buildings were probably constructed by the Fourth Cohort of Gauls, the auxiliary unit garris­ oned at Castleford, but the bath building would appear to have been erected by specialist military engineers brought in from the legionary headquarters at York in North Yorkshire (Betts 1998: 232). As with the Sussex tilemakers, not only did the legionary headquarters supply the necessary skilled builders, they also supplied the building material needed for construction, some of which was marked with their legionary tile stamp. The Sussex tilemakers may have been responsible for another technological innovation, the use

Anejos de AEspA LXXVII

Tile type

Ceramic building material: production, supply...

Brick

Tegula mammata

Opus spicatum paving brick

Half-box flue

Box flue/ hollow voussoir1

Solid voussoir

74.4

22.02

0.5

1.0

0.9

 1.1

–

29.5

5.4

–

–

60.3

Roofing (tegula/ imbrex)

105

Curved brick

Finial/ chimney

0.09

–

0.01

–

4.8

–

Early Roman North Kent group Sussex group Later Roman Calcareous group

97.4

2.5

–

–

–

 0.1

–

–

–

Reigate group

54.5

40.4

–

–

–

 5.1

–

–

–

Harrold group

75.5

5.0

–

–

–

19.5

–

–

–

  It is often not possible to differentiate small fragments of flue tile from voussoir tile. 2   Some smaller incomplete fragments may be tegula mammata. 1

Fig. 11. Percentage of types of tile supplied to Roman London from five production sources (from MOLA Oracle database).

of wooden rollers with a carved pattern to key box-flue and voussoir tiles (fig. 10). Wooden rollers were first used on the walls of wattle and daub buildings in order to provide a keyed surface for the attachment of wall plaster. The keying of daub walls was an early Roman technique, appearing in London, Colchester and St Albans before the Boudican revolt of AD 60/61 (Russell 1994: 49). The Sussex tilemakers may have been the first, or at least among the first to apply the same technique to tile. Certainly the wooden rollers they used closely resemble those employed to key daub walls. Both are characterised by deeply impressed patterns. The wooden rollers used by later tile­ makers have much shallower keying patterns. The innovative products of the Sussex tilemakers have been found in London in relatively small quantities. The majority of tiles seem to have been made using local brickearth deposits. No tile kilns have been positively identified from Roman London, but tile waster dumps have been located in the Paternoster Square area north of St Paul’s Cathedral and the Cheapside area to the east. The latter is of particular importance as some of the waster tiles have procuratorial stamps confirming the presence of a procuratorial tilework (Betts 2014: 69-70). This tilery was established to supply building material used in the construction of London’s major public buildings notably the large public baths at Huggin Hill and a major building complex in the Cannon Street area, the so-called ‘Governor’s Palace’ (Betts 1995: 223), part of which could be another baths.5

A small number of civilian tile stamps have also been recovered (Collingwood et al. 1993: 6063, 69-70) (fig. 13) indicating the activities of various private tile making concerns in the London area.6 Presumably these were established to take advantage of the increasing demand for roofing tile and brick, from the late 1st-2nd century. The small number of civilian stamps recovered from London prevents meaningful discussion of the types of tiles produced. All the examples found so far are impressed into tegula roofing tile and brick. Additional sources of tile were brought into early Roman London from tileries situated in north Kent, such as the yellow, white and cream tiles believed to have been made in the Eccles area (fig. 11). There importation was probably linked with the supply of Kentish ragstone (limestone) into London from the Maidstone area. The major­ ity of tiles arriving in London from north Kent were almost certainly brought in by boat, probably via the river Medway then along the Thames Estuary. Transport by water was ideally suited to heavy bulk items such as tile and stone. Movement of tiles by boat also seems to have been employed by the Sussex tilemakers, as many of their tiles are found on various villa sites along the south coast of England (fig. 12). It is also clear that the Sussex tilemakers distributed some of their tile by cart or waggon, principally using the major Roman road between Chi­ chester and London. Movement by road would have been needed to supply the villas at Bignor

5  Betts, I. M. 2014: Cannon Place (LYD88, CCP04, CNV08): building material, unpublished archive report, Museum of London Archaeology.

6   There is also a tile stamped...] N (in reverse) TD, and a tegula stamped...] V.

106

Anejos de AEspA LXXVII

Ian M. Betts

products of the Sussex tilemakers probable products of the Sussex tilemakers

Barking Thames

London

Croydon

Lullingstone

Cobham

1 2 3 4 5 6 7 8

Westhampnett Batten Hanger Boxgrove Storrington Arundel Tortington Newhaven Bullock Down

ay

Medw

Ewhurst

0

Alfoldean

25 km

Winchester

Bignor

2 Fishbourne

1

Wiggonholt 4

3

Chichester

6

Steyning

South Malling

5 Angmering

Southwick

7

Eastbourne 8

Fig. 12. Distribution of the products of the Sussex tilemakers (Judit Pereztegi, MOLA).

Fig. 13. London tiles with part of a procuratorial stamp lettered “p(rocuratories) p(rovinciae) BRI(tanniae) Lon(dini)” and part of a civilian stamp lettered “D(ecimi) M(…) Val(…) / D(ecimi) M(…) P(…) / (figlinae) tegul(arinae)” (Andy Chopping and author, MOLA).

and Wiggonholt, and the roadside settlement at Alfoldean. Transport by cart or waggon need not necessarily have presented a problem. It has been assumed that transport by road would have been difficult and expensive but this need not have been the case. Most tileries were situated in rural locations or at the edge of urban areas. Many farming communities would have had access to draft animals and carts, both of which could have been deployed for the carriage of tiles. Such movement would have been undertaken when they were not needed for agricultural purposes. Tile making would have been an integral part of the rural economy, along with raising livestock and the planting and harvesting of crops. The making of tiles was a seasonal activity which would have fitted in well with the demands of farming. Medieval tile making accounts suggest the clay was dug in autumn and left to weather during the winter months (Celonia and West 1967: 217-218). The evidence from tile graffiti suggests Roman tile manufacture, at least in the northern parts of the Empire, was largely confined to the summer months (Warry 2006: 121). Tile making would have fitted in well between the planting of crops in the spring and harvesting in late summer.

Anejos de AEspA LXXVII

Ceramic building material: production, supply...

LATER TILE PRODUCTION There were few, if any innovations in the production of ceramic building material used in London and most of the rest of south-east England after the mid-2nd century. Instead the later Roman period seems to have been marked by a period of retrenchment and technological stagnation. Many building material types disappear in London, such as notched wall tiles, opus spicatum paving bricks, water pipes, semi-circular and circular bricks and most tegulae mammatae (fig. 11).7 Box-flue tile (and probably hollow voussoir tile) production continued but the practice of keying with a wooden roller seems to have died out by the end of the 2nd century. Box-flue tiles were now all combed, combing having superseded the earlier practice of knife-scored keying sometime in the early 2nd century. There also seems to have been little experimentation in form types which were still in production. The only significant change in London is what appears to be a gradual decrease in roofing tile size. This reduction of the varieties of tiles used in London can be directly linked with a major reorganisation in the supply of brick and tile to the city from the mid-2nd century. During this period many tile kilns which formerly supplied London seem to have ceased production. This included the procuratorial tileworks which ended production after the completion of most of London’s major public buildings by the middle of the 2nd century. It has also been suggested that the population density was falling in the mid to late 2nd century (Swain and Williams 2008: 39), which would have manifested itself as a drop in the demand for new housing or the expansion of existing buildings. It is clear that fresh supplies of building material were still required but that these were increasingly arriving from tileries producing a much narrower range of tile types, generally just roofing tile and brick with smaller quantities of box-flue tile. This is the pattern seen when examining the types of tiles brought into London from the tileries situated at Reigate in Surrey and Harrold in Bedfordshire (fig. 11). The end of tile production in London, and at other urban locations like Canterbury in Kent, 7   The solitary tegula mammata in a later calcareous fabric type, from Harp Lane, London, is illustrated in Betts and Foot (1994: 24).

107

suggests that it became more economical to move tiles from a small number of central kilns covering a larger geographical area rather than from more numerous smaller scale tileries supplying a more limited local market. One of these more central tileries supplied tiles containing small inclusions of limestone and fragments of fossil shell. These so-called ‘calcareous tiles’, dating to around AD 140-300, have an exceptionally wide distribution across southern England (fig. 14). Many have been found on sites located on or near the coast or along navigable rivers implying movement by ship and boat. In London a large assemblage of calcareous tiles were dumped at the Roman quay at New Fresh Wharf, Lower Thames Street on the north bank of the river Thames (Betts and Foot 1994: 28-29). This may mark the location where calcareous tiles were unloaded. The tilery producing calcareous tiles, the location of which is still unknown, is highly unusual in that it concentrated on the manufacture of tegula and imbrex roof tile (fig. 11). Bricks and box-flue tiles were also manufactured but other tile types are extremely rare. Fresh supplies of roofing tile must have been required in the mid 2rd-3rd centuries. This may be because later roofs used smaller lighter tegulae and imbrices which could not easily have been used together with larger, heavier earlier types. Earlier roofing tile, which could not be reused as roofing, could still be useful such as for walling. Sometimes the flanges were deliberately knocked off for this purpose. Brick types on the other hand stayed the same so could be more readily reused, as presumably could box-flue tiles where available. In derelict and decayed buildings brick would often survived intact in the walls, whilst any roofing tile which had not been taken away for reuse would often have fallen to the ground and shattered. The large bath complex at Huggin Hill, demolished shortly  after c AD 140, would have provided a huge amount of brick and box-flue tile for reuse in other building projects, as would the forum/basilica, the largest building complex in Roman Britain (Perring 1991: 58), demolished at the begin­ning of the 4th century (Bateman 1998: 51). The range of products of the Reigate tile ind­ ustry is more typical of that found on most tileries in Roman Britain. Roofing tiles and bricks predominate, with a much smaller proportion of box-flue tile (fig. 11). It has long been suspected

108

Anejos de AEspA LXXVII

Ian M. Betts

Gosbecks near Colchester

s

Thame

London

Ebbsfleet Deerton Street

y

wa

d Me

Canterbury

Each End, Ash

Winchester Romsey

Batten Hanger Elsted

Southampton Exeter

Seaton

Newport Weymouth

Bowcombe

Brading

Worthing

Preston Park Brighton 0

50 km

Fig. 14. Distribution of tiles belonging to the calcareous tile group (Judit Pereztegi, MOLA).

that tiles from the Reigate area were used for building construction in London during the period AD 140-230 (Betts 2003: 256-257). This was recently confirmed by the results of inductivelycoupled plasma spectrometry (ICP) analysis on tiles from Reigate and London and other sites in south-east England.8 The Reigate tilery supplied tiles to nearby villa sites, such as Walton Hearth, Ashtead and posibly Abinger, roadside ­settlements such as Brentford, and possibly the temple complex at Farley Heath (fig. 15). Reigate tiles are also present at Fishbourne Roman palace and the urban settlements of Chichester and St Albans. The sites in Kent (Ebbsfleet, Rochester, Kemsley, Canterbury, Each End and possibly Snodland) may represent stockpiles of tiles held by builders merchants based in London which were carried down the river Thames by boat, rather than di­ rectly-supplied from Reigate itself. The last major movement of tile into London occurred around AD 270-350 when fresh supplies were obtained from a tilery located at Harrold in Bedfordshire, some 87 km north-west of London. The products of this kiln are widely distributed in the south Midlands and Cambridgeshire (Ungar 2009: 109). Harrold mainly supplied London with roofing tile and combed box-flue (fig. 11). There are only small amounts of brick, again probably due to the recycling of brick for the demolition of earlier structures. When production at Harrold

ceased in the mid-4th century London ceramic tile was replaced by stone roofing cut from fine grained sandstone. The same stone was used as paving. Unexpectedly in 2001 evidence for the re-­ establishment of tile making west of London was found beneath St Martins in the Field church in Trafalgar Square. The last firing of the kiln has an archaeomagnetic date of AD 400-450 (Telfer 2010: 53). This suggests the kiln was in operation either at the very end of the 4th century or in the early years of the 5th century. It is extremely difficult to determine how long Roman London continued to  be occupied into the 5th century, but there is little archaeological evidence to suggest it survived much beyond AD 409, the date when the officials of Constantine III were expelled from Roman Britain and the native population set up their own administration (Salway 1991: 434-437). What appears to be waster material from the last firing of the St Martins in the Field kiln indicates production of roofing tile and combed boxflue was still being undertaken at the very end of the occupation of Roman Britain. The kiln may have been set up to provide tile to build a structure found nearby by William Stukeley in 1722. The interpretation of this structure is still unclear. It has been suggested that it was another tile kiln (M. Biddle, pers. comm.), but it is possible the structure served some other purpose.9

8   Hughes, M. J. 2015: Fourth report on the chemical analysis of Roman box-flue tiles from the production site and villa at Ashtead and other sites in London and SE England by inductively coupled plasma spectrometry (ICP): 2014 project, unpublished archive report, Surrey Archaeological Society.

9   The arched structure drawn by Stukeley was 6 feet (1.8 m) wide, which is almost double the width of the flues used in tile kilns in Roman Britain.

Anejos de AEspA LXXVII

Ceramic building material: production, supply...

109

St Albans

s

Thame

London Brentford Ebbsfleet Ashtead

Buckland Abinger Farley Heath

Ewel Walton Heath Reigate

Rochester Kemsley

Snodland

Canterbury

Each End, Ash

ay Medw

Reigate kiln Reigate tile possible Reigate tile Fishbourne Chichester

0

25 km

Fig. 15. Distribution of tiles made at the Reigate tilery, Surrey (Judit Pereztegi and Hannah Faux MOLA).

DISCUSSION AND CONCLUSIONS From the evidence collected over the last 40 years it is clear that a wide variety of different types of ceramic building material were used in London and other areas of south-east England during the 1st-mid 2nd century. There were a number of more experimental types whose manufacture seems to have been confined to specific tileries. Falling into this category are the large thick curved bricks made by the Sussex tilemakers and the box-flue tiles with the ‘fish-tail’ projections made at the villa estate tilery at Ashtead, Surrey. The majority of early Roman ceramic building material used in London probably came from tile kilns situated in or close to London. This included the procuratorial tiles and the products of private tilemakers. Other tileries, such as those loc­ated in north Kent and Sussex, only provided in total around 10%-15% of London’s tile needs. There is a marked decrease in the variety of tiles supplied to London and elsewhere in southeast England from the mid-2nd century. This reduction coincides with a major reorganisation in the distribution and supply of tile to Roman London. Generally these later tileries, such as those at

Reigate and Harrold, produced just roofing tile, brick and combed box-flue (and possibly voussoir) tile. Despite its bulk and weight it is now clear that ceramic building material was moved consider­ able distances. Water transport utilising boats along the south coast and the river Thames was the preferred method, but other tile was moved overland by cart or waggon. Such movements increased markedly when many of London’s ­ more local kilns ceased production in the mid-2nd century. ACKNOWLEDGMENTS The author wishes to thank Andy Chopping (MOLA) for figs. 6, 9 and 13 and Judit Peresztegi and Hannah Faux (MOLA) for the maps. Lynne Lancaster (Ohio University) kindly supplied the photograph of the Sussex voussoir tile, whilst ­David Bird (Surrey Archaeological Society) ­ kindly supplied the image of the box-flue tile from Ashtead villa. Other photographs are by the author. Comments of draft copies of the text were proved by Louise Fowler (MOLA).

110

Ian M. Betts

REFERENCES Bateman, N. C. W. 1998: “Public buildings: some contrasts”, in Watson, B. (ed.) Roman London: recent archaeological work, pp. 47-57, Journal Roman Arch­ aeology Suppl. 24. Journal of Roman Archaeology, Portsmouth, R. I. Betts, I. M. 1995: “Procuratorial tile stamps from London”, Britannia, 26, pp. 207-229. Betts, I. M. 1998: “The brick and tile”, in Cool, H. E. M. and Philo, C. (eds.), Roman Castleford, Volume 1, The Small Finds, pp. 225-232, Yorkshire Archaeology 4. West Yorkshire Archaeology Service, Exeter. Betts, I. M. 2003: “Fabric analysis of the tiles”, in Masefield, R. and Williams, D. (eds.), “A Roman tilery at Doods Farm, Reigate”, Surrey Archaeolog­ ical Collections, 90, pp. 256-257. Betts, I. M. 2013: “The Roman building material”, in Cowie, R., Thorp, A. and Wardle, A. (eds.), Roman Roadside Settlement and Rural Landscape at Brent­ ford, p. 72, Archaeology studies series 29. Museum of London Archaeology, London Betts, I. M. 2013: “The Roman building material”, in Howell, I., Blackmore, L., Phillpotts, C. and Thorp, A. (eds.), Roman and Medieval Development South of Cheapside, pp. 32-35, Archaeology studies series 26. Museum of London Archaeology, London. Betts, I. M. and Smith, T. P. 2014: “Building mate­ rials”, in Watson, S. (ed.), Urban Development in the North-West of Londinium, pp. 67-70, Archaeology studies series 32. Museum of London Archaeology, London. Betts, I. M. and Foot, R. 1994: “A newly identified late Roman tile group from southern England”, Bri­ tannia, 25, pp. 21-34. Betts, I. M., Black, E. W. and Gower, J. 1994: “A corpus of Roman relief-patterned tiles in Roman Britain”, Journal of Roman Pottery Studies, 7, pp. 1-167. Brigham, T. and Crowley, N. 1992: “Reconstructing the basilica”, in Milne, G. (ed.), From Roman Basil­ ica to Medieval Market. HMSO, London. Brodribb, G. 1987: Roman Brick and Tile. Alan Sutton, Gloucester. Celoria, F. and West, H. W. H. 1967: “A standard specification for tiles in 1477”, Journal of the British Ceramic Society, 4.2, pp. 217-220. Collingwood, R. G., Frere, S. S., and Wright, R. P. 1993: The Roman Inscriptions of Britain, Volume II, Instrumentum Domesticum (Personal Belongings and the Like), Fascicule 5, Tile-stamps of the Classis Britannica; Imperial, Procuratorial and Civic TileStamps; Stamps of Private Tilers; Inscriptions on

Anejos de AEspA LXXVII

Relief-Patterned Tiles and Graffiti on Tiles. Alan Sutton, Gloucester. Dunwoodie, L. 2004: Pre-Boudican and Later Activity on the Site of the Forum, Archaeology studies series 13. Museum of London Archaeology, London. Lancaster, L. C. 2012: “A new vaulting technique for early baths in Sussex: the anatomy of a Romano-­ British invention”, Journal of Roman Archaeology, 25, pp. 419-440. Marsden, P. 1980: Roman London. Thames and Hudson, London. Perring, D. 1991: Roman London. Seaby, London. Pringle, S. 2007: “London’s earliest Roman bath-­ houses?”, London Archaeologist, 11.8, pp. 205-209. Pringle, S. 2009: “Building materials”, in Cowan, C., Seeley, F., Wardle, A., Westman, A. and Wheeler, L. (eds.), Roman Southwark Settlement and Economy: Excavations in Southwark 1973-91, pp. 187-206, Museum of London Archaeology Monograph 42. Museum of London Archaeology, London. Rowsome, P. 2008: “Mapping Roman London: identifying its urban patterns and interpreting their meaning”, in Clark, J., Cotton, J., Hall, J., Sherris, R. and Swain, H. (eds.), Londinium and Beyond, pp. 25-32, Research report 156. Council British Archaeology, York. Russell, M. 1994: “Relief patterned daub”, in Betts, I. M., Black, E. W. and Gower, J. (eds.), “A corpus of Roman relief-patterned tiles in Roman Britain”, Journal of Roman Pottery Studies, 7, pp. 47-50. Salway, P. 1991: Roman Britain. Clarendon Press, Oxford. Swain, H. and Williams, T. 2008: “The population of Roman London”, in Clark, J., Cotton, J., Hall, J., Sherris, R. and Swain, H. (eds.), Londinium and Be­ yond, pp. 33-40, Research report 156. Council British Archaeology, York. Telfer, A. 2010: “New evidence for the transition from the Late Roman to the Saxon period at St Martin-in-the-Fields, London”, in Henig, M. and Ramsay, N. (eds.), Intersections. The Archaeology and History of Christianity in England 400-1200. ­Papers in Honour of Martin Biddle and Birthe Kjlbye-­ Biddle, pp. 49-58. Archaeopress, Oxford. Unger, S. 2009: “Red or yellow? The changing colour of Roman London’s roof-line”, London Archaeolo­ gist, 12.4, pp. 107-113. Warry, P. 2006: Tegula: Manufacture, Typology And Use In Roman Britain, British Archaeological Reports British Series 417. Archaeopress, Oxford. Wroe-Brown, R. 2014: Roman Occupation South-East of the Forum, Archaeology studies series 13. Museum of London Archaeology, London.

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

TEGULAE MAMMATAE OR LATERES? ON THE PRESENCE AND USE OF TEGULAE MAMMATAE IN THE DELTA OF THE RHINE A. (GUUS) GAZENBEEK Independent researcher at StudiCo/Archaeologist at Grontmij Nederland BV, JJ Roermond (the Netherlands)

ABSTRACT: Tegulae mammatae, a term already used in antiquity, form a somewhat enigmatic group of ceramic building material used in heating systems and as insulation. The bosses after which they are named differ widely in form and size, which has not only led to confusion about the function of tegulae mammatae, but has created artificial distinctions, based on form and not on function. An example of this misinterpretation is the tegula mammata with bosses that are so small and low, that it cannot have functioned as intended, but can best be described as a later with production aides still adhering. If, how, and when tegulae mammatae were used in the Rhine delta is still a matter of study, but the first impressions are that they were first used in the first century AD by the army, and only in the form with rectangular bosses (tegula hamata). KEYWORDS: Roman brick and tile, Ceramic building material, Hypocaustum, Tegula mammata, Tegula hamata, Later, Tubulus. RESUMEN: La tegula mammata, cuyo nombre ya se usaba en la Antigüedad, forma un grupo de material cerámico de construcción usado en sistemas de calefacción y aislamiento bastante enigmático. Los mamelones de los que toma el nombre varían mucho en forma y tamaño, lo que no solo ha inducido a errores acerca de su función, sino que ha llevado también a establecer clasificaciones en virtud de los mismos, creando diferencias ficticias basadas en la forma y no en la función. Un ejemplo de esta mala interpretación es la tegula mammata con mamelones tan pequeños y bajos que no puede haber tenido la función que se le ha atribuido: es más acertado ver en ella un later con los soportes de secado aún adheridos. Cómo y cuándo la tegula mammata se haya utilizado en el delta es todavía un tema de investigación. La primera impresión es que fue introducida en la región en el siglo i d.C. por el ejército bajo la forma de mamelones rectangulares (tegula hamata). PALABRAS CLAVE: ladrillo y tejas romanos, material cerámico de construcción, Hypocaustum, Tegula mammata, Tegula hamata, Later, Tubulus.

One of the more enigmatic forms of ceramic building material are the tegulae mammatae, flat tiles with bosses mounted on the corners, which were used in constructing the inner walls of a room, with the bosses functioning as spacers or separators, creating an open space needed for the functioning of a flue system or as insulation. The form is in some way connected to baths, and is known around the Roman Empire. In its basic form, the tegula mammata has hand-formed, more or less conical, bosses m ­ ounted

on the corners. This form closely resembles the tegula hamata, which has rectangular or triangular bosses with vertical slots on the outer sides. Technically not a tegula mammata, but closely related in function, is a generally thin, rectangular tile with small notches near the corners, that, combined with spacer pins, was used in the same way as the tegula mammata. A remarkable type of tegula mammata has very small bosses that vary in number and spacing, with seemingly no clear function. It is therefore questionable if these are

112

A. (Guus) Gazenbeek

still functional tegulae mammatae. But, if not, what are they? This leads to the question what role tegulae mammatae and hamatae, tubuli and lateres played in Roman construction in the delta of the rivers Rhine, Meuse and Scheldt. TEGULAE MAMMATAE Only two texts from antiquity specifically refer to tegulae mammatae. Vitruvius mentions ‘mammatae tegulae’ in his De Architectura (VII, 4,  2) in the context of solving problems with damp in a room, so in effect as a form of insulation, and not as part of a heating system. In his Historia Naturalis, Pliny the Elder mentions the ceramic building materials produced by a workshop (Historia Naturalis, XXXV, 46, 159). One of the products mentioned are ‘ad balineas mamma­ tis’, but as Pliny gives no additional information on the form, it is uncertain whether he refers to tegulae mammatae, but it is however clear that he associates a ‘lumpy’ form of ceramic building mat­erial with baths. A tegulae mammata can be described as: a flat, rectangular, tile, on which hand-formed, more or less conical, bosses are mounted on one face, generally, but not always, on or near the four corners. The function of these bosses is to create a space between the actual wall and the tiles. In effect, the tegulae were placed on their short sides in vertical rows in front of the wall, to which they were fastened by T-spikes, and subsequently plastered. To ensure that the plaster adhered sufficiently to the tegulae, the outer side of these must be scored by combing or cutting. Their form closely resembles that of the tegulae hamatae, which have rectangular bosses mounted on each corner, generally with a vertical slot on the outer side of each boss, through which the T-spikes were placed (fig. 1). As these bosses are rather heavy in relation to the tile itself, they seemed to break off easily. This problem could therefore have led to the development of other forms such as the flat tile in combination with spacer pins or the tubulus. It is unclear whether these forms were intended solely for use as isolation, or were also used as part of a heating system. There seems to be no functional difference between the tegula mamma­ ta and the tegula hamata. Both are essentially lat­ eres with bosses mounted on the corners. The form and placement of the bosses seems to be more

Anejos de AEspA LXXVII

uniform on the tegulae hamatae than on the mam­ matae. Brodribb (1979: 397-400; 1987: 60-65), Le Ny (1992: 147 and fig. 48, 218) and Bouet (1999: 13-39) have, among others, already noted this differentiation, and have tried to categorize tegulae mammatae according to the height of the bosses, which varies between less than 1 cm and 9 cm. In addition the shapes of the bosses also vary, and rather surprisingly, also their positioning on the tile. That archaeologists therefore have had some difficulty in recognizing and correctly naming (fragments of) tegulae hamatae or mammatae, is not surprising. Even the Dutch archaeologist Bogaers, well steeped in Latin and a stickler for the correct use of it, seems to have mistaken the fragments of tegulae hamatae from Haelen Melen­ borg, for tegulae mammatae.1 It should be remembered however, that the term tegula hamata was unknown in the classical world, and was only coined in the 19th century to describe the form with rectangular bosses. So it could also be possible that Bogaers grasped the real meaning of mammatae as simply meaning ‘with bosses or studs’ and that in Antiquity no differentiation was made between the forms. Among the forms that have been described as tegulae mammatae, a group can be distinguished with bosses that are very low and flat. With a height that rarely exceeds 1,5 cm it seems highly improbable that they could have been intended for use in a heating system or as insulating material. Moreover, these tegulae often do not have four bosses mounted on each corner, but most commonly only one or two, which are more or less placed along the centre line of the tile. This would make it very difficult to mount the tegula flush with either the wall or other tegulae. And finally, it should be noted that none of the lowbossed tegulae mammatae seen by the author had the notches cut into the sides of the tegula, which would be necessary to ensure that they could be mounted without apertures between them, or had any form of scoring, which would be necessary if they were to be used in a wall. In effect, this form resembles more a later with small lumps of clay fixed to it, than a tegula mammata. But whether they can indeed be explained as such is open to debate. Of the examples from the Rhine delta 1  The personal archive of Jules Bogaers (Radboud University, Nijmegen) contains transcripts of the fieldwork, the material is kept at the Provincial Archaeological Depot of the Province of Limburg, Maastricht.

Anejos de AEspA LXXVII

TEGULAE MAMMATAE OR LATERES? ON THE PRESENCE...

113

Fig. 1. Tegulae mammatae (left), tegulae hamatae (centre) and lateres with production aids in situ (right) (author).

know to the author,2 all are loose finds, the original contexts of which are unknown. Only in Tongeren-Kielenstraat (observation by author and Clerbaut, see his article in this publication) were the pieces found in situ, in the pilae supporting the floor of a hypocaust, so essentially they were indeed used as lateres. Elsewhere this can also be seen, as for example in Meudon (Barat 2002: 232, fig. 6) or Chassenon (Coutelas 2012: figs. 6-8). It seems evident that the builders did not consider the lumps problematic, as they would otherwise have removed them before use. A more probable explanation, we believe, is therefore that these lumps are in some way related to the production of the lateres.3 Tegulae mammatae (and hamatae) that are still in situ, for example in the Stabian Baths in Pompei (Adam 2011: 292, fig. 630), give an impression as to how they were actually used in antiquity. Not surprisingly, this corresponds with 2  See Clerbaut, this volume, fig. 4. Also Sint Truiden (Melkert in prep.) and Meerssen-Herkenberg (Habets 1871: 201-246). 3   Joint paper presented at Oxford in 2015 by Tim Clerbaut and author. See also the paper of Tim Clerbaut in this volume.

what Vitruvius mentions in his advice on dealing with excessive damp in a room. He does not ­specifically mention the use in a heating-system (hypo­caust). To be functional in such a system, the height of the bosses should at least allow for enough space between the tiles and the wall so that sufficient air can pass between them, and so ensure that a heating system can function. In fact, as Adam (Adam 2011: 292) points out, the height of the bosses, around 5 cm, is insufficient to create the space needed to ensure that there is enough draft to allow a heating system to function properly. As an insulation system, however, it would function well. NO TEGULAE MAMMATAE IN THE RHINE DELTA? If our interpretation of the tegulae mammatae with low bosses is correct, then this form is simply a later with the technical aids used during its ­production still adhering. Furthermore, to the ­author’s knowledge, no tegulae mammatae with conical bosses have as yet been found in the Rhine

114

A. (Guus) Gazenbeek

delta. However, the closely related tegulae hama­ tae have been found on a number of sites, sometimes labelled as tegulae mammatae, and sometimes not at all.4 Also flat tiles with notches have been found on occasion, as have spacer pins, although in smaller numbers and on even fewer sites.5 As, until now, all the examples seen by the author are either tegulae hamatae, flat tiles with notches or lateres with flat bosses, it seems reasonable to suppose that the chances are slim that tegulae mammatae with conical bosses will turn up in the delta. As the use of tegulae mammatae and hamatae was almost certainly the same, the lack of mammatae on these sites does not mean that the constructions they were used in were absent in the delta. However, the sites on which lat­ eres with technical aids have been found should be reconsidered, as these lateres cannot be seen as an indication for a building with a heating system on or near the site. INTERPRETING FINDS OF TEGULAE HAMATAE AND RELATED FORMS IN THE RHINE DELTA In North Western Europe, and certainly in the Low Countries, tegulae mammatae, tegulae hama­ tae and tubuli are, more often than not, seen as proof, or at least as an very strong indication, of the presence of a hypocaust on the site under study, even if further evidence of the structures themselves is lacking. However, some of these sites will surely have had buildings with heated rooms or baths, as it is difficult to imagine a cas­ tellum without bathing facilities of some kind, or, for that matter, a domus or villa without at least one room with a hypocaust. In figure 2, 29 sites from the Rhine delta are listed at which building material has been collected and described in sufficient detail to allow at least a basic analysis of the types of ceramic building material used on these sites. Virtually no site has been completely excavated, nor excavated by the same institutions, so that the data is prob4   In the Archaeological depot of the Dutch province of Zuid-Holland the author encountered by chance two complete tegulae hamatae from the military site of Valkenburg, which had simply been registered as ‘brick’. 5   For flat tiles with notches see Figure 2, column ‘Plates’. No spacer pins were found at the sites mentioned in this column. They have been found in the Flavian castra of Nijmegen (Zandstra and Polak 2014: 223-224 and Afb 6.2) and in the villa of Afferden (Vermeulen-Bekkering 2006: 48-49).

Anejos de AEspA LXXVII

ably biased to some extent. Furthermore, virtually no material was found in situ, but nearly always in secondary contexts. Roofing material (tegulae and imbrices) is of course found on every site, but the percentage differs, depending on the context of the site. At rural sites it makes up more than ninety-five per cent, generally even more than ninety-eight per cent, of the total. Military, urban (including vici) or villa sites, have a higher percentage of other ceramic building material, mostly lateres but also forms that can be related to heating and insulating. On all non-rural and villae contexts all forms used in heating systems are present, with the exception however of tegulae mammatae. Tegulae hamatae are almost only found on military and urban sites, the exceptions being Houten and Naaldwijk. Houten (De Groot 2015; Gazenbeek 2015) is situated on the lower Rhine, the material from the location studied having been retrieved from a medieval rural settlement where it probably had been reused in some way. It certainly must have originated from a Roman site in the near vicinity, possibly the castellum of Fectio. The rather enigmatic site of Naaldwijk (Goossens 2012; Van der Feijst 2012) lies near the Helinium, the then estuary of the rivers Meuse and Scheldt. The site has produced a large number of stamps of the Classis Germanica, but as yet no military structures have been found, so it remains unclear if Naaldwijk was a military site or a vicus connected in some way to naval and riverine shipping. On the lower Rhine it was the army who, from around the middle of the first century onwards, were the first to use ceramic building material, so it is therefore not surprising that tegulae hamatae are found on these military sites. That they also appear on the urban sites of Nijmegen and Forum Hadriani is more interesting, as both municipia were founded around the end of the first, or very early in the second century. However, it is widely accepted that both towns where founded by Hadrian, so that a military connection in at least part of the actual building process does not seem improbable. The use of these forms, well known to the Roman military, is therefore not surprising, although it should be noted that the use here was either very late, or that these new cities incorpora­ ted existing buildings, as these forms were superseded by the tubulus in the course of the last quarter of the first century. It is interesting to note that in Tongeren, tegulae hamatae and mammatae

TEGULAE HAMATAE %

80.7

10.7

0.1

0.2

2.1

2.1

4.4

3.8

80

80

17.5

–

1.3

1.3

–

2.6

–

 

 

260

91.2

6.1

–

–

0.4

–

0.4

2.3

628

79.1

8.4

–

0.3

0.8

1.3

2.4

10.1

51.6 10.9*

0.3

3.5

10.7

4.1

18.3

18.9

OTHER %

TUBULI %

PLATES %

TOTAL HEATING %

TOTAL LATERES %

LATERES, low bosses %

TOTAL ROOFING %

115

840

N

SITE

CONTEXT

TEGULAE MAMMATAE OR LATERES? ON THE PRESENCE...

SOURCE

Anejos de AEspA LXXVII

– Matilo 1

Gazenbeek 2016a

– Matilo 2

Gazenbeek in press-a

VLEUTEN (NL) – Wachttoren

Gazenbeek 2013e

– LR 62

Gazenbeek 2012b

NIJMEGEN (NL) – Canisius

MILITARY

LEIDEN (NL)

Zandstra and Polak 2014

BEST (NL)

Bink 2010

BREDA (NL) – West

Hoegen, Koster and van Enckevort 2004

DEURNE (NL)

Hiddink 2008

2,144

DIESSEN (NL) – Rijtseweg ARCHIS, n° 408186

 

68

98.6

1.5

–

–

–

–

–

318

93.7

6.3

–

–

–

–

–

37

100

0

–

–

–

–

–

120

100

0

–

–

–

–

–

ECKELRADE (NL)

Gazenbeek 2013a

37

100

0

–

–

–

–

–

GEYSTEREN (NL) – Leeberg

ARCHIS, n° 2320081

15

100

0

–

–

–

–

–

297

99

1

–

–

–

–

–

HELDEN (NL) – Schrames

Gazenbeek 2010

589

99.1

0.8

–

–

–

–

–

HOLTUM (NL) – Noord

Gazenbeek 2012c

72

100

0

–

–

–

–

–

LIESHOUT (NL) – Beekseweg

Hiddink 2005a

428

100

0

–

–

–

–

–

MAASMECHELEN (B) – Mottekamp

Gazenbeek in prep-a

312

98.1

1.9

–

–

–

–

–

MEISE (B)

Van Liefferinge and Smeets 2013

1,368

100

0

–

–

–

–

–

NEDERWEERT (NL) – Rosveld

Hiddink 2005b

318

99.7

0.3

–

–

–

–

–

NISTELRODE (NL)

Gazenbeek 2016b

37

94.6

5.4

–

–

–

–

–

 

 

RURAL

HEERLEN (NL) – Trilandis Gazenbeek 2014b

OSS (NL) – Horzak

unpublished

– Westerveld

Wesselingh 2000

94

84

13.8

–

–

–

–

161

95.7

2.5

–

–

1.9

–

UDEN (NL) – Noord HOUTEN (NL) – Stenen Poort

2.2

Gazenbeek 2013b

57

94.7

5.3

–

–

–

–

Gazenbeek 2015

55

72.8 12.7*

–

1.8

7.3

5.4

14.5

96.4

1.9

– – –

NIJMEGEN (NL) unpublished Gazenbeek 2013c

VOORBURG (NL) – Forum Hadriani I

Gazenbeek 2009

– Forum Hadriani II

Gazenbeek 2014a

173 URBAN

– Hertogstraat – Rijn/Lekstraat

2.3

–

1.2

–

–

1.2

–

1,121

87.4 10.6*

–

0.2

1.1

0.6

1.9

0.1

2,342

82.2 10.9*

–

0.1

6.2

0.6

6.9

–

6,719

86.7 11.2*

–

0.1

1.9

0.2

2.2

–

1,148

71.1 26.7*

0.9

–

1.6

–

1.6

–

TONGEREN (B) – Kielenstraat Museum

Clerbaut in prep.

AARDENBURG (NL)

OTHER %

PLATES %

TUBULI %

TEGULAE HAMATAE %

LATERES, low bosses %

TOTAL LATERES %

TOTAL ROOFING %

N

CONTEXT

SOURCE

TOTAL HEATING %

Anejos de AEspA LXXVII

A. (Guus) Gazenbeek

SITE

116

Gazenbeek in prep. b

51

96.1

3.9

–

–

–

–

0

–

– Panheelderweg

Gazenbeek in prep. c

108

97.2

2.7

–

–

–

–

0

–

– Oud onderzoek

Gazenbeek in prep. c

126

100

0

–

–

–

–

0

–

1.8

16.8

1.3

1.2

–

NAALDWIJK (NL)

VICUS

HEEL (NL)

Gazenbeek 2012a

554

64.3

17.7

–

0.2

14.8

– ‘t Zandheultje

Gazenbeek 2012d

252

90.9

8

–

–

1.2

– Hoogwerf II

Gazenbeek in press b

147

77.6

15

–

0.7

6.1

0.7

7.5

–

– Tradepark

Gazenbeek in prep. d

356

66.3

12.9

–

0.3

20.2

0.3

20.8

–

EWIJK (NL) – Keizershoeve

unpublished

1,380

88.4

9.4*

–

–

1.7

0.5

2.2

0.1

KERKRADE (NL) – Holzkuil

Tichelman 2005

439

59.6 23.9*

–

–

16.4

–

16.4

–

MAASBRACHT (NL)

Gazenbeek in prep. e

286

57 10.8*

–

–

26.6

5.6

32.2

–

VILLA

– Hoogwerf

23.537 Fig. 2. Material according to type, as a percentage of all Roman ceramic building material from the given site. The presence of round lateres is indicated with an asterisk. Plates are flat tiles with scoring on one side and small notches near each corner and were used in combination with spacer pins. Roofing material consists of tegulae and imbrices (note that ARCHIS is the national archaeological database of The Netherlands).

seem to be absent, although the town certainly developed in the first century AD. It is possible to think of the tegulae hamatae and mammatae as forms that are connected to the military, but at this point the data is still insufficient to be more positive on this point. Tubuli are much more common, and are found on all villa, urban and military sites.6 This, however, should not be surprising, as most of the sites started or continued their development in the second century or later, when the tubulus was the standard form used in heating structures. However, although they seem to be very rare on rural sites, they do on occasion turn up amongst the ceramic building material found on these sites. On these sites lateres are also found, albeit only in very small numbers, generally less than 5 per cent of the total. As it is clear that these lateres will not  have been used as roofing material, another 6  The site Vleuten Wachttoren concerns a watchtower constructed in timber, so that tegulae, imbrices and lateres are the only forms of ceramic building material one should expect from this site.

e­ xplanation must be found. They could very well have been used in hearths built in the Roman style, and not in the traditional local style (fig. 3). The even rarer tubuli fragments on these sites could possibly also be related to these hearths, as part of a chimney. A use in this form closely resembles their intended function in a hypocaust, where they form part of the flue system to exhaust the hot air and smoke. CONCLUSION It is clear that in contemporary use the term tegula mammata refers to a very diverse group of products, which do not have the same form or function. Some, with more or less long bosses, closely resemble the tegula hamata and will have been used in similar functions in heating and insulation systems. The form with low, flat lumps, however, was never intended for such a function, but is simply a later with the aides used during its  production still adhering. The question remains why so few of these lateres are found, or

Anejos de AEspA LXXVII

TEGULAE MAMMATAE OR LATERES? ON THE PRESENCE...

117

Fig. 3. Hearth in the kitchen of a domus in Grand (France) (photo by author).

r­ecognized. Without doubt, the weak adhesion between the lump and the later plays an important role in this. These lateres have been encountered on military and urban sites, on all of which the use of ceramic building material starts in the first century A.D. on an extensive scale. But they are also found on villa sites, which generally date from the second century A.D. and later. It therefore seems unlikely that they are related to a specific period or organisation. Real tegulae mammatae with conical bosses seem to be absent on all sites in the delta. The functionally identical form of the tegula hamata however, is present on what are more or less military sites, but absent on rural and villa sites. A preliminary conclusion could be that this form was used only by the military, which preferred the version with rectangular bosses above the form with conical bosses. It is clear that many questions concerning the production and use of these forms are still unanswered.

ACKNOWLEDGMENTS First of all I would like to thank Linda Bogaert (Provincie Limburg) who first showed me the mat­ erial from Tongeren Kielenstraat, and so started a train of thought that forms the basis of this paper. Furthermore I would like to thank Twan Ernst (independent researcher, Venlo), who ferreted out a number of unpublished fragments of ‘tegulae mammatae’ from sites in the Meuse valley and very kindly shared this information with me. Also Rien Polak (Radboud Universiteit Nijmegen) deserves mention for kindly supplying me with information from the archive of Jules Bogaers, as does Marian Melkert, who shared as yet unpublished material from Sint Truiden. Special thanks however, go out to Tim Clerbaut (Ghent University), who asked me to join him in researching the material from the site of Tongeren Kielenstraat. Archaeology is quite often a family matter, and this paper forms no exception. A. Gazenbeek

118

A. (Guus) Gazenbeek

reviewed and corrected my all too rusty English, Pilar Martin Ripoll translated the abstract into Spanish, and M. Gazenbeek first introduced me to the Roman kitchen, and so to a possible explanation for lateres on rural sites. REFERENCES Adam, J.-P. 2011: La construction romaine. Matériaux et techniques. Picard, Paris. Barat, Y. 2002: “Un atelier de tuiliers d’époque romaine (IIIe s.) à Meudon (Hauts-de-Seine)”, Revue Ar­ chéologique du Centre de la France, 41, pp. 225-237. Bink, M. 2010: Best Aarle-Hokkelstraat, Fase 1. Inven­ tariserend Veldonderzoek door middel van Proefsleu­ ven, Gemeente Best-BAAC Rapport A-09.0297-I. BAAC bv, ’s-Hertogenbosch-Deventer. Bouet, A. 1999: Les matériaux de construction en terre cuite dans les thermes de la Gaule Narbonnaise, Scripta Antiqua 1. Ausonius, Bordeaux. Brodribb, G. 1979: “Tegulae mammatae”, The Anti­ quaries Journal, 59, pp. 397-400. Brodribb, G. 1987: Roman Brick and Tile. Sutton, Gloucester. Clerbaut, T. in prep: “Chapter 5. Ceramic building materials”, in Driesen, P., Bogaert, L. and Creemers, G. (eds.), Excavations Under the Museum, Atuatuca Series. Tongeren. Coutelas, A. 2012: “Les méthodes de travail pour l’étude des terres cuites architecturales retrouvées à Cassinomagus (Chassenon, Charente)”, in SF­ ECAG. Actes du Congrès de Poitiers (17-20 mai 2012), pp. 711-717. Société française d’études de la céramique antique en Gaule, Marseille. De Groot, R. 2015: Plangebied de Stenen Poort (paarden­ wei), gemeente Houten; archeologisch onderzoek: een opgraving, RAAP rapport 2936. RAAP bv, Weesp. Gazenbeek, A. E. 2009: “Bouwkeramiek en natuursteen”, in Bink, M., Franzen,P. F. J. (eds.): Forum Hadriani Voorburg. Definitief Archeologisch onderzoek. BAAC-rapport, A 05.0125, pp. 215-261. BAAC bv, ’s-Hertogenbosch-Deventer. Gazenbeek, A. E. 2010: “Bouwkeramiek”, in de Winter, J. (ed.), Archeologisch onderzoek op het plangebied Schrames te Helden. Bewoningssporen van het neolith­ icum tot de late middeleeuwen, pp. 197-217, BAACrapport A-07.0204. BAAC bv, ‘s-Hertogenbosch. Gazenbeek, A. E. 2012a: “Grofkeramisch bouwmateriaal”, in Goossens, T. A. (eds.), Van akker tot Hooghwerf. Onderzoek naar de bewoning in de ijz­ ertijd, inheems-Romeinse tijd, middeleeuwen en nieu­ we tijd op de haakwal van Naaldwijk (plangebied

Anejos de AEspA LXXVII

Hoogeland, gemeente Westland), pp. 321-354, Archol-Rapport 167. Archol, Leiden. Gazenbeek, A. E. 2012b: “Bouwkeramiek en natuursteen”, in Aarts, A. C. (ed.), Scherven, schepen en schoeiingen. LR 62: Archeologisch onderzoek in een fossiele rivierbedding bij het castellum van de Meern, pp. 91-132, Basisrapportage Archeologie 43. Gemeente Utrecht, Utrecht. Gazenbeek, A. E. 2012c: “Bouwkeramiek en natuursteen”, in Tichelman, G. (ed.), Germanen aan een Maasgeul in Holtum-Noord. Proefsleuven en opgrav­ ing in Holtum-Noord II, deelgebied Geko fase 2, Ge­ meente Sittard-Geleen, pp. 137-149, RAAP rapport 2417. RAAP bv, Weesp. Gazenbeek, A. E. 2012d: “Natuursteen en keramisch bouwmateriaal”, in van der Feijst, L. M. B. (ed.), Vechten tegen het wassende water in de Romeinse tijd. Een archeologische opgraving in plangebied Hooge­ land-Oost, ‘t Zand Heultje te Naaldwijk, pp. 109136, ADC-rapport 3186. ADC, Amersfoort. Gazenbeek, A. E. 2013a: “Bouwkeramiek en natuursteen”, in Hensen, G. (ed.), Resten van landelijke nederzettingen uit de Midden IJzertijd tot en met de Romeinse Tijd in Eckelrade. Gemeente Eijsden-Mar­ graten. Archeologisch onderzoek: een opgraving. Weesp, RAAP-rapport 2713. RAAP bv, Weesp. Gazenbeek, A. E. 2013b: “Keramisch bouwkeramiek”, in Goossens, T. A. and Meurkens, L. (eds.), Neder­ zettingen uit de vroege ijzertijd en Romeinse tijd in Uden-Noord (gemeente Uden). Een opgraving op de nieuwbouwlocatie van streekziekenhuis Bernhoven, pp. 99-115. Archol-Rapport 188. Archol, Leiden. Gazenbeek, A. E. 2013c: “Grofkeramiek en natuursteen”, in Heirbaut, E. N. A. (ed.), De zuidwesteli­ jke hoek van Ulpia Noviomagus in kaart gebracht. Deel 2 Beschrijving van de vondsten van de Rijnstraat en Lekstraat in Nijmegen-West, pp. 301-330, Archeologische Berichten Nijmegen – Rapport 42. Gemeente Nijmegen, Nijmegen. Gazenbeek, A. E. 2013d: “Grofkeramiek en natuursteen”, in Dielemans, L. (ed.), Wacht aan het wa­ ter. VLEN3-00: archeologisch onderzoek naar sporen en vondstconcentraties uit de Romeinse tijd in Vleu­ terweide, pp. 99-118, Basisrapportage archeologie 52, Gemeente Utrecht, Utrecht. Gazenbeek, A. E. 2014a: “Grofkeramiek van Voorburg-Arentsburg”, in Driessen, M. J. and Besselsen, E. (eds.), Voorburg-Arentsburg: Een Romeinse ha­ venstad tussen Rijn en Maas, vol. 2, pp. 503-545, Themata 7. University of Amsterdam, Amsterdam. Gazenbeek, A. E. 2014b: “Het grofkeramiek en natuursteen”, in Tichelman, G. (ed.), Een non-villa neder­ zetting uit de Romeinse tijd op het lössplateau bij

Anejos de AEspA LXXVII

TEGULAE MAMMATAE OR LATERES? ON THE PRESENCE...

Heerlen, Gemeente Heerlen; archeologisch onderzoek: opgravingen op bedrijventerrein Trilandis, Deel 2, pp. 237-268. RAAP rapport 2732. RAAP bv, Weesp. Gazenbeek, A. E. 2015: “Grofkeramiek en natuursteen”, in de Groot, R. (ed.), Plangebied de Stenen Poort (paardenwei), gemeente Houten; archeologisch onderzoek: een opgraving, pp. 91-133, RAAP rapport 2936. RAAP bv, Weesp. Gazenbeek, A. E. 2016a: “Grofkeramiek, natuursteen en mortel”, in Brandenburgh, C. R. and de Bruin, J. (eds.), Met de voeten in het water. Archeologisch onderzoek aan de oostzijde van castellum Matilo te Leiden, pp. 138-196. Gemeente Leiden, Leiden. Gazenbeek, A. E. 2016b. “Grofkeramisch bouwmateriaal”, in Hensen, G. and Janssens, M. P. J., Dolen door de Zwarte Molen: onbegrensde nederzettingen uit de IJzertijd tot en met de Volle Middeleeuwen. Gemeente Bernhese. Archeologisch onderzoek: een opgraving. Weesp. (RAAP-rapport 2794), 225-240. Gazenbeek, A. E. in prep. a: “Grofkeramiek en natuursteen”, in M. Smeets en M. Steenhoudt (red.) [Het archeologisch onderzoek te Maasmechelen-Mot­ tekamp]. Studiebureau Archeologie. Kessel-Lo. Gazenbeek, A. E. in prep. b: “Grofkeramiek en natuursteen”, in Wattenberghe, J. (ed.), Archeologisch onderzoek rondweg Aardenburg, gemeente Sluis, ArteFact! Rapport, Middelburg. Gazenbeek, A. E. in prep. c: “Grofkeramiek”, in van Diepen, L. (ed), Uitwerking oud-onderzoek neder­ zetting Heel, Grontmij Archeologische Rapporten. Eindhoven. Gazenbeek, A. E. in prep. d: “Grofkeramiek en natuursteen”, in Leijnse, K. (ed.), Naaldwijk TradeParc, RAAP-Rapport. RAAP bv, Weesp. Gazenbeek, A. E. in prep. e: “The building material from the villa of Maasbracht”, in Willems, W. J. H. and Vos, W. K. (eds.), Villa Maasbracht Steenakker, Analecta Praehistorica Leidensia. Gazenbeek, A. E. in press a: “Natuursteen en grofkeramiek”, in van de Feijst, L. M. B. (ed.), Park Matilo. Een archeologische begeleiding. ADC Rapport. ADC, Amersfoort. Goossens, T. A. (ed.) in press b: Van akker tot Hooghwerf. Onderzoek naar de bewoning in de ijzertijd, inheems-­ Romeinse tijd, middeleeuwen en nieuwe tijd op de haak­ wal van Naaldwijk (plangebied Hoogeland, gemeente Westland), Archol-Rapport 167. Archol, Leiden. Habets, J. 1871: Découvertes d’antiquités dans le duché de Limbourg, vol. I. Romen, Ruremonde. Hiddink, H. 2005a: Archeologisch onderzoek aan de Beek­ seweg te Lieshout (Gemeente Laarbeek, Noord-Bra­ bant), Zuidnederlandse Archeologische Rapporten 18. Archeologisch Centrum Vrije Universiteit, Amsterdam.

119

Hiddink, H. 2005b: Opgravingen op het Rosveld bij Nederweert 1. Landschap en bewoning in de IJzertijd, Romeinse Tijd en Middeleeuwen, Zuidnederlandse Archeologische Rapporten 22. Archeologisch Centrum Vrije Universiteit, Amsterdam. Hiddink, H. 2008: Archeologisch onderzoek op de Groot Bottelsche Akker bij Deurne. Bewoning uit de Steenti­ jd, IJzertijd, Romeinse Tijd, Vroege en Volle Mid­ deleeuwen, Zuidnederlandse Archeologische Rapporten 33. Archeologisch Centrum Vrije Universiteit, Amsterdam. Hoegen, R. D., Koster, A. A. and van Enckevort, H. 2004: “Voorwerpen van metaal, glas, steen en aardewerk uit de Late IJzertijd en de Romeinse Tijd”, in Koot, C. W. and Berkvens, R. (eds.), Bre­ dase akkers eeuwenoud. 4000 jaar bewoningsgeschie­ denis op de rand van zand en klei, pp. 359-376, ROB 102. ROB, Amersfoort. Le Ny, F. 1992: La production des matériaux de con­ struction de terre cuite en Gaule romaine. Thèse non publiée, Université de Rennes I (4 vol.). Melkert, M. J. A. in prep: “Natuursteen, bouwmateriaal en keramische objecten uit het Neolithicum, de IJzertijd, de Romeinse tijd en de Middeleeuwen”, in Hazen, P. L. M. (ed.), Brustem-Kapelhof (Gemeente Sint Truiden), VEC-Rapport. Tichelman, G. 2005: Het villa complex Kerkrade-Holz­ kuil, ADC Archeo projecten Rapport 155. ADC, Amersfoort. Van der Feijst, L. M. B. (eds.) 2012: Vechten tegen het wassende water in de Romeinse tijd. Een archeolo­ gische opgraving in plangebied Hoogeland-Oost, ‘t Zand Heultje te Naaldwijk, ADC-rapport 3186. ADC, Amersfoort. Van Liefferinge, N. and Smeets, M. 2013: Het arche­ ologisch onderzoek aan het Heimbeekveld te Meise, Archeo-rapport 191. Studiebureau Archeologie bvba, Kessel-Lo. Vermeulen-Bekkering, A. M. 2006: Een Romeinse villa langs de Maas bij Afferden. Gemeente Bergen (Limburg), Rapportage Archeologische Monumentenzorg 116. Rijksdienst voor het Oudheidkundig Bodemonderzoek, Amersfoort. Wesselingh, D. A. 2000: Native Neighbours. Local Set­ tlement System and Social Structure in the Roman Period at Oss (The Netherlands), Analecta Praehistorica Leidensia 32. Leiden University Press, Leiden. Zandstra, M. J. M. and Polak, M. 2014: “6. Keramisch Bouwmateriaal”, in Kloosterman, R. P. J., Polak, M. and Zandstra, M. J. M. (eds.), Opgravingen op het terrein van het voormalige Canisiuscollege in Nijmegen, 1987-1997, Vondsten uit castra en canabae I, pp. 221-258, Auxiliaria 14. Auxilia, Nijmegen.

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

THE END OF THE TEGULAE MAMMATAE? A REVIEW ON THEIR NAME, FUNCTION(ALITY) AND PRESENCE IN THE ROMAN NORTH TIM CLERBAUT Historical Archaeology Research Group (HARG) at Ghent University, Belgium

ABSTRACT: This paper intends to give a critical review of one of the most renowned forms of ­ceramic building material: the tegulae mammatae. By combining the scarce written sources with ­archaeological evidence the complexity of defining this specific group is addressed. The large differentiation, already mentioned by other authors as well, poses questions on their function(ality), especially for the subgroup with very low bosses. It is argued here that this subgroup cannot be seen as  functional tegulae mammatae as described by Vitruvius. This conclusion has important consequences for their interpretation in archaeological contexts especially in the Roman North, where they are predominantly attested. KEYWORDS: Roman brick and tile, Ceramic building materials, Hypocaustum, Tegula mammata. RESUMEN: Este artículo plantea una revisión crítica de una de las formas más interesantes en el ámbito del material de construcción cerámico: las mammatae tegulae. El estudio se dirige a la combinación de las escasas fuentes escritas con la evidencia arqueológica respecto a la complejidad de definición de este grupo específico. La gran diferenciación existente, ya mencionada por otros autores, plantea cuestiones sobre su función(alidad), especialmente para el subgrupo con salientes muy bajos. Se argumenta que este subgrupo no puede ser considerado como mammatae tegulae según la descripción de Vitruvio. Esta conclusión tiene consecuencias importantes para su interpretación en el contexto arqueológico, especialmente para el norte del mundo romano, donde se encuentran ampliamente representadas. PALABRAS CLAVE: ladrillo y tejas romanas, materiales cerámicos de construcción, Hypocaustum, Tegulae mammatae.

Ceramic building materials are an important technical development within classical architecture. Apart from other benefits, the large diversity of forms that could be shaped out of clay is endless. This gave the opportunity to architects and engineers to develop specific forms for specific needs. One of these specially designed forms is the tegula mammata (‘nippled tile’). Thanks to contemporary texts the intended function of these specific tiles is known. Nevertheless, the study of architectural remains and archaeological finds mainly from the Roman North, sketch a more diverse picture of this specific tile form. This raises important questions about their function(ality) and use throughout the Empire.

It is the intention of this paper to confront classical sources and archaeological evidence in an attempt to answer some questions concerning their function and use by explaining the large diversity attested. To do so, classical texts and arch­ aeological finds will first be addressed separately, after which the results will be confronted. The focus hereby will be mainly on the so called tegu­ lae mammatae with very short bosses which seem to be more abundant in the Roman North.1

1   For a more detailed account on this topic, please see the paper of A. G. Gazenbeek in this volume.

122

Tim Clerbaut

TEGULAE MAMMATAE IN CONTEMPORARY WRITTEN SOURCES2 It is without doubt that written sources concerning tegulae mammatae are scarce and ambiguous. Nevertheless, it is interesting to look at what contemporary writers mention so that we can come to a clear definition of this form of ceramic building material. A logical source is of course De Architectura of Vitruvius (c. 80-20 BC). He mentions tegulae mammatae in relation to solving problems with damp in a room: sin autem locus non patietur structuram fieri, canales fiant et nares exeant ad locum patentem. deinde tegulae­bipedales ex una parte supra marginem cana­ lis inponantur, ex altera parte bessalibus (laterculis) pilae substruantur, in quibus duarum tegularum anguli­sedere possint, et ita a pariete eae distent ut ne plus pateant palmum. deinde insuper erectae mam­ matae tegulae ab imo ad summum ad parietem figan­ tur, quarum interiores partes curiosius picentur ut ab se respuant liquorem item in imo et in summo supra camaram habeant spiramenta. Vitruvius, De Archi­ tectura, VII, 4, 2.

Anejos de AEspA LXXVII

rex septimum collegium figulorum instituit. Pliny, Historia Naturalis, XXXV, 46, 159.

Several authors cite this passage, stating the use of tegulae mammatae3 mentioned by the author. It is, however, debatable whether the ‘ad balin­eas mammatis’ mentioned are a­ ctually ­tegulae mammatae. When Pliny mentions the other products, an additional type and specific function are always added (e.g. storage jars for wine, piping for water, imbrices for (roof) covering). It is clear that ‘mammatis’ had a use in the construction of baths, but it is not so obvious in what form, as neither tegula nor later are mentioned but only presumed. Does this imply that these types were widely known and/or used, so that no further explanation was needed? TEGULAE MAMMATAE IN ARCHAEOLOGY: IN VARIETATE CONCORDIA?

neque adsiduitate satiant figlinarum oprea, doliis ad vina excogitatis, ad aquas tubulis, ad balineas mam­ matis, ad tecta imbricibus, coctilibus laterculis fun­ damentisque aut quae rota fiunt, propter quae Numa

When looking at tegulae mammatae in the arch­aeological record, the large variety of names given to these tiles is an important weakness. In English alone, at least five alternative names are used: ‘leg tiles’ (e.g. Gunther 1913: 224), ‘nipple tiles’ (e.g. Winbolt 1922: 104), ‘brick with knobs’ (e.g. Wright 1939: 78), ‘brick with clay stubs’ (e.g. Lethaby 1923: 25) and ‘brick with bosses’ (e.g. ­Cunliffe 1971: 43). In French the name ‘tuile à mamelons’ is most commonly used (e.g. Ferdière 2012: 20) as an alternative for tegulae mammatae. Other alternatives, in Dutch, are ‘noppentegel’ (e.g. Clerbaut and Geerts 2014) or ‘nokkentegel’ (e.g. Zandstra and Polak 2014: 224). This large diversity in names can be better under­stood when looking at the variety of forms in which tegulae mammatae appear in the archaeo­ logical record. Several authors already noted the substantial differences and tried to divide them in subgroups. Brodribb (1979: 397-400; 1987: 60-65) did this for the known examples from Britain in the eighties, and a decade later Le Ny (1992: 147 and fig. 48, 218) and Alain Bouet (1999: 13-39) did the same for material from France. All of these categorizations were based on the height of the bosses, a characteristic with a wide variation from less than 1  cm to almost 9  cm. This

2   A special thanks goes to Guido Cuyt for his help with the translation and interpretation of the Latin texts.

3   Not all versions of the text seem to agree with mammatae (‘nippled’), as some mention hamatae (‘hooked’) instead.

Indisputably the ‘mammatae tegulae’ mentioned here refer to tiles which have mammae on their surface. These tiles were specifically used to create cavity walling, with the mammae functioning as spacers. Interestingly enough, in his advice on the building of baths, Vitruvius only mentions the use tegulae mammatae as insulation in damp rooms, with no mention whatsoever about how the flue gases should be evacuated from the hypo­ caustum. The only other writer in antiquity who mentions mammatae is Pliny the Elder (23-79 AD) in his Historia Naturalis. In a passage about different products and their origin, Pliny refers to a ceramic workshop that produces a large spectrum of innovative ceramic products. Among these products are wine vessels (dolia ad vina) and different types of ceramic building materials:

Anejos de AEspA LXXVII

variation led to a differentiation between a type with short bosses (Bouet’s groupe 1 and Brodribb’s Type A) and long bosses (Bouet’s groupe 2 and Brodribb’s Type B). An important argument in making this division is its intended function as a cavity tile as described by Vitruvius. To be functional, the height of the bosses should at least be such that there is enough space between the tiles and the wall to allow for sufficient air to pass between the wall and the tile to ensure that a heating system can function. Adam (2005: 556-557) concluded on the basis of tegulae mammatae still in situ in the Stabian baths of Pompeii (fig. 1), that the height of the bosses, around 5 cm, is insufficient to create enough air draught needed for a heating system. He therefore sees the function of these tiles, at least in this case, more in the light of isolation than as part of a heating system. This interpretation is supported by Bouet (1999: 31-32), but he sees still some clear examples (e.g. the finds of Toulousain) in Gallia Narbonensis where without doubt their function is linked to heating. TEGULAE MAMMATAE WITH SHORT BOSSES IN THE ROMAN NORTH AND THEIR FUNCTION(ALITY) As discussed by Gazenbeek in this volume, a subgroup with very low bosses can be identified, the function of which as a tegula mammata can clearly be disputed based on the height and the positioning of the bosses (fig. 2, and Brodribb 1987: Annexe III).4 Alternative terminology already exists in the literature, where some authors are using tegula sine marginibus5 to point to this specific group with short bosses. THE TWO SUBGROUPS OF TEGULAE MAMMATAE AND THEIR DISTRIBUTION It is clear that at least two different subgroups can be distinguished among what in literature are mentioned as tegulae mammatae. For the sake 4   Joint paper presented at Oxford in 2015 by A. Gazenbeek and author. See also Gazenbeek in this volume for a wellillustrated discussion. 5   Non-classical term, nevertheless giving a good idea of the type described here. Already in use by several authors (e.g. Havas 2009).

THE END OF THE TEGULAE MAMMATAE?...

123

Fig. 1. Tegulae mammatae used to create cavity walling for the women’s caldarium of the Stabian Baths. Photo: S.A Jashemski.

of  the discussion, a simple division into a subgroup of functional (in the sense of Vitrivius’ ­description as cavity tile) tegulae mammatae and a subgroup of lateres with (very) short bosses was made. To get a better understanding of both subgroups and their appearance in the archaeological record, two distribution maps were drawn (fig. 3) based on intensive study of the available literature. As a base, the work of Brodribb for Britain (1987), Le Ny for the whole of France (1992), and Bouet (1999) for Southern France and adjoining regions was used and further supplemented by the author’s synthesis of the ‘tegula mammata’ finds for Belgium and the Netherlands (fig. 4) and some recent excavation reports for other regions. Despite the fact that the distribution patterns for the two groups are without doubt preliminary and incomplete, they show a remarkable difference. Whereas the subgroup with large bosses is more concentrated towards the Mediterranean, the subgroup with short bosses seems to cluster in the more northern regions of the Empire. The difference is remarkable and cannot be explained by differences in the progress. This makes us wonder about possible explanations for this clear difference. Do they reflect more than just different functions? AN ALTERNATIVE FUNCTION FOR THE SHORT BOSSES Scientific debate during and after the Oxford Conference led to a broadly supported believe that the function of this short bosses should be found elsewhere.

124

Tim Clerbaut

Anejos de AEspA LXXVII

Fig. 2. Tegulae mammatae and the observed positions of the bosses: Left: Later with short bosses from the site of TongerenKielenstraat (Belgium). Photo: Tim Clerbaut © Collectie Gallo-Romeins Museum. Right: Differentiation of the number and position of the bosses. After Brodribb 1987: 61 with personal revision.

An acceptable explanation could be linked to the production of this form and not to an intended­ function after firing. In a kiln, it is important that hot air can pass freely between the pieces stacked in the oven. Therefore, these ‘lumps’  could have had the function of kiln spacers, guaranteeing an sufficient air flow during the firing p ­ rocess. ­Wheeler already proposed this idea in 1936 (Wheeler and Wheeler 1936) as did Bouet (1999: 38). Detailed observation by several colleagues on tiles found on sites in Gallia and Germania Inferior seem to prove this point as well. More interesting however is the fact that in relation to these lumps, also small insents can be found on similar positions on the reverse side of lateres as well. Where these lumps are missing, small cavities can be seen, which seem to have been formed as a result of pressing the lumps on to the later. Jacobi (1897), who was probably one of the first to observe this, also mentions these missing bosses and the small cavities they have left. It seems certain that these lumps must have been

Fig. 3. Distribution maps of tegulae mammatae with large (below) and short (above) bosses with the region of main distribution highlighted. After Le Ny 1992; Bouet 1999; Bodribb 1987, and sites additionally added by the author.

Anejos de AEspA LXXVII

Site (province) (Smeermaas) Germania Inferior

THE END OF THE TEGULAE MAMMATAE?...

Type Rural: villa

Publication

125

Description (type)

Pauwels and Creemers 2006: 66-67 Ten loose flat bosses

Nijmegen - Hunerberg Military: Canabae legionis Germania Inferior

Zandstra and Polak 2014: 224

Six fragments of lateres with bosses

Leiden - Matilo Germania Inferior

Military: Castellum

Gazenbeek 2013

One loose flat boss

Leudal Germania Inferior

Rural: unkown

Pers. com. Twan Ernst

One fragment of a later with a flat boss

Heerlen - Thermen Germania Inferior

Urban: thermae of the vicus

collection Thermenmuseum Heerlen (unpublished)

Several fragments with bosses or indents (lateres)

Afferden - De Gening Germania Inferior

Rural: villa

Vermeulen-Bekkering 2006: 52-53

Several fragments with bosses (lateres)

Cuijk - Nutricia Germania Inferior

tegulae grave

Unpublished ROB excavation (1989)

at least two lateres with bosses

Urban (domus)

Tongeren Germania Inferior

– Kielenstraat

Clerbaut in press

– Vermeulenstraat

Unpublished

– Sint-Truiderstraat

Unpublished

Reuse

Several bossed tiles (lydion, bessales, pedales)

– Beukenbergweg

Clerbaut and Geerts 2014, 51-54

– Keverstraat

Pers. com. Peter Cosyns

Cassinomagus Gallia Aquitania

Civic thermae

Coutelas 2012: 187-190

Several tile fragments (lydion) with bosses

Gisacum (?) Gallia lugdunensis

Sanctuary

Coutelas 2012: 187-190

Several tile fragments (lydion) with bosses

Aregenua Gallia lugdunensis

Forum

Clerbaut et al. in press

Several tile fragments (uncertain) with bosses

Fig. 4. Tegulae mammatae with short bosses from the Roman North, not mentioned by Le Ny 1992, Bouet 1999 or Bodribb 1987. Composition: T. Clerbaut.

put in place shortly after the lateres were formed, when the clay was still soft enough, leaving the observed intends. However, it does not seem likely that it was the actual makers of the brick who put the bosses in  place, as an example from Tongeren-Kielenstraat shows. On this later the lump is placed over the signature made by the brick-maker. It is therefore more plausible to imagine that these lumps were placed by those responsible for the drying process. This would mean that the lumps played a role in the drying, and probably less in the firing of the lateres. A use for these lumps during drying suggests the stacking of unbaked material before firing. If the lateres were laid out flat on a drying field, the space would need to be fairly unlimited and stacking would be useless. However, when using a covered space (e.g. a shed) for drying the lateres, the space available would be limited, leaving no other choice than to increase efficiency by stacking them upright, spaced by little clay lumps. Different indications for the use of sheds on production si­tes

(e.g. Brandl and Federhofer 2010: 27, 70) are known from the Roman North, and their use for drying ceramic products is not unlikely giving the local weather conditions. Could the use of this drying method and local weather conditions explain the discrepancy between the Roman North and the Mediterranean when it comes to the distribution of this type? CONCLUSION: THE END OF THE TEGULA MAMMATA IN THE ROMAN NORTH? If this interpretation of the tegulae mammatae with short bosses is correct than this form is not a functional tegula mammata, but a later with the technical aids used during its production still adhering. It will be clear that the importance of this interpretation is relevant for the interpretation of this archaeological material and the f­ unction of any structures linked to it. After critical evaluation, the vast majority of the so called tegulae mammatae on site in the

126

Tim Clerbaut

­ oman North can be seen to be of this type. NevR ertheless, this does not in itself mean the complete end of the tegula mammata in the Roman North. Without doubt, it can be concluded that while ‘real’ tegulae mammatae are rather rare in the Roman North, they are certainly not non-existent. As for the known examples, they seem to be limited in time to the 1st century AD and part of a Mediterra­ nean influence brought to the north by the military. It is clear that many questions concerning the production and use of these forms are still unanswered. This paper only intends to give an introduction to this type of ceramic building materials and its related problems and opportunities. As an author, I hope this contribution will raise further awareness and interest for this specific type of ceramic building material and aids in future comparative studies. ACKNOWLEDGMENTS First of all I would like to warmly thank Janet­ DeLaine, Alejandra Albuerne and the entire organization of the 5th International Workshop on the Archaeology of Roman Construction for a splendid conference and presenting us with the chance to present and publish this contribution. Another word of thanks needs to go to several of our colleagues who kindly presented relevant information during or after the conference (Arnaud Coutelas, Peter Cosyns, Twan Ernst, Sandra Garside-Neville, Zoltán Havas, Lynne L ­ ancaster, Rien Polak and Peter Warry), or helped out with the interpretation of the Latin texts (Guido Cuyt). Special thanks go to Guus Gazenbeek for the vivid discussions on this topic, the joint conference presentation and his remarks on a preliminary version of this paper. REFERENCES Adam, J.-P. 2005: Roman building. Materials and tech­ niques. Routledge, London. Bouet, A. 1999: Les matériaux de construction en terre cuite dans les thermes de la Gaule Narbonnaise, Scripta Antiqua 1. Ausonius, Bordeaux. Brandl, U. and Federhofer, E. 2010: Ton + Technik, Römische Ziegel, Schriften des Limesmuseums Aalen­61. Theiss, Stuttgart. Brodribb, G. 1987: Roman Brick and Tile. Sutton, Gloucester.

Anejos de AEspA LXXVII

Brodribb, G. 1979: “Tegulae mammatae”, The Anti­ quaries Journal, 59, pp. 397-400. Clerbaut, T., Debeerst, R. and Jardel, K. in press: Etudes des terres cuites architecturales (TCA) du secteur du forum d’Aregenua (Vieux-la-romaine): ­bilan et premiers resultats, Contribution au bilan des fouilles 2014, Vieux. Clerbaut, T. and Geerts, R. C. A. 2014: “Nieuwe vondsten en inzichten met betrekking tot de ‘MHF‘stempels te Tongeren”, Signa, 3, pp. 51-54. Clerbaut, T. in prep: “Chapter 5. Ceramic building materials”, in Driesen, P., Bogaert, L. and Creemers, G. (eds.), Excavations Under the Museum, Atua­ tuca Series. Tongeren. Coutelas, A. 2012: “Les terres cuites architecturales: deux études de cas”, in Doulan, C., Laüt, L., Coutelas, A., Hourcade, D., Rocque, G. and Sicard, S. (eds.), Dossier Cassinomagus. L’agglomération et les thermes, résultats des recherches récentes (20032010) à Chassenon (Charente), Aquitania, 28, pp. 179-191. Cunliffe, B. W. 1971: Excavations at Fishbourne, II. The finds, Research Report 26. The Society of Antiquaries, London. Ferdière, A. 2012: “La production de terres cuites architecturales en Gaule et dans l’Occident romain, à la lumière de l’exemple de la Lyonnaise et des cités du nord-est de l’Aquitaine: un artisanat rural de carac­tère domanial?”, Revue archéologique du Cen­ tre de la France, 51, pp. 17-187. Gazenbeek, A. E. 2013: “Grofkeramiek, natuursteen en mortel”, in Brandenburgh, C. R. and de Bruin, J. (eds.), Met de voeten in het water. Archeologisch onderzoek aan de oostzijde van castellum Matilo te Leiden, pp. 138-196. Gemeente Leiden, Leiden. Gunther, R. T. 1913: Pausilypon. The Imperial Villa Near Naples. Hart, Oxford. Havas, Z. 2009: “Atilia Firma téglamu˝helyének termékei Pannoniában (Erzeugnisse der Ziegelwerkstatt der Atilia Firma in Pannonien)”, in Bíró, S. (ed.), Ex Officina... Studia in honorem Dénes Gabler, pp. 205-223. Mursella Régészeti Egyesület, Gyo˝r. Jacobi, L. 1897: Das Römerkastell Saalburg bei Hom­ burg vor der Höhe. Im Selbstverlage des Verfassers, Homburg vor der Höhe. Le Ny, F. 1992: La production des matériaux de construction de terre cuite en Gaule romaine. Thèse non publiée, Université de Rennes I (4 vol.). Lethaby, W. R. 1923: Londinium: architecture and the crafts. Blom, London. Pauwels, D. and Creemers, G. 2006: “Een Romeinse landelijke nederzetting te Smeermaas (Lanaken, prov. Limburg)”, Relicta, 2, pp. 49-118.

Anejos de AEspA LXXVII

THE END OF THE TEGULAE MAMMATAE?...

127

Vermeulen-Bekkering, A. M. 2006: Een Romeinse villa langs de Maas bij Afferden. Gemeente Bergen (Limburg), Rapportage Archeologische Monumentenzorg 116. Rijksdienst voor het Oudheidkundig Bodemonderzoek, Amersfoort. Wheeler, R. E. M. and Wheeler, T. V. 1936: “Verulamium: a Belgic and two Roman cities”, Reports of the Research Committee of the Society of Antiquar­ ies of London 11. The Society of Antiquaries, ­Oxford.

Winbolt, S. E. 1922: “Alfodean Roman Station”, Sus­ sex Archaeological Collections, 64, p. 81. Wright, F. S. 1939: “Report of bricks and tiles found at Highdown”, Sussex Archaeological Collections, 81. Zandstra, M. J. M. and Polak, M. 2014: “6. Keramisch Bouwmateriaal”, in Kloosterman, R. P. J., Polak, M. and Zandstra, M. J. M. (eds.), Opgravingen op het terrein van het voormalige Canisiuscollege in Ni­ jmegen, 1987-1997, Vondsten uit castra en canabae I, pp. 221-258, Auxiliaria 14. Auxilia, Nijmegen.

IV PRODUCTION, SUPPLY AND USE OF MUD-BRICK AND PISÉ DE TERRE

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

MUD BRICK AND PISÉ DE TERRE BETWEEN PUNIC AND ROMAN BEN RUSSELL*, ELIZABETH FENTRESS** * University of Edinburgh ** Independent scholar

ABSTRACT: This paper examines two building materials often overlooked in studies of Roman arch­itecture: mud brick and pisé de terre (or rammed earth), with the focus on the latter. In the first section we will examine how pisé was made and its relative benefits in comparison to mud brick. In the second section we will turn to the occurrence of these materials in North Africa, concentrating on new data from excavations at Utica. We then turn to the broader question of the spread of these construction techniques around the pre-Roman and Roman Mediterranean. Its predominately western distribution and the fact that pisé walls are found in third-century BC contexts at Kerkouane has suggested that it derived from North Africa and was diffused via Punic influence. However, as it is found in central Italy from the fourth century BC, we might suggest that its spread was a Roman phenomenon. KEYWORDS: Pisé de terre, Mud brick, Roman building techniques, Punic building techniques, Utica. RESUMEN: Este artículo examina dos materiales de construcción a menudo subestimados en los estudios de arquitectura romana: el adobe y el tapial de tierra (o tierra apisonada), con particular atención en este último. En la primera sección se examina como se preparaba el tapial y sus ventajas en comparación con el adobe. En la segunda sección se analiza la incidencia de estos materiales en el norte de África, concentrándose en los nuevos datos de las excavaciones en Utica. Sucesivamente, se trata la cuestión más amplia de la difusión de estas técnicas de construcción en el Mediterráneo en época pre-romana y romana. Su distribución principalmente occidental y el hecho de que las estructuras de tapial se documentan en contextos del siglo tercero a.C. en Kerkouane ha planteado la posibilidad que procede del Norte de África y se difunde a través de la influencia púnica. Sin embargo, visto que se encuentra en el centro de Italia desde el siglo iv a.C., podríamos sugerir que su propagación fue un fenómeno romano. PALABRAS CLAVE: tapial de tierra, adobe, técnicas de construcción romana, técnicas de construcción púnica, Utica.

Unbaked earth is still widely used for building today: in the mid-1990s it was estimated that 30% of the world’s population lived in houses built of this material, 50% of the population of developing countries, and well over half of all inhabitants of rural areas (Houben and Guillard 1994: 6). The most common of these unbaked earth techniques are mud brick and rammed earth, the latter of which is more usually known by its French

name, pisé de terre. Both of these techniques were employed in antiquity, though most archaeologists of this period will be more familiar with the former than the latter. In essence, pisé is simply earth that has been poured into formwork and compressed (fig. 1). Pliny the Elder’s brief description covers the main points: ‘... are there not in Africa and Spain walls made of earth that are called framed walls (terra parietes, quos appellant

132

Anejos de AEspA LXXVII

BEN RUSSELL, ELIZABETH FENTRESS

Fig. 1. A pisé wall under construction in northern Vietnam (Alfred Boc).

formaceos), because they are made by packing earth in a frame enclosed between two boards, one on each side, and so are stuffed rather than built’ (NH XXXV.48; other sources in de Chazelles 2010; Pesando 2011). The use of formwork is key and distinguishes pisé from other forms of earth construction, like cob or mud walling (bauge in French) that uses no such formwork or other supports, and wattle and daub (torchis in French, opus craticium in Latin), which is built up around a framework or against a wattle screen (de Chazelles 2010; Roux 2011). The French term pisé de terre is widely used in various languages for this construction technique, though rammed earth is also employed in English; the Arab term tabiya is the origin for the Spanish and Portuguese terms for this material, tapial and taipa respectively (Houben and Guillard 1994: 6-7; de Chazelles 2010: 312). The soils most suitable for pisé are gravelly sands with a small amount of clay in them. Indeed in his description of pisé walls, Varro describes them specifically as formed of ‘earth and gravel in molds (ex terra et lapillis compositis in

formis: R.R. I.14.4). Soils rich in clay shrink too much during drying to be reliable and are better suited to brick production; Williams-Ellis states that soils comprising much more than 25-30% clay should not be used for this mode of construction, while Rees recommends no more than 12.5% (1947: 45-47; cf. also Rees 1819, s.v. pisé; Houben and Guillard 1994: 109). Preparation of the soil is simple: following excavation, which is usually carried out on or close to the building site, the soil just needs to be broken up, aerated, and large stones removed. Additives, including lime, plaster, cement and bitumen, can be included to help stabilize the soil but are usually avoided (Houben and Guillard 1994: 82-83, 98-99). Straw can be used to counteract shrinkage if the soil is too clayey, but vegetal inclusions leave behind voids in the wall that can lead to damp or frost damage (Glick 1976: 150-151; Ruffin 2000: 50). Some  practitioners recommend adding coarse aggregate: Rael, in fact, gives a recipe that includes 15% clay, 23% aggregate, 30% sand and 32% silt (2009: 17). Re-used material from older pisé or even mud brick buildings can also be

Anejos de AEspA LXXVII

MUD BRICK AND PISÉ DE TERRE BETWEEN PUNIC AND ROMAN

e­ mployed, providing it is broken down ­sufficiently: in tenth-century Spain, as in walls recently excav­ ated in Almeria (fig. 2), there is sometimes more aggregate than earth, although Roman walls seem to contain far less aggregate. Since pisé is prone to damage from damp, most pisé walls are built on foundations of rubble, stone or brick, which usually protrude above ground level at least 10-20 cm. Since pisé requires sandy soil it is not surprising that this technique is not widely used in regions where the soil is predominately clayey. However, at sites where a range of soil types could have been exploited, pisé had certain benefits over other earth building techniques. It was relatively  cheap, quick, and easy; Rees describes it as fund­amentally an ‘easy, economic and convenient method’ (Rees 1819: s.v. pisé). Pisé has several clear advantages over mud brick, in particular: first, the excavation and preparation of the soil is quicker and easier; second, while mud bricks need to be shaped and left to dry, the soil used in rammed earth construction is compressed during the building process. As Herbemont puts it, in a letter preserved in Ruffin’s volume, ‘a wall made of [pisé] is made certainly at a less expense than would be required to dig clay, temper it, and mould it into bricks for the same wall’ (Ruffin 2000: 50). Labour figures provided by later sources bear this out. Rees and Williams-Ellis state that two to three builders can erect between 1.5 and 2.3 m3 of rammed earth walling in a day, provided that the soil is to hand and the conditions are satisfactory; this equates to 13-16 man-hours/m3 (Rees 1819: s.v. pisé; Williams-Ellis 1947: 42). Houben and Guillard provide a slightly higher total of 16-20 man-hours/m3, while Minke gives figures of 20-30 man-hours/m3 but this includes preparation and transport of the material (Houben and Guillard 1994: 200; Minke 2006: 60). However, we should probably expect that specialist earth builders in antiquity could have worked faster than this. Indeed, Williams-Ellis notes that a team of experienced pisé builders working in the best conditions can complete double the normal output (1947: 42). In comparison, Williams-Ellis notes that it can take a builder a day (i.e. 10-12 hours) to mould the number of mud bricks required for a similar volume of walling – 1.26 m3 (1947: 59). Houben and Guillard provide slightly higher output totals of 500-2500 mud bricks per day for 4-5 workers, while Minke suggests one worker can produce 300 average-sized mud bricks

133

Fig. 2. Wall from a 9th century AD building, Almería, Spain. The aggregates include pottery, stones, and plaster (EF).

in a day (Houben and Guillard 1994: 212-219; Minke 2006: 62-63). These totals are still just for the preparation of the material and take no account of the time required for drying, which adds up to usually between one and two weeks, or act­ ual construction. Another major advantage of pisé is its strength.1 Although pisé is laid in courses, these are often over half a metre high, and the ramming process ensures that little air or moisture is left in the wall. A covering of plaster, which is common, can make pisé walls even more resistant to erosion. As Pliny notes: ‘...do not [these rammed earth walls] last for ages, undamaged by rain, wind and fire, and stronger than any caemento?’; he also mentions pisé watch-towers, put up by Hannibal, which were still standing in Spain in his day (NH XXXV.48). Rees refers to pisé structures well over one hundred and fifty years old and Herbemont to even older buildings (Rees 1819: s.v. pisé; for Herbemont’s note, see Ruffin 2000: 49): indeed the walls of Zuwīla in the Libyan ­Sahara have been dated by radiocarbon to the 9th 1   The strength of rammed earth is due at least in part to matric suction, or apparent cohesion, which increases as they dry (Jaquin et. al  2009). For a micromorphological examination of the Islamic walls at Rirha Cammas and Roux 2016.

134

BEN RUSSELL, ELIZABETH FENTRESS

century (Mattingly et al. 2015). It is certainly possible to build two or three storey structures in pisé, provided the walls are of sufficient thickness. It is often stated that pisé walls should not be thinner than 0.40 m but this is not a rule: ­Williams-Ellis suggests that exterior walls of a single storey building should be 0.35 m wide, those of a two storey building 0.45-0.60 m, while 0.23-0.30 m is sufficient for internal dividing walls (1947: 17, 24-25, 40; on the 0.40 m figure, see Arcelin and Buchsenschutz 1985: 18). Getting the right formwork for the job is a ­vital component of pisé construction. However, as long as this formwork is strong enough to resist the ramming process and portable enough to be moved easily it need not be overly complicated, and Houben and Guillard catalogue numerous types of formwork still used worldwide (1994: 204-209). Most formwork consists of two upright panels held together by horizontal struts, usually of adjustable length, or sometimes ropes. These boards are constructed of long wooden planks of wood laid on their side and fastened together by regular sets of vertical supports. End panels can be added to this formwork as required. Rees re­ commends that the formwork should be around 3 m long and 0.8 m high but much shorter examples are also attested (Rees 1819: s.v. pisé; also Glick 1976). Formwork is usually supported by the horizontal struts between the panels, which pierce the wall and indeed leave putlog holes in it that can later be used for scaffolding (fig. 3); this is often described as either ‘crawling’ or ‘climbing’ formwork and is moved up and along the wall as it is built being. However, it is also possible to attach formwork to regularly spaced upright posts in the wall, a sort of composite form of frame and pisé construction, which does not leave putlog holes behind (Houben and Guillard 1994: 204). These posts can either be the full thickness of the wall or they can be much thinner and placed in pairs on either side of the wall, acting effectively as internal vertical struts for the formwork; this latter a solution which leaves behind characteristic vertical grooves in the wall once the posts are removed (de Chazelles 1990: 106-107, fig. 15). Needless to  say, the formwork used in pisé construction should be adjustable enough to be employed on thinner or thicker walls. This formwork is the expensive part of pisé construction. Glick has noted that in fourteenth-century Spanish sources a pair of moulds appears to cost the same as about 3.000 bricks (Glick 1976: 149-150, drawing on

Anejos de AEspA LXXVII

Fig. 3. ‘Climbing’ formwork (after Minke 2006: 50, fig. 5.2).

Hamilton 1936: 214). However, this formwork could be continually re-used, and indeed in Spain it was common to rent formwork for specific jobs rather than buy it new. CASE STUDY: EARTH CONSTRUCTION AT UTICA Unbaked building materials were certainly used widely in antiquity but distinguishing between different varieties of them archaeologically can be  problematic, especially at sites that have been ­highly disturbed by later activity. At the site of Utica, in Tunisia, however, excavations undertaken by the Tunisian-British project since 2010 have been able to identify a number of structures of various dates in both mud brick and pisé, allowing some analysis of when, and in what kinds of buildings, these different materials were employed. Mud brick was widely used in the Punic period at Utica. The earliest wall in mud brick recovered so far dates to the eighth century BC, and is found at the northern tip of the site (Imed Ben Jerbania, pers. comm.). In our own excavations mud brick walls, as yet undated, are found under the late first-century AD Roman basilica (fig. 4), while others lie beneath Roman buildings in stone, themselves built directly on top of the earlier ­Punic walls. Lézine observed that in the third century BC almost all domestic walling was in mud brick (1968: 152; 1971: 90). There is little doubt, though, that the use of the technique continued

Anejos de AEspA LXXVII

MUD BRICK AND PISÉ DE TERRE BETWEEN PUNIC AND ROMAN

Fig. 4. Pre-Roman mud brick wall beneath the probable Roman basilica at Utica (EF/Tunisian-British Utica Project).

into the Roman period: Vitruvius mentions legisl­ ation in which the magistrates of Utica required mud bricks to be dried for five years, while he himself believed that two years was a minimum (De Arch. II.3.2). These drying times are e­ xtremely long and it seems likely that Vitruvius meant months and not years here, but they do point to

135

Roman use of mud brick and also hint at some concerns with the quality of the local materials for this mode of construction. This literary evidence is supported by some archaeological finds: mud bricks can be found in the walls of the ­Maison de la Chasse, which dates from the first century BC (Ville 1961: 25 and fig. 7). However, the first Roman-period walls we recovered, in the southwest corner of the town, are of pisé on a foundation of reused ashlars, perhaps deriving from earlier walls. These appear to date from the first half of the first century BC. In the first century AD, a substantial and very elegant Roman house, known as the Maison du Grand Oecus, was built near the forum. Although its outer walls were of stone, the internal walls were of pisé on a stone socle: fig. 5 shows the collapse of a wall into a small vestibule, with its plaster still adhering to it, retaining its shape as it fell into the empty space beneath. The wall was certainly two storeys high, as we can see from the room behind it (fig. 6), where a mosaic can be seen to have collapsed from the upper storey onto a sea of melted pisé – note the pisé piers on the sides. The material in all these cases was very

Fig. 5. Utica, Maison du Grand Oecus. Fallen pisé wall with the plaster still adhering to it (EF/Tunisian-British Utica Project).

136

BEN RUSSELL, ELIZABETH FENTRESS

Anejos de AEspA LXXVII

Fig. 6. Utica, Maison du Grand Oecus. Mosaic from an upper floor collapsed onto melted pisé, pisé piers (EF/Tunisian-British Utica Project).

clean, containing nothing but a few small stones. Clay was not visibly a component of the walling, so this was not cob or mud walling. The impression gained from Utica, then, is that in post-Punic building there pisé gradually replaced mud brick, at least in very high-end construction, although for some time (up until the 1st century AD?) the two techniques appear to have co-existed. Evidence from Thysdrus confirms the pattern observed at Utica. In the early Roman period the majority of the houses used mud brick constructed on high socles of cut stone or rubble, in some case with large uprights built into them, as is typical in opus africanum (Slim 1985; Fantar 1984: 311). However pisé was used for the Maison de Lucius Verus, one of the largest and most sumptuous of the city, dating to the 2nd century AD but with substantial fourth- and fifth-century AD alter­ations (Slim 1985: 38). Pisé walls have also been documented in the second- and early thirdcentury AD houses at Acholla, notably the Maison de Neptune, where local beach sand mixed with shells was employed (Slim 1985: 38). At Volubilis we find a similar picture, with the same

overlap of techniques as observed at Utica. Mud brick was certainly used for the pre-Roman defensive wall, and for various internal walls in buildings in the Roman town (Euzennat 1989: 206 and fig. 130). However, Lenoir also identified a wall in pisé, in the Maison des Fauves, with fine painted plaster (Lenoir 1985), and the recent UCL/INSAP excavations, directed by Elizabeth Fentress with Hassan Limane, identified mud brick and pisé collapse in another building, dating to the second to third centuries AD. At Thamusida, a combination of materials can also be found in the first-century BC wall circuit, which was constructed with stone foundations, a socle in mud brick, and elevation in either pisé or possibly cob walling (since the earth used was very clayey) (Akerraz et al. 2009: 162). THE USE AND DIFFUSION OF PISÉ In the western Mediterranean, pisé construction is usually associated with Islamic practice in North Africa, diffused, as tapial, to Spain and

Anejos de AEspA LXXVII

MUD BRICK AND PISÉ DE TERRE BETWEEN PUNIC AND ROMAN

eventually France, but its first ­occurrence is thought to be Punic (de Chazelles 2010: 312-313; de Chazelles and Guyonnet 2007). In North Africa itself, however, the evidence is far from clear-cut. The only excavated pisé walls of Punic date to have been carefully documented come from third-century BC contexts at Kerkouane (Fantar 1984: 309-314). The clearest of these were found in a house at the corner of the so-called Rue du Temple and Rue des Artisans. In this case, the stone stylobates topped by the pisé have deep vertical grooves along their surface (4-6 cm deep and 12 cm wide), and were evidently constructed with the formwork attached to fixed upright posts running along each face of the wall at regular intervals (Fantar 1984: 313 and pl. 11-12; de Chazelles 1990: 106, fig. 15). The soil used for these walls contained rubble, as well as degraded mud brick, which the excavator proposed came from an earlier structure. Elsewhere on the site, pisé and mud brick are used together in walling for domestic structures: as Fantar notes, and as the evidence outlined above, from Utica and elsewhere, suggests, these techniques are not mutually exclusive (Fantar 1984: 311, 313-314). The city walls at Kerkouane also employed rammed earth but in this case the material was used between faces built out of cut stone, a technique that is not pisé, but resembles more closely the use of cob or mud walling at Lattes, in southern France, where a similar building method is found (Fantar 1984: 311; for Lattes, Roux 2003; 2011). Earth construction was certainly also used at Carthage in the Punic period, though its exact composition is unclear. Carton mentions walls of earth in the report on his excavations of the so-called Punic sanctuary, while Renault found earth walls of ­Punic date on rubble foundations made of earth mixed with lime (Carton 1929: 4; Renault 1913: 22). On the Byrsa, Ferron and Pinard mention walls made of earth rammed between roughly fired bricks (Ferron and Pinard 1955: 54). In none of these cases, however, is it clear that the technique employed was actually pisé using formwork, as opposed to other types of earth construction. It might be noted that at Carthage mud brick continued to be used well into the fifth century AD (Gallagher 1985). Outside of North Africa, the ancient sources, notably Pliny (XXXV.48) and the sixth-century AD writer Isidore of Seville (Etymologies XV.9), also connect pisé to Spain, and indeed several attempts have been made to identify early pisé use

137

in Spain, particularly in areas of Carthaginian influence. It has been suggested, for instance, that the earth walls at Fonteta/Guardamar, as early as the seventh century BC, were built in pisé, though in this case the actual technique employed a­ ppears to be closer to cob rather than rammed earth proper (de Chazelles 1990: 117; 2003; contra ­Belarte 2001, 33-34: personal observation by Fentress supports de Chazelles). De Chazelles is also skeptical about the identification of pisé in thirdand second-century BC contexts elsewhere in Spain; in many cases we have just the socles of the walls and in some cases the walls themselves appear to be made of mud brick (de Chazelles 1990: 117). Pisé, however, was certainly being used by the early to mid first century BC in Spain: Varro mentions it in his Rerum Rusticarum and since he had served in Spain in 76 and 49 BC he knew the region well (R.R. I.14.4). In practice, though, the earliest walling in Spain that can be clearly identified as genuine pisé is present in the Roman houses at Ampurias in the late first century BC (de Chazelles 1990). Here de Chazelles has shown that these pisé walls were made using moveable formwork, a technique different from that seen at Kerkouane; this is the earliest archaeologically attested evidence in Spain for this type of construction (de Chazelles 1990: 107-108). In southern France, similar claims have been made for early pisé construction. Large-scale earth construction was taking place this region from the third century BC, at the oppidum of Notre Dame de Pitié at Marignane, for example, but whether this was pisé is much less clear (Arcelin and Buchsenschutz (1985: 23) argue that it was, whereas de Chazelles (1990: 117) is much less certain). In second-century BC contexts at Martigues and at Mouriès pisé has also been claimed but again the evidence is not conclusive and on the basis of the composition of the soil de Chazelles has suggested that in both cases cob rather than pisé was used (de Chazelles 1990: 113-17; on Martigues, Nin 1988: 64). At Lattes there is no doubt that all earth walls were built in cob, using various techniques, and sometimes combined with mud bricks (Roux 2003: 2010). The case for a Punic origin for the technique seems to collapse entirely if we take Sicily and Sardinia into consideration. There, recent work has demonstrated an overwhelming use of mud brick, in Greek as much as in Punic sites (for Sicily, see Germanà and Panvini 2008; for Sardinia, pers. comm. Peter van Dommelen). Indeed, de

138

BEN RUSSELL, ELIZABETH FENTRESS

Chazelles has argued that the use of mud brick was directly tied to Phoenician settlement in Sardinia, Spain and North Africa (de Chazelles 2011). The case of Heraclea Minoa, in Sicily, seems to prove the rule: mud brick is used in many of the early houses, while pisé is found only in a patrician house dated after the Roman occupation in 132 BC (Di Natale et al. 2008). To date, discussion of the development and diffusion of pisé has focused overwhelmingly on the evidence from North Africa, Spain and France, all of which is extremely patchy. However, pisé was certainly also used in Italy. Varro mentions its use in the south of the peninsula, in the area around Tarentum (R.R. I.14.4). Archaeological evidence is mixed, however, and here, as in Spain, it would be useful to look more carefully at the evidence to determine whether we are dealing with cob or pisé. There are a number of early instances of earth structures in Italy where the use of pisé has been suggested. In some of these cases, the earth is compacted between regularly spaced posts, to which formwork could have been attached – a technique described above. This technique is very early indeed, appearing in the ninth century BC at Fidenae, where we find posts at regular intervals in the wall (Bietti-Sestieri and Di Santis 2006). Other mud walls are found in the eighth century BC at Torre di Satriano, in a building with a curved end made of mud brick (Carollo 2009), at Rome, in the building known as the Domus Regia on the slopes of the Palatine (fig. 7: Filippi 2004: 107, 115), and in the sixth century BC in the palace building at Murlo (Nielsen and Philips 1985: 60). All of these structures are rect­ angular, in contrast to the prevailing fashion for oval huts, and the orthogonal walls may have had something to do with the new building technique. There is an element of uncertainty about this arch­itecture: the roof was apparently still supported on posts, rather than on the earth walling itself. We may perhaps see this as the earliest, experimental phase of the use of pisé walling, which, on this reading, would seem to have developed in central Italy. We cannot be sure in any of these cases that we are not dealing with cob, however – although Fidenae might have a good case, because of the regular spacing of its posts, the walls at the domus regia and Torre di Satriano are narrow and slightly curvy, and photographs suggest that the former was made of clay. Furthermore, in spite of its possible appearance at Murlo, pisé does not seem to have been widely used in

Anejos de AEspA LXXVII

Fig. 7. Domus Regia. Walls of phase 1 (Dunia Filippi).

Etruria, as Marzabotto, Arezzo and Roselle all have shown abundant evidence for mud bricks (Staccioli 1967). Tantalizing evidence for a move towards pisé use is provided by the villa of the Auditorium, in the northern suburbs of Rome. Here, in the early fifth century, the stone socles of the walls were surrounded by a deposit of thick clay, interpreted by the excavators as the remains of the superstructure (D’Alessio 2006: 77). Two centuries ­later, in the very early third century, the same sort of deposit is described as clean, sandy clay (Ricci 2006: 193). Could this represent a shift from clayey cob towards pisé? Certainly by the late fourth century BC the technique had become more common: pisé walls have been found in Arpi, in Daunia, in a fourth-century house with a pebble mos­ aic (Mazzei 1996: 348), and in third-century houses in Pompeii, where they are earlier than the first attestations of opus africanum (Pesando 2007). Fregellae, too, shows pisé in the walls of a third-century atrium house (Pesando 2011). In the first century BC, it is used for internal walls at Cosa (fig. 8: Fentress 2003: 21) and at Settefines-

Anejos de AEspA LXXVII

MUD BRICK AND PISÉ DE TERRE BETWEEN PUNIC AND ROMAN

139

Fig. 8. Cosa, tablinum wall of the House of Diana (Adam Rabinowitz).

tre (Regoli 1985: 64). In both cases only a few centimetres remained above the stone socles, but the preparation of the fallen plaster showed the imprint of the scoring of the mud walls prior to the application of the plaster – note that there was no trace of the regular imprint that would have been created by mud brick. The assertion that pisé was the standard Roman building technique, particularly in rural contexts, cannot be supported by much excavation evidence, because it has generally dissolved into the clean layer of earth that covers them – a layer which is notably lacking in the clay lumps characteristic of the disintegration of mud brick. However, if the stone socles, regularly preserved to around 50 cm, had supported walls in stone rather than in pisé, the ruins of ­Roman buildings would resemble far more closely the massive piles of rubble left in the collapse of medieval buildings. Although the early Italian evidence is patchy, and much of it still needs to be tested, it is clear that the idea of pisé as a distinctly Punic building technique should be called into question. In fact, the diffusion of pisé appears to be a more obviously Italian phenomenon, particularly if we trace the use of this material between the first century BC and the second century AD. At Delos, destroyed in 86 BC, it is significantly found in a building owned by an individual with a Roman name, Quintus Tullius (Zarmakoupi 2015: ­10-11).2 This seems to imply the presence of an Italian construction team on the site, hardly impossible on an island on which Italian trade was so dominant. Note too, that like the House of the Grande 2   Mantha Zarmakoupi pers. comm. An attempt to locate other such houses in Tang 2005 failed because of the author’s reluctance to fill in the ‘building techniques’ field in her catalogue.

Fig. 9. Verulamium. Socle of wall and fallen wall plaster in Insula XXI, Building 2 (after Frere 1983: pl. 22b).

Oecus, in Utica, the house of Quintus Tullius had an upper floor: it is possible that pisé was used here because of its greater strength compared to rubble masonry. In Africa, we have noted the use of pisé in high-status buildings of the Roman ­period, but it is also found in the legionary settlement of Lambaesis, where there was likely to be an Italian element in the builders (Claire-Anne de Chazelles, pers. comm.). In France, pisé is much rarer than wattle and daub but is used at a number of inland sites in the first and into the second centuries AD: at Cravant, Bram, Cavaillon, and Orléans, for example (de Chazelles and Guyonnet 2007: 109; Coulon and Joly 1985: 93-94). In Britain, clear use of pisé was identified by Frere in Building XXI.2 at Verulamium, an elite house datable to the late second century AD (Frere 1983: 237). Not only are the putlog holes visible along the top of the stone socles in this house, but the back of the fallen wall plaster preserves the chevron pattern made on the surface of the pisé as keying for the plaster; the impression left in the plaster shows no signs of mud bricks (fig. 9). Pisé is also attested at Canterbury in the late first and second centuries AD (Frere and Stow 1983: 72),

140

BEN RUSSELL, ELIZABETH FENTRESS

while at London, it appears in high-status houses of Flavian date (Perring 1985: 154; Perring 1991). Meates also identified possible pisé walls in the residential block of the late first-century AD villa at Farningham in Kent and in a late s­ econd-century context in the temple associated with the nearby Lullingstone villa (1974: 5; 1979: 61). This material also appears to have been employed in military contexts: what appear to be pisé walls were used for the apodyterium of the extra-mural baths at Castell Collen fort in Radnorshire, where holes left by the formwork could be seen (Alcock 1956; Wacher 1985: 143), while at Colchester it appears in the early barracks buildings (Crummy 1977: 73). In general, in Britain as in France, the most widespread earth construction technique was wattle and daub, a common pre-Roman building technique, but pisé and even mud brick were used quite extensively in the South East and  were evidently introduced during by the Romans (Wacher 1985). PISÉ AND OPUS CAEMENTICIUM Pisé is very much a vernacular construction technique, requiring particular soil types and clim­atic conditions. However, particularly in its use of formwork it also resembles another, more famous building technique, Roman opus caementicium. While opus caementicium is usually faced in stone or brick, and is composed of a high proportion of aggregate, the method of building up concrete foundations between shuttering closely resembles pisé construction. This is not to argue that these techniques are identical. Pisé walls tend to have stone foundations, with the formwork employed to build the elevations, whereas shuttering is used in foundations of structures made of opus caementicium, with the brick or stone faces of the elevation effectively acting as formwork. How­ ever, some broad similarities remain and these have been stressed in earlier scholarship (on this, Golebiowski 2009: 145-151). Pisé was mentioned as a possible precursor to concrete by Delbrück, even if he regards it as a Phoenician building technique, introduced to the western Mediterranean by the Carthaginians in the third century BC (1912: 86-87). Even earlier, Ford had described tapial, a later iteration of pisé, as ‘a sort of African or Phoenician concrete’ (Ford 1845: 238; Ford 1837: 537-538). Blake also notes the importance of pisé as a mode of construction employing formwork that clearly predates concrete,

Anejos de AEspA LXXVII

though she is sceptical of the diffusionist model proposed by Ford and Delbrück: ‘this is an obvious enough procedure to have developed quite independently.’ (Blake 1947: 325). Blake is right, at least to a certain degree – pisé did develop quite independently elsewhere: it is well documented in China, for instance, after the Three Kingdom ­period (AD 221-581), though a form of rammed earth was certainly also used earlier in the region (Houben and Guillard 1994: 13; also Jaquin 2008: 384-387). However, around the Mediterranean only pisé and opus caementicium make extensive use of formwork, and it is interesting to note that the development of modern béton (concrete) in France was directly connected to the ‘rediscovery’ of, and then experimentation with, pisé construction, notably by Cointeraux in the mid to late eighteenth century (Rael 2009: 10). As Collins has put it, ‘the importance of pisé in the development of [modern] concrete construction lay not so much in the material used as the technique employed’ (2004: 21; see also Fentress 2003: 21). One can wonder if the attribution, by many of the early scholars of Roman architecture, of the pisé technique to Phoenicians or ‘Africans’ was not an assertion that it was ‘other’ – foreign and, in particular, primitive. As such, some of the similarities between it and the very Roman opus caementicium was less obvious than it might have been if its possible Italian origins had been taken into account. One reason for this, of course, is that its use seems to have survived much better in North Africa than it did in Italy itself: like Buddhism, it died out in the land of its origin, replaced by the omnipresent mortared rubble and caementicium techniques. Its rich afterlife was left in the hands of the North Africans. Our assertion that it was, at least initially, a building technique more common in central Italy than elsewhere in the western Mediterranean, leads to interesting possibilities. We have seen it used in high-status buildings from Britain to North Africa. It could be argued that, because building techniques are perhaps the most conservative of practices, they have the most to tell us about the origins of the people who use them. Gaggiotti has argued that the use of pavimentum punicum spread to central Italy through the use of Carthaginian captives taken during the second Punic war (Gaggiotti 1987; see also Fentress 2013: 174). A similar argument, with different captives, might be used to explain the use of pisé in Kerkouane. However, this is far less likely in the cases

Anejos de AEspA LXXVII

MUD BRICK AND PISÉ DE TERRE BETWEEN PUNIC AND ROMAN

of Delos, Heraclea Minoa or the North African buildings built under Roman rule. Here we must imagine that immigrants included building teams whose work was preferred by the new settlers. The long co-existence of mud brick alongside pisé in cities such as Carthage works as a proxy for the mixed community of the Roman city. However, it is also worth asking why pisé, rather than mud brick, became the dominant technique in the med­ ieval Maghreb, spreading, in southern Morocco and Mauretania, far outside areas of Roman influence. Perhaps the answer simply lies in its greater efficiency? ACKNOWLEDGEMENTS This paper owes much to the help and suggestions of Claire-Anne de Chazelles and Fabrizio Pesando, whose work on pisé walls in North Africa and Spain on the one hand and Italy on the other has done much to bring the technique to our attention. Martin Millett provided useful cit­ations for British instances of pisé. Marta Lorenzon and Martin Michette offered a number of helpful suggestions regarding the practicalities of pisé construction. REFERENCES Akerraz, A., El Khayari, A., and Papi, E. 2009: “L’habitat maurétano-punique de Sidi Ali ben Ahmed – Thamusida (Maroc)”, in Helas, S. and Marzoli, D. (eds.), Phönizisches und punisches Städtewesen, pp. 147-170. Von Zabern, Mainz. Alcock, L. 1956: “Castell Cohen excavations 1956”, Transactions of the Radnorshire Society, 26, pp. 10-21. Arcelin, P. and Buchsenschutz, O. 1985: “Les données de la protohistoire”, in Lasfargues, J. (ed.), ­Architectures de terre et de bois. L’habitat privé des provinces occidentales du monde romain. Antécédents et prolongements: Protohistoire, Moyen Age et quelques expériences contemporaines. Actes du 2e congrès archéologique de Gaule méridionale (Lyon 2-6 Novembre 1983), pp. 15-28, Documents d’Archéologie Française 2. Maison des Sciences de l’Homme, Paris. Bietti Sestieri, A. M. and De Santis, A. 2001: “L’edificio della I età di ferro di Fidene (Roma): posizione nell’abitato, tecnica costruttiva, funzionalità in base alla distribuzione spaziale dei materiali e degli arredi”, in Rasmus Brandt, J. and Karlsonn, L.

141

(eds.), From Huts to Houses. Transformations of ­Ancient Societies, pp. 211-221. Astrom, Stockholm. Blake, M. 1947: Ancient Roman Construction in Italy from the Prehistoric Period to Augustus. A Chronological Study Based in Part upon the Material ­Accumulated by the Late Dr. Esther Boise van Deman. Carnegie Institution, Washington D. C. Cammas, C. et Roux, J.-C. 2016: “Étude des matériaux de construction en terre crue des sites antiques de Rirha (Maroc)”, Archéopages, 42, 68. Carollo, G. 2009: “La residenza ad abside: la struttura, l’organizzazione degli spazi, le fasi”, in Osanna, M., Colangelo, L. and Carollo, G. (eds.), Lo spazio del potere. La residenza ad abside, l’anaktoron, l’episcopio a Torre di Satriano, pp. 19-32. Osanna, ­Venosa. Carton, L. 1929: Sanctuaire punique découverte à Carthage. Geuthner, Paris. Chazelles, C.-A. de 1990: “Les constructions en terre crue d’Empúrias à l’époque romaine”, Cypsela, 8, pp. 101-118. Chazelles, C.-A. de 1997: Les maisons en terre de la Gaule méridionale. Mergoil, Montagnac. Chazelles C.-A. de and Guyonnet, F. 2007: “La construction en pisé du Languedoc-Roussilon et de la Provence, du Moyen-Âge à l’époque moderne (XIIIe-XIXe s.)”, in Guillard, H., Chazelles, C.-A. de and Klein, A. (eds.), Échanges transdisciplinaires sur les constructions en terre crue, 2: la terre massive, May 2005, Villefontaine (Isère), France, pp. 109139. Éditions de l’Espérou, Montpellier. Chazelles, C.-A. de 2010: “Quelques pistes de recherches sur las construction en terre crue et l’emploi des terres cuites architecturales pendant l’Âge du fer dans le basin occidental de la Méditerranée”, in Tréziny, H. (ed.), Grecs et indigènes de la Catalogne à la Mer Noire, pp. 309-318, Bibliothèque d’archéologie Méditerranéenne et Africaine 3. Centre Camille Jullien, Aix-en-Provence. Chazelles, C.-A. de 2011: “La construction en brique crue moulée dans les pays de la Méditerranée, du Néolithique à l’époque romaine. Réflexions sur la question du oulage de la terre”, in Chazelles, C.-A. de, Klein, A. and Pousthomis, N. (eds.), Les cultures constructives de la brique crue. Échanges transdisciplinaires sur les constructions en terre crue, 3, pp. 153-165. Éditions de l’Espérou, Toulouse. Collins, P. 2004: Concrete: The Vision of a New Architecture. McGill-Queen’s University Press, Montreal. Coulon, G. and Joly, D. 1985: “Le centre”, in Lasfargues, J. (ed.), Architectures de terre et de bois. L’habitat privé des provinces occidentales du monde romain. Antécédents et prolongements: Protohistoire, Moyen Age et quelques expériences contemporaines.

142

BEN RUSSELL, ELIZABETH FENTRESS

Actes du 2e congrès archéologique de Gaule méridionale (Lyon 2-6 Novembre 1983), pp. 93-102, Documents d’Archéologie Française 2. Maison des Sciences de l’Homme, Paris. Crummy, P. 1977: “Colchester: the Roman fortress and the development of the colonia”, Britannia, 8, pp. 65-105. D’Alessio, M. T. 2006: “La villa e il quartiere servile, fasi 1-3 (500-350/300)”, in Carandini, A., D’Alessio, M. T. and Di Giuseppe, H. (eds.), La fattoria e la villa dell’Auditorium, pp. 103-158. “L’Erma” di Bretschneider, Roma. Delbrück, R. 1912: Hellenistische Bauten in Latium, II. Trübner, Strasbourg. Di Natale, E., Lanzarone, F. and Canicatti, F. E. 2008: “Eraclea Minoa: architetture in terra cruda”, in Germanà, M.-L. and Panvini, R. (eds.), La terra cruda nelle costruzioni. Dalle testimonianze Archeologiche all’architettura sostenibile. Atti della Giornata di Studi (Caltanissetta, Museo Archeologico Contrada Santo Spirito, 29 giugno 2007), pp. 99-108. Nuova Ipsa, Caltanissetta. Euzennat, M. 1989: Le limes de Tingitane. La frontière méridionale. CNRS, Paris. Fantar, M. 1984: Kerkouane. Cité punique du Cap Bon (Tunisie), I. Institut National d’Archéologie et d’Art, Tunis. Fentress, E. 2003: Cosa V. An Intermittent Town. Excavations 1991-1997. The University of Michigan Press, Ann Arbor, Mich. Fentress, E. 2013: “Strangers in the city. Élite communication in the Hellenistic central Mediterranean”, in Prag, J. and Quinn, J. (eds.), The Hellenistic West, pp. 157-179. Cambridge University Press, ­Cambridge. Ferron, J. and Pinard, M. 1955: “Les fouilles de Byrsa, 1953-1954”, Cahiers de Byrsa, 5, pp. 31-81. Filippi, D. 2004: “La Domus Regia”, Workshop di Archaeologia Classica, 1, pp. 101-121. Ford, R. 1845: Hand-Book for Travellers in Spain. John Murray, London. Frere, S. 1983: Verulamium Excavations, II. The Society of Antiquaries of London, London. Frere, S. and Stow, S. 1983: Excavations in the St George’s Street and Burgate Street Areas, The Arch­ aeology of Canterbury 7. Kent Archaeological Society, Maidstone. Gaggiotti, M. 1987: “Pavimenta Punica marmore Numidico constrata”, in Mastino, A. (ed.), L’Africa Romana. Atti del V Convegno di studio (11-13 dicembre 1987), pp. 215-221. Università di Sassari, Dipartimento di Storia, Sassari. Gallagher, J. B. 1985: “Mudbrick at Roman Carthage”, in Senay, P. (ed.), Carthage VIII. Actes du

Anejos de AEspA LXXVII

Congrès international sur Carthage (Trois Rivière, 10-13 octobre 1984), pp. 64-112, Cahier des Études Anciennes (deuxième Partie) 18. Université du Québec, Québec. Germanà. M.-L. and Panvini, R. (eds.) 2008: La terra cruda nelle costruzioni. Dalle testimonianze Archeologiche all’architettura sostenibile. Atti della Giornata di Studi (Caltanissetta, Museo Archeologico Contrada Santo Spirito, 29 giugno 2007). Nuova Ipsa, Caltanissetta. Glick, T. F. 1976: “Cob walls revisited: the diffusion of tabby construction in the western Mediterranean world”, in Hall, B. and West, D. (eds.), On PreModern Technology and Science: Studies in Honor of Lynn White, Jr., pp. 147-159. Undena, Malibu. Gobelowski, J. 2009: Rammed Earth Architecture’s Journey to the High Hills of the Santee and its Role as an Early Concrete, MSc Thesis, Clemson University & The College of Charleston. Hamilton, E. J. 1936: Money, Prices, and Wages in Val­ encia, Aragón and Navarre, 1351-1500. Harvard University Press, Cambridge, Mass. Houben, H. and Guillard, H. 1994: Earth Construction: A Comprehensive Guide. Intermediate Technology, London. Jaquin, P. A. 2008: Analysis of Historic Rammed Earth Construction, PhD Thesis, Durham University. Jaquin, P. A., Augarde, C. E., Gallipoli, D. and Toll, D. G. 2009: “The strength of unstabilised rammed earth materials”, Géotechnique, 59, 5, 487490. Lenoir, M. 1987: “Le Maroc”, in Lasfargues, J. (ed.), Architectures de terre et de bois. L’habitat privé des provinces occidentales du monde romain. Antécédents et prolongements: Protohistoire, Moyen Age et quelques expériences contemporaines. Actes du 2e congrès archéologique de Gaule méridionale (Lyon 2-6 Novembre 1983), pp. 47-59, Documents d’Archéologie Française 2. Maison des Sciences de l’Homme, Paris. Lézine, A. 1968: Carthage-Utique. Études d’architecture et d’urbanisme. CNRS, Paris. Lézine, A. 1971: “Utique. Note d’archéologie punique”, Antiquités africaines, 5, pp. 87-93. Mattingly, D., Edwards, D., and Sterry, M. 2015: “The origins and development of Zuwīla, Libyan Sahara: an archaeological and historical overview of an ancient oasis town and caravan centre”, Azania: Archaeological Research in Africa, 12, pp. 1-49. Mazzei, M. 1996: “Appunti per lo studio della casa nella Daunia Antica”, in D’Andria, F. and Mannino, K. (eds.), Ricerche sulla casa in Magna Grecia e

Anejos de AEspA LXXVII

MUD BRICK AND PISÉ DE TERRE BETWEEN PUNIC AND ROMAN

in Sicilia. Atti del Colloquio (Lecce, 23-24 giugno 1992), pp. 335-354. Congedo, Galatina. Meates, G. 1974: “Farningham Roman villa II”, Arch­ aeologia Cantiana, 88, pp. 1-21. Meates, G. 1979: The Lullingstone Roman Villa, Vol. I: The Site. Kent Archaeological Society, Maidstone. Minke, G. 2006: Building with Earth. Design and Technology of a Sustainable Architecture. Birkhäuser, Basel. Nielsen, E. and Phillips, K. M. Jr. 1985: “Poggio Civitate (Murlo)”, in Stopponi, S. (ed.), Case e palazzi d’Etruria, pp. 64-69. Electa, Milano. Nin, N. 1988: “Deux villages bâtis en terre”, Dossiers d’Histoire et d’Archéologie, pp. 58-68. Perring, D. 1985: “La Bretagne (2): Londres et les villes du sud-est”, in Lasfargues, J. (ed.), Architectures de terre et de bois. L’habitat privé des provinces occidentales du monde romain. Antécédents et prolongements. Protohistoire, Moyen Age et quelques expériences contemporaines. Actes du 2e congrès archéologique de Gaule méridionale (Lyon 2-6 Novembre 1983), pp. 47-59, Documents d’Archéologie Française 2. Maison des Sciences de l’Homme, ­Paris. Perring, D. 1991: The Archaeology of Roman London, II: Early Developments of Roman London West of the Walbrook, CBA Research Report 70. Council for British Archaeology, London. Pesando, F. 2007: “Case di età medio-sannitica nella Regio VI: tipologia edilizia e apparati decorativi”, in Guzzo, P.-G. and Guidobaldi, M.-P. (eds.), Nuove ricerche archeologiche nell’area vesuviana, pp. 159176, Studi della Soprintendenza archeologica di Pompei 25. Electa Napoli, Napoli. Pesando, F. 2011: “L’ars struendi nella precettistica catoniana (agr. 14)”, in Roselli, A. and Velardi, R. (eds.), L’insegnamento delle technai nelle culture antiche (Ercolano, 23-24 marzo 2009), pp. 85-94, Quaderni di AION sez. Filologico-letteraria 15. Serra, Pisa-Roma. Rael, R. 2009: Earth Architecture. Princeton Architectural Press, New York. Rees, A. 1819: The Cyclopaedia; or, Universal Dictionary of Arts, Sciences, and Literature. Longman, Hurst, Rees, Orme, & Brown, London. Regoli, E. 1985: “Tecnica e tipologia delle costruzioni” in Carandini, A. (ed.), Settefinestre. Una villa schiavistica in Etruria romana, Vol. I, pp. 61-73. Panini, Modena. Renault, J. 1913: “Les bassins du Trik Dar-Saniat”, Cahiers archéologique tunisienne, 1, pp. 1-116. Ruffin, E. 2000: Nature’s Management: Writings on Landscape and Reform, 1822-1859 (ed by J. Temple

143

Kirby). University of Georgia Press, Athens, Ga.London. Ricci, G. 2006: “Periodo 3. La villa dell’Acheloo (300225 A.C.)”, in Carandini, A., D’Alessio, M. T. and Di Giuseppe, H. (eds.), La fattoria e la villa dell’Auditorium, pp. 191-201. “L’Erma” di Bretschneider, Roma. Roux, J.-C. 2003: “Exemples de mise en œuvre de murs en terre massive, dans la ville portuaire de Lattes (Hérault), aux Ve et première moitié du IV s av. n. è.”, in Échanges transdisciplinaires sur les constructions en terre crue. 1. Terre modelée, découpée ou ­coffrée. Matériaux et modes de mise en œuvre, pp. 263-273. Éditions de l’Espérou, Montpellier. Roux, J-C. 2010: “Les techniques constructives en bauge”, Lattara, 21.1, pp. 219-288. Slim, H. 1985: “La Tunisie”, in Lasfargues, J. (ed.), Architectures de terre et de bois. L’habitat privé des provinces occidentales du monde romain. Antécédents et prolongements. Protohistoire, Moyen Age et quelques expériences contemporaines. Actes du 2e congrès archéologique de Gaule méridionale (Lyon 2-6 Novembre 1983), pp. 35-45, Documents d’Archéologie Française 2. Maison des Sciences de l’Homme, ­Paris. Staccioli, R. A. 1967: “Sulla struttura dei muri nelle case della città etrusca di Misano a Marzabotto”, Studi Etruschi, 35, pp. 113-126. Tang, B. 2005: Delos, Carthage, Ampurias. The Housing of Three Mediterranean Trading Centres, Analecta Romana Instituti Danici Suppl. 36. “L’Erma” di Bretschneider, Roma. Ville, G., 1961: “La maison et la mosaïque de la chasse à Utique”, Karthago, 11, pp. 18-76. Wacher, J. S. 1985: “La Bretagne (1): généralités”, in Lasfargues, J. (ed.), Architectures de terre et de bois. L’habitat privé des provinces occidentales du monde romain. Antécédents et prolongements: Protohistoire, Moyen Age et quelques expériences contemporaines. Actes du 2e congrès archéologique de Gaule méridionale (Lyon 2-6 Novembre 1983), pp. 143-152, Do­ cuments d’Archéologie Française 2. Maison des Sciences de l’Homme, Paris. Williams-Ellis, C. 1916: Building in Cob, Pisé, and Stabilized Earth. Country Life Limited, London. Zarmakoupi, M. 2015: “Les maisons des négociants italiens à Délos: structuration de l’espace domestique dans une société en mouvement”, Cahiers «Mondes Anciens», 7, December 2015, URL: http:// mondesanciens.revues.org/1588 (consulted January 2016).

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

L’INDUSTRIE DE LA BRIQUE CRUE DANS LA COLONIA LUGDUNUM. TYPOLOGIE, APPROVISIONNEMENT ET ORGANISATION DE LA PRODUCTION BENJAMIN CLÉMENT Université Lumière Lyon 2/ARCHEODUNUM

ABSTRACT: Mud brick was the most commonly used material in domestic architecture of the ­roman colony of Lugdunum. Clay masonry is very well preserved in the form of mud brick found in fire debris layers which allows a wide corpus of mud bricks to be collected in order to establish a typology. The recent excavations reveal different ways of clay extraction for the development of mud bricks and of filled-frame construction. Analysis of the extraction pit has led to several questions about the organisation of the mud brick industry in Lyon. Were the extraction pits of a single parcel used for only one or for several building sites? As the production of mud-bricks takes time and space, was it directly done on the building site, or should we consider a different organisation of the ­production? The discovery of extraction pits in the streets poses the question of the juridical status of this particular area during the building phase. KEYWORDS: Mud-bricks, Typology, Building materials, Roman construction industry, Organization of production, Lyon, Gaul, Antiquity. RÉSUMÉ: La terre crue constitue le matériau le plus communément employé dans l’architecture domestique de Lugdunum. La découverte de maçonneries en terre préservées par les incendies, ainsi que de briques crues ou fragments de torchis dans les niveaux de démolition, a permis d’obtenir un vaste corpus, nécessaire à la mise en place d’une typo-chronologie de ces matériaux de construction. D’autre part, les fouilles récentes ont révélé différents types de structures d’extraction de terre destinées à la production d’adobes, ou à l’édification de maçonneries. L’analyse de ces fosses d’extraction pose une série de questions quant à l’organisation de l’industrie de la brique crue à Lugdunum. Les fosses ouvertes sur une parcelle donnée sont-elles destinées à un ou plusieurs chantiers de construction? Puisque la production d’adobes requiert du temps et de la place, cette opération se déroulait-elle directement sur le chantier, ou doit-on considérer une organisation différente? La découverte de fosses d’extraction au centre des rues pose enfin la question du statut juridique de ces espaces particulier durant un chantier de construction. MOTS-CLÈS: Terre crue, Brique crue, Typologie, Matériaux de construction, Industrie de la construction antique, Organisation de la production, Lyon, Gaule, Antiquité. RESUMEN: La tierra cruda representa el material más común empleado en la arquitectura doméstica de Lugdunum. El descubrimiento de estructuras en tierra preservados por los incendios, así como de ladrillos crudos o fragmentos de bajareque en los niveles de demolición, ha permitido obtener un amplio corpus, fundamental para la realización de una tipo-cronología de estos materiales de construcción. Por otra parte, las excavaciones recientes han revelado la presencia de diferentes tipos de contextos de extracción de tierra destinados a la producción de adobes o a la construcción de estructuras en tierra. El análisis de estas fosas de extracción plantea una serie de cuestiones en relación con la organización de la industria del ladrillo crudo a Lugdunum. ¿Las fosas abiertas en cada parcela se destinan a una o más obras de construcción? ¿Visto que la producción de adobes requiere de tiempo y sitio, las operaciones se llevaron a cabo directamente en la obra o tenemos que imaginar una orga-

146

Benjamin Clément

Anejos de AEspA LXXVII

nización distinta? El descubrimiento de fosas extractivas en el centro de calles plantea finalmente la cuestión de la condición jurídica de estos espacios durante la ejecución de una obra edilicia. PALABRAS CLAVE: tierra cruda, ladrillo crudo, tipología, material de construcción, industria de la construcción antigua, organización de la producción, Lyon, Galia, Antigüedad.

La brique crue, qui commence à être systématiquement employée à Lyon suite à la fondation coloniale en 43 av. J.-C., constitue sans doute le matériau de construction le plus couramment employé dans l’architecture domestique de la colonie jusqu’à la fin du iiie siècle. L’opus caementicum et la brique cuite ne sont utilisés que pour réaliser les solins, qui soutiennent une élévation systématiquement édifiée en opus latericium ou craticium. Paradoxalement, ce type de matériaux ne constitue que rarement un sujet d’étude pour les archéologues lyonnais (Desbat 1981: 55-80; Desbat 1985: 75-83), alors qu’ils sont systématiquement présents lors des fouilles archéologiques urbaines. En effet, depuis une trentaine d’années ces opérations préventives ont permis de renouveler notre connaissance sur ce sujet, notamment grâce à la découverte de nombreux restes de maçonneries en terre encore en place (incendiées ou non), ou de fragments de briques crues. La reconnaissance des fosses d’extraction de terre, en lien avec les différentes phases de chantier d’une maison, d’un îlot, voire d’un pan entier de la colonie, permet également d’apporter de nouvelles données sur l’art de bâtir en terre. C’est à travers l’analyse des traces d’outils, de la forme et des dimensions, ou encore de l’emplacement et de la relation stratigraphique de ces structures d’extraction avec les niveaux de chantier (construction ou démolition), qu’il est possible de s’interroger précisément sur les artisans briquetiers de Lyon, leurs savoir-faire, mais encore sur l’organisation de cette industrie emblématique de la construction urbaine antique.

EXTRACTION DE LA TERRE À BÂTIR L’une des particularités de la colonie de Lyon est d’avoir fourni un nombre important de fosses d’extraction de terre, utilisée pour la mise en œuvre des maçonneries. Si la découverte de fosses d’extraction en Gaule n’est pas une nouveauté, le fait est qu’à Lyon, ce sont de véritables carrières à ciel ouvert qui ont été creusées, afin de fournir la quantité de matériau nécessaire au bon déroule-

ment des grands chantiers de construction qui rythment l’histoire de la colonie. La terre employée provient de plusieurs gisements qui ont pu être localisés avec précision sur le territoire de Lugdunum (fig. 1). Les bancs de lœss/lehm1 sont principalement concentrés au sommet du plateau de Fourvière, ainsi que sur la colline de la Croix-Rousse. Ce sédiment est sans doute la formation la plus exploitée pour réaliser des briques crues, car présent au sein du pomerium de Lyon et accessible directement depuis les parcelles à bâtir. L’argile, et les dépôts fluvio-lacustres, constituent le second type de terre à bâtir. Ils se concentrent plutôt dans la plaine alluviale de Vaise, et sur les rives de la Presqu’île. Leur emploi semble coïncider avec l’extension de l’emprise urbaine à partir du milieu du ier siècle apr. J.-C. La carrière de l’Hôpital Fourvière (Lyon 5e)2 Cette opération archéologique a permis d’apporter une documentation récente et de bonne qualité sur les structures d’extraction de terre (fig. 2). Elles ont pu être mises en évidence sur le site à chaque période chronologique, en grand nombre, avec un registre de formes diversifiées. Les traces d’extraction sont présentes dès la première occupation de ce secteur (état 1 et 2), dans les années 30/20 av. J.-C. Il s’agit de fosses de dimensions variées (entre 2 et 5 m de diamètres), de forme régulière en surface, mais dont le profil présente des surcreusements ou un aspect piriforme, ce qui indique une extraction en creusements successifs. Leur profondeur est assez importante, jusqu’à 1 m environ. Il est intéressant de noter qu’elles se concentrent principalement dans l’espace dédié à la voirie, qui est mise en place lors de cette phase de construction. La destruction de ce premier état par un incen­ die violent, survenu dans les années 10 apr. ­J.-C., 1   Les loess et Lehm correspondent à des dépôts éoliens de sable et limon. 2  Fouille effectuée sous la direction de T. Silvino, pour Archeodunum (Silvino 2010).

Anejos de AEspA LXXVII

L’INDUSTRIE DE LA BRIQUE CRUE DANS LA COLONIA LUGDUNUM...

Fig. 1. Carte géologique de Lyon, avec l’emplacement des carrières de terre à bâtir (Fond de carte: BRGM).

147

148

Benjamin Clément

Anejos de AEspA LXXVII

Fig. 2. Hôpital Fourvière, état 3 (5/10 apr. J.-C.), plan général du site avec l’emplacement des fosses d’extraction de terre à bâtir (DAO: D. Tourgon D., B. Clément).

entraîne une reconstruction complète de l’îlot. En amont du chantier de construction, les parcelles ont été percées de nombreuses fosses d’extraction, de très grandes dimensions, qui prennent des formes diverses (fig. 3). Certaines s’apparentent même à des carrières de dimensions modestes. La majeure partie est représentée par des fosses polylobées, à parois irrégulières et fond relativement plat. La plus importante (F176) présente une longueur de 12 m pour une largeur de 4 m environ et une excroissance au sud de 3 m. Sa profondeur atteint presque 2 m pour une capacité d’environ 120 m3! Le second type de fosse correspond à une grande tranchée (F147) reconnue sur 12,10 m de long pour une largeur de 1,70 m (fig. 4). Elle s’élargit au nord pour atteindre environ 3 m. Cette

tranchée présente un profil quasi rectangulaire et sa profondeur atteint 1,40 m. Son creusement a permis d’extraire environ 110 m3 de terre à bâtir. Ces fosses présentent un comblement identique, réalisé en une seule fois, et constitué des restes de construction (tegulae, imbrices, briques crues, associées à un important mobilier céramique) issus de l’incendie qui a ravagé cette partie de la co­ lonie. Les fosses observées résultent donc d’une extraction de matière par creusement unique. Elles ont ensuite été réutilisées comme fosses dépotoirs afin d’assainir les parcelles dans l’optique de la reconstruction du quartier. Elles ont permis d’extraire un volume de matière exceptionnel par rapport aux autres exemples connus dans la colonie,

Anejos de AEspA LXXVII

L’INDUSTRIE DE LA BRIQUE CRUE DANS LA COLONIA LUGDUNUM...

Fig. 3. Hôpital Fourvière, état 3 (5/10 apr. J.-C.), fosses polylobées d’extraction de terre à bâtir (clichés: Archeodunum).

Fig. 4. Hôpital Fourvière, état 3 (5/10 apr. J.-C.), tranchée d’extraction de terre à bâtir (clichés: Archeodunum).

149

150

Benjamin Clément

laissant penser à la mise en place d’une véritable carrière d’extraction préalable à l’ouverture du chantier de construction. Notons enfin que les fosses d’extraction les plus récentes correspondent une nouvelle fois à la reconstruction de l’îlot qui se déroule au milieu du ier siècle apr. J.-C. Elles présentent les mêmes caractéristiques morphologiques que celles des états précédents. Leur comblement, de nature détritique, semble indiquer qu’elles ont été réutilisées comme dépotoir. Les parcelles à bâtir étant déjà en partie occupées par les constructions précédentes, ces structures d’extraction ont été creusées de préférence sous les portiques et dans la rue. Le clos du Verbe-Incarné (Lyon 5e)3 Cette opération, effectuée de 1979 à 1987, a permis l’exploration d’un quartier d’habitation de la colonie où de nombreuses fosses d’extraction de loess, présentant des formes diversifiées, ont pu être mises en évidence. Toutefois, elles n’ont pas nécessairement été documentées sur le terrain (dessin en plan et coupe), ou décrites dans les rapports de fouille, ce qui complique leur interprétation. Certaines structures particulières, ou exceptionnelles par leur forme ou leur dimension, ont cependant fait l’objet de descriptions détaillées, que nous présenterons dans cette partie. Extraction dans le cadre d’un chantier. Une fosse d’extraction bien documentée a pu être observée au sein de la domus au Faune Flûteur, dégagée sous le sanctuaire du culte impérial (Îlot XXIV) (fig. 5). Implantée au cœur de la parcelle dans les années 20/15 av. J.-C., cette fosse de forme pseudo-circulaire (diamètre de 2,50 m) présente un creusement unique, piriforme, d’une profondeur de 1,50 m. Elle a permis d’extraire environ 10 m3 de loess avant d’être comblée par des restes de construction et un dépotoir domestique. Cette fosse participe sans doute au chantier de construction de la domus, afin de fournir la terre nécessaire au moulage des briques crues employées dans les cloisons et les foyers. Au moins quatre autres structures similaires ont été repérées sous les branches du portique de rue, mais n’ont malheureusement pas été documentées. L’analyse de la 3   Fouille réalisée sous la direction de B. Mandy, pour le Service Archéologique de la ville de Lyon (Mandy 1979; Mandy 1986).

Anejos de AEspA LXXVII

céramique les rattache à la phase postérieure. Elles ont sans doute été creusées à la suite de l’abandon de la domus et de la création du Sanctuaire. En parallèle de ces fosses tibériennes creusées dans la maison du Faune Flûteur, une carrière en tranchée a été implantée au centre de l’îlot XXIV (Mandy 1979: 16-19). Elle présente un véritable front de taille qui se développe en ligne droite sur près de 30 m de long, à l’ouest du soubassement du temple. L’accès à cette structure se faisait par un plan incliné permettant d’atteindre le fond de la tranchée, à 2,25 m de profondeur. D’une largeur de 2,30 m en moyenne, elle était limitée au sud par une remontée de moraine contre laquelle elle s’appuie, vérifiant ainsi sa destination comme carrière de loess. Cette structure, qui correspond sans doute à la plus grande carrière mise au jour à Lyon, a permis d’extraire au minimum 150 m3 de loess. À celle-ci s’ajoute la découverte de huit fosses de plan ovoïde et de profil piriforme. D’un diamètre compris entre 0,80 et 1,60 m, elles étaient conservées sur moins d’un mètre de profondeur, partiellement arasées par la construction du sanctuaire. À l’instar des autres fosses d’extraction de loess, ces fosses étaient comblées par un important mobilier céramique, associé à des déchets d’artisanat métallurgique et de tabletterie. À l’instar de la fouille de l’hôpital Fourvière, une grande quantité de fosses d’extraction sont également présentes dans les espaces dédiés à la voirie, notamment dans les rues de Cybèle et de la  Fontaine. Implantées dès l’époque augustéenne, elles ont pour la plupart été creusées lors de la reconstruction de l’ensemble du quartier à l’époque claudienne. Ainsi, la rue de Cybèle va être perforée de plusieurs fosses à creusement multiples, permettant d’extraire environ 100 m3 de terre à bâtir, avant que la rue soit pavée de dalles de granite (fig. 5). Extraction ponctuelle. À côté de ces carrières à ciel ouvert qui ponctuent les premières phases de construction du quartier, plusieurs fosses isolées, souvent implantées dans les maisons ellesmêmes, ont été mises au jour à partir du milieu du ier siècle apr. J.-C. L’exploration de la maison du Laraire (Îlot XIV – parcelle 4) a permis d’observer une fosse d’extraction de terre implantée au centre de la pièce principale de la domus. Lors de la réfection de la maison, à l’époque néronienne, le sol de la pièce 10 a été perforé par une fosse à creusements multiples (fig. 6). Elle présente un

Anejos de AEspA LXXVII

L’INDUSTRIE DE LA BRIQUE CRUE DANS LA COLONIA LUGDUNUM...

151

Fig. 5. Clos du Verbe-Incarné. a) fosse polylobée d’extraction découverte dans la maison au Faune Flûteur (cliché: SAVL); b) vue général des fosses d’extraction dans la rue de Cybèle (cliché: SAVL); c) coupe de la rue de Cybèle (d’après Mandy 1984).

plan de forme irrégulière, pour un diamètre de 4,50 m et une profondeur de 1,50 m. Elle a permis d’extraire une quantité d’environ 35 m3 de terre. Une structure similaire a été mise en évidence lors de la fouille des pièces 11 et 12 de la maison à la Banquette Chauffante (XIV-3). Ces espaces, sur hypocauste, sont repris au milieu du iie siècle, et une grande fosse aux dimensions de la pièce est creusée sur près de 1,40 m de profondeur.

Ces fosses, qui semblent répondre à un même phénomène, ont sans doute été employées pour extraire la quantité de terre nécessaire à la restauration de la domus, sans pour autant en modifier ses composantes principales. Une fois la fosse comblée par un dépotoir céramique et de terre cuite architecturale, le sol en béton de la pièce a été systématiquement reconstruit. Pour le cas de la maison du Laraire, le sol est doté d’un em-

152

Benjamin Clément

Fig. 6. Clos du Verbe-Incarné, état IV (milieu du Ier siècle apr. J.-C.). a) plan de la maison du Laraire; b) vue de la fosse d’extraction creusée dans la pièce 10 (cliché: SAVL).

blema en opus sectile, indiquant ainsi que cet espace  conserve son rôle de pièce principale de la maison.

Anejos de AEspA LXXVII

Extraction en galerie. La structure la plus surprenante a été découverte au sein de l’îlot XV. Une galerie longitudinale souterraine, traversant une partie de l’îlot, pourrait être interprétée comme une structure d’extraction de terre à bâtir (fig. 7). Constituée d’un puits d’accès creusé à partir de la parcelle 4, au pied du mur de terrasse qui la sépare de la parcelle 3, la galerie part ensuite en direction de l’ouest à environ 4 m sous les niveaux d’occupations contemporains. Le puits, profond d’environ 2,50 m pour une largeur de 1,50 m, donne accès au couloir par plusieurs marches taillées directement dans le loess, sur le bord sud. L’accès se fait ainsi perpendiculairement au conduit. Les marches font environ 25 cm de hauteur, mise à part la dernière qui atteint 80 cm. La galerie présente une largeur de 0,90 m pour une hauteur de 2 m à son départ (fig. 7-8a). Elle se poursuit sur 12 m avant d’être obstruée par un éboulement qui colmate la totalité du conduit. Notons la présence d’un renfoncement de 0,50 m de profondeur dans la paroi sud dont la destination reste hypothétique (espace lié à la circulation des ouvriers, stockage des outils?). Cette galerie présente un profil à bord droit et fond plat. Son plafond est grossièrement semi-circulaire, permettant ainsi de mieux répartir les charges. Aucune trace de montant en bois n’a pu être observée, ce qui laisse supposer que la galerie n’a pas été étayée. De nombreuses traces liées au creusement sont présentes sur les parois et permettent de reconnaître l’outil qui a été utilisé (fig. 8b). L’association de traces obtenues par un objet pointu et un objet tranchant laisse supposer qu’elle a été excavée à l’aide de houes ou de pioches, voire d’un outil qui présenterait deux parties actives différentes, comme le bidens. Notons également la présence d’une petite niche ovoïde, aux parois rubéfiées, qui devait recevoir une lampe à huile et ainsi assurer l’éclairage de la cavité (fig. 8c). Cette structure est comblée par des niveaux de colluvionnement qui ont livré un abondant matériel (céramique et verre) permettant de dater son abandon dans la première moitié du iiie siècle apr. J.-C. Associé à ce mobilier, un dépôt d’objets métalliques a été découvert. Il est principalement constitué d’outils liés au travail du bois et de la pierre (ripes, hache, ciseaux, fers à rabot, pierre à affûter les gouges...) (fig. 8d). La destination de cette tranchée reste hypothétique: extraction de matériaux? Passage de ­voleurs en direction de la Villa aux Mosaïques

Anejos de AEspA LXXVII

L’INDUSTRIE DE LA BRIQUE CRUE DANS LA COLONIA LUGDUNUM...

153

Fig. 7. Clos du Verbe-Incarné, état VI (iiie siècle apr. J.-C.). a) plan de l’îlot XV avec l’emplacement de la galerie d’extraction (d’après Mandy 1986); b) vue de la galerie (cliché SAVL); c) coupe de la galerie (d’après Mandy 1986).

(îlot XXV), située plus à l’ouest de l’autre côté de la rue, comme semble le penser le fouilleur (Mandy 1986)? La qualité technique de l’ouvrage, associé à l’absence de place dans le quartier à cette

époque pour extraire de la terre à bâtir, semble plutôt orienter l’interprétation vers l’extraction de matériaux. À ce propos, on rappellera le creusement de pièces supplémentaires à partir du iiie

154

Benjamin Clément

Anejos de AEspA LXXVII

Fig. 8. Clos du Verbe-Incarné, état VI (iiie siècle apr. J.-C.). a) vue du comblement de la galerie; b) traces d’outils sur les parois; c) logette dans la paroi servant à recevoir une lampe à huile; d) dépôt d’outils découvert dans le comblement de la galerie (cliché et dessin: SAVL).

siècle sur le site, afin de dégager des espaces de vie supplémentaires sur des parcelles déjà bien occupées (Delaval 1995; Delaval et Thirion 2012:

101-125). La présence d’un dépôt d’outils métalliques liés aux artisans de la construction, constitué au moment du comblement de la galerie,

Anejos de AEspA LXXVII

L’INDUSTRIE DE LA BRIQUE CRUE DANS LA COLONIA LUGDUNUM...

semble également orienter son interprétation vers l’extraction. En effet, ce dépôt fait écho à une pratique cultuelle sans doute réalisée par des fabri ou des structores, plutôt que par des voleurs, pour consacrer la fermeture de cette cavité. Quoi qu’il en soit, cette tranchée a permis de retirer au minimum 80 m3 de terre à bâtir, sans toucher aux constructions déjà en place sur le site. Elle doit sans doute être liée aux derniers réaménagements du quartier qui interviennent dans la première moitié du iiie siècle apr. J.-C.

155

2009). Réparties anarchiquement sur l’emprise de fouille, elles présentent des formes irrégulières ou polylobées, avec des dimensions moyennes comprises entre 1,20 × 2,50 m et 4,50 × 6 m, pour une profondeur d’environ 1 m. Ces fosses ont été creusées à l’occasion de la reconstruction de ce secteur du vicus de Vaise, à la suite d’un incendie violent qui intervient au cours de l’époque tibérienne. Elles ont permis d’extraire près de 150 m3 de terre à bâtir, sans doute employées pour édifier les murs des entrepôts et des bâtiments artisanaux construits au milieu du ier siècle.

Autres carrières connues à Lyon LA TYPOLOGIE DES BRIQUES CRUES D’autres mentions de fosses de grandes dimensions qui pourraient être rattachées à l’extraction de terre sont signalées sur différents sites de la colonie. Sur la colline de Fourvière, la reprise des fouilles sur le Pseudo-Sanctuaire de Cybèle a révélé plusieurs fosses – notamment dans le sondage D1 – qui pourraient être rattachées à l’extraction, ainsi qu’une tranchée (Desbat 2000). Cette dernière a été creusée dans la rue avant la réfection complète du quartier au début de l’époque augustéenne (20/15 av. J.-C.). Elle présente une largeur de 3,30 m, avec un creusement à bords évasés et fond plat, d’une profondeur de 1 m en moyenne. Cette tranchée est proche dans sa morphologie de celle découverte sur le site de l’Hôpital Fourvière. Lors des fouilles de l’Îlot Central de l’Antiquaille, une grande fosse creusée dans la rue à l’époque augustéenne a été observée en limite de chantier. Elle est interprétée par le fouilleur comme une fosse d’extraction de loess, liée au chantier de construction des grandes domus augustéennes implantées sur le site. Rue Henry Le Chatelier, sur la colline de Fourvière, une fosse similaire, de près de 8 m de diamètre, vient perforer le fossé protohistorique (Goudineau 1989). De forme polylobée, elle présente une profondeur d’environ 1,80 m; elle a été entièrement comblée par un dépotoir domestique d’époque augustéenne. Enfin, signalons la découverte de plusieurs fosses d’extraction de loess lors des fouilles de la basilique St-Just (Reynaud 1981). Hors de la colonie, si aucune fosse de ce type n’a été observée dans le quartier des Kanabae ou de Condate, les fouilles menées par M. Monin sur le site de la rue du Mont d’Or à Vaise, ont cependant mis au jour plusieurs fosses d’extraction datées de la première moitié du ier siècle (Monin

Plusieurs modules de briques crues ont été employés successivement ou simultanément dans la colonie (Desbat 1981: 55-80; Desbat 1985: 7583). Ces modules semblent varier selon le type de maçonnerie dans laquelle la brique va être mise en œuvre. Notre analyse s’appuie sur celle déjà réalisée (Desbat 1985: 83), à laquelle nous ajouterons les éléments de notre propre corpus. Ce dernier se compose de 311 fragments de briques pour un nombre minimum d’individu de 226, répartis sur 11 sites différents. Analyse modulaire Trois modules fondés sur l’évolution de la longueur et de la largeur des briques semblent se dégager de notre analyse: Le type A (fig. 9a), avec un module moyen de 44 × 29 × 7,5 cm, correspond exactement au module lydien décrit par Vitruve (II, 3, 3). Ce type de brique est essentiellement employé pour l’élévation des solins maçonnés constituant les murs porteurs des bâtiments construits entre la fin du  ier siècle av. et le milieu du ier siècle apr. J.-C. Seul le site du clos de la Solitude semble encore  employer ce type de briques à la fin du ier siècle, pour l’élévation d’un solin situé au rez-dechaussée d’une boutique. Signalons également leur mise en œuvre dans les architectures à pans de bois, où elles sont disposées verticalement sur deux cloisons de la Maison A, rue Auguste Comte. Le type B (fig. 9b), avec un module moyen de 28,5 × 14 × 7 cm, est le plus employé dans la colonie. Il correspond à un demi-module de type A, avec une longueur d’un pied; une largeur d’un de-

156

Benjamin Clément

Anejos de AEspA LXXVII

Fig. 9. Les briques crues de la colonie de Lyon. A) briques de type A associées à une architecture en opus latericium (clichés: A. Desbat); b) briques de type B associées à une architecture en opus craticium.

Anejos de AEspA LXXVII

L’INDUSTRIE DE LA BRIQUE CRUE DANS LA COLONIA LUGDUNUM...

mi-pied, et une épaisseur d’un palmus environ. À la différence du précédent, ce type de brique a été principalement mis en œuvre dans les cloisons à pans de bois. Il intervient dès l’époque augustéenne et au moins jusqu’à la fin du iie siècle sur l’ensemble du territoire colonial. Le type C présente un module de 21 × 12 × ×  6  cm. Observé ponctuellement dans des cloisons à pans de bois, ce module correspond à une longueur de deux tiers de pied; une largeur d’un tiers de pied pour une épaisseur proche d’un palmus. Le type C, qui apparaît au milieu du ier siècle, semble avoir été employé essentiellement au cours du iie siècle à Lyon. L’emploi du module lydien pour les briques crues est assez courant dans le monde médite­ rranéen. Il se rencontre essentiellement dans le monde étrusque – à Arezzo, Fiesole, Gela, Pyrgi, Rusellae – ou punico-romain – St Monique, Kerkouane, Utique ou encore Volubilis (De Chazelles 1997: 63‑64). En Gaule Narbonnaise, ce module est peu employé; on le retrouve uniquement à Olbia ou à Narbonne, alors qu’il est absent d’autres colonies romaines comme Nîmes ou Arles. Signalons que ce module est employé à Avenches, en Suisse, pour les murs périmétraux des domus du ier siècle apr. J.-C. (Morel 2001: 33). Les briques de type B, correspondant à une demi-brique lydienne, semblent peu employées dans le bassin méditerranéen. Une fois encore, les exemples les plus nombreux nous viennent d’Étrurie, où des briques similaires ont été observées à Misano ou Veii (Lugli 1957). À l’instar du module A, le module B a été repéré en Gaule sur le site d’Avenches, au ier siècle (Morel 2001: 14). Enfin, le module C, également fondé sur le pied romain, semble également rarement attesté en Gaule méridionale, puisqu’une seule occurrence est connue à Glanum, au ier siècle av. J.-C. (De Chazelles 1997: 60). Il est également attesté en Gaule Germaine, dans la cité de Metz, où il est employé dans des cloisons à pans de bois datées du milieu du ier siècle apr. ­J.-C. (Heckenbenner et al. 1992: 9-35). Les comparaisons avec les autres cités employant l’adobe soulignent également l’homogénéité des modules mis en œuvre à Lyon. À Glanum, au moins quatre modules différents ont été employés pour le plan d’urbanisme du début du ier siècle av. J.-C. Le même constat peut être avancé à Nîmes, où cinq modules distincts ont été observés sur le site de la Propriété Solignac, au ier siècle apr. J.-C. (De Chazelles 1997: 60).

157

SYNTHÈSE Ces trois modules, fondés sur le pied romain, ont été employés à Lyon entre le milieu du ier siècle av. J.-C. et la fin du iiie siècle apr. J.-C. Les types A et B apparaissent dès la fondation coloniale alors que le type C semble être employé seulement à partir du milieu du ier siècle, époque à laquelle va progressivement disparaître le type A. On constate une bonne homogénéité des dimensions des briques, avec des écarts types faibles – inférieur à 2 cm – ce qui dénote un soin particulier porté au matériau, ou un contrôle de la production stricte, afin de produire des briques homogènes plus faciles à mettre en œuvre. Si les longueurs et les largeurs des briques évoluent selon leur destination, l’épaisseur reste stable, avec une valeur moyenne de 7 cm (un palmus). Cette valeur semble s’affiner légèrement pour le type C, avec une épaisseur plus proche des 6 cm. Concernant la nature de la terre employée, le type A est exclusivement moulé avec du loess, mis à part les éléments recueillis sur la villa de StLaurent-d’Agny, réalisés à partir d’une argile limoneuse locale, et les briques observées dans le premier état de la maison aux Xenia, ces dernières n’ayant pas été conservée à l’issu de la fouille (Delaval et al. 1995). Cette préférence pour le loess doit être relativisée, les sites ayant livré ce type de matériaux étant tous situés sur la colline de Fourvière, où le loess affleure en de nombreux points. Toutefois, l’analyse des briques de type B montre que les adobes produites entre la fondation coloniale et le milieu du ier siècle sont exclusivement en loess, quelle que soit la localisation du site où elles sont mises en œuvre: colonie, Kanabae, Vaise. Ainsi, les bâtiments augustéens de la place Valmy, et tibériens de la rue Auguste Comte, sont édifiés avec des briques de type B constituées de loess. En revanche, ces mêmes sites emploient des briques de type B, réalisées à partir de limons locaux – ­limons de débordement rue Auguste Comte, et limons argileux lacustres place Valmy – à l’occasion des reconstructions effectuées à la fin du Ier ou au début du iie siècle. Quant aux briques de type B employées sur la colline de Fourvière aux ier et iie siècles, elles sont toujours façonnées à partir de loess. Les briques de type C, employées plus ponctuellement dans la colonie, semblent suivre le même schéma. Les éléments recueillis sur les sites de la rue des Farges ou du clos du Verbe-Incarné sont constitués de loess. À l’inverse, les adobes découvertes dans les couches de démolition du

158

Benjamin Clément

iie  siècle

de la rue Auguste Comte sont réalisées à partir d’une argile alluviale grossière. Signalons également des briques de type C observées dans une maçonnerie du clos de la Solitude, qui sem­ blent avoir été réalisées à partir d’un matériau brassé et anthropisé, proche des remblais antiques présents sur le site. ORGANISATION DE LA PRODUCTION DES BRIQUES CRUES: FABRICATION, STOCKAGE ET DISTRIBUTION Une production organisée? Au terme de notre analyse des briques crues de la colonie de Lugdunum, l’un des points marquants est l’homogénéité des modules employés dans l’architecture domestique. Ainsi, la pro­ duction d’adobes nous apparaît extrêmement encadrée, avec le recourt à des briques lydiennes (type A) pour les murs porteurs, et des briques de type  B (1/2 module lydien) pour les cloisons à pans de bois entre le milieu du ier siècle av. et le milieu du ier siècle apr. J.-C. À partir de cette date, le type A semble rapidement abandonné au profit du type B, parfois accompagné du type C, suite au développement de l’architecture à pans de bois sur solins maçonnés. Pour C.-A. De Chazelles, l’unicité des modules de briques crues, ainsi que de la terre employée est un indice d’une production organisée, et non du travail de structores au cas par cas, selon la nature des chantiers (De Chazelles 1997: 66‑69). La présence d’artisans spécialisés dans le façonnage, le séchage et le stockage dans de bonnes conditions des briques crues semble évidente dans le cas de la colonie de Lyon, où la demande en matériaux devait être extrêmement importante lors des grands plans d’urbanisme. Ce type d’organisation semble d’ailleurs indiqué a silentio par Pline, lorsqu’il mentionne la paternité des premières briqueteries d’Athènes, qu’il attribue à deux frères: Euryalus et Hyperbius (Plin. nat. VII, 57, 4). Un indice plus sûr nous est donné par l’édit du Maximum, promulgué sous le règne de Dioclétien, qui mentionne les salaires comparés des ouvriers moulant des briques crues et cuites (Giacchero 1974: section 7, 15-16). Cette mention prouve qu’au ive siècle, en ville, la production d’adobes était réalisée par des artisans spécialisés. Toutefois, il faut garder à l’idée la possibilité d’une production restreinte faite par un groupe

Anejos de AEspA LXXVII

familial, dans le cas de travaux limités (foyer, réparation de parois, bouchage de porte...), ou en contexte rural. De la carrière au chantier de construction Du point de vue morphologique, la quasi-totalité des vestiges d’extraction prend la forme de fosses polylobées profondes (souvent à creusements multiples) ou de tranchées à bords droits et fond plat. Mis à part la galerie observée sur le site du clos du Verbe-Incarné, l’exploitation de la terre se déroule systématiquement à ciel ouvert et les fosses sont creusées directement sur les parcelles à bâtir. Leur comblement est généralement unique, d’origine détritique, ce qui atteste leur réutilisation comme dépotoirs. Ce phénomène est encore plus marqué entre la fin du ier siècle av. et la première moitié du ier siècle apr. J.-C., période où le ramassage des déchets n’est pas encore mis en place au sein de la colonie (Desbat 2003: 117-120). D’après les vestiges de carrières de loess mis au jour dans la colonie de Lyon, on constate tout d’abord une surreprésentation des structures d’extraction sur le plateau de Fourvière, où le loess est facilement accessible. Toutefois, ces structures ont été essentiellement creusées entre l’épo­ que augustéenne et le milieu du ier siècle apr. J.-C. Les fosses postérieures au milieu du ier siècle, souvent de petites dimensions et disposées à l’intérieur des maisons, devaient être destinées à des réparations ponctuelles. Ce phénomène pourrait indiquer le déplacement des carrières à partir de cette époque, où le tissu urbain semble trop dense pour y implanter de véritables fosses d’extraction, capables d’assurer la production de briques pour un îlot entier. Il faut alors imaginer que des carrières ont été ouvertes en bordure de la colonie: soit dans le vallon de Trion, riche en limon argileux; soit plus à l’est, dans le secteur de St-Foyles-Lyon; soit dans une zone comprise entre Ecully et Champagne au mont d’Or, où de grands gisements de loess ont été observés. Sur la plaine de Vaise, pourtant riche en terre à bâtir de bonne qualité, il est étonnant qu’aussi peu de traces de ce type d’activité n’aient été mises en évidence. Si l’emploi de limons argileux provenant de ce secteur semble bien assuré, on peut supposer que les structures d’extraction devaient se situer en bordure des espaces urbanisés, sans doute au sein de domaines agricoles qui occupaient une bonne partie de ce secteur.

Anejos de AEspA LXXVII

L’INDUSTRIE DE LA BRIQUE CRUE DANS LA COLONIA LUGDUNUM...

Les modalités de l’implantation des fosses dans les parcelles révèlent des informations quant à la succession des équipes sur le chantier de construction. Le creusement des fosses suit systématiquement la délimitation des parcelles, et semble donc intervenir après les premiers travaux de terrassement qui devaient être effectués à l’échelle d’un îlot. L’exemple de l’Hôpital Fourvière est sans doute le plus parlant, puisque les structures d’extraction sont implantées anarchiquement de part et d’autre du mur de limite de parcelle M 30 (fig. 10). Cette répartition semble indiquer que l’extraction se déroulait parcelle par parcelle, et nécessitait donc l’accord du propriétaire. En revanche, l’implantation des fosses d’extraction ne tient pas compte du futur lotissement de la parcelle: sur les sites de l’Hôpital Fourvière, du clos du Verbe-Incarné, ou encore de la rue des Farges, il est fréquent que des murs porteurs traversent ces fosses, nécessitant d’augmenter la profondeur des maçonneries – parfois jusqu’à 2 m – afin d’assurer la stabilité de l’édifice. Ceci tend à démontrer que deux équipes différentes se succèdent sur le terrain, sans pour autant connaî­tre, ou tenir compte de leurs tâches respectives.  Cet argument confirme d’ailleurs l’existence d’artisans spécialisés dans la production de briques crues, différents des équipes de structores qui interviennent dans un second temps pour la construction des maçonneries. Enfin, la découverte de fosses d’extraction au sein même de la voirie, rue et portiques, pose une question d’ordre juridique, concernant le statut de ces espaces particuliers lors des travaux de construction d’un îlot, ou de réaménagement d’une parcelle. L’ouverture de fosses dans une rue oblige à sa fermeture temporaire, au moins au niveau d’une parcelle, si ce n’est au niveau d’un quartier, afin de permettre le bon déroulement des travaux. C’est donc la question du statut de la voie au moment du chantier qui est posée. Dans la législation romaine, la construction et l’entretien du portique, ainsi que des collecteurs d’eaux usées, sont à la charge du propriétaire de la parcelle, et non de la colonie (Alberti 2008: 109-115; Saliou 2008: 63-68; Saliou 2012: 15-29). Le creusement de fosses dans le portique serait donc, comme celles implantées dans les parcelles, à la charge du propriétaire. Concernant le statut de la rue à Rome, la «table d’Héraclée»4 nous indique 4   Table de loi découverte à Héraclée, en Italie du sud. Elle porte une succession d’extraits de lois romaines (Tabula Heracleensis, Crawford 1996: 355-391).

159

Fig. 10. Fondations maçonnées suivant le profil des fosses d’extraction. a) clos du Verbe-Incarné, état I (40-20 av. J.-C.) (cliché: SAVL); b) Hôpital Fourvière, état 4 (10 apr. J.-C.) (cliché: Archeodunum).

qu’elle était à la charge des riverains jusqu’à mi-largeur de la chaussée en face de la parcelle habitée (Saliou 2012: 21-22). En cas de non-réalisation des travaux, une entreprise pouvait être désignée aux frais du riverain. L’ensemble de l’entretien à l’échelle de l’Urbs était à la charge d’un magistrat, mais les travaux eux-mêmes pouvaient être effectués soit par le riverain, soit par une entreprise, et concernaient généralement une portion importante de voie. Dans le cas qui nous intéresse, bien que cette loi soit promulguée pour Rome, il est fort probable qu’une organisation similaire prévalait à l’entretien des rues dans la colonie de Lyon. Lors de ces travaux, qui ont dû

160

Benjamin Clément

Anejos de AEspA LXXVII

atteindre des dimensions colossales au milieu du ier siècle, à l’occasion du pavage de certaines rues, les voies devaient être ponctuellement fermées, permettant ainsi l’exploitation des ressources sous-jacentes. Mais cette exploitation était-elle soumise à l’accord du propriétaire en charge de la portion de voie en face de son immeuble, ou à l’autorisation accordée par le magistrat en charge des rues? La question reste ouverte. Cependant, le recourt fréquent à cette pratique, que ce soit à l’occasion de grands plans d’urbanisme, ou d’une simple reconstruction de parcelle, indiquerait plutôt l’intervention d’un particulier, sans doute avec accord préalable d’un magistrat. Organisation et gestion de la production L’analyse de ces structures d’extraction pose une série de questions quant à l’organisation du travail des briquetiers de la colonie. Tout d’abord, si l’on calcule le volume total de terre extraite sur le site de l’Hôpital Fourvière, pour les niveaux ­augusto-tibériens, on obtient une quantité minimale d’environ 500 m3 pour une surface correspondant à une parcelle, sans prendre en compte la terre issue des travaux de terrassement préalables. Ensuite, en développant notre raisonnement à partir de briques de type A, employées à cette époque sur le site pour l’élévation des murs porteurs avec un module de 44  ×  29  ×  8 cm, soit 0,010 m3, un tel volume permet de mouler environ 50.000 briques, et de réaliser 400 m de mur.5 Si l’on produit plutôt des briques de type B (29 × 14 × ×  7  cm), présentant un volume de 0,0028 m3, 200.000 briques au minimum ont pu être moulées à partir de ces fosses, ce qui correspond à environ 770 m de cloison en opus craticium.6 Le nombre extrêmement élevé de briques crues produites semble largement dépasser les besoins de construction des parcelles en travaux. À titre d’exemple, la maison augustéenne à l’Opus Spicatum, découverte sur le site du Pseudo-Sanctuaire de Cybèle, occupe une parcelle d’une superficie similaire à celle observée sur le site de l’Hôpital Fourvière (60 pieds de côté) (fig. 11). Ses 5   Il faut environ 1300 briques pour un mur d’adobe de 10 m de long, 0,50 m de large et 2,60 m de haut (soit 13 m3), permettant d’atteindre 3,50 m avec un solin maçonné de 0,90 m. 6   Il faut environ 2600 briques pour une cloison de 10 m de long, 3,50 m de haut et 0,25 m de large (soit 6,5 m3), en comptant un poteau de 0,20 × 0,15 m disposé tous les 0,60 m selon la disposition la plus classique à Lyon (Desbat 1985: 83).

Fig. 11. Plan et élévation de la maison à l’Opus Spicatum, Pseudo-Santuaire de Cybèle, état 2 (d’après Desbat 2004).

élévations réalisées en briques d’adobe de type A, reposent sur des solins maçonnés de 0,90 m de haut en moyenne. A. Desbat restitue un étage pour cette domus, qui se développe sur les trois ailes de la maison, ainsi que sur les boutiques donnant sur la rue de Cybèle (Desbat 2004: 221231). Un rapide calcul permet d’arriver à 140 m linéaires de mur au rez-de-chaussée, auxquels on peut encore ajouter 100 m pour les élévations de l’étage. Ainsi, la maison est matérialisée par 240 m linéaires de maçonnerie de 3,50 m de haut en moyenne. En se fondant sur les calculs précédents, il a fallu environ 30.000 briques crues pour édifier

Anejos de AEspA LXXVII

L’INDUSTRIE DE LA BRIQUE CRUE DANS LA COLONIA LUGDUNUM...

161

Fig. 12. Briqueterie encore en activité dans le sud de l’Inde. a) préparation de la terre; b et c) moulage d’une brique crue; d) séchage des briques directement sur le sol (clichés: M. Demierre).

cette domus, en plus des moellons et du mortier employés dans les solins. Ces résultats, même s’ils doivent être pris à titre indicatif, sont bien en deçà de la quantité de terre à bâtir extraite sur la parcelle 2 de l’Hôpital Fourvière. Il faut sans doute envisager qu’une partie de la terre servît à produire des matériaux destinés aux parties de la colonie bâties sur un terrain naturel morainique ou rocheux, et où il n’y avait pas d’accès direct à cette matière première. Rappelons que cette estimation

ne prend pas en compte la terre issue des travaux de terrassement préalables à la construction, qui ont dû générer une quantité non négligeable de matériau. Une dernière question reste en suspens: sachant qu’il faut un espace assez conséquent pour mouler et faire sécher ces briques d’adobe, ontelles été réalisées sur place comme cela est souvent envisagé dans la littérature archéologique? En effet, si l’on calcule la surface qu’occupent au sol,

162

Anejos de AEspA LXXVII

Benjamin Clément

lors de la première étape de séchage, les 30.000 briques crues de type A nécessaire à l’édification d’une simple domus, on obtient une surface de près de 3.600 m².7 Cette surface est équivalente à celle occupée par un îlot d’habitation de la colonie de 120 × 270 pieds, soit 3.500 m² environ. Si l’on considère que le moulage et le séchage étaient effectués sur place, le stockage sur la tranche des briques crues, avant de les entreposer pour un séchage définitif par empilement sous forme de haies (couvertes ou non) (fig. 12), devait alors être la solution pour gagner de la place, déjà bien entamée par le creusement des fosses elles-mêmes. Cependant, l’absence de traces au sol des différentes activités liées à la fabrication des adobes, même ténues – aire de moulage, aire de séchage, aire de stockage marquée par des trous de poteau, accès à l’eau (citerne, puits?) –, semble aller à l’encontre de l’hypothèse d’une production globale directement sur les parcelles à bâtir. Même s’il faut rester prudent sur ce point, car des travaux de terrassement ont très bien pu faire disparaître ces traces, le caractère saisonnier, et le temps long de production semblent aller dans le sens de cette hypothèse. Le moulage des briques, à l’instar des chantiers de construction, se déroulait entre les mois d’avril et d’octobre, et exigeait une préparation de la terre, étape particulièrement longue (De Chazelles 1997: 66‑68), ce qui devait bloquer la parcelle à bâtir quasiment pour une saison. En effet, en considérant le temps de creusement, de préparation de la terre, de moulage et de séchage des briques8 (au moins deux mois), l’exploitation de 500 m3 de loess devait durer au minimum quatre à cinq mois. Ainsi, il semble difficile d’envisager l’immobilisation de parcelles à bâtir, voire d’îlots entiers, pendant un laps de temps aussi long, ce qui compliquerait le calendrier des redemptores qui devraient attendre que le terrain soit dégagé pour débuter leur travail. Une organisation différente doit donc être envisagée. L’extraction étant effectuée sur place, la terre pourrait alors être transportée vers des ateliers (grands entrepôts?) où les briques seraient moulées, séchées puis stockées en vue d’être redistribuées sur les chantiers de la colonie. Toutefois, il semble peu probable que les briques employées dans la 7   Sachant qu’une brique de type A possède une superficie de 0,12 m². 8  Vitruve (II, 3, 2) préconise un temps de séchage d’au moins 2 ans, faute de quoi les murs montés avec des briques incomplètement séchées n’auraient pas permis l’adhérence des enduits.

construction fassent l’aller-retour entre le chantier et l’atelier. On peut également envisager qu’une fois extraite, la terre était moulée sur place. Les mouleurs devaient attendre que les adobes soient suffisamment sèches pour être transportées vers des entrepôts où elles terminaient leur séchage avant d’être redistribuées sur les chantiers de construction de la colonie. Dans tous les cas, les briquetiers ont dû avoir recours aux vectores – charretiers qui transportaient les matériaux de construction jusqu’au chantier – afin de transporter leurs matériaux à travers la colonie. BIBLIOGRAPHIE Alberti, G. 2008: «La communication entre la rue et la maison: quelques exemples de Gaule Belgique et des Germanies», dans Ballet, P., Dieudonné-Glad, N. et Saliou, C. (dir.), La rue dans l’Antiquité. Définition, aménagements et devenir de l’Orient méditerranéen à la Gaule, pp. 109-115, Archéologie & Culture Rennes. Presses Universitaires de Rennes, Rennes. Crawford, M. H. 1996: Roman Statutes. Institute of Classical Studies, London. De Chazelles, C.-A. 1997: Les maisons en terre de la Gaule méridionale, Monographies Instrumentum 2. Mergoil, Montagnac. Delaval, E. 1995: L’habitat privé de deux insulae de la ville haute de Lugdunum (Lyon) sous le Haut-Empire Romain, Thèse de doctorat. Université d’Aix-Marseille, Aix-en-Provence. Delaval, E., Bellon, C., Chastel, J., Plassot, E. et Tranoy, L. 1995: Vaise, un quartier de Lyon antique, Documents d’archéologie en Rhône-Alpes. Série lyonnaise 11. Service régional de l’archéologie, Lyon. Delaval, E., Thirion, Ph. 2012: «Feuille 6 – Verbe-Incarné», dans Lenoble, M. (dir.), PCR Atlas topographique de Lugdunum, pp. 101‑215, Rapport de PCR 2012. DRAC Rhône-Alpes, Lyon. Desbat, A. 1981: «L’architecture de terre à Lyon à l’époque romaine», dans Walker, S. (dir.), Récentes recherches e archéologie gallo-romaine et paléochrétienne sur Lyon et sa région, pp. 55-80, B.A.R International Series 108. B.A.R., Oxford. Desbat, A. 1985: «La région de Lyon et de Vienne», in Lasfargues, J. (dir.), Architectures de terre et de bois: l’habitat privé des provinces occidentales du monde romain. Actes du 2e Congrès archéologique de Gaule méridionale (Lyon, 2-6 novembre 1983), pp. 75-83, Documents d’archéologie française 2. Maison des sciences de l’homme, Paris.

Anejos de AEspA LXXVII

L’INDUSTRIE DE LA BRIQUE CRUE DANS LA COLONIA LUGDUNUM...

Desbat, A. 2000: Le Pseudo-Sanctuaire de Cybèle, Lyon 5e, Rapport de fouille programmée. SRA Rhône-Alpes, Lyon. Desbat, A. 2003: «La gestion des déchets en milieu urbain: l’exemple de Lyon à la période romaine», dans Ballet, P., Cordier, P. et Dieudonné-Glad, N. (dir.), La ville et ses déchets dans le monde romain: rebuts et recyclage. Actes du colloque de Poitiers (19-21 septembre 2002), pp. 117-120. Mergoil, Montagnac. Desbat, A. 2004: «Une nouvelle maison à atrium, augustéenne, à Lyon (Fourvière)», Revue Archéologique de l’Est, 53, pp. 221‑231. Giacchero, M. 1974: Edictum Diocletiani de pretiis: Edictum Diocletiani et collegarum de pretiis rerum venalium. In integrum fere restitutum e Latinis Graecisque fragmentis editit Marta Giacchero. Istituto di storia antica e scienze ausiliarie, Genova. Goudineau, C. (dir.) 1989: Aux origines de Lyon, Documents d’Archéologie en Rhône-Alpes. Série lyonnaise 1. Circonscriptions de antiquités historiques, Lyon. Heckenbenner, D., Brunella, Ph., Leroy, M., Milutinovic, M. et Thion, P. 1992: «Le quartier de l’Arsenal à Metz (Moselle): topographie urbaine et évolution architecturale durant l’Antiquité», Gallia, 49.1, pp. 9‑35. Lugli, G. 1957: La tecnica edilizia romana, con particolare riguardo a Roma e Lazio. Bardi, Roma. Mandy, B. 1979: Clos du Verbe-Incarné, 24 rue Roger Radisson, Lyon 5e, Rapport de fouilles. SRA Rhône-Alpes, Lyon.

163

Mandy, B. 1986: Clos du Verbe-Incarné, 24 rue Roger Radisson, Lyon 5e, Rapport de fouille. SRA Rhône-Alpes, Lyon. Monin, M. 2009: Rue du Mont d’Or, Lyon 9e, Rhône, DFS de fouille préventive. SRA Rhône-Alpes, Lyon. Morel, J. 2001: «L’insula 12 et les quartiers adjacents à Avenches. Approche architecturale et urbanistique», Bulletin Pro Aventico, 43, pp. 10‑65. Reynaud, J.-F. 1981: Fouilles archéologique de SaintJust, Rapport de fouille programmée. SRA Rhône-Alpes, Lyon. Saliou, C. 2008: «La rue dans le droit romain classique», dans Ballet, P., Dieudonné-Glad, N. et Saliou, C. (dir.), La rue dans l’Antiquité. Définition, aménagements et devenir de l’Orient méditerranéen à la Gaule. Actes du Colloque de Poitiers (7-9 septembre 2006), pp. 63-68. Presses Universitaires de Rennes, Rennes. Saliou, C. 2012: «Le déroulement du chantier à Rome et dans le monde romain durant la période républicaine et le Haut Empire: une approche juridique», dans Camporeale, S., Dessales, H. et Pizzo, A. (dir.), Arqueología de la construcción III. Los procesos constructivos en el mundo romano: la economía de las obras (París, 10-11 de diciembre de 2009), pp. 15-29, Anejos de Archivo Español de Arqueología 57. CSIC, Mérida. Silvino, T. 2010: Lyon, Hôpital Fourvière, 8-10 rue Roger Radisson, Rapport de fouille. SRA Rhône-Alpes, Lyon.

V LIFTING MACHINES IN THEORY AND PRACTICE

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

TAKING A BEARING ON HERO’S ANTI-CRANE AND ITS UN-WINDLASS: THE RELATIONSHIP BETWEEN HERO OF ALEXANDRIA’S MOBILE AUTOMATON AND GRECO-ROMAN CONSTRUCTION MACHINERY, ARTILLERY AND WATER-LIFTING MACHINES DUNCAN KEENAN-JONES, IAN RUFFELL, EUAN MCGOOKIN University of Glasgow, Glasgow, Lanarkshire, United Kingdom

ABSTRACT: Hero’s treatise on building automata exploits ideas and terminology from machines used in Greco-Roman construction and other contexts. Hero’s drive mechanism inverted the function and load-bearing capacity of cranes and the key element, the exeliktra (‘unwinder’), encapsulated that inversion. Bearings were crucial to the success of these and other mechanical devices, but their characteristics have been neglected. A range of bearings, including the knōdax, khoinikis and khelōnion, began to take shape in construction machinery in the sixth century BCE, and was later extended to areas such as artillery. Its terminology remained stable into the imperial period, notwithstanding (partial) translation into Latin. In his automata, Hero prefers the more specialised iron knōdax and suggests that friction was particularly influential in that choice. Hero is consistent and demonstrates mechanical expertise and experience in his design, but both his statements and his ­silences allow us to draw some inferences about his intended audience. KEYWORDS: Crane, Windlass, Pulley, Mechanics, Automata, Vitruvius, Bronze, Iron, Friction, Exeliktra, Knōdax, Cnodax, Khoinikis, Khelōnion, Water screw, Water wheel, Paconius, Chersiphron, Metagenes. RESUMEN: El tratado de Herón sobre la construcción de mecanismos automáticos aprovecha las ideas y la terminología de máquinas empleadas en la construcción grecorromana y la de otros contextos. El mecanismo de accionamiento de Herón invierte la capacidad y la función del soporte de carga de las grúas y del elemento clave, el exeliktra. Los soportes fueron cruciales para el éxito de estos y otros dispositivos mecánicos, pero sus características se han descuidado. Una gama de soportes, incluidos los knōdax, khoinikis y khelōnion, empezaron a tomar forma de maquinaria de construcción en el siglo vi a.C. y, más tarde, se extendió a áreas como la artillería. Su terminología se mantuvo estable en el período imperial, a pesar de su traducción (parcial) al latín. En sus mecanismos, Herón prefiere el más adapto knōdax de hierro y sugiere que la fricción fue particularmente determinante en esa elección. Herón es preciso y demuestra conocimiento mecánico y experiencia en su diseño, pero sus afirmaciones y silencios nos permiten sacar algunas conclusiones sobre su público. PALABRAS CLAVE: grúa, torno, polea, mecánica, mecanismos automáticos, Vitruvio, bronce, ­ ierro, fricción, Exeliktra, Knōdax, Cnodax, Khoinikis, Khelōnion, tornillo de agua, rueda de agua, h Paconius, Chersiphron, Metagenes.

168

Duncan Keenan-Jones, Ian Ruffell, Euan McGookin

INTRODUCTION1 ... πᾶν μέρος τῆς μηχανικῆς ἐν αὐτῇ τῇ αὐτομα­ τοποιητικῇ παραλαμβανόμενον διὰ τῶν κατὰ μέρος ἐν αὐτῇ ἐπιτελουμένων. ... every part of mechanics is taken over in the very practice of automata-making, through the individual parts which are executed in it. (Hero, Aut. 1.1)

In his opening statement of On the making of automata (Περί αὐτοματοποιητικῆς, peri automa­ topoiētikēs), Hero makes a number of claims to present the making of automata as a worthy object of study. As well as the striking effects that could be achieved (see especially Tybjerg 2003) the making of automata required machines from all branches of ancient mechanics and a “sophisticated variety of craftsmanship” (τὸ ποικιλὸν τῆς δημιουργίας Aut. 1.1). This claim is, no doubt, one way in which he seeks to aggrandise his work and his own status, but it is equally clearly not without some force. Even a cursory glance at the treatise shows explicit comparisons with specialised devices used in artillery (Aut. 2.6, 13.9) and obvious use of hydraulic elements, to name but two aspects of ancient engineering. This paper does not set out to map fully the positioning of the peri automatopoiētikēs in relation to the entirety of ­ancient mechanics, but to explore Hero’s claim by focussing on one area of comparison: the relationship between Hero’s mobile automaton, particularly its drive elements, and some machines used in Hellenistic and Roman construction, namely the cranes and transport devices used to move ancient stone architectural elements. Hero, writing in the first century CE, discusses the preparation of a mobile (ὑπάγον, hupagon) and a stationary (στατόν, staton) automaton, both of which are powered by a falling weight that drives a number of axles for different purposes. Either can be tailored to the user’s own preferences or preferred story (Aut. 1.2, 1.8). Of the two devices, the mobile automaton is, Hero claims (again not without reason), the more sophis­ticated and difficult device to build and it is treated at considerably greater length in the first 1   Abbreviations of authors and texts are taken from LSJ and OLD. Readers should note the abbreviation of Hero’s Pneumatica as ‘Spir.’ Philo’s and Hero’s Belopoeica are cited with Diels-Schramm section numbers and Thévenot and Wescher pages respectively.

Anejos de AEspA LXXVII

book (Aut. 1-19) of the treatise. The automaton takes the form of a model of a shrine set on top of four columns. Its basic function is to move for a distance, halt, perform various displays involving the shrine and its associated figurines, and then move back. Hero pays particular attention to the drive system that creates the motion and presents a variety of configurations that offer different patterns of movement (rectangular, circular, ‘snake-like’), which in turn provide further insights into how Hero understood the mechanics of his drive. We shall be arguing that the power mechan­ ism of the mobile automaton, and particularly the mechanism causing the motion of the mobile automaton, is in fact the exact opposite of the cranes described by Hero and other writers such as Vitruvius. We discuss how this similarity is reflected in Hero’s choice of terminology in these two applications and we explore the relationship further through modelling the respective loadbearing capacities of the two devices. We go on to examine in more detail the bearings used in both devices. The bearings were crucial to the success of the automaton, and to many other mechanical devices such as cranes and water-lifting machines. The exact meaning of the terms used for these bearings, and the differences between them, has not, to our knowledge, been systematically invest­ igated. As we shall see, Hero’s choice of bearing in the automata displays clear understanding of the practical challenges presented by these devices, but also displays an intriguing mix of affinities across specialisms within ancient engin­ eering. CRANES AND WINDLASSES Roman Cranes The basic principles of Roman cranes are wellknown, both from detailed depictions such as the relief from the tomb of the Haterii (fig.  1A) – which inspired the full-scale (10.4 m-long mast) reproduction (fig. 1B) described by MeighörnerSchardt (1990) – and from ancient texts. A good starting point is Hero’s Mechanica, or ‘book on the lifting of heavy objects’ as it is called in the Arabic translation (Drachmann 1963: 21), the only version in which it is wholly preserved. Although this presents some challenges for direct comparison of Hero’s descriptions of cranes and

Anejos de AEspA LXXVII

Taking a bearing on Hero’s Anti-Crane and its Un-windlass...

169

Fig. 1. The relief from the tomb of the Haterii (A De Agostini Picture Library / G. Nimatallah/Bridgeman Images. Used with permission) and two modern reconstructions of it: B. A near full scale physical reconstruction (described in Meighörner-Schardt 1990). (Base Image: Qualle, commons.wikimedia.org/wiki/File:Roemerkran.jpg, used under a Creative Commons AttributionShare Alike 3.0 Unported license creativecommons.org/licenses/by-sa/3.0/deed.en). C. SolidWorks model with fixtures and loadings for static Finite Elements Analysis (FEA). D. Simulated Von Mises stresses and E. simulated displacements (exaggerated in the diagram) of the SolidWorks model under use.

170

Duncan Keenan-Jones, Ian Ruffell, Euan McGookin

automata, this difficulty is offset by the transliteration of some key Greek terms2 into Arabic by the translator, Qustā ibn Lūqā (Drachmann 1963: 21), and the preservation of numerous quotations of Hero’s lost Greek text in Pappus of Alexandria. Two of the five means laid out by Hero (Mech. 2.1) for lifting heavy weights, the windlass and the  compound pulley, were applied to ­multiply the input force to the crane, while simultaneously increasing the distance over which this input for­ce had to be applied. The windlass (ἄξων ἐν περιτροχίῳ, axōn en peritrokhiōi, Pappus 8.52) wound heavy ropes (ὅπλα, hopla), around an axle (ἄξων, axōn) to lift a very large weight (μεγάλα βάρη, megala barē) using a smaller force (ἐλάσσονι βίαι, elassoni biai), as shown in Figure 1. The input force was applied to the larger diameter circle  (περιτροχίον, peritrokhion) in this case the treadmill, in order to produce a greater force over a smaller distance around the smaller diameter circle (the axle), and thus raise a heavier weight (Hero, Mech. 2.10). The compound pulley (πολύσπαστος, polyspastus) applied to the rope produced the same effect. The form of both mach­ines, and the method of their construction, is carefully described, e.g. for the windlass: ... ξύλον δεῖ λαβεῖν εὔτονον τετράγωνον (καθάπερ δοκίδα) καὶ τούτου τὰ ἄκρα σιμώσαντα στρογγύλα ποιῆσαι καὶ χοινικίδας περιθεῖναι χαλκᾶς συναραρυίας τῶι ἄξονι, ὥστε ἐμβληθείσας αὐτὰς εἰς τρήματα στρογγύλα ἐν ἀκινήτωι τινὶ πήγματι εὐλύτως στρέφεσθαι τῶν τρημάτων τριβεῖς χαλκοῦς ἐχόντων ὑποκειμένους ταῖς χοινικίσι· καλεῖται δὲ τὸ εἰρημένον ξύλον ἄξων. περὶ δὲ μέσον τὸν ἄξονα περιτίθεται τύμπανον ἔχον τρῆμα τετράγωνον ἁρμοστὸν τῶι ἄξονι, ὥστε ἅμα στρέφεσθαι τόν τε ἄξονα καὶ τὸ περιτρόχιον. Ἡ μὲν οὖν κατασκευὴ δεδήλωται, χρεία δ’ ἐστὶν ἡ μέλλουσα λέγεσθαι. ὅταν γὰρ βουλώμεθα μεγάλα βάρη κινεῖν ἐλάσσονι βίαι, τὰ ἐκδεδεμένα ἐκ τοῦ βάρους ὅπλα περιθέντες περὶ τὰ σεσιμωμένα τοῦ ἄξονος, καὶ ἐμβαλόντες σκυτάλας εἰς τὰ ἐν τῶι περιτροχίωι τρήματα, ἐπιστρέφομεν τὸ περιτρόχιον κατάγοντες τὰς σκυτάλας, καὶ οὕτως εὐκόπως κινηθήσεται τὸ βάρος ὑπὸ ἐλάσσονος δυνάμεως τῶν ὅπλων περὶ τὸν ἄξονα ἐπειλουμένων [ἢ καὶ διαμηρυομένων ὑπό τινος πρὸς τὸ μὴ ἅπαν τὸ ὅπλον περικεῖσθαι τῶι ἄξονι]. It is necessary to take a hard, square piece of wood (like a beam), smooth its ends and make 2  E.g. περιτρόχιον (bari-trakīn, 2.1), μάγγανον (manġanum, 2.3), τύλος (tûlus, 2.5, 18, 19), ὑπομόχλιον (ibûmahliûn, 2.2, ˘ 9,  29), The Arabic text with parallel German translation is available in the standard Teubner edition of Hero by Schmidt et al.

Anejos de AEspA LXXVII

them rounded and to put bronze bands (kho­nikides) around them, close-fitted to the axle (fig. 2), so that when they are put into round holes in some rigid frame they turn easily. The holes have bronze liners (tribeis) set underneath the bands (khoinikides). Around the middle of the axle should be set a drum with square hole to fit the axle, so that the axle and windlass (peritrokhion) turn together.

So the construction has been made clear; its use is as follows. For when we want to move large weights by a smaller force, we put the ropes connected to the weight around the round grooves of the axle and put rods into the holes in the windlass, and we turn the windlass by pulling down the rods. And thus the weight will move easily by means of a lesser force, as the ropes wind around the axle [or are in fact arranged in loops by someone, so that all the rope does not lie on the axle]. Hero, Mechanica 2.1 [=Pappus 8.53]

Finite Element Analysis (FEA) of Crane Load Capacity To estimate just how heavy this weight could be, we conducted a static linear Finite Element Analysis (FEA) by using the Simulation feature in the computer-aided design (CAD) package SolidWorks.3 The model for the FEA (fig. 1C) was based on the crane from the tomb of the Haterii (now in the Museo Gregoriano Profano in the Vatican, fig. 1A).4 The load was borne by the rope, by the mast of the crane and also by the treadmill windlass. Too much of the windlass was hidden by the treadmill to model it effectively, so it was not considered. The load borne by the rope was applied to the mast in order to determine whether the strength of the rope or the mast was the limiting factor for the load capacity of the crane.   Premium 2015 x64 edition SP 2.1   Dimensions were measured on an image of the relief in ImageJ, software (US National Institutes of Health, https:// imagej.nih.gov/ij/) using a mast length of 16m (ultimately derived by assuming the men in the treadmill to be 1.6m high) to set the scale (O’Connor 1993). The scale, however, is clearly variable throughout the image (Meighörner-Schardt 1990; O’Connor 1993) so the dimensions must be viewed as plausible approximations. The dimensions measured did not yield any convincing multiples of Roman or Oscan feet. The angle between the two beams making up the mast was assumed to  be 17 degrees, following Meighörner-Schardt (1990: 50, Abb. 4), since this angle is not visible in the side-on view shown on the relief. 3 4

Anejos de AEspA LXXVII

Taking a bearing on Hero’s Anti-Crane and its Un-windlass...

Fig. 2. The stages in the creation of an axle for a windlass according to Hero, Mech. 2.1.

The tensile strength of the rope was assumed to be 16 MPa, equal to that of modern Egyptian palm fibre rope, still made using ancient techniques (Cotterell and Kamminga 1992: 225). Eight measurements of the rope carrying the load depicted on the relief gave an average diameter of 9 cm. The diameter was quite variable, with a ­total range of 7-12 cm, and was probably the item of the crane least to scale. While ropes of this size are known from antiquity, such as 6 cm (O’Connor 1993: 48) and 9 cm (UC30841 in the Petrie Museum, University College London) Egyptian ropes, using these in a compound pulley would create problems with friction (Adam 1994: 47). We assumed a rope diameter of 5 cm. Since stress is equal to force divided by area, in this case the cross-sectional area of the rope (Cotterell and Kamminga 1992: 66-67), a 5 cm diameter rope could bear a maximum load of 3.2 tonnes. This is a little weaker than the minimum values for modern natural fibre ropes (O’Connor 1993: 48). This force (31.4 kN in total, fig. 1C) was applied to the top ends of the masts. Gravity was also applied throughout the model. Wood is a complex material to simulate since its properties vary considerably within a single piece due to changes in growing conditions over time.5 The wood was simulated as a linear elastic6 5  Kretschmann, D. E. 2010: “Chapter 5 Mechanical Properties of Wood.”, General Technical Report FPLGTR-190. U.S. Department of Agriculture, Forest Service, Forest Products Laboratory, Madison, WI. 5-1 – 5-46. http:// www.fpl.fs.fed.us/documnts/fplgtr/fplgtr190/chapter_05. pdfhkcUUgCjw&bvm=bv.94911696,d.d24: 5-1. 6  Pirinen, M. 2014: “Ductility of Wood and Wood Members Connected with Mechanical Fasteners.” Masters of Science thesis, Aalto University, p. 22. https://aaltodoc.aalto. fi/bitstream/handle/123456789/13138/master_Pirinen_ Matti_2014.pdf.

171

and orthotropic6 material. The Mohr-Coulomb failure criterion was used since most wood failures are brittle in nature. The long axes of wooden elements in the crane were assumed to be parallel to the grain, since this would simplify manufacture, and would be essential given the lengths of the two long masts and the cross-beam. This would also result in most of the force acting parallel to the grain, the strongest axis of the wood, in the beams when the crane was near vertical (fig. 1C). The strength of the beams would have reduced the closer the beam got to the horizontal and the more the forces acted perpendicular to grain, where the wood was much weaker. There is little data on wood selection for ancient cranes, but timbers considered strong and used as structural timbers in buildings (mentioned by Vitruvius, 2.9.5-17, Veal in Press) and in ships (Allevato et al. 2010) included evergreen Oak (Quercus, especially), alpine fir (Abies), spruce (Picea) or larch (Larix). Oak was chosen for the simulation, and European Oak (Quercus ilex) was modelled by North American White Oak (Quercus alba).7 Given these modelling and data issues, the results of the Finite Element Analysis should be considered as indicative but not definitive. The mast was simulated in a static position, which would occur when the load was being moved vertically but not laterally, at the angle of the mast shown on the relief (80o). The beam at the base of the crane was assumed to be resting on the ground, and was modelled as a hinge fixed in place (Meighörner-Schardt 1990: 53, Abb. 4). The different wooden elements (bodies) making up the crane were modelled with bonded contact (i.e. as if welded together). On the relief, at the top-most brace on the load side, there are four opposing guy ropes, one front and one rear on each beam of the mast. At the second brace from the top, there is one rear brace on each of the two beams. The arrangement at the fourth brace is irregular and unlikely: only the beam closest to the viewer has a guy rope. Both Meighörner-Schardt (1990) and Adam (1994: 46, fig. 95) reconstruct a symmetrical arrangement of guy ropes. Given this irregularity, and the fact that ropes cannot be modelled in SolidWorks Simulation, the guy ropes were 7   Due to the paucity of data on the mechanical properties of European natives of these genera in all three directions (axial, radial and transverse), data could only be found for members of these genera native to North America (Kretschmann 2010: see n. 6), which were used in the simulation.

172

Duncan Keenan-Jones, Ian Ruffell, Euan McGookin

Anejos de AEspA LXXVII

modelled as a constraint where the beams meet, on the meeting edges, allowing m ­ otion only vertically and not laterally. The mesh elements were 0.22681787 m and there were 4 Jacobian points. The h-adaptive method was used, where potential weak points identified in an initial simulation are reinvestigated with a higher-resolution mesh. Results of the FEA are shown in fig. 1D-F. The simulated Von Mises stresses reached a maximum of 1.8 × 106 N/m2 just below the guy ropes (fig. 1D) where the mast beams are almost at their thinnest and weakest. As a result, this area had the minimum Factor of Safety (FOS) within the model (3.74512, fig. 1F) and was the most likely to fail. Since this FOS was still greater than 1, the FEA suggests that the estimated load strength of the rope could be borne by the mast and that the rope (or the windlass) was the factor limiting the load capacity of the crane. In addition, in reality the presence of lower guy ropes would have spread the stresses at the top of the mast further down, where the beams are thicker and stronger. Another area of high stress was where the masts were joined to the lower cross-beam. This join would have needed to be well constructed. The mast would have been displaced (bent) towards the load by a maximum of 1.2 mm in the centre of the mast (fig. 1E). This shows that the lower guy ropes were well-positioned, since they would have pulled the mast away from the load to counteract the displacement in the upper part of the crane. Hero’s Anti-Crane Three of Hero’s means for lifting heavy weights in the Mechanica, the windlass, the pulley and the screw, were used by Hero within his automaton. The windlass was used in the drive mechanism for the mobile automaton, but the terminology is different, reflecting its opposite function. In Hero’s drive mechanism, a very light weight unwound a cord (σπάρτος, spartos) as it descended to turn a pair of wheels (fig. 3). The drive mechanism of the automaton necessarily sought to solve the opposite problem to the crane. The automaton was built to be as light as possible (Aut. 2.8). The weight of our CAD model of the mobile automaton, as estimated by SolidWorks, is only 24-27 kg (depending on the configuration adopted), plus up to 8.5 kg of millet or mustard seed used to slow the falling weight, the falling weight required to move it even less, a far cry

Fig. 3. Schematic representation of Hero’s mobile automaton and its drive mechanism.

from the 3.2 tonne lifting capacity of our modelled crane. Hence, in the automaton the force was applied to the smaller circle, producing a lesser force applied over a greater distance on the larger

Anejos de AEspA LXXVII

Taking a bearing on Hero’s Anti-Crane and its Un-windlass...

circle, the wheels of the automaton, thus increasing its range and keeping speeds safe and manageable. The automaton’s single-sheaved pulley (τροχίλος, trokhilus) served only to transfer the direction of the force (Aut. 17.2): there was no multiplication of the input force as in the crane. There was instead an interest in extending the automaton’s range, by adding an additional drum and axle (Aut. 18). The word for rope in the automation, spartos, can often mean a small rope (σπαρτίον, spartion) or cord (LSJ s.v. σπάρτος A.I.1) especially when used in the feminine, as it is always by Hero (cf. also σπάρτη). It seems Hero is differentiating the smaller spartos of the automaton, carrying a smaller force, from the larger ropes (ὅπλα, hopla) of the crane. This very different conception of the windlass is probably reflected in the use of a different and, outside Hero, almost unattested term for smaller circle on the axle of the automation: exeliktra (ἐξελίκτρα, fig. 4). This object is never defined, glossed or described, in stark contrast to the detailed instructions for constructing the means of lifting described in the Mechanica (reproduced above). This reflects the greater amount of know­ ledge, perhaps of “every part of mechanics”, Hero assumes of the reader of the peri automatopoiētikēs compared to the reader of the Mechanica. Exelik­ tra is clearly derived from the verb ἐξελίσσω (exelissō), meaning to unroll, and thus emphasizes the unwinding of the cord (like the English term bobbin). Hero also uses the term in the Pneumatica (Spir. 2.32), where it again means a bobbin made to rotate by the unwinding of a cord. Thus the exeliktra is actually an un-windlass. Hero’s automaton is notable in that its motion could be programmed through the arrangement of the cords. Changing the direction of winding of the cord on the exeliktra (after securing the cord around a knob) resulted in a change in the direction of motion (Aut. 6.1-3). Insertion of lengths of ­ halasmata), gummed together slack (χαλάσματα, k in loops (μηρύματα, mērumata) for storage, resulted in pauses in the motion (Aut. 2.10-11). Philo uses a related term, a wooden exeliktron (ἐξέλικτρον, Bel. 38 [p. 67]) in a passage that uses much of the same terminology as Hero’s description of the use of the exeliktra. καὶ οὕτως χάλασμα λαβόντων τῶν τόνων τούς τε  ἀγκῶνας ἐξαιρεῖν καὶ τὰ τῶν τόνων μηρύματα περιελεῖν ὄντα ἄφθαρτα καὶ ἀσινῆ καὶ λιπάναντας εἰς ἐξέλικτρον ξύλινον συντιθέναι.

173

Fig. 4. A reconstruction of the axle and exeliktra assembly of Hero’s mobile automaton.

And so when the sinews go slack, take the arms out and pull off the coils of sinew if they are undamaged and without kinks, oil them and assemble them on the wooden bobbin (exeliktron).

The greasing may have been accomplished by soaking in oil (Bel. 27 [p. 61 Th]). The use of this object is not immediately clear. The cords may have been wound around the exeliktron for storage while they absorb the oil and regenerate, or to fac­ ilitate their reinstallation into the artillery device. Either way, the sinews had just been unwound, and will have needed to be unwound  again from the exeliktron prior to, or during, reinstallation. One or both of these unwindings is probably the reason for naming the device an exeliktron. The only o ­ ther instance of exeliktron that we have found is in a list of women’s hair-care equipment (PCorn. 29.3) Hero is clearly heavily influenced by Philo, both in the Pneumatica and above all in the automata treatise where he explicitly draws on Philo in relation to the stationary automaton (Aut. 21). It is not unlikely that this terminology is taken over from his predecessor, albeit with a change of gender. This may not be particularly significant, since it is not always clear why Hero switches between different forms (e.g. diminutive and nondiminutive forms) in the peri automatopoiētikēs. It is even possible that the use of an exeliktra or ex­ eliktron in mobile automata goes back to Philo in the third century BCE, but that is beyond the scope of our current evidence. What is clear is that Hero is working in an existing tradition of mobile automata (1.2, 1.7), although he acknow­ ledges a much greater debt to specific predecessors when he turns to the stationary variety.

174

Duncan Keenan-Jones, Ian Ruffell, Euan McGookin

Another ancient un-windlass is recorded that offers suggestive parallels to the operation, if not the name, of Hero’s exeliktra. Some kind of continuity of thought in relation to this sort of device seems probable, but it also draws attention to some problems which Hero’s mobile automaton seems designed to overcome. In the first century BC, a certain Paconius, apparently inspired by Metagenes (see below), attempted to transport a base for the colossal statue of Apollo in the Temple of Artemis at Ephesus by encasing the ends in wheels roughly twice as large as the base (Vitr. 10.2.13-14). Then he joined the wheels with rods (or spindles, fusi) placed around the base and wound (involvit) a rope around them. When pulled by oxen, the rope unwound (explicaretur), causing the base and wheels to rotate and gaining a mechanical advantage of 1:2 (Coulton 1977: 143) by its operation as an (un-)windlass. Unfortunately the wheel did not travel straight, but swerved to one side, presumably because the loop being unwound was not perfectly centred on the base. Hero’s drive mechanism avoided the worst of this defect of Paconius’ machine since, in order to convey the automaton, the exeliktra had to be mounted on bearings (discussed below) fixed in the sides of the case. There was still an issue, however, since an increasing amount of the force used to propel the automaton was lost the more the rope departed from the centre of the exelik­ tra, and the angle the cord was inclined the vertical increased. The loss in driving force due to this inclination is represented by the following relationship: FL = FD (1 − cosΘ) (1) Here Θ is the cord inclination angle, FD is the total drive force and FL is the force loss (see fig. 4). It is very likely that the exeliktra was intended to keep the cord confined within a narrow space located at the centre of the axle, where this inclination angle (Θ1) and the resulting loss of force would be minimised. Again, we could speculate that Hero had learnt from Paconius’ error, or other similar errors. The automaton also shared with Paconius’ un-windlass the problem that the range before rewinding of the rope was required was limited by the length of the rope multiplied by the mechanical advantage (on this problem for Paconius see Coulton 1977: 143).

Anejos de AEspA LXXVII

BEARINGS Knōdax (Κνώδαξ) Bearings were crucial to the success of the automata (2.3). Two forms of mountings for rotating parts are set out initially by Hero: a knōdax (κνώδαξ) rotating within an [em]pyelis ([ἐμ] πυελίς), used for mounting wheeled axles, and an axle surrounded by a khoinikis (χοινικίς), used for figurines (2.3). In fact, Hero gives no explicit example of the latter in the treatise, but he does employ khoinikides for the wheels used in achieving snake-like motion, the only wheels to rotate independently of their axle (11.2). Both the knōdax and khoinikis are journal bearings, where a cylindrical shaft sits inside a cyl­ indrical shell, one of which rotates relative to the other. The bearings differ primarily in size and method of attachment to the axle. Hero stresses that smooth rotation is critical for the automaton to function effectively, so all these bearings are lubricated with oil (2.4). Other types of bearing are known in the period. Trunnion-mounted bronze balls, roughly 5 cm in diameter, are known from a bearing in the ships recovered at Nemi in Italy (Ucelli 1940 as cited in Dowson and Hamrock 1981: 8-10). Although roughly contempor­ ary with Hero, these bearings were not employed by him, despite this drive to reduce friction. If he did know of them, they may have been difficult to manufacture in the smaller sizes required for the automaton. If similarly made of bronze, their small size may also have accentuated the melting problems associated with bronze described below. The knōdax is an unusual term, although not as rare as the ἐξελίκτρα, and seems particularly associated with technical, especially mechanical, contexts in a broad sense. Lexicographic works are not particularly helpful: Hesychius (κ3157) defines the knōdax in vague terms as ‘the centre of an axle’, but more helpfully suggests that it is a point (γνώμων). As well as other (elsewhere un­ attested) uses in the singular, he also suggests that in the plural it refers to the bellows of bagpipes (κ3155). So the knōdax denotes a pin, usually, but not exclusively, one of a pair, on which an axle or similar shaft rotated. In the case of the mobile automaton, it is the ἐξελίκτρα’s axle which rotated on or around them, as well as other wheeled axles in different configurations. Closely related is the use in a rotating axle and drum used to extend the range (Aut. 18). There is one instance of a single

Anejos de AEspA LXXVII

Taking a bearing on Hero’s Anti-Crane and its Un-windlass...

knōdax used for a rotating figure, the Nike on the apex (Aut. 13.7-8). In Hero’s Pneumatica, the use of knōdakes is somewhat different. The Pneumatica has vertical, rotating pins extending to the floor of a device, as part of a hinge or similar, each time described as ‘turning on’ a knōdax or knodakes (στρεφέσθωσαν ἐν κνωδακίοις, Spir. 1.38; ἐν κνώδακι στρεφόμενος, Spir. 2.3). “Around knōdakes” (περὶ κνώδακα) at 13.8 seems to describe a very similar configuration; the prepositions are used apparently interchangeably for the same movement at Aut. 11.810. But there are also fixed knōdakes on which spherical vessels are hung and around which they rotate (Spir. 2.4), a process for which the otherwise unattested verb κνωδακίζω is used. In the case of Hero’s famous description of a steam engine (aeolipile), rotation takes place around one fixed pipe and one fixed knōdax (Spir. 2.11). Hero also discusses a wind turbine which this time involves an orthodox axle, but again with the axle rotating around iron knōdakes in a mobile frame (κινούμενος περὶ κνώδακας σιδηροῦς ἐν πήγματι δυναμένῳ μετάγεσθαι, Spir. 1.43). It is unclear whether these knōdakes are fixed or rotating. Although the latter use comes close to the technique adopted in the mobile automaton, there is still an important distinction, which is that this axle is not being used to drive wheels and effect propulsion. This understanding of the knōdax as a fixed axis of rotation also seems to be reflected in non-mechanical texts, such as Sextus Empiricus (M. 10.51), who includes spheres (probably celestial)8 turning on knōdakes along with other instances such as the potter’s wheel (and indeed axles) where (rotational) motion takes place, but there is no change of position. This is some distance from Hero’s use of the knōdax in the drive of the mobile automaton. For that, we need to look elsewhere, namely to the techniques of Greek and Roman construction. Cnodax is used by Vitruvius (10.2.11-12) to describe the shafts upon which large stone architectural elements rotated during their transport by Chersiphron and his son Metagenes during the

8   See also S. E. M. 10.93, with a similar list; the sphere is also mentioned again at 10.52. An imaginary knōdax is also used in a curse (SEG 47.1291.27) for a god occupying the ‘pole’ of heaven (ὁ τὸν κνώδακα τοῦ οὐ(ρα)νοῦ κατέχων); the term κνωδακοφύλακες appears in the magical papyri with apparently the same sense (Preisendanz no. 4). The only other notable appearance of knōdax is anatomical, referring to rotation around a fixed pivot ([Galen] 14.720 and 723).

175

construction of the temple of Artemis at Ephesus. The cnodax was inserted into the end of columns, or of architraves housed in large wooden wheels, in order to mount them on a frame which was used to pull them as they were rolled from the quarry to the construction site. Cnodax is only found in Latin in these passages from Vitruvius (Callebat and Fleury 1995: 318) and he almost certainly derived it from knōdax used in the treatise published by Chersiphron and Metagenes (Vitr. 7.pr.12). Given that the temple was completed c. 560 BC (Coulton 1977: 143), Chersiphron and Metagenes’ treatise would be considerably earlier than any of our extant technical works, but there is some other evidence for the use of this bearing as a fixed axis of rotation at such an early date, namely for the rotating kyrbeis on which early Athenian laws were inscribed (Aristophanes of Byzantium, fr.76 = Et. Gud. p. 355, Et. Magn., p. 547, Suda κ 2745, Stroud 1979). Marks of transport in the manner of Chersiphron and Metagenes also survive on the ends of column drums and architraves, as well as on the sides of the architraves, of two temples at Selinous dated c. 530-460 BC (Koldewey and Puchstein 1899: 119-120, 125; Coulton 1977: 143-144). Evidence for κνώδαkες in surviving columns would appear similar to that for dowels or for the iron reinforcing found in weaker columns (see Amici, this volume),9 and there may be further examples that have not been recognized. Knōdakes are also found in water screws (κοχ[λ(ιῶν)], P. Lond 3.1177.233-4), otherwise known as Archimedes screws. This passage records payments made to a certain Mareis, a locksmith (kleidopoios, κλειδοποιός), which indicates that these are both specialised and precise pieces of equipment. Exact comparisons in terms of cost are tricky, given that quantities for other items in the papyrus are not always specified, but 6.5 minas and 26 drachmas for two knōdakes, two rings (krikoi) and an unspecified number of, but (in context) presumably two, sockets is significant. Vitruvius (10.6.3) calls the shafts inserted into a water screw styli, and this provides further support that the term knōdax specified a pointed pin. Judging from Vitruvius and the water screw, it appears that the knōdax was a pin thinner than, and inserted into, an axle (fig. 5). This then ­rotated

9

  We owe this suggestion to Lisa Fentress.

176

Duncan Keenan-Jones, Ian Ruffell, Euan McGookin

Anejos de AEspA LXXVII

easily by means of this arrangement. Moreover, the wheels are not hollow but project. So, given that the sockets (holmiskoi) move around knōdakes, and the wheels are raised, with the axles running very freely, the stretching will be unimpeded, even if the rope is pulled by hand. Oribasius 49.23.19-21

Fig. 5. A reconstruction of the bearing arrangement of the axle of Hero’s mobile automaton.

within a socket. In the peri automatopoiētikēs, knōdakes rotate in empyelides (ἐμπυελίδες, Aut. 2.3), empyelidiα (ἐμπυελιδια, Aut. 10.1 [twice], 11.9 and especially 26.2) or simply a pyelis (πυελίς, Aut. 5.3), although in that case it is specified as being ‘in the sides’, ἐν τοίχοις. Outside of Hero, we see other terms used: holmos (ὅλμος, SB 18 13873 = SB 10 10739) or its diminutive holmiskos (ὁλμ[ίσ]κος, P Lond 3.1177) both meaning socket, or armilla (Vitr. 10.11) meaning ring, but never a khoinikis, according to the surviving evidence. There is a little more information on how knōdakes may be attached in machines in Oribasius’ description of a stretching machine (a medical rack) the trispaston (or trispastus, cf. Vitr. 10.2.3). This account explicitly refers to a change in technique from rotation around a fixed pivot towards rotation on axles socketed in the walls of the device (and presumably using a standard kh­ oinikis bearing, as discussed below). Ἀριστίων δ’ ὁ τοῦ Πασικράτους υἱὸς ὑπήντησε τῷ  πατρὶ ἠγνοηκότι τὴν ἀρχαίαν τοῦ ὀργάνου κατασκευήν· ἐν γὰρ τῇ ἀρχαίᾳ, φησίν, ὀργανοποιίᾳ οὐκ ἐν κοιλότησι πλευρῶν ἐκινοῦντο οἱ ἄξονες, ἀλλὰ περὶ κνώδακας προσηλωμένους τοῖς πλευροῖς κοιλότητας ἔχοντας ἑλικοειδεῖς, ὥσπερ ἔστι θεάσασθαι τὸ τοιοῦτον γινόμενον ἐπὶ τῶν ὑδραγωγῶν ὀργάνων, διὰ τὴν κατασκευὴν ῥᾳδίως στρεφομένων. (20) οὐδὲ μὴν οἱ τροχοὶ κοῖλοί εἰσιν, ἀλλ’ ὑπερέχοντες. (21) ἐπείπερ οὖν ὁλμίσκοις περὶ κνώδακας κινοῦνται, καὶ οἱ τροχοί εἰσι μετέωροι, εὐχερέστατα τῶν ἀξόνων κινουμένων, ἀνεμπόδιστοι αἱ τάσεις ἔσονται, κἂν χειρὶ ἕλκηται ὁ κάλος· Aristion the son of Pasikrates disagreed with his father, who was unaware of the old arrangement of the device. For in ancient engineering, he says, axles did not move in holes in the sides [sc. of devices], but around knōdakes that were fixed in the sides with threaded holes, just as it is possible to see happening with water-lifting devices, which turn

Here, if the text is reliable, it seems that the wheels themselves are socketed (with the hol­ miskoi) and rotate around the knōdakes, which appear to be screwed into the walls of devices.10 They are thus analogous to the fixed knōdakes discussed in the Pneumatica and to another unique machine in which knōdakes are used, namely the mechanism for a repeating catapult described by Philo of Byzantium, where a crucial component is a cylinder moving on knōdakes (ὀχούμενον ἐν κνώδαξι, Bel. 75.56), which are not sockets, despite LSJ. By contrast, the sockets in Hero’s automaton are set in the walls of the device, which is consistent with Oribasius’ account of the later method (although he may be referring to axles used without any intervening knōdax, but instead with a khoinikis, as discussed below). ­Hero’s usage is thus consistent with the practice in moving architectural members, but reverses the configuration of knōdakes preferred in other ­machines. Oribasius’ account (or, rather, that of his source) also indicates that for this machine knōdakes should be made of iron. A papyrus refers to old knōdakia [κνωδ(άκια)] being handed over to a khalkeus, but that may denote a generic metalworker rather than a bronzesmith specifically (SB 18 13873 = SB 10 10739 cf Pintaudi 1985: 92 and Hero Bel. 21 [p. 98], where iron is finished in the khalkeia). Certainly Hero specifies iron for two different devices (Aut. 2.3, Spir. 1.43); he also specifies iron for the sockets (Aut. 2.3), as does Vitruvius (10.6.3) for the sockets (foramina) supporting the pins (styli) in water screws, which is supported by archaeological evidence (Oleson 1984: 183-184, 221, 250, 295). Iron was proba10  The ἔχοντας is awkward and suggests that the knōdakes have screw-like sockets, which could conceivably allow them to be fastened to the sides, but we are taking it more loosely to refer to their relationship to the sockets into which they are screwed. Habermann (2000: 232 n.º 471) prefers Bussemaker and Daremberg’s reading κοιλότητα ἔχοντες ὁρμισκοειδῆ (their apparatus only indicates that they have emended ἔχοντας): thus (‘[axles] ... with a socket-like hole’). This would indicate a similar wall-mounting for the knōdakes but the tautology and coinage of ὁλμισκοειδής do not seem very plausible.

Anejos de AEspA LXXVII

Taking a bearing on Hero’s Anti-Crane and its Un-windlass...

bly  preferred to bronze due to the thinness of the  knōdax. Iron was more complex to smelt than  bronze and had to be forged rather than cast in Hero’s day (Craddock 2008: 107-109), but quenched iron was much harder than the hardest bronze (Rehder 1992: 42), and was also less dense (Cotterell and Kamminga 1992: 68, Table 3.1). Hence, Philo (Bel. 23, 33 [pp. 60, 65]) recommends the use of wrought iron for thin parts in artillery devices, since defects in thin bronze items rendered them liable to shatter, while Hero (Bel. 21 [p. 98]) specified well-forged pure iron for small levers (epizygides, mentioned below) subject to great forces in Roman artillery. Similarly, iron was always preferred for architectural clamps, dowels and structural elements (Cooper 2008: 240-243). Another advantage may have been iron’s higher melting point, making it less likely to soften, deform, or fail through repeated friction-based annealing. Deformation would lead to asymmetry in the rotation of the bearing and increased contactinduced friction. This, in turn, would cause more deformation in softer metals and alloys and quickly worsen the performance of the bearing. Friction, indeed, seems to be a repeated concern in the decision to deploy knōdakes as bearings. Thus Hero prescribes the materials used for  knōdakes and empyelides (iron), the need for smoothness in their manufacture and the importance of lubrication. To these we may add the small diameter and probable pointed end of the bearing. These would have resulted in a small reduction in the amount of friction in the bearings through reduction of the contact surface area within the bearing socket. Also, the radius of the bearing reduces as it tapers to a point, which causes a reduction in the friction moment generated by the rotation of the axle. In two cases, Hero emphasises the speed required from such devices: the wind turbine (1.43) and the device to create bird noises (ταχέως στρέφεσθαι, Spir. 2.4). We may speculate that something similar may also be required in the repeating catapult, although that is not explicit (Ph., Bel. 56 [p. 75]). In the case of the manual force needed for the trispaston, Aristion’s view was that the knōdakes made the machine sufficiently easy to turn (εὐχερέστατα ... κινουμένων and ἀνεμπόδιστοι 49.23.21) where his father Pasikrates had opted to multiply the force by adding pulleys (which Oribasius’ source – perhaps Heli­odorus argues is more effective). Hero appears to have friction in mind in his discussions of different configurations for ‘snake-

177

like’ motion. Here, instead of the single exeliktrapowered axle mounted on knōdakes, the two wheels rotate independently around a fixed axle, with cords wrapped directly around each wheel’s khoinikis (here an extended nave or hub). The mat­erial of the khoinikis is unspecified (bronze at Aut. 2.3, albeit with a different application). Hero comments that these khoinikides do not rotate effectively (δυσχερῶς ἐπιστρέφονται, Aut. 11.8), preferring two independent axles mounted and rotating on knōdakes on the model of the usual drive mechanism (Aut. 11.8-10). Here, there is a clear belief in the greater efficiency of the knōdax bearing, which he argues should be used for all rotating elements in automata. The use of knōdakes in this context therefore has affinities with other machines, such as the wind turbine, repeating catapult or trispaston, where low friction has been a major operational requirement and where loads tended to be smaller. Even in the uses described by Vitruvius, the considerable forces on the knōdax would still have been significantly less than those in the crane eventually used to lift the stone. The column shaft, or the wooden wheel around the architrave, is in contact with the ground and bearing the weight of the stone. The knōdax only attached the stone element to the frame used to pull it along. This method was only used over a short distance of level ground (10.2.12) perhaps suggesting it was at the limit of the knōdax’s operational abilities. Khoinikis (Χοινικίς) Clearly, then, the choice to use the knōdax, whether in complex machines or in various construction contexts was, as far as the evidence ­allows, a choice closely associated with specific requirements and specialised applications. The khoinikis, by contrast, was a much more generalpurpose bearing. This is why it was the bearing used in Hero’s account of the windlass (Mech. 2.1, above) and it is the bearing to be seen in most technical contexts. Such a distinction is supported by textual, epigraphic and papyrological evidence, in all of which the term is much more common than the knōdax, and much more widely distributed, as well as by the limited archaeological evidence. In textual sources, khoinikis is a much more widely attested  term than knōdax, and in a wider range of

178

Duncan Keenan-Jones, Ian Ruffell, Euan McGookin

a­ uthors. Its association with axles was evidently well understood in general terms in the ancient world. Outside of technical contexts, there is a certain lack of clarity about the precise configuration of the khoinikides and their relation to either wheel or axle, but this in fact reflects both the broad range of contexts in which khoinikides were used. The primary meaning of khoinikis, and the meaning used by Hero in the peri automatopoiētikēs, is a cylindrical shell around a cylindrical shaft or similar object, one of which is rotating relative to the other. A wide range of machines are known to have used them. Most extensively documented are washers used in torsion springs that powered  Hellenistic and Roman artillery: bronze washers were placed at the top and bottom of bundles of sinew which passed through the washers and were kept under tension by iron levers (epizygides). In addition to technical treatises, espec­ially by Philo (Bel. esp. 10, 16, 24-25 [pp. 53, 57, 60-61 Th]) and Hero (Bel. esp. 9, 15, 20-2 [pp. 82-83, 91, 96-99]), a dozen examples survive from all over the Mediterranean world, dating from the third century BC to the second century AD.11 They are bronze cylindrical shells with a flange at one end for attachment to the main frame of catapult, which had been cast and then finished on a lathe. In some examples the cylinder widens as it approaches the flange, as do the flangeless washers depicted in medieval manuscripts of Hero’s Cheiroballistra (Baatz 1978: 12, fig. 10). The arch­ aeological finds agree with the picture from the written sources, except that Vitruvius (10.11.5) specifies an ovoid cross-section, or perhaps square (also shown in a relief from Pergamon [Marsden 1971: Pl. 3]), to allow greater room for the torsion spring (Baatz 1978: 16). In a machine of a very different type- the surveying instrument, dioptra – a khoinikis plays a role at the top of the pedestal or column on which the main apparatus rotates (Dioptr. 3) and as smaller washers (khoinikidia) in relation to plates adjusted by screws at either side of that apparatus (Dioptr. 4). Also attested in papyri are khoinikides in olive press (Chr. Wilck. 176.8) as well as unspecified water-lifting machines (PLouvre 1.11.15, 26; POxy. 2778.13-15). Wooden bearings in Roman­bilge pumps also contained internal 11  Baatz 1978: 3-7 and plate III; Wilkins 2003: 25-27; Wilkins, A. “The Xanten-Wardt Roman Torsion Catapult and Catapult Parts from Carlisle,” n.d. http://romanarmy.net/pdf/ The%20Xanten-Wardt%20and%20Carlisle%20catapult%20 finds.pdf

Anejos de AEspA LXXVII

c­ ylindrical bronze linings secured to the wood by means of flanges (Galli 1996). In epigraphic contexts, khoinikides refer mainly to sockets in which hinges rotate (IEleusis 1177.263-4 [IG II2 1672]; IG XI,2 165.11; IG XI.2 287.A103; IDélos 403.12). The most common use of all in manuscript sources, however, outside of technical texts, is as the ubiquitous term for either the hub or the bearing of axles in wheeled vehicles – carts and chariots. As such vehicles are not the main concern of technical texts, this use is not well represented there, the principal exceptions being Hero’s discussion of axle bearings for snake-like motion in Aut. 11 and in his description of an odometer, which is fitted to the wheel’s khoinikis (Dioptr. 34). Nor is there good documentary evidence, with the very clear exception of PCair. Masp. 3.67303, an agreement for the lease of a cart. That this is the regular term seems inescapable: it is frequently used to provide explanatory glosses either for more poetic or more generic forms in this connection (πλῆμναι, σύριγγες or ὀπαί: see, for example, scholia on Hom., Il. 5.726, A., Th. 153 and E., Hipp. 1234; Hesych. π 2563; Suda π 1756). As has been seen already in Hero’s treatment in the peri automatopoiētikēs, however, the relationship between khoinikides in the wheeled axle is not without its complications. In general, the term khoinikis does not specify which of the two, shaft or socket, is rotating. ­Often the shaft is rotating, similar to the way knōdakes are deployed in the automaton drive and in the rolling column. Thus the rotating axle is borne by a fixed khoinikis in Aut. 2.3, and the torsion spring is surrounded by a fixed washer (Hero, Bel. 20 [p. 96] Ph., Bel. 24, 30 [pp. 60, 63]) in earlier catapult designs (Marsden 1971: 53, n. 30). Sometimes the khoinikis itself is rotating while the shaft is fixed, such as the nave of a wheel (Aut. 11.3, 11.7), or in the dioptra (Dioptr. 3) or the serrated outer shell rotating around the fixed central pin of a rotary drill (Cels., Med. 8.3). The khoinikis is mostly independent of the movement of the axle. In Mech. 2.1, however, Hero uses the term in a slightly different sense, to mean the bronze cladding around a wooden axle where it touches the socket. Later catapult designs were similar, in that the khoinikis moved with the torsion spring (Marsden 1971: 53, n.º 30). Where wheeled vehicles are concerned, the further complication is that axles could either be fixed, so that wheels rotated around them (the configuration just described) or free to rotate, so that the entire

Anejos de AEspA LXXVII

Taking a bearing on Hero’s Anti-Crane and its Un-windlass...

assembly (wheels plus axles) rotated (Landels 2000: 180-181). This range of uses may explain why non-technical texts had such problems in defining the relationship of axles, wheels and khoini­ kides, such that many of our book sources refer to axles turning ‘in/on’ or ‘through’ khoinikides (which works for many axles, but not vehicular ones).12 Others seem to refer to fixed axles, where the central hub is still described as a khoinikis (Σ Hom., Il. 5.723 (ἐμβάλλω}; Σ Hom., Il. 5.726a, Σ Α., Th. 371gh (ἐντίθημι). With those qualifications, there appears to be considerable consistency in the terminology used for this kind of element used as a bearing or washer. An exception may be the ‘light bronze cylinders’ (κύλινδροι χαλκοῖ κοῦφοι) of Hero’s Cheiroballistra W 129 (Marsden 1971: 208, 242 n.º 4), which are functionally similar. There are strong arguments for supposing that this work is spurious and reflects later Roman developments in artillery (see Baatz 1978: 9-14, with references). In Latin, khoinikis was usually translated as modi­ olus, both in the sense of the nave of a wheel (e.g. Vitr. 10.9.2; Plin. Nat. 9.8) and as a washer (Vitr. 10.11.5, 10.12.1 [both in artillery]), as well as in the more extended and specialised use in the ­outer shell of the trepanning drill (Cels. 8.3, which makes the translation explicit; for the Greek, see Gal. 10.448 and 19.129; also χοινίκιον Gal. 14.783). Overall, apart from the modiolus also ­being used for the meaning of a cup or bucket (e.g. Vitr. 10.4.3, 10.5.1), the core semantic fields of the two terms were closely aligned. In comparison with the discussions of the knōdax, there is less (but by no means no) direct treatment of the need to reduce friction or for smooth turning of axles with the khoinikis as a bearing, but there is a concern for both load and wear. It is clear that the khoinikides were a focus of wear, with papyri that emphasize replacement kh­ oinikides or express concern for whether a ­machine comes with its khoinikides or not. Thus the author of Chr.Wilck. 176 complains that the hire of an olive-oil press did not include khoinikides that needed to be funded out of private means, while conversely the lease of a cart (PCair.Masp. 67303, above) emphasizes that such critical fittings are included and POxy. 2778 emphasizes that replacement khoinikides are available (for wear at the axle 12  Using ἐνστρέφω or simply στρέφω: see, for example, Et. Magn.: 676.41; Ps-Zonaras, Lexicon: 1554; Σ Dem. 22.175a; Σ Hom., Il. 5.726.

179

end, see also Σ A. Th. 205c, Σ S. El. 745). For these applications, a different choice was made in terms of materials. Where specified, khoinikides were almost always built of, or clad in, bronze, across a range of uses (Hero, Aut. 2.3, Mech. 2.1, Bel. 20 [p. 96], Dioptr. 3, 4; Ph. Bel. 24 [p. 60], Vitr. 10.12.1). Where the khoinikides are used as bearings, this is usually specified as rotation of bronze against bronze (Mech. 2.1; Dioptr. 3, 4). Exceptionally, in large catapults, the khoinikides could be made of wood (Hero, Bel. 20 [p. 96]) with bronze plating (Marsden 1971: 53 n. 30) and the rotating drill, for its specific function, was according to Celsus made of iron. The load on the axle can be seen to be a consideration in a number of cases. Since the khoin­ ikis surrounds the axle or shaft, where they are horizontally mounted, the lower part of the kh­ oinikis is always bearing the weight of the axle or shaft. The torsion engine provides a different kind of wear, and it is here that we see explicit reference to padding that seeks to cushion the forces exerted on the khoinikis (Ph., Bel. 16 [p. 57]). This was, then, the most common bearing in the Greek and Roman world, used through to late antiquity, but not without its problems, with the need for wear and replacement a factor in users’ minds. Khelōnion (Χελώνιον) This picture is, however, complicated in a further set of evidence which uses the different term khelōnium (χελώνιον) to refer to the sockets at each end of the axles of windlasses. In manuscript sources, this use is confined to Vitruvius (cheloni­ um: 10.2.2, 5; 10.3.2). He also uses this term of specific applications of windlasses in artillery and siege engines (10.12.1, 10.5.4). Vitruvius does not, however, refer to a khoinikis in these contexts. It is possible that this represents another element in the assembly and that Vitruvius has simply failed to mention it. Thus, Hero’s account of the crane, with which we began (Mech. 2.1), descri­bes the sockets into which his khoinikis-clad windlass fits as “circular holes with bronze linings under [or adjacent to] the khoinikides” (τρήματα στρογγύλα...τῶν τρημάτων τριβεῖς χαλκοῦς ἐχόντων ὑποκειμένους ταῖς χοινικίσι). It is possible that the term refers to such liners fixed in the sockets, but it is likely, however, that here, at least, the terms are referring to essentially the same kind of element as the khoini­ kis as outlined above.

180

Duncan Keenan-Jones, Ian Ruffell, Euan McGookin

The chelonium of Vitruvius’ crane must have enclosed the axle of the windlass at the front and top at least, and probably all around the axle’s circumference, to prevent the windlass being pulled forwards and upwards by the rope. Thus, the khelōnion of the crane was very likely cylindrical like the khoinikis. Vitruvius (10.2.8) also used khelōnion for the two sockets on the single-beam crane (see Fleury 2009: 186, fig. 3) from which compound pulleys were attached by rope, suggesting a cylindrical hole. In contrast to the khoin­ ikis as described by Hero in Mech. 2.1, the khēlonion does not itself rotate. Evidence from papyri provides some support for interchangeable terms for much the same kind of element. As we have seen, PLond. 1177 provides evidence for knōdakes. It also repeatedly refers to water-lifting machines that use khelōnia (179, 208, 221) – which Habermann (2000: 207-208) interprets to be sockets for either a horizontal or ver­tical axle – but never khoinikides. Other papyri similarly refer to machines with khelōnia without khoini­ kides, but in terms reminiscent of the ones already mentioned (compare PLond. 1177.208 with PLou­ vre 11.24-6 and POxy. 2778.13-15). This would support the idea of interchangeable terminology. Further help might be had from the use of khelōnion in other contexts. As well as its primary meaning of tortoise (or crab) shell, khelōnion could mean an arched human back. In extant Greek mechanical texts, khelōnion is exclusive to works on artillery, where it denotes objects that always support or bear another object, but the other object can be supported either on the inner, concave surface or the outer, convex surface of the shell (or arched back). Vitruvius (10.11.7-8) followed Philo (Bel. 28 [p. 61]) in using khelōnion for the entire slider,13 which Hero called a diōstra (δίωστρα) (Marsden 1971: 161 n. 28; Schiefsky 2005: 258-262). This slider contained a hemicylindrical groove to hold the bolt, as reconstructed byMarsden (1971: 203, fig. 13) and attested in remains of artillery found at Xanten-Wardt.14 Conversely, Hero reserved the term khelōnion for the small raised block or plate on which the trigger was mounted (Marsden 1971: 161 n. 28; Schiefsky 2005: 258-259), stating that the reason for this name is that the block projects from the rod on which it is mounted (χελώνιον (ἦν γὰρ 13  Some prefer to read χηλήν here instead of χελώνιον (Schiefsky 2003: 267 citing Callebat and Fleury 2003: 233). 14   Wilkins n.d. (see n. 10 above).

Anejos de AEspA LXXVII

καὶ ὑψηλότερον τοῦ ἐπικειμένου κανόνος) Bel. 6 [pp. 77-78]). This block seems to have had no groove, although it does have a round peg fixed on its upper surface, possibly by inserting it into a hole. This hole is small and off-centre, and seems unlikely to be a concave element leading to the use of the term khelōnion for this block. In another artillery design, Hero (Bel. 17 [p.  93]) uses khelōnion to mean a pad located where one moving part comes to rest on another. In the artillery piece called the scorpion, Vitruvius also used chelonium to mean the block or cushion on which the rear part of the case rested, which was itself supported by another beam (10.10.5). He glosses chelonium here as pulvinus or pillow. This seems not to have had a groove in which the case rested, since, at a width of ¾ of the diameter of the hole for the torsion spring, the chelonium was thinner than the case, which was probably as wide as the spring hole (Marsden 1971: 194-195, n.º 8, figs. 1-2). This pad may have rotated on a pin to facilitate the movement of the case, if the reconstruction of Fleury (Callebat and Fleury 1995: Pl. XXIV) is to be preferred to that of Marsden (1971: diagram 10). These uses in artillery contexts suggest a hemi­spherical element, either concave or convex (although a pad or a cushion might perhaps be seen in either terms). A recess is also suggested by its other main mechanical use in antiquity, where it refers to a lock or specifically the socket (ὀχῆας] νῦν τὰς λεγομένας βαλάνους τὰς ἐν τῷ λεγομένῳ χελωνίῳ κατ’ ἀντικρὺ τῆς κλειδὸς, ‘bolts: what are now called balanoi which are in the khelonion right against the key’ Σ Hom. Od. 21.47) and is most well-known from documentary contexts (Hellenistic inscriptions from Delos [IDélos 316, 338, 354, 403, 461, 1412, 1417, 1442, 1443, 1426; XI.2, 287]; but also papyri: PTebt.46.17, BGU 1028.20,26, POxy. 113.4. See Habermann 2000: 207 n. 341). This would perhaps be most consistent with a cylindrical rather than hemispherical recess, but the lack of detail makes it difficult to draw too many inferences. In many cases then, the khelonion is clearly a cylindrical element, but could be hemicylindrical. This raises the possibility that the khelonion mentioned in the water lifting devices may be identical to the probably hemicylindrical bearings (fig. 6B) known from Vitruvius’ description of the water wheel (10.4.1-2) and from the bearings supporting the axles of water-lifting wheels at Rio Tinto (Oleson 1984: fig. 115 and, now in the Huelva Muse-

Anejos de AEspA LXXVII

Taking a bearing on Hero’s Anti-Crane and its Un-windlass...

um, figs. 123, 126, perhaps reconstructions). Without any epigraphic evidence, it is difficult to draw any certain conclusions, but it may well be that a khelonion would be the more appropriate term for a hemicylindrical socket than a khoinikis. One might speculate that this might in turn explain the distinction between the water-lifting device of P.Lond. 1177.179, with its khelonion, in comparison with other water-lifting machines described as having khoinikides. If so, it is curious that Vitruvius used the term khelonion for the crane but not for the water-lifting wheel. Similarly, as far as we can tell, our examples of the khoinikis suggest a cylindrical shell open at both ends (especially in Aut. 11 and other uses where it means the hub of a wheel, as well as the artillery washers). But the idea of a socket drawing its name for the concave inner surface of a tortoise shell suggests the khelōnion may have been closed at one end in at least some of its configurations. These may, however, be overly precise distinctions and it would be preferable to see khoinikis and khelōnion as overlapping (but not identical) terms. Hero himself seems to have made a distinction, using khelōnion only in its convex sense, and only in artillery contexts, while using khoinikis for sockets. The square axles (Fig. 6A, where the square hole for the axle is visible in the centre of the wheel) with rounded ends bearing the Rio Tinto water wheels were constructed similarly to those of Mech. 2.1 (quoted above) but entirely from bronze (Oleson 1984: 256, figs. 116, 122, 123, 127, 128). Some of these bronze axles had nearly worn through, and it may have been a propensity for this type of failure that led Vitruvius (10.4.1-2), in a description almost identical to Mech. 2.1, to specify iron linings (described as lamna instead of khoinikides/modioli) for the axles and bearings of water lifting devices. In contrast to the crane, water lifting devices would be rotating almost constantly, leaving them very liable to the heating/ annealing positive feedback loop described earlier. CONCLUSION Schiefsky (2005: 254, 269) argued that the invention of torsional artillery in the third century BC led to the creation of a detailed terminology that remained stable down into the first century BC, in contrast to other fields such as medicine. The evidence for the language of construction is more diverse and in many respects less detailed,

181

Fig. 6. A. Remains of the wooden wheel and bronze axle (not shown) of a water wheel from Roman mines at Rio Tinto (now on display in the British Museum Nr. 1889,0622.1. © The Trustees of the British Museum. Used with permission). B. A reconstruction of the hemicylindrical bearing of the axle.

but shows consistency over a long term. We have suggested that precise terms for the knōdax type of bearing were in use as early as the 6th century BC, and also remained stable down to the same period (at least to the second century CE). Other bearings, the khoinikis and khelōnium, cannot be traced back so far, but can be attested as being in established use in some construction contexts by the Hellenistic period (for axle bearings [particularly if we can trace the scholia ultimately back to that period] and hinge-sockets; and for door locks respectively). It is possible that these terms were appropriated from construction terminology to

182

Duncan Keenan-Jones, Ian Ruffell, Euan McGookin

describe parts of artillery. The careful usage of bearing terminology continued into Latin, with terms either being transliterated or, in the case of khoinikis/χοινικίς/modiolus, being consistently translated. Whether the chelonium refers to a somewhat different sort of bearing from the kh­ oinikis cannot be fully determined at this time. Here, at least, there may be some overlap, but (from what textual evidence there is) the khoinikis remained the predominant term and is attested in documents down to the sixth century CE. The knōdax was a thin, probably pointed pin, often made of iron, which allowed larger cylindrical or spherical shapes to rotate. Either the pin was fixed into the end of a larger object and rotated (together with it) in a socket (as in the peri automatopoiētikēs) or the larger object rotated directly on fixed pins. There was more variety in the nomenclature of the sockets, but in no instance was a khoinikis used with a knōdax. The khoinikis was a (usually bronze) cylindrical shell around a shaft, one of which was rotating relative to the other. The khelōnion or tortoise-shell used in Vitruvian cranes was clearly some kind of a cylindrical socket, but may also have been hemicylindrical in some contexts. It probably fulfilled the same function as the khoinikis even if not identical in form. The khelōnion clearly had a wide variety of uses dependent on whether the shell’s inner (concave) or outer (convex) surface was figuratively in view. We have argued that both in terms of its use of bearings and in terms of its drive mechanism Hero’s automaton has continuities with devices used in construction. The drive was the exact opposite of a crane both in function and in loadbearing capacity, and this relationship to the windlass is reflected in the rare and undoubtedly specialised term he used for it, the exeliktra. The use of the knōdax as the bearing for rolling elements can also be best paralleled, on current evidence, within the arena of construction. This is not an exclusive relationship, however, and we have used relationships with other types of ancient devices, from artillery to medicine, to explain Hero’s drive mechanism. Hero’s mechanical expertise and experience is evident in his design of the automata, particularly in the choice of materials, and possibly shape, and consideration of friction in the performance of bearings. Hero specified iron for the knōdax, whose thinness may have been beyond bronzemaking technology at the time, and its socket. Their small size, the ensuing concentration of

Anejos de AEspA LXXVII

friction, as well as, in the automaton at least, greater loads, made resistance to heat-induced deformation more critical. Hero specified bronze for the thicker axle and its khoinikis, which were often more lightly loaded in the automaton. This design process is similar to the modern Failure Modes and Effects Analysis (FMEA). Hero is precise and consistent in his use of terminology in both the Mechanica and the peri automatopoiētikēs, and in expressing a preference for one type of axle configuration in the peri automatopoiētikēs over another. Yet the differences between the two configurations are not fully explained, even at the level of basic design, and the explanation of the advantages of the knōdax come almost in passing. This reticence raises interesting questions about the character of Hero’s intended readers. A full examination of that problem would necessitate going beyond the drive mechanism alone, but we may tentatively suggest a bifurcated audience. Practitioners would be fam­ iliar with, in some cases, quite specialised termin­ ology and would be able to fill in the blanks both in Hero’s account of the automata and in his reasoning for taking certain design decisions. Nonspecialists would, for different reasons, be impressed by this display of expertise, as Hero draws aside the curtain to reveal the working of wonders (thaumata). If the functioning devices excite visual wonder, the treatise itself works both implicitly and explicitly to stake a claim for mechanical skill and sophistication, which, for Hero, is equally the point of the art of making automata. ACKNOWLEDGEMENTS We wish to thank the organizers for the opportunity to participate in this workshop. This work was funded by a Leverhulme Trust Research Project Grant “Hero of Alexandria and his Theatrical Automata”. Dr. Keenan-Jones’ attendance at the workshop was supported by Classics at the University of Glasgow. All images were created by D. Keenan-Jones unless otherwise stated. PRIMARY SOURCES Hero of Alexandria Schmdt, W., Nix, L., Schöne, H. and Heiberg, J. L. (eds.) 1899-1926: Heronis Alexandrini Opera Omnia. Teubner, Leipzig (reprinted 1975: Teubner, Stuttgart).

Anejos de AEspA LXXVII

Taking a bearing on Hero’s Anti-Crane and its Un-windlass...

Belopoeica Diels, H. and Schramm, E. (eds.) 1918: Herons Belo­ poiika. Schrift vom Geschützbau. Königl. Akademie der Wissenshaften, Berlin. Oribasius Raeder, J. (ed.) 1933: Collectionum Medicarum Reli­ quiae, Libri XLIX-L, Libri Incerti, Eclogae Medica­ mentorum. Teubner, Leipzig-Berlin. Philo of Byzantium Belopoeica Diels, H. and Schramm, E. (eds.) 1919: Philons Belo­ poiika (Viertes Buch der Mechanik). Akademie der Wissenschaften, Berlin. Sextus Empiricus Adversus Mathematicos Mutschmann, H. (ed.) 1914: Sexti Empirici Opera, II, Adversus Dogmaticos, Libros Quinque (Adv. mathem. VII-XI) Continens. Teubner, Leipzig. Vitruvius Krohn, F. (ed.) 1912: Vitruvii De Architectura. Teubner, Leipzig.

REFERENCES Adam, J.-P. 1994: Roman Building. Batsford, London. Allevato, E., Russo Ermolli, E., Boetto, G. and Di Pasquale, G. 2010: “Pollen-Wood Analysis at the Neapolis harbour site (1st – 3rd century AD, Southern Italy) and its archaeobotanical implications”, Journal of Archaeological Science, 37, pp. 23652375. Baatz, D. 1978: “Recent finds of ancient artillery”, Britannia, 9, pp. 1-18. Callebat, L. and Fleury, P. (eds.) 1995: Dictionnaire des termes techniques du De Architectura de Vitruve, Alpha-Omega 123. Olms-Weidmann, Hildesheim. Cotterell, B. and Kamminga, J. 1992: Mechanics of Pre-Industrial Technology: An Introduction to the Mechanics of Ancient and Traditional Material Cul­ ture. Cambridge University Press, Cambridge (New edition). Cooper, F. A. 2008: “Greek engineering and construction”, in Oleson, J. P. (ed.), Oxford Handbook of En­ gineering and Technology in the Classical World, pp. 225-255, Oxford University Press, Oxford-New York. Coulton, J. J. 1977: Ancient Greek Architects at Work: Problems of Structure and Design. Cornell University Press, Ithaca, N. Y.

183

Craddock, P. T. 2008: “Mining and metallurgy”, in Oleson, J. P. (ed.), Oxford Handbook of Engineering and Technology in the Classical World, pp. 93-120. Oxford University Press, Oxford-New York. Dowson, D. and Hamrock, B. J. 1981: “History of ball bearings”. NASA technical memorandum 81689. NASA. http://ntrs.nasa.gov/archive/nasa/casi.ntrs. nasa.gov/19810009866.pdf. Drachmann, A. G. 1963: The Mechanical Technology of Greek and Roman Antiquity: A Study of the Liter­ ary Sources. Munksgaard, Copenhagen. Fleury, P. 2009: “Dénomination générique, dénominations particulières. L’exemple des machines de levage en latin”, Voces, 8. http://revistas.usal.es/index.php/1130-3336/article/view/5543, accessed 09/04/15. Galli, G. 1996: “Roman flanged pump bearings: further finds in the harbour of Ponza (Pontine Islands, Italy)”, International Journal of Nautical Archaeol­ ogy, 25.3-4, pp. 257-261. Glare, P. W. (ed.) 1982: Oxford Latin Dictionary. Clarendon Press, Oxford. [OLD] Habermann, W. 2000: Zur Wasserversorgung einer Metropole im kaiserzeitlichen Ägypten. Neuedition von PLond. III 1177. Text – Übersetzung – Kommen­ tar, Vestigia 55. Beck, Munich. Koldewey, R. and Puchstein, O. 1899: Die griechis­ chen Tempel in Unteritalien und Sicilien. Asher, Berlin. http://digi.ub.uni-heidelberg.de/diglit/koldewey1899bd1. Landels, J. G. 2000: Engineering in the Ancient World. University of California Press, Berkeley (Second Edition). Liddell, H. G. and Scott, R. 1940: A Greek-English Lexicon (revised and augmented throughout by Sir Henry Stuart Jones with the assistance of. Roderick McKenzie. Clarendon Press, Oxford. [LSJ] Marsden, E. W. 1971: Greek and Roman Artillery: Technical Treatises. Clarendon Press, Oxford. Meighörner-Schardt, W. 1990: “Zur Rekonstruktion Eines Römischen Bockkranes”, Journal of Ro­ man Military Equipment Studies, 1, pp. 43-60. O’Connor, C. 1993: Roman Bridges. Cambridge University Press, Cambridge. Oleson, J. P. 1984: Greek and Roman Mechanical Wa­ ter-Lifting Devices: The History of a Technology, Phoenix 16. Reidel, Dordrecht-Lancaster. Pantelia (ed.) 2015: Thesaurus Linguae Graecae: A Digital Library of Greek Literature, University of California at Irvine. http://stephanus.tlg.uci.edu (accessed 10/06/15). Pintaudi, R. 1995: “Spigolature III”, Zeitschrift für Papyrologie und Epigraphik, 58, pp. 89-92.

184

Duncan Keenan-Jones, Ian Ruffell, Euan McGookin

Rehder, J. E. 1992: “Iron versus bronze for edge tools and weapons: a metallurgical view”, The Journal­ of the Minerals, Metals & Materials Society 44.8, pp. 42-46. Robertson, N. 1986: “Solon’s axones and kyrbeis, and the sixth-century background”, Historiai, 35.2, pp. 147-176. Schiefsky, M. J. 2005: “Technical terminology in Greco-Roman treatises on artillery construction”, in Fögen, T. (ed.), Antike Fachtexte. Ancient Techni­ cal Texts, pp. 253-270. De Gruyter, Berlin. Stroud, R. S. 1979: The Axones and Kyrbeis of Drakon and Solon. University of California Press, BerkeleyLos Angeles.

Anejos de AEspA LXXVII

Tybjerg, K. 2003: “Wonder-making and philosophical wonder in Hero of Alexandria”, Studies in ­History and Philosophy of Science, Part A 34.3, pp. 443-466. Ucelli, G. 1940: Le navi di Nemi. La Libreria dello Stato, Roma. Veal, R. in press: “Detecting building timbers in the archaeological record: charcoal from selected late Republican and early imperial contexts”, in Hurst, H. (ed.), Santa Maria Antiqua. Wilkins, A. 2003: Roman Artillery. Shire Publications Ltd., Princes Risborough.

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

LIFTING BLOCKS, 1ST - 5TH CENTURY AD: THE INCLINED PLANE GIANGIACOMO MARTINES*, MATTHIAS BRUNO**, CINZIA CONTI*** * Former Regional Director of the Ministry of National Heritage, Culture and Tourism, in Friuli Venezia Giulia, Italy ** Archaeologist and expert in ancient marbles and quarries, Rome *** Archaeologist at the Soprintendenza Speciale per il Colosseo, il Museo Nazionale Romano e l’Area archeologica di Roma, Ministry of National Heritage, Culture and Tourism, Director of the Laboratory of Restoration

ABSTRACT: According to modern Mechanics, the inclined plane is classified as a simple machine. Hero of Alexandria wrote about it in Mechanics. In the second book, Hero illustrates five simple machines that offer several advantages. In the third book, he presents how to lift down a heavy block from a quarry by a special funicular railway. On an inclined plane a load is always safe. Dating to the end of the first century AD is the Haterii treadwheel crane, a políspaston, used in the construction of the Colosseum, whose blocks weigh at most between 5 and 8 tons. But the crane’s power was very much bigger. Going up inside Trajan’s Column, one observes the exceptional size of the blocks: the pedestal’s last block weighs 72.33 tons, while the capital weighs 44.66 tons. To move them, a combination of five of Hero’s machines may have been used, as well as the inclined plane. KEYWORDS: Hero of Alexandria, Pappus of Alexandria, Políspaston, Haterii treadwheel crane, Trajans’ Column, Column of Marcus Aurelius, Column of Arcadius. RESUMEN: La mecánica moderna considera el plano inclinado una máquina sencilla. Herón de Alejandría escribió sobre ello en su obra Mechanica. En el segundo libro Herón presenta las cinco máquinas sencillas que ofrecen diversas ventajas. En el tercer libro, Herón describe cómo hacer descender un bloque pesado de una cantera mediante un funicular especial. Sin embargo, en el plano inclinado la carga está siempre segura. A finales del i siglo d.C., un testimonio de este tipo de máquinas es la grúa de los Haterii, un polispasto, que fue utilizado en la construcción del Coliseo, cuyos bloques pesan como máximo entre las 5 y las 8 toneladas. Sin embargo, la fuerza de la grúa era mucho mayor. Subiendo por el interior de la Columna Trajana, se observa la excepcional dimensión de los bloques: el último del basamento pesa 72,33 toneladas, mientras que el capitel pesa 44,66. Para su movimiento podemos deducir la utilización de una combinación de las cinco máquinas de Herón y también el empleo del plano inclinado. PALABRAS CLAVE: Herón de Alejandría, Papón de Alejandría, polispasto, grúa de los Haterii, Columna Trajana, Columna de Marco Aurelio, Columna de Arcadio.

In memory of Vanni Mannucci INTRODUCTION Pliny the Elder tells us about the construction of the temple of Diana at Ephesus, and how the epistyle blocks were laid on the columns. The arch­itect Chersiphron built a ramp that went up to the height of the capitals: (...) molli clivo super

capita columnarum exaggerato – “(...) constructing a gently graded ramp which reached the upper surfaces of the capitals of the columns.”1 In 1  Pliny, Naturalis Historia, 36.96; English trans. Eichholz, 77. The temple had been rebuilt in marble in about 560 BC, then was destroyed by a fire in 356 BC. The central lintel

186

GIANGIACOMO MARTINES, MATTHIAS BRUNO, CINZIA CONTI

­ egalithic architecture it was quite usual to use m the inclined plane, being a natural way to lift huge blocks of stone in quarries, on roads and at building sites. In the history of architecture, we do not know when the inclined plane ceased to be regularly used. In antiquity the inclined plane was much studied to determine the mathematical formula. This shows that this system continued to be of interest to engineers up to the fourth century AD. However, the formulas calculated in ancient times continued to be inaccurate until Galileo Galilei and Simon Stevin, separately, found the right one.2 On the other hand, in Greek architecture of the classical period and in Roman times, cranes and other machines were regularly used to lift weights, by multiplying the force applied. We do not know the maximum load they could lift. As to this aspect, we can only speculate, taking into account the weight of the blocks and examining any tool marks left on individual monuments. INCLINED PLANE IN ANCIENT SCIENCE In his first book of Mechanics, Hero of Alexandria discusses the inclined plane3 and the theory for lifting a cylinder, which in the context is a column (fig. 1A).4 He considers the contact point between the inclined line and the circle and draws a perpendicular that cuts the circle into two un­ equal parts. The two circular segments, in dark in the figure, form the weight that is balanced on the inclined plane, while to hold the entire circle in place, we must apply a force equal to the crescent. An increase in the inclination of the inclined plane also increases the crescent. Hero’s explanation is the most accurate in antiquity. Pappus, on the other hand, considers strength Γ needed to move block A along a horizontal plane, which depends on friction (fig. 1B).5 He weighed over 41 tons (Coulton 1974: 12 note 63; Fleury 1993: 140; Martines 2000: 32). 2   Sinopoli 2015: III and 144-165. 3  Hero, Mechanica 1.23. 4  In fig. 1, the plan and Hero’s formula are taken from Glagett 1959: 41-42, 49-50. See also Drachmann 1963: 47-49. 5  Pappus Alexandrinus, Collectio Mathematica 8, Proposition 9. Pappus’s solution can be easily understood in Cuomo 2000: 114 fig. 3.2. See also Duhem 1905, I: 184-186; Downey 1947-1948: 197-200. Friction was studied by Leonardo da Vinci in the Madrid I, Atlantico and Arundel Codices: Galluzzi, 1989: 16-18. In fig. 1 Pappus’s formula is taken from Cohen and Drabkin 1958: 196 note 3; the value for sliding friction, oak on oak, wet, is taken from Ormea 1988, I: 175.

Anejos de AEspA LXXVII

then places a sphere equivalent in volume to A on an inclined plane and calculates the increase in Γ to keep it in balance on the inclined plane. Like Hero, Pappus, too, considers a vertical line from the point where a circle touches an inclined plane and on that point he sets a scale with a beam the length of the radius, as Archimedes does in his Method. Having placed weight Γ at the centre of the circle, force Δ keeps the sphere balanced on the inclined plane. An increase in the inclination also increases Δ. Pappus’s intuition may have come to him with a plane inclined at 30°, as in figure 1B, a really intuitive solution. Though erroneous, the assumptions made by Hero and Pappus are developed logically. In the diagram (fig. 1C), we can compare the formulas of Hero, Pappus and Galileo.6 Along the abscissa, we have the degree of the slope, and, along the ordinate, the part of the weight needed to support the weight itself on the inclined plane. Hero’s curve follows a path similar to Galileo’s. The formula presented by Pappus becomes impossible after 45 degrees but Pappus and Galileo coincide at 30°. We may note that between 15° and 42°, the three curves are included in the same field: Alexandrian mechanics could be proud of their formulas because the excess strength, compared to Galileo, corresponds more or less to friction. According to modern Mechanics, the inclined plane is classified as a simple machine.7 The inclined plane could be made of wood or simply earth, and did not require the construction of any device: so Hero did not consider it a machine. In Hero’s treatise, blocks in a sled on rollers slide on “planed boards on the ground, smeared with grease; so the burdens are moved without the need for great power”.8 This technique was used to move the great stone blocks of ancient Egypt.9 Hero and Pappus were both from Alexandria and their knowledge was based on the construction 6   Galileo’s solution for the inclined plane is found in Galilei, Opere, II, Le Mecaniche: 178-186; also Discorsi intorno a due nuove scienze attenenti alla Mecanica e i Movimenti Locali, Giornata terza Teoremi 3-5, Opere, VIII: 215-221; Galilei, On Motion and on Mechanics: 169-177. The correct solution for the inclined plane had already been identified in the 13th century by Jordanus de Nemore, using another method, unknown to Galileo: Moody and Clagett 1952: 190-191, Proposition 1.10. See also D’Alessandro and Napolitani 2013. 7   Stevin 1605: 34-41, 99-101. Galilei, On Motion and on Mechanics: 169-177, wrote on the inclined plane and the screw. Varignon 1725, I: 1. See also Mach 1883, 22-31; Duhem 1905, 182-193; Renn 2013; Sinopoli 2015. 8  Hero, Mechanica 1.21. Drachmann 1963: 46-47. 9   Choisy 1904: 117-121; Clarke and Engelbach 1990: 8495; Arnold 1991.

Anejos de AEspA LXXVII

LIFTING BLOCKS, 1ST - 5TH CENTURY AD: THE INCLINED PLANE

187

Fig. 1. Solving the problem of inclined plane: A. According to Hero; B. According to Pappus; C. Comparing the solutions of Hero, Pappus and Galileo.

traditions of the pharaohs. Herodotus tells us of a road built to carry stone blocks to the Great Pyramid of Giza: it took ten years to build and another twenty for the pyramid itself. The road was a perfect work with a smooth stone surface.10 Jean-Claude Golvin and Jean-Claude Goyon wrote a book in which they discuss the papyri and the archaeological remains of roads and ramps, as well as the experiments performed in the early twentieth century by Georges Legrain.11 Lifting  Herodotus, Historiae 2.124. Clayton and Price 1988: ch. 1.   Golvin and Goyon 1987: 88-107. See also Nicholson and Shaw 2000; Parra Ortiz 2008. 10

11

weights by means of an inclined plane is analogous to transporting them on water, in which case the weight is lightened by Archimedes’ buoyancy principle.12 Nature can come to man’s aid. In the second book, Hero illustrates five simple machines,13 which offer different advantages: the pulley14 changes the direction of force and a compound pulley system15 lightens the load to be 12  Archimedes, Opera, II, De Corporibus Fluitantibus, 1.57. Wirsching 2007; Wirsching 2000: 273-283. 13  Hero, Mechanica 2.1.1-5. 14   Drachmann 1963: 67-69. 15   Drachmann 1963: 53-55.

188

GIANGIACOMO MARTINES, MATTHIAS BRUNO, CINZIA CONTI

lifted. The lever,16 the capstan,17 the wedge,18 and the screw19 can be used to multiply the force applied. Theoretically, Hero’s engines have no limits. Lynne Lancaster has produced plausible reconstructions for the raising of Trajan’s Column,20 and Manolis Korres21 and Jean-Pierre Adam22 for the Mausoleum of Theodoric in Ravenna. In the third book, Hero presents how to lower a heavy block from a quarry by means of a special funicular railway (fig. 2A). The block is moved slowly by a team of oxen, ropes, pulleys and two carts loaded with small weights.23 The manuscript Add. 23,390 of the British Museum, has an illustration in the fol. 43 r, with three captions: “the mountain”, “the weight”, and “p” for pulley.24 This short passage is followed by a description of how a column can be raised by means of a counter­balance astride a main wall (fig. 2B).25 Counterbalances and quarry roads are two specific themes discussed by Hero. To conclude our review of Hero’s treatise, we can distinguish two criteria for lifting a load: first by suspension, and secondly by raising it while one end remains on the ground. The risks of suspension include the ropes breaking and the wooden poles buckling, but on an inclined plane, a load is always safe. THE HATERII CRANE Dating from the end of the first century AD is the Haterii tread wheel crane (fig. 3),26 a políspas­ ton, used in the construction of the Colosseum, whose blocks weigh between 5 and 8 tons;27 but the crane could lift much greater loads.28 If we take a close look at the bas-relief, we can see a crane with a double mast: in fact there are two parallel and overlapping wood-stanchions in   Drachmann 1963: 52-53.   Drachmann 1963: 50-51. 18   Drachmann 1963: 55-56. 19   Drachmann 1963: 56-59. 20   Lancaster 1999: 419-439. 21   Korres 1997: 219-258. 22   Adam 2012. 23  Hero, Mechanica 3.9. Drachmann 1963: 106-107. 24  Cureton and Rieu 1871: 619-620, n.1337 cod. Add. 23,390. Drachmann 1963: 107 fig. 42. 25  Hero, Mechanica 3.10. Drachmann 1963: 107-109. 26   Sinn and Freyberger 1996: 51-59. 27  Bruno et al. 1998: the authors examined the weights of the blocks used to build the Colosseum and the Arch of Septimius Severus. 28   Coulton 1974: 1-19, particularly 13 note 73: “(...) the real lifting capacity could still be of the order of 20 to 30 tons”. 16 17

Anejos de AEspA LXXVII

the relief, duo tigna.29 The mast is operated by means of seven guylines, two on the right in front of the crane, Latin antarii funes, and five on the left, behind the crane, retinacula.30 The ropes are slack because the machine has stopped and the construction already finished. Fixed to the top of the mast is a standing purchase block, summa tro­ clea, but we cannot see the ima troclea since it is hidden by the treadwheel.31 The ima troclea is attached to the load to be lifted. During the lifting the summa troclea remains stationary while the ima troclea moves with the load. The políspaston can lift a weight with a force that is less that the weight itself. Let us see how. Under the standing purchase block we can see 8 vertical lengths of rope – five clearly visible in the foreground and three others in the background – as previously observed by Blümner, and then by Sinn and Freyberger.32 The lengths of rope in the foreground are part of the compound pulley system shown in figure 4A. Since there are eight lengths of rope, the system is a double one as in figure 4B. Evidently, there are two haulage ropes.33 To understand how it works, we shall make use of a notion of Analytical Mechanics. In figure 4B, at the bottom, the ima troclea, namely the runner block, holds load W. If we divide the load by eight ropes equally, for the system to be in balance, each rope is pulled by force W/8. Thus, the compound pulley system divides the load according to the number of lengths of rope that run betwenn ima and summa troclea. Now, to calculate the lifting capacity of the compound pulley system, we need to know the diameter of the rope and the fibre used. As an   As regards Latin nomenclature, see Fleury 1993: 95-128.   Each guyline is connected to the mast by means of a stropped-block, and the strop is made by three turns of rope. The biggest stropped-blocks hold the stanchion in the foreground. On the blocks on the left, we see a small length of pull rope being lowered between the two stanchions. This analysis of the ropes qualifies the previous one by Martines 1998-1999: 261-275, especially 264-266. 31   We can see 4 pin heads on the cheek of the standing purchase block, 3 square and 1 circular above, on which a worker rests his foot. The top pin holds the strop, i.e. the rope connecting the purchase block to the mast of the crane; but the strop is hidden. The three pins below hold the pulleys, orbiculi. 32  Blümner 1884: 122 and fig. 11. Sinn and Freyberger 1996: 54; note 16 indicates another little rope on the right coming down from the block of the overlying antarius funis, but it is simply a little rise of background. 33   By shining oblique artificial light, we can see one and maybe two lengths of rope, below the worker’s foot, coming out of the standing purchase block and being lowered between the two stanchions of the mast. 29

30

Anejos de AEspA LXXVII

LIFTING BLOCKS, 1ST - 5TH CENTURY AD: THE INCLINED PLANE

189

Fig. 2. Hero of Alexandria, Mechanics: A. Ms. Add. 23,390, British Museum, early 17th century, fol. 43 r, a stone block being lowered down a mountain road (photo and © British Museum, London); B. Ms. Or. 51, Bibliotheek der Rijksuniversiteit, Leiden, 1445 ca., p. 66, a column being raised on its end using a counterweight (photo and © Bibliotheek der Rijksuniversiteit, Leiden).

example, we have chosen a famous rope, the rope which is carved on the pedestal of Trajan’s Column, binding the trophy of arms: diameter 5.5 cm, perhaps made of hemp (fig. 4C). For this sort of rope, a safe working load could be two tons.34 So, in the double pulley block (fig. 4B above), each haulage rope can hold two tons, and the eight lengths of rope, together, can hold load W of 16 tons. QED: the compound pulley system uses a little force to balance a great load, while the mast holds the rest of the load, with a buckling force acting on it. The compound pulley system balances the load, but it is the wheel that produces movement. The two haulage ropes are wound around the axis of a giant tread wheel, which is operated by five men (fig. 4D).35 We cannot see the winding   Zignoli 1951-1956, III: 98, 5.1.2.  Outside the head-wheel, two men hold two ropes connected to it: evidently their job is to start and stop the machine. Sinn and Freyberger: 54 note 13. 34

35

axis that corresponds to the axis of the tread wheel, at the back of the image. The wheel is a capstan that multiplies the weight force of the five men by the radius of the wheel itself. The radius may be calculated as the height of the men in the wheel, e.g. 170 cm. In proportion, the external radius is 3.4  m, and the internal radius, where the men tread, 3.0 m. The five men can exercise of force equivalent to their combined weights, i.e. 400 Kg, with a wheel moment of 1200 Kgm, i.e. 5 men × 80 Kg × 3 m. The m ­ oment is the product of a force multiplied by its distance from a certain point, in this case the wheel centre. The difference between the diameter of the wheel, 3 m, and the axis of rotation, by hypothesis, of 22 cm, or 1 palmus, creates movement, like in a scale  with two different weights on beams that  are  inversely proportional to the weights: 400 Kg × 3 m > 4000 Kg × 0.22 m. So the crane had the power to lift 16 tons by applying a force of just 400 Kg. The advantage, i.e. the ratio of the load lifted and force used is 40

190

GIANGIACOMO MARTINES, MATTHIAS BRUNO, CINZIA CONTI

Anejos de AEspA LXXVII

We can work out the lifting capacity of the machines in Hero’s Mechanics. At the beginning of Book 1, he presents a manual winch, the Greek Barulkós, with 6 cogwheels, and shows how a child that turns a crank, producing a force of 5 talents, can lift a load of 1000 talents.38 The load of 1000 talents probably corresponded to 26 tons.39 In this passage, Hero expresses poetic wonder created by mechanics in the minds of man, the Greek thaumázein.40 The same limit is also to be found in the third book, where Hero presents his most complex pulley system, which  could lift a load of 1000 talents with a force of five.41 Numerous passages from Hero in Greek have been preserved by Pappus. In particular, he compared the wonder of the Barulkós with Archimedes’s famous statement: “Give me a place to stand on, and I will move the Earth”.42 According to Plutarch, Archimedes invented a políspaston too, a special machine to move singlehandedly a threemasted merchantman that was lying aground, to  the great astonishment of King Hiero of ­Syracuse.43 The compound pulley system, Latin multae trocleae, is an astonishing object because it ‘cuts’ the load, while the tread wheel, Latin tympanum, multiplies a small applied force by the radius of the wheel. In the fourth book of De Rerum Natu­ ra, “The Nature of Things”, dedicated to sensory perceptions, Lucretius writes: Múltaquae pér trocleás/et tímpana póndere mág­ no Cómmovet átque leví/sustóllit máchina nísu

Fig. 3. Haterii funerary temple and a crane, Museo Gregoriano Profano, Vatican City, circa 120 A.D., marble slab 104.1 × 131.9 cm, bas-relief depth 5.7 cm: particular (photo and © Vatican Museums, Vatican City).

to 1. Perhaps there were two tread wheels, placed symmetrically on either side of the two stanchions to keep the crane in balance, which could lift 32 tons.36 Moreover, by using bigger wheels and more resistant ropes, this power could be increased. According to Pliny, the strongest ropes were made of esparto grass, the most famous ­being the spartum of Spain.37 Let us examine the lifting capacity of these machines according to an ancient source. 36  Meighörner-Schardt and Blumenthal 1989: 20-21; Dienel and Meighörner 1997: 26. 37  Pliny, Naturalis Historia, 19.26-30. See also Hübner 1899; Coarelli 2008: 41; Ucelli 1996: 268, 283 note 5.

and a machine by its blocks and tread wheels moves many bodies of great weight and uplifts them with small effort.44

 Hero, Mechanica, 1.1. Drachmann 1963: 22-32.   For the equivalence between talents and kilos: Fleury 1993: 352 Annexe B. In Hero’s Baroulkos, the 5 talents of the child turning the crank probably regard the moment of force exerted by the child’s arm. 40   Martines 1983. 41  Hero, Mechanica, 2.23. Drachmann 1963: 86-87. 42  Pappus, Collectio Mathematica 8, Proposition 10; English trans. by Büttner 2013: 87; also Cuomo 2000: 117. Hultsch 1877: 114-123. 43  Plutarch, Lives, Marcellus 14.7-9. 44   Lucretius, De Rerum Natura 4.905-906; English trans. Rouse. 38

39

Anejos de AEspA LXXVII

LIFTING BLOCKS, 1ST - 5TH CENTURY AD: THE INCLINED PLANE

191

Fig. 4. A study on the Haterii crane.

COLUMNS OF TRAJAN, MARCUS AURELIUS AND ARCADIUS: BLOCKS OF EXCEPTIONAL SIZE In Trajan’s Column, the blocks are exceptional in size (fig. 5):45 the last block of the pedestal weighs over 72 tons, while the capital weighs over 44 tons.46 Of even greater weight were the corresponding blocks used in the Column of Marcus   Martines 2001: Pl. 86 v.   Martines 2000, 75-76 Pl. 1. Martines 2008: 50 with tab.

Aurelius:47 the 4th pedestal block weighs over 95 tons, the capital 79. To lift them, a combination of five of Hero’s machines may have been used, as well as the inclined plane,48 which perhaps was not only used in quarries or for transportation, but also on construction sites. To give us an idea of the weights of the blocks in Trajan’s Column, we can compare them to similar weights in the contemporary world (fig. 6):

45

47

46

48

  Martines 2013: Pl. 1. Beckmann 2011.   Martines 2000: 34-35, 77-78 Pl. 2.

192

GIANGIACOMO MARTINES, MATTHIAS BRUNO, CINZIA CONTI

Anejos de AEspA LXXVII

Fig. 5. Columns of Trajan and Marcus Aurelius: weight of blocks and drums, those over 31 tons in red (from Martines 2000 and 2001).

the capital weighs the same as a big shunting locomotive, and the top drum the same as a small locomotive. These locomotives can only be lifted by exceptionally big cranes, like the Ursus in Trieste (1931)49 or the latest Astaldi gantry cranes (2007).50

Between the drums of the Columns of Trajan and Marcus Aurelius there are dowels sealed by cast lead; in the Column of Arcadius in Costantinople51 there are lewis holes on the torus drum (fig. 7); the same is true of the fourth block of Trajan’s pedestal.52 These facts indicate that the

  Tatò 2004: 74; Angiolini 2007, 24-25, 36; Rosato 2008.   A common crane of Astaldi fleet is: Cimolai Technology spa Rail-Mounted Gantry Crane, lifting capacity 120 tons, cable diameter 20 mm.

51  Becatti 1960: 151-264; Konrad 2001: 319-401; Taddei 2009: 46-56. The survey and photos of Arcadius’ Column were made on September 2012, by Conti and Bruno. 52   Martines 2000: 31 note 39.

49 50

Anejos de AEspA LXXVII

LIFTING BLOCKS, 1ST - 5TH CENTURY AD: THE INCLINED PLANE

Fig. 6. A. A big shunting locomotive, 46 tons, and a small shunting locomotive, 21 tons; B. The Ursus crane, Trieste Harbour, lifting capacity 150 tons (photo G. Nicotera, Trieste); C. The Ursus crane at work (from Tatò 2004).

drums needed to be suspended in order to place them in their final positions. However, this height may have been reached by means of the inclined plane. TWO EXERCISES: BUILDING TRAJAN’S COLUMN The first drawing that follows is an exercise on the inclined plane using counterbalances, following Hero’s treatise and applied to the building of Trajan’s Column (fig. 8). We assume that each drum has been hollowed out in the quarry to make the inner staircase.53 This would make the drum a third lighter than the corresponding solid block. The drum is brought down from the quarry on a sled (fig. 8A), then shipped by sea and   Martines 2000: 19-39; Martines forthcoming.

53

193

along the Tiber until put in place, as if the sled were a bunting bag. Our plane has a 15° slope because it is at this incline that the formulas of Hero and Pappus coincide (fig. 1C). On top, for example, we place the 10th drum, which weighs 28 tons (fig. 8B and E). The two counterbalances needed to raise it are two big boxes laden with small stones, weighing 4 tons each. The calculations are reported in Figure 8E. There may be more than two scaffoldings with counterbalances along the ramp so that shorter ropes may be used. The ramp, which consists of two walls of stone blocks with earth inside, is gradually built up to the height of the drum on top of which the other drum has to be placed (fig. 8F). The greatest height is that of the statue of Trajan, exactly 38.57 m from the bed surface of the column, which requires a ramp with a 144 metre long base (486 pedes), stretching all the way to the façade of Palazzo Valentini in Via IV Novembre. When the two counterbalances reach the ground, (fig. 8F) the 10th drum is raised to its position in the column, on top of the 9th drum (fig. 8C). Now the sled is removed and the drum placed in its definitive position. This ‘hopping’ operation – an expression used by Rabun Taylor –54 is extremely precarious. To place the drum perfectly in its final position we use a crane mounted on the top of an inclined plane (fig. 8F).55 The haulage ropes are pulled by a team of oxen, which move a load that is the same weight as the drum (fig. 8D) – the load might be the next drum to be positioned or the blocks used to contain the earth ramp. As regards the earth ramp and blocks, we must remember that numerous scenes on Trajan’s Column show earthworks and stone blocks (fig.  8F particular), by which means the legion­ aries quickly built a new castrum. We can imagine   Taylor 2003: 115-129, particularly 118.  In figures 8D and F, we designed two distinct and converging masts at the top: they are each as described by Hero in Mechanica 3.2; Drachmann 1963: 97-99. In Hero’s description, the mast is a robust circular tree trunk, reinforced by a tight helix rope, which a worker can climb comfortably. It rests on a wooden base on which it can be swivelled and tilted. In our exercise, each mast has a compound pulley system at the top, like that in figures 4B and C. Its advantage is the ratio between the weight to be lifted and the force used. In our exercise the advantage given by the inclined plane at 15° and that of the 2 cranes on top of the plane are similar, but the first has two decisive advantages over the second: first, it guarantees a greater degree of safety; secondly, the ropes do not have to be so long. 54

55

194

GIANGIACOMO MARTINES, MATTHIAS BRUNO, CINZIA CONTI

Anejos de AEspA LXXVII

Fig. 7. Column of Arcadius, Constantinople, 401-421 A.D.: A. Sketch of the torus drum; B. The torus drum.

that at the time the Column was built, the area to the North was still free of buildings; in fact, the Forum was only completed after the Column’s colossal blocks had been laid. So, as regards the width of the ramp, there would have been no particular limits, except for the two libraries at the sides of the Column, about 25 meters apart; while the widest parts of the Column are the podium, 6.21 m, and the abacus, 4.29.56 As for the structure and the height of the walls to keep the ramp in place, we may assume they were similar to the one in the Forum of Augustus, which is 33.5  m high at the topmost point, and 1.3  m thick,57 a very slender structure yet still stable after so many centuries. The structure is built in parallelepipedal blocks, united by dovetailed oak cramps horizontally,58 and without any dowel vertically.59   Martines 2000: 84-86 Pl. 9.   Bauer 1985: fig. 3. 58   Ganzert 1985: 206-207 and drawing 1, Pl. 79; Ganzert and Kockel 1988: 167 n. 56. 59   Bauer 1985: 230. 56 57

Our exercise has come to an end and we believe we would have been given a good mark by Hero at the Museum of Alexandria, but we do not know what sort of mark we would have got from Apollodorus. In comparing the inclined plane and the other machines, we must forget about the concept of ‘progress.’ “It is apparent from essays by Alistair Crombie and Paolo Rossi that the idea of progress was defined only in the modern age, at about the same time that modern science was being consolidated” (Evandro Agazzi).60 Thus, the inclined plane was evidently not an outdated machine in the age of Trajan, of Constantine, or of Arcadius, but it was the usual means for transporting blocks from quarries, on journeys by land, and perhaps on some construction sites. Second exercise: As regards lifting marble blocks, there are many ways to hook a crane onto 60

  Agazzi 1976: 9; Crombie 1976: 15-36; Rossi 1976.

Anejos de AEspA LXXVII

LIFTING BLOCKS, 1ST - 5TH CENTURY AD: THE INCLINED PLANE

195

Fig. 8. An exercise on Hero’s “mountain”: Trajan’s Column built using an inclined plane.

a drum, but the principal are two: first by means of projecting bosses around the drum,61 like in fig.  8F, and secondly by using lewises placed in  characteristic hollows in the marble block,62 like in fig. 7A. The lewis produces huge shearing stress, like a wedge, which risks breaking the seat of the lewis itself in the marble block. So in fig.  8F, we have drawn projecting bosses on the outer cylinder of the drums as a pure hypothesis. Hero writes about the lewis in his third book,63 just before the inclined plane, warning of the risk of cracking the heavy blocks, but gives no information on the lewis’ safe lifting capacity.64 So, we 61  Orlandos 1968: 16 bis, 87-92; Ginouvès and Martin 1985, I: 121 and Pl. 33.1. 62  Lugli 1957, I: 228-230; Orlandos 1968: 16 bis, 97-98; Ginouvès and Martin 1985, I: 122-123 and Pl. 33.5. 63  Hero, Mechanica 3.5-8. Drachmann 1963: 101-106. 64  In Trajan’s Column, the 4th block of the base has a trapezoidal lewis hole, in the western corner: it is 24 cm deep, 11 wide, with a trapezoid section over 18.5 and under 21 (Martines 2000: 31 note 39). Matthias Bruno found similar holes in the travertine blocks in the façade of the Colosseum, measuring 23, 6, 20, and 23 respectively. A lewis, corresponding to a trapezoidal lewis hole, found on Trajan’s Column, could

carried out a second exercise, using lewises only in a safe and special way (fig. 9). First of all, let us take a close look at the surface of contact between two drums, the 9th and 10th (fig. 9A). Four dowels are inserted into very precise holes at the base of the 10th drum and are then lowered into the corresponding holes on top of the 9th drum.65 The holes underneath are wider and, after assembly, are sealed with a casting of lead, through grooves carved on top of the 9th drum.66 safely raise a capital weighing 4-5 tons, two lewises a lintel weighing 8-10 tons, three a long lintel of 12-15 tons, and four a large drum of 16-20 tons, according to engineers Carriero, Sabbadini, and Peroni. Inglese and Pizzo 2015. 65   On the drums of Trajan’s Column, the dowel holes are square and have the following measurements: on the underside of the drum on top, they are 45-50 mm wide and 105-115 deep; on the topside of the drum underneath, they are 85-110 and 60-75 respectively; the pin, which was perhaps made of bronze, was square. It fitted lightly into the hole above, and was no less that 160 mm high, Martines 2000: 26. 66   Two pins have been preserved on the shaft of Trajan’s Column, with their lead seals, in scenes 24 and 33, Martines 2000: 83 Tab. 7.

196

GIANGIACOMO MARTINES, MATTHIAS BRUNO, CINZIA CONTI

Anejos de AEspA LXXVII

Fig. 9. Second exercise on the construction of Trajan’s Column: drums being laid using lewises, step by step.

Now we describe the process step by step. In figure 9B at the highest point of the inclined plane, the 10th drum is slid off the sled on small rollers and placed on the top of the 9th drum. In figure 9C on the top of the 10th drum, we insert three lewises on the left only, and we tighten the ropes so half of the drum is raised, while the other­ half rests on rollers. In actual fact, we just have to tighten the ropes to lighten the load on the rollers underneath. Once these rollers are removed (fig. 9D), the drum is lowered on one side (fig.  9E), while two of the four dowels fit into their corresponding holes. Figure 9F: The same operation is repeated on the other side to remove the rollers from the centreline. Finally, the drum is placed into its final position (figs. 9G and H). In this way each block always rests on something, from the quarry to its final resting position. Each solution contains a new element: in this case, counterbalances are placed on the other side of the Column to pull the drum up above the one below it and then bring it to rest on top.

Martin Beckmann said that the staircase in Trajan’s Column was the first of its kind in architecture.67 Certainly, the column was the tallest and most slender building of its time and even the way it was raised was perhaps the first of its kind, at least in Rome. If the inclined plane was used for Trajan’s Column (fig. 10)68 certainly up to and including the torus, then the last two lines of the epigraph:69 AD DECLARANDVM · QVANTAE · ALTITVDINIS MONS · ET · LOCVS TANTIS · OPERIBVS · SIT · EGESTVS

  Beckmann 2011: 55-67.   For the ancient topography of the area covered by the inclined plane, see Meneghini 1996: 47-88, particularly 74-78; Meneghini 1998: 127-148, particularly 134-148; Meneghini 2001: 245-268, particularly 246-248; Del Signore 2008. 69   Lepper and Frere 1988: 203-207; Martines 2001: Pls. 74 and 75; Conti forthcoming. 67 68

Anejos de AEspA LXXVII

LIFTING BLOCKS, 1ST - 5TH CENTURY AD: THE INCLINED PLANE

197

ACKNOWLEDGEMENTS

Fig. 10. Hero’s “mountain” on Trajan’s Forum (authors’ drawing on Google Earth).

in order to show how lofty had been the mountain – and the site for such mighty works was nothing less – which had been cleared away.70

may have been used figuratively, and understood as such by those working in the construction site. The mons and the locus might not, therefore, be the ‘saddle’ between the Capitol and Quirinal, i.e. the excavated area between two hills, but simply an ephemeral hill and construction site: namely an eighth hill of Rome built artificially by Apollodorus for Trajan and then removed. In the epigraph, the apparent contradiction of celebrating a demolition would be read in the positive light of an elaborate construction.

  English trans. Lepper and Frere 1988: 20.

70

Vanni Mannucci was an architect working for the Soprintendenza Archeologica di Ostia Antica. He studied Roman construction techniques, introduced new methods of restoration for ruins and mosaics, and worked on safeguarding the Port of Trajan. He would have taken part in this study with us but died prematurely in 1996. We are grateful to professor Pier Gabriele Molari, former Professor of Machine Design at the Alma Mater Studiorum University of Bologna, for discussing our research with us and examining together the Haterii crane before the original bas-relief. We also thank the archaeologists Giandomenico Spinola, Eleonora Ferrazza and Sabina Francini, director and functionaries respectively of the Department of Greek and ­Roman Antiquities, Vatican Museums, for illustrating to us the series of funerary reliefs in the Mausoleum of the Haterii; engineers Alessandra Carriero & Fabio Sabbadini, and Marco Peroni, for discussing the lifting capacity of lewises; Astaldi Company Manager Dott. Olivio Angelini, for information on Astaldi bridge cranes; Professor Filippo Coarelli, who gave us a length of rope made of esparto grass from the mountains of Cil­ ento in Campania; Dott. Luigi Fozzati, an underwater archaeologist and Soprintendente per i Beni Archeologici del Friuli Venezia Giulia, for the information provided on ancient shipping ropes; Professor Mark Wilson Jones, for helping us find figure 2A at the British Library; the Director of the Biblioteca dell’Istituto di Matematica “Guido Castelnuovo” Dr. Adele Piccolomo, for giving us permission to consult the original works of Simon Stevin. Figure 1 is by Giangiacomo Martines; figures 3, 8-10 are by the authors. All the graphics were designed by the architect Filippo M. Martines and the layout is by architect Roberta Zaccara; figure 6B was provided by photographer Giorgio Nicotera of the Soprintendenza per le Belle Arti e il Paesaggio del Friuli Venezia Giulia; figure 6C was provided by the Director of the Arch­ivio di Stato di Trieste Dr. Claudia Salmini, whom we all thank. Figure 7A was designed by Cinzia Conti; the photos in figure 7B are by Matthias Bruno; figure 10 was designed by the a­ uthors using a Google Earth photo. Finally, thanks also to Dr. Esther Barrondo, Escuela Española de Historia y Arqueología en Roma, and Mr Fred Moffa of the British Insti-

198

GIANGIACOMO MARTINES, MATTHIAS BRUNO, CINZIA CONTI

tute of Rome, for their clear and faithful translations, respectively, into Spanish and English. PRIMARY SOURCES Archimedes 1910-1915: Opera, ed. and Latin trans. J. L. Heiberg. Teubner, Leipzig, 3 vols. Galileo Galilei 1890-1909: Le Opere, Edizione Nazionale, ed. A. Favaro. Barbèra, Firenze, 21 vols. Galileo Galilei 1960: On Motion and On Mechanics, English trans. I. E. Drabkin and S. Drake. University of Wisconsin Press, Madison. Hero Alexandrinus 1988: Les Mécaniques, ed. Arabic text and French trans. C. M. B. Carra de Vaux. Imprimerie Nationale, Paris. 1894; Ist. edition Jour­ nal Asiatique 1893. 1: 386-472, and Journal Asia­ tique 1893. 2: 152-192, 227-269, 461-514; reprint Héron d’Alexandrie, Les Mécaniques ou l’Élévateur des Corps Lourds, A. G. Drachmann and D. R. Hill (eds.). Les Belles Lettres, Paris. Hero Alexandrinus 1900: Mechanica, ed. Arabic text and German trans. L. Nix and W. Schmidt. Teubner, Leipzig. Herodotus 1920-1930: Historiae, English trans. A. D. Godley. Harvard University Press, Cambridge, Mass.-London. (Loeb), 4 vols. Lucretius 1982: De Rerum Natura, English trans. W. H. D. Rouse and M. Ferguson Smith. Harvard University Press-Heinemann, Cambridge, Mass.-­ London (Loeb). Pappus Alexandrinus 1876-1878: Collectio Mathe­ matica, ed. and Latin trans. F. Hultsch. Weidmann, Berlin, 3 vols. Pappus Alexandrinus 1933: La Collection mathéma­ tique, French trans. P. Ver Eecke. Desclée de Brouwer, Paris, 2 vols. Pliny 1962: Natural History, Books 36-37, English trans. D. E. Eichholz, Harvard University PressHeinemann, Cambridge, Mass.-London. (Loeb). Plutarch 1914-1926: Vitae Parallelae: Marcellus, English trans. B. Perrin, V. Harvard University PressHeinemann, Cambridge, Mass.-London. (Loeb, 11 vols.).

REFERENCES Adam, J.-P. 2012: “La couverture à dalle monolithique du mausolée de Théodoric à Ravenne”, in Gargiani, R. (ed.), L’architrave, le plancher, la plate-forme. Nou­ velle histoire de la construction, pp. 86-89. Presses poly­techniques et universitaires romandes, Lausanne.

Anejos de AEspA LXXVII

Agazzi, E. (ed.) 1976: Il concetto di progresso nella scienza. Feltrinelli, Milano. Angiolini, E. (ed.) 2007: Ursus. Simbolo dell’archeolo­ gia industriale del mare. Luglio, Trieste. Arnold, D. 1991: Building in Egypt. Pharaonic Stone Masonry. Oxford University Press, New York-­ Oxford. Bauer, H. 1985: “Ricerche sul muro perimetrale e sul portico del Foro di Augusto”, in Roma. Archeologia nel centro. I. L’area archeologica centrale, pp. 229240, Lavori e studi di archeologia pubblicati dalla Soprintendenza Archeologica di Roma 6. De Luca, Roma. Becatti, G. 1960: La Colonna coclide istoriata. Proble­ mi storici iconografici stilistici. “L’Erma” di Bretschneider, Roma. Beckmann, M. 2011: The Column of Marcus Aurelius. The Genesis and Meaning of a Roman Imperial Monument. University of North Carolina Press, Chapel Hill. Blümner, H. 1884: Technologie und Terminologie der Gewerbe und Künste bei Griechen und Römern. Teubner, Leipzig. Bruno, M., Gorgoni, C. and Pallante, P. 1998: “I marmi dell’Arco di Settimio Severo: composizione struttrale, volumetria e analisi archeometriche”, in Pensabene, P. and Panella, C. (eds.), Arco di Costan­ tino tra archeologia e archeometria, pp. 157-169. “L’Erma” di Bretschneider, Roma. Büttner, J. 2013: “The lever, the balance a beautiful proof ”, in Di Pasquale, G. and Parisi Presicce, C. (eds.), Archimedes. The Art and Science of Invention, pp. 86-91. Giunti, Firenze. Choisy, A. 1904: L’art de bâtir chez les Égyptiens. Rouveyre, Paris (reprint Forni, Sala Bolognese 1977). Clarke, S. and Engelbach, R. 1990: Ancient Egyptian Construction and Architecture. Dover, New York (1st edition Oxford University Press, London 1930). Coarelli, F. 2008: La Colonna di Marco Aurelio. The Column of Marcus Aurelius. Colombo, Roma ­(English trans. H. Patterson). Clayton, P. A. and Price, M. J. 1988: The Seven Won­ ders of the Ancient World. Routledge, London. Cohen, M. R. and Drabkin, I. E. 1958: A Source Book in Greek Science. Harvard University Press, Cambridge, Mass. Conti, C. Forthcoming: “Colonna Traiana: ordinatio dell’epigrafe”, in Ferrandes, A. F. and Pardini, G. (eds.), Le regole del gioco. Tracce Archeologi Rac­ conti. Studi in onore di Clementina Panella. Quasar, Roma. Coulton, J. J. 1974: “Lifting in early Greek Architecture”, The Journal of Hellenic Studies, 94: pp. 1-19.

Anejos de AEspA LXXVII

LIFTING BLOCKS, 1ST - 5TH CENTURY AD: THE INCLINED PLANE

Crombie, A. 1976: “Alcuni atteggiamenti nei confronti del progresso scientifico: Antichità, Medioevo, inizi dell’Era moderna”, in Agazzi, E. (ed.), Il concetto di progresso nella scienza, pp. 15-36. Feltrinelli, Milano. Cuomo, S. 2000: Pappus of Alexandria and the Mathe­ matics of Late Antiquity. Cambridge University Press, Cambridge-New York-Melbourne-Madrid. Cureton, W. and Rieu, Ch. 1871: Catalogus codicum manuscriptorum orientalium qui in Museo Britannico asservantur. Pars secunda. Codices Arabici Muham­ madani. Museo Britannico, London. D’Alessandro, P. and Napolitani, P. D. 2013: “Archimedes in the 12th-16th Centuries”, in Di Pasquale, G. and Parisi Presicce, C. (eds.), Archimedes. The Art and Science of Invention, pp. 136-143. Giunti, Fi­ renze. Del Signore, R. 2008: Palazzo Valentini. L’area tra an­ tichità ed età moderna: scoperte archeologiche e pro­ getti di valorizzazione. Ediart-Provincia di Roma, Roma. Dienel, H.-L. and Meighörner, W. 1997: Der Tre­ tradkran. Deutsches Museum, München. Downey, G. 1947-1948: “Pappus of Alexandria on architectural Studies”, Isis, 38, pp. 197-200. Drachmann, A. G. 1963: The mechanical Technology of Greek and Roman Antiquity. A Study of the Liter­ ary Sources. Munksgaard-University of Wisconsin Press-Hafner, Copenhagen-Madison-London. Duhem, P. 1905: Origines de la Statique. Hermann, ­Paris, 2 vols. Fleury, Ph. 1993: La mécanique de Vitruve. Presses Universitaires de Caen, Caen. Galluzzi, P. 1989: Leonardo e i proporzionanti, Letture Vinciane 28. Giunti Barbèra-Biblioteca Leonardiana-Comune di Vinci, Firenze-Vinci. Ganzert, J. 1985: “Der Mars-Ultor-Tempel auf dem Augustusforum”, Römische Mitteilungen, 92, pp. 201-219. Ganzert, J. and Kockel, V. 1988: Augustusforum und Mars-Ultor-Tempel, in Hofter, M., Heilmeyer, W.D., La Rocca, E. and Martin, H. G. (eds.), Kaiser Augustus und die verlorene Republik, pp. 149-199. Staatliche Museen Preußischer Kulturbesitz, Berlin. Ginouvès, R. and Martin, R. 1985-1988: Dictionnaire méthodique de l’architecture grecque et romaine, Collection de l’École Française de Rome 84. Boccard-“L’Erma” di Bretschneider, Roma, 3 vols. Golvin, J.-C. and Goyon, J.-C. 1987: Les bâtisseurs de Karnak. CNRS, Paris. Hübner, E. 1899: “Campus spartarius”, in Wissowa, G. (ed.), Paulys Real-Encyclopädie der classischen Altertumswissenschaft, III. 2, pp. 1447-1448. Metzlerscher, Stuttgart.

199

Hultsch, F. 1877: “De Heronis Mechanicorum reliquiis in Pappi collectione servatis”, in Commentationes philologae in honorem Theodori Mommseni scripse­ runt amici. Weidmann, Berolini. Inglese, C. and Pizzo, A. 2015: I tracciati di cantiere: disegni esecutivi per la trasmissione e diffusione delle conoscenze tecniche. Gangemi, Roma. Konrad, C. B. 2001: “Beobachtungen zur Architektur und Stellung des Säulenmonumentes in Istanbul-­ Cerrahpașa-‘Arkadiossäule’ ”, Istanbuler Mitteilun­ gen, 51: pp. 319-401 Korres, M. 1997: “Wie kam der Kuppelstein auf den Mauerring? Die einzigartige Bauweise des Grabmals Theoderichs des Grossen zu Ravenna und das Bewegen schwerer Lasten”, Römische Mitteilungen, 104: pp. 219-258. Lancaster, L. 1999: “Building Trajan’s Column”, American Journal of Archaeology, 103: pp. 419-439. Lepper, F. and Frere, S. 1988: Trajan’s Column. A New Edition of the Cicorius Plates. Sutton, Glou­ cester. Lugli, G. 1957: La tecnica edilizia romana con partico­ lare riguardo a Roma e Lazio. Bardi, Roma, 2 vols. Mach, E. 1883: Die Mechanik in ihrer Entwickelung his­ torisch-kritisch dargestellt. Brockhaus, Leipzig. Meighörner-Schardt, W. and Blumenthal, H. 1989: Räderwerk Römerkran. Bouvier, Bonn. Martines, G. 1983: “La struttura della Colonna Traiana: un’esercitazione di meccanica alessandrina”, Prospettiva, 32, pp. 60-71. Martines, G. 1998-1999: “Macchine da cantiere per il sollevamento dei pesi, nell’antichità, nel Medioevo, nei secoli xv e xvi”, Annali di architettura, 10-11, pp. 261-275. Martines, G. 2000: “L’architettura”, in Scheid, J. and Huet, V. (eds.), Autour de la Colonne Aurélienne, pp. 19-88, Bibliothèque de l’École des Hautes Études. Sciences Religieuses 108. Brepols, Turnhout. Martines, G. (ed.) 2001: Colonna Traiana. Corpus dei disegni 1981-2001. Quasar, Roma. Martines, G. 2008: “Les tambours de la colonne de  Trajan à Rome”, in Gargiani, R. (ed.), La co­ lonne. Nouvelle histoire de la construction, pp. 42-53. ­Presses polytechniques et universitaires romandes, Lausanne. Martines, G. (ed.) 2013: Columna Divi Marci. Corpus dei disegni 1981-1996. Modus, Roma. Martines, G. Forthcoming: “Description of the structure”, in Mitthof, F. and Schörner, G. (eds.), Co­ lumna Traiani – Trajan’s Column: A Victory Monu­ ment and War Report in Pictures. International Conference Marking the 1900th Anniversary of its Dedication (Vienna, May 9-12, 2013). Historisch-

200

GIANGIACOMO MARTINES, MATTHIAS BRUNO, CINZIA CONTI

Kulturwissenschaftliche Fakultät Universität Wien, Wien. Meneghini, R. 1996: “Templum Divi Traiani. Con appendici di L. Billi, S. Coletti, M. Milella”, Bullettino della Commissione Archeologica Comunale di Roma, 97, pp. 47-88. Meneghini, R. 1998: “L’architettura del Foro di Traiano attraverso i ritrovamenti archeologici più recenti”, Römische Mitteilungen, 105, pp. 127-148. Meneghini, R. 2001: “Il Foro di Traiano. Ricostruzione architettonica e analisi strutturale. Con appendici di A. Ballarin, G. Berti e E. Bianchi”, Römische Mitteilungen, 108, pp. 245-268. Moody, A. and Clagett, M. (eds.) 1952: The Medieval Science of Weight (Scientia de Ponderibus). Trea­ tises ascribed to Euclis, Archimedes, Thâbit ibn Qur­ ra, Jordanus de Nemore and Blasius of Parma, English trans. University of Wisconsin Press, Madison. Nicholson, P. and Shaw, I. 2000: Ancient Egyptian Materials and Technology. Cambridge University Press, Cambridge-New York-Melbourne-Madrid. Orlandos, A. K. 1966-1968: Les matériaux de construction e la technique architecturale des anciens Grecs, Ècole Française d’Athènes. Travaux et Mémoires 16 and 16 bis. De Boccard, Paris. Ormea, G. B. 1988: Manuale pratico per l’ingegnere ci­ vile. Kappa, Roma, 2 vols. Parra Ortiz, J. M. 2008: Historia de las pirámides de Egipto. Complutense, Madrid. Renn, J. 2013: “The equilibrium controversy”, in Di Pasquale, G. and Parisi Presicce, C. (eds.), Archime­ des. The Art and Science of Invention, pp. 144-149. Giunti, Firenze Rosato, G. (ed.) 2008: Langer Heinrich dal 1915 a oggi. Erga, Genova.

Anejos de AEspA LXXVII

Rossi, P. 1976: “Sulle origini dell’idea di progresso”, in Agazzi, E. (ed.), Il concetto di progresso nella scien­ za, pp. 37-87. Feltrinelli, Milano. Sinn, F. and Freyberger, K. S. 1996: Die Grabdenk­ mäler, II. Die Ausstattung des Hateriergrabes (Vati­ kanische Museen. Museo Gregoriano Profano ex La­ teranense). Von Zabern, Mainz. Sinopoli, A. 2015: Il problema dell’equilibrio da Aristo­ tele a Varignon. Franco Angeli, Milano. Stevin, S. 1605: Hypomnémata Mathématica, IV, De Sta­ tica. Ex officina Ioannis Patii, Lugodini Batavorum. Taddei, A. 2009: “La Colonna di Arcadio a Costantinopoli. Profilo storico di un monumento attraverso le fonti documentarie dalle origini all’età moderna”, Nea Rhome, 6, pp. 46-56. Tatò, G. (ed.) 2004: L’evoluzione delle strutture portuali della Trieste moderna tra ‘800 e ‘900. Archivio di Stato di Trieste-Biblioteca Statale di Trieste-Soprintendenza archivistica per il Friuli Venezia Giulia, Trieste. Taylor, R. 2003: Roman Builders. A Study in Architec­ tural Process. Cambridge University Press, Cambridge-New York. Ucelli, G. 1996: Le navi di Nemi. Istituto Poligrafico e Zecca dello Stato, Roma (1st. ed. 1940). Varignon, P. 1725: Nouvelle Mécanique ou Statique dont le projet fut donnée en 1687. Ouvrage posthume de M. Varignon. Jombert, Paris. Wirsching, A. 2007: Obelisken trasportieren und auf­ richten in Ägypten und in Rom. GmbH, Norderstedt. Wirsching, A. 2000: “How the obelisks reached Rome. Evidence of Roman double-ship”, The International Journal of Nautical Archaeology, 29.2, pp. 273-283. Zignoli, V. 1951-1956: “Fune”, in Perucca, E. (ed.), Dizionario d’ingegneria, III, pp. 96-99. Utet, Torino, 5 vols.

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

THE RESTORATION OF THE COLUMNS OF THE TEMPLUM CASTORIS DURING VERRES’ PRAETORSHIP: THE MACHINA AND ORGANISATION OF THE BUILDING SITE PAULINE DUCRET Université Paris 8 - Vincennes - Saint Denis

ABSTRACT: During his praetorship, Verres was in charge of reviewing the restoration work of the temple of the Castores in the Forum Romanum. As he required the builders to straighten some columns, a new building site was initiated. According to Cicero, the work was done by the use of a “machina” which allowed the builders to have these columns taken down and replaced. The aim of this study is to figure out which kind of machine was used there, a machine which had to be more complex than the ones described in theoretical texts like Vitruvius’ De Architectura, and, furthermore, to understand how the use of such a machine impacted the organisation of the building site: it appears that a specialised team was specifically employed to supply and use the machine, and this particularity may have changed the terms of the contract with the builders. KEYWORDS: Roma, Roman Republic, Temple of the Castores, Cicero, Machine, Building site, Skilled workers. RESUMEN: Durante su pretura, Verres se encargaba de la revisión de los trabajos de restauración del templo de los Castores en el Foro Romano. Tras su requerimiento a los constructores para enderezar algunas columnas, se inició una nueva obra de construcción. Según Cicerón, el trabajo se realizó con el uso de una machina que permitió a los constructores remover estas columnas y reemplazarlas. El objetivo de este estudio es averiguar qué tipo de máquina se utilizó en la obra, una máquina que tuvo que ser más compleja de las que se describen en los textos teóricos como el De Architec­ tura de Vitruvio. Por otra parte, se pretende entender cómo el uso de esta máquina afectaba a la organización de la obra: es probable que un equipo especializado se empleara específicamente para el suministro y el uso de la máquina, particularidad que puede haber cambiado los términos del contrato con los constructores. PALABRAS CLAVE: Roma, República, Templo de los Castores, Cicerón, máquina, obra, mano de obra especializada.

Very little literary evidence exists regarding how machines were used for construction during the Roman period, and even less about how these building sites were organised. Vitruvius and Hero describe various types of lifting machines which can be used in construction, but these remain at a theoretical level.1 Only Cicero offers a brief de1   Vitruvius devotes the tenth book of his De Architectura to the use and construction of machines, with a special chapter on lifting machines. As for Hero of Alexandria, he describes in his Mechanica different ways of lifting heavy objects.

scription of the effective use of a “machine” in a building context, in his Second Speech against Verres.2 During his praetorship, Verres was in charge of checking the restoration work on the ­Temple of the Castores, in the Forum Romanum. In 74 BC, the Temple of the Castores currently standing was the second version of the temple, which L. Caecilius 2  Cicero, Second Speech against Verres. 1st Book: the Urban Praetorship, L-LVII, 130-150.

202

Pauline Ducret

Metellus Dalmaticus had rebuilt in 117 BC. According to Cicero, when Verres checked the work on this temple, the restoration was finished and the work of high quality. Verres, however, decided to reject it and to require new work, allegedly to make some of the columns plumb. This actually provided an occasion for one of the numerous extortions committed by Verres. That is why Cicero relates in some detail this restoration work on the columns, giving us an example of a machine being used in a building context. THE WORK AND THE USE OF THE MACHINA According to Cicero, the purpose of this new work was both to make four of the columns of the temple plumb, and also to re-whiten them: Etenim quid erat operis? Id quod vos vidistis. Omnes illae columnae, quas dealbatas videtis, machi­ na apposita, nulla impensa deiectae iisdemque lapidi­ bus repositae sunt. How much, after all, was there to do? Exactly what you yourselves, gentlemen, saw done. All these columns, that you can see freshly whitened, were taken down and replaced by means of a machine leaning against them, without further expense and with the same stones.3

Nielsen, who excavated and studied the Republican temple, suggests that these could have been columns from the western colonnade (fig. 1), as the temple was built on a swampy area and could have sunk towards the north-west (Nielsen 1992: 114). We may be able to identify even more precisely which columns were restored. The expression “quas dealbatas videtis” (“that you can now see freshly whitened”) might simply refer to the fact that the Senators used to see these columns – belonging to one of the most important temples of the republican Forum Romanum – almost daily, but, had the trial not been cancelled, the members of the jury would have been sitting in the tribunal of this particular temple when listening to Cicero’s speech; thus, we understand from this expression that Cicero wanted to show to the Patres the exact columns which had been restored. Only the 3  Cicero, 2 Verr. I, LV, 145. All translations are freely adapted from the Loeb edition.

Anejos de AEspA LXXVII

first columns of the north-west colonnade were not hidden by the wall of the cella and so visible to the Senators during the speech. The columns that were restored were therefore probably the four first ones. Cicero insists that the main task for the builders in this work was to remove and reset three of these columns, stone by stone, which implied the use of a special machine. To get an idea of the type of machine which could have been used in this work, we need to evaluate the average weight of each “stone”. Precise information about the columns of the Metellan temple is missing, but we can compare our temple with that built in Cora around 100 BC which, according to Nielsen, reflected the temple at Rome, although it was smaller in size (Nielsen 1992: 114 and 124). The columns of the Cora temple were 99 cm in diameter, and 8.70 m high; they were Corinthian columns made of travertine, as probably were those of the Metellan temple at Rome. We can thus calculate the minimum weight that the machine had to lift (fig. 2). The columns probably exceeded 1 m in diameter, and were at least 9 m high. That means that each column weighed at least 17 tons. If the columns were monolithic, the machine had thus to be able to lift a little this weight. However, the text implies that the columns were not monolithic but made of several drums (we do not know the exact number). So, for example, if they were made of three different drums, that would have meant that the minimum weight to lift was about 5.7 t; and with a maximum of six drums, the weight of each stone would have been 2.8 t. These are only estimated figures, but they allow us to assess what kind of machine the builders could have used. As a matter of fact, Vitruvius provides a description of three different categories of lifting machine used in construction. According to Ph. Fleury, the only one of these which could have lifted such heavy loads, would have been a crane with at least a double hoisting cable and a drum or treadmill (Fleury 1993: 96-112). This may be the type of machine being used here, even if we cannot be sure about this point due to the lack of evidence on this subject. However, other difficulties remain. This kind of machine could lift heavy loads, but required a lot of space to be operated. First of all, several cables had to be fixed into the ground for the machine, which could have been problematic in

Anejos de AEspA LXXVII

THE RESTORATION OF THE COLUMNS OF THE TEMPLUM CASTORIS...

203

Fig. 1. The late republican Temple of the Castores (Nielsen 1992: 114).

Minimum density of travertine: 2500 kg/m3 Minimum volume of one single column: V = p × 0.502 × 8.7 ≈ 6.8 m3 8.7 m

Fig. 2. Evaluation of the minimum weight of each drum the machine had to lift.

Minimum weight of one single column: M ≈ 6.8 × 2500 ≈ 17 t Minimum weight of one single drum whether 3 drums ≈ 5.7 t whether 6 drums ≈ 2.8 t 1.0 m

such a busy area as the forum. Even more problematic may have been the probable continued existence of an entablature and the roof of the colonnade, which would have prevented the stones from being placed vertically directly onto the top of the columns by the machine. The block and tackle cannot have been anywhere but above the lifted stone, so that we can hardly imagine the last drums being placed between the rest of the column and the roof (or the archi-

trave); unless the “machina” was not only a machine but a whole system combining a crane with shoring and scaffolding. The stones could have been first lifted up to the right level by the machine and then, thanks to the scaffolding, moved horizontally to their exact place in the column, while the shoring was maintaining the rest of the column. Finally, the scaffolding could have been used by the plasterers  who had to stucco the four columns once plumbed. In short,

204

Pauline Ducret

this restoration work required a highly complex machine, even if we do not know what it was exactly, and the workers had to adapt it to suit the constraints of the site. Therefore, only workers with specific skills could have set and used such a machine. A TEAM SPECIALISED IN THE USE OF THE “MACHINE” The text implies that a team was specifically employed to work with the machine: [...] tantum operis in ista locatione fuit quantum pau­ cae operae fabrorum mercedis tulerunt, et manuspre­ tium machinae. [...] there was no more in this contract than the wage of some workers and the cost of the skilled labour for the machine.4

Cicero clearly differentiates two main costs related to the builder: the salary of some “fabri”, probably mainly masons and plasterers, and the labour costs linked to the machine. The former could have been day labourers, since they seem to have been paid separately (mercedis fabrorum) but the latter were paid altogether (manuspretium). Thus, they were probably a team, perhaps from a workshop specialised in the use of machines in a construction context. In fact, another text by Cicero implies the existence of such workshop teams specialised in specific building tasks:

Anejos de AEspA LXXVII

whole work.6 Therefore, Cillo’s must have been a secondary role, and he was probably employed to undertake only part of these works. His team seems to have been digging a tunnel in Venafrum when the accident occurred; they could thus have been engaged by Cicero and Quintus for the same kind of task, probably to execute the dug parts of the aqueduct. Furthermore, Cicero mentions “conservi” and “discipuli” working with Cillo. The first term implies that they were all slaves, including Cillo himself, belonging to the same familia; as for the second one, which refers to pupils formed by Cillo and his companions, it shows that the workshop was organised under a strict hierarchy. Cillo was thus in charge of a team of skilled slaves belonging to an owner who does not appear in our text. On the contrary, Cicero seems to have dealt directly with Cillo without asking the permission of his master but we cannot know from this text alone if they were directly employed by Quintus, or lent or let to him by the owner.7 Concerning the work on the Temple of the Castores, the team in charge of the machine could have been quite comparable to Cillo’s workshop, although we have no information about the workers’ status. Indeed, both of these teams were employed for specific tasks on a much bigger building site; Cillo’s team was specialised in digging tasks, whereas that employed for the Temple of the Castores may have specialised in building machines. As a matter of fact, the machine was provided by this specific team, as Cicero highlights that no timber was supplied by the main builder: Nam illo non saxum, non materies advecta est.

Cillonem arcessieram Venafro, sed eo ipso die quattuor eius conservos et discipulos Venafri cunicu­ lus oppresserat. I had sent for Cillo from Venafrum, but, that very day at Venafrum, a tunnel had fallen in, crushing four of his fellows and pupils.5

We are in a private context, as Cicero is writing to his brother Quintus about one of his brother’s properties. In this villa, Cicero and Quintus are planning to build an aqueduct, and the man referred to as Cillo was clearly summoned by the orator for this purpose. However, the brothers had already contracted with a builder for the

 Cic., 2 Verr. I, LV. 147  Cic., Q. fr., III, 1, 3

4 5

That was all: there was no stone and no timber brought there.8

That means that the workshop had its own machines and scaffolding, which were brought to the building sites where they were erected according to the specific constraints of the work. Therefore the specialised workshop probably supplied both the machine and the labour.

6   This builder, employed as a redemptor by Cicero’s brother, was named Mescidius. Cicero cites him twice in this letter, both times regarding the construction of a private aqueduct, but in two different villas (Q. fr., III, 1, 1-3). 7   We have a similar case in Att., XIV, 3, 1: Cicero writes in this letter that he is awaiting an architect named Corumbus in his Tusculan villa where work was then in progress. This Corumbus is one of Cicero’s friends’ slaves. 8  Cic., 2 Verr. I, LV. 147.

Anejos de AEspA LXXVII

THE RESTORATION OF THE COLUMNS OF THE TEMPLUM CASTORIS...

LEX OPERIS (CONTRACT) AND THE EMPLOYMENT OF SPECIALISED TEAMS The fact that this team was employed separately from the rest of the labour can explain a clause in the lex operis, the contract established between Verres and the builder regarding the restoration of the columns, which has clearly proven problematic to the editors: Si pupillo opus redimitur, mihi praeda de manibus eripitur. Qui est igitur remedium? Quod? Ne liceat pupillo redimere. (...) Operae pretium est legem ip­ sam cognoscere (...). Lex operi facivndo. Qvi de L. Marcio M. Perpenna censoribvs... socivm ne admittito neve partem dato neve redimito. If the ward secures the contract, my prey is snatched from my grasp. How do we stop that, then? How? Why, let us prohibit the ward from bidding for it. (...) It is worth your while, gentlemen, to note the text of the contract (...) Text of the Contract. “The one who from the censors Lucius Marcius and Marcus Perperna... must not take a partner nor give or sublet a part of the works.9

The global meaning of the text is clear: Junius, “the ward”, was forbidden to bid for the contract because of this clause, whose purpose consists precisely in excluding him from the work (“let us prohibit the ward from bidding for it”). He is thus “the one who from the censors...” (L. Marcius and M. Perperna were censors in 86); in other words, the builder in charge of the previous restoration work on the temple which Verres rejected (a verb is definitively missing). However, the second part of the clause is far less clear. Some scholars, following the Loeb edition, translate “must not take him as partner nor allow him to share in the undertaking nor himself secure the contract”. Nevertheless, it is grammatically difficult, as the three verbs are in the active form. Furthermore, this translation implies that the subject changes between the second and third verbs, which is most unlikely.10 Thus, to understand this clause without altering the text or by Cic., 2 Verr. I, LV. 142-143.   Some editors thus add “eum” before “socium”, and “ei” twice after “neve”. The Loeb edition does not accept these emendations, but explains its translation as follows: “It can only be conjectured that the complete clause added by Verres prohibited in general terms the exact action now taken by the guardians of Iunius” (Loeb edition, 1928: 276-277).  9 10

205

passing the grammar, it is necessary to consider that the works done on the columns required the employment of a specialised team because of the machine. Therefore, the builder in charge of the work on the columns had to either have such a team in his possession or hire one. The clause proposes three different contractual possibilities for this second option: to “take a partner”, to “give a part of the works” or to “sublet a part of the works”. As Junius was personally forbidden to do so, and did not seem to have had such specialised workers in his possession, he was simply unable to perform the required work on the ­columns. In conclusion, thanks to Verres, who organised an extortion based on these works, and Cicero, who was never reluctant to describe Verres’ actions in detail when they were not completely legal, here is a very rare piece of evidence of the use of a machine in a building context. This text is however more than a simple testimony to the Roman capability of using complex engineering solutions on building sites. In fact, if we are correct in our interpretation, the exclusion clause shows how important subletting or association could be in the building economy, as an ability to resort to it could even exclude someone from the work. Cicero names only important people in his speech: members of the upper class of Roman society or contractors linked to them but, finally, it was the team of workers, perhaps of slaves, who were the key in this restoration work. Without them, and without their skills and the materials they brought with them, nothing could be performed, so it is a pity that we have no more information about them or their exact status, and that we cannot tell whether they belonged to an independent workshop or to the family of an important house. PRIMARY SOURCES Callebat, L. (ed. trad.) 1986: Vitruvius, De Architectu­ ra X. Les Belles Lettres, Paris. Greenwood, L. H. G. (trad.) 1928: Cicero VII, The Ver­ rine Orations, Volume I, Against Caecilius. Against Verres Part 1; Part 2, Books 1-2. Loeb Classical Library 221. Harvard University Press, Cambridge, Mass. Shackleton Bailey, D. R. (ed. trad.) 1999: Cicero XXIX, Letters to Atticus, Volume IV. Loeb Classical

206

Anejos de AEspA LXXVII

Pauline Ducret

Library 491. Harvard University Press, Cambridge, Mass. Shackleton Bailey, D. R. (ed. trad.) 2002: Cicero XXVIII, Letters to Quintus and Brutus. Letter Frag­ ments. Letter to Octavian. Invectives. Handbook of Electioneering: Loeb Classical Library 462. Harvard University Press, Cambridge, Mass.

REFERENCES Fleury, P. 1993: La mécanique de Vitruve. Centre d’études et de recherche sur l’antiquité, Caen. Nielsen, I. (ed.) 1992: The Temple of Castor and Pol­ lux. The Pre-Augustan Temple Phases with Related Decorative Elements. De Luca, Rome.

VI PRACTICAL SOLUTIONS TO ENGINEERING PROBLEMS

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

COSTRUIRE IN TERRENI PALUDOSI: SISTEMI DI FONDAZIONE E BONIFICA IN USO IN ETÀ ROMANA IN ITALIA SETTENTRIONALE FRA TRADIZIONE E INNOVAZIONE CATERINA PREVIATO Dipartimento dei Beni Culturali: Archeologia, Storia dell’Arte, del Cinema e della Musica Università degli Studi di Padova (Italy)

RIASSUNTO: Fino ad oggi i sistemi di fondazione in uso in età antica sono stati raramente analizzati in modo approfondito, nonostante il loro ruolo di primaria importanza nelle dinamiche strutturali degli edifici. Il presente contributo mira a colmare in parte questa lacuna, e si pone l’obiettivo di fornire una panoramica dei sistemi di fondazione e bonifica adottati in età romana in Italia settentrionale, area occupata per la maggior parte da una pianura alluvionale caratterizzata da terreni paludosi e cedevoli e dalla presenza diffusa di acque superficiali e sotterranee. L’analisi dei sistemi di fondazione diffusi in questa regione mette in luce come le maestranze attive in età romana furono in grado, da un lato, di introdurre innovazioni tecniche importate da altri contesti, dall’altro di adottare le tradizioni costruttive locali, sviluppate nel corso del tempo per risolvere le problematiche legate alle caratteristiche del contesto geo-ambientale, avendo come unico scopo quello di ottenere edifici solidi e duraturi. PAROLE CHIAVE: Fondazioni, Terreni paludosi, Bonifica geotecnica e idraulica. ABSTRACT: Until now, the foundation systems used in ancient times have only rarely been closely examined, despite their key role in the structural dynamics of buildings. This paper aims at filling this gap by way of the analysis of foundations and reclamation systems that were used in the Roman period in Northern Italy, a region occupied for the most part by an alluvial plain characterized by soft wet soils with a low bearing capacity, and by the presence of numerous rivers, marshes and a high ground-water level. The analysis of the foundation systems of this region shows that the builders of the Roman period were able, on the one hand, to introduce technical knowledge imported from other contexts and, on the other hand, to adopt local building techniques, intentionally developed to solve problems linked to features of the terrain, with the aim of obtaining solid and lasting constructions. KEYWORDS: Foundation systems, Wetland, Geotechnical and hydraulic reclaim. RESUMEN: Los sistemas de cimentación empleados en la Antigüedad se han estudiado raramente en profundidad, a pesar de la importancia fundamental en las dinámicas estructurales de los edificios. Esta contribución intenta llenar, en parte, este vacío y tiene como objetivo ofrecer un panorama general de los sistemas de cimentación y drenaje utilizados en época romana en Italia septentrional, área ocupada en su mayoría por terrenos aluviales y pantanosos y por la presencia de agua superficial y subterránea. El análisis de los sistemas de cimentación difundidos en esta región pone en evidencia como la mano de obra activa en época romana fue capaz, por un lado, de introducir novedades técnicas importadas de otros contextos y, por otro lado, de emplear tradiciones constructivas locales desarrolladas al momento para solucionar problemas ligados con las características geo-ambientales del territorio, con el objetivo de obtener edificios sólidos y duraderos. PALABRAS CLAVE: cimentaciones, terrenos aluviales, drenaje geo-técnico e hidráulico.

210

Anejos de AEspA LXXVII

Caterina Previato

In epoca antica era conoscenza diffusa e acquisita che la stabilità, la solidità e la durata nel tempo di un edificio erano garantite innanzitutto dalla presenza di fondazioni costruite in modo adeguato. Altrettanto noto era il fatto che, per realizzare fondazioni in grado di sostenere il peso di un edificio, in fase di progettazione era necessario tenere conto del contesto ambientale in cui ci si trovava ad operare, e in particolare del tipo di terreno su cui si andava a costruire, nonché della sua capacità portante. Entrambi questi concetti emergono in più parti del trattato di Vitruvio, che fornisce varie indicazioni su come realizzare le fondazioni.1 Egli fa un’importante distinzione tra edifici costruiti in zone ove è possibile raggiungere il solidum, cioè il terreno compatto e resistente,2 ed edifici realizzati in zone palustri o su terreni di riporto.3 Nel primo caso, l’autore afferma che è sufficiente che le fondazioni raggiungano il terreno solido e che eventualmente si approfondiscano al suo interno, fino ad una profondità proporzionale alle dimensioni dell’edificio soprastante.4 Nel secondo caso invece, Vitruvio prescrive che prima della posa delle fondazioni si provveda a delle operazioni preliminari, funzionali a consolidare il suolo. In particolare, egli consiglia che il terreno su cui poggerà l’edificio sia scavato, vuotato e confitto con pali di legno di salice, di olivo o di rovere bruciati, disposti in modo molto fitto, e che lo spazio tra i pali sia riempito con carboni.5 Questa tec1   Vitr. 1, 3, 2 (concetto di firmitas); 1, 5, 1; 1, 5, 3; 1, 5, 6; 1, 5, 7 (fondazioni delle mura); 2, 9, 10 (palificate lignee); 3, 4, 1-2 (fondazioni dei templi); 5, 3, 3 (fondazioni dei teatri); 5, 12, 4-5 (fondazioni di moli e strutture portuali); 6, 7, 7 e 6, 8, 1 (fondazioni di edifici privati). 2   Il termine solidum viene più volte utilizzato da Vitruvio quando parla di fondazioni (Vitr. 1, 5, 1; 3, 4, 1; 6, 8, 1). Secondo alcuni studiosi la regola di scavare fino a raggiungere il terreno “solido” deriverebbe da un’indicazione contenuta nell’opera Mechanikè syntaxis di Filone di Bisanzio: l’espressione ad solidum corrisponderebbe infatti all’espressione greca mechrì pétras, cioè “fino alla roccia”. È stato ipotizzato però che con il termine solidum Vitruvio indichi non tanto la roccia, quanto più il suolo compatto argilloso (cfr. Gros 1990: 127; Fleury 1990: 135; Galliazzo 1994: 333; Gros et al. 1997: 91, nota 212 e 321, nota 140; Nardelli 2003: 946, in particolare nota 22). 3   Vitruvio effettua questa distinzione quando tratta delle fondazioni dei templi (Vitr. 3, 4, 1-2). La necessità di realizzare le fondazioni in modo diverso a seconda del contesto in cui ci si trova ad operare (montagna / pianura o zona paludosa) ritorna anche nel libro V, in relazione alle fondazioni dei teatri (Vitr. 5, 3, 3). 4   Vitr. 1, 3, 2; 1, 5, 1 e 3, 4, 1-2. 5   Vitr. 3, 4, 1-2. L’impiego di carbone a livello di fondazione è attestato già nel vi secolo a.C. nell’Artemision di Efeso, le cui fondazioni, come ricordato da Plinio, poggiavano su uno strato di carbone cui era sovrapposto un livello di velli di lana (Plin. nat. XXXVI, 95).

nica di consolidamento e costipamento dei terreni paludosi per mezzo di pali lignei è ricordata anche in un altro passo del trattato vitruviano, in cui l’autore esalta le proprietà del legno di ontano, utilizzato nelle fondazioni degli edifici della città di Ravenna, dove omnia opera et publica et privata sub fundamentis eius generis habeant palos.6 In altri punti del suo trattato l’autore fornisce precise indicazioni sulle modalità di realizzazione delle fondazioni, che devono essere costruite con materiali accuratamente selezionati e impiegati senza risparmio di mezzi,7 e che devono avere solidissima structura8 e profondità e spessore proporzionali alle dimensioni delle strutture soprastanti.9 Precise prescrizioni circa le modalità con cui dovevano essere realizzate le fondazioni degli edifici si ritrovano anche in un altro trattato, e cioè nel libro I dell’Opus agriculturae di Emiliano Palladio, in cui l’autore fornisce dettagliate indicazioni sulle dimensioni delle fondazioni, che devono avere uno spessore maggiore rispetto alle strutture soprastanti (mezzo piede in più per lato), e sulla profondità delle trincee di fondazione, che varia a seconda del tipo di terreno su cui ci si trova a costruire.10 Un accenno alla necessità di realizzare le fondazioni in modo diverso a seconda che si sia in presenza di un terreno solido o poco resistente è presente anche nelle epistole di Seneca.11 I testi citati rivelano dunque in modo chiaro come in età romana vi fosse piena consapevolezza del fondamentale ruolo svolto dalle fondazioni nel garantire la firmitas di un edificio, delle modalità con cui esse dovevano essere realizzate, e soprattutto dello stretto legame tra tipo di fonda­ zione e terreno su cui si andava a costruire. Al momento della posa delle fondazioni, un attento esame del contesto geo-ambientale in cui si operava era infatti assolutamente necessario, se non si voleva incorrere in problemi quali quelli verifica  Vitr. 2, 9, 10.   Vitr. 1, 3, 2.  8   Vitruvio parla di solidissima structura in relazione alle fondazioni di mura (Vitr. 1, 5, 1) e templi (Vitr. 3, 4, 1-2). Il concetto si ritrova anche nel libro VI, in riferimento alle fondazioni degli edifici privati (Vitr. 6, 7, 7 e 6, 8, 1). Con solidissima structura egli si riferisce probabilmente all’opera cementizia (Gros 1990: 128).  9  Queste generiche indicazioni dimensionali vengono fornite per le fondazioni di mura (Vitr. 1, 5, 1) e templi (Vitr. 3, 4, 1-2). A differenza di Emiliano Palladio (cfr. infra), Vitruvio non fornisce misure precise sulla profondità che le trincee di fondazione devono raggiungere. 10   Pallad. I, 8, 2. 11  Sen. epist. LII, 5.  6  7

Anejos de AEspA LXXVII

Costruire in terreni paludosi: sistemi di fondazione...

tisi a Nicea e raccontati da Plinio, dove il teatro, costruito su di un terreno imbibito d’acqua e cedevole in prossimità del lago Ascania, utilizzando una pietra poco resistente e sfaldabile, era soggetto a crolli e spaccature, e necessitava pertanto di continui interventi di rinforzo a livello di fonda­ zione,12 o a Fidene dove, secondo il racconto di Tacito, un anfiteatro costruito in un’area alluvionale in prossimità del Tevere crollò perché le sue fondazioni non poggiavano su terreno solido.13 Degno di nota è il fatto che nel testo vitruviano viene fatta una distinzione tra le fondazioni in terreni “solidi”, e le attività e le opere da attuarsi nelle zone paludose o caratterizzate da terreni poco resistenti e funzionali al consolidamento del terreno, quali le palificate lignee. La stessa distinzione tra fondazioni in terreni solidi e fondazioni in zone paludose si ritrova anche in trattati ben più recenti, come quello di Andrea Palladio del 1570 che, sulla scia forse proprio del trattato di Vitruvio, distingue tra gli interventi pre-costruttivi funzionali a consolidare i terreni paludosi, tra cui annovera le palificate lignee, e le normali fondazioni.14 Purtroppo, confrontare le informazioni fornite dai trattatisti antichi con il dato archeologico non sempre è possibile, in quanto le informazioni edite circa questi aspetti del costruito antico sono in numero piuttosto ridotto. In genere infatti, le fondazioni degli edifici antichi sono poco studiate, dal momento che si tratta di elementi “fan­ tasma”, situati sottoterra e pertanto raramente visibili; ancora minori sono le informazioni disponibili relative all’insieme delle opere che possono essere raggruppate sotto la denominazione “interventi pre-costruttivi”, necessarie in presenza di terreni a scarsa capacità portante.15 In questo senso, l’Italia settentrionale rappresenta un’eccezione, in quanto in questa regione le  Plin. epist. X, 39, 48.  Tac. ann. IV, 62-63. A proposito del teatro di Nicea e dell’anfiteatro di Fidene, si veda anche Nardelli 2003: 946-947 e Giuliani 2006: 162-163. 14   Il tema delle palificate lignee è affrontato nel capitolo VII de I quattro libri dell’architettura di A. Palladio, dedicato alle “qualità del terreno ove s’hanno da poner le fondamenta” (Magagnato e Marini 1980: 19-22). La distinzione tra interventi di preparazione del terreno e fondazioni (Préparation du terrain / Construction des fondations) è presente anche in Ginouvès 1992: 8-10. 15   Gli autori che trattano il tema delle fondazioni in uso nel mondo classico non sono molto numerosi. Per quanto riguarda il mondo greco, cfr. Martin 1965: 308-322 e Murolo 1967: 9-32; per il mondo romano: Ginouvès 1992: 7-17; Giuliani 2006: 161-180; Cartapati 2007: 159-175; Lancaster 2008: 259-260; Adam 2011: 115-166 e 137-138.

211

strutture di epoca romana, soggette a intensa spoliazione, si conservano spesso solo a livello di fondazione, e permettono quindi di analizzare in modo dettagliato le strutture poste alla base degli edifici antichi e le opere di bonifica e sottofondazione preliminari alla costruzione delle strutture. L’Italia settentrionale inoltre, in antico come oggi occupata per la maggior parte da una pianura alluvionale caratterizzata da terreni paludosi, acquitrinosi o umidi, e quindi in genere a scarsa resistenza, nonché dalla presenza di una falda freatica posta a ridotta profondità e spesso affiorante, proprio per le sue caratteristiche geo-morfologiche e ambientali costituisce un osservatorio privilegiato per analizzare le diverse tipologie di interventi finalizzati a migliorare le caratteristiche del terreno destinato ad ospitare le costruzioni.16 Probabilmente proprio le caratteristiche del contesto ambientale che caratterizza la regione determinarono infatti in età romana un’enorme diffusione degli apprestamenti funzionali a migliorare le condizioni del terreno su cui si andava a costruire, come si evince da un’analisi dei dati archeologici ad oggi disponibili. Tali apprestamenti, che si distinguono per la varietà dei materiali e delle soluzioni tecniche adottate, trovano confronto in altre regioni della penisola italica, del bacino Mediterraneo e delle province d’Oltralpe, e costituiscono quindi indizi preziosi per ricostruire le dinamiche di cantiere e l’identità delle maestranze attive in Italia settentrionale in età romana. Data la potenzialità informativa del contesto, in questo contributo si cercherà pertanto, alla luce dei dati a disposizione, di individuare le peculiarità dei sistemi di fondazione e bonifica in uso in Italia settentrionale in età romana, nonché di definire l’origine dei saperi tecnici applicati e le dinamiche della loro diffusione all’interno della regione.

12

13

FONDAZIONI E SOTTOFONDAZIONI IN LEGNO Da una rassegna del materiale edito, appare evidente che in Italia settentrionale ebbero ampia 16   Per un inquadramento geologico e idrogeologico della Pianura Padana, cfr. Alemani 1996. Le caratteristiche ambientali dell’Italia settentrionale erano ben note agli antichi, come emerge dalle descrizioni di numerosi autori quali Strabone (V, 1, 7), Polibio (II, 16-6-15), Plinio il Vecchio (nat. III, 119-122),Vitruvio (1, 4, 11; 2, 9, 10-11).

212

Caterina Previato

diffusione i sistemi di fondazione che prevedevano l’impiego di elementi lignei (fig. 1).17 Le testimonianze archeologiche finora note documentano una notevole varietà di soluzioni, che rientrano essenzialmente in tre gruppi: le palificate, le palafitte e gli zatteroni.18 Il più diffuso è senza dubbio il sistema delle palificate lignee, ricordato anche da Vitruvio. Tale sistema prevede l’inserimento, nel terreno alla base di strutture ed edifici, di pali di legno appuntiti, che venivano infissi in verticale per mezzo di macchine battipalo. Spesso le punte dei pali erano dotate di puntazza, cioè di una lamina di ferro fucinato, che le proteggeva al momento dell’infissione nel terreno.19 In Italia settentrionale molto spesso, dal momento che il terreno solido era impossibile da raggiungere perché situato ad un’eccessiva profondità, i pali erano infissi fino ad una profondità relativamente ridotta, ed erano utilizzati per costipare il terreno e aumentarne la resistenza, e per evitare il cedimento delle strutture soprastanti. In questo caso si è soliti parlare di “palificate sospese”, perché i pali non scaricano il peso direttamente sul terreno solido, ma traggono la loro portanza dall’attrito laterale tra pali e terreno.20 In alcuni casi, le fondazioni delle strutture murarie venivano costruite direttamente sopra i pali che, posti a distanze ravvicinate, venivano a costituire una sorta di piattaforma. Strutture di questo tipo, cioè semplici palificate, sono state individuate in numerosi siti dell’Italia settentrionale, come per esempio Aquileia, Concordia Sagittaria, Altino, Oderzo, Milano e Como.21 In altri casi, al di sopra dei pali venivano collocate delle tavole o delle travi disposte in senso orizzontale, che costituivano il vero piano di posa delle strutture. In questo caso si è soliti parlare di 17  Il censimento delle fondazioni in legno dell’Italia settentrionale è tuttora in corso. I dati raccolti fino ad ora sono presentati nella tabella di Fig. 1. 18   La distinzione tra queste tre tipologie di apprestamenti lignei si ritrova nel manuale di G. Pegoretti. L’autore infatti distingue tra palificate semplici, “castelli di legname”, cioè palafitte, e zatteroni o graticole (Pegoretti 1863: 126-130). Anche V. Righini distingue tra palafitte e zatteroni (Righini 1991: 205). 19  Le puntazze di età romana avevano di solito forma conica o piramidale, erano lunghe da cm 35-40 a cm 80, ed erano fissate al palo con dei chiodi, servendosi di orecchie o ali forate che costituivano parte della puntazza (cfr. Galliazzo 1994: 338-339). 20   Cartapati 2007: 162; Antico Gallina 2011a: 109. 21  Per una completa rassegna delle attestazioni delle strutture in legno finora censite si veda la tabella di Fig. 1, con bibliografia di riferimento.

Anejos de AEspA LXXVII

“palafitte”. Apprestamenti di questo tipo sono stati individuati ad Aquileia, Treviso e Ivrea, per citare solo qualche esempio. Un altro tipo di struttura lignea di cui invece non parlano le fonti ma che trova riscontro nel dato archeologico è quello degli zatteroni, che prevede la posa alla base di strutture ed edifici di elementi lignei, di solito pali o travi, disposti in orizzontale e assemblati tra loro, a costituire una sorta di zattera su cui far poggiare la costruzione. Zatteroni in legno sono stati finora individuati ad Aquileia, Concordia Sagittaria e Corte Cavanella (Rovigo). Le caratteristiche dimensionali (estensione, sviluppo verticale, etc.) dei tre tipi di apprestamenti descritti sono molto variabili, e dipendono dalle dimensioni e dal carico degli edifici che erano chiamati a sostenere, nonché dalla natura del terreno su cui si impostavano.22 Le palificate, le palafitte e gli zatteroni possono infatti avere un’estensione lineare, e pertanto essere situati solo sotto le strutture murarie di un edificio, di solito sotto le strutture portanti, come osservato ad esempio nell’edificio rustico di Corte Cavanella (Rovigo), oppure avere un’estensione areale, come riscontrato invece nel quartiere residenziale riportato in luce sotto l’ex cinema Garibaldi a Treviso. Talvolta nello stesso contesto si assiste alla compresenza di diversi sistemi di sottofondazione. A Treviso per esempio, nel medesimo quartiere residenziale sono stati individuati una palafitta, composta da un reticolo di pali verticali posti a distanze regolari con sovrapposte una serie di tavole e travi orizzontali e, poco distante, uno zatterone costituito da una serie di travi disposte in orizzontale e tra loro allineate.23 Il numero e la distanza tra gli elementi lignei (pali o travi) impiegati nei diversi tipi di apprestamenti è altrettanto variabile, così come il materiale ad essi interposto e sovrapposto, benché un’analisi completa ed esaustiva dei rinvenimenti sia alquanto difficile, in quanto in rari casi in letteratura si ritrova un’accurata descrizione di queste strutture. Le specie lignee maggiormente utilizzate sono la quercia e il rovere, affiancate da ontano, pino e abete.24 22   Nelle palificate di fondazione dei ponti i pali sono di solito allineati in 3 o più file longitudinali e posti a distanze regolari, che possono variare tra i 30 e i 110 cm e oltre (Galliazzo 1994: 339). 23   Marcassa 1996; Tirelli 1999: 7-9. 24   Cfr. Tabella di Fig. 1 e Antico Gallina 2011a: 80.

Anejos de AEspA LXXVII

Costruire in terreni paludosi: sistemi di fondazione...

Località

Edificio

Tipo apprestamento

Descrizione

Tipo legno

Struttura sostenuta

Aquileia

Basilica forense

palificata

–

–

muro perimetrale nord

Aquileia

Basilica forense

palificata

–

–

muro perimetrale ovest (tratto nord)

Aquileia

Edificio p.c. 523 (fondo Tuzet)

palificata

–

–

Aquileia

Edificio p.c. 523 (fondo Tuzet)

palafitta

–

Aquileia

Porto fluviale, sponda occidentale

palafitta

Aquileia

Porto fluviale, sponda occidentale

palafitta

Aquileia

Porto fluviale, sponda occidentale

palafitta

Aquileia

Magazzini del porto fluviale

Aquileia

Aquileia

213

Cronologia

Bibliografia

i

d.C.?

Previato 2015: 245

i

d.C.?

Previato 2015: 245

muri A, E ed E1

età romana

Previato 2015: 245

ontano

muro B

età romana

Previato 2015: 247

–

–

muro di sponda

i

a.C.?

Previato 2015: 246

pali fitti, non molto grossi

ontano e pioppo

muro di sponda

i

d.C.

Previato 2015: 246

–

ontano

muro ad ovest del muro di sponda

età romana

Previato 2015: 246

palafitta

–

–

muri

Ponte del Cristo

palafitta

pali fitti, squadrati

rovere

–

età romana

Galliazzo 1994: n. 262; Previato 2015: 247

Ponte p.c. 441

palafitta

–

–

piloni

prima età imperiale

Galliazzo 1994: n. 264; Previato 2015: 247

pioppo e rovere

piloni e spalle

fine ii-iii d.C.

Galliazzo 1994: n. 265; Previato 2015: 247

età romana

Previato 2015: 247

seconda metà Previato 2015: 246 i d.C.

Aquileia

Ponte p.c. 281

palafitta

pali verticali (pioppo) su cui poggia un doppio tavolato (rovere), con tavole legate da grappe in legno a coda di rondine; pietre tra le teste dei pali

Aquileia

Edificio p.c. 281

palafitta

–

–

muri in prossimità del ponte

Aquileia

Mura M2, settore nord-est

palafitta

–

–

mura

seconda metà iii d.C. Previato 2015: 246 prima metà iv d.C.

Aquileia

Mura M2, p.c. 397/5

palafitta

“graticola” di pali

–

torrione

seconda metà iii d.C. Previato 2015: 55 prima metà iv d.C.

Aquileia

Mura M2, area del porto fluviale

palafitta

“grosse squadrate palafitte” lunghe anche 4 m

rovere

struttura a U

seconda metà iii d.C. Previato 2015: 246 prima metà iv d.C.

Aquileia

Mura M2, torrione TTT2

zatterone

travi orizzontali

–

torrione

seconda metà iv d.C. Previato 2015: 248 inizio v d.C.

Aquileia

Mura M2, torrione TTT3

zatterone

travi orizzontali

–

torrione

seconda metà iv d.C. Previato 2015: 248 inizio v d.C.

214

Anejos de AEspA LXXVII

Caterina Previato

Località

Edificio

Tipo apprestamento

Descrizione

Tipo legno

Struttura sostenuta

Aquileia

Mura M3, area del porto fluviale

palificata

“reticolato” di pali

–

mura

seconda metà Previato 2015: 245 v d.C.

Aquileia

Mura M3, p.c. 644

palificata

“doppia graticola” di pali

pioppo

mura

seconda metà Previato 2015: 245 v d.C.

Aquileia

Edificio terminale?, p.c. 555/1

palafitta

–

–

muri

Aquileia

Strutture di sponda, località S. Stefano, p.c. 239/1

zatterone

“ossatura” di travi

–

muro

Aquileia

Necropoli di Ponte Rosso

palificata

–

rovere

monumenti sepolcrali

età romana

Previato 2015: 246

Aquileia

Necropoli località Colombara

palificata

–

–

monumenti sepolcrali

età romana

Previato 2015: 246

Torre di Pordenone

Villa romana

palificata

–

–

muri perimetrali

Torre di Pordenone

Banchina fluviale

palificata

pali

rovere

banchina

Concordia Sagittaria

Concordia Sagittaria

Concordia Sagittaria

Ponte di via San Pietro

Mura, a nord-est dell’abitato

palificata

pali “addensati“

fitta palificata

–

–

Previato 2015: 247

età imperiale? Previato 2015: 248

ii

d.C.

Conte et al. 1999: 40; Frassine 2013: 101

età romana

Conte et al. 1999: 41; Frassine 2013: 101

piloni

fine i a.C. inizio i d.C.

Croce da Villa 1987: 395; Galliazzo 1994: n. 448; Antico Gallina 2011a: 113

mura

Sandrini 1987: 410; Bonetto 1998: 35; Di Filippo Balestrazzi 1999; età augustea Croce da Villa 2001b: 169; Croce da Villa 2009: 152; Frassine 2013: 101

–

Sandrini 1987: 410; Croce da Villa 2001b: 169; Croce da Villa 2008: 166 e 168

–

zatterone

6 strati di pali disposti orizzontalmente e in senso alternato, il cui diametro si riduce procedendo dall'alto verso il basso

–

muro dell’edificio età augustea scenico

Magazzini di Concordia piazza Cardinal Sagittaria Costantini

zatterone

platea composta da tavole disposte in orizzontale e collegate da travi trasversali

–

pavimento in cotto con 2 livelli di preparazione

i

d.C.

Antico Gallina 2011a: 123; Frassine 2013: 102

Magazzini di Concordia piazza Cardinal Sagittaria Costantini

palafitta

pali verticali con sovrapposti pali disposti in orizzontale

–

muri perimetrali

i

d.C.

Antico Gallina 2011a: 123

Terme

–

età romana

Bibliografia

palificata

Concordia Sagittaria

Terme

palificata

Cronologia

i-ii

d.C.

Di Filippo Balestrazzi 1994: 189; Di Filippo Balestrazzi 1989: 129-131; Frassine 2013: 101

Anejos de AEspA LXXVII

Costruire in terreni paludosi: sistemi di fondazione...

Località

Edificio

Tipo apprestamento

Descrizione

Tipo legno

Struttura sostenuta

Concordia Sagittaria

Approdo fluviale connesso ai magazzini

palificata

–

–

molo in blocchi di trachite

Concordia Sagittaria

domus di via I maggio

zatterone

pali disposti in orizzontale

–

Altino

Molo lungo il canale Sioncello

palificata

3 file di pali a sezione quadrangolare

Altino

Bachina fluviale ad est del Museo

palificata

Altino

Porta-approdo

Altino

Necropoli lungo la via Annia

Altino, Ca’ Tron

Ceggia

215

Cronologia

Bibliografia

d.C.

Antico Gallina 2011a: 123

–

metà i a.C.

Vigoni 1996: 286

–

–

età augustea

Tirelli 1999: 12

–

–

gradinata

età repubblicana

Tirelli 1999: 12

palificata

–

–

–

prima metà i a.C.

Tirelli 1999: 17

palificata

pali

rovere

monumento funerario e recinto limitrofo

età romana

Antico Gallina 2011a: 119

Ponte sulla via Annia

palificata

560 pali; sopra i pali riporto limoso ricco di ghiaia (spessore m 0,70)

rovere

ponte

i

a.C.

Susana 2004: 71-73

Ponte

palafitta?

palificata “a graticcio”

quercia o pino

ponte

i-ii

abete

pile in opera quadrata

San Donà di Piave

Ponte in località Flumicetto

palafitta?

palificata “a graticcio”; pali con diametro di 30 cm; pietre pressate tra le teste dei pali

Arzignano

Ponte in località Costo (Vicenza)

palificata

pali con diametro di 30 cm

rovere

spalla sinistra

Oderzo

Banchina fluviale in via delle Grazie

palificata

triplice fila di pali per larghezza complessiva di 1 m; ciottoli tra le teste di pali

rovere

banchina in opera quadrata

palafitta

reticolo a maglie larghe composto da pali verticali con sovrapposte travi e tavole disposte in orizzontale; sopra strato di ghiaia misto a torba e ceramica, sigillato da uno spesso riporto di limo

–

Treviso

Area residenziale ex cinema Garibaldi

Treviso

Area residenziale ex cinema Garibaldi

zatterone

sequenza di grosse travi allineate disposte in orizzontale e fossa riempita di frammenti laterizi

Adria

Edificio sull’argine d’Agosta

palificata

–

i

d.C.

Galliazzo 1994: n. 445

fine i - inizio Galliazzo 1994: n. 452 ii d.C.

–

Galliazzo 1994: n. 458

d.C.

Tirelli 1987: 366-367; Trovó 1996

domus

prima metà i a.C.

Marcassa 1996; Tirelli 1999: 7-9

–

hortus

prima metà i a.C.

Marcassa 1996; Tirelli 1999: 7-9

–

basi di pilastri in laterizio

a.C.

Uggeri 1973: 175

i

i

216

Anejos de AEspA LXXVII

Caterina Previato

Località

Edificio

Corte Cavanella Edificio rustico (Rovigo)

Tipo apprestamento

Descrizione

zatterone

–

palificata

pali lignei disposti in orizzontale a formare una piattaforma omogenea

palafitta

pali verticali infissi

palificata

pali con puntazze di ferro quasi coniche; tra le teste dei pali ciottoli pressati saldati da una colata di piombo

Pavia

Ponte sul Ticino

Olginate

Ponte sul fiume Adda

Corte Edificio rustico, Cavanella darsena (Rovigo)

Ravenna

Verona

Ivrea

Tra via D’Azeglio e via Moriga

Ponte Postumio

Banchina portuale

Tipo legno

Struttura sostenuta

Cronologia

Bibliografia

–

muri in trachite euganea

fine i a.C. - inizio i d.C.

Sanesi Mastocinque 1987; Frassine 2013: 102

pilastri di sostegno della tettoia metà i d.C.? della darsena in mattoni sesquipedali

Sanesi Mastocinque 1987

–

–

strutture abitative?

abitato vissuto dalla fine del v sec. a.C. al ii sec. a.C.

Sermond Montanari 1983

–

piloni

età augustea?

Galliazzo 1994: 457

palificata

pali con diametro tra i 15 e i 30 cm, che occupano un'area larga circa 1 m

–

pile in trachite

età augustea

Galliazzo 1994: 301; Antico Gallina 2011a: 113

palificata

passoni o tronchi

quercia

al di sopra dei pali, gettata in cementizio

palafitta

tronchi squadrati (base 25 cm) lunghi m 3,17 e disposti su 3 file, infissi nella sabbia e muniti di puntazze di ferro; sopra i pali assito ligneo spesso 8 cm

–

banchina in opera cementizia rivestita di lastre lapidee

età romana

Finocchi 1980; Frassine 2013: 103

–

–

età tardorepubblicana?

Caporusso et al. 2007: 89-91

rovere

cavea

età augustea

Antico Gallina 2011a: 126-127

Milano

Banchina fluviale

–

in corrispondenza dell'emiciclo esterno: pali alti 80-130 cm con diametro di 25 cm, posti a una distanza di 30 cm; in corrispondenza dell'anello più interno della cavea centinaia di pali appuntiti alti 80-120 cm e con diametro di 20-22 cm; ciottoli tra le teste dei pali

Milano

Teatro

palificate sospese

–

iv

d.C.?

Galliazzo 1994: 293; Antico Gallina 2011a: 112

Anejos de AEspA LXXVII

Costruire in terreni paludosi: sistemi di fondazione...

217

Località

Edificio

Tipo apprestamento

Descrizione

Tipo legno

Struttura sostenuta

Cronologia

Bibliografia

Milano

Mura

palificata

–

–

mura

età repubblicana

Bonetto 1998: 35

Milano

Mura

palificata

–

–

mura

età imperiale

Bonetto 1998: 35

età tardoimperiale

Bonetto 1998: 35

a.C.-i d.C.

Passi Pitcher e Bishop 1991: 49; Antico Gallina 2011a: 123

Laumellum

Mura

palificata

–

–

mura

Bedriacum

Domus in prossimità del fiume Oglio

palificata

–

–

muri portanti

i

in fine età prossimità repubblicana della parte - inizio età più esterna augustea della cavea

Maggi 1993: 18; Antico Gallina 2011a: 137 Antico Gallina 2011a: 123

Como

Teatro

palificata

–

–

Como

Edificio D (suburbio)

palificata

–

ontano

fondazioni muro esterno

fine i - inizio ii d.C.

–

muri nel settore settentrionale

età augustea Nardelli 2003: 957

Parma

Anfiteatro

palificata

–

Fig. 1. Tabella riassuntiva delle fondazioni e sottofondazioni in legno dell’Italia settentrionale finora censite.

Per quanto riguarda la prescrizione vitruviana circa la posa, tra le teste dei pali, di carbone, in Italia settentrionale non vi sono casi a me noti che attestino quest’usanza. Molto più diffusa sembra essere invece l’abitudine di porre, tra i pali, materiale lapideo, e cioè pietre o ciottoli. Nel caso ad esempio delle fondazioni delle pile del ponte Postumio di Verona, gli interstizi tra i pali sono occupati da ciottoli pressati e saldati tra loro da una colata di piombo. Ciottoli sono stati individuati anche tra le teste dei pali di fondazione del teatro di Milano. Tra i pali di fondazione del ponte di San Donà di Piave, località Fiumicetto, si trovano invece pietre pressate. Sebbene nuove ricerche e censimenti siano necessari per definire un quadro dettagliato dei ritrovamenti, appare evidente che i sistemi lignei di bonifica e sottofondazione (palificate, palafitte e zatteroni) trovarono ampia diffusione in Italia settentrionale in età romana. Essi risultano particolarmente diffusi in aree fluviali o perifluviali, così come in zone paludose o depresse, e in generale in zone caratterizzate dalla presenza di acqua, sia superficiale, sia ipogea. Numerosi sono i casi di strutture lignee poste alla base di piloni di ponti,25 come riscontrato ad Aquileia, Concordia, Altino (Ca’ Tron), Ceggia, 25  A proposito dell’uso di palificate e palafitte nelle fondazioni dei ponti di età romana cfr. Galliazzo 1994: 337-342.

San Donà di Piave, Arzignano, Verona, Pavia, Olginate. Altrettanto numerosi sono i casi di strutture lignee poste alla base di banchine e moli fluviali, come verificato in contesti pubblici ad Aquileia, Altino, Oderzo, Ivrea, Milano, e talvolta in contesti privati, come nella villa di Torre di Pordenone. Apprestamenti lignei di vario tipo sono presenti anche alla base di complessi architettonici pubblici di dimensioni e peso notevoli, come cinte murarie (Altino, Concordia Sagittaria, Milano), basiliche civili (Aquileia) ed edifici termali (Concordia Sagittaria), nonché teatri (Milano, Como) ed anfiteatri (Parma), edifici sottoposti ad un grande carico sia statico che dinamico (fig. 2). Degno di nota il caso del teatro di Concordia Sagittaria, in cui un muro in pietra alto 3,5 metri, probabilmente pertinente all’edificio scenico, poggia su uno zatterone composto da almeno 6 strati sovrapposti di pali disposti in orizzontale e in senso alternato. Più raro, ma comunque attestato, sembra essere l’impiego di palificate, palafitte e zatteroni in edifici privati.26 Questo fatto ben si spiega dal mo26  Tra le attestazioni note vi sono uno zatterone ligneo individuato sotto una domus di Concordia Sagittaria situata in via I Maggio (Vigoni 1996: 286) e una domus alla periferia del vicus di Bedriacum, posta in prossimità dell’antico alveo del fiume Oglio, i cui muri portanti poggiano su pali (Passi Pitcher e Bishop 1991: 49).

218

Caterina Previato

Anejos de AEspA LXXVII

LE STRUTTURE AD ANFORE

Fig. 2. Concordia Sagittaria. La palificata presso l’angolo nord-orientale delle mura di età augustea (da Sandrini 1987).

mento che gli apprestamenti lignei prevedevano da un lato un grande dispendio di materiale pregiato, qual era il legno, dall’altro necessitavano di operatori specializzati in possesso di competenze  specifiche nella posa in opera dei singoli elementi.27 A livello cronologico possiamo affermare che palificate, palafitte e zatteroni trovarono diffusione già in età repubblicana, nelle fasi di vita più antiche delle colonie dell’Italia settentrionale, e restarono quindi in uso per tutta l’età imperiale e anche oltre.28

27   Antico Gallina 2011a: 135. Per essere messo in opera sotto forma di pali, il legno doveva essere giunto a maturazione, per raggiungere la quale servivano molti anni. Una quercia infatti raggiunge la maturità in 140-180 anni, un abete in 80100 anni, un pino o un olmo in 80-90 anni e un pioppo in 3040 anni (Antico Gallina 2011a: 80). 28   In Italia settentrionale le attestazioni successive all’età romana di palificate, palafitte e zatteroni lignei sono numerosissime. Apprestamenti di questo tipo sono stati individuati a Ravenna, Milano, Ferrara, Comacchio e Venezia (cfr. Righini 1991; Antico Gallina 2011a: 119-121, 145, 147).

Un secondo sistema funzionale a migliorare le caratteristiche del terreno che trovò ampissima diffusione in Italia settentrionale è quello delle cosiddette “strutture ad anfore” o “sistemi ad anfore”,29 cioè accumuli di anfore poste a riempire avvallamenti naturali o trincee e fosse artificiali scavate nel terreno, posti alla base di strutture murarie, piani pavimentali, aree scoperte e talvolta interi edifici, così come in prossimità di strade, argini e banchine. Tali apprestamenti sembrano aver avuto grande diffusione sia nell’architettura privata, sia in contesti pubblici, come nel caso dell’anfiteatro di Padova.30 Nella maggior parte dei casi le anfore venivano disposte in verticale, perlopiù capovolte. In altri casi, i contenitori erano posti in orizzontale, di solito con il puntale di una infilato nell’imboccatura dell’altra, a costituire una sorta di condotto. Dal momento che nella Pianura Padana la falda ha gradiente piezometrico dell’ordine dell’1‰ e pendenza generale intorno al 3%, per svolgere una funzione di drenaggio i condotti di anfore disposte in orizzontale dovevano essere particolarmente lunghi.31 I ritrovamenti di questi accumuli di anfore in Italia settentrionale sono andati moltiplicandosi negli ultimi anni, e hanno permesso di analizzarne in dettaglio le caratteristiche e le modalità di messa in opera e di funzionamento.32 Proprio gli studi condotti in anni recenti hanno permesso di capire che le strutture ad anfore in epoca antica svolgevano essenzialmente due funzioni, in quanto venivano utilizzate a scopo di bonifica geotecnica (opere di bonifica a scopo di fondazione o costipamento) o idraulica (opere di bonifica a scopo di aerazione; isolamento termico; infiltrazione; drenaggio). In alcuni casi essi svolgevano contemporaneamente più funzioni.33 Per comprendere quali, di fondamentale importanza è 29   Questi apprestamenti sono stati a lungo definiti, in modo erroneo, “drenaggi”, benché essi non siano sempre utilizzati per drenare, cioè per sottrarre acqua al terreno. Si preferisce quindi adottare le definizioni “sistema ad anfore” o “struttura ad anfore” introdotta da M. V. Antico Gallina, più generica e priva di sfumature interpretative (cfr. Antico Gallina 1996 e Antico Gallina 2011c: 180-183). 30   Pesavento Mattioli et al. 1999. 31   Antico Gallina 1996: 81. 32  Per una sintesi dei ritrovamenti finora effettuati cfr. Antico Gallina 1996 e Pesavento Mattioli 1998. 33  Per un’analisi delle possibili e diverse funzioni svolte dalle strutture ad anfore cfr. Antico Gallina 1996; Lunardi 1998; Frassine 2013; Mazzocchin 2013: 51-59.

Anejos de AEspA LXXVII

Costruire in terreni paludosi: sistemi di fondazione...

analizzare il contesto in cui si inseriscono, e cioè le caratteristiche idrogeologiche del sito, esaminando il tipo di terreno posto sotto e a lato dei contenitori, il materiale interstiziale e il materiale posto all’interno delle anfore, tutti elementi cui si  è cominciato a prestare attenzione solo di recente.34 Questi particolari apprestamenti, mai citati né dalle fonti antiche né a livello epigrafico, trovarono ampissima diffusione in Italia settentrionale, perlopiù in contesti caratterizzati da terreni imbibiti o saturi e a scarsa resistenza, e in aree perifluviali. Strutture ad anfore sono state individuate ad Aquileia, Concordia Sagittaria, Altino, Oderzo, Padova, Vicenza, Verona, Cremona, Milano35 e in moltissimi altri centri urbani, con una particolare concentrazione di casi nella Regio VIII, nella Regio X e nella Regio XI 36 (fig. 3). Il periodo di massima diffusione di questo sistema è quello compreso tra il i secolo a.C. e il i secolo d.C., e coincide con quello del massimo sviluppo monumentale ed economico dei centri urbani dell’Italia settentrionale. Vi sono anche delle attestazioni più tarde, risalenti alla media e avanzata età imperiale, ma sono meno numerose. L’enorme fortuna che conobbero le strutture ad anfore trova sicuramente motivazione da un lato, nelle loro caratteristiche fisiche e nelle loro proprietà meccaniche, in quanto le anfore sono materiali leggeri, con peso specifico ridotto, che ben si prestano a svolgere funzioni di alleggerimento, che offrono una maggiore superficie di attrito laterale rispetto ai pali di legno, e che sono soggetti a compressione, ma non a taglio né a torsione. D’altro canto, il successo di questo tipo di apprestamenti va ricercato nell’econo­ micità del sistema, che prevedeva l’impiego di materiali, quali le anfore, di riuso e di facile approvvigionamento, soprattutto in un momento storico, quale il periodo tra il i secolo a.C. e il i secolo d.C., di grande fervore economico, in cui le merci circolavano perlopiù in contenitori di questo tipo. 34   Cfr. Antico Gallina 1996; Antico Gallina 2011b; Antico Gallina 2014. 35   Per Aquileia: Maselli Scotti 1998; Previato 2015: 232239. Per Concordia: Croce Da Villa e Sandrini 1998. Per Altino: Tirelli e Toniolo 1998. Per Oderzo: Tirelli et al. 1998; Cipriano e Ferrarini 2001. Per Padova: Pesavento Mattioli 1992; Pesavento Mattioli 1998; da ultimo, Cipriano e Mazzocchin 2011, con bibliografia precedente. Per Vicenza: Mazzocchin 2013. Per Verona: Cavalieri Manasse 1998. Per Cremona: Passi Pitcher 1998. Per Milano: Bruno 1998. 36   Antico Gallina 1996: 70.

219

Fig. 3. Verona. Struttura ad anfore rinvenuta presso l’ex Convento dei Cappuccini (da Cavalieri Manasse 1998).

LE SOTTOFONDAZIONI PLURISTRATIFICATE Infine, per completare il quadro dei sistemi di fondazione e bonifica diffusi in età romana in Italia settentrionale, è necessario ricordare anche un terzo sistema meno noto ma largamente utilizzato per migliorare le condizioni del terreno su cui si andava a costruire, e cioè quello delle cosiddette sottofondazioni pluristratificate.37 Dietro questa definizione è celato un particolare apprestamento, che prevede la posa, alla base di strutture murarie, ma anche di strutture idrauliche e pavimenti, di livelli alternati di materiali accuratamente selezionati, funzionali a stabilizzare il terreno, ad evitare cedimenti delle strutture soprastanti e a contrastare l’umidità e la risalita dell’acqua presente nel sottosuolo. Le soluzioni adottate sono molteplici, e i materiali utilizzati variano a seconda del contesto preso in esame. Di solito però si riscontra un’al37   A proposito delle sottofondazioni pluristratificate, cfr. Previato 2012; Bonetto e Previato 2013.

220

Anejos de AEspA LXXVII

Caterina Previato

Fig. 4. Sulla sinistra: Padova, via San Francesco. Sottofondazione pluristratificata a livelli alternati di limo-argilla e frammenti di laterizi (da Bonato et al. 2010). Sulla destra: disegno ricostruttivo.

ternanza di livelli di sedimenti fini limoso-argillosi poco permeabili, e di sedimenti grossolani o materiali permeabili, quali ghiaia o frammenti ceramici e laterizi. I materiali si trovano disposti per livelli sub-orizzontali di spessore centimetrico all’interno di fosse o trincee, o talvolta riportati sopra terra, in questo caso con il duplice scopo di consolidare e rialzare il piano di calpestio. Questo tipo di apprestamenti, sicuramente molto meno noto rispetto alle strutture lignee e alle strutture ad anfore, è in realtà attestato in moltissime città dell’Italia settentrionale (Aquileia, Oderzo, Concordia Sagittaria, Altino, Treviso, Padova, Verona, Ravenna, Milano, Cremona, Bedriacum), con numerose varianti per quanto riguarda i tipi di materiali impiegati38 (fig. 4). Sebbene le attestazioni più antiche risalgano al iii secolo a.C., questa tecnica sembra aver trovato massima diffusione in Italia settentrionale tra il ii secolo a.C. e il i secolo d.C., con una maggior concentrazione di casi tra il i secolo a.C. e il i secolo d.C. Le sottofondazioni pluristratificate furono utilizzate soprattutto in ambito privato, probabilmente perché si tratta di un sistema costruttivo economico, che prevedeva l’impiego di materiali di facile approvvigionamento e che non necessitavano di lavorazione.

IL CASO DI AQUILEIA

38   Per una rassegna dei ritrovamenti in Italia settentrionale, si rimanda a Previato 2012 e alla tabella di fig. 4 in Bonetto e Previato 2013.

39  Per una più approfondita e completa rassegna delle attestazioni aquileiesi di strutture lignee, cfr. Previato 2015: 243-248, con bibliografia di riferimento, e in particolare Tab. 9.

Gli apprestamenti funzionali al miglioramento delle caratteristiche del suolo finora descritti trovarono ampia diffusione in età romana nelle zone di pianura dell’Italia settentrionale. Spesso all’interno del medesimo centro urbano si assiste alla diffusione e alla compresenza di più sistemi, anche nella stessa fase cronologica. Emblematico è il caso della colonia di Aquileia, città situata nel settore nord-orientale della penisola italica, dove sono attestati sia apprestamenti lignei di sottofondazione (palificate, palafitte e zatteroni), sia strutture ad anfore, sia sottofondazioni pluristratificate (fig. 5). Per quanto riguarda le sottofondazioni in legno, per quanto noto, esse trovarono diffusione esclusivamente in contesti di tipo pubblico, soprattutto in zone prossime ai corsi d’acqua e nel settore orientale della città.39 Palafitte sono state individuate infatti nella zona del porto fluviale, dove sono presenti alla base della banchina della sponda orientale e occidentale e sotto i muri perimetrali dei vicini magazzini costruiti nel i secolo d.C. lungo la sponda occidentale. Più a nord, palafitte sono state individuate alla base della cinta muraria M2, realizza-

Anejos de AEspA LXXVII

Costruire in terreni paludosi: sistemi di fondazione...

Fig. 5. Aquileia. Carta di distribuzione dei diversi sistemi di fondazione e bonifica riportati in luce nell’area urbana (elaborazione dell’autrice).

221

222

Caterina Previato

ta in età tardo-imperiale in prossimità del fiume Natissa. Le palafitte furono utilizzate anche come sostegno delle pile di alcuni ponti che attraversavano i corsi d’acqua circostanti Aquileia, tra cui quello della p.c. 441, situato in prossimità del porto fluviale, e quello della p.c. 281 (fig. 6). Apprestamenti lignei di sottofondazione, e in particolare palificate, sono presenti inoltre sotto le fondazioni dei muri perimetrali della basilica forense, situata in una zona depressa circa al centro della città, a sud della piazza del foro. Nello stesso contesto urbano, si ha notizia inoltre di numerosi ritrovamenti di strutture ad anfore.40 Questi apprestamenti, secondo quanto si può dedurre dai pochi dati editi, si ritrovano indistintamente in contesti di natura pubblica e privata, e si concentrano nel settore occidentale del centro urbano, che si distingue per essere un’area particolarmente bassa e paludosa.41 Tra i numerosi ritrovamenti, basti ricordare i depositi individuati nel settore nord-occidentale della città, in località Santo Stefano, sottoposti ai pavimenti di due ambienti appartenenti probabilmente a dei magazzini, e quelli individuati sotto il basolato del decumano di Aratria Galla, strada che attraversa la città in senso est-ovest, situata a sud del foro cittadino (fig. 7). Sempre ad Aquileia, numerose sono infine le attestazioni di sottofondazioni pluristratificate.42 Questo tipo di apprestamenti sembra aver trovato diffusione soprattutto in ambito residenziale, come riscontrato nella domus delle Bestie ferite e nella domus di Tito Macro dei fondi ex Cossar, anche se non mancano testimonianze in contesti pubblici: il condotto dell’acquedotto sottoposto al lastricato forense poggia infatti su una sottofondazione di questo tipo (fig. 8). L’ampia diffusione di una così grande varietà di opere funzionali al miglioramento delle caratteristiche del suolo all’interno della medesima città ben si spiega con le caratteristiche del contesto idrogeologico aquileiese. La città di Aquileia infatti in antico era situata a breve distanza dal 40  Per una più approfondita e completa rassegna delle attestazioni aquileiesi di strutture ad anfore, cfr. Previato 2015: 232-239, con bibliografia di riferimento, e in particolare Tab. 6. 41  Strutture ad anfore sono state individuate anche nel suburbio della città antica, in contesti di difficile lettura e comprensione per l’assenza di dati di rinvenimento editi. 42  Per una più approfondita e completa rassegna delle attestazioni aquileiesi di sottofondazioni stratificate, cfr. Previato 2015: 223-231, con bibliografia di riferimento, e in particolare Tab. 5.

Anejos de AEspA LXXVII

Fig. 6. Aquileia. Il ponte della p.c. 281, costruito su palafitte (da Galliazzo 1994).

Fig. 7. Aquileia. Struttura ad anfore individuata in prossimità del decumano di Aratria Galla (da Lopreato 1980).

Fig. 8. Aquileia. Sottofondazione pluristratificata di un muro della domus delle Bestie ferite (da Previato 2015).

Anejos de AEspA LXXVII

Costruire in terreni paludosi: sistemi di fondazione...

mare, in un’area di pianura caratterizzata, come dimostrato da recenti indagini, da un’alternanza di ambienti emersi e ambienti umidi (paludi e acquitrini), da una falda acquifera posta a ridotta profondità, e di conseguenza da suoli umidi e a bassa resistenza. Oltre all’acqua presente nel sottosuolo, vi era inoltre un’abbondanza di acque superficiali, che fecero di Aquileia una vera e propria città d’acque. La colonia infatti, sorta alla confluenza di due fiumi,43 era dotata di una serie di vie d’acqua (canali, fossi, etc.) che formavano un circuito idroviario, che venne ampiamente sfruttato dalla città per i collegamenti con il mare Adriatico e con l’entroterra. I diversi tipi di opere di bonifica e sottofondazione furono utilizzati senza soluzione di continui­ tà nelle diverse fasi di vita della città, dall’età repubblicana alla tarda età imperiale, nonostante si osservi una particolare concentrazione di apprestamenti risalenti al periodo compreso tra la fine del i secolo a.C. e il i secolo d.C., momento in cui la città conobbe un notevole sviluppo urbanistico e edilizio. La frequente contemporaneità di utilizzo all’interno del centro urbano di sistemi di bonifica e sottofondazione di tipo diverso si può spiegare con scelte operate dalle maestranze sulla scia delle necessità imposte dal contesto idrogeologico e dal carico dell’edificio da costruire, nonché senza dubbio delle competenze in loro possesso e della qualità e quantità di materiale edilizio di cui potevano disporre nei diversi cantieri. Del tutto analoga la situazione di Milano, dove per l’età romana si ha testimonianza sia di apprestamenti lignei (palafitte, palificate), sia di strutture ad anfore, sia di sottofondazioni a sedimenti, in alcuni casi pluristratificate e del tutto simili a quelle aquileiesi.44 ORIGINE E MODALITÀ DI DIFFUSIONE DEI SISTEMI DI FONDAZIONE E BONIFICA IN USO IN ITALIA SETTENTRIONALE I casi sin qui analizzati appaiono molto significativi, e permettono di apprezzare le approfondite conoscenze geotecniche di cui i costruttori attivi in età romana in Italia settentrionale erano 43   Si tratta del Natiso cum Turro citato da Plinio (nat. 3, 18, 126). 44  A proposito del caso di Milano, cfr. Antico Gallina 1996: 89-95; Bruno 1998; Antico Gallina 2011a: 107-196; Antico Gallina 2011b; Antico Gallina 2011c: 189-191.

223

in possesso. Essi infatti furono in grado di scegliere ed applicare sistemi di bonifica e di consolidamento dei suoli talvolta anche molto complessi, selezionati a seconda del contesto in cui si trovarono ad operare. Questa abilità dei costruttori dell’Italia settentrionale di operare scelte mirate in funzione del contesto, trova preciso riscontro in quanto riferito dagli autori antichi e in particolare da Vitruvio che, come detto, ricordava l’importanza di valutare attentamente, a seconda della natura loci in cui si svolgeva l’attività costruttiva, quali fossero i materiali e le tecniche più adeguate da applicare, per garantire la firmitas degli edifici. I dati sin qui presentati evidenziano il fatto che le particolari caratteristiche geomorfologiche ed idrografiche dell’Italia settentrionale misero senza dubbio alla prova i costruttori romani, che si rivelarono però all’altezza della situazione e furono in grado di ovviare, con soluzioni davvero originali e ingegnose, ai problemi statici legati alla presenza di terreni paludosi e caratterizzati da scarsa resistenza e permeabilità e alla diffusa presenza di acque superficiali e ipogee. Ma qual è l’origine delle conoscenze e delle tecniche che i costruttori di età romana applicarono così diffusamente in Italia settentrionale? Un’analisi a più largo raggio dimostra che tutti i sistemi sopra descritti sono in realtà attestati anche in numerosissimi altri contesti italici e mediterranei, così come nelle province d’oltralpe. Per quanto riguarda i sistemi di sottofondazione con elementi lignei, senza andare troppo lontani, all’origine della diffusione di questa tecnica, ma soprattutto delle palificate e delle palafitte, è facile porre le esperienze precedenti maturate in ambito locale già in età protostorica. Numerose infatti sono le attestazioni in Italia settentrionale di strutture risalenti perlopiù all’età del Ferro poggianti su sottofondazioni in legno, come riscontrato per esempio ad Aquileia,45 Concor­ dia,46 Spina,47 Adria,48 San Basilio di Ariano Po­ 45  Ad Aquileia scavi presso l’Essiccatoio nord hanno rivelato l’esistenza di un insediamento datato alla prima età del ferro e caratterizzato sia nella prima (ix-viii sec. a.C.) che nella seconda fase (vi sec. a.C.) da una serie di capanne costruite su palafitte in legno di quercia (Maselli Scotti 2004). 46   Interessante, per il sito di Concordia, la bonifica datata alla prima età del ferro individuata nel corso dello scavo di via Fornasatta, nell’area Coop, composta da elementi lignei orizzontali e verticali (Di Filippo Balestrazzi 1999: 236; Bianchin Citton 2001: 101). 47   A proposito di Spina, cfr. Righini 1990: 259 e Bacchetta 2003: 23, con bibliografia precedente. 48  Palafitte sono state individuate nell’orto Lodo alla Bettola, nel cortile Ornati vicino al cinema San Francesco, nel Giardino Pubblico presso l’ospedale, nell’area dell’attuale

224

Caterina Previato

Anejos de AEspA LXXVII

Fig. 9. Aquileia, Essiccatoio nord. Strutture di sottofondazione in legno dell’abitato protostorico (da Maselli Scotti et al. 1996).

lesine,49 Padova50 e in altri siti dell’area padana, molti dei quali caratterizzati da una continuità insediativa dall’età protostorica all’età romana (fig. 9). In questo caso, nonostante queste tecniche di sottofondazione siano attestate in età romana sia in altre aree della penisola italica, sia nelle province, soprattutto in siti che presentano affinità geomorfologiche con quelli dell’Italia settentrionale, tenendo conto anche della cronologia delle attestazioni, appare evidente che i costruttori giunti in Italia settentrionale non fecero altro che sfruttare le conoscenze pregresse delle popolazioni locali e adottare tecniche in uso da secoli nel territorio, già testate e rivelatesi vincenti, adattandole alle loro necessità e utilizzandole su larga scala.51 Museo Archeologico e dietro l’abside della chiesa di S. Maria Assunta della Tomba (cfr. Dallemulle 1977: 169-170; Bacchetta 2003: 24-25, con bibliografia). 49   De Min 1987: 85-86; Bacchetta 2003: 25-26. 50   Nell’area dell’ex albergo Storione sono stati individuati i resti di 3 abitazioni con fondazioni costituite da travi lignee disposte in orizzontale (Leonardi 1988; Malnati 1999: 172; Bacchetta 2003: 27, con bibliografia). 51   Uno degli esempi più noti è sicuramente quello del teatro di Marcello a Roma, situato in prossimità di un’ansa del Tevere, quindi in una zona perifluviale caratterizzata da un

A testimonianza di ciò sta il fatto che i sistemi lignei di sottofondazione, diffusi in Italia settentrionale a partire dall’età protostorica, restarono in uso senza soluzione di continuità in fase di romanizzazione, per tutta l’età imperiale e anche successivamente, come si riscontra ad esempio nella città di Venezia, dove palafitte sono state individuate sotto edifici databili tra il vii e il xii secolo d.C.52 Diverso è il caso dei sistemi ad anfore. Questo tipo di apprestamenti infatti trovò diffusione in Italia settentrionale esclusivamente in età romana, a partire dal i secolo a.C., e conobbe enorme fortuna soprattutto tra il i secolo a.C. e il i secolo d.C. Se questo dato cronologico trova in parte motivaterreno poco resistente, e costruito su una fondazione a platea in opera cementizia poggiante su palificata lignea (Fidenzoni 1970: 55; Nardelli 2003: 955). Numerosissime sono inoltre le attestazioni dell’uso di palificate e palafitte nelle province: tra queste basti ricordare le palificate sottoposte alla banchina fluviale di Narbonne in Francia, dell’inizio del i d.C., o quelle sottoposte alla banchina di Londinium lungo il Tamigi, realizzata all’inizio del ii secolo d.C., o ancora le palificate sottoposte al podio del tempio del “Cicognier” di Aventicum (Avenches, Svizzera), risalenti all’età flavia (Finocchi 1980; Gros et al. 1997: 323, nota 147). 52   Righini 1991: 205.

Anejos de AEspA LXXVII

Costruire in terreni paludosi: sistemi di fondazione...

225

Fig. 10. Rodi. Struttura ad anfore individuata lungo il litorale nord-occidentale dell’isola (da Maiuri 1921-22).

zione nella grande disponibilità di anfore di riuso determinata dal grande boom economico che caratterizza la regione in questo periodo, è evidente che l’impiego in età romana di questo tipo di manufatti a scopo di bonifica è l’esito di esperienze maturate in altri contesti culturali e che la sua diffusione in Italia settentrionale è strettamente legata al dominio romano. Analogamente, anche in molti altri contesti italici (Roma,53 Ostia54) e mediterranei (Francia,55 Spagna,56 Africa57) dove questi sistemi di bonifica sono attestati, spesso molto simili dal punto di vista geomorfologico e ambientale all’Italia settentrionale, questa tecnica appare caratteristica dell’età romana, e si diffonde soprattutto tra i secolo a.C. e i secolo d.C., per poi essere sempre 53   Per la città di Roma, il ritrovamento più noto è senza dubbio quello effettuato da H. Dressel nel 1878 al Castro Pretorio, quando vennero ritrovate circa 280 anfore poste in verticale e capovolte (Manacorda 1998: 9-10). In seguito, a Roma sono state riportate alla luce altre strutture ad anfore (cfr. Mazzocchin 2013: 57-59, con bibliografia). 54   Cfr. Zevi 1972; Antico Gallina 2011c: 191-192. 55   Cfr. Laubenheimer 1998. 56   Cfr. Bernal et al. 2005; Antico Gallina 2001c. 57  Per l’Africa, si vedano i casi citati da M. V. Antico Gallina (Antico Gallina 1996: 105).

meno utilizzata a partire dal ii secolo d.C.58 È quanto si riscontra ad esempio nella Gallia meridionale, dove la maggior parte delle strutture ad anfore si data al periodo compreso tra il i secolo a.C. e il i secolo d.C., o in Illyria, dove un consistente numero di ritrovamenti risale allo stesso periodo. Anche in questo caso però, l’invenzione del sistema non si deve ai costruttori romani. Un’analisi a più largo raggio in termini geografici e cronologici condotta da M. V. Antico Gallina ha infatti evidenziato la presenza di strutture ad anfore più antiche, risalenti ad un periodo compreso tra il v e il iii secolo a.C.59 Le attestazioni finora note si collocano in territorio greco (Rodi, Thasos, Atene), magno-greco (Metaponto), nelle colonie greche orientali (Histria) e occidentali (Massalia) (fig. 10). Alla luce degli evidenti legami culturali tra queste diverse aree, nonché della coerenza crono  Antico Gallina 2011c: 202.  Con antecedenti ancora più antichi, appartenenti al mondo orientale (Korsabad, regno dei Sargonidi) e risalenti al vii secolo a.C. (Antico Gallina 1996: 100-106, con bibliografia di riferimento). 58

59

226

Caterina Previato

logica tra i ritrovamenti, si può ipotizzare che la tecnica delle bonifiche con anfore sia stata “inventata” in contesti di cultura greco-orientale, e sia stata quindi adottata, sviluppata e massicciamente utilizzata in tutto il bacino del Mediterraneo in età romana, quando conobbe diffusione forse per il tramite diretto di maestranze di cultura greca e magno-greca. È quanto si può a buon ragione ipotizzare almeno per il settore orientale dell’Italia settentrionale, caratterizzato, fin dall’età protostorica, da stretti contatti e legami con il mondo greco e magno-greco, da sempre favoriti dalla presenza del mare Adriatico, canale di collegamento diretto con le regioni del Mediterraneo grecoorientale.60 Infine, veniamo al terzo ed ultimo sistema di bonifica/sottofondazione considerato in questa sede, e cioè le sottofondazioni pluristratificate. Questo tipo di apprestamento, per quanto noto, in età romana sembra essere caratteristica esclusiva dell’Italia settentrionale, dove fa la sua comparsa nel iii-ii secolo a.C., per poi raggiungere la massima diffusione tra il i secolo a.C. e il i secolo d.C., con numerose varianti. Anche in questo caso però esistono degli antecedenti assimilabili ai casi cisalpini, e cioè attestazioni di sottofondazioni che prevedono l’utilizzo di sedimenti posti in trincee sottostanti le strutture murarie, per consolidare e stabilizzare il terreno sottoposto al carico della costruzione.61 Ampiamente diffusa nel Mediterraneo era l’abitudine di costruire i muri al di sopra di trincee riempite esclusivamente di sabbia, materiale selezionato allo scopo di ripartire ugualmente le pressioni sul piano di posa e forse, come alcuni sostengono, per svolgere una funzione anti-sismica.62 Sottofondazioni di questo tipo sono state individuate in territorio greco (Samo, Olympia), orientale (Tebe,63 Euesperides, Troia, Magnesia al Meandro) e magno-greco (Paestum), e presentano cronologie molto più alte rispetto a quelle dell’Italia settentrionale, in quanto si collocano in un arco cronologico che va dal vi al iii secolo a.C. (fig. 11). Inoltre, sottofondazioni ancor più simili a quelle dell’Italia settentrionale, in quanto caratterizzate dall’alternanza di sedimenti a diversa gra60  A proposito dei legami tra l’Italia settentrionale e il Mediterraneo greco-orientale, cfr. Braccesi 1971; Braccesi 2000. 61   Per una rassegna delle attestazioni note, cfr. Bonetto e Previato 2013, con bibliografia di riferimento. 62   Cfr. Murolo 1967: 20; Bonetto e Previato 2013: 253-254. 63   Murolo 1967: 13.

Anejos de AEspA LXXVII

Fig. 11. Troia. Sottofondazione di un muro del Tempio di Atena Ilias (da Dörpfeld 1968).

nulometria, sono documentate in area greca (Basse) e magno-greca (Metaponto e Paestum). Tali evidenze si datano al v-iv secolo a.C. Queste attestazioni, sebbene non perfettamente identiche a quelle dell’Italia settentrionale, appaiono molto interessanti, e permettono di ipotizzare che anche le sottofondazioni a sedimenti siano un sapere di importazione, introdotto in età romana alla luce di esperienze maturate in contesti greco-orientali e magno-greci, e rielaborato in modo originale dai costruttori dell’Italia settentrionale, sfruttando i materiali di più facile approvvigionamento di cui potevano disporre in quantità, come i laterizi o la ghiaia. Per concludere, ciò che emerge in maniera evidente dai dati sin qui presentati, è che in età romana in Italia settentrionale esistevano saperi tecnici e conoscenze geotecniche approfondite, e che i costruttori attivi in quest’area furono in grado, spinti dalla necessità di ovviare alle problematiche ambientali e avendo come unico obiettivo quello di costruire edifici solidi e duraturi, di far proprie, elaborare e applicare su larga scala conoscenze ed esperienze di diversa matrice culturale, rielaborandole in modo originale, e adattandole al contesto in cui si trovarono ad operare. Dalla rasse-

Anejos de AEspA LXXVII

Costruire in terreni paludosi: sistemi di fondazione...

gna delle opere di sottofondazione e bonifica prese in esame, trapela infatti la sensazione che il linguaggio costruttivo che trovò manifestazione in Italia settentrionale non sia propriamente di matrice “romana”, ma sia piuttosto l’esito di un connubio di culture costruttive diverse, di matrice locale e di importazione, rielaborate a seconda delle necessità costruttive imposte dal contesto ambientale e dei materiali da costruzione di più facile approvvigionamento. L’analisi degli apprestamenti di fondazione e bonifica in uso in Italia settentrionale porta infatti a credere che con l’arrivo dei Romani nella regione non vi sia stata un’imposizione diretta dei saperi tecnici e delle tradizioni costruttive in uso a Roma, quanto più che si sia verificato un lento processo di confronto, assimilazione e scambio di conoscenze tra le maestranze provenienti dall’Italia centrale e quelle di origine locale o provenienti da altri contesti culturali, come quello greco e magno-greco, che probabilmente si trovarono ad operare contemporaneamente nel medesimo cantiere o in cantieri limitrofi nei diversi centri urbani della regione, soprattutto nelle prime fasi di vita delle colonie. Benché definire in modo preciso l’origine delle diverse maestranze, così come le modalità, i luoghi e i tempi in cui vennero tra loro a contatto sia un procedimento molto complesso, che necessiterà in futuro di nuove ricerche e approfondimenti, si può affermare senza dubbio che proprio il confronto e la fusione tra saperi tecnici di diversa origine portò alla nascita della cultura costruttiva della Cisalpina romana, destinata a conservare a lungo nel tempo una sua specifica identità.64 REFERENCES Adam, J.-P. 20116: La construction romaine. Picard, ­Paris. Alemani, P. 1996: “Lineamenti geologici ed idrogeologici della Pianura Padana”, in Antico Gallina, M. V. (a cura di), Acque interne: uso e gestione di una risorsa, pp. 9-28. ET, Milano. Antico Gallina, M. V. 1996: “Valutazioni tecniche sulla cosiddetta funzione drenante dei depositi di anfore”, in Antico Gallina, M. V. (a cura di), Acque interne: uso e gestione di una risorsa, pp. 67-112. ET, Milano. 64   A proposito di questo tema, si veda anche il contributo di Jacopo Bonetto in questo volume.

227

Antico Gallina, M. V. 2011a: Archeologia del legno. Uso, tecnologia, continuità in una ricerca pluri-disciplinare. EDUCatt, Milano. Antico Gallina, M. V. 2011b: “Bonifiche geotecniche e idrauliche con anfore: teoria e pratica di un fenomeno”, The Journal of Fasti Online, 226. Antico Gallina, M. V. 2011c: “Strutture ad anfore: un sistema di bonifica dei suoli. Qualche parallelo dalle Provinciae Hispanie”, Archivo Español de Arqueología, 84, pp. 179-205. Antico Gallina, M. V. 2014: “Dalla topografia al diritto. Sistemi ad anfore e mutamenti verticali del suolo”, in Atlante tematico di topografia antica. Roma, città romane, assetto del territorio, pp. 233246, Atta 24. “L’Erma” di Bretschneider, Roma. Bacchetta, A. 2003: Edilizia rurale romana, Flos Italiae 4. All’insegna del Giglio, Firenze. Bermond Montanari, G. 1983: “Ravenna, via Morigia”, in Bermond Montanari, G. (a cura di), Ravenna e il porto di Classe. Vent’anni di ricerche archeologiche tra Ravenna e Classe, pp. 52-54. University Press Bologna, Imola. Bernal, D., Sáez, A. M., Montero, R., Díaz, J. J., Sáez, A., Moreno, D. e Toboso, E. 2005: “Instalaciones fluvio-marítimas de drenaje con ánforas romanas. A propósito del embarcadero flavio del Caño de Sancti Petri (San Fernando, Cádiz)”, SPAL – Revista de Prehistoria y Arquelogia de la Universidad de Sevilla, 14, pp. 179-230. Bianchin Citton, E. 2001: “Il villaggio della tarda età del bronzo e l’abitato protourbano”, in Croce Da Villa, P. e Di Filippo Balestrazzi, E. (a cura di), Concordia tremila anni di storia, pp. 97-109. Esedra, Concordia Sagittaria. Bonato, S., Destro, C., Mazzocchin, S., Tomaello, E. e Tuzzato, S. 2010: “Nuovi dati sull’abitato di ix-viii sec. a.C. e sull’edilizia pubblica romana da uno scavo presso il ponte di San Lorenzo a Padova”, Archeologia Veneta, 33, pp. 8-43. Bonetto, J. 1998: Mura e città nella Transpadana romana. Fondazione A. Colluto, Portogruaro. Bonetto, J. e Previato, C. 2013: “Tecniche costruttive e contesto ambientale: le sottofondazioni pluristratificate ad Aquileia e nella Cisalpina”, in Cuscito, G. (a cura di), Le modificazioni del paesaggio nell’altoadriatico tra pre-protostoria e medioevo, pp. 231-264, Antichità Altoadriatiche 76. Editreg, Trieste. Braccesi, L. 1971: Grecità adriatica: un capitolo della colonizzazione greca in occidente. Pàtron, Bologna. Braccesi, L. (a cura di) 2000: Studi sulla grecità di Occidente, Hesperìa 10. “L’Erma” di Bretschneider, Roma. Bruno, B. 1998: “Le strutture di bonifica con anfore rinvenute a Milano nello scavo dell’Università Cat-

228

Caterina Previato

tolica”, in Pesavento Mattioli, S. (a cura di), Bonifiche e drenaggi con anfore in epoca romana: aspetti tecnici e topografici. Atti del seminario di studi (Padova, 19-20 ottobre 1995), pp. 259-264. Panini, Modena. Busana, M. S. 2004: “La via Annia nella Tenuta di Ca’ Tron. Il ponte lungo il percorso interno”, in La via Annia e le sue infrastrutture. Atti delle Giornate di Studio (Ca’ Tron di Roncade, 6-7 novembre 2003), pp. 68-80. Antiga, Cornuda (Treviso). Caporusso, D., Donati, M. T., Masseroli, S. e Tibiletti, T. 2007: Immagini di Mediolanum. Archeologia e storia di Milano dal v secolo a.C. al v secolo d.C. ET, Milano. Cartapati, E. 2007: “Fondazioni”, in Carbonara, G. (a cura di), Atlante del restauro VIII, I, pp. 159-175. Utet, Milanofiori Assago (Milano). Cavalieri Manasse, G. 1998: “Banchi d’anfore romane a Verona: nota topografica”, in Pesavento Mattioli, S. (a cura di), Bonifiche e drenaggi con anfore in epoca romana: aspetti tecnici e topografici. Atti del seminario di studi (Padova, 19-20 ottobre 1995), pp. 185-196. Panini, Modena. Cipriano, S. e Ferrarini, F. 2001: Le anfore romane di Opitergium. Lions club di Oderzo, Oderzo. Cipriano, S., Mazzocchin, S. 2011: “Bonifiche con anfore a Padova: note di aggiornamento alla cronologia e alla distribuzione topografica”, in Tra protostoria e storia. Studi in onore di Loredana Capuis, pp. 331-367, Antenor Quaderni 20. Quasar, Roma. Conte, A., Salvadori, M. e Tirone, C. 1999: La villa romana di Torre di Pordenone. Tracce della residenza di un ricco dominus nella Cisalpina orientale. Quasar, Roma. Croce Da Villa, P. 1987: “Concordia”, in Buchi, E. (a cura di), Il Veneto nell’età romana, II, pp. 391-423. Banca popolare di Verona, Verona. Croce Da Villa, P. 1998: “Concordia”, in Tesori della Postumia. Archeologia e storia intorno a una grande strada romana alle radici dell’Europa, pp. 478-482. Electa, Milano. Croce Da Villa, P. 2001a: “Le mura di cinta”, in Croce da Villa, P. e Di Filippo Balestrazzi, E. (a cura di), Concordia. Tremila anni di storia, pp. 146-158. Esedra, Concordia Sagittaria. Croce Da Villa, P. 2001b: “I complessi termali”, in Croce da Villa, P. e Di Filippo Balestrazzi, E. (a cura di), Concordia. Tremila anni di storia, pp. 168170. Esedra, Concordia Sagittaria. Croce Da Villa, P. 2008: “Proposte interpretative sulle terme di Iulia Concordia”, Quaderni di Archeologia del Veneto, 24, pp. 165-174.

Anejos de AEspA LXXVII

Croce Da Villa, P. 2009: “Concordia Sagittaria (Venezia): analisi del tratto nord orientale della cinta muraria”, Quaderni di Archeologia del Veneto, 25, pp. 151-162. Croce Da Villa, P. e Sandrini, G. M. 1998: “Concordia Sagittaria (VE)”, in Pesavento Mattioli, S. (a cura di), Bonifiche e drenaggi con anfore in epoca romana: aspetti tecnici e topografici. Atti del seminario di studi (Padova, 19-20 ottobre 1995), pp. 113-128. Panini, Modena. Dalle Mulle, U. 1977: “Topografia ed urbanistica dell’antica Adria”, Aquileia Nostra, 48, pp. 165-192. De Min, M. 1987: “L’abitato arcaico di S. Basilio”, in De Marinis, R. (a cura di), Gli Etruschi a nord del Po, II, pp. 84-91. Campanotto, Udine. Di Filippo Balestrazzi, E. 1989: “Concordia Sagittaria: Quartiere Nord-Ovest. Relazione preliminare della campagna 1998”, Quaderni di Archeologia del Veneto, 5, pp. 118-144. Di Filippo Balestrazzi, E. 1994: “Il teatro romano di Concordia”, in Spettacolo in Aquileia e nella Cisalpina romana, pp. 183-206, Antichità Altoadriatiche 41. Arti Grafiche Friulane, Udine. Di Filippo Balestrazzi, E. 1999: “Le origini di Iulia Concordia”, in Cresci Marrone, G. e Tirelli, M. (a cura di), Vigilia di romanizzazione. Altino e il Veneto orientale tra ii e i sec. a.C. Atti del Convegno (Venezia, S. Sebastiano, 2-3 dicembre 1997), pp. 229-257, Studi e ricerche sulla Gallia cisalpina 11. Quasar, Roma. Dörpfeld, W. 1968: Troja und Ilion. Ergebnisse der Ausgrabungen in den vorhistorischen und historischen Schichten von Ilion. 1870-1894. Zeller, Osnabrück. Fidenzoni, P. 1970: Il teatro di Marcello. Liber, Roma. Finocchi, S., 1980: “Banchina romana su palafitta trovata a Ivrea nell’alveo della Dora”, in Studi di Archeologia dedicati a Pietro Barocelli, pp. 241-242, Soprintendenza Archeologica del Piemonte, Torino. Fleury, P. 1990: Vitruve, De l’Architecture. Livre I. Les belles lettres, Parigi. Frassine, M. 2013: Palus in agro. Aree umide, bonifiche e assetti centuriali in epoca romana, Agri centuriati suppl. 1. Serra, Pisa-Roma. Galliazzo, V. 1994: I ponti romani. Canova, Treviso. Ginouvès, R. 1992: Dictionnaire méthodique de l’architecture greque et romaine, Tome II, Eléments constructifs: supports, couvertures, aménagements intérieurs. École française d’Athènes-École française de Rome. Atene-Roma. Giuliani, C. F. 2006: L’edilizia nell’antichità. Carocci, Roma. Gros, P. (a cura di) 1990: Vitruve, De l’Architecture. Livre III. Les belles lettres, Parigi.

Anejos de AEspA LXXVII

Costruire in terreni paludosi: sistemi di fondazione...

Gros, P., Romano, E. e Corso, A. (a cura di) 1997: Vitruvius, De Architectura. Einaudi, Torino. Lancaster, L. C. 2008: “Roman engineering and construction”, in Oleson, J. P. (a cura di), The Oxford Handbook of Engineering and Technology in the Classical World, pp. 256-284. Oxford University Press, Oxford. Laubenheimer, F. 1998: “L’eau et les amphores. Les systèmes d’assainissement en Gaule romaine”, in Pesavento Mattioli, S. (a cura di), Bonifiche e drenaggi con anfore in epoca romana: aspetti tecnici e topografici. Atti del seminario di studi (Padova, 1920 ottobre 1995), pp. 47-70. Panini, Modena. Leonardi, G. 1988: “Gli abitati”, in Chieco Bianchi, A. M. e Tombolani, M. (a cura di), I Paleoveneti, pp. 39-42. Programma, Padova. Lopreato, P. 1980: “Aquileia: lo scavo a S-O del Foro romano. Gli ambienti tardo-antichi e la Basilica Forense”, Aquileia nostra, 50, pp. 21-96. Lunardi, P. 1998: “Significato e obiettivi del consolidamento dei suoli oggi e nell’antichità”, in Pesavento Mattioli, S. (a cura di), Bonifiche e drenaggi con anfore in epoca romana: aspetti tecnici e topografici. Atti del seminario di studi (Padova, 19-20 ottobre 1995), pp. 37-46. Panini, Modena. Magagnato, L. e Marini, P. (a cura di) 1980: I quattro libri dell’architettura, A. Palladio. Il Polifilo, Milano. Maggi, S. 1993: “Como romana: la forma urbis problemi e proposte di studio”, in Novum Comum 2050. Atti del Convegno celebrativo della fondazione di Como romana, pp. 163-184. Società archeologica comense, Como. Maiuri, A. 1921-22: “Una fabbrica di Anfore Rodie”, Annuario della Scuola Archeologica Italiana di Atene e delle missioni italiane in Oriente, 4-5, 1924, pp. 249-269. Malnati, L. 1999: “Note sull’edilizia residenziale preromana a Oderzo e nell’Italia orientale”, in Cresci Marrone, G. e Tirelli, M. (a cura di), Vigilia di romanizzazione. Altino e il Veneto orientale tra ii e i sec. a.C. Atti del Convegno (Venezia, S. Sebastiano, 2-3 dicembre 1997), pp. 171-191. Studi e ricerche sulla Gallia cisalpina 11. Quasar, Roma. Manacorda, D. 1998: “Introduzione”, in Pesavento Mattioli, S. (a cura di), Bonifiche e drenaggi con anfore in epoca romana: aspetti tecnici e topografici. Atti del seminario di studi (Padova, 19-20 ottobre 1995), pp. 9-12. Panini, Modena. Marcassa, P. 1996: “La bonifica”, in Tirelli, M. (a cura di), “Lo sviluppo di un settore urbano di Treviso dalla fase di romanizzazione all’età moderna attraverso i primi risultati dello scavo dell’ex cinema Garibaldi”, Quaderni di Archeologia del Veneto, 12, pp. 31-33.

229

Martin, R. 1965: Matériaux et techniques. Picard, P ­ arigi. Maselli Scotti, F. 1998: “Bonifiche e drenaggi con anfore ad Aquileia”, in Pesavento Mattioli, S. (a cura di), Bonifiche e drenaggi con anfore in epoca romana: aspetti tecnici e topografici. Atti del seminario di studi (Padova, 19-20 ottobre 1995), pp. 107-111. Panini, Modena. Maselli Scotti, F. 2004: “Aquileia prima di Roma. L’abitato della prima età del ferro”, in Cuscito, G. e Verzár-Bass, M. (a cura di), Aquileia dalle origini alla costituzione del Ducato longobardo: topografia, urbanistica, edilizia pubblica, pp. 19-38, Antichità Altoadriatiche 59. Editreg, Trieste. Maselli Scotti, F., Mandruzzato, L. e Tiussi, C. 1996: “Essicatoio Nord, scavo 1996”, Aquileia Nostra, 67, pp. 267-272. Mazzocchin, S. 2013: Vicenza. Traffici commerciali in epoca romana. Editreg, Trieste. Murolo, M. G. 1967: La tecnica delle costruzioni presso gli antichi. Parte I. Oriente e Grecia. Volume I. Fondazioni e murature, Roma. Nardelli, M. 2003: “Natura loci e aedificatio. Il rapporto fra terreno e strutture negli edifici per spettacoli romani in Italia”, in Tosi, G. (a cura di), Gli edifici per spettacoli nell’Italia romana, pp. 941-960. Quasar, Roma. Passi Pitcher, L. 1998: “Cremona e i banchi d’anfora: riflessioni sulla topografia storica”, in Pesavento Mattioli, S. (a cura di), Bonifiche e drenaggi con anfore in epoca romana: aspetti tecnici e topografici. Atti del seminario di studi (Padova, 19-20 ottobre 1995), pp. 129-134. Panini, Modena. Passi Pitcher, L. e Bishop, J. 1991: “Calvatone (CR). Località Costa di St. Andrea. Vicus di età romana: saggio 3”, Notiziario della Soprintendenza Archeologica della Lombardia, pp. 49-50. Pegoretti, G. 1863: Manuale pratico per l’estimazione dei lavori architettonici, stradali, idraulici e di fortificazione per uso degli ingegneri ed architetti. Monti, Milano. Pesavento Mattioli, S. (a cura di) 1992: Anfore romane a Padova: ritrovamenti dalla città. Panini, Mo­ dena. Pesavento Mattioli, S. (a cura di) 1998: Bonifiche e drenaggi con anfore in epoca romana: aspetti tecnici e topografici. Atti del seminario di studi (Padova, 1920 ottobre 1995). Panini, Modena. Pesavento Mattioli, S., Mazzochin, S. e Pavoni, M. G. 1999: “I ritrovamenti di anfore presso l’anfiteatro romano di Padova”, Bollettino del Museo civico di Padova, 88, pp. 7-44. Previato, C. 2012: “Tecniche costruttive utilizzate nelle case di Aquileia: le sottofondazioni pluristratifi-

230

Caterina Previato

cate”, in Bonetto, J. e Salvadori, M. (a cura di), L’architettura privata ad Aquileia in età romana, pp. 165-180. Antenor Quaderni 24. Padova University Press, Padova. Previato, C. 2015: Aquileia. Materiali, forme e sistemi costruttivi dall’età repubblicana alla tarda età imperiale, Antenor Quaderni 32. Padova University Press, Padova. Righini, V. 1990: “Materiali e tecniche da costruzione in età preromana e romana”, in Susini, G. (a cura di), Storia di Ravenna I. L’evo antico, pp. 257-296. Marsilio, Venezia. Righini, V. 1991: “Materiali e tecniche da costruzione in età tardoantica e altomedievale”, in Carile, A. (a cura di), Storia di Ravenna II. Dall’età bizantina all’età ottoniana. Territorio, economia e società, pp. 193-221. Marsilio, Venezia. Sandrini, G. M. 1987: “Concordia Sagittaria – Area delle Terme. Scavi 1987”, Aquileia Nostra, 58, pp. 409-411. Sandrini, G. M. 2001: “Le infrastrutture commerciali”, in Croce da Villa, P. e Di Filippo Balestrazzi, E. (a cura di), Concordia. Tremila anni di storia pp. 188-192. Esedra, Concordia Sagittaria. Sanesi Mastrocinque, L. 1987: “L’insediamento di Corte Cavanella di Loreo”, in Buchi, E. (a cura di), Il Veneto nell’età romana, II, pp. 291-300. Banca popolare di Verona, Verona. Tirelli, M. 1987: “Oderzo”, in Buchi, E. (a cura di), Il Veneto nell’età romana, II, pp. 357-389. Banca popolare di Verona, Verona. Tirelli, M. 1999: “La romanizzazione ad Altinum e nel Veneto orientale: pianificazione territoriale e interventi urbanistici”, in Cresci Marrone, G. e Tirelli,

Anejos de AEspA LXXVII

M. (a cura di), Vigilia di romanizzazione. Altino e il Veneto orientale tra ii e i sec. a.C. Atti del Convegno (Venezia, S. Sebastiano, 2-3 dicembre 1997), Studi e ricerche sulla Gallia cisalpina 11, pp. 5-31. Quasar, Roma. Tirelli, M., Ferrarini, F. e Cipriano, S. 1998: “Oderzo (TV): strutture di bonifica con anfore presso il molo fluviale e la necropoli sud-orientale”, in Pesavento Mattioli, S. (a cura di), Bonifiche e drenaggi con anfore in epoca romana: aspetti tecnici e topografici. Atti del seminario di studi (Padova, 1920 ottobre 1995), pp. 135-156. Panini, Modena. Tirelli, M. e Toniolo, A. 1998: “Altino. Strutture ad anfore in aree a diversa funzionalità”, in Pesavento Mattioli, S. (a cura di), Bonifiche e drenaggi con anfore in epoca romana: aspetti tecnici e topografici. Atti del seminario di studi (Padova, 19-20 ottobre 1995), pp. 87-106. Panini, Modena. Trovò, R. 1996: “Canalizzazioni lignee e ruota idraulica di età romana ad Oderzo (Treviso)”, Quaderni di Archeologia del Veneto, 12, pp. 119-134. Uggeri, G. 1973: “Un insediamento romano a carattere industriale (Relazione preliminare degli scavi sull’argine d’Agosta, 1971-1973)”, Musei ferraresi, 3, pp. 174-186. Vigoni, A. 1996: “Via I maggio – Banca popolare Friuladria”, in La protostoria tra Sile e Tagliamento. Antiche genti tra Veneto e Friuli pp. 286-289. Esedra, Padova. Zevi, F. 1972: “Ostia (Roma). Scoperte archeologiche effettuate nei mesi di Settembre e ottobre 1968 nell’Isola Sacra, presso la sponda della Fiumara grande tra il ponte della Scafa e l’ansa”, Notizie degli Scavi di Antichità, s. viii, xxvi, pp. 404-431.

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

FACING STRUCTURAL PROBLEMS IN ANCIENT TIMES. A STRUCTURAL ASSESSMENT DURING THE CONSTRUCTION OF THE MAUSOLEUM OF HADRIAN* PAOLO VITTI Università di Roma Tre

ABSTRACT: The foundations of the mausoleum of Hadrian were designed in order to transfer properly the weight of the above massive structure on the alluvial deposits along the banks of the river Tevere. However, the entrance sector subsided while it was under construction. Analysis of the masonry reveals two stages of the subsidence. The first caused the masonry that had been damaged to be demolished and the area that had subsided to be given a new layout, to reduce weight on the foundation. In a second stage, while the construction of the new layout was being implemented, there was a second severe subsidence, with cracks. At this stage a sounding was carried out in order to assess the extent of the structural damage. Once the mausoleum was completed, this area did not undergo further damage, as we detect by the remains of the holes made to anchor the marble veneer, which are perfectly horizontal. KEYWORDS: Mausoleum, Hadrian, Subsidence, Ancient structural assessment, Sounding, Foundation, Roman concrete, Ashlar masonry. RESUMEN: Los cimientos del mausoleo de Adriano fueron diseñados para descargar correctamente el peso de la estructura sobre los depósitos aluviales a lo largo de las orillas del río Tíber. Sin embargo, el sector de la entrada se hundió mientras se encontraba en construcción. El análisis de la fábrica revela dos fases de hundimiento. La primera causó la demolición de la fábrica que había sido dañada y el replanteo de un nuevo diseño para el área que se había hundido, con el objetivo de reducir el peso sobre la cimentación. En una segunda etapa, mientras se estaba realizando la construcción del nuevo proyecto, hubo un segundo fuerte hundimiento, con grietas. En esta etapa se realizó un sondeo con el fin de evaluar la entidad del daño estructural. Una vez completado el mausoleo, el área en cuestión no sufrió ulteriores daños, ya que por los restos de los orificios hechos para anclar el revestimiento de mármol se detecta que son perfectamente horizontales. PALABRAS CLAVE: mausoleo, Adriano, hundimiento, valoración estructural antigua, sondeo, ­cimentación, hormigón romano, sillería.

FOREWARD Transferring the weight of a building to the ground is one of the crucial issues of any construction. It is likely that it was one of the major

problems discussed by engineers while the site for the construction of the Mausoleum of Hadrian was being set. The monumental building is the same height as the Pantheon, but instead of hosting a hollow interior, like the famous rotunda, it

*  I would like to express my gratitude to the organizers of  the 5th International Workshop on the Archaeology of Roman Construction for accepting my paper. This work would never have been carried out without the benevolence of the Director of Castel Sant’Angelo Maria Picarreta and Aldo Mastroianni, who have eased access and permission for my

surveys and studies. Fulvio Cairoli Giuliani, Renato Perucchio, Alessandro Viscogliosi and my brother Massimo have all contributed with their observations to my understanding of the structure. The text has been generously proofread by Donna Logan. All images and drawings are by the author.

232

Paolo Vitti

Anejos de AEspA LXXVII

Fig. 1. Plan of the north-west sector of the city of Rome. In dashed line the hypothetical itinerary of the funus imperatorum.

Anejos de AEspA LXXVII

Facing structural problems in ancient times...

is a solid and extremely heavy structure.1 The Horti Domitiae located in the Ager Vaticanus were chosen as the site for construction not only because they were an imperial property, but also because of their close proximity to the Campus Martius and their position on the itinerary of the funus imperatorum2 (fig. 1). The location of the building on the right river bank involved the construction of a bridge, which was completed in 134,3 five years prior to the Mausoleum (fig. 2). Here the river forms a tight meander, which must somehow have influenced the layering of the sand deposits along the right bank, deposits which are visible even today. Furthermore, the entire area, including most of the Campus Martius, was rich in alluvial deposits with low weight-bearing capacity. For this reason the designers laid the Mausoleum on an immense concrete platform. If built on compressible soil, heavy and solid structures like the Mausoleum can crack with even minimum movement. In fact, elements of possible settlement can be detected on the drum of the Mausoleum, which is inclined towards the river. Measurements made with the total station on the imprints left by the ashlars on the concrete core reveal that the south side, the one facing the river, is lower than the opposite side. The drum and the square basement are cracked. Cracks can also be found in the radiating walls of the south side of the basement (fig. 12-B) and in the lower drum, both on the inner corridors and on the outer surface. When did the settlement occur and the cracks appear? They could have happened long after the completion of the building. J. Heyman suggests that the period of a settlement is very long and that it can continue many years after a construction is completed, thus the ones in the Mausoleum may well belong to some years after it was inaugurated.4 They could also have developed after an earthquake, when horizontal forces typically create tensile stresses that cannot be supported by Roman masonry.5 In this case they could date to decades after the completion of the building. 1   The Mausoleum was composed of three main volumes: the lower solid concrete drum with the entrance corridor and the spiral staircase; the upper drum, made of a concrete ring and a central tower, which hosted two chambers; the rotunda, located at the top of the construction. There was a 85 × 85 m square basement all around the lower drum which was formed by a terrace supported by radiating walls. See Vitti 2014. 2   See Valli 2014. 3   CIL, VI, 973. 4   Heymann 1995, I: 24-25. 5  The mechanical properties of mortared masonries are based on the structural properties of such materials as Roman

233

Fig. 2. Plan of the Mausoleum and the Pons Aelius.

However, cracks can also appear during construction, if a part of the structure subsides locally. In this case, some action must be undertaken to repair the damage and to correct possible defects that could jeopardize the quality of construction. What follows is the evidence of severe damage that occurred during the first stages of construction of the Mausoleum of Hadrian and how the ancient architects faced it.

concrete, stone and fired-brick masonries, which have good resistance to compression but low resistance to tension.

234

Paolo Vitti

Anejos de AEspA LXXVII

Fig. 3. Axonometric cutaway section of the Mausoleum. A) outer vestibule; B) entrance corridor; C) atrium; D) foundation; E) radiating walls of the square basement; F) radial ramp; G) lower drum; H) upper vestibule and air shaft; I) upper drum; L) rotunda.

THE ENTRANCE SECTOR: EVIDENCE OF A LAYOUT DIFFERENT FROM THE ONE THAT WAS EXECUTED The Mausoleum of Hadrian stands on a massive concrete platform, measuring 85 × 85 m, as deep as 8 m. This foundation was made with caementa of leucitite melilite, i.e. of magmatic stones extremely resistant to compression. The foundation was intended to support a structure 44 m high with no less than 80,000 m³ of Roman concrete plus the ashlar masonry of the exterior elevation and marble decoration (fig. 3). The concrete is formed of horizontal layers of rubble and mortar. Lighter materials were used towards the top of the structure. A survey of the lower sector of the Mausoleum, from the entrance to the spiral ramp that led to the burial room, has revealed important details about the building phases. This sector of the Mausoleum was organised with an outer vestibule (fig. 3-A), located at the entrance, which gave room to a corridor (fig. 3-B), 14,50 m long, that abutted a wider and higher rectangular room, the atrium (5.80 × 6.50 m, fig. 3-C) The atrium had a wide apse set precisely at the end of the visual axis

from the gate and a square niche on the left. The spiral ramp was located on the right side of the atrium (fig. 3-F). Some traces on the floor of the atrium show that this room is the result of a second design phase (fig. 4). The evidence can be summarised as follows: First. A number of cuttings trace the design of the elevation on the floor, some of which partially belong to a different elevation from the present day layout: –– a straight cut runs at the centre of the corridor  and marks the axis of the entrance area (fig. 5-A). The cut was used to trace the main axis of the lower level, from south to east.6 –– a curved cut, having the same dimension and form as the end apse of the atrium, is located towards the centre of the atrium (fig. 5-B). The end of the cut, towards the spiral corridor, is linked to other cuts, identified with the 6   This cut corresponds to the setting of the geometry of the Mausoleum. Similar cuts were used to mark the elevation of the lower drum on the travertine blocks on which the ashlar masonry was founded.

Anejos de AEspA LXXVII

Facing structural problems in ancient times...

235

Fig. 4. View of the atrium floor.

letters (C) and (D) in fig. 5. Cut (C) is oriented in the same way as the end wall of the atrium, while cut (D) is inclined, as its extension is aligned to the left wall of the spiral corridor (fig. 5-E). Second. Part of the floor surface, to the left side of the atrium, is chiselled (fig. 5-area highlighted in dark grey). It could be interpreted as the levelling of the travertine blocks to lay a course of blocks, which is missing. This levelling

occurred during the positioning of the blocks in order to create a homogenous level from the outer vestibule to the atrium. Since the chiselling is interrupted on the north end, it is likely that the levelling of the travertine surface was conducted from south to north, but was never completed. Third. Dowel holes show that some blocks were removed after being placed on the chiselled surface. These holes appear both in the area that

236

Paolo Vitti

Anejos de AEspA LXXVII

Fig. 5. Plan of the atrium. A) straight cut marking the axis of the entrance sector; B) curved cut having the same radius as the apse; C) straight cut; D) oblique cut; E) extension of the oblique aligned to the left wall of the spiral corridor; F and G) dotted line showing hypothetical position of ashlar wall that was dismantled; H) lift shaft.

Anejos de AEspA LXXVII

Facing structural problems in ancient times...

237

was chiselled and in the southeast corner of the atrium. Fourth. Iron clamps connecting adjacent blocks are visible on the floor surface of the atrium. It is unusual to find clamps, which are generally used only in ashlar walls to give additional soundness to the construction, on a floor. These clamps were, therefore, intended to be covered by other blocks, which were then removed. They are located in the area that was chiselled and in the southeast corner of the atrium. Fifth. The end arch of the barrel vault covering the entrance corridor, towards the atrium, is not straight, i.e. the two end walls that support the arch are not aligned (fig. 6). Sixth. The travertine blocks, where the apse of the atrium rises, may have been a later addition since they are all aligned on a straight line and have a different pattern from the other blocks of the atrium (fig. 5-area highlighted in light grey). All these elements show that a portion of the  ashlar wall that forms the entrance corridor (fig. 3-B) was built, but then subsequently demolished before the construction was completed. This wall ended with an apse, as large as the corridor, which was never finished. The traces also show that the spiral corridor was intended to begin precisely at the very end of the apse, as proposed in fig. 7-A. The reason for interrupting the construction and demolishing part of it (fig. 7-B) may be due to major damage that occurred while the wall was being erected. It is likely that the foundation subsided because the soil settled due to the increased weight of the area towards the atrium. THE RECONSTRUCTION OF THE END SECTOR OF THE ENTRANCE CORRIDOR Evidence for this subsidence can be gathered on the lateral walls of the corridor. The horizontal joints of the blocks are all inclined from the exterior towards the core of the Mausoleum. The difference between one end and the other is about 17.9 cm (i.e. 1.45%). The subsidence did not occur all at once, but in two stages, since the two uppermost rows of ashlars (including the square cornice under the impost of the barrel vault, highlighted in grey in fig. 8-A) which were

Fig. 6. View of the end arch of the entrance corridor, highlighting the irregular geometry caused by the misalignment of the abutments.

given a wedged shape in order to regain horizontality after the wall subsided, are also inclined. There is sufficient evidence to reconstruct the two stages. First stage. The lateral walls of the corridor were built up to 4.60 m from the foundation and were planned to create a corridor ending with an apse located according to the circular cut on the travertine wall (fig. 5-B) with the spiral corridor straightening just at the end of the apse (fig. 8-A). The chiselled surface shows that the end apse was not built when the subsidence occurred (fig. 8-B). The north end of the wall, where the apse was to be built, subsided by 11.3 cm. As a consequence the design of the north end of the entrance corridor was modified, and the atrium was designed to decrease the weight on the foundation. It can be calculated that the masonry was reduced by c. 230 m³, around 460 tons. The dimension of the atrium was adjusted to give the room proper symmetry. The irregular geometry of the end arch of the entrance corridor shows that reshaping of the wall was influenced by the dimensions of the ex-

238

Paolo Vitti

Anejos de AEspA LXXVII

Fig. 7. Plan with the two phases of the entrance sector: A) Original plan of the entrance corridor highlighting the portion of the wall that had been built when the first subsidence occurred; B) Blocks dismantled (in dark grey) after subsidence and area demolished to enlarge the end of the corridor (light grey); C) new travertine blocks used as foundation of the atrium walls (grey); D) location of the crack that appeared with the second subsidence.

Anejos de AEspA LXXVII

Facing structural problems in ancient times...

239

Fig. 8. Section with the two phases of the entrance sector: A) Reconstruction of the original layout of the entrance corridor. Dashed line shows parts that were not built when the first subsidence occurred. In grey the square cornice at the impost of the barrel vault; B) First subsidence produced a drop of 11,3 cm towards the atrium; C) Construction of the atrium after demolition of the existing masonry (white dashed line). Addition of two layers of ashlar masonry on the wall that had subsided. They have a wedged shape to give the vault an horizontal impost. D) second subsidence. The drop between the two ends of the corridor wall is 6,6 cm.

240

Paolo Vitti

Fig. 9. View of the west wall of the entrance corridor. Dashed line highlights the corner of the wall towards the atrium as resulted after the demolition of the blocks, which was compulsory for the construction of the atrium. A) surfaced picked to give a vertical line to the wall that had subsided; B) surface under the square cornice which was picked to place the marble cladding.

isting blocks7 (fig. 9), the sum of which resulted in the non-alignment of the two wall ends of the corridor. In fact, the western wall remained 27 cm shorter than the eastern one. The widening of the atrium area can thus be associated with the de­ molition of the existing concrete foundation (fig. 7-light grey area) and the extension of the travertine foundation in order to include the north walls of the atrium. The exclusion of the innermost part of the apse from having travertine ashlars as foundation (figs. 5 and 7-C) can be explained by the fact that the apse had a podium for a statue, which was 1 m ca. above the floor of the atrium. 7   The blocks were cut according to the vertical joints of the existing ashlar wall, in order to have the corner blocks big enough as to guarantee strength and stability to the ends of the wall.

Anejos de AEspA LXXVII

Fig. 10. View of the end apse of the atrium. A) Dashed line highlights the crack. B) Dotted line highlights limit of the masonry which was built when the wall subsided and the crack appeared.

Second stage. The construction of the ashlar wall of the atrium proceeded along the corridor together with the completion of the masonry that had subsided. However, the two layers of blocks which were added above the masonry that had subsided were given, as said, a wedged shape in order to make the impost of the barrel vault horizontal (fig. 8-C). When the new ashlar masonry of the atrium had reached the level of the impost of the barrel vault of the entrance corridor, a new subsidence occurred (fig. 8-D). The masonry moved downwards another 6.6 cm, which is the difference between the two ends of the wall, measured at the impost of the vault. However, the second settlement of the soil along the corridor caused the masonry to crack. A crack appeared in the floor of the entrance corridor (fig. 7-D) and in the newly built wall of the apse, exactly at its centre (fig. 10-A). This crack is still visible, even though repairs and patches (fig. 5-hatched surface) have made it less evident. It is visible all through the travertine floor and goes up to the impost of the apse, but does not continue in the

Anejos de AEspA LXXVII

Facing structural problems in ancient times...

half-dome of the apse nor is it visible in the intrados of the barrel vault above the entrance corridor. This evidence shows that the crack occurred before the apse and the barrel vault were built.

ASSESSING THE DAMAGE It is good practice when cracks appear to proceed with investigations to identify possible construction inaccuracies and to check the entity of  damage. Since cracks are likely to develop whenever the soil settles, if the damage is severe, investigations may well take place through the excavation of pits, which can give a broader understanding of the failure. Documentation of soundings in antiquity is lacking, because once done, they were usually backfilled. Soundings of more recent periods are indeed recorded in archival documents. A good example is the one carried out during the completion of the works on the new basilica of Saint Peter’s in Rome. Gian Lorenzo Bernini, to accomplish the wish of Urban VIII to have two three-storey belfries siding Maderno’s façade, designed and nearly achieved a spectacular construction that would have been as high as the dome of the basilica.8 The third level of the southern belfry was nearly accomplished when cracks began to appear on the facade of the church and on the benediction loggia. The new pope, Innocent X, created a technical commission whose first task was to request for a “tasto”, a sounding, to examine the foundations of the façade, as Bernini suggested himself.9 The pit was opened in 1645; it was as deep as 17.6 m and also tunnelled across the foundations for another 15.84 m, along the cracks which were visible at the footing of the façade. Was the tower sinking because of its excessive weight and inadequate foundations or was it just temporarily settling? Cardinal Virgilio Spada, the head of the commission and a well-known protector of Bernini, decided that it was just a settlement that would end soon thereafter. Borromini, indeed, had understood that the foundations of the belfry made some 30 years before by his master, Maderno, were not intended to support a weight which was six times higher than the one he had originally 8   See McPhee 2002, particularly pages 95-120, where the damage caused by the construction of the two belfries is examined thoroughly. 9   McPhee 2002: 97.

241

planned. These foundations were laid on 42 masonry piers that in turn transferred the weight onto wooden piles inserted into the clayish soil which were insufficient to bear the immense weight of a three storey belfry. However, Cardinal Spada proposed to replace the third tier with a lighter structure and wait for the towers to settle for some years until the process was over.10 What is worth noting is that when the cracks appeared a commission of experts decided to assess the condition of the foundations through investigation pits, and to propose actions that included the lightening of the construction as well as waiting for the consolidation of the soil before making further decisions. It is likely that similar discussion must have taken place in the Mausoleum of Hadrian. The modification of the project after the first subsidence reveals that builders took into consideration the fact that the settlement of the foundation was related to the weight of the structure. Once the critical area that was subsiding had been identified, it was decided to decrease the weight above it through a reduction of the volume of masonry. If the cross-section of the Mausoleum is considered, we find that above the area where the atrium was created, there was an upper vestibule  covered by a pyramidal ventilation shaft (fig. 3-H). Both the upper vestibule and the ventilation shaft were built with marble ashlars instead of travertine blocks and were key elements of the project,11 thus they cannot be considered as a second phase design, like the atrium. It is possible that the addition of the atrium was meant to create a hollow volume over the area that had subsided. Such a decision shows that the builders were confident that the damage was not severe and that the construction could be continued. When the second settlement occurred and the crack appeared, a more precise evaluation of the on-going movements became necessary. The wall had subsided to a lesser extent, but the crack was considerable both in extension (it split the corridor in two) and size, being no less than 3 cm wide. It is also likely that a severe crack on the southwest side of the drum (fig. 11) may have developed at the same time as the one in the entrance sector. This crack extends to one of the few stillpreserved peperino blocks of the ashlar facing, but it ends before reaching the travertine course 10 11

  McPhee 2002: 113.   Vitti 2016.

242

Anejos de AEspA LXXVII

Paolo Vitti

the entrance sector may thus identify a much wider sector that was affected by the settlement of the soil (fig. 12-C), which needed to be assessed before further action was taken. Similar problems must have been faced also during the first stages  of construction of the Pantheon, where the strengthening of the structure can still be detected around the rotunda and in the pronaos.12 However, in the Mausoleum, we know that in order to assess the extension of the damage and the quality of the foundations in the area that was subsiding a sounding was opened between the atrium and the exterior crack. It was intended to verify the condition of the masonry, its correct execution and identify the sector that had subsided. THE SOUNDING

Fig. 11. View of the lower drum of the Mausoleum with a crack which does not extend all through the height of the lower drum. A) ashlar layer separating the lower drum from the upper drum.

Fig. 12. Plan of the ground level of the Mausoleum. A) location of the crack in fig. 11; B) cracks on the radiating walls; C) hypothetical sector that had subsided.

which seals the lower part of the drum, showing that it appeared when the upper drum had not yet been built. The crack on the drum and the one in

The investigation was set to the west of the atrium. In order to reduce the damage on the existing ashlar masonry, an opening was cut into the rectangular niche on the west side of the atrium. Today this opening still survives, since it was reused to give access to the lift, built some time after 1500 (fig. 5-H). What can be seen in the lift shaft is a tunnel that develops parallel to the atrium, in a northern direction (fig. 13). The floor is buried under a thick level of debris, and slopes to the north. The tunnel was dug in the concrete mass which formed the drum of the Mausoleum, but left the rear of the travertine wall of the atrium untouched (fig. 14), with the possible exception of  the blocks corresponding to the lift shaft (fig.  ­13-C), which were chiselled. This difference highlights two separate phases: one brought about with care aiming to remove concrete without damaging the ashlar masonry, the other coarsely achieved in order to create the lift shaft. The way the concrete was removed is not accidental. The travertine blocks, as noted earlier, are all intact. Oblique light, in fact, highlights the fact that the chisels used to pick away the concrete, which obviously left marks on it, preserved the surface of the travertine blocks undamaged. Here we can still see the signs of the original work of the stone masons who used a toothed chisel; only a few scratches document the removal of the concrete (fig. 15). Sometimes the red coating that was painted on the stone surface during construc-

12

  Recently analysed by C. F. Giuliani. See Giuliani 2015.

Anejos de AEspA LXXVII

Facing structural problems in ancient times...

243

Fig. 13. Plan and section of the sounding. 1/5 stages of the tunnelling into the concrete mass; A) foundation of the Mausoleum made with leucititic melilite lava, topped by a layer of bipedales; B) opus caementicium corresponding to the filling of the cut made to create the atrium; C) lift shaft; D) niche made in the first stage of tunnelling, located at the same height as the opening from the atrium.

244

Paolo Vitti

Anejos de AEspA LXXVII

Fig. 16. View of the innermost part of the sounding. A) Leucitite melilite lava opus caementicium; B) tuff and marble/travertine opus caementicium filling a cut in (A); C) Bipedales placed above (A) which topped the foundation of the Mausoleum.

tion is still preserved.13 Only at the corners of the blocks was some of the hard concrete not removed (fig. 14-A).

Concrete was picked away following the layering of construction. Masons were asked, in fact, to locate the different work stages of the concrete, which roughly correspond to every two courses of ashlars. The sounding was made in five stages: 1) the opening of the travertine blocks (fig. 13-1), and the cutting of concrete from east to west; the excavation was conducted horizontally for 4 m ca. (fig. 13-D); 2) the sounding was lowered and directed to the north (fig. 13-2); 3) the sounding was continued in a northerly direction and the roof of the tunnel was lowered to the level of the first work-stage above the foundations (fig. 13-3); 4) further extension of the sounding along the rear of the travertine wall of the atrium, until the last travertine block was reached and a vertical cut into the concrete foundation has been identified (fig. 13-4). This cut was made into the foundation after the first subsidence, to give room for the construction of the atrium. The first phase concrete foundation had caementa made of leucititic melilite lava (dark grey colour) and was topped by a course of bipedales (fig. 13-A), while the second phase concrete had tuff (brown) and marble/travertine (white) caementa (figs. 13-B and 16). It was at this stage of the sounding where the diggers met a crack that was between 3 and 5 cm in width. The crack is oriented east to west and affects both the travertine blocks and the concrete mass. It is clear that by carrying out the sounding two significant elements for the assessment were gathered: the identification of the enlargement of

13   Red coating on ashlars of 1st and 2nd century imperial buildings at Rome has been thoroughly analysed by L. Tucci, who suggests that it was used “to mark the blocks that were approved for placement in the building” (Tucci 2001: 604).

However, interpretation of the function of red paint on stone blocks is still discussed. As suggested to me by F. C. Giuliani, it could be interpreted as a way to identify areas of the contact surfaces between the stones that are not perfectly levelled.

Fig. 14. View of the sounding from the lift shaft, showing the backside of the ashlar wall of the atrium. A) remains of opus caementicum on a travertine block; B) travertine block with a rough surface, C) squared block with a hole for the ferrei forfices (lifting device); D) opus caementicium surface corresponding to the tunnelling phase 3.

Fig. 15. View of block in fig. 14-C, highlighting the original work with a toothed chisel and few scratches caused by the removal of concrete.

Anejos de AEspA LXXVII

Facing structural problems in ancient times...

Fig. 17. View of the lower surface of the bipedales in fig. 15-C as resulted after careful removal of opus caementicium. A) stamp; B) dog footprint; C) scratches made by the pick for removal of opus caementicium.

the atrium as well as the identification of the crack which was likely related to the one in the atrium and the one on the drum. However, the sounding was continued for a few metres under the bipeda­ lis-course that topped the original foundation (fig. ­13-5). Concrete was once again so carefully removed that only a few scratches damaged the brick surface and all of the brick-stamps are perfectly visible (fig. 17). The aim for this further extension of the sounding must have been to reach the crack at the centre of the apse, since the excavators, having arrived at the end of the travertine blocks, began to curve towards the apse, in an eastern direction. However, this additional work was soon aborted, on the basis that the ­information that had been gathered was sufficient for assessing the extent of the damage and deciding if it would affect or not the subsequent steps of c­ onstruction. The fact that the tunnel was never back-filled may be due to the need to leave it accessible for checking any further damage during and after completion of the work. CONCLUSION Foundation settlements can lead to displacements that may cause high strains in some portions of a structure and eventually cause a building to collapse. When analysing the settlement of the Mausoleum of Hadrian we must take into account that the huge cylindrical mass and its thick concrete platform were solid and stiff enough to transfer the stress generated by its own weight to the soil. The weak point of the construction was the alluvium of the Tiber on which the construction was set. Typically in such cir-

245

cumstances, Romans used to lay foundations of piles to limit foundation movements. The use of piles to transfer the immense weight of more than 140,000 m³ of concrete must then be supposed. This solution had previously been employed in the construction of the Theatre of Marcellus, an imposing construction built close to the Tiber River. In 1928, during restoration work, a pit was dug to investigate the foundation of the theatre; it brought to light the head of circular wooden piles lying under the 6.35 m thick concrete foundation.14 Further evidence for the use of piles under the Mausoleum come from the construction in 1890 of the new river embankments, which caused the two ends of the Pons Aelius to be demolished. Square wooden piles, headed with iron tips, were found under the concrete foundations of the piers15 and are now preserved at Castel Sant’Angelo.16 It is thus more than likely that the whole construction was laid on square piles to avoid settlement of the soil. When the entrance corridor subsided, two key observations must have been considered. First that the settlement was not due to an underestimation of the foundation but to a settlement of the soil, since at this stage of construction the weight lying on the immense foundation was minimal. Second, that a solid mass, such as the drum of the Mausoleum, could crack without jeopardizing the stability of the construction. Today our understanding of mechanical behaviour is discussed in terms of stresses. Consolidation of the soil when it has low weight-bearing capacity is predictable and the danger for it to happen is proportional to the surface occupied by the foundation: the wider it is, the higher the risk of uneven settlements, especially when diversity in the layering of the alluvium is more than likely. Differential settlement produces tensile bending stresses in the foundation that crack a structure. This means that a building will not collapse if an equilibrium is found and maintained through the following steps of construction. The question then depended entirely on the weight-bearing capacity of the soil. When cracks appeared most probably the construction was monitored to check if the settlement was over and once it was considered over, the construction was completed. 14 15 16

  Fidenzoni 1970: 106 and fig. 29; Rossetto 1995.   Borsari 1892: 412-428.   Iron tips measure 22-29 cm and were 90 cm long.

246

Anejos de AEspA LXXVII

Paolo Vitti

Fig. 18. Plan and section of the entrance sector. Above: pattern of the holes for the clamps used to fix the marble cladding. A) horizontal line of the marble cladding; B) inclined line of the ashlar masonry. Below: C) area that was chiselled to level the travertine floor after subsiding the northern end of the entrance corridor.

Whatever might have been the thoughts around the settlement and the solutions that were adopted after the sounding, the construction of the entrance sector was achieved with no further damage given that both the barrel vaults of the entrance corridor and of the atrium are not cracked. Once the construction of the Mausoleum was over, the marble cladding was laid in the area that had subsided. The holes left by the clamps in the travertine blocks are perfectly horizontal17 (fig. 18) and give a definitive answer to the expertise of the technicians in charge of the construction, who successfully completed such a formidable structure. After the marble decoration of the walls was accomplished, the final stage of

17   Since the square cornice at the impost of the vault had also subsided it was chiselled in order to be adapted to the horizontal pattern of the marble cladding. See Vitti and Vitti 2015.

construction was the marble flooring set on travertine foundation blocks. Since they were inclined because of the subsidence, they were all chiselled to become horizontal, as can be argued by the fact that the area on which the marble slabs were laid was not chiselled (fig. 18-C).

REFERENCES Borsari, L. 1892: “Delle recenti scoperte relative al ponte elio ed al sepolcro di Adriano”, Notizie degli Scavi di Antichità, pp. 412-428. Ciancio Rossetto, P. 1995: “Indagini e restauri nel Campo Marzio meridionale: Teatro di Marcello, Portico d’Ottavia, Circo Flaminio, Porto Tiberino”, in Archeologica Laziale, 12. Dodicesimo incontro di studio del Comitato per l’archeologia laziale, pp. 93-101. Consiglio Nazionale delle Ricerche, Roma.

Anejos de AEspA LXXVII

Facing structural problems in ancient times...

Fidenzoni, P. 1970: Il teatro di Marcello. Liber, Roma. Giuliani, C. F. 2015: “Problemi strutturali dell’avancorpo del Pantheon”, Annali della Pontificia Accademia di Belle Arti e Lettere dei Virtuosi al Pantheon, 14 (2014), pp. 385-409. Heyman, J. 1995: The Stone Skeleton. Structural Engineering of Masonry Architecture. Cambridge University Press, Cambridge. McPhee, S. 2002: Bernini and the Bell Towers. Architecture and Politics at the Vatican. Yale University Press, New Haven, Conn. Tucci, L. 2011: “Red-painted stones in Roman architecture”, American Journal of Archaeology, 115.4, pp. 589-610. Valli, B. 2014: “La nascita del funus imperatorum, dai precedenti di Silla e Cesare ad Augusto”, in Abbondanza, L., Coarelli, F. and Lo Sardo, E. (eds.), Apo-

247

teosi, da uomini a dei. Il mausoleo di Adriano, pp. 149-169. Munus, Roma. Vitti, P. 2014: “Il Mausoleo di Adriano. Costruzione e architettura”, in Abbondanza, L., Coarelli, F. and Lo Sardo, E. (eds.), Apoteosi, da uomini a dei. Il mausoleo di Adriano, pp. 243-267. Munus, Roma. Vitti, P. 2016: “Il Mausoleo di Adriano e il culto dinastico. L’evidenza architettonica”, in Gasparini, V. (ed.), Vestigia Miscellanea di studi storico-religiosi in onore dell’80° anniversario di Filippo Coarelli, pp. 675-688. Steiner, Stuttgart. Vitti, P. and Vitti, M. 2015: “I rivestimenti marmorei del Mausoleo di Adriano (Castel Sant’Angelo)”, in Angelelli, C. and Paribeni, A. (eds.), Atti del XX colloquio dell’Associazione italiana per lo studio e la conservazione del mosaico (Roma, 19-22 marzo 2014), pp. 103-116. Scripta Manent, Tivoli.

VII INFRASTRUCTURE AND ORGANISATION OF CONSTRUCTION

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

LES TECHNIQUES ET LES ÉTAPES DE LA CONSTRUCTION DES SALLES DE SOUTÈNEMENT DES THERMES DE LONGEAS (CHASSENON, FRANCE) ARNAUD COUTELAS*, DAVID HOURCADE** * Chercheur associé UMR AOROC (CNRS-ENS Paris) ** Chercheur associé Institut Ausonius (CNRS/Université Bordeaux Montaigne)

RÉSUMÉ: Les thermes publics de Longeas, à Chassenon (Charente, France), sont des thermes doubles impériaux du iie s. ap. J.-C. Les 25 ou 26 salles de soutènement voûtées de leur rez-de-chaussée technique sont exceptionnellement bien conservées. L’étude minutieuse du bâti a permis d’y repérer l’intégralité des traces laissées dans le mortier des voûtes et des piédroits par le travail des ouvriers, par leurs outils, voire par eux-mêmes ou leurs vêtements: peinture de chantier, empreintes de truelle, traces de doigts, empreintes d’entrait des cintres, marques de sciage des planches de coffrage, graffiti gravés ou peints, etc. Ces données permettent de restituer avec une précision inhabituelle les multiples techniques et les nombreuses étapes de la construction de ces salles de soutènement. Par ailleurs, cette étude qui confirme l’usage primordial d’éléments en bois pour le chantier permet de s’interroger sur leur réutilisation durant le processus. MOTS-CLÉS: Techniques de construction, Thermes gallo-romains, Salles voutées, Bois de construction, Signes rouges. ABSTRACT: The Longeas public baths, at Chassenon (Charente, France), are double imperial-type baths built during the 2nd c. AD. The 25 or 26 vaulted basement rooms are exceptionally well preserved. A meticulous survey was conducted of all of the marks left in the mortar of the vaults and side walls by the workers’ activity, their tools or even the workers themselves or their clothing: painted signs for the construction process, trowel marks, finger marks, imprints of tie beams for scaffolding, saw marks on shuttering boards, graffiti, etc. This provides an unusual understanding of the construction techniques and stages used for these rooms. Also, this survey confirms the predominant use of timber and wood in the builder’s yard and enables questions to be asked about their reuse during the process. KEYWORDS: Building techniques, Gallo-Roman baths, Vaulted rooms, Timber, Red signs. RESUMEN: Las termas públicas de Longeas, en Chassenon (Charente, Francia), son unas termas dobles imperiales del s. ii d.C. Las 25 o 26 salas abovedadas de sostenimiento de su planta son excepcionalmente bien conservadas. El estudio minucioso del edificio permitió documentar la totalidad de los rastros dejados en el mortero de las bóvedas y paredes por el trabajo de los obreros; signos pintados para el proceso de construcción, marcas de la llana, marcas de dedos, huellas de vigas del andamiaje, marcas de los tableros de encofrado, grafiti, grabados o pintadas. Estos datos proporcionan una comprensión inusual de las técnicas de construcción y las etapas de obra para la construcción de estas habitaciones. Además, este estudio confirma el uso predominante de la madera y plantea algunas preguntas sobre su reutilización en el proceso de edificación. PALABRAS CLAVE: técnicas constructivas, termas galo-romanas, salas abovedadas, madera para la obra, signos rojos.

252

Anejos de AEspA LXXVII

ARNAUD COUTELAS, DAVID HOURCADE

INTRODUCTION Les thermes de Longeas Les thermes de Longeas sont des thermes doubles de type impérial construits au IIe s. ap. J.‑C. (Hourcade 1999; Hourcade et al. 2012). Erigés au centre de l’ensemble monumental de l’agglomération secondaire de Cassinomagus,1 ils couvrent une superficie de près de 12 500 m2 et leur plan, presque parfaitement symétrique, s’inscrit dans un carré d’environ 120 m de côté (fig. 1). Étroitement associé aux pratiques cultuelles, cet édifice qui servait de «bains d’entrée» au sanctuaire des Chenevières était placé sous la protection des Numina Augustorum, de Mars Grannus Victor et de Dea Cobrandia (Hourcade et Maurin 2013). Il fonctionnait vraisemblablement comme des bains à la fois hygiéniques et thérapeutiques.2 Conservés par endroits sur près de 10 m de hauteur,3 ils sont construits sur deux niveaux: un étage destiné au public et un rez-de-chaussée technique réservé au personnel de service. Ce dernier est composé d’espaces de travail à ciel ouvert4 et de vingt-cinq ou vingt-six salles voûtées de soutènement5 (fig. 2). Construites à la suite des murs porteurs, ces salles servaient à compenser la déclivité du sol naturel6 et à permettre l’installation, à 1   Cette bourgade d’environ 140 ha, qui a donné naissance au village actuel de Chassenon en Charente, était située aux confins occidentaux de la cité des Lémovices (Doulan et al. 2012; Doulan et al. 2014). L’ensemble monumental, composé du sanctuaire des Chenevières, de l’édifice de spectacle de la Léna, de l’aqueduc de Longeas, des temples jumeaux et des thermes de Longeas, occupait à lui seul près de 18 ha. 2   Une fois entrés, les utilisateurs pouvaient en effet choisir entre deux circuits: un hygiénique et dextrogyre, au nord, et un thérapeutique et sinistrogyre, au sud (Hourcade 1999; Hourcade et al. 2012). La présence de deux grands bassins d’eau chaude, dans l’angle sud-ouest du corps du bâtiment (Pic1 et Pic2), conduit en effet à penser que la partie méridionale de l’édifice était aussi utilisée à des fins de santé et de bien-être. 3  Si son état de conservation en fait l’un des édifices balnéaires les mieux préservés en France, ses dimensions en font surtout un des plus grands édifices thermaux de Gaule romaine (Doulan et al. 2014). 4   Trois cours de chauffe – Sv1, Sv2 et Sv3 – sont connues (fig. 1). Elles permettaient de faire fonctionner les 11 praefurnia destinés à chauffer l’air des hypocaustes et l’eau des chaudières. Un douzième foyer est installé dans la cour nord, au pied du bassin du caldarium. 5   Les prospections géophysiques menées sur le site n’ont en effet pas permis de confirmer avec certitude la présence d’une salle de soutènement sous la salle V (Hourcade et al. 2012). L’existence de la salle voûtée 25 reste donc hypothétique. 6  La pente sud/nord est relativement forte. Le dénivelé entre la cour1, au sud, et la cour 2, au nord, atteint en effet 6 m (fig. 1). Du fait de la déclivité du terrain, la hauteur des salles augmente du sud au nord et passe de 1 à 4 m environ.

l’étage, d’un sol horizontal destiné à la circulation du public. À l’issue du chantier de construction de l’édifice, seules neuf de ces salles étaient encore accessibles.7 Les autres, dont le rôle était purement technique, étaient scellées.8 Ces espaces, aujourd’hui pour la plupart accessibles, ont livré un ensemble exceptionnel de traces préservées pendant près de 2000 ans. Elles nous renseignent sur le déroulement de la construction et, plus spécifiquement, sur toutes les étapes de la construction de ces salles et de leur voûte. Problématiques et méthode L’analyse du chantier originel de construction est l’un des principaux axes de la recherche menée ces dernières années sur les thermes de Longeas. La construction des salles de soutènement constitue une étape importante dans ce chantier, insérée entre, d’une part, la fin de la construction des murs périphériques et porteurs et, d’autre part, l’habillage interne de l’édifice avec la mise en place des sols, des hypocaustes, des bassins, etc. Cette étape du chantier est particulièrement intéressante car il s’agit d’une étape complexe, relativement technique et a priori réalisée dans un laps de temps restreint. Cependant, au regard du volume important de bois à mettre en œuvre pour les cintres et les planches de coffrage des voûtes, et devant la diversité d’appareil des piédroits, on peut se demander si les salles n’ont pas été réalisées lots après lots ou si plusieurs équipes ne sont pas intervenues, se répartissant le travail entre les différentes parties de l’édifice. Ainsi, de nombreuses questions se posent concernant la chronologie, la dynamique et l’organisation de cette étape de la construction des thermes. Pour tenter d’y répondre, nous avons décidé de concentrer en premier lieu notre étude sur les salles cendriers: les salles 15 à 20, situées sous le tepidarium central T2 de l’étage. Elles for7   Au centre, les salles 15 à 20 servaient de cendriers aux foyers des cours de chauffe voisines. Les sondages réalisés ont en effet prouvé qu’on y rejetait temporairement les cendres des praefurnia avant évacuation (Hourcade 1999). Elles étaient séparées en deux groupes, traversés par le couloir 26. Au nord, la salle 05 servait sans doute de remise. Au sud, la salle 24 servait à collecter les eaux de ruissellement de la cour sud et à les rediriger vers l’égout central. 8   Rares sont les niveaux de sol de ces salles de soutènement réellement fouillés. Un sondage, réalisé en 1996 dans la salle 07 a permis de montrer que le sol de travail correspondait simplement au substrat – l’impactite – mis à nu (Hourcade 1999).

Anejos de AEspA LXXVII

Les techniques et les étapes de la construction...

253

Fig. 1. Situation de Chassenon (Cassinomagus) et plans des thermes de Longeas, niveau supérieur ouvert aux curistes en haut et niveau inférieur dédié au service, avec les salles de chauffe et les salles de soutènement numérotées de 1 à 26 (Doc. Hourcade, Doulan, Sicard, Aupert et Laüt). Légende: C: caldarium, salle du bain chaud; D/U: destrictarium-unctorium, nettoyage et onction; F: frigidarium, salle froide; G: galerie; La: latrines; N: natatio, piscine froide; P: palestre; Pa: pièce d’angle sud ouest; Pic: piscine chaude; S: sudatio, étuve sèche; Sv: cour de chauffe du rez-de-chaussée; T: tepidarium, salle tiède; V: vestibule intermédiaire ou unctorium d’entrée.

254

Anejos de AEspA LXXVII

ARNAUD COUTELAS, DAVID HOURCADE

Fig. 2. Salle de soutènement n° 19 (cliché A. Coutelas).

ment le groupe le plus important avec six pièces –au lieu de trois ou quatre ailleurs–, auxquelles s’ajoute un couloir central (n° 26). De plus, il s’agit des salles les plus aisément accessibles de nos jours et parmi les seules qui l’étaient durant l’Antiquité. Nous y avons observé9 les traces laissées volontairement ou involontairement par les artisans lors du chantier: traces de peinture de chantier, empreintes du passage à la truelle, empreintes de doigts ou de vêtement, graffiti peints ou gravés, marques de sciage et empreintes des marques de tâcheron gravées sur les planches. Elles nous permettent de mieux appréhender le déroulement du chantier de construction de cette partie des thermes. 9  Y ont été mis en œuvre les techniques habituelles de l’archéologie du bâti, s’appuyant sur des relevés pierre à pierre, des relevés issus de photographies redressées et sur un relevé par photogrammétrie 3D des voûtes. Les premiers ont été effectués par L. Calamy, architecte, les seconds avec l’aide du topographe N. Morelle et le dernier par P. Mora, ingénieur de la cellule Archéotransfert d’Archéovision (UMS SHS 3D n° 3657 Pessac).

Les résultats obtenus pour les six salles centrales ont ensuite servi de référentiel pour l’analyse des autres salles, lesquelles ont permis de compléter les observations et les interprétations, notamment sur la circulation et la réutilisation des éléments de coffrage. L’étude des salles 21, 22 et 23, situées sous le destrictarium/unctorium D/U1, a posé problème. Les deux premières sont en effet inaccessibles et la troisième ne l’est que par un conduit percé par les Modernes au sein du béton de la chambre de chaleur de l’hypocauste.10 De plus, dans de nombreuses autres pièces, notamment dans les salles 01, 02, 03, 10 et 11, la totalité des piédroits et/ou de la voûte n’est pas visible, à cause d’effondrements ou de la présence de remblais. Ces inconvénients rendent impossible l’étude statistique de l’ensemble des données.

10   Elle n’est pas entièrement vidée, ce qui oblige à ramper pour se déplacer et ne permet pas d’étudier les piédroits.

Anejos de AEspA LXXVII

Les techniques et les étapes de la construction...

LES ÉTAPES DU CHANTIER Les premiers et principaux résultats concernent la restitution des différentes étapes de la construction des salles de soutènement. Elles sont présentées ci-dessous selon leur enchaînement logique. Construction des murs périphériques Les «murs périphériques» sont les murs principaux et porteurs des pièces qui accueillent les séries de trois ou quatre salles de soutènement. S’ils portent des traces liées à la construction des salles voûtées, ils portent aussi celles de leur propre chantier de construction. Les aménagements les plus lisibles sont les trous de boulin. Ils sont visibles dans les salles nord, situées en bas de pente, puisqu’au sud la roche affleure et les murs périphériques apparaissent directement posés dessus. Ces aménagements n’ont, en soi, aucune originalité, avec un vide à la place d’un moellon et une pierre plus longue en linteau. Seul trait remarquable, on a pu associer cette étape de mise en œuvre des trous de boulins –et probablement des boulins– avec une pause dans l’érection du mur est de la pièce F2.11 Cette connivence entre pause et mise en œuvre d’un élément technique est somme toute logique, mais elle est souvent difficile à percevoir au sein des murs (Coutelas 2009). Les thermes de Longeas en offrent en réalité un beau florilège. On observe aussi quelques séries de petites cavités. Dans le mur ouest de la salle 16, il s’agit de quatre petites cavités rectangulaires faites dans le mortier frais, positionnées en diagonale, sur quatre rangs, toujours au niveau de l’angle supérieur droit d’un moellon. Il pourrait s’agir de plusieurs empreintes d’un même objet métallique. On trouve trois autres empreintes similaires en parement du mur est de la salle 19. Elles sont toutefois moins ordonnées et se démarquent par la présence d’ocre sur leurs parois internes. Enfin, les marques au niveau de la salle 08 sont aussi placées en diagonale, mais se distinguent par leur section carrée, de quelques centimètres de côté. Malgré cette diversité, leur interprétation pourrait être identique: il pourrait s’agir de tests destinés à vérifier la rapidité et la qualité de la prise du mor11   Une pause du même acabit s’observe aussi dans le mur oriental des salles de soutènement sous le caldarium C et le laconicum/sudatio S.

255

tier de chaux; la peinture permettant de les signaler et leur répétition sur plusieurs assises permettant de contrôler l’homogénéité du séchage. Marquage de la position des piédroits Une fois les murs périphériques construits, pouvait débuter la construction des salles de soutènement voûtées. La première étape consistait à placer les maçonneries qui servaient de piédroit. Pour ce faire, les maçons indiquaient parfois en amont sur les murs porteurs l’emplacement futur des piédroits. En effet, on observe de temps en temps des traits verticaux peints à l’ocre et systématiquement placés dans l’axe d’un parement de piédroit. Assez longs, ils mesurent plusieurs dizaines de centimètres. Nous en avons trouvé trois dans les salles cendriers orientales 18 à 20. Tous peints sur le mur «périphérique» oriental de la pièce T2, ils servaient à positionner le parement sud du muret nord de ces salles de soutènement. Il est possible que d’autres aient été tracés, mais ils ont été masqués par la suite par les murets. Les observations effectuées dans les autres salles nous ont permis de voir qu’il n’y avait pas forcément de régularité dans le choix des indications. Un trait est visible dans l’angle sud-est de la salle 04, ici pour aider à appréhender l’épaisseur du piédroit, plus que son emplacement. La salle 06 montre trois traits, dans les angles sud-ouest, nord-ouest et nord-est; les traits occidentaux sont utiles à la définition et à la régularité de l’épaisseur du piédroit. La salle 07 n’en possède qu’un dans l’angle nord-est, utile au placement du mur. La salle 08 en laisse voir deux, dans les angles sud-ouest et nord-est. Finalement, sur ces trois salles (06, 07 et 08), il apparaît que les traits à l’ocre sont utilisés non pas seulement pour le placement des murs libres, mais surtout pour l’indication de l’emplacement des parements, même pour les piédroits collés aux murs porteurs. Cela ne se dément pas ailleurs. La salle 09 présente des traits dans les angles nord-est et sud-est. Nous n’en avons pas vu en salle 10 et il n’est pas possible de vérifier s’il y en a en salle 11 (présence de remblais). Il y en a un dans l’angle sud-est de la salle 12, dans l’angle nord-est de la salle 13 et dans les angles sud-est et sud-ouest de la salle 14. Pour ces trois salles, les murs et piédroits pouvaient ainsi être parfaitement placés. Le trait au sud-ouest de la salle 14 est associé à un autre, horizontal, situé à 55 cm sous le sommet du piédroit

256

Anejos de AEspA LXXVII

ARNAUD COUTELAS, DAVID HOURCADE

sud, soit grossièrement 2 pieds en dessous. Un autre trait horizontal a été trouvé dans l’angle sud-est de la salle 13, sur le mur oriental, 30 cm (soit environ 1 pied) sous le sommet du piédroit sud. On aimerait interpréter ces signes comme des indications de niveau, mais rien ne l’assure (cf. infra). La marque de peinture horizontale placée sur le mur périphérique ouest en salle 17 se situe à la même altitude que le sommet du piédroit nord. Il pourrait s’agir ici d’une marque indiquant le niveau à atteindre pour les murs de la salle de soutènement. Construction des piédroits Une fois les repères placés, les piédroits ont été montés. Les murs de fermeture des salles cendriers, donnant sur le couloir central, ont été érigés en même temps, mais seulement jusqu’à la hauteur du sommet des piédroits, ce qui correspond presque toujours au toit du linteau en briques de la baie d’accès. Les parements de toutes ces maçonneries ont été laissés bruts, offrant à la vue de très nombreuses traces faites par les outils et par les artisans eux-mêmes. Traces de truelles. On est tout d’abord frappé par la multitude de traces liées au lissage grossier du mortier à la spatule ou à la truelle. Ces traces sont toutefois le plus souvent partielles, ou sinon les empreintes de la lame paraissent visiblement déformées par le geste du maçon. Néanmoins, rien que pour les six salles centrales, nous avons repérés vingt-cinq empreintes de ce qui pourrait être la forme exacte de la lame de la truelle. Sans surprise, leurs formes et dimensions rappellent fortement nos outils modernes et contemporains. Il en est sans doute de même pour l’épaisseur de la lame. Nous savons qu’il existait différents types de truelles à l’époque romaine. Classés en fonction de la forme de la lame, les six principaux types sont: ovales pointues (A), elliptiques (B), triangulaires (C), rhombiques (D), rectangulaires (E) et trapézoïdales (F) (Gaitzsch 1980, 133‑147; Manning 1985, 26‑27).12 Dans les thermes de Longeas, les traces laissées dans le mortier suggèrent l’usage de truelles proches des types C (triangulaire), C2 (triangulaire à bout arrondi) et B4 (feuille de laurier). Les angles, mesurés au niveau de la pointe, 12  Il est difficile de les distinguer chronologiquement ou géographiquement en l’absence d’étude systématique.

varient de 30 à 55°, sans qu’il y ait réellement de dimensions privilégiées.13 Toute une gamme de truelles semble donc avoir été utilisée. Empreintes de doigts. Ce traitement grossier des joints, au mortier débordant, fait que d’autres traces laissées par les artisans dans le mortier ont été conservées. Les plus nombreuses sont les empreintes des doigts des ouvriers. Il s’agit, à de rares exceptions près, de traces légères, laissées par un simple contact plus que par un appui contre le mur, ce que prouve notamment la découverte fréquente de l’empreinte digitale encore partiellement visible dans le mortier. Parfois les traces sont un peu plus profondes et l’on peut restituer la position de la main, indiquant généralement que l’ouvrier s’est appuyé contre le mur afin d’effectuer un geste en partie basse du piédroit: lissage des joints ou ramassage de ses outils. On note d’ailleurs que la très grande majorité des marques est située en partie haute des piédroits. Plus encore, pour les salles centrales, les concentrations sont les plus nombreuses dans les angles, vers l’entrée et le muret de fermeture, sans doute parce que le travail y était moins aisé ou parce que le matériel y était stocké. Empreintes de textile. Jusqu’à nos travaux, une seule empreinte de tissu imprimé dans le mortier était connue au sein des salles de soutènement (Coutelas 2012). Désormais, ce sont vingt-huit nouvelles empreintes, ou lots d’empreintes, qui ont été répertoriées. Cet ensemble particulièrement exceptionnel correspond à des empreintes de coude ou de genoux. Parfois isolées, elles sont généralement alignées et groupées par deux, trois ou plus. Les plus nombreuses sont les empreintes de genoux, laissées lorsque l’ouvrier était en position fléchie, placé sur le sol et appuyé contre le mur. Les empreintes situées à un niveau plus élevé que le sol nous indiquent que l’ouvrier travaillait sur un échafaudage.14 En réalité, parmi tous ces enfoncements, seuls dix-sept permettent véritablement d’observer le moulage du tissu (fig. 3).15 Il n’y a qu’un seul indice de forme – un pli –, mais pas de lisière, ni de couture ou de trace de découpe. Les fils sont d’un   La plupart dépassent cependant les 40 °C.   C’est le cas notamment en salle 04, où les traces ont été repérées à 40 cm au-dessus du niveau des trous de boulin. 15   Bien que réduit, ce lot a néanmoins permis à Delphine Henri, doctorante de l’Université François-Rabelais de Tours, d’en mener une analyse détaillée. 13 14

Anejos de AEspA LXXVII

Les techniques et les étapes de la construction...

257

Fig. 3. Empreinte de tissu dans la salle 07 (cliché A. Coutelas).

diamètre assez important et leur torsion indique une probable production domestique. Les «armures» – façon dont ces fils se croisent – sont parmi les plus simples. Sans surprise, les tissus euxmêmes sont grossiers ou moyens.16 Les différences de qualité de tissu rencontrées pourraient refléter la hiérarchie de statut des travailleurs, chef d’équipe ou ouvriers. Les caractéristiques intrinsèques des textiles permettent de proposer quelques «recollages». Ici, il s’agit de regrouper les empreintes faites par les mêmes vêtements. Plusieurs regroupements ont pu être faits entre différentes empreintes dans la salle 07. Surtout, des rapprochements ont pu être faits entre plusieurs salles. C’est le cas pour des empreintes de genou des salles 04, 07 et 20, avec pour tissu une toile aux mêmes caractéristiques. C’est aussi le cas entre une empreinte de genou de la salle 08 et une empreinte de coude de la salle 20: dans ce dernier cas, la tunique et les braies (?) semblent avoir été taillés dans le même tissu. 16  Un tissu est considéré comme grossier lorsqu’on rencontre moins de 8 fils par centimètre, dans un sens ou dans les deux sens. Il est dit moyen s’il existe entre 8 et 15 fils par centimètre.

Les indices de séquences de construction. Certains piédroits ont nécessité la mise en œuvre d’échafaudage. C’est le cas des salles nord, où certains murs atteignent plus de 3 m de hauteur. Un niveau de trous de boulin classiques est alors visible. On l’observe par exemple pour les piédroits nord et sud de la salle 04. Mais ce n’est pas le cas partout. En salle 09, par exemple, le piédroit sud présente un alignement de trous de boulin habituels, avec encadrement de moellons, tandis que le piédroit nord se contente de ce que nous appellerons par commodité des «trous de pieu», c’est-àdire qu’aucune pierre n’est venue encadrer et protéger le boulin. Ces «trous de pieu» s’observent ailleurs. Dans les salles 03 et 04, ils semblent former un second niveau d’échafaudage, à moins qu’il ne s’agisse d’un système de renforcement de la structure en bois. Cette technique a été mise en œuvre dans les piédroits nord et sud en salle 04, alors qu’elle n’est présente que dans le piédroit nord pour la salle 03. Les trous sont globalement à l’aplomb des trous de boulin, trois à cinq rangs de moellons plus haut, soit environ 60 à 75 cm d’écart. Des variations dans l’aspect du mortier et des moellons du piédroit nord de la salle 03 signalent

258

ARNAUD COUTELAS, DAVID HOURCADE

plusieurs volumes de maçonnerie et sont donc révélatrices de pauses ou d’étapes dans le chantier de construction de ce mur. En partant du bas, le premier volume de maçonnerie visible correspond à trois rangs d’opus vittatum17 couverts par une assise de briques. Puis c’est un volume de quatre rangs d’opus vittatum qui est disposé sur une hauteur de 69 cm. C’est alors qu’est installé le premier niveau de boulins, lequel coïncide par ailleurs au sommet des piédroits de l’accès latéral, menant à la salle 02. La construction reprend avec cinq rangs d’opus vittatum, pour une hauteur de 63 cm seulement, de moellons de petites dimensions. Cette séquence correspond aussi à la mise en œuvre de l’arc de décharge en briques du passage. On note alors une nouvelle pause dans la construction, afin peut-être de laisser le linteau de briques reposer et durcir. La construction reprend avec un volume d’opus vittatum de 90 cm de hauteur, constitué de six assises de moellons. Les empreintes de pieux du second niveau d’échafaudage sont situées entre la première et la deuxième assise de ce volume. Enfin, un dernier volume très légèrement en retrait de trois rangs d’opus vittatum vient terminer la construction du mur. Cette séquence n’est pas généralisable à l’ensemble des piédroits des salles septentrionales, loin de là. Il suffit de se tourner vers le piédroit sud de la même salle pour observer une toute autre séquence. L’hypothèse de deux équipes travaillant en même temps, mais pas nécessairement de concert, est possible (cf. infra), à moins que la même équipe mette à profit les temps de pause du premier piédroit pour avancer la construction du second selon un procédé différent. L’existence  de  divergences dans le choix des matériaux ou de leurs rythmes va dans le sens de la première hypothèse, comme le laisse par exemple supposer la salle 10 dont le piédroit nord est construit sans brique alors que le piédroit sud l’est en opus mixtum.18 Nous ne discuterons pas en détail de toutes les séquences observées. Il convient seulement de 17  Nous emploierons, pour simplifier, le terme d’opus vittatum pour tous les appareils de petits moellons quadrangulaires des piédroits, bien que ceux-ci n’aient pas toujours la même qualité de taille ni de mise en œuvre. Quelques portions d’assises sont parfois érigées avec des pierres plates disposées en épi ou de chant (opus spicatum); il ne s’agit toutefois ici que d’un procédé opportuniste pour ne pas être obligé d’évacuer des moellons aux qualités non requises. 18   Globalement à un rang de briques pour trois rangs de moellons.

Anejos de AEspA LXXVII

comprendre que la plupart des arrêts ou des pauses dans la maçonnerie de ces piédroits sont liés à la mise en place d’un élément technique: échafaudage, cintre, volume en retrait, etc. Résultat plus remarquable encore, ces arrêts de chantier paraissent avoir été volontairement marqués et signalés par des traits peints à l’ocre. En effet, un certain nombre de ces tracés horizontaux ocres ont été observés soit sur le piédroit uniquement, soit à la fois sur le piédroit et le mur périphérique. Toutes les salles centrales en ont livré, de même que les salles 04, 06, 08, 09 et, peut-être aussi, 13 et 14. L’angle sud-ouest de la salle 04 est celui qui en présente le plus: au moins quatre, dont trois s’étalent sur le piédroit et le mur porteur (fig. 4). Grâce à eux, on peut restituer la séquence de construction de ce piédroit de la manière suivante: un premier volume de sept à huit rangs d’opus vittatum (de 105 cm d’épaisseur), un deuxième volume avec un seul rang d’opus vittatum à moellons grossiers (20 cm), un troisième fait aussi d’un seul rang d’opus vittatum à moellons grossiers, un quatrième avec quatre rangs d’opus vittatum (60 cm d’épaisseur, soit environ 2 pieds), un cinquième avec deux rangs d’opus vittatum (30 cm, soit environ 1 pied) et enfin un sixième consistant en sept rangs d’opus vittatum. La première marque horizontale est tracée dans l’angle à la base du troisième volume. Les trous de boulin sont situés au sein du premier rang de moellons du quatrième volume. Les «trous de pieu» et la seconde marque rouge sont à la base du cinquième volume. Enfin, la troisième marque se situe à la base du sixième volume. Une dernière marque, limitée au seul piédroit et a priori volontairement presque entièrement effacée, complète l’ensemble. Elle est située deux rangs au-dessus de la toute première marque, soit juste au-dessus du rang de moellons accueillant les trous de boulin. La première remarque qui s’impose est que les traits horizontaux sont placés à des endroits bien spécifiques: à la base d’un volume ou d’une séquence de maçonnerie. Le seul trait placé au sein d’un volume est, nous venons de le voir, en partie  effacé; il pourrait donc s’agir d’une erreur. Puisque d’autres salles fournissent le même type de résultat, il est probable que ces traits soient des  éléments de repère, sans doute des témoins d’avancement du chantier. Ils permettaient peutêtre aux ouvriers de signaler à partir de quel niveau a débuté leur dernière séquence de maçonne-

Anejos de AEspA LXXVII

Les techniques et les étapes de la construction...

259

Fig. 4. Angle sud-ouest de la salle 04, avec le mur de piédroit sud à gauche montrant divers arrêts de construction (lignes transparentes) et plusieurs tracés horizontaux peints à l’ocre (flèches horizontales). Entre parenthèses: une marque rouge en partie effacée. Les flèches verticales indiquent les «trous de pieu» (doc. A. Coutelas).

rie et aux contremaitres de suivre l’avancée de leur travail.19 Deux nuances doivent cependant être apportées à ces conclusions. Dans les salles centrales, nous n’avons pas pu établir de lien entre les marques identifiées et une quelconque séquence de construction. On sait qu’aucun échafaudage n’a été utilisé pour leur aménagement, voire que les maçonneries ont été érigées en une seule fois. Néanmoins, plusieurs tracés horizontaux y ont été retrouvés. La plupart étaient peints entre 29 et 33 cm en dessous du sommet du piédroit. Dans ce cas, on peut se demander si ces traits n’étaient pas, contrairement aux autres, des repères d’avancement, mais plutôt des repères de niveau, indiquant que la construction devait se terminer environ 1 pied plus haut. Par ailleurs, on a pu remarquer que certaines marques ne se résumaient pas à un seul trait horizontal. Dans les salles 04, 06 et 09, elles peuvent 19   C’est aussi une des hypothèses avancées pour expliquer la présence des marques triangulaires rouges découvertes dans la «Palestra» de la villa d’Hadrien, à Tivoli (Attoui 2012).

prendre l’apparence de deux traits orthogonaux, limités au seul piédroit, avec dans la quasi totalité des cas la hampe placée à gauche et vers le bas. Cette marque pourrait avoir été tracée à l’aide d’une équerre, le trait dessiné à l’intérieur lorsque la marque est distante de 3 cm du mur périphérique, ou à l’extérieur sinon. Les tests de prise. L’intérêt des arrêts de chantier et des pauses dans la construction est multiple. Pour les murs de ces salles de soutènement, ils sont essentiellement liés à la mise en place d’un élément technique et à la nécessité de laisser reposer la maçonnerie. Ce second point s’explique par le durcissement lent du mortier de chaux et de sable employé dans la maçonnerie. Ne pas attendre que le mortier ait fait suffisamment prise entraine le risque d’une déformation du mur suivi d’un effondrement de la salle. Cette attente a été généralement respectée puisque, parmi toutes les maçonneries étudiées, seul le piédroit nord de la salle n°04 témoigne d’une légère déformation. Un renflement de quelques centimètres, sur une bande d’environ 1,20 m de

260

ARNAUD COUTELAS, DAVID HOURCADE

long, a en effet été observé dans la partie orientale du mur. Certaines empreintes nous semblent justement en lien avec la surveillance du phénomène de prise et durcissement du mortier. Les plus fréquentes, localisées dans les salles 09, 10, 12 et 19, sont des cavités aux parois internes parfois couvertes d’ocre. Elles sont réalisées avec un outil de section carrée, de 1 cm de côté, dont l’extrémité débute de façon pyramidale et finit en pointe circulaire de grande finesse. Cet outil indéterminé, qui a parfois été très profondément enfoncé dans le parement, jusqu’à 7 cm, rappelle divers objets en métal. Parmi les outils antiques, il pourrait correspondre à certains poinçons fins utilisés pour le travail du métal (Duvauchelle 2005, 153, pl. 5, n.° 33-37 et pl. 6, n.° 41-42), ou à des mèches pointues utilisées pour le travail du bois (Duvauchelle 2005, 180, pl. 32, n°173). De même, certaines alênes employées pour la confection du cuir pourraient convenir (Duvauchelle 2005, 191, pl. 43, n°235, 237-240). Les hypothèses de compas et de stylet sont moins probables puisque ces derniers, notamment, présentant habituellement des sections circulaires. Ces cavités sont rarement esseulées, leur densité peut même être forte. Il pourrait s’agir là aussi de vérification de la prise du mortier. La présence d’ocre pourrait indiquer que le maçon souhaitait retrouver facilement les endroits de test. Des comptes et/ou une validation du chantier? Un graffito peint en rouge a été retrouvé dans l’angle nord-est de la salle 08, à hauteur de la sixième assise du mur de piédroit oriental (fig. 5). Haute de 13 cm et longue de 24 cm, l’inscription a été peinte lors du chantier. Elle correspond d’ailleurs à un arrêt temporaire, puisque la partie supérieure de certaines lettres a été recouverte par le mortier de l’assise supérieure. Elle a donc été rédigée entre la pose des sixième et septième assises du mur. La lecture est délicate, mais on devine trois groupes de caractères: IX II B. La présence d’un trait horizontal au dessus de l’inscription laisse penser que les premières lettres capitales sont en fait des chiffres. Si c’est le cas, l’inscription pourrait être retranscrite de façon suivante: 9 2 B. Le sens à donner à cette série est délicat.20 Il pourrait s’agir d’indication de volume ou de

  Il est rediscuté infra, note 43.

20

Anejos de AEspA LXXVII

quantité: peut-être des journées de travail ou le nombre d’ouvriers présents. Si c’est le cas, on pourrait alors comprendre le texte de la façon suivante: 9 (ouvriers), 2 (jours) – ou l’inverse. Mais il est aussi possible que l’un des deux chiffres corresponde au code identifiant l’équipe d’ouvriers concernée,21 auquel cas, on pourrait comprendre: (équipe) 9, 2 (jours) – ou l’inverse. Le sens à donner à la lettre B est plus problématique encore. C’est en vain que l’on chercherait à y voir le témoin d’une date ou d’une indication calendaire, sur le modèle des marques peintes retrouvées dans les Thermes de Trajan à Rome (Volpe 2002; Volpe 2008; Volpe et Rossi 2012). Les possibilités sont nombreuses, mais rien n’interdit d’y voir une abréviation pour b(onum) ou b(ene), voire b(ene fecit). Si tel est le cas, il s’agirait d’une inscription peinte par le contremaître passé vérifier la qualité du travail réalisé et la rapidité de son exécution. Par ailleurs, on a aussi pu repérer quelques inscriptions tracées dans le mortier frais. On en trouve notamment plusieurs dans l’angle sud-est de la salle 16, à hauteur de la sixième assise du mur de piédroit. Elles ont été faites à la fin de la construction du mur, vraisemblablement par la même main. Le premier ensemble est composé de trois séries de chiffres superposées: XXXI, soit 31, sur la première ligne; IIIIS, soit 4,5, sur la seconde et de nouveau IIIIS, soit 4,5, sur la troisième.22 Les lettres mesurent en moyenne 3 cm de hauteur. Leur lecture est aisée, même si la dernière est relativement effacée ou maladroite. Au même niveau, mais distante des précédentes de 30 cm, une autre inscription est formée de lettres de 2,5 cm de haut. Il s’agit, là encore, de chiffres correspondant probablement à une somme ou à un compte: IIS, soit 2,5. Le sens de ces comptes n’est pas clair, mais il pourrait s’agir d’une indication de volume ou de quantité: nombre d’ouvriers présents, jours travaillés ou lots de matériel utilisé. La dernière inscription gravée dans le mortier frais a été découverte dans l’angle sud-ouest de la salle 18, entre la deuxième et la troisième assise supérieure du mur de piédroit sud. Elle est asso21  On sait en effet que des chiffres servent parfois de marques de tâcherons pour identifier le travail réalisé par une équipe, tel les graffiti “XX” et “XIX” découverts sur des blocs calcaires à Bordeaux (ILA Bordeaux, n.° 386 et 387, p. 578 = IRB, 1, n.° 845-846, pp. 603-604). 22  La lettre S doit certainement être comprise comme l’abréviation de semis (moitié).

Anejos de AEspA LXXVII

Les techniques et les étapes de la construction...

261

Fig. 5. Graffito peint de la salle 08 (les traits ont été retravaillés par les auteurs afin de permettre une meilleure lecture en noir et blanc; doc. D. Hourcade et A. Coutelas).

ciée à une marque de peinture rouge horizontale indiquant un arrêt temporaire de chantier. Les lettres mesurent environ 8 cm de haut et 3 cm de large. La lecture ne semble pas poser de problème. On restitue les lettres capitales BVL. La présence d’une barre relativement horizontale qui croise le sommet des lettres V et L incite à penser que, comme ailleurs, ces derniers caractères sont des chiffres. L’inscription pourrait alors être retranscrite de façon suivante: B 5 50. Le sens à donner à cette inscription est aussi délicat que pour les précédentes, mais au vu des hypothèses formulées pour le graffito peint de la salle 08, on pourrait envisager la restitution suivante: b(ene fecit), 5 (jours), (équipe) 50 ou 50 (ouvriers) – ou toute composition équivalente. Positionnement des entraits Les piédroits construits, il fallait ensuite positionner les cintres pour la mise en œuvre de la voûte. Pour cela, on devait d’abord installer les entraits, poutres sur lesquelles reposaient les cin­tres destinés à supporter les planches de coffrage de la

voûte. Le nombre de cintres dépendait de la longueur de la salle. Dans les salles cendriers (15 à 20), ainsi que dans les salles 06 à 08, il en fallait onze. Ils y sont séparés en moyenne de 100 cm. La distance entre le cintre situé à l’extrémité de la voûte et le parement du mur périphérique est de 30 à 45 cm, soit 1 pied à 1 coudée. Dans les salles 03 et 04, plus longues, on en a, en revanche, repéré dix-sept. L’observation minutieuse des maçonneries a permis de révéler des marques de positionnement préalables de ces entraits, tracées sur le mur à l’ocre rouge par l’artisan afin de repérer l’endroit où installer les cintres. Toutes les salles centrales en ont livré, mais la salle 16 est la mieux pourvue. Ailleurs, nous avons pu observer une nette distorsion des occurrences, avec sept marques dans la salle 08, deux dans la salle 09, une dans la salle 12, mais aucune dans les autres. Notons que six des marques de la salle 08 se suivent – du sixième au onzième entrait. Tous les entraits n’étaient donc pas préalablement positionnés à l’aide de marques peintes. Quand elles sont présentes, elles s’apparentent à un court trait vertical tracé à l’ocre rouge, parfois seulement une incision, correspondant plus ou moins à l’axe central de la future

262

Anejos de AEspA LXXVII

ARNAUD COUTELAS, DAVID HOURCADE

poutre. Le décalage parfois repéré laisse penser que si l’emplacement était globalement indiqué, le positionnement précis de l’entrait pouvait fluctuer de quelques centimètres. Les entraits étaient placés directement sur l’arase des piédroits. Certains ajustements pouvaient être effectués au besoin, comme par exemple pour le piédroit nord de la salle 15 où le dernier rang a été enlevé, probablement suite à une erreur de chantier. Quelques entraits reposaient aussi sur une brique placée sur l’arase du mur. Bien qu’il s’agissait parfois de solutions ponctuelles, c’est en revanche une pratique employée systématiquement pour tous les entraits du piédroit sud de la salle 10. Dans la moitié ouest de la salle 09, une petite pierre plate d’impactite locale était utilisée comme calage pour chaque entrait. A la différence des briques, ces pierres sont systématiquement en débord par rapport au parement du mur. Les entraits reposaient sur le sommet du mur sur une longueur de 10 cm en moyenne, rarement moins de 7 cm,23 jamais plus de 20 cm. Le mortier des voûtes a fréquemment moulé l’extrémité des entraits et l’on constate que la grande majorité de ces pièces de bois présentait une section circulaire, d’un diamètre de 10 cm environ, avec une extrémité taillée, en lame plus qu’en pointe, sinon à fond plat. Parfois, les sections sont rectangulaires, d’environ 10 cm de large pour 20 cm de haut; l’extrémité étant alors chaque fois plate. La longueur que nous avons pu restituer pour ces pièces de bois, dans l’hypothèse où elles faisaient effectivement la jonction entre les deux piédroits, est comprise entre 235 et 270 cm, plus généralement vers 245255 cm. Seule la salle 24, la plus étroite, nécessitait des pièces de bois de longueur réduite, de 155 cm environ. La plus large, la salle 05, aurait nécessité des pièces de bois d’environ 320 cm de longueur. Si certaines empreintes sont parfaitement nettes, le profil de l’entrait étant intégralement moulé par le mortier, d’autres sont beaucoup moins claires, voire inexistantes. Cette différence s’explique par le soin apporté à la protection de certains entraits, avec l’aménagement d’une sorte de cadre autour de la poutre, lors du versement des premiers volumes du blocage de la voûte (cf. infra). La plupart de ces aménagements sont grossiers et emploient de moellons d’impactite mal équarris en montants et en linteau. Pour quelques entraits cependant, on peut noter le

  3 cm est le minimum observé.

23

choix de la brique en linteau. Les espaces ainsi créés sont chaque fois plus importants que la pièce de bois elle-même, ce qui avait sans doute l’avantage de permettre un retrait plus aisé de la poutre au moment du décoffrage. Positionnement des planches Une fois les cintres installés, les planches servant au coffrage de la voûte pouvaient être disposées. Bien que seules en subsistent les empreintes laissées dans le mortier, celles-ci sont parfois d’une telle qualité qu’elles rendent possible l’identification de leurs modes de sciage et de préparation (cf. infra). Plus important, elles permettent aussi de bien comprendre la technique de mise en œuvre du coffrage. Les planches, qui étaient a priori posées à l’aide del cales (couchis), se chevauchaient, aussi bien latéralement que longitudinalement, les unes les autres afin d’empêcher l’écoulement du mortier. Les empreintes repérées ne correspondent donc pas à celles des planches dans leur totalité et on peut considérer que leur longueur, notamment, devait probablement être supérieure de quelques dizaines de centimètres. Quelques empreintes ont néanmoins permis de connaître les dimensions exactes de certaines planches. En effet, leur mise en place s’effectuait, a priori, chaque fois à partir d’une extrémité de la salle vers son opposé. Après moulage par le mortier, seule la section non recouverte des premières planches nous apparaît, avec un effet de relief. En revanche, à l’autre extrémité, la dernière planche est posée en surface sur le coffrage et est donc moulée intégralement par le mortier. En pratique, il s’avère toutefois que l’installation des planches n’est pas toujours aussi régulière, notamment pour les salles au développement le plus important. Ainsi, dans la salle 04, dont la voûte a gardé l’empreinte de cinquante-deux planches, on peut remarquer que les plus longues sont placées en premier sur les piédroits et près des murs périphériques, alors que les plus petites, parfois ajustées grossièrement, se situent au centre. En définitive, on retiendra que seules les empreintes «entièrement en creux» permettent de connaître les dimensions exactes des éléments de bois utilisés. Les longueurs des planches varient dans des proportions importantes, mais on a pu constater l’usage préférentiel d’éléments de grandes dimensions, supérieurs à 3 voire 5 m. Dans la salle 04 où

Anejos de AEspA LXXVII

Les techniques et les étapes de la construction...

toutes les empreintes ont pu être mesurées, la longueur des planches y est comprise entre 0,58 et 10,18 m, avec de nombreux éléments plus petits que 2 m ou de longueur comprise entre 4,50 et 6,50 m. Les largeurs des planches sont presque toutes inférieures au pied romain. Elles varient généralement entre une dizaine et une vingtaine de centimètres. Il est d’ailleurs possible que les ouvriers antiques aient employé des demi-planches.24 Enfin, l’épaisseur exacte des planches est parfois lisible. Elle est comprise entre 0,7 et 3,9 cm. La nature du bois utilisé pour ces éléments de coffrage reste indéterminée. Néanmoins, l’empreinte en coupe de l’extrémité de deux planches de la salle 20, où des concrétions calcaires se sont formées par infiltration, permet de voir clairement qu’il s’agissait d’un bois à zone poreuse, très vraisemblablement du chêne. Mise en œuvre de la voûte Sur le coffrage de planches est coulé un opus caementicium fait de mortier de chaux et de sable englobant des éclats d’impactite. La clé de voûte est constituée de briques et le volume coulé, ou appliqué sur le coffrage, se limite à une simple «couronne» de 40 cm d’épaisseur sur laquelle viendra plus tard un blocage beaucoup plus fruste. Lorsqu’il est mis en œuvre, le mortier se présente telle une masse visqueuse, plus ou moins fluide selon la quantité d’eau ajoutée. Bien que les traces de quelques infiltrations de mortier entre les planches ont souvent été repérées, il n’y pas lieu de penser que les ouvriers ont rencontré de gros soucis d’écoulement de matière. Le mortier semble en effet assez compact. Le fait que les planches reposaient généralement sur l’extrémité des entraits plutôt que directement sur l’arase du mur, permet de penser que la construction de la couronne débutait par la pose de pierres sur le sommet des piédroits. De fait, les retombées des voûtes sont des zones assez grossières, peu travaillées, sauf peut-être lorsque les pierres servaient aussi à aménager des cavités d’encadrement des entraits. Parfois ces zones sont mieux préparées, avec des moellons de plus grandes dimensions et placés en épi, comme par exemple en salle 09 et en salle 24. 24  C’est du moins l’hypothèse formulée par Christelle Belingard au vu des cernes de croissance asymétriques repérées sur les empreintes de quelques planches des salles 19 et 20.

263

Cette zone située entre la première planche et l’arase du piédroit a parfois été retravaillée à la truelle. On y trouve aussi de petites cavités qui nous semblent être, à l’instar de celles des piédroits, des tests de prise. Il est fort probable que cette zone non coffrée de la voûte ait pu offrir un secteur accessible aux vérifications de prise de l’opus caementicium, avant toute étape de décoffrage. Décoffrage La phase de décoffrage intervenait une fois la voûte suffisamment consolidée par la prise du mortier. De la sorte, on pouvait récupérer les pièces de bois pour leur éventuel remploi car elles n’étaient jamais abandonnées dans les salles. Cette étape devait elle aussi présenter une certaine complexité. En effet, si l’enlèvement du cintre était parfois préparé en amont par l’aménagement de cavités d’encadrement des entraits – qui permettaient de faire glisser la pièce de bois hors de son logement –, la plupart du temps l’entrait était englobé dans le mortier de la voûte, rendant son extraction plus difficile. De plus, l’enlèvement du cintre n’impliquait pas automatiquement la chute des planches du fait de l’adhérence entre le bois et le mortier de chaux de la voûte. Enfin, l’évacuation des éléments en bois devait parfois poser problème puisqu’aucune des salles ou groupes de salles de soutènement, à l’exception des salles 5, 15 à 20, 24 et 26, ne comportaient d’accès extérieur de plain-pied. Les traces liées au chantier de décoffrage ont été retrouvées dans toutes les salles. A chaque fois, il s’agit des empreintes laissées dans le mortier par l’outil utilisé afin de dégager les pièces de bois de leur «gangue». Ainsi, on a pu retrouver le négatif d’une pointe en pic, à section carrée aux angles arrondis, mesurant environ 2 × 2 cm. Une partie de la tête de cet outil s’est imprimée au niveau du négatif d’une planche de coffrage de la voûte de la salle 13 (fig. 6), probablement au moment du décoffrage de la pièce de bois. On distingue le sommet de la tête en forme de feuille de laurier et son œil servant à l’emmanchement au centre. Plus qu’une ascia, outil spécifique au travail du bois, il s’agit sans doute plutôt d’un outil de maçon à percussion lancée, muni d’un manche, de type «marteau pic». Le décoffrage des planches a laissé beaucoup moins de traces que le dégagement des cintres. C’est dans les salles 17 et 20 que l’on trouve le plus d’empreintes de coups au pic. Elles se re-

264

ARNAUD COUTELAS, DAVID HOURCADE

Anejos de AEspA LXXVII

Fig. 6. Empreinte de la tête de l’outil utilisé pour le décoffrage des planches, salle 13, par-dessus une empreinte de triangle d’arrachage (de sciage) (cliché A. Coutelas).

trouvent surtout autour des entraits, secondairement à la retombée de la voûte. Ces marques, qui paraissent très fraiches et sont très profondes, indiquent que le mortier était donc encore très humide et n’avait pas eu le temps de faire totalement prise, ce qui a particulièrement dû gêner la phase de décoffrage. L’effet de succion du mortier était encore important, nécessitant ainsi d’user plus que de coutume des outils de percussion pour arracher les pièces de bois de la maçonnerie. Au vu de ces observations, il est tentant de croire que le décoffrage des voûtes des salles cendriers a commencé au sud, pour se poursuivre ensuite vers le nord, salle après salle. Excepté les salles cendriers 15 à 20, le couloir central 26 et les salles n° 05 et 24, toutes les salles de soutènement ont donc été scellées à l’issue de leur construction. Cette fermeture définitive n’est pas intervenue à une extrémité de la salle, forcément cloisonnée par un des murs porteurs des thermes, mais par le bouchage d’une ouverture laissée dans la voûte. Il s’agit en fait d’une cheminée de sortie qui avait été aménagée pour permettre d’évacuer les planches et les cintres du cof-

frage des voûtes une fois le mortier suffisamment sec, ainsi que la sortie des derniers ouvriers. Il semble qu’il n’y avait qu’un seul de ces «puits d’extraction» pour chaque lot de trois ou quatre salles. Ainsi, pour les salles 09 à 11, cette cheminée a été retrouvée à peu près au centre de la voûte de la salle 10, à hauteur des passages transversaux (fig. 7). De section carrée et mesurant 53 × 54 cm, sans aménagement particulier visible, cette ouverture de chantier était donc assez large pour permettre d’extraire une planche de coffrage vers l’étage. Au sein d’un même groupe de salles de soutènement, le déplacement des planches d’une salle à l’autre était facilité par le fait que les passages entre les salles étaient construits de biais. De plus, leur alignement –attesté pour les salles 01 à 04 et 09 à 11 et qui devait être la règle ailleurs–25 aidait encore plus à la manutention des planches les plus longues.26 25  L’inversion que l’on observe entre l’orientation du passage entre les salles 12 et 13 et celle du passage entre les salles 13 et 14 résulte en fait d’une mauvaise restauration de la maçonnerie par les découvreurs du site. 26   La salle 4 ayant livré les dimensions les plus importantes pour les planches (jusqu’à 10,18 m), nous avons tenté d’estimer la longueur maximale d’une planche pouvant circuler par le

Anejos de AEspA LXXVII

Les techniques et les étapes de la construction...

Fig. 7. Puits d’extraction des planches dans la salle 10 (cliché A. Coutelas).

265

266

Anejos de AEspA LXXVII

ARNAUD COUTELAS, DAVID HOURCADE

La salle 23 permet d’observer la technique de bouchage de ce type de cheminée d’extraction. L’ouverture initiale, de 44 × 44 cm, était de section carrée. Elle ne semble pas avoir bénéficié d’aménagement particulier, mais on note tout de même la présence de gros moellons vers la retombée de la voûte, peut-être des renforts tardifs. Le bouchage est composé d’un volume de moellons d’impactite et de fragments de terres cuites architecturales – des tegulae placées de chant surtout –, pris dans un mortier de chaux et de sable siliceux, de coloration moutarde. Au moment de sa mise en œuvre, ce volume reposait peut-être sur un léger coffrage, de type branches de clayonnage, si l’on en croit les quelques restes de concrétions carbonatées de section circulaire que l’on a retrouvé. Pour le groupe de salles 01 à 04, la cheminée est aménagée à l’extrémité orientale de la salle 01, dans le premier tiers de la moitié nord de la voûte et non pas dans son axe, vraisemblablement pour des raisons de sécurité et de stabilité. Cette salle 01 est sans doute la dernière construite du groupe. L’ouverture rectangulaire a été cette fois bouchée par un bloc d’impactite retaillé en forme de coin trapézoïdal. Le blocage au-dessus de la voûte Enfin, avant que ne soient coulés les sols des pièces de l’étage, fréquentées par les baigneurs et les curistes, la dernière étape de la construction des salles de soutènement correspondait à la mise en place du blocage recouvrant toute les voûtes et permettant de réaliser un niveau plan à l’altitude souhaitée. Les écorchés des salles 01 et 02 permettent de lire très clairement cette technique. On remarque une alternance de remblais pierreux (sans liant) et de chapes, plus fines, de mortier de chaux. Selon les endroits, on note cinq à neuf couches de remblais, formant ainsi comme une sorte de fondation multicouches, laquelle est ensuite recouverte par une chape de mortier scellant le tout et servant de sol de travail aux ouvriers chargés d’aménager le sol de la pièce de l’étage, à savoir ici le frigidarium F2. passage entre les salles 03 et 04, sachant que le puits d’extraction se trouve en salle 01. Nos essais, par nature empiriques, indiquent une longueur maximale d’environ 8,50 m. Si ces hypothèses sont exactes, on doit donc en conclure que quelques planches n’ont pas pu être extraites en l’état et qu’elles ont dû être découpées avant évacuation.

LA GESTION DES PLANCHES DE COFFRAGE Comme on le voit, la volonté de récupérer les planches de coffrage et les autres éléments en bois a eu une incidence sur l’architecture des salles de soutènement, notamment par l’aménagement de «puits d’extraction» au sein de plusieurs voûtes. La question qui se pose est de savoir si cette phase de récupération est intervenue à la fin du chantier général, ou si elle a eu lieu durant la campagne d’érection des salles de soutènement, de sorte que les planches employées pour le coffrage de la voûte d’une salle ont été réutilisées pour le coffrage d’une autre? Dans le but d’étudier le remploi éventuel des planches, nous avons cherché à recenser tous les éléments qui pourraient permettre de les caractériser et de les identifier. Pour ceci, nous avons tenté de réaliser un catalogue des planches sur la base de leurs dimensions et des «marques distinctives» qu’elles recèlent parfois: marques de tâcheron, triangles de sciage et défauts de planche. Nous avons vérifié en détail la position, la morphologie et les dimensions des différentes marques, ainsi que  les longueurs, visibles et parfois réelles, des planches. Une telle minutie était nécessaire car les différentes marques se ressemblent souvent et, devant leur nombre, on s’aperçoit que leurs dimensions et leur emplacement peuvent trouver un écho sur d’autres empreintes laissées par des planches en réalité distinctes. Nous présenterons tout d’abord ces différentes «marques distinctives» avant de discuter les résultats issus de ce catalogue. Les marques de tâcheron Les «marques de tâcheron» sont des lettres gravées dans les planches qui se retrouvent en relief dans le mortier de la voûte. Elles sont de plusieurs types (fig. 8): A, N, P, R ou B, parfois s­ eulement un trait vertical ou oblique. Nous postulons qu’il s’agit de signatures laissées par le ou les artisans afin que leur travail puisse être comptabilisé. Néanmoins, bien que le marquage du bois puisse paraître évident, par analogie avec ce qui se pratique avec d’autres matériaux de construction, comme la pierre ou les terres cuites architecturales, les témoignages de cette habitude sont rares pour la période antique.27 D’autres hypothèses doivent donc être 27  Nous remercions Christelle Belingard d’avoir attiré notre attention sur ce point. Seuls les pieux de fondation semblent parfois marqués d’un chiffre, comme pour le quai de

Anejos de AEspA LXXVII

Les techniques et les étapes de la construction...

267

Fig. 8. Quatre exemples de marques de tâcheron gravées sur des planches de coffrage. Un A (1), un P (2), un N (3) et un R ou un B (4) (doc. A. Coutelas).

évoquées. On pourrait d’abord penser qu’il s’agit de marques de lot – destinées à identifier un client ou une utilisation précise, par exemple – mais la découverte de plusieurs marques sur une même planche ne milite pas pour cette interprétation.28 Une autre hypothèse, liée au sciage des planches, est également possible. En effet, l’iconographie antique indique – si ce n’est pas une simple convention – que deux tâcherons étaient nécessaires pour le sciage de long.29 Si une marque pourrait correspondre à une équipe, deux marques30 pourraient indiquer l’in­ tervention d’une deuxième brigade après retour­ nement de la grume ou de la bille de bois. Malheureusement, cette interprétation ne permet pas d’expliquer les cas rencontrés de planches portant trois marques. Enfin, il est aussi possible que ces Rezé ou le cirque d’Arles. Pour ce dernier exemple, des timbres de bûcheron frappés à froid pourraient aussi être présents (Sintès et al. 1990). 28   Ainsi, dans la salle 13, quatre planches étaient incisées de trois marques de tâcheron chacune, selon quatre cortèges différents. 29   On pense notamment à la peinture de la «Procession de charpentiers», mise au jour à Pompéi en VI, 7, 8. 30   Les planches avec deux marques de tâcheron sont aussi fréquentes que celles avec une seule marque.

marques n’aient été réalisées que lors du chantier par les maçons, pour les aider à gérer les importants stocks de planches pour les différents coffrages, ou pour noter chaque utilisation en cas de réemploi. La répartition des marques, ou du moins des planches marquées, au sein des différentes salles est inégale. Or, la conservation différentielle des voûtes ne permet pas d’expliquer à elle seule ce fait. Parmi les salles cendriers, c’est la salle 17 qui possédait le plus de planches marquées: quatre (et six marques en tout). Dans le reste de l’édifice, toutes les autres salles en possèdent. Ainsi, dans la salle 14 sont conservées les empreintes de douze planches gravées, soit quinze marques. Dans la salle 13 on compte même dix-huit planches et trente-deux marques. Une telle différence est difficile à expliquer et l’étude de la répartition des symboles ne nous ait malheureusement d’aucune utilité. Nous ne savons pas si toutes les planches étaient marquées ou non, mais si on imagine que la plupart l’étaient, peut-être que les planches étaient généralement disposées avec les marques tournées vers le haut et que leur (hypothétique) réutilisation nécessitait de les retourner parfois. Toutefois, le plus probable reste que les artisans

268

ARNAUD COUTELAS, DAVID HOURCADE

n’apposaient leur signature que sur un nombre limité de planches, représentatif de leur travail. Les triangles de sciage Les planches étaient généralement obtenues par le sciage de long de billes ou de poutres posées en oblique sur un tréteau. Deux ouvriers procédaient au sciage, l’un positionné en hauteur, l’autre à terre. Une fois la première partie sciée jusqu’au tréteau, la pièce de bois était basculée ou retournée et le sciage reprenait depuis l’autre bout. La rencontre des deux plans de sciage ne se faisant pas toujours à la perfection, la planche était alors désolidarisée du reste de la pièce d’un coup sec, laissant une marque triangulaire, du fait de l’orientation inverse entre les passages de la scie.31 Les formes exactes et les dimensions de ces triangles ne sont bien évidemment pas normalisées.32 Parfois équilatéraux ou trapus, rarement effilés, ils sont le plus souvent isocèles (fig. 9). Leur hauteur correspond généralement à la largeur de la planche, soit 10 à 20 cm environ, mais elle est parfois plus réduite.33 Au total, quarante-sept triangles ont été repérés dans les salles cendriers et quatre-vingt-onze supplémentaires dans le reste des pièces. Proportionnellement, c’est la salle 07 qui en contient le plus, mais c’est dans les salles 10 et 13 qu’ils sont les plus nombreux. Leur position sur la planche34 et leur orientation35 ont été notées afin de faciliter leur indivi31  La position du triangle sur la planche n’est pas obligatoirement centrée. Elle dépend de la longueur de la bille et du moment choisi par les ouvriers pour retourner la pièce de bois. D’ailleurs, rares sont les éléments repérés vers le milieu de la planche; ils sont généralement localisés vers un des tiers de la longueur. En revanche, lorsque le triangle se trouve très près de l’extrémité d’une planche – comme c’est le cas pour certaines planches des salles 07, 08 et 15 – tout indique qu’il y a eu découpe de l’objet. La question est alors de savoir si elle est intervenue immédiatement après le sciage ou, plus tard, pour la mise en œuvre, voire lors de la récupération de la planche. 32   Toutes les planches n’en sont d’ailleurs pas pourvues. Ce fait peut surprendre, mais il pourrait démontrer soit que la qualité du sciage manuel était telle qu’il n’a pas laissé de traces, soit que les planches «lisses» ont été en fait retaillées dans des planches plus grandes – l’autre morceau portant la trace du triangle –, soit, encore, qu’une autre méthode de sciage était aussi employée. 33   Le plus petit triangle repéré mesure 6,5x3 cm. 34   Afin de pouvoir travailler à partir d’une distance réelle, non altérée par le chevauchement des planches, la position est calculée à partir du bord de l’empreinte; celui collé à un mur périphérique quand cela était possible. On n’oubliera pas cependant qu’une même planche présente normalement deux triangles d’arrachage différents; un sur chaque face. 35  Cette information est importante car elle permet de repérer les éventuels triangles «miroirs»; traces laissées lors du

Anejos de AEspA LXXVII

dualisation et d’étudier l’éventuelle réutilisation des éléments de bois. La question de la réutilisation des planches et de leur circulation Au final, tous ces indices permettent-ils d’attester la réutilisation et la circulation des planches d’une salle à l’autre? La réponse est délicate. Globalement, les résultats sont en effet assez ténus et les éventuels rapprochements incertains. Ainsi, deux marques de tâcherons semblent avoir été faites par le même ouvrier dans les salles 05 et 14, mais rien ne permet d’affirmer qu’il s’agit bien des mêmes planches. Il en est de même pour deux marques découvertes en salles 04 et 17 et pour deux autres en salles 11 et 13. On notera néanmoins que, dans ces trois cas, le remploi des planches est d’autant plus probable que les pièces dans lesquelles elles ont été découvertes appartiennent à des groupes distincts de salles de soutènement, c’est-à-dire construites par des équipes distinctes ou à des moments différents. Pourtant, quelques résultats semblent réellement positifs et pourraient confirmer de façon plus certaine la circulation des planches. Ainsi, quelques correspondances ont été révélées dans les salles cendriers grâce à la comparaison des triangles de sciage. C’est le cas de la deuxième planche de la salle 16 et de la cinquième planche de la salle 20 où les empreintes du triangle sont identiques et sont placées au même endroit. C’est également le cas pour la deuxième planche de la salle 15 et la vingt-huitième de la salle 20.36 La correspondance est surtout parfaite pour le triangle de la sixième planche de la salle 15 et celui de la première planche de la salle 20.37 De même, la comparaison des marques de tâcheron a permis d’identifier une autre planche utilisée pour plusieurs coffrages. Cette planche, longue de plus de 190 cm de long, est marquée d’un P resserré, haut de 20 cm, à environ 50 cm sciage à la fois sur la planche prélevée et sur la face de la bille. C’est peut-être le cas pour quelques couples de planches des salles 04 et 15, 06 et 07, et enfin 11 et 20. 36   Dans ces deux cas de «couples», les longueurs de planche ne correspondent pas, mais on peut supposer qu’un chevauchement occulte une partie des dimensions de l’une d’entre elle, voire qu’elles ont dû être retaillées avant remploi. 37   Les dimensions des triangles sont les mêmes à un ou deux centimètres près. Leur emplacement, sur la largeur et dans la longueur de la planche, à 353 cm d’un bord, coïncident. Enfin, les planches mesurent environ 5,50 m toutes les deux.

Anejos de AEspA LXXVII

Les techniques et les étapes de la construction...

269

Fig. 9. Triangle de sciage (salle 17) (cliché A. Coutelas).

d’un de ses bords (voir fig. 8). On en retrouve l’empreinte dans les salles 06, 11 et 18. Les résultats sont donc minces, mais ils existent et ils prouvent, selon nous, qu’il y a eu circulation des planches d’un espace de construction à l’autre.38 Les difficultés que l’on rencontre pour les suivre de salle en salle peuvent s’expliquer par le fait que ces objets pouvaient, d’une utilisation à l’autre, être soit retournés, en partie salis de mortier, soit recoupés, soit diversement chevauchés... La première conclusion de cette étude de distribution des marques est que les six salles cendriers n’ont vraisemblablement pas été couvertes en même temps, mais probablement plutôt trois par trois: 15 à 17, d’une part, et 18 à 20, de l’autre. Cela n’est pas étonnant, puisque les autres salles semblent elles aussi généralement construites par groupe de trois comme l’indique le plan du bâtiment. Néanmoins, les autres salles étant plus grandes, elles ont forcément nécessité soit des lots plus importants de planches, soit des planches 38   L’emploi d’une même planche est donc assuré entre les salles 15 et 20 d’une part, et entre les salles 06, 11 et 18 d’autre part.

plus longues, soit des chevauchements moins importants entre celles-ci. La seconde conclusion est que le soin apporté à la récupération de ces éléments en bois était, au moins pour partie, conditionné par le besoin de leur réemploi au sein même du chantier. Le cas des pièces 06 à 08 donne un bon exemple de l’importance parfois accordée par les ouvriers à la récupération et au remploi de ces matériaux de construction en bois. Ce trio de salles se distingue en effet par l’absence de «puits d’extraction» dans leurs voûtes. On pourrait en déduire que les éléments de leurs coffrages n’ont pas été récupérés, voire que, puisque l’une des planches de la salle 06 se retrouve également dans les salles 11 et 18, les bâtisseurs ont terminé la construction des salles de soutènement par elles. En réalité, il existe bien une ouverture qui aurait pu servir à l’évacuation des planches, mais elle ne se situe pas au niveau des voûtes, mais dans les parois. Ainsi, au nord de la salle 07, un trou longtemps considéré comme moderne perce grossièrement le mur porteur qui, à l’étage, sépare les pièces F2 et D/U2. Le tunnel creusé dans l’opus caementicium débute depuis la salle 4 par une baie parfaitement maçonnée dans le mur de piédroit sud (fig. 10). Or cet aménage-

270

ARNAUD COUTELAS, DAVID HOURCADE

ment n’aurait pas de sens si les deux groupes de salles ne communiquaient déjà pas entre eux au moment de leur construction.39 De toutes ces observations, on déduit d’abord que l’extraction des planches du coffrage des salles 06 à 08 par le mur nord a été planifiée non seulement avant la construction des murs piédroits de ces trois salles et de ceux de la salle 04, mais aussi avant la pose de leurs voûtes. On peut également en conclure que la construction de la salle 04 – voire 01 à 03 – devait être contemporaine, ou de très peu postérieure, à celle des salles 06 à 08, faute de quoi le besoin d’aménager un passage ne se serait pas fait sentir. Néanmoins, on peut légitimement se demander pourquoi les bâtisseurs ont préféré attaquer et percer le soubassement d’un des murs principaux et porteurs de l’édifice plutôt qu’aménager un puits d’extraction au sein de l’une des voûtes comme ils l’ont fait partout ailleurs. Aucune contrainte architectonique ne permet d’expliquer ce choix. La pièce de l’étage, située au-dessus des salles 06 à 08, est en effet le destrictarium/ unctorium D/U2. Or, à l’instar de la pièce D/U1, construite au-dessus des salles 21 à 23, il s’agissait d’une simple salle sèche construite sur hypocauste. Pour l’heure, la seule raison que nous puissions envisager est que, pour des raisons de coûts ou de gain de temps, les bâtisseurs avaient le besoin impérieux de réutiliser les planches des coffrages des salles 06 à 08 pour réaliser ou, plutôt, pour compléter ceux des voûtes des salles 01 à 04. Pour cela, ils ont été prêts à changer leurs méthodes habituelles de fonctionnement et à percer un puissant mur d’opus caementicium pour cela. LA QUESTION DE LA PRÉSENCE DE PLUSIEURS ÉQUIPES L’identification minutieuse des planches et l’étude fine de la répartition des marques considérées comme significatives nous ont permis d’attester la circulation des éléments de bois au sein du chantier, mais qu’en est-il des hommes eux-mêmes? Les données recueillies sont-elles suffisantes pour identifier et individualiser le travail d’un groupe d’ouvriers et repérer les zones du chantier sur lesquelles il est intervenu? Il nous semble que c’est le cas.

39   On note d’ailleurs que les passages entre les salles 06/07 et 07/08 ne sont pas alignés, mais tous deux dirigés vers l’extrémité nord de la salle 07.

Anejos de AEspA LXXVII

En effet, la compilation et la localisation de toutes les techniques, marques et traces identifiées permettent de repérer certaines pratiques récurrentes, voire systématiques, ou au contraire isolées. Ainsi, il semble tout abord évident que les six salles cendriers centrales ont été réalisées par une même équipe, ou du moins un groupe travaillant avec les mêmes techniques et les mêmes habitudes. En effet, pour les salles 15 à 20, on remarque que l’emplacement des futurs murs piédroits a chaque fois été marqué, que les parements sont toujours construits en opus vittatum simple, que l’emplacement des futurs entraits a fréquemment été indiqué et qu’un même stock de planches a été utilisé pour le coffrage des voûtes puisque l’une d’elles a circulé de la salle 15 à la salle 20. Parallèlement, on note que la même technique de réalisation des murs et que la même habitude d’emploi de la peinture de chantier s’observent dans le groupe des salles 06 à 08. Or, puisqu’une même empreinte de planche a été retrouvée dans les salles 06 et 18 et que la marque d’un même tissu semble avoir été laissée à la fois dans les salles 07 et 20, il paraît très probable que la même équipe d’artisans est intervenue pour construire les salles 15 à 20 et 06 à 08. Dans les autres groupes de salles,40 les résultats sont moins homogènes ou moins clairs. Ainsi, les salles 05 et 24, isolées toutes les deux, brillent par leurs nombreuses originalités.41 Le groupe de salles 01 à 04 se distingue aussi quelque peu des précédentes: il est composé de quatre pièces au lieu de trois, toutes de grande hauteur; quelques piédroits présentent une assise de briques, absente ou rare ailleurs; les niveaux de trous de boulin sont complétés avec des trous de pieux; les traces de peintures sont très peu nombreuses. En fait, dans ce groupe, seule la salle 04 en porte plusieurs. Or, puisque c’est aussi dans cette pièce que l’on a trouvé une empreinte de tissu équivalente à celle repérée dans les pièces 07 et 20 et que l’on a déjà noté la singularité du trou creusé dans le soubassement d’un mur porteur pour faire communiquer les salles 07 et 04, il est fort pro40   On rappelle que les salles 21 à 23 n’étant pas entièrement dégagées et accessibles, elles sont exclues de la discussion. 41   La salle 05 est beaucoup plus large que les autres, alors que la salle 24 présente un aménagement de bas de voûte singulier, à larges pierres en épis. Ces deux salles ne portent d’ailleurs pas de trace de peinture.

Anejos de AEspA LXXVII

Les techniques et les étapes de la construction...

271

Fig. 10. Baie aménagée dans le piédroit sud de la salle 04 (cliché A. Coutelas).

bable que cette dernière a été aménagée par l’équipe chargée de la construction des espaces 06

à 08 et 15 à 20. Dans ce cas, les salles 01 à 03 auraient été érigées par un autre groupe d’artisans.

272

Le groupe de salles 12 à 14 est moins évident à associer aux autres car il présente des caractéristiques proches de celles des salles centrales, mais avec beaucoup moins d’occurrences de trace de peinture et aucun élément déterminant. La présence d’un rang de briques dans le piédroit central, entre les salles 13 et 14, rappelle plutôt le groupe 01 à 03. Enfin, le groupe 09 à 11 se distingue très fortement du reste de l’édifice et les données semblent mélangées. Seule la pièce 09 montre des marques de peinture, notamment en équerre comme dans les salles 04 et 06. On serait donc tenter de la rapprocher du travail effectué par l’équipe chargée des salles 06 à 08 et 15 à 20. Cependant, le piédroit nord de la salle 09 montre deux rangs de briques et les entraits reposent sur des pierres en saillies, soit deux caractéristiques originales. Parallèlement, la salle 11 montre la même empreinte de planche que celle découverte en salles 06 et 18, mais aucune trace de peinture. De plus, le mur de séparation entre les salles 10 et 11 est en opus mixtum. Enfin, la salle 10, caractérisée par un mur sud élevé en opus mixtum, une absence de traces de peinture et la présence de briques en support des entraits, s’écarte de toute comparaison. Au vu de ces remarques, nous pouvons supposer que la confection de la majorité des salles est l’œuvre d’une seule et même équipe, intervenue dans les pièces 04, 06 à 08 et 15 à 20. Les salles 01 à 03 et 12 à 14, où les différences de techniques repérées ne sont que minimes, pourraient avoir été construites soit par une autre équipe respectant plus ou moins strictement les mêmes protocoles, soit par la même équipe n’ayant pas eu besoin d’autant de repères ou étant intervenue avec plus de liberté. En revanche, les salles 09 à 11 montrent de trop nombreuses différences pour que l’on puisse affirmer qu’il s’agit strictement de la même équipe ou encore de la même étape de chantier.42 Quant aux salles 05 et 24, l’originalité de leurs techniques de mise en œuvre empêche toute comparaison, mais elle pourrait dépendre directement des spécificités architecturales et du rôle particulier de ces espaces.43  Néanmoins, la découverte en salle 11 d’une planche aussi employée dans les salles 06 et 18 assure que la construction de ces salles voûtées appartient elle aussi au chantier originel. Peut-être que l’anomalie rencontrée témoigne en fait d’une séquence particulière? Peut-être s’agit-il des premières ou des dernières salles érigées? 43   Si cette hypothèse est exacte, il faudrait donc en déduire que les graffiti retrouvés dans les salles 08 et 18 concernent les 42

Anejos de AEspA LXXVII

ARNAUD COUTELAS, DAVID HOURCADE

CONCLUSIONS Cette étude des salles de soutènement des thermes de Longeas à Chassenon a livré nombre de résultats inattendus. Grâce à l’exceptionnel état de conservation du site, des pratiques longtemps ignorées, ou extrêmement mal documentées, sont désormais attestées et identifiées pour l’époque antique. Ainsi, toutes les étapes de la construction nous sont maintenant connues, même les plus inattendues comme l’emploi de la peinture ocre pour signaler à même les murs l’emplacement futur de différents éléments ou pour marquer l’avancement du chantier. Les «tests de prise» sont aussi une découverte car cette pratique, si elle semble logique, n’avait jamais été mise en évidence sur d’autres sites. L’observation minutieuse des parois a aussi permis de découvrir d’inhabituelles –et pourtant très nombreuses– marques laissées, parfois volontairement, par les artisans. Les traces d’outils –principalement de truelles, mais aussi de marteaux pics employés pour l’enlèvement des cintres et le décoffrage des voûtes–, sont présentes dans toutes les salles de soutènement dans des proportions rarement vues ailleurs. Les empreintes de doigts, de coudes et de genoux nous renseignent certes sur les positions et les gestes exécutés, mais ils créent aussi une proximité inattendue et rare avec ces acteurs anonymes de la construction d’un bâtiment monumental. Par ailleurs, l’étude de l’ensemble de ces empreintes nous fournit aussi des indices probants sur la circulation des artisans et des matériaux en bois à l’intérieur du chantier. L’image qui se dessine désormais du chantier de construction de ces salles de soutènement est relativement claire. Il semble avoir été réalisé par étapes – un groupe de salles après l’autre – et principalement par une même équipe.44 En effet, la réelle unité qui se dégage des salles 04, 06 à 08 et 15 à 20 paraît indiquer l’intervention d’un même groupe d’ouvriers, dans un laps de temps réduit. Les autres salles, où les empreintes et les marques mêmes équipes. L’un des deux chiffres ne pouvant alors correspondre à l’identifiant du groupe d’artisans, ces inscriptions indiqueraient le total de journées/homme utilisées, respectivement 18 et 250. Malheureusement, ces deux résultats sont trop différents pour être acceptés sans réserve. 44   Si tel est le cas, la durée de la construction a dû être relativement longue et l’on comprend mieux pourquoi la stratigraphie et le mobilier archéologique associé semblent indiquer que le bâtiment thermal a été achevé en plusieurs étapes au cours du iie s. ap. J.-C. (Hourcade et al. 2012).

Anejos de AEspA LXXVII

Les techniques et les étapes de la construction...

laissées sont différentes, pourraient avoir été aménagées soit par une autre équipe, soit durant une autre phase du chantier originel. BIBLIOGRAPHIE Attoui, R. 2012: «Segni di cantiere nella «Palestra» di  Villa Adriana a Tivoli», dans Camporeale, S., Dessales, H. et Pizzo, A. (dir.), Arqueología de la construcción I. Los procesos constructivos en el mundo romano: Italia y provincias occidentales (Mérida, ­25-26 de octubre de 2007), pp. 49-66, Anejos de Archivo Español de Arqueología 50. CSIC, Mérida. Coutelas, A. (dir.) 2009: Le mortier de chaux, Collection «Archéologiques». Éditions Errance, Paris. Coutelas, A. 2012: «Les mortiers de chaux et de sable: produits d’un artisanat et témoins du chantier de construction», Aquitania, 28, pp. 171‑178. Doulan, C., Hourcade, D., Guédon, St., Laüt, L., Rocque, G. et Sicard, S. 2014: «Cassinomagus (Chassenon, Charente): une «agglomération de confins» de la cité des Lémovices?», dans Bedon, R. (dir.), Confinia. Confins et périphéries dans l’Occident romain. Actes du colloque international (Limoges, 19 et 20 octobre 2012), pp. 311-343. Presses Universitaires de Limoges, Limoges. Doulan, C., Laüt, L., Coutelas, A., Hourcade, D., Rocque, G. et Sicard, S. (dir.) 2012: «Cassinomagus: l’agglomération et ses thermes. Résultats des recherches récentes (2003-2010) à Chassenon (Charente)», Aquitania, 28, pp. 99-298. Duvauchelle, A. 2005: Les outils en fer du Musée Romain d’Avenches, Documents du Musée romain d’Avenches 11. Association pro Aventico, Avenches. Gaitzsch, W. 1980: Eiserne Römische Werkzeuge: Studien zur römischen Werkzeugkunde in Italien und den

273

nördlichen Provinzen des Imperium Romanum, BAR International series 78. B.A.R., Oxford. Hourcade, D. 1999: «Les thermes de Chassenon (Charente), l’apport des fouilles récentes», Aquitania, 16, pp. 153-181. Hourcade, D., Doulan, C., Perrot, X., Bobée, C. et Soulas, S. 2012: «Plan et chronologie des thermes de Longeas (Chassenon, Charente): nouveau bilan», Aquitania, 28, pp. 131-148. Hourcade, D. et Maurin, L. 2013: «Mars Grannus à Cassinomagus (Chassenon, Charente)», Aquitania, 29, pp. 137-153. Manning, W. H. 1985: Catalogue of the Romano-British Iron Tools, Fittings and Weapons in the British Museum. British Museum Publications, London. Sintès, C., Daudibertières, C. et Bremond, J. 1990: Carnets de fouilles d’une presqu’île, Catalogue d’exposition, Revue d’Arles 2. Musée d’Arles, ­ Arles. Volpe, R. 2002: «Un antico giornale di cantiere delle Terme di Traiano», Römische Mitteilungen, 109, pp. 377‑394. Volpe, R. 2008: «Le giornate di lavoro nelle iscrizioni dipinte delle Terme di Traiano», dans Caldelli, M. L., Gregori, G. L. et Orlandi, S. (dir.), Epigrafia 2006. Atti della XIVe rencontre sur l’épigraphie in onore di Silvio Panciera con altri contributi di colleghi, allievi e collaboratori, pp. 453‑466, Tituli 9. Quasar, Roma. Volpe, R. et Rossi, F. M. 2012: «Nuovi dati sull’esedra SudOvest delle Terme di traiano sul Colle Oppio : percorsi, iscrizioni dipinte e tempi di costruzione», dans Camporeale, S., Dessales, H. e Pizzo, A. (a cura di), Arqueología de la construcción III. La economía de las obras (Paris, 10-11 de Diciembre de 2009), pp. 69-81, Anejos de Archivo Español de Arqueología 69. CSIC, Madrid-Mérida.

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

TRACCE DI CANTIERE DALL’AREA DEL CAPITOLIUM DI BRESCIA: EVIDENZE ARCHEOLOGICHE E MATERIALI DAI RECENTI SCAVI* ANTONIO DELL’ACQUA Università Cattolica del Sacro Cuore di Milano / Eberhard Karls Universität Tübingen

RIASSUNTO: Si presentano i risultati delle recenti indagini archeologiche condotte nell’area del Capitolium di Brescia (Brixia), città romana della Regio X Venetia et Istria. Grazie agli scavi della Soprintendenza archeologica della Lombardia condotti dal 2009 al 2014 sono stati raccolti nuovi dati relativi all’area sacra che dal ii sec. a.C. e fino all’età flavia conobbe quattro fasi costruttive. In questa sede si analizzano le tracce di cantiere e i materiali archeologici riconducibili alla ristrutturazione della fase repubblicana e a quella tra lo smantellamento del santuario augusteo e il rifacimento in età flavia. Nello specifico, sono stati individuati un’officina per la lavorazione dei metalli, con i relativi strumenti utilizzati, e uno strato di scaglie in pietra di médolo e in pietra di Vicenza relativi alla fase di cantiere del santuario repubblicano; un graffito su intonaco con le modanature della cornice della porta maggiore del tempio e incisioni per la definizione della griglia modulare sono invece riconducibili alla ricostruzione del tempio in epoca flavia. PAROLE CHIAVE: Cantiere, Capitolium, Brescia, Brixia, Graffito, Modulo, Officina, Scalpello, Utensili di carpenteria. ABSTRACT: The aim of this paper is to illustrate the traces of the building site found in different zones in the area of the Capitolium of Brescia (Brixia). The area of the Capitoline temple was occupied by a religious building renovated four times from the 2nd century BC to the Flavian period. During the most recent excavations archaeologists have found traces of a workshop for the manufacturing of iron tools, layers of Vicenza stone flakes and two chisels used by the stonecutters dated to the phase between the dismantling of the oldest structures and the rebuilding of the Republican temple. A graffito on the Augustan wall refers to the moulding on the main door of the central cella of the latest temple, together with some engravings that seem to indicate the module used to sketch out the building plan on the ground. KEYWORDS: Building site, Capitolium, Brescia, Brixia, Graffito, Module, Workshop, Chisel, Wood-working tools. RESUMEN: Se presentan los resultados de las recientes investigaciones arqueológicas llevadas a cabo en el área del Capitolium de Brescia (Brixia), ciudad romana de la Regio X Venetia et Istria. Tras las excavaciones de la Soprintendenza Archeologica della Lombardia, realizadas entre los años 2009 y 2014, se han recopilado nuevos datos relativos al área sacra que desde el ii a.C. y época Flavia presenta cuatro diferentes fases constructivas. En esta ocasión, se analizan las evidencias vinculadas *  Il lavoro che qui si presenta deve molto alla disponibilità di alcune persone senza le quali non sarebbe stato possibile ­realizzarlo. A Filli Rossi e a Serena Solano, funzionari della Soprintendenza Archeologica della Lombardia, devo l’autorizzazione allo studio dei risultati delle indagini archeologiche e la fiducia accordatami nell’affidarmene la pubblicazione; Pier Luigi Dander, che ha eseguito lo scavo, mi ha molto gentilmente fornito la documentazione e più volte illustrato il contesto. Sono grato a Francesca Morandini, Responsabile servizio

c­ ollezioni e parco archeologico del Comune di Brescia, per aver sempre garantito la massima disponibilità ad accedere alle aree archeologiche. Molto devo anche all’amico Francesco Franzoni, che con pazienza ha ascoltato a lungo le mie speculazioni. A  Furio Sacchi, Francesca Bonzano e Johannes Lipps sono grato per la lettura delle bozze. Le foto di scavo e dei materiali vengono pubblicate su concessione del Ministero dei Beni e delle attività Culturali e del Turismo – Soprintendenza Archeologia della Lombardia.

276

Antonio Dell’Acqua

Anejos de AEspA LXXVII

con la obra y los materiales de la reforma de la fase republicana, la fase de demolición del santuario augusteo y su reconstrucción en época Flavia. Concretamente, se han documentado una oficina para la producción de metales, los instrumentos empleados y un nivel de lascas de piedra de “médolo” y piedra de Vicenza que pertenecen a la fase de obra del santuario republicano; un grafito sobre enlucido con las molduras de la cornisa de la puerta principal del templo e incisiones para la definición de la malla modular de la reconstrucción del templo de época Flavia. PALABRAS CLAVE: obra, Capitolium, Brescia, Brixia, grafito, módulo, oficina, puntero, Utensilios de carpintería.

L’area su cui sorge il Capitolium di Brescia si estende alle pendici del Colle Cidneo, a nord del foro romano e del decumano massimo, nella zona orientale della città moderna. Sebbene noto dal xvii secolo,1 lo scavo dell’area venne realizzato solo più tardi, tra il 1823 e il 1826.2 Ulteriori indagini hanno interessato l’area a più riprese nel corso degli anni tra il dopo guerra e gli anni Novanta, fino agli interventi del 2009-20113 e dell’ottobre 2014.4 A seguito di questi scavi è oggi possibile ripercorre le quattro fasi che hanno interessato l’area con interventi che ne hanno modificato l’assetto a seconda delle esigenze e delle epoche; è stato inoltre possibile individuare alcune delle fasi di cantiere che hanno accompagnato gli interventi di distruzione e ricostruzione, di cui qui si presentano i primi dati. Essi si riferiscono, in particolare, a due fasi costruttive e riguardano solo alcune attività legate alla cantierizzazione dell’area: tracce di attività artigiana, sbozzatura, utensili, disegni preparatori. L’AREA DEL CAPITOLIUM DI BRESCIA: DAL SANTUARIO DI CULTO INDIGENO AL TEMPIO CAPITOLINO Il primo santuario viene edificato intorno al secondo venticinquennio-metà del ii sec. a.C. e si presentava come una struttura sopraelevata rispetto alla pista di età protostorica che scorreva in direzione est-ovest. Esso consisteva in due ambienti allineati che si affacciavano su un’area scoperta lastricata. Alla terrazza era annesso un edi1  L’erudito locale Ottavio Rossi effettuò il primo scavo attorno ad una colonna rimasta integra che emergeva dal suolo. Rossi 1683: 18. 2   Il locale Ateneo di Scienze e lettere promosse lo scavo per intero dell’edificio, riportando in luce tutto il tempio, di cui si conservano anche limitate porzioni delle murature, le pavimentazioni delle celle e i materiali architettonici relativi all’alzato. I risultati delle indagini vennero pubblicati nel volume Museo Bresciano illustrato 1838. Cfr. Labus et al. 1838. 3   Per i risultati cfr. Rossi 2014. 4   I cui dati sono ancora inediti. Si presentano in questa sede alcuni ritrovamenti.

ficio, ad est, in tecnica mista (alzati lignei, battuti interni in limo argilloso) costituito da un lungo corridoio con tre ambienti distinti separati da tramezzi lignei. Di questa prima sistemazione, che forse monumentalizza un più antico luogo di culto protostorico, sono visibili una porzione del muro di delimitazione della terrazza, alcune lastre in médolo5 che rivestivano la piazza e pochi frammenti di intonaco dipinto.6 Nel secondo quarto del i sec. a.C. le strutture del primo santuario vengono rasate7 e viene costruito un nuovo complesso che prevedeva, sul fondo di una vasta terrazza, quattro tempietti su un unico alto podio,8 e sulla piazza due strutture a tholos interpretate come edicole circolari utilizzate per l’esposizione di signa.9 Già da tempo gli studi hanno evidenziato l’apporto di maestranze di provenienza centro-italica nella realizzazione di questo santuario che con la sua architettura ricca di influssi ellenistici si colloca pienamente nel solco della tradizione tardo repubblicana.10 Le indagini effettuate tra il 2009-2011 hanno consentito di evidenziare, in maniera più ampia rispetto a quanto noto in passato, la fase di ristrutturazione del santuario in età augustea. Documenti epigrafici attestano il legame tra Brescia  5  Si tratta di un calcare variamente marnoso di colore grigio-giallastro o grigio plumbeo, che tende a scheggiarsi per la ricchezza di noduli di selce scura e resti di fossili. È utilizzato più come pietra grezza che come pietra da taglio. Il termine mèdolo, localmente, indica calcari a strati sottili, separati da letti scistosi, così da fornire conci quasi regolari. Cfr. Rodolico 1965: 104.  6   Cfr. Dander 2014a, per i dati di scavo; Sacchi 2014a, per la ricostruzione architettonica di questa fase; Mariani 2014, per gli intonaci.  7   Dander 2014b per i dati di scavo.  8   Sacchi 2014b. L’edificio tardo repubblicano vanta una lunga tradizione di studi per la cui bibliografia si rimanda a Sacchi 2014: 201, note 7-13.  9  Sacchi 2014b: 202-203. Dell’ edificio di culto si conservano in ottime condizioni gli interni delle aule con i pavimenti (da ultimo Morandini et al. 2014 e Morandini 2015 con bibliografia precedente), la decorazione pittorica (Bianchi 2014), articolata su più registi, con in basso un motivo a drappo e nella fascia mediana riquadri decorati da finti marmi, e alcuni elementi architettonici. 10   Cavalieri Manasse 2002.

Anejos de AEspA LXXVII

TRACCE DI CANTIERE DALL’AREA DEL CAPITOLIUM DI BRESCIA...

e Ottaviano/Augusto che è ricordato con la qualifica di pontifex in una epigrafe del 44 a.C.;11 nell’8  a.C. la città divenne colonia civica Augus­ ta.12 Il progetto augusteo dell’area sacra consistette in una ristrutturazione dell’edificio più antico con la stuccatura degli elementi architettonici, i cui finanziatori furono Lucius Hostilius Fronto e Caius Clodius Merga ricordati in un’iscrizione rinvenuta nella cella più occidentale nel 1992.13 Nei recenti scavi14 si è anche appurato che in questa fase vennero inserite due ali porticate, con colonne doriche, che enfatizzarono la prima, la seconda e la terza cella, mentre la quarta ad ovest venne in parte occultata dal porticato; venne modificato l’accesso al pronao, ora possibile solo tramite due scalette, e di fronte al podio vennero disposte sei strutture rettangolari, documentate solo a livello di fondazione, su cui forse trovavano posto statue bronzee ed equestri come nel foro di Pompei riprodotto nell’affresco dai praedia di ­Giulia ­Felice.15 L’ultima fase si riferisce al cosiddetto rifacimento di epoca flavia quando tutto il complesso costituito dal Capitolium, foro e Basilica fu ricostruito e ammodernato. Il rinvenimento di un’ara con dedica a Giove Optimo Massimo, Giunone Regina e Minerva nel dicembre 2011 costituisce elemento probante della funzione del tempio come capitolium urbano.16 Nell’ultima fase, dopo la rasatura delle strutture antiche, il tempio mantenne alcune caratteristiche degli edifici precedenti: posto su un’alta terrazza, fu ingrandito riproponendo un corpo centrale con tre celle separate, una facciata con pronao esastilo, un unico podio, due bracci porticati. Tradizionalmente si attribuisce il rifacimento del Capitolium a Vespasiano che inaugurò il tempio nel 73 d.C.;17 come ricordato dall’iscrizione posta sul fregio del pronao.18 Lo studio del materiale architettonico ha però rilevato che i lavori potrebbe essere stati avviati già negli anni Sessanta del i sec. d.C. e che in epoca Flavia si procedette al completamento dell’edificio e al rifacimento del Foro e della Basilica i cui elemen  I. It. 84; Gregori 2000: 253.  Gregori 2000: 254. Ad Augusto si deve anche l’avvio della costruzione dell’acquedotto che portava l’acqua dalla Val Trompia, poi completato da Tiberio. I. It. 85; Gregori 2000: 254 13   Rossi e Garzetti 1995: 83. 14   Dander 2014c. 15   Sacchi 2014c: 287-288. 16   Gregori 2014: 319-320. 17  Cioè dopo la battaglia di Bedriacum svoltasi quattro anni prima. 18   I. It. 88. Garzetti 1979. 11

12

277

ti architettonici riproducono quelli del Capitolium e sono più tipicamente di produzione flavia.19 È questa l’ultima fase di monumentalizzazione dell’edificio che dopo la fine del i sec. d.C. non conobbe altri interventi, se non alcune limitate operazioni di rifacimento delle pavimentazioni delle celle in età severiana, in seguito ad un incendio,20 nell’ambito di una più vasta ristrutturazione che interessò soprattutto il teatro.21 IL CANTIERE DEL SANTUARIO REPUBBLICANO Le più antiche tracce di cantiere presenti nell’area sono riferibili alla seconda fase del santuario tardo repubblicano (prima metà del i sec. a.C.), rinvenute nella zona meridionale del settore di scavo 422 (fig. 1). In particolare i resti si riferiscono a modeste costruzioni, forse in legno o in argilla, con telaio ligneo e zoccolo in pietre a secco, in cui compaiono blocchi reimpiegati dal precedente edificio23 (fig. 2). Si tratta forse di baracche o laboratori adibiti ad attività complementari al cantiere, tra cui una struttura a pianta quadrata (27,41 × 4,4 m) probabile officina per la fusione e la fucinatura da cui provengono uno scalpello e il becco di un mantice. Lo studio delle scorie metalliche, di numero piuttosto esiguo, induce a ritenere che l’officina, una volta terminato il cantiere, non fu abbandonata, ma smantellata.24 I materiali delle fasi di lavorazione sono in prevalenza chiodi di ferro, uno scalpello, le terminazioni a punta di due grossi strumenti di lavoro dei carpentieri, e pochi resti di piombo, la cui fusione era funzionale alla realizzazione di grappe in uso nell’architettura. Come è stato evidenziato nello studio di questi reperti da parte di Elisa Grassi,25 l’officina di forgia di Bre  Dell’Acqua 2014.   Tracce di restauro sono state notate nella pavimentazione della cella centrale. Cfr. da ultimo Angelelli e Dell’Acqua 2014: 371-373, con bibliografia precedente. 21  Frova et al. 1975: 64-66. Sulla ristrutturazione del teatro cfr. Cavalieri Manasse 1979. Il periodo degli Antonini e dei Severi coincide con la presenza contemporanea e per più generazioni di esponenti bresciani nel Senato di Roma. Cfr. Gregori 2000: 316. 22   Dander 2014a: 191-192. 23   Il precedente edificio, nonché il più antico, risale al ii secolo a.C. Cfr. Dander 2014b: 167-170; Sacchi 2014: 171-178. 24   Secondo E. Grassi “il fatto che non siano conservati in situ nemmeno i residui dell’ultimo processo di lavorazione qui effettuato suggerisce che l’officina non sia stata semplicemente abbandonata nel momento in cui non fu più utilizzata, ma che sia stata volontariamente smantellata al completamento dei lavori di rinnovamento edilizio”. Grassi 2014: 196. 25   Grassi 2014: 195-198. 19 20

278

Antonio Dell’Acqua

Anejos de AEspA LXXVII

Fig. 1. Pianta dell’area con indicazione dei settori di scavo. (Archivio Soprintendenza Archeologia della Lombardia, autore P. L. Dander).

scia trova confronti con altri santuari, sia greci sia romani, dove la presenza di attività legate alla lavorazione di metalli poteva essere connessa o alla produzione di oggetti votivi26 oppure di strumenti ed elementi per le architetture durante la fase di cantiere. Un esempio molto antico di questa tipologia di officina, operante durante la costruzione dell’edificio religioso, è l’impianto metallurgico individuato nell’area del tempio di Apollo Daphneforos ad Eretria, dove, tuttavia, esso rimase in uso anche dopo il completamento dell’edificio.27 Ulteriori esempi analoghi al rinvenimento bresciano sono il santuario magno greco di Marasà (Locri Epizefiri)28 e quello gallo-romano dedicato a Mars Mullo di Allones nel nord ovest della Bre­tagna.29 26   Cfr. Grassi 2014 e i molti esempi citati, con relativa bibliografia, a nota 6, p. 197. 27   Huber 1997. 28   L’officina fu impiantata contemporaneamente alle fasi iniziali del cantiere quando vennero lavorati sul posto i blocchi di calcare bianco impiegato per la costruzione del tempio ionico. Rubinich 2010: 390-391; sull’architettura del tempio cfr. Costabile et al. 2006. 29  Durante la fase di cantiere furono approntate una officina per la lavorazione dei metalli e accanto una per la lavorazione dei materiali lapidei, diversi a seconda della

A questa stessa fase di cantiere è riconducibile quanto rinvenuto nell’ottobre 2014 nel settore 1 (fig. 1): nell’area antistante l’estremità ovest del podio è stato messo in luce un cumulo di blocchetti di médolo sbozzati a colpi di martellina in forma tronco-piramidale, funzionali alla composizione del paramento del podio. Il procedimento di preparazione degli elementi sembra partire da grezzi lastroni calcarei, con almeno una o due facce regolari, ricavati dalla collina retrostante. Ammassi di piccole scaglie visibili più a sud sembrano far pensare allo svolgimento in situ di questa attività (fig. 3).30 Insieme a queste testimonianze, è stata diffusamente evidenziata la presenza di uno strato di scaglie minute di pietra di Vicenza, con spessore decisamente più elevato ad est, da destinazione d’uso (calcari duri per le fondazioni, materiale tufaceo per l’ornato architettonico scolpito). BrouquierReddé e Gruel 2004: 310-312, fig. 27; Loiseau 2012. 30   Il rinvenimento ricorda per analogia quanto documentato archeologicamente nel complesso horreario a Testaccio (età neroniana-flavia): una postazione di lavoro, nei pressi di un muro, dotata di una soglia in travertino utilizzata come sedile con blocchi di tufo di medie dimensioni a destra e un cumulo di pietre piramidali a sinistra. Cfr. Serlorenzi 2010: 116.

Anejos de AEspA LXXVII

TRACCE DI CANTIERE DALL’AREA DEL CAPITOLIUM DI BRESCIA...

279

Fig. 2. Resti della struttura interpretata come fucina, vista da Nord (foto Archivio Soprintendenza Archeologia della Lombardia).

Fig. 3. Settore 1, scavo 2014, strati di scaglie di pietra di Vicenza (foto Archivio Soprintendenza Archeologia della Lombardia).

cui provengono due scalpelli in ferro (fig. 4, a-b), strumenti impiegati nelle operazioni di finitura degli elementi architettonici in calcarenite, effettuata probabilmente in corrispondenza dei depositi stessi. Si tratta di un calcare fossilifero estratto nella parte nord-orientale dei monti Berici, caratterizzato da una elevata porosità e ridotta resistenza alla compressione, facilmente lavorabile, ma poco durevole.31 La pietra fu utilizzata in larga quantità tra la fine del i sec. a.C. e gli inizi del i d.C. ad esempio a Padova,32 Este,33 Vicenza,34 Verona35 e Milano.36 Le esportazioni non mostrano di avere raggiunto le zone dell’odierno Piemonte e numericamente limitate sono le attestazioni nell’VIII Regio, tranne che a Reggio Emilia, dove si conosce un certo numero di attestazioni.37 A Brescia questo litotipo è impiegato, oltre che nel santuario repubblicano per la realizzazione dell’ornato architettonico,38 anche per la realizzazione di elementi statuari e monumenti funerari,39

fino alla sua sostituzione intorno all’età augustea quando sembra prendere avvio lo sfruttamento delle locali cave di Botticino distanti pochi chilometri dal centro abitato.40 L’insieme dei dati amplia il quadro delle conoscenze relative al santuario repubblicano e quelle inerenti il processo di lavorazione dei materiali destinati all’architettura in età repubblicana. Già in passato era stato rilevato il contributo di maestranze alloctone sia per la realizzazione delle strutture murarie sia per la decorazione pittorica, molto probabilmente chiamate dall’Italia centra­ le;41 inoltre era stato messo in evidenza come i materiali lapidei provenienti dall’area veneta costituissero la principale materia prima per l’edilizia non solo a Brescia ma anche a Milano e nelle regioni limitrofe.42 Il ritrovamento delle scaglie di lavorazione e i relativi strumenti utilizzati sembra indicare che il materiale raggiungesse il cantiere edilizio quanto meno a livello di semilavorato: ciò presuppone o lo spostamento di maestranze dall’area veneta oppure, data la qualità scultorea leggermente

31   In generale sul litotipo e le sue varietà cfr. Cornale 1994; sulle caratteristiche di lavorabilità Alberti 1994. 32   De Vecchi e Lazzarini 1994. 33   Fogolari 1975: 179. 34   Buchi 1987. 35   Bianco 2008: 209-211. 36   Bugini e Folli 2000; Bugini e Folli 2012; Sacchi 2012: 48, cat. nn. 1-4, 6, 19-22, 24-25, 51-55, 57-58, 62. 37   Cavalieri Manasse 2006: 129. 38   La decorazione architettonica del santuario repubblicano di Brescia, considerata la sua unicità nel panorama dell’Italia settentrionale, ha avuto una lunga tradizione di studi; ultimo in ordine di tempo quello di Giuliana Cavalieri Manasse che ha integrato i precedenti lavori con i materiali emersi negli scavi del 1992-1997. Cavalieri Manasse 2002. 39  Mi limito a ricordare un piccolo capitello di lesena ionico italico, pertinente ad un monumento funerario della metà del I sec. a.C. (Cavalieri Manasse 1997: 249-250, fig. 6), e una statua di togato seduto la cui cronologia oscilla tra la fine del i sec. a.C. e la prima metà del i sec. d.C. (Cavalieri Manasse 1997: p. 250).

  Zezza 1982: 26-27; Bugini e Folli 2014: 185-186.   L’ipotesi era stata formulata da Maria Pia Rossignani secondo cui i committenti si sarebbero rivolti alle più antiche colonie cisalpine per procurarsi progettisti e maestranze abili (Rossignani 1986: 235-236). Già Frova aveva distinto tra il materiale architettonico, che si inserisce “agevolmente nel panorama dell’ornamentazione coeva della Cisalpina”, e le pitture la cui complessità e la perizia tecnica “suggeriscono più che un’influenza centro italica, ... l’ipotesi di artigiani venuti dal Lazio o dalla Campania” (Frova 1979: 217). In merito alla realizzazione di murature in opera quasi reticolata, M. Torelli ha evidenziato che fuori dall’area di Roma, e della Regio I, si tratta di un fenomeno eccezionale che presuppone l’intervento di squadre specializzate provenienti da quell’ambito geografico (Torelli 1980: 152-153). 42   Cavalieri Manasse 2006: 129 con bibliografia precedente; Cavalieri Manasse 2013: 103. 40

41

280

Antonio Dell’Acqua

inferiore,43 la presenza di manodopera locale affiancata da un artigiano proveniente da un centro di antica tradizione culturale44 e quindi l’esistenza in loco di botteghe lapicide in grado, già in questa fase, di lavorare e scolpire un litotipo diverso dalla locale pietra di Botticino.45 Per quanto riguarda gli elementi in ferro rinvenuti, in totale sono quattro gli scalpelli trovati, due nel 2014 nel settore 1, altri due nell’officina durante gli scavi del 2009. I primi due46 sono di dimensioni e consistenza maggiori con una lunghezza di 17,5 cm uno e 16,5 l’altro (fig. 4, a-b); nella fucina, invece, i due scalpelli misurano rispettivamente 15,5 cm e 11 cm (fig. 4, c-d), ma tutti e quattro sono caratterizzati dall’avere la punta piatta. Appare evidente che essi facevano parte di un set di attrezzi da lavoro degli scalpellini e che le diverse dimensioni, comunque piuttosto contenute, fossero legate alle esigenze di scolpire elementi architettonici, anche in zone di maggior dettaglio, in una fase di rifinitura piuttosto che di lavorazione di grandi blocchi che avrebbe invece richiesto utensili di maggiori dimensioni e resistenza. A parte gli esemplari musealizzati, come quelli a Londra o a Chesterford e Worlington47 o gli esemplari conservati al Museo del marmo di Carrara, proviene da scavo una porzione di scalpello trovato nel 2014 nel corso di indagini archeologiche condotte dall’Israel Antiquities Authority nella porzione meridionale del Western Wall a Gerusalemme,48 in un contesto di cantiere duran43  Analogo è il fenomeno che è stato riscontrato nella romana Autun dove i lapicidi locali erano soliti lavorare i calcari, mentre dopo l’introduzione del marmo, la qualità dei prodotti risentì dell’imperizia e della mancanza di pratica. Cfr. Brunet-Gaston 2008: 493-494. 44  Forse di provenienza cisalpina o centroitaliaca. Cfr. Cavalieri Manasse 2002: 103. La presenza di manodopera mista, straniera e indigena, è attestata nella bottega che realizza il monumento dei Giuli a Glanum e altri sepolcri lungo la valle del Rodano (Kleiner 1977: 680-681) ed ipotizzata a Roma dove le prime costruzioni templari marmoree avrebbero impegnato botteghe formate da artigiani greci e locali (Gros 1976: 65). 45   Si tratta di un calcare a grana fine del Lias inferiore, resistente e facile da lavorare, molto apprezzato per la somiglianza con il marmo lunense. Le cave si collocano a circa una decina di chilometri a nordest di Brescia, nella località di Botticino, da cui deriva il nome; da qui il trasporto poteva avvenire sia tramite la strada che collegava Brixia a Verona, sia per via fluviale, data la presenza di corsi d’acqua e navigli nella zona orientale della città. Rodolico 1965: 106. 46   Rinvenuti nel settore 4, US 324, RR 62. 47   Strong and Brown 1976: 160-161, fig. 269. 48  La notizia è stata riportata sui giornali israeliani (Haaretz, Times of Israel), oltre che sul sito internet della Biblical Archaeology Society: http://www.biblicalarchaeology.

Anejos de AEspA LXXVII

te la fase di realizzazione del tempio voluto da Erode il Grande sulla collina di Moriah;49 l’anno precedente nella zona meridionale della stessa città a Qiryat Ha-Yovel è stata rinvenuta una cava di calcare locale e tra i materiali quattro porzioni di scalpelli.50 Si possono poi ricordare altri due particolari ritrovamenti. Il primo è il relitto di Porto Novo, in Corsica, che trasportava un carico di marmo di Carrara; tra i materiali, sono stati rinvenuti gli attrezzi degli scalpellini – alcuni personali con le iniziali dei nomi incise – tra i quali un set di scalpelli di varia forma e dimensioni.51 Materiali di cantiere sono stati rinvenuti anche durante lo scavo del ponte a Chalon-sur-Saône, costruito nel iii sec. d.C. sulla Loira, tra cui fili a piombo, asce, grappe metalliche e scalpelli.52 Nell’ambito dei ritrovamenti del 2009, sempre nel settore 4 dello scavo (US 250), si segnala un ulteriore oggetto (fig. 5), in pietra e di forma ovale (RR 61), dotato di un piccolo gancio, di peso pari a 562 gr.53 Non è chiara la funzione dell’oggetto, ma al momento si possono avanzare due ipotesi. Seppure ovale, l’elemento sarebbe potuto servire per tendere un filo in una soluzione molto rudimentale di “filo a piombo”, anche se di norma esso è dotato di un peso di forma conica con la punta verso il basso; con la stessa funzione si potrebbe attribuirlo ad un archipendolo, strumento utilizzato per determinare una direzione orizzontale o verificare l’orizzontalità di una retta. Come ulteriore ipotesi si potrebbe ritenere che, considerate le dimensioni molto ristrette dell’anello, l’oggetto potesse essere un contrappeso per una stadera. La tradizione di pesi di pietra è ampiamente attestata in ambito etrusco e numerosi sono gli esemplari noti,54 ma anche in età romana essi erano impiegati come contrappeso nelle bilance, anche se ben più famosi sono quelli in piombo figurati.55 Essi potevano essere realizzati in pietre calcaree, in marmo – come quelli di Luni – oppure in una varietà specifica detta lapis aequipondus,56 per il tipo d’uso al quale era generalmente destinata, ovvero una varietà di serpentina peridotitica a struttura compatta e di una org/daily/biblical-sites-places/temple-at-jerusalem/ancientchisel-unearthed-at-western-wall/ 49   Bahat 2011: 46-48; Netzer 2006: 220-221. 50   Oz 2014. 51  Bernard et al. 1998: 61-62, fig. 6. 52   Bonnamour 2000: 293-302. 53   Misure: h 10 cm, diametro ca 7,5 cm. 54   Cattani 2001. 55   In generale cfr. Corti 2001. 56   Gnoli 1988: 181.

Anejos de AEspA LXXVII

TRACCE DI CANTIERE DALL’AREA DEL CAPITOLIUM DI BRESCIA...

281

Fig. 4. Strumenti da lavoro rinvenuti nell’area, scalpelli e un gancio (foto autore).

certa durezza di colore verde scuro, nerastro, che consentiva di realizzare pesi perfettamente lisciati, generalmente con un polo piatto, come gli esemplari presso il Museo comunale di Gubbio.57 È evidente, comunque, che l’eccentricità della forma potrebbe giustificarsi con il ritrovamento in un contesto di cantiere, dove forse la necessità di verificare la regolarità dei piani o misurare quantità limitate di prodotti indusse qualcuna delle maestranze ad approntare alla bell’e meglio un strumento che garantisse la verticalità della corda o indicasse il peso delle materie prime. Infine, si segnale anche il rinvenimento, nel settore 4, di un gancio in ferro forse riconducibile ad una bilancia (fig. 4, e).58 I dati nella loro complessità sembrano indicare un procedimento di allestimento del cantiere in linea con la tradizione costruttiva dell’epoca e, trattandosi di un edificio intra-urbano, esso coincise con l’area stessa della fabbrica.59 L’applicazione dell’indagine stratigrafica ha agevolato l’individuazione di strati con caratteristiche analoghe a quelle delle UUSS rinvenute nel settore 1 dello scavo costituite da scaglie litiche, anche se non in tutti i contesti è possibile con facilità risalire alle motivazione che ne hanno causato la formazione. La pratica di riutilizzare le scaglie di lavorazione del materiale architettonico di un edificio in strati   Ciani 1995: 405-408.  Le dimensioni sono tali da rendere improbabile un’associazione col peso in pietra, dotato di un anello troppo piccolo rispetto al gancio. 59   Cfr. Giuliani 2006: 250.

Fig. 5. Peso in pietra rinvenuto nel settore 4 (foto Archivio Soprintendenza Archeologia della Lombardia).

preparatori, allo scopo di economizzare l’uso di materie prime, è attestato nel santuario di Locri Epizefiri;60 in ambito romano, invece, si può citare, ad esempio, il tempio di Vesta a Roma dove attorno al podio è stato rinvenuto uno strato di 4 cm di spessore composto da scaglie di marmo lunense per cui sono state avanzate ipotesi legate alla fase di cantiere: o materiale di scarto della struttura più antica, rinterrato in situ in chiave rituale, oppure i resti della lavorazione degli elementi architettonici realizzato sul posto.61 Sempre a Roma è noto il contesto del Mausoleo di Augusto costruito su un’area bonificata dove la depressione del terreno venne risolta utilizzando malta mista a scaglie di marmo e travertino, “interpretabili come residui di lavorazione dei blocchi del paramento esterno del tamburo”.62 Più o meno contemporaneo è il Tempio C di Grumentum, relativo alla fase tardo-repubblicana/giulio-claudia, nel corso del cui scavo è stato individuato uno strato di scaglie di lavorazione dei cubilia, miste a frammenti di laterizi, formatosi durante le attività edilizie relative alla costruzione del tempio.63 Infine, si possono segnalare gli spessi strati di scaglie marmoree rinvenuti dalla missione italiana nel corso degli scavi presso la stoà-basilica di Hiera-

57

58

 Costabile et al. 2006: 25, fig. 24.   Caprioli 2007: 101. 62  Agnoli et al. 2014: 219. 63   Candelato e Perretti 2009: 65. 60 61

282

Antonio Dell’Acqua

polis di Frigia, scaglie risultanti dalla lavorazione in situ degli elementi architettonici.64 IL CANTIERE DI ETÀ FLAVIA L’analisi del complesso del Capitolium di Brescia ha permesso di seguire il processo costruttivo nella fase di ristrutturazione a partire dalla seconda metà del i sec. d.C., dalla cantierizzazione dell’area fino all’elevazione delle strutture. In questa fase di ristrutturazione si decise di ampliare le dimensioni del santuario, dovendo tuttavia tener conto di molti vincoli: il colle a nord, le strutture abitative e termali preesistenti ad ovest,65 l’ingombro del teatro ad est66 e lo scorrere del decumano a sud.67 Il livello della piazza venne alzato di 2,5 m, interrando i resti degli edifici più antichi; non potendo estendersi verso sud, il progettista optò per l’arretramento delle strutture verso est di 12,5 m e verso nord di 4 m andando a sbancare una parte del colle Cidneo. La superficie complessivamente occupata è pari a 4020 mq, con un ingombro in larghezza di 60 m e in profondità di 67 m. Anche in questa fase è attestata la presenza di un’officina nell’area ipogea del santuario dell’aula dei Pilastrini,68 ma a differenza di quella precedente, essa era dedita alla realizzazione di oggetti per la riparazione degli elementi decorativi dell’edificio templare e di amuleti a cui si riferisce una piccola lastra trovata nel 2001;69 inoltre, il ritrovamento di uno scarico di materiali,70 in un condotto nella zona ovest, in cui sono stati rinvenuti materiali riferibili alla media età imperiale, lascia intuire che un’officina rimase sempre attiva nell’area santuariale, anche se non è al momento possibile localizzarla.71 Non si doveva trattare di una 64   Gli scavi si riferiscono alle missioni 2009-2010 a cui ha partecipato anche l’Università Cattolica di Milano. I dati, tuttora inediti, mi sono stati forniti dal Prof. Furio Sacchi e dalla Dott. Elisa Grassi. 65   Rossi e Garzetti 1995: 80; Rossi 2005: 19; Morandini 2012: 86-87. 66  L’edificazione del teatro risale ad età augustea e fu oggetto di almeno altri due interventi fino all’epoca severiana. Cfr. da ultimo Basso 2013: 70, con bibliografia precedente. 67   Antichi edifici sul foro: 15-16. 68   Miazzo 2002: 453-456. 69   Gagetti 2014. 70  Si tratta di un accumulo di materiale successivo alla chiusura del tempio, costituito da frammenti ceramici, anforari, vitrei, spilloni in osso, materiali bronzei e metallici (pesi, chiavi e ornamenti). Rossi e Miazzo 2001: 430. 71   Rossi e Miazzo 2002. L’officina nell’aula dei pilastrini fu poi smantellata poiché l’ambiente venne ristrutturato e destinato ad altre funzioni in età antonina-severiana. Cfr. Rossi 2007: 205-214.

Anejos de AEspA LXXVII

bottega specializzata o dedita alla fusione di grandi bronzi, quanto piuttosto di una sorta di laboratorio dedito alla manutenzione del santuario e delle suppellettili.72 Nel corso degli scavi tra il 2009 e il 2011 nel settore 2 sottostante Casa Pallaveri (fig. 1) è stato rinvenuto un graffito (figg. 6-7) realizzato sull’intonaco del muro di fondo della fase augustea. L’individuazione nella zona antistante di due buche per palo allineate nord-sud, quasi a ridosso del limite orientale del portico, ha fatto ritenere agli scavatori che, durante la fase di cantiere di età flavia, in questa zona fosse stata predisposta una struttura, appoggiata su pali, per riportare, e forse anche verificare con fili, le misure e la regolarità delle modanature da riprodurre sui blocchi semilavorati.73 Simile al rinvenimento bresciano dovrebbe essere un graffito su un arco dal Palatino citato da Wilson Jones come inedito.74 Il graffito, in scala 1:1 (fig. 7), si compone di una fascia inferiore alta 29,6 cm e una superiore di altezza crescente dai 13,5 ai 14 cm, scandita in rettangoli di dimensioni variabili (fig. 8); sulla destra, infine, si osservano due profili di cornici identiche (fig. 9) poste a 30 cm tra loro. L’altezza massima tra la linea superiore e quella corrispondente al piano di calpestio antico (relativo alla fase tardo-repubblicana e augustea) è di 89 cm, mentre è di 47,5 rispetto alla riga inferiore. Nel 2014 avevo già avuto modo di studiare e pubblicare il graffito75 avanzando alcune considerazioni. All’epoca era visibile una porzione di soli 1,347 m, e si era ipotizzato che si trattasse di un passus, pari 1,479 m, e che il graffito continuasse, come poi il restauro ha effettivamente messo in evidenza. Dopo il restauro e la pulitura della superficie si è notato che il graffito si estende per un massimo di 4,72 m in lunghezza e 0,89 m in altezza, ma va tenuto conto che in origine continuava sul lato sinistro e forse anche in altezza,76 mentre non ci sono altri segni sul lato destro. Va anche 72   Indirizzano verso questa interpretazione il ritrovamento di frammenti di statue, un dito, resti di capigliatura, chiodi, elementi in bronzo dorato, tasselli o porzioni di lamina di bronzo, scolature di fusione, una spatolina, uno scalpellino e un martello, due crogioli. Rossi e Miazzo 2002: 435. 73   Dander 2014c: 315. Non sono noti nello specifico i dati di queste buche (distanza dal muro e distanza tra loro) che potrebbero essere da riferire anche ad impalcature. 74  Anche questo graffito sarebbe inciso su una parete stuccata. Wilson Jones 2000: cap. III, nota 34. 75   Dell’Acqua 2014: 341-343. 76   In alcuni punti si notano tratti dell’incisione proseguire verso l’alto per alcuni centimetri oltre la riga orizzontale superiore.

Anejos de AEspA LXXVII

TRACCE DI CANTIERE DALL’AREA DEL CAPITOLIUM DI BRESCIA...

283

Fig. 6. Graffito rinvenuto sul muro di fondo del portico occidentale della fase augustea. Il tracciato nero evidenzia le linee altrimenti non visibili perché solo lievemente incise sull’intonaco della parete (foto ed elaborazione autore).

Fig. 7. Rilievo del graffito (elaborazione autore).

rilevata la scarsa precisione nel tracciamento delle linee, soprattutto in quella superiore, con uno scarto di 0,5 cm tra l’inizio e la fine.77 Tenuto conto della parzialità di quanto conservato e dell’approssimazione con cui sono tracciate le linee, risulta che dalla risega del muro augusteo e fino alla riga superiore la superficie si può teoricamente dividere in 3 p.r. e che la prima riga indichi la mezzeria.78 La fascia con i rettan77   Questa mancanza di precisione si nota in molti punti: oltre al fatto che la fascia superiore non ha un’altezza costante, un’analisi ravvicinata consente di verificare che alcune linee si sovrappongono, altre non sono ortogonali e i tratti che compongono le modanature delle due cornici sono molto imprecisi. 78  Non si tratta di una linea di mezzeria esatta perché dovrebbe essere a 45,4 cm dal muro sottostante, invece eccede di 2 cm.

goli, per quanto di andamento crescente verso sinistra, sembrerebbe corrispondere in altezza a circa un semipes, mentre la fascia sottostante, che presenta una ulteriore riga verticale a circa 1  m dal profilo più interno, equivale ad 1 p.r. La linea inferiore individuerebbe, dunque, due metà pari ciascuna ad 1 ½ p.r., mentre quella in verticale non ha un significato al momento chiaro.79 La fascia superiore comprende 42 rettangoli di dimensioni variabili. In passato la si era interpretata come una sorta di riga graduata, o mensa mensuraria, di cui si conoscono altri esemplari in

79   Potrebbe costituire di per sé un altro asse di simmetria, così da ottenere una sorta di piano cartesiano, ma non essendo nota l’estensione a sinistra risulta difficile determinare i rapporti tra le due metà individuate dal segmento.

284

Antonio Dell’Acqua

Anejos de AEspA LXXVII

Misure dei rettangoli (da dx) in cm

Misure dei rettangoli in piedi romani

21,5

0,7

13

0,43

2

0,06

16

0,54

prende appieno come mai il valore dei singoli rettangoli muti costantemente, anziché corrispon­ dere a misure standard; inoltre alcuni valori ritornano più di una volta, come 21,581 cm che si ripete quattro volte e risulta sempre dalla somma di tre rettangoli, oppure 16/16,5 cm che compare tre volte ma come rettangolo singolo. Procedendo verso sinistra, e superata la linea verticale che interseca la fascia inferiore, i rettangoli interi si dilatano e si incontrano due volte il valore 18 cm, una volta 19 cm e una 20 cm,82 mentre le somme dei rettangoli a gruppi di tre una volta danno 21,5 cm, una volta 21 cm e una volta 19. Si nota, poi, una relazione tra i rettangoli e i profili delle cornici che corrispondono a quella collocata sulla porta della cella centrale. Essi sarebbe potuto servire come modello di riferimento fornito alle maestranze per effettuare le misurazione e gli eventuali controlli: una parte degli ­studiosi ritiene che queste raffigurazione “rappresentano veri e propri progetti di particolari architettonici, in scala al vero; per altri, sarebbero delle rappresentazioni oggettive di parti realizzate; per altri ancora dei modelli grafici, in scala, utilizzati per il controllo delle fasi di realizzazione”.83 Nelle facciate delle tombe 826 e 633 di Petra analizzate da May Shaer sono state individuate incisioni degli elementi decorativi dell’esterno quali cornici “in their approximate proportions”,84 successivamente realizzate secondo le competenze proprie di ciascun scalpellino, come pure sono state trovate incisioni delle cornici del tempio in quello di Bacco a Baalbek.85 La relazione tra rettangoli e profilo sembra derivare dal fatto che il semipes è anche il modulo di base per la cornice (fig. 10): la mensola è lunga 2M e alta 1M;86 il primo listello, la gola rovescia e il secondo listello sono compresi in 1M e tra loro sono in rapporto 1:3;87 la gola convessa – corri-

3,5

0,11

5,5

0,18

6,3

0,21

8,5

0,28

16,5

0,55

14,5

0,48

8

0,27

sextans ?

semuncia

semipes

6,5

0,25

16,5

0,55

8

0,27

17,5

0,6

7,5

0,25

palmus

7,5

0,5

palmus

7,5

0,25

palmus

18

0,6

14,5

0,48

semipes

7

0,23

palmus (?)

18

0,6

6,5

0,2

8,5

0,28

6,5

0,21

19

0,64

5,5

0,18

9

0,3

6

0,2

19,5

0,65

bes

5

0,16

sextans (?)

9,5

0,3

triens

15

0,5

25,5

0,86

5

0,16

3,5

0,11

5

0,16

20

0,67

4,5 10 4,5 15

0,15

sextans (?)

sextans (?)

0,33 0,15

sextans (?)

0,5

semipies

Fig. 8. Tabella con le dimensioni dei rettangoli del graffito (v. fig. 7).

materiali diversi, raffigurate su rilievi funerari o in formato epigrafico.80 Se così fosse, non si com  Adam 1984: 42-44; Ruiz de la Rosa 1987: 111-112.

80

  Il valore è prossimo ad un dodrans che è pari a ¾ di p.r.   Valore prossimo al bes che corrisponde a 2/3 di p.r. 83   Inglese 1999: 44; si veda anche Ruiz de la Rosa 1987 che affronta lo studio geometrico-proporzionale e quello delle incisioni degli edifici dall’antichità al medioevo. 84   Schaer, M. 2003: The Decorative Architectural Surfaces of Petra, tesi di dottorato discussa presso la Technische Universität di Monaco, 62, fig. 7. 85  Kalayan 1971: 269-274. 86   In assenza di un rilievo diretto della cornice della porta centrale, così come dei restanti frammenti ricollocati nella ricostruzione ottocentesca, si fa riferimento ai disegni delle tavole nel volume Museo bresciano illustrato. Nello specifico, per la cornice della porta maggiore vd. Tav. IX, fig. II, da cui risulta che la mensola ha una larghezza minimo di 15,6 cm con andamento incostante. 87   Sommando le dimensioni reali delle modanature (primo listello cm 2,5, gola rovescia cm 7,5, secondo listello cm 2,5) risulterà uno scarto di ca 2,5 cm rispetto al modulo, ma si 81

sextans (?)

82

Anejos de AEspA LXXVII

TRACCE DI CANTIERE DALL’AREA DEL CAPITOLIUM DI BRESCIA...

285

Fig. 9. Foto del graffito relativo alla sottocornice della porta centrale del tempio. Le incisioni sono ricalcate in nero per una maggiore facilità di lettura (foto autore).

Fig. 10. L’immagine mostra la relazione che sembra intercorrere tra la fascia dei rettangoli e i profili ipotizzando l’utilizzo di un modulo pari a un semipes (elaborazione autore).

spondente nella realtà al kyma ionico – il terzo listello e la gola concava sono pure inscrivibili in 1M. Alla luce di questa relazione, sembra ipotizzabile che l’alternanza non propriamente regolare rammenda l’irregolarità in questa porzione del graffito e il fatto che la fascia di rettangoli tenda a diventare più alta procedendo verso sinistra. L’incongruenza tra i dati potrebbe derivare dall’approssimazione con cui il disegno teorico è stato riportato sulla parete.

di rettangoli in prevalenza larghi 21,5 cm e 16/16,5 cm indichi la successione delle mensole e dei cassettoni della cornice: sulla base dei rilievi ottocenteschi, infatti, la mensola risulta larga minimo 15,6 cm, ma la dimensione non è costante, mentre la distanza tra due elementi risulta essere, se calcolata sulla corona, di 22 cm (corrispondente ad un dodrans ovvero ¾ di p.r.). Si potrebbe dunque avanzare l’ipotesi secondo cui questa fascia, la cui

286

Antonio Dell’Acqua

lunghezza oggi non è definibile completamente, avrebbe potuto costituire il tracciato di riferimento per il dimensionamento della cornice della porta centrale che in origine raggiungeva circa 6,7 m di lunghezza totale; inoltre, all’interno di questo schema geometrico molto sommario si hanno anche le indicazioni minime per calcolare le dimensioni degli elementi che vanno a decorare le gole inferiori per i quali sembra impiegato un modulo più piccolo pari ad un palmo (7,4 cm), contenuto tre volte in un dodrans (fig. 11). Applicando questo teorema alla cornice si nota come in corrispondenza delle mensole si collochino tre baccellature semilunate, mentre nello spazio compreso tra due mensole se ne possono collocare tre coppie; la larghezza degli archetti trilobati nel kyma lesbio corrisponde a due palmi, mentre gli ovuli del kyma ionico ad uno, così come la distanza tra due sgusci di due ovuli separati da una punta di freccia (fig. 12). Da un punto di vista filologico, il graffito può definirsi Παραδειγμα, termine usato dagli scrittori antichi quali Erodoto,88 Polibio89 e Plutarco,90 per indicare un campione o un prototipo a scala reale.91 La ricerca ha negli ultimi anni accresciuto la campionatura di questo tipo di documentazione relativa alle fasi di cantieri.92 Casi celebri sono: l’incisione sulla pavimentazione marmorea del Tempio ionico di Pergamo in cui è raffigurato il fusto completo con lo studio della rastremazione e l’asse di simmetria,93 o il disegno dell’architrave della cella che Schwandner ha individuato sulla parete occidentale dello stesso edificio.94 In epoca romana la prassi di realizzare sull’edificio stesso i disegni preparatori delle membrature architettoniche, in scala reale o ridotta, accomuna cantieri sparsi in diverse zone dell’impero, trattandosi, a prescindere dalle tradizioni costruttive indigene, di un modus operandi pratico e comunemente noto. Ulteriori esempi di tracciati di cantiere sono stati individuati nei templi di Zeus a Gerasa95 e di Baalbek;96 su un piccolo tempio tetrastilo di

 Erodoto, Historia, 2.82.2.  Polibio, Historiae, 1.59.8 90  Plutarco, Moralia. 91   Ruiz de la Rosa 1987: 120-123; Inglese 2012. 92   Sui tracciati di cantiere Inglese e Pizzo 2014. 93   Bohn 1896: Taf. XXXVI. 94   Schwandner 1990: 95-97, Abb. 12-16. 95   Si tratta della cornice del tempio. Cfr. Kalayan 1988: 30. 96  Sul lastricato del tempio è incisa la metà esatta della cornice. Cfr. Kalayan 1971: 269-274; Mattern 2001: 86-87, Abb. 30. 88 89

Anejos de AEspA LXXVII

Bziza;97 in quelli di Apollo a Didyma98 e di Athena Polias a Priene;99 nella platea dell’Anfiteatro di S. Maria Capua Vetere;100 sulle colonne del tempio di Adriano101 e in quello di Vespasiano a Roma;102 sul mausoleo di Augusto;103 sul muro della parodos del teatro di Terracina;104 nel foro romano (in una fase relativa ai rifacimenti tetrarchici);105 nel Mausoleo di Adriano, in seguito ad un cambiamento nel progetto;106 nelle terme di Pollena Trocchia come segnalato in un recente studio.107 Un’ulteriore acquisizione degli scavi effettuati nell’autunno 2014 è una coppia di lastre (fig. 13) in pietra di Botticino che rivestono la pavimentazione del podio di età repubblicana, entrambe conservate in situ, poste in corrispondenza della scaletta che consentiva l’accesso al pronao del tempio in questa fase.108 I tracciati (fig. 14a), non facilmente leggibili a causa delle condizioni ambientali in cui si trovano, consistono in una serie di incisioni ortogonali tra loro, distribuite a distanza regolare così che sembra potersi ricostruire una griglia composta da quadrati di lato 16  × 16 cm (ca 0,54 p.r.). Si osservano, inoltre, cinque archi di circonferenze di dimensioni variabili:109 il  minore corrisponde ad una circonferenza di raggio 1,12 m inscrivibile in un quadrato di lati 2,2  ×  2,2 m (7,5  ×  7,5 p.r.), quadrato che a sua volta risulta composto da 14 moduli di lato 16 × 16 cm (fig. 14b). Per quanto riguarda le altre circonferenze, non si è in grado al momento di proporre valide interpretazioni circa il loro signi97   Sono presenti due incisioni: un dettaglio della cornice con sima e fascia a dentelli e un’incisione più lunga di m 8. Cfr. Mattern 2001: 87, Abb. 31-32. 98   Haselberger 1983: 111-119, Abb. 2. Sul muro occidentale dell’adyton si trova l’incisione della trabeazione e di una porzione dello spiovente in scala 1:1. 99   È raffigurato l’intero timpano. Cfr. Wilson Jones 2000: fig. 3.12; Mattern 2001: 89-90, Abb. 36. 100  Dove si trova il disegno dell’arcata del primo ordine della facciata. Cfr. De Franciscis 1959: 399-402, fig. 1. 101   Claridge 1982: 27-30. Per un caso analogo da Ostia cfr. Kockel 2014: 233-239, Abb. 9. 102   Rockwell 1989: 90. 103   Sono due incisioni raffiguranti una cornice maggiore e una più piccola. Cfr. Mattern 2001: 89, Abb. 34, con bibliografia precedente. 104   Mattern 2001: 89, Abb. 35, con bibliografia precedente. 105   Giuliani e Verduchi 1987: 154-155. 106   Cfr. Vitti in questo volume. 107   I dati sono stati presentati in un poster da Josef Souček in occasione di questo convegno. 108  Dander 2014b: 193-194. In seguito all’allestimento dell’area museale della IV cella, le due lastre non sono più visibili. 109   Le circonferenze misurano: 2,7 m; 2,6 m; 2,30 m; 2,25 m; 1,12 m.

Anejos de AEspA LXXVII

TRACCE DI CANTIERE DALL’AREA DEL CAPITOLIUM DI BRESCIA...

287

Fig. 11. Ipotesi della relazione che intercorre tra il graffito e la cornice reale nell’ipotesi in cui i rettangoli indichino la distanza tra le mensole (elaborazione autore).

Fig. 12. Indicazione del modulo usato per gli elementi decorativi delle gole della sottocornice: 1 palmo = 7,4 cm (elaborazione autore).

Fig. 13. Lastre con incisioni ortogonali e cinque circonferenze (foto Archivio Soprintendenza Archeologia della Lombardia).

ficato: esse potrebbero riferirsi ad altre proporzioni che fino ad ora non sono state riconosciute, oppure si potrebbe ritenere che essere furono rea-

lizzate nella fase preparatoria del cantiere, forse come tentativi progettuali, poi risultati essere troppo sproporzionati rispetto alla superficie disponibile. Già in passato si era ipotizzato che il complesso capitolino di Brescia potesse essere inserito entro un reticolo formato da quadrati di lato 7,5 × 7,5110 p.r. come il tempio di Poggio Casetta a Bolsena e quello della Magna Mater sul Palatino;111 tale modulo andava a costituire una griglia di 30  ×  34 entro cui si inserisce tutto il complesso santuariale: il valore 7,5 p.r. è divisibile per l’ampiezza della facciata settentrionale, pari a 135 p.r., e per la larghezza massima di tutta l’area considerando anche i portici laterali (202 p.r.). L’arco della circonferenza di raggio 1,12 m inscrivibile nel quadrato di lati 2,2 m = 7,5 p.r. sembra confermare oggi quella ipotesi (fig. 14b). Nell’insieme, le incisioni potrebbero indicare che i progettisti utilizzarono i resti delle strutture 110 111

  Dell’Acqua 2014: 350, fig. 15.   Barresi 1990: 261, fig. 2; 276-277, fig. 13.

288

Anejos de AEspA LXXVII

Antonio Dell’Acqua

Fig. 14a. Il rilievo della lastra (autore P. L. Dander).

più antiche, oramai dismesse, come piano di calcolo per impostare lo schema modulare del complesso templare, partendo da una griglia più piccola – quadrati di 16 × 16 cm – per poi giungere a quella definitiva che utilizza il modulo di 7,5 p.r. (fig. 15). In altri contesti, invece, sono stati rinvenute le linee principali dell’edificio tracciate direttamente sul terreno: ad esempio, nel foro romano (in una fase relativa ai rifacimenti tetrarchici),112 nel Mausoleo di Adriano, in seguito ad un cambiamento nel progetto,113 oppure, ancora, nelle terme di Pollena Trocchia come segnalato in un recente studio.114 CONCLUSIONI I dati relativi ai cantieri che si sono succeduti nell’area del Capitolium di Brescia, seppur in una fase del tutto preliminare di interpretazione e stu  Giuliani e Verduchi 1987: 154-155.   Cfr. Vitti in questo volume. 114   I dati sono stati presentati in un poster da Josef Souček in occasione di questo convegno. 112 113

dio, sono utili per ricostruire più concretamente quel complesso processo che ruota attorno alla costruzione di un edificio.115 Nel caso bresciano si constata una marcata continuità di occupazione dell’area che da un lato ha comportato l’occultamento delle tracce più antiche, dall’altra ha favorito l’accumulo dei materiali che, ridotti a macerie, sono stati riutilizzati per colmare i dislivelli, secondo modalità di cantiere volte ancora ad economizzare i costi.116 Si 115   Si vedano le considerazione in merito alla complessità delle dinamiche di cantiere in Amici 2008; sull’approvvi­ gionamento dei materiali Russel 2013: 201-255. 116  In epoca romana il riutilizzo di macerie di edifici religiosi (tralasciando la fase tardo antica) sembra poco normato e soprattutto poco indagato (Lippolis 2008: 39; Marano 2012: 64, 66), a differenza dei contesti greci (Lippolis c.s.). Tacito (hist. 4.53.1) e Svetonio (Vesp. 8), ad esempio, ricordano che le macerie dell’antico tempio di Giove Capitolino, dopo la distruzione per mano dei sostenitori di Vitellio, furono scaricate nelle paludi attorno a Roma. Tra le accuse rivolte a  Verre da Cicerone c’è anche quella di aver ristrutturato il tempio dei Castori non con materiali nuovi ma con quelli delle fasi precedenti, che per consuetudine sarebbero spettati all’appaltatore dei lavori per il suo guadagno (Cic., Verr. 2.56.148). Diversa, invece, la normativa relativa alle statue di culto che dovevano essere depositate in una favissa.

Anejos de AEspA LXXVII

TRACCE DI CANTIERE DALL’AREA DEL CAPITOLIUM DI BRESCIA...

289

Fig. 14b. Ricostruzione della quadrettatura basata su moduli di 16 × 16 e della circonferenza di raggio 1,12 m. Questa si iscrive in un quadrato di lato 2,2 m = 7,5 p.r. che corrisponde al modulo usato per l’impostazione di tutto il complesso (elaborazione autore).

constata in primis che la cantierizzazione dell’area ha sempre rispettato una data superficie, corrispondente grossomodo a quella interessata dalle strutture sacrali più antiche: l’allargamento procede gradualmente, prima verso i lati (età augustea), poi verso nord e est (c.d. età flavia) avendo come costante il mantenimento della viabilità urbana, in particolare del decumano massimo che garantiva il collegamento tra est e ovest nel tratto urbano. I limiti imposti dalla topografia devono aver richiesto un’attenta pianificazione dell’area di cantiere in tutte le quattro fasi, tenendo conto che si sarebbe dovuto prevedere un percorso viario Si veda Estienne 2009. Ringrazio per le indicazioni bibliografiche Yuri Marano.

per consentire il trasporto dei materiali dalle zone di approvvigionamento, e una zona di stoccaggio dei materiali. La limitatezza dello spazio sembra avere avuto come conseguenza che le attività artigianali si siano svolte sul cantiere stesso e le strutture siano poi state dismesse man mano che procedevano i lavori: ciò ha richiesto l’utilizzo di strutture mobili e facilmente trasferibili, non specializzate ma funzionali più che altro alla manutenzione degli attrezzi. In tutte e quattro le fasi i materiali lapidei risultano provenire da est; è quindi probabile che in questa zona della città ci fossero strutture atte allo scarico dei materiali e all’immagazzinamento. Per le fasi più antiche non si conosce alcun elemento che possa essere collegato ad una struttura portuale: i rinvenimenti del 1959 in via Mantova

290

Anejos de AEspA LXXVII

Antonio Dell’Acqua

Fig. 15. Ricostruzione dell’ipotetica griglia modulare basata su di un modulo 7,5 in relazione ai tracciati sulle lastre (elaborazione autore).

si riferiscono ad una banchina, realizzata con materiale funerario di recupero, per la cui realizzazione si sono proposte datazione che oscillano tra l’età severiana117 e quella gota118 o longobarda;119 solo ipotetica, invece, un’altra struttura portuale sul Mella e sul fiume Grande Superiore, ad ovest   Mirabella Roberti 1963: 279.   Ruggiu Zaccaria 1969: 145.   Lusuardi 1984: 513.

della città, realizzata sempre con materiale funerario riutilizzato.120 Si è visto come le specie lapidee utilizzate siano prettamente locali (in alcuni casi recuperati dallo sbancamento della collina) o provenienti al massimo dall’area del vicentino;121 l’impiego dei marmi colorati è invece attestato solo nella fase

117 118

120

119

121

  Rossi 1991: 151-154.   La distanza è di circa un centinaio di km.

Anejos de AEspA LXXVII

TRACCE DI CANTIERE DALL’AREA DEL CAPITOLIUM DI BRESCIA...

c.d. flavia per la decorazione interna delle celle.122 Scelte di questo tipo rientrano, in generale, in una consolidata prassi volta alla riduzione dei costi, soprattutto in età augustea quando il rifacimento, che comportò una modifica planimetrica della struttura con l’introduzione delle ali porticate, vide addirittura l’utilizzo degli stucchi per la decorazione, anziché un apparato decorativo in pietra,123 probabile indizio di una minore disponibilità economica. Per la fase repubblicana, l’utilizzo della pietra vicentina è invece in linea con quanto accade altrove in Cisalpina, in particolare a Milano dove i materiali più antichi (fine ii-inizi i sec. a.C.) sono i quattro capitelli di via Bocchetto che la critica ha associato ad un aedes, probabilmente commissionato da “mercatores italici o personaggi di qualche potente gens romana che aveva interessi in Nord-Italia, in sintonia con i maggiori esponenti dell’aristocrazia locale”.124 Come per l’ex centro insubre, anche per Brescia si può riprendere l’ipotesi secondo cui l’utilizzo della pietra di Vicenza sia accompagnato dall’impiego di manodopera proveniente dalla zona occidentale del Veneto, e non da Aquileia, dove i materiali più antichi – e relativi a questa fase cronologica – sono scolpiti in calcare di Aurisina e i capitelli monolitici.125 Ciò conferma la capacità e la forza economica della committenza, che impegna ingenti risorse per il rinnovamento architettonico della colonia latina la cui élites aveva iniziato a godere dei privilegi previsti dallo ius Latii concesso nell’89 a.C. alle comunità della Gallia Cisal­pina.126 Al di là della provenienza delle maestranze, mi pare degno di nota l’effetto prodotto dalla migrazione degli artigiani nella locale realtà bresciana: seppur solo a livello ipotetico, si può ritenere che a partire da questo episodio si sia sviluppata in città una tradizione di botteghe lapicide locali che raggiunse una completa maturità nel corso della seconda metà del i sec. a.C. e si emancipò del tutto a partire dalla tarda età augustea, quando i calcari veneti vennero soppiantati dalla pietra di Botticino, litotipo che verrà usato massicciamente d’ora innanzi.   Cfr. Angelelli e Dell’Acqua 2014.  Le membra architettoniche vennero scalpellate e le superfici predisposte per l’applicazione della nuova decorazione plastica. Per la bibliografia si veda la prima parte. 124   Da ultimo Sacchi 2012: 47-50. 125   L’ipotesi in Sacchi 2012: 49. Per i materiali di Aquileia: Cavalieri Manasse 1978: 44-109, cat. nn. 1-83. 126   Per le conseguenze, e le relative considerazioni, della costituzione a colonia latina fittizia si rimanda a Bandelli 1990: 263-264. 122

123

291

L’impiego praticamente quasi esclusivo del Botticino,127 a partire dal rifacimento della prima età imperiale del Capitolium, è al momento in controtendenza rispetto al tradizionale utilizzo di materiali marmorei più nobili per edifici di tale importanza nel i sec. d.C., così come è noto, ad esempio, sia in Spagna sia nelle Gallie.128 Risulta difficile pensare che la scarsità di marmo, lunense o greco, a Brescia sia da imputare a difficoltà di approvvigionamento in relazione alle vie di comunicazione, dal momento che a Verona questo materiale, come altri dalla Grecia, vi giungeva e veniva impiegato nei principali edifici pubblici della città.129 È forse da domandarsi se queste scelte non avessero un legame col processo di privatizzazione delle cave di marmo avviato da Augusto e con l’impossibilità, da parte della committenza, di acquistare la quantità di marmo necessario per un complesso di queste dimensioni; oppure si può ipotizzare, più semplicemente, che i proprietari delle locali cave di Botticino avessero finanziato la costruzione dell’edificio imponendo l’impiego del proprio materiale e ottenendone in cambio vantaggi al momento non quantificabili.130 L’elemento di maggiore novità emerso dai recenti scavi è costituito dalle incisioni relative alla progettazione e al disegno dell’apparato architettonico dell’edificio di età imperiale. Il rinvenimento del graffito sulla parete intonacata, delle buche nel terreno e delle incisioni sulle lastre nella stessa area – coincidente con l’ala porticata occidentale della fase augustea – designano la zona come quella in cui si impiantò l’équipe dei progettisti e dell’architetto che elaborarono il nuovo assetto planimetrico e volumetrico del tempio. Le evidenze relative al capitolium flavio sembrano costituire la prova archeologica di quel procedimento che viene definito icnographia da Vitruvio nel De Architectura (1, 2, 2), intendendo con questo termi127   La pietra locale non sarà mai sostituita nelle architetture cittadine neanche in età severiana durante il rifacimento del teatro e solo nel Foro, per ora, risultano impiegate basi attiche in marmo e colonne di Bardiglio, Cipollino e, forse, in granito della Troade per il quale cfr. Sacchi et al. 2010. 128   Pensabene 2004: 197-199; Pensabene e Mar 2010: 280297. Questa scelta non può che avere una spiegazione ideologica, preferendosi il marmo italico a quello locale per il prestigio che il primo aveva in relazione al suo impiego nei principali edifici di committenza imperiale a partire dall’età augustea, secondo una classificazione gerarchica preventiva che escludeva di massima le pietre locali da programmi statuari, sacrali e iconografici. Mansuelli 1975: 152. 129   Terracina 1985-87: 643-655. 130   Si ricorda l’epigrafe dedicata verso la metà del II sec. d. C. ad una divinità ignota dal senatore M. Nonius Macrinus in località Botticino e proprio i Nonii potrebbero aver avuto legami economici con le cave. Gregori 1990-99: 327.

292

Anejos de AEspA LXXVII

Antonio Dell’Acqua

ne l’iter con cui si imposta il tracciato dell’edificio sul terreno.131 È questa la prima delle sette operazioni principali in cui si articola il cantiere, secondo la proposta di ricostruzione del processo costruttivo avanzata da Janet DeLaine nel 2008 nel primo workshop di questa serie.132 In particolare, le incisioni ritrovate a Brescia sembrano riferirsi al sistema modulare attestato nell’antichità che può essere considerato parte propedeutica alla raffigurazione sul terreno vera e propria e che prende avvio dalla determinazione di una unità di misura o modulo.133 Come afferma M. Wilson Jones, che considera tipica la procedura descritta da Vitruvio, dopo una prima fase, in cui viene definita l’idea generale di carattere, dimensione e sito dell’opera, ne segue una che prevede lo schema preliminare e la scelta delle proporzioni che dovevano governare l’opera.134 Oltre ai dati geometrici e di interpretazione modulare sopra esposti, si può sottolineare come la localizzazione in questa area palesi un dato in termini di cronologia relativa dell’edificio capitolino: se il graffito, che si ritiene rappresenti le modanature della porta, è stato tracciato su un muro della fase augustea collocato su lato ovest, allora la realizzazione della facciata del tempio e del suo apparato architettonico deve precedere la costruzione dei portici. Quando fu inciso il graffito le maestranze dovevano essere impegnate a lavorare sulla facciata e l’edificio doveva essere ancora sprovvisto delle ali porticate. Sembra dunque confermarsi, anche sulla base di questi dati, quanto avanzato a livello ipotetico sulla base dell’analisi della decorazione architettonica, e cioè che l’avvio della ricostruzione dell’edificio capitolino vada fatta scalare verso gli anni centrali del i sec. d.C.135 Si potrebbe dunque ritenere che l’iscrizione posta sul fregio del pronao, che indica chiaramente, sulla base delle cariche citate, una datazione al 73 d.C., sancisca la fine di un primo lotto di lavori e, verosimilmente, quello che aveva riguardato l’allestimento della terrazza capitolina e comprendente varie fasi: lo smantellamento del santuario repubblicano, la parziale eliminazione delle macerie e il loro livel131   Circa il problema esegetico del termine in Vitruvio cfr. Bartoli 1978. Si veda anche Gros 1997: 83, nota 146. 132   Ovvero dell’ “initial concept”. DeLaine 2008: 324. 133  Vitr., De Architectura 1.2.2. 134   Wilson Jones 2000: 58-63. 135  Così come già ritenuto da Mirabella Roberti che sottolineava l’uso dell’opera listata per le murature delle celle, impiegata già da quest’epoca (Mirabella Roberti 1963: 259, nota 1). Per la tecnica utilizzata cfr. Lugli 1957: 518; Mirabella Roberti 1963: 259. Per la datazione della decorazione architettonica Dell’Acqua 2014: 329.

lamento o utilizzo per colmare le celle; la predisposizione delle nuove strutture di fondazione e la delimitazione del nuovo spazio occupato dalla terrazza; la costruzione delle murature delle celle e del podio e la realizzazione degli elementi che componevano i portici laterali e il colonnato della facciata. È verosimile ritenere che, come sarà consuetudine in età medievale quando, durante la costruzione delle cattedrali, i lavori saranno avviati a partire dall’abside e dal presbiterio, anche in un edificio templare antico l’obiettivo primario fosse quello di portare a termine la cella e in un secondo momento provvedere allo spazio esterno. I quattro anni che separano l’iscrizione del fregio dall’assunzione di potere da parte di Vespasiano sembrano davvero pochi per comprendere tutte le fasi di un progetto che doveva prevedere già dall’inizio la costruzione della terrazza capitolina, del Foro e della Basilica, edifici che furono realizzati in tempi diversi, così come lasciano pensare sia considerazioni di ordine tecnico e progettuale sia l’analisi dei materiali, ma che mantennero fede ad un progetto unitario. BIBLIOGRAFIA Adam, J.-P. 2001: L’arte di costruire presso i romani: materiali e tecniche. Longanesi, Milano. Agnoli, N., Carnabuci, E., Caruso, G. e Loreti, E. M. 2014: “Il Mausoleo di Augusto. Recenti scavi e nuove ipotesi ricostruttive”, in Abbondanza, L., Coarelli, F. e Lo Sardo, E. (a cura di), Apoteosi. Da uomini a dei. Il Mausoleo di Adriano, pp. 214-229. Palombi, Roma. Alberti, E. 1994: “Le pietre tenere vicentine nei secoli”, in Cornale, P. e Rosanò, P. (a cura di), Le pietre tenere del vicentino. Uso e restauro, pp. 125-164. Associazione Artigiani della Provincia di Vicenza, Vicenza. Amici, C. M. 2008: “Dal monumento all’edificio: il ruolo delle dinamiche di cantiere”, in Camporeale, S., Dessales, H. e Pizzo, A. (a cura di), Arqueología de la Construcción I. Los procesos constructivos en Italia y en las provincias romanas: Italia y provincias occidentales (Mérida, 26-27 de octubre de 2007), pp. 17-38, Anejos de Archivo Español de Arqueología 50. CSIC, Mérida. Angelelli, C. e Dell’Acqua, A. 2014: “La decorazione delle aule”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 369-387. All’Insegna del Giglio, Firenze. Antichi edifici sul foro. Percorsi archeologici in Palazzo Martinengo a

Anejos de AEspA LXXVII

TRACCE DI CANTIERE DALL’AREA DEL CAPITOLIUM DI BRESCIA...

Brescia (a cura di A. Morandini, F. Rossi). Provincia di Brescia, Brescia. Bahat, D. 2011: Atlante topografico di Gerusalemme. Messaggero, Padova. Bandelli, G. 1990: “Colonie e Municipi delle regioni transpadane in età repubblicana”, in La città nell’Italia settentrionale in età romana. Morfologie, strutture e funzionamento dei centri urbani delle regiones X e XI. Atti del Convegno organizzato dal Dipartimento di scienze dell’Antichità dell’Università di Trieste e dall’École française de Rome (Trieste, 1315 marzo 1987), pp. 251-277, Collection dell’École française de Rome 130. Università di Trieste-École française de Rome, Trieste-Roma. Barresi, P. 1990: “Schemi geometrici nei templi dell’Italia centrale”, Archeologia Classica, 42, pp. 251-285. Bartoli, M. T. 1978: “Orthographia, ichnographia, scaenographia”, Studi e documenti di architettura, settembre 1978.8, pp. 197-208. Basso, P. 2013: “Gli edifici per spettacoli”, in Cavalieri Manasse, G. e Basso, P. (a cura di), Storia dell’architettura del Veneto. L’età romana e tardo antica, pp. 68-87. Marsilio, Venezia. Bellandi, G. 2014: “Nuove sepolture tardoantiche-altomedievali dall’area del Capitolium”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 467-468. All’Insegna del Giglio, Firenze. Bernard, H., Bessac, J.-Cl., Mardikian, P. e Feugère, M. 1998: “L’épave romaine de marbre de Porto Novo”, Journal of Roman Archaeology, 11, pp. 53-81. Bohn, R. 1896: Die Theater-Terrasse, Altertümer von Pergamon 4. Speemann, Berlin. Bianchi, B. 2014: “La decorazione pittorica del santuario repubblicano di Brescia”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 223-259. All’Insegna del Giglio, Firenze. Bianco, M. L. 2008: “La decorazione architettonica lapidea e marmorea”, in Cavalieri Manasse, G. (a cura di), L’area del Capitolium di Verona. Ricerche storiche e archeologiche, pp. 169-214. Soprintendenza per i beni archeologici del Veneto, Verona. Bonnamour, L. 2000: “Les ponts romains de Chalonsur-Saône. Étude préliminaire de la pile n° 3”, Gallia, 57, pp. 273-306. Brogiolo, G. P. 2006: “The control of public space and the transformation of an early medieval town: a re-examination of the case of Brescia”, in Bowden, W., Gutteridge, A. e Machado, C. (a cura di), Social and Political Life in Late Antiquity, pp.  251-283, Late antique archaeology 3.1. Brill, Leiden-Boston.

293

Brouquier-Reddé, V. e Gruel, K. 2004: “Le sanctuaire de Mars Mullo chez les Aulerques Cénomans (Allonnes, Sarthe) Ve s. av. J.-C. – IVe s. apr. J.-C.: état des recherches actuelles”, Gallia, 61, pp. 291-396. Brunet-Gaston, V. 2008: “Le marbre de Carrare dans les programmes architecturaux d’AugustodunumAutun (France)”, in Camporeale, S., Dessales, H. e Pizzo, A. (a cura di), Arqueología de la Construcción I. Los procesos constructivos en Italia y en las provincias romanas: Italia y provincias occidentales (Mérida, 26-27 de Octubre de 2007), pp. 490-507, Anejos de Archivo Español de Arqueología 50. CSIC, Mérida. Buchi, E. 1987: “Le strutture economiche del territorio”, in Storia di Vicenza. I. Il territorio, la preistoria, l’età romana, pp. 145-157. Pozza, Vicenza. Bugini, R. e Folli, L. 2000: “I materiali lapidei utilizzati in epoca augustea”, in Milano tra l’età repubblicana e l’età augustea. Atti del Convegno di studi (Milano, 26-27 marzo 1999), pp. 439-440. Comune di Milano, Milano. Bugini, R. e Folli, L. 2012: “Le calcareniti venete negli edifici di Mediolanum”, in Atti del VII congresso nazionale AIAr (Modena, 22-24 febbraio 2012), pp. 714-719. Pàtron, Bologna. Bugini, R. e Folli, L. 2014: “Le cave romane in Lombardia”, in Bonetto, J., Camporeale, S. e Pizzo, A. (a cura di), Arqueología de la Construcción IV. Las canteras en el mundo antiguo: sistemas de explotación y procesos productivos (Padova, 22-24 de noviembre de 2007), pp. 179-188, Anejos de Archivo Español de Arqueología 69. CSIC, Mérida. Candelato, F. e Perretti, T. 2009: “Considerazioni preliminari sulla stratigrafia archeologica dell’area di scavo ubicata presso il lato orientale del cosiddetto Tempio C di Grumentum (PZ)”, in Mastrocinque, A. (a cura di), Grumentum Romana. Convegno di studi (Grumento Nova, 28-29 giugno 2008), pp. 63-77. Porfidio, Moliterno. Caprioli, F. 2007: Vesta aeterna. L’Aedes Vestae e la sua decorazione architettonica. “L’Erma” di Bretschneider, Roma. Cattani, M. 2001: “I pesi in pietra in Etruria padana”, in Corti, C. e Giordani, N. (a cura di), Pondera. Pesi e misure nell’antichità, pp. 89-94. Museo della bilancia, Campogalliano. Cavalieri Manasse, G. 1978: La decorazione architettonica romana di Aquileia, Trieste, Pola, 1. L’età repubblicana augustea e giulio claudia. Associazione nazionale per Aquileia, Aquileia. Cavalieri Manasse, G. 1979: “La decorazione architettonica del teatro romano”, in Brescia romana. Materiali per un Museo, I-II, pp. 111-145. Comune di Brescia, Brescia.

294

Antonio Dell’Acqua

Cavalieri Manasse, G. 1997: “Note sull’edilizia funeraria romana di Brescia e Verona”, in Mirabella Roberti, M. (a cura di), Monumenti sepolcrali romani di Aquileia e nella Cisalpina, pp. 243-273, Antichità Altoadriatiche 43. Editreg, Trieste. Cavalieri Manasse, G. 2002: “Architetture ellenisticoitaliche in Cisalpina: le testimonianze del santuario bresciano”, in Rossi, F. (a cura di), Nuove ricerche sul Capitolium di Brescia. Scavi, studi e restauri, pp. 95-116. Et, Milano. Cavalieri Manasse, G. 2006: “Materiali architettonici di tradizione ellenistico-italica a Feltre”, in Bianchin Citton, E. e Tirelli, M. (a cura di), ... ut rosae ... ponerentur. Scritti di archeologia in ricordo di Giovanna Luisa Ravagnan, pp. 125-136, Quaderni archeologia del Veneto, serie speciale 2. Quasar-Canova, Treviso. Cavalieri Manasse, G. 2013: “Le testimonianze più antiche della documentazione architettonica in pietra”, in Cavalieri Manasse, G. e Basso, P. (a cura di), Storia dell’architettura nel Veneto. L’età romana e tardo antica, pp. 98-103. Marsilio, Venezia. Ciani, N. 1995: “Pesi”, in Matteini Chiari, M. (a cura di), Museo comunale di Gubbio. Materiali archeologici, pp. 405-408. Electa, Milano. Claridge, A. 1982: “Le scanalature delle colonne”, in Cozza, L. (a cura di), Tempio di Adriano, pp. 27-30. De Luca, Roma. Cornale, P. 1994: “Origine geologica delle pietre tenere”, “Zona di estrazione, cave e ciclo produttivo”, in Cornale, P. e Rosanò, P. (a cura di), Le pietre tenere del vicentino. Uso e restauro, pp. 13-38. Associazione artigiani della provincia di Vicenza, Vicenza. Corti, C. 2001: “Pesi e contrappesi”, in Corti, C. e Giordani, N. (a cura di), Pondera. Pesi e misure nell’antichità, pp. 191-212. Museo della bilancia, Campogalliano. Costabile, F., Martorano, F. e De Nittis, V. 2006: “Il tempio ionico di Marasà a Locri Epizefiri. Per un atlante dell’architettura samia di Occidente”, Polis. Studi internazionali sul mondo antico, 2006.2, pp. 17-70. Coulton, J. J. 1977: Greek Architects at Work. Elek, London. Dander, P. L. 2014a: “La fase di santuario del II sec. a.C.: i dati di scavo”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 167-170. All’Insegna del Giglio, Firenze. Dander, P. L. 2014b: “Il santuario tardo-repubblicano: i dati di scavo”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 191194. All’Insegna del Giglio, Firenze. Dander, P. L. 2014c: “La ristrutturazione augustea: i dati di scavo”, in Rossi, F. (a cura di), Un luogo per

Anejos de AEspA LXXVII

gli dei. L’area del Capitolium di Brescia, pp. 289-292. All’Insegna del Giglio, Firenze. Dander, P. L. 2014d: “Il Capitolium flavio: i dati di scavo”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 313-194. All’Insegna del Giglio, Firenze. De Franciscis, A. 1959: “Osservazioni sul disegno d’arco dell’Anfiteatro campano di S. Maria Capua Vetere”, Atti dell’Accademia dei Lincei, serie VIII, vol. XIV, pp. 399-402. De Vanna, L. 2014: “L’area del Capitolium tra tardo antico ed alto medioevo: case, botteghe e sepolture”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 411-432. All’Insegna del Giglio, Firenze. De Vecchi, G. e Lazzarini, L. 1994: “Marmi e pietre di Padova romana”, in Zampieri, G. e Cisotto, M. (a cura di), Padova romana. Testimonianze architettoniche nel nuovo allestimento del Lapidario del Museo Archeologico, pp. 108-110. Electa, Milano. DeLaine, J. 2008: “Conclusions”, in Camporeale, S., Dessales, H. e Pizzo, A. (a cura di), Arqueología de la Construcción I. Los procesos constructivos en Italia y en las provincias romanas: Italia y provincias occidentales (Mérida, 26-27 de Octubre de 2007), pp. 319-326, Anejos de Archivo Español de Arqueología 50. CSIC, Mérida. Dell’Acqua, A. 2014: “Nuovi dati sull’architettura”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 321-359. All’Insegna del Giglio, Firenze. Estienne, S. 2010: “Simulacra deorum versus ornamenta aedium. The status of divine images in the temples of Rome”, in Mylonopoulos, J. (a cura di), Divine Images and Human Imaginations in Ancient Greece and Rome, pp. 257-271, Religions in the Graeco-Roman World 170. Brill, Leiden-Boston. Fogolari, G. 1975: “La protostoria delle Venezie”, in Popoli e civiltà dell’Italia antica, IV, pp. 61-222. Biblioteca di Storia patria, Roma. Frova, A. 1979: “Tipologie e forme architettoniche degli edifici pubblici”, in Brescia romana. Materiali per un Museo, II, pp. 211-240. Comune di Brescia, Brescia. Frova, A., Rossignani, M. P. e Cavalieri Manasse, G. 1975: “Il Capitolium e la decorazione architettonica romana di Brescia”, in Atti del convegno internazionale per il XIX centenario della dedicazione del Capitolium e per il 150° anniversario della sua scoperta (Brescia, 27-30 settembre 1975), pp. 53-66. Ateneo di Brescia, Brescia. Gagetti, E. 2014: “Una forma di fusione per amuleti?”, in Rossi, F. 2014 (a cura di), Un luogo per gli

Anejos de AEspA LXXVII

TRACCE DI CANTIERE DALL’AREA DEL CAPITOLIUM DI BRESCIA...

dei. L’area del Capitolium di Brescia, pp. 405-407. All’Insegna del Giglio, Firenze. Garzetti, A. 1979: “Dedica del Tempio di Vespasiano”, Brescia romana. Materiali per un Museo, I-II, pp. 190-193. Comune di Brescia, Brescia. Giuliani, C. F. 2006: L’edilizia nell’antichità. Carocci, Roma (nuova edizione). Giuliani, C. F. e Verduchi, P. 1987, L’area centrale del foro romano. Olschki, Firenze. Gnoli, R. 1988: Marmora romana. Edizioni dell’Elefante, Roma. Grassi, E. 2014: “L’officina di forgia”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 195-198. All’Insegna del Giglio, Firenze. Gregori, G. L. 1990-1999: Brescia romana. Ricerche di prosopografica e storia sociale, I. I documenti, II. Analisi dei documenti. Quasar, Roma. Gregori, G. L. 2000: Brescia romana. Ricerche di prosopografia e storia sociale, II. Analisi dei documenti. Quasar, Roma. Gregori, G. L. 2014: “Una dedica per la triade capitolina”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 319-320. All’Insegna del Giglio, Firenze. Gros, P. 1976: Aurea Templa. Recherches sur l’architecture religieuse de Rome à l’époque d’Auguste, Bibliothèques des Écoles Françaises d’Athènes et de Rome 231. École française de Rome, Roma. Gros, P. 1997 (a cura di): Vitruvio. De Architectura (traduzione e commento di A. Corso e E. Romano). Einaudi, Torino. Grüben, G. 1966: Die Tempel der Griechen. Hirmer, Berlin. Haselberger, L. 1983: “Die Werkzeichnung des Naiskos im Apollontempel von Dydima”, in Blauplanung und Bautheorie der Antike, pp. 111-119, Diskussionen zur archäologischen Bauforschung 4. Deutsches archäologisches Institut, Berlin. Huber, S. 1997: “Activité métallurgique dans le sanctuaire d’Apollon à Erétrie”, in Gills, C., Rinsberg, C. e Sjöberg, B. (a cura di), Trade and Production in Premonetary Greece. Production and the Craftsman. Proceedings of the 4th and 5th International Workshops, Athens, 1994 and 1995), pp. 173-184. Åström, Jonsered. Inglese, C. 1999: “I tracciati di cantiere incisi nel Mausoleo di Augusto e sul Pantheon a Roma: ipotesi di lettura”, in Migliari, R. (a cura di), Strumenti del Dottorato di Ricerca, 1. Geometria ed architettura, pp. 43-62. Gangemi, Roma. Inglese, C. 2012: “All’origine del disegno architettonico esecutivo: Συνγραϕαι (Syngraphai), Παραδειγμα

295

(Paradigma) ed Αναγραϕευζ (Anagrapheus)”, in Carlevaris, L. e Filippa, M. (a cura di), Elogio della teoria. Identità delle discipline del disegno e del rilievo. Atti del 34° Convegno internazionale dei docenti della rappresentazione (Roma, 2012), pp. 275-284. Gangemi, Roma. Inglese, C., Pizzo, A. 2014: I tracciati di cantiere di epoca romana. Progetti, esecuzioni, montaggi. Gangemi, Roma. Kockel, V. 2014: “Fragmente marmorner Architekturverkleidungen aus dem sog. Macellum in Ostia”, Römische Mitteilungen, 120, pp. 227-242. Labus, G., Basiletti, L. e Vantini, R. 1838: Museo bresciano illustrato. Tipografia della Minerva, Brescia. Lippolis, E. 2008: “Lo spazio per votare e altre note di topografia sulle agorai di Atene”, Annuario della Scuola Archeologica Italiana di Atene e delle missioni italiane in oriente, 84, serie III, 6, tomo I, 2006, pp. 37-62. Lippolis, E. c.s.: “Restauro e reimpiego nelle poleis della Grecia: esempi e forme di una prassi negata”, in Selinus 2011. Ricerche ed esperienze nel Mediterraneo di età greca. Atti del Convegno (Selinunte, 20-23 ottobre 2011). Loiseau, Ch. 2012: “Les métaux dans les constructions publiques romaines: applications architecturales et structures de production (ier-iiie siècle ap. J.-C.)”, in Camporeale, S., Dessales, H. e Pizzo, A. (a cura di), Arqueología de la Construcción III. Los procesos constructivos en el mundo romano: la economía de las obras (Paris, 10-11 de diciembre de 2009), pp. 117129, Anejos de Archivo Español de Arqueología 69. CSIC, Madrid-Mérida. Lugli, G. 1957: La tecnica edilizia romana con particolare riguardo a Roma e Lazio. Bardi, Roma. Lusuardi Siena, S. 1984: “Sulle tracce della presenza gota in Italia: il contributo delle fonti archeologiche”, in Magistra barbaritas. I barbari in Italia, pp. 509-558. Scheiwiller, Milano. Kalayan, H. 1971: “Notes on assembly marks, drawings and models concerning the Roman period monuments in Lebanon”, Les Annales Archéologiques Arabes Syriennes, 21, pp. 269-274. Kleiner, F. S. 1977: “Artists in the Roman world. An itinerant workshop in Augustan Gaul”, Melanges de l’École Française de Rome. Antiquité, 89, pp. 661-696. Mansuelli, G. A. 1975: “Note sulla cultura di Brescia romana”, in Atti del Convegno internazionale per il XIX centenario della dedicazione del Capitolium e per il 150° anniversario della sua scoperta (Brescia, 27-30 settembre 1975), pp. 151-156. Ateneo di ­Brescia, Brescia.

296

Antonio Dell’Acqua

Mariani, E. 2014: “Il primo santuario: intonaci dipinti dai nuovi scavi”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 183185. All’Insegna del Giglio, Firenze. Marano, Y. 2012: “Fonti giuridiche di età romana (i sec. a.C.-vi sec. d.C.) per lo studio del reimpiego”, in Cuscito, G. (a cura di), Riuso di monumenti e reimpiego di materiali antichi in età postclassica: il caso della Venetia, pp. 63-84, Antichità Altoadriatiche 74. Editreg, Trieste. Mattern, T. 2001: Gesims und Ornament. Zur stadtrömischen Architektur von der Republik bis Septimius Severus. Scriptorium, Münster. Miazzo, L. 2002: “Strumenti e scarti di lavorazione”, in Rossi, F. (a cura di), Nuove ricerche sul Capitolium di Brescia. Scavi, studi e restauri, pp. 452-457. Et, Milano. Mirabella Roberti, M. 1963: “Archeologia e arte di Brescia romana”, in Treccani degli Alfieri, G. (a cura di), Storia di Brescia, pp. 233-316. Morcelliana, Brescia. Morandini, F. 2012: “Abitare a Brescia in età flavia”, in Panazza, P. F. e Morandini, F. (a cura di), Divus Vespasianus. Pomeriggio di studio per il bimillenario della nascita di Tito Flavio Vespasiano imperatore romano, pp. 84-109. Ateneo di Brescia, Brescia. Morandini, F. 2015: “I pavimenti del santuario tardorepubblicano di Brescia”, in Angelelli, C. e Paribeni, A. (a cura di), Atti del XX colloquio dell’associazione italiana per lo studio e la conservazione del mosaico (Roma, 19-22 marzo 2014), pp. 375-382. Scripta manent, Tivoli. Morandini, F., Massara, D. e Slavazzi, F. 2014: “I pavimenti del santuario tardo-repubblicano”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium a Brescia, pp. 207-218. All’Insegna del Giglio, Firenze. Netzer, E. 2006: L’architettura di Erode. Il grande costruttore. Messaggero, Padova. Oz, L. 2014: “Jerusalem, Qiryat Ha-Yovel”, Hadashot Arkheologiyot. Excavations and Surveys in Israel, 126 (online). Pensabene, P. 2004: “Roma e le capitali provinciali. Contributi per lo studio dell’architettura e della decorazione architettonica in marmo nella Hispania romana”, in De Arbulo, J. R. (a cura di), Simulacra Romae. Roma y las capitales provinciales del Occidente Europeo. Estudios arqueológicos, pp. 175-199. Generalitat de Catalunya, Tarragona. Pensabene, P. e Mar, R. 2010: “Il tempio di Augusto a Tarraco”, Archeologia Classica, 61, n.s. 11, pp. 241-307.

Anejos de AEspA LXXVII

Rockwell, P. 1989: Lavorare la pietra: manuale per l’archeologo, lo storico dell’arte e il restauratore. NIS, Roma. Rodolico, F. 1965: Le pietre delle città d’Italia. Le Monnier, Firenze. Rossi, O. 1693: Le memorie bresciane. Opera istorica et simbolica di O.R. in Brescia per Bartolomeo Fontana. Rossi, F. 1991: “Brescia. Via Carducci. Necropoli e strutture medievali”, Notiziario Soprintendenza archeologica della Lombardia, pp. 151-154. Rossi, F. 2005: “Domus romane a Brescia. Un primo inquadramento dei contesti residenziali urbani”, in Borgiolo, G. P., Rossi, F. e Morandini, F. (a cura di), Dalle domus alla corte regia. S. Giulia a Brescia: gli scavi dal 1980 al 1992, pp. 13-34. All’Insegna del Giglio, Firenze. Rossi, F. 2007: “Brixia tra età tardorepubblicana e I sec. d.C. Nuovi dati dall’area del Capitolium”, in Brecciaroli Taborelli, L. (a cura di), Forme e tempi dell’urbanizzazione nella Cisalpina (ii sec. a.C.-i sec. d.C.). Atti delle giornate di studio (Torino, 4-6 maggio 2006), pp. 205-214. All’Insegna del Giglio, Borgo San Lorenzo. Rossi, F. 2014 (a cura di): Un luogo per gli dei. L’area del Capitolium a Brescia. All’Insegna del Giglio, Firenze. Rossi, F. 2014: “Sequenze cronologiche e culturali nell’area del Capitolium tra protostoria e prima romanizzazione”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 153-164. All’Insegna del Giglio, Firenze. Rossi, F. e Garzetti, A. 1995: “Nuovi dati sul santuario tardo-repubblicano di Brescia”, in Cavalieri Manasse, G. e Roffia, E. (a cura di), Splendida civitas nostra. Studi in onore di A. Frova, pp. 77-94. Quasar, Roma. Rossi, F. e Miazzo, L. 2002: “L’officina di un artigiano nel Capitolium di Brescia”, in Bronzi di età romana in Cisalpina. Novità e riletture. Atti della XXXII Settimana di studi aquileiesi, 28-30 maggio 2001, pp. 427-438, Antichità Alto Adriatiche 51. Editreg, Trieste. Rossignani, M. P. 1986: “Monumenti pubblici e privati di età tardo-repubblicana nei centri urbani della Lombardia”, in Atti del 2° Convegno archeologico regionale (Como, 13-15 aprile 1984), pp. 215-239. Società archeologica comense, Como. Rubinich, M. 2010: “Locri Epizefiri: resti di un’officina metallurgica nell’area del santuario di Marasà”, in Lepore, L. e Turi, P. (a cura di), Caulonia tra Crotone e Locri. Atti del convegno internazionale (Firenze, 30 maggio-1 giugno 2007), pp. 389-398. Firenze University Press, Firenze.

Anejos de AEspA LXXVII

TRACCE DI CANTIERE DALL’AREA DEL CAPITOLIUM DI BRESCIA...

Ruggiu Zaccaria, A. 1969: “Indagini sull’insediamento longobardo a Brescia”, Contributi dell’Istituto di Archeologia, 2, pp. 110-150. Ruiz de la Rosa, J. A. 1987: Traza y simetría de la arquitectura en la antigüedad y medioevo. Universidad de Sevilla, Sevilla. Russell, B. 2013: The Economics of the Roman Stone Trade. Oxford University Press, Oxford. Sacchi, F. 2012: Mediolanum e i suoi monumenti dalla fine del ii sec. a.C. all’età dei Severi. Vita e pensiero, Milano. Sacchi, F. 2014a: “La prima fase edilizia del santuario (ii secolo a.C.)”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 171178. All’Insegna del Giglio, Firenze. Sacchi, F. 2014b: “La seconda fase edilizia del santuario (prima metà del i sec. a.C.)”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 201-205. All’Insegna del Giglio, Firenze. Sacchi, F. 2014c: “La terza fase edilizia del santuario (l’età augustea)”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 293302. All’Insegna del Giglio, Firenze. Sacchi, F., Dell’Acqua, A., Bugini, R. e Folli, L. 2011: “I portici del Foro di Brescia”, in Maggi, S. (a cura di), I complessi forensi della Cisalpina romana: nuovi dati. Atti del convegno (Pavia, 12-13 marzo 2009), pp. 115-129. All’Insegna del Giglio, Firenze. Solano, S. 2014: “Un edificio seminterrato ai piedi del colle Cidneo: casa o sacello cenomane?”, in Rossi, F. (a cura di), Un luogo per gli dei. L’area del Capitolium di Brescia, pp. 43-48. All’Insegna del Giglio, Firenze. Strong, D. e Brown, D. 1976: Roman Crafts. Duckworths, London.

297

Serlorenzi, M. 2010: “La costruzione di un complesso horreario a Testaccio. Primi indizi per delineare l’organizzazione del cantiere edilizio”, in Camporeale, S., Dessales, H. e Pizzo, A. (a cura di), Arqueología de la Construcción II. Los procesos constructivos en Italia y en las provincias romanas: Italia y provincias orientales (Siena, 13-15 de noviembre de 2008), pp. 105-126, Anejos de Archivo Español de Arqueología 57. CSIC, Madrid-Mérida. Schwandner, E. L. 1990: “Beobachtungen zur hellenistischen Tempelarchitektur von Pergamon”, in Hermogenes und die hochhellenistische Architektur. Internationales Kolloquium in Berlin vom 28. bis 29. Juli 1988, pp. 85-102. Von Zabern, Mainz. Terracina, F. 1985-87: “Importazione di marmi bianchi per uso architettonico nella Cisalpina: il caso di Verona”, Quaderni di Studi Lunensi, 10-12, pp. 643-659. Tognazzi, I. 1997: “Il botticino e la pietra bresciana nell’architettura moderna e contemporanea”, in Porteri, A. e Simoni, C. (a cura di), Il marmo bresciano. Territorio, vicende, economia, pp. 191-206. Grafo, Brescia. Torelli, M. 1980: “Innovazioni nelle tecniche edilizie romane tra il I sec. a.C. e il i sec. d.C.”, in Tecnologia, economia e società nel mondo romano. Atti del Convegno (Como, 27-29 settembre 1979), pp. 139-159. Banca Popolare Commercio e Industria, Como. Wilson Jones, M. 2000: Principles of Roman Architecture. Yale University Press, New Heaven, Conn.London. Zezza, M. G. 1982: “I materiali lapidei locali impiegati in età romana nell’area compresa tra il Ticino e il Mincio”, Atti della Società Italiana di Scienze Naturali, Museo Civico di Storia Naturale di Milano, 123, pp. 3-177.

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

THE ‘LOWER AGORA’ OF PERGAMON. THE ORGANISATION OF A MAJOR BUILDING SITE IN ROMAN ASIA MINOR BURKHARD EMME*, ARZU ÖZTÜRK** (†) * Freie Universität Berlin, Institut für Klassische Archäologie, Berlin ** Mimar Sinan Güzel Sanatlar Üniversitesi, Arkeoloji Bölümü, İstanbul

ABSTRACT: The Lower Agora of Pergamon was excavated in the early 20th century by Wilhelm Dörpfeld. The building was originally dated to the mid-2nd century BC. However, recent excavations prove that the building has to be dated roughly to the Augustan period instead. Therefore, the complex yields important evidence for the development of Pergamene building technology in the transitional phase between the late Hellenistic and the early Imperial periods. This paper focuses on the organisation of a major building site in Asia Minor with special regards to the production of material away from the site and the division of labour on the building site itself. Close observations of the extant architecture as well as architectural pieces make it possible to arrive at a better understanding of the building process and its social and historical implications. KEYWORDS: Pergamon, Lower Agora, Building material, Building technology, Organisation of building site. RESUMEN: El Ágora de la ciudad baja de Pérgamo fue excavada al comienzo del siglo xx por Wilhelm Dörpfeld. El edificio fue originalmente fechado a mediados del siglo ii a. C. Sin embargo, las recientes excavaciones demuestran que su construcción puede vincularse, aproximadamente, con una fecha de época de Augusto. Por lo tanto, es necesario revisar las pruebas para el desarrollo de la tecnología de construcción en Pérgamo en la fase de transición entre el final del periodo helenístico y los primeros períodos imperiales. Este documento se centra en la organización de una obra de construcción importante en Asia Menor, teniendo especialmente en cuenta la producción de material fuera del lugar y la división del trabajo en la misma obra. Se aportan observaciones sobre la arquitectura existente y elementos arquitectónicos que permiten una mejor comprensión del proceso de construcción y sus implicaciones sociales e históricas. PALABRAS CLAVE: Pérgamo, ágora de la ciudad baja, material de construcción, tecnología de la construcción, organización de la obra.

INTRODUCTION In an influential essay published in 1924 the German archaeologist and architect Armin von Gerkan stressed the importance of architectural remains as a primary historical source (Gerkan 1959). However, whereas the interpretation of arch­itecture as a static background of social inter­ action has been preponderant in scholarly literature, the implications of the actual process of building have often been neglected. It is only in

recent years that scholars have focussed on a historical interpretation of building technology and construction processes. Two complementary perspectives are possible. On the one hand, a close reading of architectural remains can lead to a detailed understanding of the building itself, its history starting with its construction, and its later alterations. On the other hand, a comparison of different buildings can provide us with an understanding of a broad architectural landscape, its relationship to either local or regional traditions,

300

BURKHARD EMME, ARZU ÖZTÜRK

or innovations and transfer of knowledge from different regions respectively. On this level changes in building technology might indicate technological advance as well as social or even political alterations. An interesting case is the spread of ‘Roman’ or Italian building technology in the Eastern Mediterranean during the early stage of Roman rule in the east. In fact, a remarkable technological transfer can occasionally already be observed in the time of Augustus. A conspicuous example is the palace of King Herod the Great at Jericho that was partially built in opus reticulatum (Netzer 2001: 231-232). This feature closely relates the palace to contemporaneous constructions in central Italy. Therefore, it has been suggested that the responsible workmen were sent to Judaea by Augustus himself. Similarly, in Limyra the alleged cenotaph for Caius Caesar was constructed out of ashlar with a backfill of opus caementicium (Ganzert 1984: 94-100 Pl. 18-19). In this case, the whole construction is very similar to grave monuments in Rome, like those along the Via Appia. In the case of Pergamon the introduction of Italian building technology and the overall development of the city’s building industry has so far not been thoroughly discussed. The reason for this is a gap in the material record. As a centre of the Hellenistic kingdom of the Attalids, a considerable quantity of Pergamene architecture was built in the first half of the 2nd century BC (Conze 1913: 214-232; Wulf 1994: 142-151). On the contrary, the majority of Roman imperial architecture in Pergamon is dated to the later 1st or the 2nd century AD (Wulf 1994: 154-168; Radt 1999: 47). Since both phases mark distinctive stages in the architectural development of the city, it is usually possible to discern architectural features of both periods according to their building technology. In this respect, Pergamon does not differ much from other sites in Asia Minor. Typical features of Hellenistic monumental architecture in Pergamon include the use of local stone material (andesite) for visible structures such as columns etc., the use of clamps and dowels for joining architectural members and the absence of lime-based mortar and brick (Conze 1913: 153-155). Marble was used rarely for highly prestigious royal building projects such as the Great Altar or the colonnades in the sanctuary of Athena dedicated by Eumenes II (Schrammen 1906: 15-45; Radt 1999: 165). In contrast, major buildings of Roman imperial Pergamon show the full range of building technologies usually associated with ‘Roman’ architecture. This includes the

Anejos de AEspA LXXVII

use of brick and mortar in the case of the ‘Red Hall’, and, far more regularly, the use of opus caementicium. Arches and vaults were used for substructures, for example in case of the Temple of Trajan. And finally the use of marble and, more specifically, of monolithic marble columns is widespread among Pergamene buildings of Roman imperial date for the porticoes surrounding the ­Temple of Trajan and the Asklepieion, for ­example. This observation of two different periods in the development of Pergamene architecture automatically leads to the question when exactly Pergamene architects decided to adopt new building technologies and what other changes resulted from this transformation. However, due to the above-mentioned gap in the chronology of Pergamene architecture, for a long time answering this question seemed hardly possible. In fact, a scholar reading his way through the publications of the Pergamon excavation might come to the conclusion that there was hardly any building activity between the later 2nd century BC and, say, the later 1st century AD. It is exactly this gap where the Lower Agora comes into play. Based on its position within the Hellenistic city walls, the complex was originally dated to the mid-2nd century BC as well (Dörpfeld 1902: 26). However, as new excavations have shown, the building has to be dated roughly to the late 1st century BC/early 1st century AD instead, i.e. roughly to the Augustan period (Emme et al. 2014: 129). Thus, the Lower Agora becomes an important example of Pergamene architecture in the intermediate phase between the Hellenistic period and the Roman imperial era. The Lower Agora of Pergamon was excavated in the years 1900-1902 by the German archaeologist and architect Wilhelm Dörpfeld. In its original state the building consisted of a rectangular square surrounded by porticoes on all four sides (figs. 1-2). Behind these a small number of rooms was placed mainly in the north-eastern and the western part of the building. In the northern and, presumably, the western aisle there was an upper floor. Due to the sloping terrain, the southern and eastern part of the building had a basement floor with rooms behind a colonnade opening to the outside of the building. Today the complex is in a rather poor state of preservation. Only small sect­ ions of the back walls are preserved to a considerable height, especially in the north-western corner, where the natural rock was cut deeply to provide the necessary building ground. On the

Anejos de AEspA LXXVII

The ‘Lower Agora’ of Pergamon...

301

Fig. 1. Pergamon, Lower Agora and nearby structures.

other hand, in the southern and eastern part of the building only small sections of the basement walls still exist. Little has remained of the colonnade as well and most architectural members were found out of their original position. One reason for this is that the central square was used as a building ground for a church in the early by­ zantine era (Dörpfeld 1902: 31-35; Rheidt 1991: 226-228). Like in many other places the architectural pieces of the original porticoes were re-used as building material for the Byzantine construction. Finally, immediately after its excavation by Dörpfeld, several rooms of the Agora were rebuilt in order to provide a small museum as well as ­depot facilities for the finds of the Pergamon ex-

cavation (Dörpfeld 1902: 23). Following this tradition the place is still used as a repository today. Dörpfeld’s work was published only in two short preliminary reports (Dörpfeld 1902; Dörpfeld 1904). The first report includes a plan of the building by the hand of the Greek architect Panagiotis Sursos as well as a reconstruction of the façade of the central court and a section drawing (figs. 2-3). In addition, a considerable number of photographs were taken illustrating the progress of the work. Although most of the photos remained unpublished they are still accessible in the archives of the German Archaeological Institute in Athens where they form an important source of information for the current project.

302

Anejos de AEspA LXXVII

BURKHARD EMME, ARZU ÖZTÜRK

Fig. 2. Pergamon, Lower Agora. Plan (Dörpfeld 1902: plan 2).

Fig. 3. Pergamon, Lower Agora. Section drawing (Dörpfeld 1902: plan 2).

SOME METHODOLOGICAL CONSIDERATIONS The preliminary character of Dörpfeld’s publication causes several problems. For example, Sursos’ plan mixes features of a reconstruction and a present-state plan. Also, in the case of the

reconstruction of the façade not a single architectural piece was published in its actual state. Therefore, it is hardly possible to test Dörpfeld’s reconstruction of the building against the archaeological evidence. Moreover, it is clear that the focus of the excavators was the reconstruction of the ­alleged Hellenistic building in its original state,

Anejos de AEspA LXXVII

which was supposed to be a part in the masterplan of the extension and embellishment of the city under king Eumenes II. Consequently later phases were only treated cursorily by the exca­ vators. Therefore, it is the aim of the current project to consider and document all extant architectural pieces of the structure. This approach is essential in order to discern differences in style and/or technology that might point to different construction phases or workshops engaged in the building process respectively. Two problems arise from this method. On the one hand, it seems difficult to base general assumptions on a small material ­basis. This becomes evident with regard to the different groups of architectural pieces. For example, the number of extant capitals is very limited. In the case of the colonnade of the upper storey there is only one capital left, although the old photographs prove that several pieces were found in the course of Dörpfeld’s excavations (Dörpfeld 1902: Pl. 5). With only one piece remaining, it is unclear whether the absence of a dowel hole on the upper side of this capital is a general feature of the construction or rather a unique phenomenon of this individual piece. The question of the capitals of the lower storey is more difficult and cannot be discussed here in full. Suffice it to say that two types of capitals have been found. Whereas one type is a standard Doric capital, the  second is a typically Pergamene type with a cavetto-­shaped echinus (‘Tuscan capital’ – Rumscheid 1994: 304). According to their dimensions both types could be attributed to the lower storey. Since both types are preserved in two pieces each, it is not possible to determine which one originally belonged to the Agora. But even when there is a considerable number of pieces left, their study and documentation can pose problems. For example, with approximately 45 discernible pieces, the number of extant column drums of the lower storey allows some general considerations concerning the construction of this part of the building. However, since about one half of these drums were re-erected on the original stylobate or integrated into the construction of a modern depot building, it is currently impossible to study their original surfaces with regard to lifting holes, dowel holes etc. (fig. 4). The same is true for large sections of the stylobate, where columns have been re-erected. This leads to yet another problem concerning the question of columns. The accessible section of the sty-

The ‘Lower Agora’ of Pergamon...

303

lobate illustrates that the column drums were ­usually joined with only one dowel that was positioned centrally. However, a small number of columns with matching dimensions has two dowel holes. Currently there is no way to decide whether to exclude these pieces from the reconstruction of the building or not. A similar problem is the question of the attribution of the entablature. In his reconstruction, Dörpfeld assumed that the same type of Doric entablature was used for both, the upper as well as the lower storey of the façade (Dörpfeld 1902: 20 fig. 2; here fig. 10). From a technical point of view, this seems possible since both sections of the ­façade share the same system of interaxial spaces. However, all other parts of the upper storey were made of limestone, instead of the andesite that was used for the lower storey (see below). Due to the different material of the columns it seems possible that the entablature of the upper storey was made of limestone as well. In this case no parts of the entablature of the upper storey would have survived. In fact, other two-storeyed porticoes in Pergamon usually show two different types of entablature (Bohn 1885: 40). This question that so far has not been clearly solved obviously has an impact on the interpretation of the extant pieces and their allocation to different parts of the façade. Yet another more general problem lies in the interpretation of the differences between the individual pieces of one series of architectural members as either indicating chronological developments such as later repairs, or rather being the result of different workshops working contemporarily side by side. A good example for this problem is the stylistic execution of the architrave blocks. In this case it is possible to discern two different groups of blocks (fig. 5). The more elaborate group shows a decoration with a narrow head band on top of the triglyph and overlapping corners on the upper edge (‘Triglyphenöhrchen’Rumscheid 1994: 313-314). In contrast, the simple version does not show these details. Since both groups have the same dimensions and are made of the same material, their attribution to the Lower Agora is equally justified. However, it seems currently impossible to decide whether the different execution of the ornaments is a result of different workshops or rather indicates a later repair. Apart from all these imponderables it is still possible to discuss a number of aspects concerning the architecture of the ‘Lower Agora’ and ­especially its building process.

304

BURKHARD EMME, ARZU ÖZTÜRK

Anejos de AEspA LXXVII

Fig. 4. Pergamon, Lower Agora. Re-erected columns in front of modern depot building.

Fig. 5. Pergamon, Lower Agora. Architraves with differing stylistic features.

Anejos de AEspA LXXVII

BUILDING MATERIAL The architecture of the two-storeyed porticoes was systematically composed of two different types of stone, andesite and limestone. In contrast to the architecture of later Roman Imperial Pergamon, marble was not used for any part of the construction. The obvious absence of marble was explicitly one of the reasons for Dörpfeld to date the Lower Agora to the Hellenistic era (Dörpfeld 1902: 26). In fact, the use of ‘local’ ­materials for certain elements such as columns relates the Lower Agora with many of its Hellenistic predecessors, although marble was already being used for comparable constructions such as the porticoes in the sanctuary of Athena in the 2nd century BC (Bohn 1885: 73). Most parts of the walls as well as the colonnade of the lower storey were made of pink ande­ site. Generally, this stone is a common feature of both Hellenistic and Imperial constructions in Pergamon. However, whereas the use of this stone for architectural pieces such as columns was the rule for Hellenistic stoai, Roman imperial porticoes usually show a façade of white marble (Radt 1999: 218-219, 234-236). Also, there are considerable differences between both eras with regard to the construction of the columns. While the Hellenistic architects generally tended to build up a column out of several drums, the marble columns of the Roman imperial era were usually carved out of a single piece. This difference even applies to the rare cases where marble was used for Hellenistic buildings such as the stoai in the sanctuary of Athena that can be dated epigraphically to the reign of king Eumenes II (Bohn 1885: 34, 73). Since the natural rock of the Pergamene acropolis consists of andesite as well, it is commonly assumed that this kind of stone was quarried locally, i.e. directly on the building site when the ground for the building was cut from the natural rock. However, a closer observation of the building site proves that the Agora was built on a slope of soft andesite rock of a pale grey colour. It is definitely possible to discern this material from the pink andesite, that was used for most sections of the visible architecture of the same building. In contrast it seems reasonable to assume that the material gained from cutting the natural rock in the western part of the building site was used mainly for the backfilling of the terrace on the southern and the eastern side of it (figs. 2-3; Conze 1913: 151). Thus, it becomes

The ‘Lower Agora’ of Pergamon...

305

clear that the pink andesite was quarried away from the building site proper. It is possible, for example, that this stone was quarried on the northern slope of the Acropolis hill where quarries were observed during recent surveys (Conze 1913: 151; Pirson 2012: 187-190). In addition to the variety of stone, timber was used in considerable quantities for the constructive parts of the Agora. Especially interesting is the construction of the entablature, where timber and stone were used side by side. On their rear side the architrave blocks show a narrow recess (figs. 6-7). In his reconstruction Dörpfeld rightly pointed out that this feature was supposed to ­allow the installation of a second beam in order to support the cross beams for the floor of the first storey (Dörpfeld 1902: 21 fig. 2). With regard to the material of these beams Dörpfeld’s reconstruction with its uniform hatching implies that he thought them to be of stone as well. However, no such piece has been identified either by Dörpfeld or by recent investigations. With a reconstructed thickness of approximately 25 cm, it seems more plausible to assume that these beams were made of wood. Given the sequence of the different steps in the building process this observation indicates that workmen specialised in both materials must have worked side by side (fig. 7; Hoepfner 1996: 28). Furthermore, a comparison with earlier constructions in Pergamon illustrates that the combination of wood and stone within the constructive element of the entablature is highly unusual in the Hellenistic period. N ­ ormally the architrave and triglyph frieze were carved from two separate blocks with the architrave having the full thickness necessary to support the crossbeams lying on top of it (fig. 6). The remarkable solution found for the construction of the Agora obviously results from a preference for wood instead of stone. The extensive use of timber presumably follows a Helle­n­ istic tradition. It is generally assumed that the Pergamene kingdom controlled considerable resources of woods in the Pergamene hinterland, especially the forests of the modern Kozak mountains north of Pergamon. In an honorary decree from Miletus dating back to the 2nd century BC, king Eumenes II is honoured for a donation of timber for the construction of the local gymnas­ ium (Herrmann 1965: 80). It is clear that the king must have had considerable supplies of wood at his disposal if he could offer it so deliberately to other poleis within his empire. It is likely that the

306

BURKHARD EMME, ARZU ÖZTÜRK

Anejos de AEspA LXXVII

Fig. 6. Pergamon, comparison of entablature of Hellenistic portico buildings.

timber for the woodwork of the Lower Agora was produced in the Kozak mountains as well. In contrast to the lower storey the upper storey of the Lower Agora was built of limestone. This material was quarried on the southern side of the Pergamene plain and was used in Pergamon from the 4th century BC onwards (Pirson 2012: 243). At first glance it seems possible that in case of the upper storey a different material was used for constructional reasons. However, both materials share the same specific weight ranging between 2.5-2.8 gr/cm3 (andesite) and 2.6-2.9 gr/ cm3 (limestone) respectively. Thus, it becomes clear that in the case of the Lower Agora limestone was not used for its constructional characteristics. This leads to the conclusion that the alter­nation in material probably has aesthetic reasons. In fact, a comparison with contemporary buildings in Pergamon illustrates that, in the case of the Agora, limestone was used as a substitute for the more expensive white marble. For example, the peristyle of the nearby house of the

­ onsul Attalos combines a lower colonnade of C andesite with an upper storey made of marble (fig. 1; Dörpfeld 1907: 167-189; Radt 1999: ­97-100). In fact, the overall impression of a two-coloured façade with a shining white upper storey must have been very similar in both cases. It seems possible that the colourful impression was meant to recall the more ambitious constructions of Augustan Rome, namely the Forum of Augustus with its wide range of coloured marbles. An interesting aspect of the organisation of the building site is the position of the different materials within the building with regard to their original place of production. Obviously, the distance to the place of production of the different materials becomes longer during the building process. At a first stage a considerable amount of material was necessary in order to create the terrace of the central court. In this case efforts of transport were minimized presumably by re-using the material cut from the western part of the emerging terrace. At a second stage, pink andesite

Anejos de AEspA LXXVII

The ‘Lower Agora’ of Pergamon...

307

Fig. 7. Pergamon, Lower Agora. Reconstruction of entablature.

was used for the architecture of the lower storey and the back walls. This stone was presumably quarried either within the city or at least very close to it. After the construction of the lower colonnade and the walls other materials had to be transported over longer distances, like timber and limestone. It seems plausible to assume that these chains of delivery were taken into account from the very beginning of the building process. Presumably, the production of the different materials started more or less at the same time in order to ensure that every kind of material was available on the building site when it was required. The same can be assumed for the production of rooftiles that were necessary in large quantities. Closely related to the observation of different materials is the question of contractors. In the

case of the stone material this question is closely linked to the phenomenon of contractors’ marks. These marks are a common feature of Hellenistic buildings in Pergamon such as the gymnasium, the sanctuary of Demeter or the city walls (Bachmann 2009: 218-222). They usually consist of one or two letters, sometimes combined to form a liga­tion. These marks are regularly observed on surfaces of individual ashlars. It is a common assumption that the marks were used to specify the quarry and/or the contractor that delivered the stone material to the building site (Bachmann 2009: 221). In a recent study of the phenomenon Martin Bachmann rightly pointed out that during the period of its major extension in the first half of the 2nd century BC a considerable number of contractors was necessary in order to meet the

308

Anejos de AEspA LXXVII

BURKHARD EMME, ARZU ÖZTÜRK

r­equirements of the growing Attalid capital (Bachmann 2009: 220). In contrast to these examples the extant original walls of the Lower Agora do not bear masonry marks at all. A small number of loose blocks with marks that were found in the area of the Agora may be explained as spolia that were re-used in the construction of the Byzantine church on the spot. The absence of masonry marks indicates that the organisation of quarrying and/or delivery of the stone material must have differed from the one used in Hellenistic times. It seems reasonable to suppose that in the time of the construction of the Agora, there was only one contractor responsible for the delivery of stone material – or possibly two, given the two different types of stone that were used. CONSTRUCTION Due to the sloped terrain, levelling the building site was already a permanent challenge for Pergamene architects in the Hellenistic period. Large terraces with retaining walls were needed for the big public structures and sacred buildings. In the Hellenistic period Pergamene architects developed a retaining system with small rooms for the necessary support of the terraces. Hence, two parallel walls including a distance between them were divided into several sections by crosswalls. In this case, a series of small rooms was formed in order to support and enlarge the terrace above (figs. 2-3; Klinkott 1991). These rooms have usually no function and are sometimes filled with earth, small stones or debris etc. As a typical feature of the Hellenistic Pergamene building technology, there is a narrow channel on the slope side of these walls in order to provide a drainage system for the upper terrace (fig. 3). This feature is referred to as peristasis in Pergamene inscriptions. The façade of the retaining system for a lower terrace has usually two architectural forms: there is either a stoa or a series of piers in front of these rooms. A typical example for the combination with a stoa is the retaining system between the upper and the middle terrace of the gymnasium in Pergamon which was built under the reign of king Eumenes II (Radt 1999: 114-118, fig. 68). The best example for a Hellenistic retaining system with piers is the southern terrace wall of the Demeter sanctuary which is also dated to the era of Eumenes II (Radt 1999: 180, fig. 126). Nonetheless, in Roman imperial Pergamon vaulting

systems were used for the support of the terraces. The most prominent example is the terrace of the Temple of Trajan (Radt 1999: 212-214). The terrace of the Lower Agora was made by constructing two retaining systems both in the north and in the south of the central court (figs. 2-3). Both systems are similar to the upper terrace of the Hellenistic gymnasium: In the north, there was a series of small and irregular rooms and on the southern side of these rooms, a peristasis existed. On the level of the central court there was a stoa in front of the south face of the peristasis. On the southern side of the Agora, a peristasis was built in first place. To the southern face of this peristasis, a stoa was erected on the lower level of the street running along this side of the building (figs. 2-3). Both of these retaining systems bear structural similarities to the northern part of the middle terrace of the Pergamene gymnasium. Therefore, the retaining wall technology of the Lower Agora belongs to the Hellenistic building technology for terracing. Whereas the method of terracing c­ learly points to technological ideas of earlier centuries other constructional details illustrate that the Agora was different from its Hellenistic predecessors. This is specifically valid for the construction of the entablature. It has been mentioned already that the entablature of the porticoes of the Lower Agora was made of one single block with a recess for the insertion of a wooden beam on its rear side. This construction differs much from the one that was regularly used in the 2nd century BC when architrave and triglyph frieze were generally made out of two different blocks (fig. 6). The reasons that led to the unique solution used for the Agora can hardly be guessed. Presumably the use of wood for some of the constructive parts was less expensive than the use of stone (see above). BUILDING PROCESS The original walls of the Lower Agora were built out of ashlars with a mud-filling. Clamps and dowels were used only for joining the architectural members of the colonnades. The columns show a system of usually one dowel that was placed centrally in case of the smaller columns of the central court (ca. 52 cm diam.). Two dowels were used for the bigger drums of the outer façade on the southern side (approx. 60 cm diam., i.e. 2 feet). Pour channels indicate the use of lead.

Anejos de AEspA LXXVII

Clamps were used for connecting blocks on the horizontal level. Hellenistic clamps are long and narrow. For instance, π-shaped iron clamps were used at the extraurban sanctuary of Mamurt Kale in the early Hellenistic period (Bachmann 2011: 79, 468 Kat. 3.50). Here the clamps were 18.5 cm long, 2.2 cm wide and 3.4 cm high. In comparison, the π-shaped clamps used for the Temple of Trajan were wider and higher (Stiller 1895: 21-22, 27-28). At the Lower Agora, architraves and cornices were connected with π-shaped clamps, as well. They are averagely 18 cm long, 2 cm wide and 3 cm high. These narrow and long clamps correspond to those used for Hellenistic buildings. With regard to the building process the system of dowel holes used for the entablature is especially interesting. In this case, the dowel holes were placed on the bottom edge of the upper block (architrave, and geison blocks respectively) leaving an opening on the vertical side. Pry holes that are regularly placed next to the dowel holes of the corresponding lower course indicate that the upper blocks were pushed into their final ­po­sition rather than being inserted from above (fig. 7). Furthermore, the placement of the dowel holes on the upper blocks alternates between ­either the left or the right side of each block. A similar system has been observed e.g. at the cella walls of the Parthenon in Athens (Orlandos 1966: 116-118 fig. 134). In this case the alternation of the position of the dowel holes is the result of the construction process of the wall. While the first course of ashlars was laid starting from the eastern side, the next course started from the western side so alternating. In the case of the Lower ­Agora, however, the blocks in question do belong to the same course, i.e. the geison zone above the architrave. Therefore, with regard to the organisation of the work the observation of alternating dowel holes indicates that at least two teams were working parallel at the same stage of the building process, i.e. the installation of the geison blocks. Presumably these two teams were starting their work from the same corner of the building for example in the north-western angle. In this case, one group of workmen would have been working on the northern colonnade advancing eastwards while the second group would have been working contemporaneously on the western colonnade progressing southwards. With regard to the two-storeyed architecture of the building, the use of lifting machines poses another major question about the organisation of

The ‘Lower Agora’ of Pergamon...

309

the building site. At the Lower Agora, the lifting method with lewises is only occasionally obser­ ved; only Doric cornices as well as some of the double half-columns of the upper floor show central lewis holes on their top (fig. 8). In the case of the cornice blocks three different systems of lifting can be deduced. Most common among both entablature blocks as well as column drums is a type of long narrow lewis holes. On average this type measures 11 × 1.5 cm approximately and has a double-slanted lewis socket. The same type of lewis holes can be observed for example on a leaf capital from the inner order of the North portico of the Athena temenos from the era of Eumenes II (Aylward 2009: 315-317, fig. 6). Therefore, it seems that this type follows a Hellenistic tradition. A second type of lewis holes has distincti­ vely larger dimensions of 13  ×  4-5 cm. Finally, some pieces of the same series of blocks do not bear lewis holes at all (fig. 8). One fragment of a single architrave block that was recut into a geison might indicate that the larger version of the lewis hole was not part of the original building process but rather used during a repair at a later time. With two different systems remaining it seems possible that the two alternative ways of lifting were used contemporaneously for the construction of the colonnade. This observation might again indicate that two (or more) teams were working parallel on the site. Whereas one team used a crane with a lewis machine, the second team might have lifted the blocks by simply putting a rope around them. In both cases the capacity of these lifting machines can be calculated to approximately 900 kg at the least on the basis of the architrave blocks. Due to the larger number of extant pieces, the columns of the upper storey are significant for the question of lifting as well. Each column of the upper storey was composed of two or three drums (Dörpfeld 1902: 20-22). The bottom drums are easily discernible due to their unfluted rear side. With regard to the question of lifting, two corresponding features can be observed (fig. 10). On the one hand, the lower drums are much taller with a height ranging between 1.70 and 1.88 m. In comparison, the majority of the upper drums with a fluted rear side range in size between 1.02 and 1.25 m. Two very short pieces have a height of only 0.75 m or 0.51 m respectively. On the other hand it seems striking that only the upper drums bear lewis holes. Even though the state of preservation does not provide clear evidence for

310

Anejos de AEspA LXXVII

BURKHARD EMME, ARZU ÖZTÜRK

Fig. 8. Pergamon, Lower Agora. Architrave blocks showing different types of lewis holes. Left to right: no lewis hole; small type; large type. Above right: Column drum of the upper storey, small type (not to scale).

total

With lewis hole

No lewis hole

Lewis hole unclear

Height

Lower drums

 7

0

4

3

1.70-1.88 m

Uppers drums

13

6

2

5

0.51-1.25 m

Placement unclear (fragments)

 3

0

0

3

Unclear

Fig. 9. Technical details of column drums of the upper storey.

all pieces, none of the four pieces that are sufficiently well preserved has a lewis hole (fig. 9). ­According to their bigger dimensions as well as their position within the overall construction, one might deduce that the installation of the lower drums was done by erection rather than by actually lifting these pieces (fig. 10). In turn, this observation indicates that the architectural pieces of the upper storey were not lifted from the ground of the central court. Contrarily, it seems plausible

that the large drums were delivered directly to the upper storey by carts or similar devices using a direct connection to the main road like a ramp (fig. 1). A lifting machine that was used to move the smaller drums might then have been placed directly on the upper floor. Again, it seems plausible that at least two groups of workmen were engaged in the construction of the upper storey’s colonnade. While one group could erect the bottom drums a second group might have been con-

Anejos de AEspA LXXVII

The ‘Lower Agora’ of Pergamon...

Fig. 10. Pergamon, Lower Agora. Reconstruction of building process of upper storey (based on: Dörpfeld 1902: fig. 00).

Fig. 11. Pergamon, Lower Agora. Detail of architrave with unfinished triglyph on the edge.

311

312

BURKHARD EMME, ARZU ÖZTÜRK

cerned with lifting the second (and eventually the third) drum on top of it. This organisation might have had two advantages: firstly it would have helped speed up the building process; secondly it would have minimized the weight of the blocks that had to be lifted by a lifting machine. Based on the dimension of the largest drum with a lewis hole (1.25 m) the capacity of this machine can be calculated to approximately 1.2 tons. This is not much more than the calculated weight of an architrave block (see above). Due to these rather modest dimensions it is obvious that the lifting machines used for the construction of the Lower Agora must have been much smaller than those used for lifting the monolithic columns and similar architectural pieces in the Temple of Trajan, the Asklepieion, the Red Hall and other constructions of Roman imperial Pergamon. As has been pointed out already it is likely that the material for architectural members of pink andesite was quarried away from the building spot. A further investigation of the architectural ornamentation indicates that the same is true for the actual process of carving the architrave blocks including the architectural ornamentation. In this case, the usual treatment of the triglyphs on the edges was left unfinished (fig. 11). This observation implies that the architectural decoration was carved mainly on the ground with only marginal parts left for execution after the blocks had been put into their final position. However, this final treatment was not executed. Also, the direction of lead channels in the upper storey makes clear that the lead was filled in generally from the rear side of the blocks, i.e. from the inside of the gallery. Both details imply that no scaffolding was used for the construction of the façade. Instead the floor of the upper storey was used as a working platform during the building process. It was here that a lifting machine would have been placed rather than on the ground floor of the central court. CONCLUSIONS Due to the results of recent excavations the Lower Agora of Pergamon can be dated to the late 1st century BC/early 1st century AD. In spite of this dating the architecture of the Lower ­Agora has many resemblances with earlier buildings from the era of the Attalid kingdom. Many fea-

Anejos de AEspA LXXVII

tures of its building technology indicate that arch­ itects and workmen were following the local Hellenistic tradition. This is true for the materials used as well as the technical treatment of the arch­itectural pieces and the composition of walls and colonnades. Several different observations are possible with regard to the division of labour for all stages of the building process. Working on the site started­with levelling the ground. Contemporaneously the production of the building material started in several different places, as can be deduced from the different distances of their location to the building site. The carving of the architectural members took place on the ground. This demonstrates that there was a differentiation in the working process between finishing the blocks and putting them into their final position. And even this last step seems to have been divided partially among two different groups of workmen as is indicated by the lifting holes of the upper storey columns. In this case it is clear that only the upper drums of the columns were put up by using lifting devices. This might point to two different teams erecting the lower or the upper drums respectively (fig. 10). A close cooperation between different groups of workmen can be deduced also from the construction of the entablature. In this case wood and stone were closely entangled in such a way that workers specialised in both materials must have been working side by side (fig. 7). At first glance, it does not seem surprising that there was a high degree in the division of labour at a building of the size of the Agora. Presumably many of the steps in the working process were executed by rather small teams of workmen. This becomes apparent by the relatively modest size of the architectural pieces. For example, in our own fieldwork we have found that it is possible effectively to move an architrave block with only four workmen by using a prybar and wooden rolls. Also, with regard to the rather limited capacity of lifting machines or to the absence of scaffolding, it seems that efficiency was a major aim of the organisation of the building process. Furthermore, some features might indicate that financial resources were rather limited. In contrast to contemporary buildings such as the house of the Consul Attalos the stone that was used was quarried ‘locally’ within the Pergamene plain. Marble or similarly expensive materials that are a common feature of public architecture of later Roman imperial Pergamon were not used.

Anejos de AEspA LXXVII

The ‘Lower Agora’ of Pergamon...

313

Fig. 12. Pergamon, Lower Agora. Northeastern corner with secondary construction of brick vaults and piers instead of columns. (Courtesy of the German Archaeological Institute: D-DAI-ATH-Pergamon-0208).

Also, the efforts of lifting were reduced by dividing architectural pieces such as columns into several drums. The lifting machines were accordingly modest and presumably small in size. The thickness of the architrave blocks was so small that it was necessary to add a wooden beam on their rear side. The combination of stone and wood alongside other features indicates that the building was constructed by using the first floor as a platform for the following procedures rather than using a separate ephemeral construction like scaffolding. Generally, it seems that the original structures of the Lower Agora have much more in common with Hellenistic predecessors than with public architecture of the following centuries of the ­ ­Roman imperial era. In the original construction no use was made of new materials such as limebased mortar and brick nor of new technologies such as vaults, opus caementicium etc. Since all of these features can be observed in Pergamene architecture of the later 1st and the 2nd century AD, it is clear that the Lower Agora is a late example of a Hellenistic tradition in Pergamene building technology – or rather the last one, in fact. It is therefore very interesting to find that the building was repaired at a later time by extensively using typically ‘Roman’ technologies. Apparently huge

parts of the complex must have become unstable or even collapsed at a certain point, probably in the mid-2nd century AD. The most conspicuous part of the following rebuilding was the installation of a cross-vault in the north-western corner of the porticoes that was constructed out of brick with a backfill of opus caementicium (fig. 12). In addition a set of piers was built in order to stabilise the back wall as well as the colonnade of the northern portico (fig. 2: hatched parts). These piers were made of small ashlars with a filling of mortar and gravel. Even though it is not possible to date these measures exactly, a date in the second half of the 2nd century AD seems likely. In any case, it is clear that the period between the original construction of the Lower Agora and its extensive rebuilding must have been a time of fund­amental change in Pergamene architecture and building technology. Hopefully, more research in the development of Pergamene architecture during the early Roman period will provide us with a better understanding of the process that led to an adaptation of Roman building technology that is characteristic for many Pergamene buildings of this time. In the case of the Lower Agora, however, it seems remarkable that both architects as well as workshops stuck to t­ raditional

314

BURKHARD EMME, ARZU ÖZTÜRK

Hellenistic building technologies more than three generations after the end of the royal building program in the middle of the 2nd century BC. REFERENCES Aylward, W. 2009: “Lewises in Hellenistic and Roman building at Pergamon”, in Bachmann, M. (ed.), Bautechnik im antiken und vorantiken Kleinasien, pp. 309-322, Byzas 9. Ege Yayınları, Istanbul. Bachmann, M. 2009: “Hellenistische Steinmetzmarken im westlichen Kleinasien”, in Andrassy, P., Budka, J. and Kammerzell, F. (eds.), Non-Textual Marking Systems, Writing and Pseudo Script from Prehistory to Modern Times, pp. 215-231, Lingua Aegyptia – Studia Monographica 8. Seminar für Ägyptologie und Koptologie, Göttingen. Bachmann, M. 2011: “Pergamenische Architektur und Bautechnik”, in Grüßinger, R., Kästner, V. and Scholl, A. (eds.), Pergamon, Panorama der Antiken Metropole, pp. 75-81. Imhof, Petersberg. Bohn, R. 1885: Das Heiligtum der Athena Polias Nikephoros, Altertümer von Pergamon 2. Speman, Berlin. Conze, A. 1912: Stadt und Landschaft, Altertümer von Pergamon 1. Reimer, Berlin. Emme, B. et al. 2014: “Neue Forschungen zur Unteren Agora”, Archäologischer Anzeiger, 2014.2, pp. 122-131. Dörpfeld, W. 1902: “Die Arbeiten zu Pergamon 19001901. Die Bauwerke”, Athenische Mitteilungen, 27, pp. 16-35. Dörpfeld, W. 1904: “Die Arbeiten zu Pergamon ­1902-1903. Die Bauwerke”, Athenische Mitteilungen, 29, pp. 114-116. Dörpfeld, W. 1907: “Die Arbeiten zu Pergamon ­1904-1905. Die Bauwerke”, Athenische Mitteilungen, 32, pp. 167-189. Ganzert, J. 1984: Das Kenotaph für Gaius Cäsar in Limyra. Architektur und Bauornamentik, Istanbuler Forschungen 35. Wasmuth, Tübingen.

Anejos de AEspA LXXVII

Gerkan, A. v. 1959: “Die gegenwärtige Lage der archäologischen Bauforschung in Deutschland”, in Boehringer, E. (ed.), Von antiker Architektur und Topographie. Gesammelte Aufsätze von Armin von Gerkan, pp. 9-13. Boehringer, Stuttgart. Herrmann, P. 1964: “Neue Urkunden zur Geschichte Milets im 2. Jh. v. Chr.”, Istanbuler Mitteilungen, 15, 1965, pp. 71-117. Hoepfner, W. 1996: “The architecture of Pergamon”, in Dreyfuss, R. and Schraudolph, E. (eds.), Pergamon. The Telephos-Frieze from the Great Altar 2, pp. 23-58. University of Texas Press, Austin. Klinkott, M. 1991: “Hellenistische Stützmauerkonstruktionen in Pergamon”, in Hoffman, A. (ed.), Bautechnik der Antike. Internationales Kolloquium in Berlin vom 15.-17. Februar 1990, pp. 131-136. Von Zabern, Mainz. Netzer, E. 2001: Hasmonean and Herodian Palaces at Jericho. Final Report of the 1973-1987 Excavations I. Stratigraphy and Architecture. Israel Exploration Society, Jerusalem. Orlandos, A. K. 1966: Les matériaux de construction et la technique architecturale des anciens grecs. De Boccard, Paris. Pirson, F. 2012: “Pergamon – Bericht über die Arbeiten in der Kampagne 2011”, Archäologischer Anzeiger, 2012.2, pp. 175-274. Rheidt, K. 1991: Die Stadtgrabung II. Die byzantinische Wohnstadt, Altertümer von Pergamon 15, 2. De Gruyter, Berlin-New York. Rumscheid, F. 1994: Untersuchungen zur kleinasiatis­ chen Bauornamentik des Hellenismus. Von Zabern, Mainz. Schrammen, J. 1906: Der Große Altar. Der obere Markt, Altertümer von Pergamon 3, 1. Reimer, ­Berlin. Stiller, H. 1895: Das Traianeum, Altertümer von Pergamon 5, 2. De Gruyter, Berlin. Wulf, U. 1994: “Der Stadtplan von Pergamon”, Istanbuler Mitteilungen, 44, pp. 135-175.

VIII THE ORGANISATION OF CONSTRUCTION IN REPUBLICAN AND PRE-REPUBLICAN ITALY

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

QUANTIFYING MONUMENTALITY IN A TIME OF CRISIS. BUILDING MATERIALS, LABOUR FORCE AND BUILDING COSTS IN LATE REPUBLICAN CENTRAL ITALY DOMINIK MASCHEK Department of Classics, Ancient History and Archaeology School of History and Cultures University of Birmingham

ABSTRACT: The paper tries to go beyond conventional typology or stylistic analysis, looking at the huge Late Republican sanctuaries of Central Italy mainly as sources for economic and social history. Such an evaluation cannot be achieved by just looking at ground plans or available building space. Instead, only the reconstruction of both labour force and building costs leads to a valid estimate of a given building’s importance in its respective historical context. The first results of this approach are discussed for the three building phases of the monumental extraurban sanctuary at Tusculum, modeling both the seasonal demand for resources and manpower and the hypothetical building costs for each phase. Taken together, all these aspects and analytical steps result in a new and highly complex model of the building industry in Central Italy from the late second to the first century BC. KEYWORDS: Monumental sanctuary, Tusculum, Opus caementicium, Euergetism, Slave labour. RESUMEN: El artículo intenta traspasar la tipología convencional o el análisis estilístico, mirando a los grandes santuarios tardo-republicanos del centro de Italia, principalmente como fuentes para la historia económica y social. Esta evaluación no se puede lograr analizando solamente planimetrías o los espacios construidos disponibles. En su lugar, solo la reconstrucción de los costes de la mano de obra y de la construcción conduce a una estimación válida de la importancia de un edificio determinado en su respectivo contexto histórico. Los primeros resultados de esta aproximación se discuten en el caso de las tres fases de construcción del santuario monumental extraurbano en Tusculum, definiendo tanto la demanda temporal de los recursos y la mano de obra como los costes hipotéticos de construcción para cada fase. En su conjunto, todos estos aspectos y etapas analíticas dan como resultado un nuevo y muy complejo modelo de la industria de la construcción en el centro de Italia desde finales del ii siglo a.C. hasta el i a.C. PALABRAS CLAVE: santuario monumental, Tusculum, Opus caementicium, evergetismo, mano de obra de esclavos.

Since the studies of Richard Delbrueck at the beginning of the last century, the monumental Late Republican architecture of Central Italy has periodically been scrutinized by various eminent arch­ aeologists and architects.1 However, the main lines of historical interpretation were already ­ firmly

e­stablished at the very beginning by Delbrueck himself: he saw Late Republican architecture as a creation of Roman capitalism, driven by a high degree of technical rationalisation and, above every­thing else, by innumerable crowds of cheap slaves (Delbrueck 1912: 177-180). This judgement

1   E.g. Delbrueck 1912; Fasolo and Gullini 1953; Kähler 1958; Drerup 1966; Rakob and Heilmeyer 1973; Gros 1976; Rakob 1976; Coarelli 1977; Gros 1978; Lauter 1979; Coarelli

1987; Zevi 1996; Caliò 2003; Wallace-Hadrill 2008: 103-210; D’Alessio 2010; Rous 2010; D’Alessio 2011; La Rocca 2012; Maschek 2014; Stek 2014.

318

Dominik Maschek

Anejos de AEspA LXXVII

Fig. 1: Tusculum, sanctuary, site plan; after Quilici and Quilici Gigli 1995: fig. 2.

is mostly accepted even today, some hundred years after Delbrueck and in spite of the fact that detailed studies on the effort and costs of Late Republican building are still long overdue. In the following I will thus try to sketch out how such approaches can offer us vital new evidence for the socioeconomic and political history of the last generations of the Roman Re­public. It may be indicative to start with a short glance at the bigger picture. In the first half of the second century BC, the erection of monumental buildings, mainly sanctuaries, became a widespread phenomenon in Central Italy. In the second half of the century this building activity boomed, running parallel to both an enormous wave of urbanisation in Latium and Campania, and a general refashioning of urban centres and infrastructure. During this period the great sanctuaries of Gabii, Terracina, Praeneste, Tivoli and Pietrabbondante were built, to name just a few of the most prominent examples. In certain regions of Central Italy, this boom was violently stopped by the Social War. However, in other regions, and especially in Latium, there are not only instances of survival, but also of ongoing investment in construction.2

It may be assumed that this continuity can provide crucial evidence for the economic capacities of Latian communities in the first half of the 1st century BC. Thus, it is exactly such a case of continuous building activity which forms the core of this paper. As a case study, I will analyze a single building with different phases of construction, spanning the time from the second half of the second century BC to the middle of the first century BC. Based on the archaeological remains the use of building materials, labour force and building costs will be estimated, before I conclude with some general thoughts about the building industry in Late Republican Central Italy. The point of  departure and reference is a huge sanctuary ­located close to the city walls of Tusculum in the Alban Hills, just 20 kilometers from Rome (figs. 1-2. The building was published, dated and correctly identified as a sanctuary only in the 1990s by Lorenzo Quilici and Stefania Quilici Gigli, ­after Canina and others had interpreted it as an imperial villa (Quilici and Quilici Gigli 1995). The complex is situated on a slope, facing southwestwards, with an imposing view over the countryside of Tusculum. It was built in three

2   For a detailed chronological summary on the monumental Late Republican sanctuaries in Latium see Rous 2010: 109-119.

For a general comparison of Latium with Samnium, Campania and Rome see Maschek (2016).

Anejos de AEspA LXXVII

Quantifying monumentality in a time of crisis...

319

Fig. 2: Tusculum, sanctuary, groundplan Phases 1-2b; after Quilici and Quilici Gigli 1995:

chronologically distinct phases (figs. 3-5). Building Phase 1 dates to the second half of the second century BC and consists of a long wall in opus incertum, containing a vast rectangular terrace. In the second quarter of the first century BC this structure was enlarged by a Nymphaeum, consisting of several rooms and a cryptoportico; this is building Phase 2a. Briefly afterwards, in building Phase 2b, both the containing wall and the Nymphaeum were integrated into a huge complex of substructures and vaulted rooms, virtually doub­ ling the size of the original terrace.3 Based on the

observation of building materials, dimensions and elevations, the construction effort for each of the three building phases in man-hours and mandays was estimated, based on 19th century building manuals and following the groundbreaking studies of Janet DeLaine (DeLaine 1997; DeLaine 2000; DeLaine 2006). By correlating this estimate with both the available space on the site and the extant evidence for the yearly period used for concrete construction (ca. March/April to October/November), based both on Front. Aq. 122123 and the dipinti in the baths of Trajan (Volpe

3  On the chronology see Quilici and Quilici Gigli 1995: 525-532; Ghini 2002: 196-197; Dupré Raventós and Ribaldi

2004: 216; Gorostidi Pi and Ribaldi 2008: 78-79; Ceccarelli and Marroni 2011: 585-586.

320

Dominik Maschek

Fig. 3: Tusculum, sanctuary, Phase 1 (drawing: D. Maschek).

Fig. 4: Tusculum, sanctuary, Phase 2a (drawing: D. Maschek).

Anejos de AEspA LXXVII

Anejos de AEspA LXXVII

Quantifying monumentality in a time of crisis...

321

Fig. 5: Tusculum, sanctuary, Phase 2b (drawing: D. Maschek).

2002; Volpe and Rossi 2012), the probable ma­ ximum number of workers can be assigned to any  given step of the construction process (see fig. 6).4 In the following, I will focus on the schedule and scope of the different building phases. Due to the monument’s current state of preservation only the extant foundations, substructures, walls and vaults have been taken into account. For the sake of consistency, any hypothetical upper­structures in ashlar masonry were ­excluded. Thus, the following results always have to be understood as an estimate of minimum costs. In Phase 1 (150-50 BC) the main effort was directed towards the erection of the terracing wall and the associated earthworks. It is obvious that the demand for workers was highly dynamic in the two years of construction which can be assigned to the various tasks (see fig. 7). This is especially true for the unskilled workforce. For excavating and terracing you would have needed a peak of fifty unskilled workers at the very beginning of the building process, in April of the first 4   Calculations are mainly based on Pegoretti 1843 (for a critical assessment of Pegoretti’s manual see Barker and Russell 2012; Russell 2013: 30-35, 228-232).

year. This number then declined by more than 50% during the foundation works between April and November in the first year. By contrast, the number of skilled labourers constantly stayed at the same level. In the second year the building of the foundations started again in April, and the numbers rose to the same level as in the previous year. In the final stage of construction, with the building of the terracing wall between May and September, the total numbers climbed even higher; at this stage there was a remarkable increase in the demand for skilled workers, from under ten to over twenty. The second building phase, Phase 2a (75-50 BC), encompassed both the Nymphaeum and a second vast terrace immediately to the west of the platform of Phase 1. With the Nymphaeum a new aspect was added to the building process, namely the vaulting in opus caementicium. Again, we observe a clear peak in the demand for unskilled workers right at the beginning of the first year (see fig. 8). From April to October, this number declined by more than 50% during the foundation works. However, in October the building of the walls started, with the total number of skilled and

322

Anejos de AEspA LXXVII

Dominik Maschek

all values in man-days Excavating and terracing

Phase 1 unskilled

Phase 1 skilled

Phase 2a unskilled

Phase 2a skilled

Phase 2b unskilled

Phase 2b skilled

Digging foundations and throwing out (< 1,6 m)

27,03

0,00

19,07

0,00

323,21

0,00

Digging foundations > 1,6 m

103,62

0,00

70,49

0,00

1.337,13

0,00

Load into baskets

61,41

0,00

45,80

0,00

839,91

0,00

Raising spoil from foundations > 1,6 m

18,42

0,00

13,74

0,00

251,97

0,00

Carry earth over 50 m (medium distance on site)

214,94

0,00

160,31

0,00

2.939,68

0,00

Terracing (additional to removal of spoil) Supervision TOTAL (Excavating & terracing)

1.522,89

0,00

873,80

0,00

0,00

0,00

194,83

0,00

118,32

0,00

569,19

0,00

2.143,15

0,00

1.301,54

0,00

6.261,09

0,00

425,13

0,00

303,50

0,00

3.849,03

0,00

0,00

15,35

0,00

11,45

0,00

209,98

Foundations Slaking lime {3 months in advance) Shoring foundations Lay core for foundations

3.697,46

336,13

2.519,67

229,06

32.507,18

2.955,20

Mixing mortar, foundations

512,77

0,00

366,06

0,00

4.642,47

0,00

Carry materials over 50 m (medium distance on site) and fill

233,96

0,00

167,02

0,00

2.118,23

0,00

Saw timbers (shoring foundations)

0,00

58,60

0,00

65,64

0,00

1.163,31

486,93

41,01

335,63

30,62

4.311,69

432,85

4.931,12

451,10

3.388,38

336,77

43.579,58

4.761,33

Slaking lime (3 months in advance), walls

347,97

0,00

205,40

0,00

4.657,21

0,00

Mixing mortar, walls

534,17

0,00

315,31

0,00

7.149,23

0,00

0,00

43,39

0,00

64,48

0,00

1.021,04

Supervision TOTAL (Foundations) Walls

Saw timbers (scaffolding) Scaffolding, erect

0,00

18,32

0,00

18,72

0,00

402,00

Scaffolding, uprights

0,00

50,51

0,00

75,06

0,00

1.188,57

Walls, bonding courses

0,00

0,00

0,00

0,00

556,93

1.113,86

Lay core for walls

1.184,08

2.368,15

580,37

1.160,74

15.203,09

30.406,17

Lay walls, facing

144,61

289,23

420,97

841,94

12.586,94

25.173,87

Carry materials over 50 m (medium distance on site}, walls

191,50

0,00

113,04

0,00

2.563,00

0,00

Raise materials for walls

65,91

0,00

24,73

0,00

812,40

0,00

5upervision

246,82

276,96

165,98

216,09

4.352,88

5.930,55

2.715,06

3.046,56

1.825,80

2.377,04

47.881,68

65.236,07

0,00

0,00

69,51

0,00

3.068,38

0,00

TOTAL (Walls) Vaults Slaking lime (3 months in advance), vaults Saw timbers (centering)

0,00

0,00

0,00

1,21

0,00

251,05

Prepare and erect centering, vaults

0,00

0,00

0,00

2,02

0,00

418,41

Mixing mortar, vaults

0,00

0,00

106,70

0,00

4.710,24

0,00

Lay core for vaults

0,00

0,00

261,74

523,49

11.601,82

23.203,64

Carry materials over 50 m (medium distance on site), vaults

0,00

0,00

38,25

0,00

1.688,62

0,00

Raise mortar for vaults and fill in

0,00

0,00

12,74

0,00

742,70

0,00

Supervision

0,00

0,00

55,84

53,20

2.488,02

2.496,10

TOTAL (Vaults)

0,00

0,00

475,28

579,91

21.231,40

26.369,20

20.003,78

6.995,32

14.355,00

6.587,42

TOTAL

Fig. 6: Tusculum, sanctuary, labour demand Phases 1-2b (D. Maschek).

244.824,91 192.733,21

Anejos de AEspA LXXVII

Quantifying monumentality in a time of crisis...

323

Fig. 7: Tusculum, sanctuary, hypothetical schedule for Phase 1 (D. Maschek).

Fig. 8: Tusculum, sanctuary, hypothetical schedule for Phase 2a (D. Maschek).

unskilled workers rising vigorously, reaching the level of over thirty and over forty, respectively. These were the masons and their assistants. With the same workforce the building of the walls was continued in April and May of the following year. During the final construction of the vaults, the number declined to under twenty in both the skilled and the unskilled workforce between May and the end of June. The work was finished in July. These results can be put into comparison only by finally looking at the third and by far largest building phase, the monumental substructures of Phase 2b (c. 50 BC). It is absolutely clear that the demand for workers during this project must have been many times higher than in the previous building phases. In the peak periods more than the tenfold number of workers was needed. However, with the proposed schedule also this monumental complex could theoretically have been completed in two years (see fig. 9). Again, there appears the

characteristic peak of unskilled workers at the very beginning of the building project, during the excavation works in April. Starting in September, the building of the walls caused a dramatic increase in the demand for skilled workers which stayed on this level until the end of the building season in November. In the second year, from April to May, the walls could be completed with the same labour force. However, there was a final significant decrease in numbers from June to September, when the vaults were built. In comparison, these three hypothetic schedules are not at all identical, but they share one important aspect: in all three phases, by far the largest number of unskilled workers was required at the beginning of the first year, in a comparatively short period in April. In all three models this huge initial number significantly decreases between April and November. However, the next increase in the demand for unskilled labour shows

324

Dominik Maschek

Anejos de AEspA LXXVII

Fig. 9: Tusculum, sanctuary, hypothetical schedule for Phase 2b (D. Maschek).

a striking difference between the models: in building Phase 1 it reaches its peak in the second year, between May and September. In Phase 2a and Phase 2b, however, the peaks are located in both the first and the second year, in October and Novem­ber and in April and May, respectively. This is extremely important, because it can serve as indirect evidence for the provenience of the unskilled workforce. Based on written sources from early Imperial times, Peter Brunt (Brunt 1980) already came to the conclusion that the big building projects in Rome must have been largely dependent on the seasonal employment of wage labourers.5 The proposed analysis of the three Late Republican building phases of the sanctuary at Tusculum not only points into the same direction, but it also gives a valuable indication of the unskilled workers’ backgrounds. Most importantly, the distribution of the peaks in the demand for unskilled labour can be correlated with information from Roman agricultural calendars about the course of the year on small farms (CIL VI 2305 = ILS 8745; Frayn 1979: 47-48; Goodchild 2007: 298-302, 397-401). The most labour-intensive phase in agriculture began only in May, when the grain fields were cleared; in June, the hay was mown; in July, barley and beans were brought in; and, finally, in August wheat and cereals were harvested. Thus, the largest peak in the demand for unskilled workers in all three models­ is located right before the beginning of the most demanding months in agriculture. It is therefore extremely probable that, between 150 and 50 BC, the unskilled workers for excavation and terracing came mainly from the farms around Tusculum. 5   Cf. Brunt 1966 (contra Casson 1978); Treggiari 1980; Skydsgaard 1983; Martin 1989: 69-71; Anderson 1997: 119127; DeLaine 1997: 197.

In this light we should also reconsider the following, well-known passage in Cicero’s letters to Atticus (Cic. Att. 14.3.1): “ecce autem structores nostri ad frumentum profecti, cum inanes redissent, rumorem adferunt magnum Romae domum ad Antonium frumentum omne portari.” Cicero wrote this letter from Tusculum on 9 April 44 BC, obviously in the context of Marcus Antonius’ preparations for civil war (Havas 1992: 56). The meaning of “ad frumentum” has been either connected with the corn-dole (e.g. Brunt 1966: 16; Brunt 1980: 81 n. 2; Anderson 1997: 35. 125) or with the purchase of wheat (Casson 1978: 47 n.º 17). However, our model for seasonal labour demand, integrating both agriculture and construction, suggests a different reading: Cicero’s structores could well have been away helping in the harvest (Crook 1967: 195; Casson 1978: 47; Gros 1983: 439), e.g. of barley.6 Even more important are the conclusions to be drawn from the distribution of unskilled labour­over the remaining periods of time. In Phase 1 it is absolutely obvious that the second peak in demand lies in the second year, exactly in the most intense period of the agricultural cycle. Thus, we can presume that in Phase 1 the unskilled workers of the second year were not the same people as those of the first year. Due to the  schedule it is even highly doubtful that they were coming from the farmsteads at all. In contrast, the building Phases 2a and 2b show an altogether different distribution, which can be called 6   The harvesting of barley in April would not have been impossible under certain conditions, see Migeotte 2009: 69: “Harvesting took place at the start of the dry season, barley in April and May (at least in warmer regions) and wheat in May and June.” On the importance of seasonal labour in Roman agriculture cf. White 1970: 335, 367-368 and Erdkamp 2008: 424-433. I owe the reference to Cicero to Pauline Ducret.

Anejos de AEspA LXXVII

Quantifying monumentality in a time of crisis...

a­ nticyclical when compared with the agricultural calendar. In both phases, the second largest number of unskilled workers is needed in October and November of the first year, and in April and May of the second year – both periods lay outside the most intensive agricultural season, so the potential of Tusculum’s peasantry could again be exploited. This brings us to an issue of central importance, namely the most probable proportion of free labour and slave labour in the three building phases. Since the first studies of Delbrueck the driving forces behind the boom of the Late Republican sanctuaries have been identified with a mixture of abundant wealth and cheap slaves.7 However, the seasonal peaks in the demand for labour on the construction site of the sanctuary at Tusculum clearly indicate that this predominant interpretation does not match the complexity of historical situations. A slave-owner naturally would have had to calculate the upkeep of his slave workforce for 365 days a year, and despite­ some short periods with a very high employment of workers there is only a very small number of approximately 10% of the total workforce which could have been employed over the whole course of the year in each of the sanctuary’s three building phases. So the crucial questions must be:  how cheap or expensive was the employment of slaves in relation to freeborn workers, and how  big were the respective differences between the three building phases of the sanctuary at Tusculum? The last part of this paper is dedicated to some tentative answers. First of all it is necessary to look at the sources for the progression of wages­ and prices in Italy in the second and first century BC. For building Phase 1, which can be dated to the second half of the second century BC, Cato’s De Agricultura is the most vital source because it was written in the years around 150 BC. The ­daily  cost for a slave is given by Cato at 2 sesterces (Cato Agr. 22.3; Frank 1933: 165-166, 188190. Cf. Cato Agr. 56-60; Frank 1933: 167). Several other snippets of information indicate that 7  Delbrueck 1912: 179-180; Lugli 1957: 363-444; Rakob 1976: 372-373; Coarelli 1977; Boëthius 1978: 136-215; Torelli 1980; Rakob 1983; Torelli 1983: 247-250; Adam 1984: 79-90; Pfanner 1989: 172-174 (with a strong emphasis on rationalisation and specialised production); Anderson 1997: 145-151; D’Alessio 2007; D’Alessio 2010: 52, 54-55. For a thoroughly critical revision of opus caementicium and its socio-political contexts in Late Republican Rome see recently Mogetta 2015.

325

at  the same time free unskilled workers earned between 2.5 and 3 sesterces a day (Frank 1933: 188, 200). Finally, Cato tells us about 4 sesterces as a reasonable daily wage for a skilled craftsman (Cato Agr. 21.5; Kay 2014: 286-287). The subsequent building Phases 2a and 2b have both been dated to the second quarter of the first century BC. For this period we have an information from Cicero’s speech for Sextus Roscius that a slave cost 3 sesterces a day (Cic. Rosc. Am. 28; Frank 1933: 384). Regarding the wages of free-born workers, Philipp Kay has recently shown that we must reckon with a massive rate of inflation for the Late Republic (Kay 2014: 102-104). This also had an impact on the wages for unskilled free workers: in Cato’s days a reasonable wage was ­between 2.5 and 3 sesterces, whereas by the time of Cicero it had climbed to 4 sesterces (Kay 2014: 284). This means an average increase in wages of 60% between 150 and 50 BC. Using this as a reference we can also postulate that the daily wage of a skilled worker must have risen from 4 to 6.5 sesterces over the same period. When we finally integrate these average wages into the calculation, it becomes even more unlikely that the peaks in the hypothetic schedules of construction at Tusculum were filled by huge crowds of slaves (see fig. 10). First of all, there is a significant difference between seasonal free workers and the employment of slaves, mainly due to the fact that a slave requires an upkeep for 365 days a year. If only slaves would have been employed, the labour force for building Phase 1 would have cost 120,000 instead of 40,000 sesterces; the labour force for building Phase 2a would have cost 190,000 instead of 60,000 sesterces; and finally, the labour force for building Phase 2b would have cost 5 million instead of 1.2 million sesterces. Thus, the total reliance on slave labour would have led to costs three or four times higher than those of a seasonal free workforce. However, from the same calculation it also becomes clear that it makes perfect sense to recruit a comparatively small, permanent workforce of approximately 10% from slaves, as well as to fill the largest short-time peaks in labour demand with unskilled free labour, if available. To conclude: The three building phases of the sanctuary at Tusculum provide us with a great amount of most vital information about the process and the social dimensions of construction work in Late Republican Latium. Firstly, we have seen that the terracing of the second century BC

326

Anejos de AEspA LXXVII

Dominik Maschek Workers Workers Workers Workers unskilled skilled unskilled skilled core core seasonal seasonal (slave) (slave) (free) (free)

PHASE 1

PHASE 1

Year 1 - Jan-Apr

900

360

0

0

Year 1 - Jan-Apr

Year 1 -Apr

80

32

960

96

Year 1 -Apr

2.120

848

7.420

2.544

Year 1 - End Apr-Nov

Year 1 - End Apr-Nov

Workers Workers Workers Workers unskilled skilled unskilled skilled core core seasonal seasonal (slave) (slave) (free) (free) 1.125

720

9.540

100

64

848

4.140 368

2.650

1.696

22.472

22.472

Year 1 - Nov-Dec

460

184

0

0

Year 1 - Nov-Dec

575

368

4.876

2.116

Year 2 - Jan-Apr

900

360

0

0

Year 2 - Jan-Apr

1.125

720

9.540

4.140

Year 2 -Apr Year 2 - End Apr-Sept

190

76

665

228

1.430

572

10.368

12.012

Year 2 - Oct-Dec

1.040

416

0

0

TOTAL (HS)

7.120

2.848

19.413

14.880

Year 2 -Apr

44.261

Workers Workers Workers Workers unskilled skilled unskilled skilled seasonal seasonal core core (slave) (free) (free) (slave)

PHASE 2a

238

152

2.014

874

Year 2 - End Apr-Sept

1.788

1.144

15.158

6.578

Year 2 - Oct-Dec

1.300

832

11.024

4.784

TOTAL (HS)

8.900

5.696

75.472

45.472

PHASE 2a

Workers Workers Workers Workers unskilled skilled unskilled skilled core core seasonal seasonal (slave) (slave) (free) (free)

Year 1 - Jan-Apr

810

540

0

0

Year 1 - Jan-Apr

0

0

14.850

Year 1 - Apr

45

30

1.100

98

Year 1 - Apr

60

65

825

480

1.512

1.008

12.096

3.276

2.016

2.184

27.720

16.128

564

611

7.755

4.512

0

0

7.590

4.416

Year 1 - End Apr-Oct Year 1 - Oct-Nov

423

282

7.896

9.776

Year 1 - Dec

414

276

0

0

Year 1 - End Apr-Oct Year 1 - Oct-Nov Year 1 - Dec

8.640

Year 2 - Jan-Apr

810

360

0

0

Year 2 - Jan-Apr

0

0

14.850

8.640

Year 2 - Apr-May

225

150

4.200

5.200

Year 2 - Apr-May

300

325

4.125

2.400

Year 2 - May-End June

456

494

6.270

3.648

0

0

33.495

19.488

3.396

3.679

117.480

68.352

342

228

1.976

3.458

Year 2 - July-Dec

Year 2 - May-End June

1.827

1.827

0

0

TOTAL (HS)

6.408

4.701

27.268

21.808

Year 2 - July-Dec 60.185

Workers Workers Workers Workers unskilled skilled unskilled skilled core core seasonal seasonal (slave) (slave) (free) (free)

PHASE 2b

TOTAL (HS)

PHASE 2b Year 1 - Jan-Apr

Year 1 - Jan-Apr

18.630

2.700

0

0

0

0

217.080

1.449

210

22.512

3.140

Year 1 - Apr

1.932

455

16.884

9.996

Year 1 - End Apr-Sept

32.292

4.680

114.192

49.686

Year 1 - End Apr-Sept

43.056

10.140

376.272

222.768

Year 1 - Sept-Nov

11.799

1.710

68.628

176.358

Year 1 - Sept-Nov

15.732

3.705

137.484

81.396

Year 1 - Dec

9.522

1.380

0

0

Year 1 - Dec

0

0

110.952

65.688

128.520

Year 2 - Jan-Apr

18.630

2.700

0

0

Year 2 - Jan-Apr

0

0

217.080

128.520

Year 2 - Apr-May

14.904

2.160

86.688

222.768

Year 2 - Apr-May

19.872

4.680

173.664

102.816

Year 2 - June-Sept

21.942

3.180

70.808

168.116

Year 2 - June-Sept

29.256

6.890

255.672

151.368

Year 2 - Oct-Dec

18.216

2.640

0

0

Year 2 - Oct-Dec

0

0

212.256

125.664

TOTAL (HS)

147.384

21.360

TOTAL (HS)

192.9071

Workers Workers Workers Workers unskilled skilled unskilled skilled core core seasonal seasonal (slave) (slave) (free) (free)

Year 1 - Apr

362.828 620.068 1.151.640

135.540

109.848

25.870 1.717.344 1.016.736 2.869.798

Fig. 10: Tusculum, sanctuary, hypothetical costs of construction Phases 1-2b (D. Maschek).

was connected to a requirement for seasonal labour­completely different from the great substruc­ tures and vaults of the first century BC. Secondly, specialized slave labour was in many situations actually a poor choice when compared to the employment of unskilled workers from the agricultural sector. The results of archaeological survey in the ager Tusculanus indicate that from the late second century to the Early Empire the agricultural landscape around Tusculum was dominated by huge estates (Valenti 2003, for critical remarks see Launaro 2011: 63 with n.º 36), most probably driven by a considerable servile workforce. In the

framework of our model, this would of course imply that also many of the unskilled workers in April could theoretically have been slaves, although only temporarily leased out by their owners­. Such an arrangement should thus be seen as an alternative way of public spending for Late Republican municipal élites. Thirdly, the high demand for skilled masons in the enormous third building phase could presumably only be satisfied by the effects of vigorous urbanisation in the late second century BC (cf. Gabba 1972; Patterson 2006; De Ligt 2012; Mayer 2012: 22-60). Finally, the scope of finan-

Anejos de AEspA LXXVII

Quantifying monumentality in a time of crisis...

cial investment is simply staggering, chiefly the minimum 2 million sesterces needed for the third building phase, which comes up to five times the census requirement for a Roman knight.8 Taken together, all these aspects lead us to a new and highly complex model of the building industry in Central Italy during the last decades of the second century BC. The urban élites’ ever growing concern for public spending had a massive impact both on the expanding urban communities and on the availability of temporary employment. Thus, the estimates of labour force and building costs not only add to our understanding of socioeconomic processes connected to Late Republican monumental architecture, but also to a proper assessment of monumentality in a time of political crisis. REFERENCES Adam, J.-P. 1984: La construction romaine. Matériaux et techniques. A. et J. Picard, Paris. Anderson, J. C., Jr. 1997: Roman Architecture and Society. John Hopkins University Press, BaltimoreLondon. Barker, S. and Russell, B. 2012: “Labour figures for Roman stone-working. Pitfalls and potential”, in Camporeale, S., Dessales, H. e Pizzo, A. (a cura di), Arqueología de la construcción III. La economía de las obras (Paris, 10-11 de diciembre de 2009), pp. 83‑94, Anejos de Archivo Español de Arqueología 69. CSIC, Madrid-Mérida. Barresi, P. 2003: Province dell’Asia Minore. Costo dei marmi, architettura pubblica e committenza, Studia Arch­ aeologica 125. “L’Erma” di Bretschneider, Roma. Boëthius, A. 1978: Etruscan and Early Roman Architecture. Harmondsworth, New York (2nd ed.). Brunt, P. A. 1966: “The Roman mob”, Past & Present, 35, pp. 3-27. Caliò, L. M. 2003: “La scuola architettonica di Rodi e l’ellenismo italico”, in Quilici, L. and Quilici Gigli, S. (eds.), Santuari e luoghi di culto nell’Italia antica, pp. 53-74, Atlante Tematico di Topografia Antica 12. “L’Erma” di Bretschneider, Roma. Casson, L. 1978: “Unemployment: the building trade, and Suetonius, Vesp. 18”, The Bulletin of the American Society of Papyrologists, 15, pp. 43-51. 8  For the scope of investment in building in Late Republican Central Italy see Gabba 1972; Torelli 1983; Jouffroy 1986: 15-61, 320-398; Pobjoy 2000; Patterson 2006: 125-183; Wallace-Hadrill 2008: 103-143; Rous 2010: 242-250; Stek 2013: 337-350 with bibliography.

327

Ceccarelli, L. and Marroni, E. 2011: Repertorio dei santuari del Lazio, Archaeologica 164. G. Bretschneider, Roma. Coarelli, F. 1977: “Public building in Rome between the second Punic war and Sulla”, Papers of the British School at Rome, 45, pp. 1-23. Coarelli, F. 1987: I santuari del Lazio in età repubblicana. NIS, Roma. Crook, J. 1967: Law and Life of Rome. Cornell University Press, London. D’Alessio, A. 2007: “La diffusione degli impianti a sostruzione cava nell’architettura italica di età tardorepubblicana. Considerazioni su due casi di Pozzuoli e Roma”, in Malacrino, C. G. and Sorbo, E. (eds.), Architetti, architettura e città nel Mediterraneo antico, pp. 217-234. Mondadori, Milano. D’Alessio, A. 2010: “Fascino greco e “attualità” romana: la conquista di una nuova architettura”, in La Rocca, E. and Parisi Presicce, C. (eds.), I giorni di Roma. L’età della conquista, Mostra Roma, Musei Capitolini, marzo 2010-settembre 2010, pp. 49-64. Skira, Milano. D’Alessio, A. 2011: “Spazio, funzioni e paesaggio nei santuari a terrazze italici di età tardo–repubblicana. Note per un approccio sistemico al linguaggio di una grande architettura”, in La Rocca, E. and D’Alessio, A. (eds.), Tradizione e innovazione. L’elaborazione del linguaggio ellenistico nell’architettura romana e italica di età tardo-repubblicana, pp. 51-86, Studi Miscellanei 35. “L’Erma” di Bretschneider, Roma. DeLaine, J. 1997: The Baths of Caracalla. A Study in the Design, Construction, and Economics of Largescale Building Projects in Imperial Rome, Journal of Roman Archaeology Suppl. 25. Journal of Roman Archaeology, Portsmouth, R. I. DeLaine, J. 2000: “Bricks and mortar. Exploring the economics of building techniques at Rome and Ostia­”, in Mattingly, D. J. and Salmon, J. (eds.), Economies beyond Agriculture in the Classical World, pp. 271-296, Leicester-Nottingham Studies in Ancient Society. Routledge, London-New York. DeLaine, J. 2006: “The cost of creation. Technology at the service of construction”, in Lo Cascio, E. (ed.), Innovazione tecnica e progresso economico nel mondo romano, Atti degli incontri capresi di storia dell’economia antica (Capri, 13-16 aprile 2003), pp. 237252, Pragmateiai 10. Edipuglia, Bari. Delbrueck, R. 1912: Hellenistische Bauten in Latium II. Trübner, Straßburg. De Ligt, L. 2012: Peasants, Citizens and Soldiers. Studies in the Demographic History of Roman Italy 225 BC–AD 100. Cambridge University Press, Cambridge.

328

Dominik Maschek

Drerup, H. 1966: “Architektur als Symbol. Zur zeitgenössischen Bewertung der römischen Architektur”, Gymnasium, 73, pp. 181-196. Dupré Raventós, X. and Ribaldi, R. 2004: “Il santuario extraurbano di Tusculum. A proposito dell’intervento di scavo del 1997”, in Religio. Santuari ed  ex voto nel Lazio meridionale. Atti della giornata di studio, pp. 212-223. Comune di Terracina, Terracina. Erdkamp, P. 2008: “Mobility and migration in Italy in the second century B.C.”, in De Ligt, L. and Northwood, S. (eds.), People, Land, and Politics. Demographic Developments and the Transformation of ­Roman Italy 300 B.C. – A.D. 14, pp. 417-449, Mnemosyne Suppl. 303. Brill, Leiden-Boston. Fasolo, F. and Gullini, G. 1953: Il santuario della Fortuna Primigenia a Palestrina. Istituto di Archeologia, Roma. Frank, T. 1933: An Economic Survey of Ancient Rome. Volume 1: Rome and Italy of the Republic. John Hopkins University Press, Baltimore. Frayn, J. M. 1979: Subsistence Farming in Roman Italy. Fontwell, Sussex. Gabba, E. 1972: “Urbanizzazione e rinnovamenti urbanistici nell’Italia centro-meridionale del i sec. a.C.”, Studi Classici e Orientali, 21, pp. 73-112. Ghini, G. 2002: “Il santuario extraurbano di Tusculum”, in Cappelli, G. and Pasquali, S. (eds.), Tusculum. Luigi Canina e la riscoperta di un’antica città, pp. 194-202. Campisano, Roma. Goodchild, H. 2007: Modelling Roman Agricultural Production in the Middle Tiber Valley, Central Italy, Doctoral dissertation. University of Birmingham. Gorostidi Pi, D. and Ribaldi, R. 2008: “Il santuario extraurbano di Tusculum”, in Ghini, G. (ed.), Santuari e luoghi di culto dell’antichità nei castelli romani e prenestini, pp. 72-85. Pescara. Gros, P. 1976: “Les premières générations d’architectes hellénistiques à Rome”, in L’Italie préromaine et la Rome républicaine 1, Mélanges offerts à Jacques Heurgon, pp. 387-410, Publications de l’École française de Rome 27. École française, Rome. Gros, P. 1978: Architecture et société à Rome et en Italie centro-méridionale aux deux derniers siècles de la République, Collection Latomus 156. Latomus, Bruxelles. Gros, P. 1983: “Statut social et rôle culturel des architectes (période hellénistique et augustéenne)”, in Gros, P. (ed.), Architecture et société. De l’archaïsme grec à la fin de la république romaine. Actes du colloque international organisé par le CNRS et l’École française de Rome, Rome 2-4 décembre 1980, pp. 425-452. CNRS-École française, Paris-Rome.

Anejos de AEspA LXXVII

Havas, L. 1992: “Work organization in Cicero’s letters”, Acta Classica Universitatis Scientiarum Debreceniensis, 28, pp. 51-63. Jouffroy, H. 1986: La construction publique en Italie et dans l’Afrique romaine. AECR, Strasbourg. Kähler, H. 1958: “Das Fortunaheiligtum von Palestrina – Praeneste”, Annales Universitatis Saraviensis, 7.3-4, pp. 189-240. Kay, Ph. 2014: Rome’s Economic Revolution. Oxford University Press, Oxford. La Regina, A. 1976: “Il Sannio”, in Zanker, P. (ed.), Hellenismus in Mittelitalien, Kolloquium Göttingen 5.-9. Juni 1974, pp. 219-248. Vandenhoeck und Ruprecht, Göttingen. La Rocca, E. 2012: “La pietrificazione della memoria. I templi in età medio-repubblicana”, in Marroni, E. (ed.), Sacra Nominis Latini. I santuari del Lazio arcaico e repubblicano, Atti del Convegno Internazionale, Palazzo Massimo, 19-21 febbraio 2009, pp. 37-88. Arbor sapientiae, Napoli. Launaro, A. 2011: Peasants and Slaves. The Rural Population of Roman Italy (200 BC to AD 100). Cambridge University Press, Cambridge-New York. Lauter, H. 1979: “Bemerkungen zur späthellenistischen Baukunst in Mittelitalien”, Jahrbuch des Deutschen Archäologischen Instituts, 94, pp. 390459. Lugli, G. 1957: La tecnica edilizia romana con particolare riguardo a Roma e Lazio. Bardi, Roma. Martin, S. D. 1989: The Roman Jurists and the Organization of Private Building in the Late Republic and Early Empire, Collection Latomus 204. Latomus, Bruxelles. Maschek, D. 2014: “Der Tempel neue Kleider? Rezeptionsästhetische und semantische Aspekte von Bauornamentik im spätrepublikanischen Mittelitalien”, in Lipps, J. and Maschek, D. (eds.), Antike Bauornamentik. Grenzen und Möglichkeiten ihrer Erforschung, pp. 181-202, Studien zur antiken Stadt 12. L. Reichert, Wiesbaden. Maschek, D. 2016: “Zwischen Stabilität und Kollaps. Mittelitalische Elitenkultur und die ‚Krise’ der römischen Republik”, in Gilhaus, L. et al. (eds.), Elite und Krise in antiken Gesellschaften, Deutsch-französisches Forschungsatelier für Nachwuchswissenschafter, Bonn 2014, pp. 59-82. F. Steiner-Verlag, Stuttgart. Mayer, E. 2012: The Ancient Middle Classes. Urban Life and Aesthetics in the Roman Empire 100 BCE250 CE. Harvard University Press, Cambridge, Mass.-London. Migeotte, L. 2009: The Economy of the Greek Cities from the Archaic Period to the Early Roman Empire. University of California Press, Berkeley.

Anejos de AEspA LXXVII

Quantifying monumentality in a time of crisis...

Mogetta, M. 2015: “A new date for concrete in Rome”, Journal of Roman Studies, 105, Available on CJO 2015 doi:10.1017/S007543581500043X. Patterson, J. R. 2006: Landscapes and Cities. Rural Settlement and Civic Transformation in Early Imperial Italy. Oxford University Press, Oxford. Pegoretti, G. 1843: Manuale pratico per l’estimazione dei lavori architettonici, stradali, idraulici e di fortificazione. Monti, Milano. Pfanner, M. 1989: “Über das Herstellen von Porträts. Ein Beitrag zu Rationalisierungsmaßnahmen und Produktionsmechanismen von Massenware im späten Hellenismus und in der römischen Kaiserzeit”, Jahrbuch des Deutschen Archäologischen Instituts, 104, pp. 157-257. Pobjoy, M. 2000: “Building inscriptions in Republican Italy. Euergetism, responsibility, and civic virtue”, in Cooley, A. (ed.), The Epigraphic Landscape of Roman Italy, pp. 77-92, Bulletin of the Institute of Classical Studies Suppl. 73. Institute of Classical Studies, London. Quilici, L. and Quilici Gigli, S. 1995: “Un grande santuario fuori la porta occidentale di Tusculum”, in Archeologia laziale, 12. Dodicesimo incontro di studio del Comitato per l’archeologia laziale, pp. 509534, Quaderni di archeologia etrusco-italica 23-24. Consiglio Nazionale delle Ricerche, Roma. Rakob, F. and Heilmeyer, W.-D. 1973: Der Rundtempel am Tiber in Rom. Von Zabern, Mainz. Rakob, F. 1976: “Hellenismus in Mittelitalien. Bautypen und Bautechnik”, in Zanker, P. (ed.), Hellenismus in Mittelitalien, Kolloquium Göttingen 5.-9. Juni 1974, pp. 366-376, Vandenhoeck und Ruprecht, Göttingen. Rakob, F. 1983: “Opus caementicium und die Folgen”, Römische Mitteilungen, 90, pp. 359-372. Rous, B. D. 2010: Triumphs of Compromise. An Analysis of the Monumentalisation of Sanctuaries in Latium­in the Late Republican Period (Second and First Centuries BC), Doctoral dissertation. Universiteit van Amsterdam. Russell, B. 2013: The Economics of the Roman Stone Trade. Oxford University Press, Oxford. Skydsgaard, J. E. 1983: “Public building and society in ancient Rome”, in de Fine Licht, K. (ed.), Città e architettura nella Roma imperiale. Atti del seminario del 27 ottobre 1981 nel 25° anniversario dell’Accademia di Danimarca, pp. 223-227, Analecta Romana Instituti Danici Suppl. 10. Odense University Press, Odense.

329

Stek, T. D. 2013: “Material culture, Italic identities and the romanization of Italy”, in DeRose Evans, J. (ed.), A Companion to the Archaeology of the Roman­ Republic, pp. 337-353. Wiley-Blackwell, Chichester. Stek, T. D. 2014: “Monumental architecture of nonurban cult places in Roman Italy”, in Ulrich, R. B. and Quenemoen, C. K. (eds.), A Companion to Roman Architecture, pp. 228-247. Wiley-Blackwell, Chichester. Torelli, M. 1980: “Innovazioni nelle tecniche edilizie romane tra il i secolo a.C. e il i secolo d.C.”, in Tecnologia, economia e società nel mondo romano, Atti del Convegno di Como (27-29 settembre 1979), pp. 139-162. Banca Popolare Commercio e Industria, Como. Torelli, M. 1983: “Edilizia pubblica in Italia centrale tra guerra sociale ed età augustea: ideologia e classi sociali”, in Cébeillac-Gervasoni, M. (ed.), Les “Bourgeoisies” municipales italiennes aux iie et ier siècles av. J.-C., Centre Jean Bérard, Institut Français de Naples 7-10 décembre 1981, pp. 241-250. CNRSCentre Jean Bérard, Paris-Napoli. Treggiari, S. M. 1980: “Urban labour in Rome: mercennarii and tabernarii”, in Garnsey, P. (ed.), NonSlave Labour in the Greco-Roman World, pp. 48-64, Cambridge Philological Society Suppl. 6. The Cambridge Philological Society, Cambridge. Valenti, M. 2003: Ager Tusculanus. IGM 150 III NEII NO, Forma Italiae 41. Olschki, Firenze. Volpe, R. and Rossi, F. M. 2012: “Nuovi dati sull’esedra sud-ovest delle Terme di Traiano sul Colle Oppio. Percorsi, iscrizioni dipinte e tempi di costruzione”, in Camporeale, S., Dessales, H. e Pizzo, A. (a cura di), Arqueología de la construcción III. La economía de las obras (Paris, 10-11 de diciembre de 2009), pp. 69-81, Anejos de Archivo Español de Arqueología 69. CSIC, Madrid-Mérida. Volpe, R. 2002: “Un antico giornale di cantiere delle Terme di Traiano”, Römische Mitteilungen, 109, pp. 377-394. Wallace-Hadrill, A. 2008: Rome’s Cultural Revolution. Cambridge University Press, Cambridge. White, K. 1970: Roman Farming. Thames and Hudson, London. Zevi, F. 1996: “Le élites municipali, Mario e l’architettura del tempo”, Cahiers du Centre Gustave Glotz, 7, pp. 229-252.

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

NETWORKS AND WORKSHOPS. CONSTRUCTION OF TEMPLES AT THE DAWN OF THE ROMAN REPUBLIC PATRICIA S. LULOF Amsterdam Centre for Ancient Studies and Archaeology, University of Amsterdam

ABSTRACT: Recent research into specific find contexts in Rome and its satellite cities provides the oppor­tunity to trace connections between urban sanctuaries and workshops, and thus to shed light on the organization process of construction and decoration of temples in Central Italy. Especially at the transition from the Regal to the Republican period, when temple building changed rapidly and became both widespread and monumental, networks emerge. Architecture and decoration converged into two very different systems employed at a network of sites in and around Rome, aligned with the end of the so-called First Phase and the rise of the Second Phase in decorative roof systems. New research in the field of technology, such as petrographic analyses and advanced 3D modelling applied as in experimental archaeology, in combination with traditional methods, such as stylistic analysis, iconographic interpretation and the study of ancient building techniques, have shown a pattern of ­affinities between the workshops responsible for the construction of the temples and their terracotta roofs. KEYWORDS: Chaîne opératoire, Sanctuaries, Archaic architecture, Building techniques, Satricum, Central Italy. RESUMEN: La investigación reciente en contextos específicos de Roma y sus ciudades satélite ofrece la oportunidad de trazar algunas conexiones entre santuarios y talleres urbanos, y, en consecuencia, arroja luz sobre la organización del proceso de construcción y decoración de templos en la Italia central. Particularmente, los contactos emergen en la transición entre el período regio y el período republicano, cuando la construcción de templos cambió rápidamente y se convirtió en un hecho generalizado y monumental. Arquitectura y decoración confluyeron en dos sistemas muy diferentes, empleados en una red de yacimientos en Roma y en sus alrededores, entre el final de la llamada Primera Fase y el comienzo de la Segunda Fase de decoración de techos. Nueva investigaciones en el campo de la tecnología, como por ejemplo el análisis petrográfico y el modelado 3D aplicado en la arqueología experimental, en combinación con los métodos tradicionales, como el análisis estilístico, la interpretación iconográfica y el estudio de las antiguas técnicas de construcción, han evidenciado un patrón de afinidades entre los talleres responsables de la construcción de los templos y sus techos de terracota. PALABRAS CLAVE: cadena operativa, santuarios, arquitectura arcaica, técnicas de construcción, Satricum, Italia central.

INTRODUCTION Recent research into specific find contexts in  Rome and its satellite cities at the transition from the Regal to the Republican period ­provides the opportunity to trace connections between

urban sanctuaries and workshops, and thus to shed light on the organization process of construction and decoration of temples in Central Italy. Analysis of the chaîne opératoire may reveal the organization of workshops and the different

332

Anejos de AEspA LXXVII

Patricia S. Lulof

Fig. 1. Exploded view of 3D Model of the temple of Caprifico, Latium.

social groups involved in the building process, completion, maintenance, destruction and rebuilding of temples according to developments in construction technology.1 When the organization of the workshop is unraveled, comparative analysis reveals the origin, diffusion and proliferation of these building techniques. A comparative ­approach enables ascribing innovations and differences in building techniques to the intervention of travelling artisans and architects or outside elite forces, or perhaps to a local invention.2 Stone foundations, columns of wood and stone, stone entablature and wooden roof structures, terracotta roof decoration, in fact, all materials and their function in the building structures, have to be studied in detail in order to argue for a specific reconstruction. This reveals the chaîne opératoire of temples. Re-contextualisation is also particularly useful to investigate when one realizes that the use of religious architecture to express élite identities or power relationships tends to be most prominent during the formative stages of a state or society, in this case Archaic Italy and the rise

of Rome. As the social relationships of power changed through time, so did the physical appearance of the monumental buildings.3 Analysis of the building process of the monuments, with the application of 3D modeling and Virtual Reality (fig. 1), may explain how different technologies played a part in the construction and the transformation of temples at the dawn of the Roman Repub­lic.4 TEMPLE CONSTRUCTION AT THE END OF THE 6TH CENTURY BC: A ROMAN NETWORK SYSTEM There are two successive manifestations in arch­itecture and decorative systems prominent at the dawn of the Roman Republic that I would like to discuss: the temples built at the time of the legendary last king of Rome and the temples that replaced them in the early Roman Republic. Both groups of temples represent a rare phenomenon   Lulof 2014a: 113-125; Hopkins 2012: 111-127.   Frischer 2008: v-xxii; Lulof et al. 2013: 333-337; Lulof 2014b: 28-29. 3

  Audouze 2002: 277-306; Dobres 2010: 35-49. 2   Leroy-Gourhan 1945; Nijboer 1998; Cifani 2010: 124-146. 1

4

Anejos de AEspA LXXVII

Networks and Workshops. Construction of temples...

333

Veü-Roma-VeUetri decorative system

References

  l.  V eii Portonaccio

Winter 2009, Roof 5-2

  2.  Veii Piazza d'Armi

Belelli Marchesini 2011: 283-284

 3. Veii Campetti

Belelli Marchesini 2011 : 283-284

 4. Veii Comunita

Belelli Marchesini 2011 : 283-284

 5. Veii Cardo-Decumano

Belelli Marchesini 2011 : 283-284

  6.  Rome, Capitoline S slope-Giardino Romano-Well

Winter 2009: Roof 5-3; Mura Sommella 2010: 91

  7.  Rome, S. Omobono

Winter 2009: Roof 5-4

  8. Rome, Forum – temple Castori – Regia IV – Lacus Iutumae – Velia

Winter 2009: Roof 5-5, 5.B.3.a, and 5.E.l.a; La Grande Roma 1990: 143

  9.  Rome, Palatine hill SW comer – Casa di Livia

Winter 2009: Roof 5-6; La Grande Roma 1990: 91

10.  Rome, Palatine hill – favissa Casa dei Grifi

Winter 2009: 5.C.3.a, 5.D.l.g; Carlucci 2011 a: 208

11.  Rome, Comitium – Curia Hostilia

Winter 2009: 5.D.l.c, 5.D.2.c; La Grande Roma 1990: 53-54

12.  Velletri SS. Stimmate

Winter 2009: Roof 5-7

13. Praeneste

Gatti 20 1 1 : 246

14. Gabii

Information M. Fabbri

Rome-Caprifico variant decorative system

References

15.  Rome, S. Omobono repair or second roof

Winter 2009: Roof 5-9

16.  Rome, Palatine hill SW comer

Winter 2009: Roof 5-l O, and 5.C.3.a; Carlucci 2011 a: 208

17.  Rome, Forum – Regia/temple Castori

Winter 2009: 5.A.l.b, 5.A.l.e

18. Caprifico

Winter 2009: Roof 5-8

19. Ficana

Winter 2011 a: 219, with previous bibl.

20. Ardea

Subvariant, Ceccarelli 2011 a: 253-254 Fig. 2. Lists of roofs, representing temples (530-520 BC).

of building projects completed in a very short period of time.5 The roofs with a specific decorative system (fig. 2), known as the Veii-Rome-Velletri system, are dated around 530 BC, and consisted of revetment plaques with chariot races, galloping warriors, triumphal processions, seated assemblies and banquet scenes, as well as plain female head antefixes between lion-head spouts. ­Sphinxes ­decorated the corners and the most prominent decorative elements, of course, were the central acroterial statues of Athena and Herakles placed between volutes. The same moulds were used at all three of the sites. Six different temples in Rome carried roofs of the Veii-Rome-Velletri system. Roofs of the same system were exported to Velletri, Praeneste and Gabii.6 The iconography and style of the figurative and decorative details, such as the newly introduced chariot races, reveal the presence of East Greek artisans working in the workshops responsible for the system. Given the close   Lulof 2014a: 113-125; Winter 2011c: 296-302.   Winter 2009: 311-395 (Chapter 5); Winter 2010: 113-122; Winter 2011a: 216-219; Winter 2011b: 289-295; Winter 2013: 907-908; Lulof 2011a: 259-266; Carlucci 2011a: 202-215; Carlucci 2011b: 220-237; Mura Sommella 2011: 197; and most recently, Lulof 2014a: 113-125.

r­ elation between the two cities, it would be safe to suggest that Veii’s workshop with Ionian artisans was the first to develop a new decorative system, which then later moved to Rome, where an expanding artistic market connected with the exploding urbanization attracted workshops from outside of Rome. Perhaps there is some truth in the famous quotation of Pliny that the Tarquins invited Vulca of Veii to Rome to decorate the temple of Jupiter Capitolinus.7 Around 520 BC, the Rome-Caprifico variant was introduced. It appears four times in Rome, and was probably produced by the same workshop responsible for the Veii-Rome-Velletri system accord­ing to the results of recent research.8 The same enterprise exported roof terracottas of this variant to Ficana, Caprifico/Cori, and Ardea. Clays and inclusions are exactly the same as used in the previous group, as is the use of the same formula for mixing clays and inclusions. It used slightly different moulds for the figurative reliefs, but the morphology was exactly the same as in the

5 6

7  Pliny, NH, 12.151-157, 28.2, 33.16; Plutarch, Public. 13; Lulof 2000: 207-219. 8   Winter 2009: 311-395 (Chapter 5); Winter 2010: 113-122; Winter 2011a: 216-219; Winter 2011b: 289-295; Winter 2013: 907-908; and most recently, Lulof 2014a: 113-125.

334

Patricia S. Lulof

Anejos de AEspA LXXVII

Fig. 3. Map showing sites with roofs of the Roman Networks decorative system.

decorative system of its predecessor, as are the modules and sizes of the terracotta decorative elements and the tiles. This could indicate that one designer or architect was responsible for all of these roof terracottas. The group of temples with this type of roof were built in a very short time span of perhaps 10 years: the first five in Veii, six or seven in Rome alone, and one each in Velletri, Praeneste and Gabii (fig. 3).9 Both systems have an interchangeable format so that the somewhat later Caprifico variant could have been used for repairs, or as an entire replacement roof, for at least two sites in Rome, as well as to decorate perhaps five newly built temples inside and outside of Rome. The roof terracottas were all executed by one ­Roman workshop and artistic and technical centre. The earlier roofs in Veii were perhaps the forerunners and the Veientine workshop the creator of the moulds.10   Gatti 2011: 246, fig. 5.  Winter et al. 2009; Winter 2011b: 289-296; Colonna 2008; Carlucci 2011b: 229-230.  9

10

We know little of the organization of the specialized terracotta workshops, but it seems they worked closely together with the architect or architects responsible for the structure itself. All temples seem to have had the same shallow roof slope with an inclination of 12 degrees. All revetment plaques have nail holes that were created before firing (fig. 4). This indicates that the height of the rafters and horizontal beams were known before the production of the plaques, and that both roof and wooden substructure were adapted to each other. Foundations excavated at Veii, Rome and Velletri have a very similar plan, almost square, measuring c. 12 by c. 14 meters, with the same module found in the ground-plan. Four columns sit in front of the cella, and there are closed walls at the back and the long sides. The commissioning and construction of these temples must have been well-organized, obviously from within Rome. Architects must have been responsible for the construction of the  temples in Rome but must also have been travelling to the sites in question, taking with

Anejos de AEspA LXXVII

Networks and Workshops. Construction of temples...

335

Fig. 4. Revetment plaque of the temple of Caprifico, showing nail holes.

them the ready-made roof terracottas as building kits.11 The Roman Network system, as I would like to name it, was as quickly abandoned as it had begun. Whether these temples were destroyed, around 500 BC, is only in some cases stratigraphically demonstrated (Rome Sant’Omobono, the Roman Forum, Gabii, Velletri). Other sites have not been thoroughly investigated or are still under study. It is, however, certain that the system ceased to be used and that a new decorative system replaced it, with new, monumentalized, architecture emerging everywhere in the region, at exactly the dawn of the Roman Republic.12 TEMPLE CONSTRUCTION AT THE BEGINNING OF THE 5TH CENTURY: THE SECOND PHASE DECORATIVE SYSTEM A greater divide between the previous period and the one now introduced is unimaginable.   Lulof 2006; Lulof 2014a.  On the chronology of the emergence of the second decorative roof system in central Italy: Lulof 2007a: 23-24, following Della Seta 1918 and Andrén 1940 (Second Phase, with previous bibliography); on the architecture: Hopkins 2012: 111-127; Cifani 2014: 15-28. 11

­ rchitecture, plan lay-out, construction techA niques, roof system and roof decoration and its morphology completely changed, and practically all temples in Central Italy had, from then on, the same appearance. Soon after the inauguration of the colossal temple of Jupiter Optimus Maximus in Rome (fig. 5), the new Second Phase decorative roof system in temple building was introduced, more or less simultaneously with the construction of many other “grand” temples, as in the Roman Forum, and in Southern Etruria, as at Veii, Pyrgi and Tarquinia. But especially in Latium, many temples were built or re-built in the Late Archaic period with the Second Phase decorative system, such as at Ardea, Satricum, Lavinium and Lanuvium.13 This decorative system, with floral revetment plaques, full-figure antefixes, high-relief columen and mutulus plaques, and large-scale ridge-pole statues soon spread all over Central Italy, clearly connected to the change towards monumenta­ lity and structured urbanization in Latium Vetus. Based on the choice of subjects in roof decoration, such as the Gigantomachies, Amazons, the mythical Mischwesen like Sirens and Hippocamps, and

12

13  Winter 2009: 4, 493; Winter 2013: 909-910; Carlucci 2006: 2-21; Ceccarelli 2012: 108-119; Lulof 2014a: 119-123; Lanuvium: Lulof 2016.

336

Anejos de AEspA LXXVII

Patricia S. Lulof

Fig. 5. 3D Model of the temple of Jupiter Optimus Maximus, Rome.

the Dionysian overtone in the many Satyrs, Silens and Maenads, it seems obvious that other terracotta workshops than the previously discussed Rome-based workshops, played an important role in the roofing industry around 500 B.C.14 The workshops active in this period show a different artistic and technical input, mostly reflecting trends from Campania and Sicily, like the introduction of columen and mutuli plaques, open pediments, and antefixes in tongue frames.15 This change may provide a clue to understanding the political, cultural and historical setting of the building activities in this period. Ancient sources give us independent evidence that there were West Greek coroplasts active­in Rome after 500 BC: mentioned are Damophilos and Gorgasos.16 A CASE STUDY: SATRICUM, THE LATIN TEMPLE AND CHAÎNE OPÉRATOIRE 3D reconstructions of the temples of Satricum, currently under study at the Netherlands Institute for Advanced Study, have provided us with detailed knowledge of these new building techniques, which were otherwise untraceable. 14   Pairault Massa 1992 and Torelli 1997. New considerations: Strazzulla 2011: 32-44 and Lulof 2012: 427-440. 15   Lulof 1996: 204-206 (on techniques); Lulof 2011. 16  Pliny, NH,, 35.154; Lulof 2007b: 24-25.

The targeted 4D reconstruction forms part of a transparent and annotated model that shows the diachronic change (the fourth dimension) of one of the most important sanctuaries in Latium, dedicated to Mater Matuta.17 Three different temples­were built successively on the same site, using and re-using their foundations and carrying three completely different decorative roof systems, each one representing a different artistic background of the successive architects and terracotta workshops (Etruscan, Campanian and Latin). The sanctuary provides the unique opportunity to trace the changes in building techniques and practices over a long diachronic sequence.18 The Latin temple of Satricum is one of the most complete temples in Latium with a fullfledged Second Phase roof, dated around 500 BC; it belongs to the largest building on the acropolis, Temple II. In total the roof consisted of thirty different classes of roof decoration, among which spectacular ridge-pole statues, giant hippocamps as lateral acroteria, and large pedimental plaques with twenty figures in relief, depicting scenes from the Trojan Saga and Herakles’ labours.19 17  On the Research group at NIAS: www.nias.knaw.nl/ theme-groups/current-theme-groups/biographies-ofbuildings-virtual-futures-for-our-cultural-past 18  Colonna 2005; Knoop and Lulof 2007: 32-42; Lulof 2011a: 23-32. 19   Lulof 2012: 427-440.

Anejos de AEspA LXXVII

Networks and Workshops. Construction of temples...

In the 3D reconstruction of the temple and its  roof, every step of the building process, the ­choices made by those involved and the chaîne opératoire has been checked with modern architects. The reconstruction process and the data collection revealed several important details.20 In this paper I will discuss several construction techniques that show the importance of following chaîne opératoire during the experimental building of the Latin temple of Satricum and what it can say about the origin of workshops and craft communities. Analysis of pollen in the region of Latium showed that the main wood used for construction was oak. Tuff stone quarries have not been found in the direct area, but tufa is well represented in the Alban Hills nearby. Petrographical analysis of the clay used for the terracotta decorative elements has shown it employed the same formula for the mixing of clays and inclusions as the roofs in Rome mentioned earlier. Perhaps we could speculate about the place of manufacture of the roof, as it is now known that finished roofs were transported by sea and by river.21 In re-assembling the temple block-by-block and reconstructing the roof tile-by-tile, the building module was discovered, measuring c. 32 cm, apparent in all decorative elements (tiles, antefixes, revetment plaques, sima blocks, statues and acroteria), but also traceable in the plan of the temple. The ground plan and the foundations show that different types of tufa stone (white and brown) were used; apparently the ground plan of the earlier Temple I was used in the foundations of the larger successor (fig. 6). The ground plan has a peristasis with four by eight columns and a clearly Greek-oriented design with a peculiar adaptation to the lay-out: a slight narrowing towards the front of the building (fig. 7).22 The cella walls were constructed with vertical wooden beams, as holes in the foundations clearly indicate, most probably supporting mudbrick or walls made of rammed earth (pisé). They did not have to carry a lot of weight, although they could have done so.23 The 12 degree roof slope and a 1.5 m overhang have always been assumed. Some kind of extra support must have been necessary, because the roof-corner, with its heavy lateral decoration  Lulof et al. 2013: 333-337 and Lulof 2014b: 28-29.   Lulof 1996: 182-203; Lulof 2006: 235-243. 22   De Waele 1981: 42-45, Foglio 18-20; Colonna 2005. 23   Minke 2012: 52-69.

337

Fig. 6. Tufa blocks from the foundations of the Latin temple of Satricum (Photo Anneke Dekker).

20 21

Fig. 7. Plan of the Latin temple of Satricum.

338

Patricia S. Lulof

and acroteria in terracotta, is structurally unsound. Architects suggested wooden struts to take care of this problem, and in fact, we have found these struts in votive models (fig. 8). Votive models also show there must have been rectangular openings in the cella walls directly under the roof. These openings provided the necessary light inside the cella, and the necessary air circulation for the interior construction of the wooden roof (fig. 9).24 The presence of nail-holes (made before ­firing!) in the revetment plaques indicated the height of the rafters and horizontal beams, which formed a simple truss-system strong enough to carry the weight of the roof, which, in the case of this new system, became even heavier than before, possibly more than 50 tons. Again, the measurements of the wooden beams must have been known during production of the decorative elements (fig. 10).25 The Campanian-Greek influence in structure, manufacture, morphology, iconography, artistic capabilities in the workshops, and perhaps even the architects, is apparent everywhere. The details in construction technique and solutions for difficulties in the roof system all indicate a Campanian or Sicilian background. One last example shows this: the roof had pedimental roofs with tiles and antefixes (i.e. the floor of the pediment is treated as a small roof), which is a Campanian influence, as the earliest examples are to be found in Cumae and Capua. All in all there were five different types of antefixes: four of them, sirens alternated with typhons and satyr-and-maenad pairs, decorating each lateral side of the temple. The helmeted Juno Sospita heads and Silen head antefixes probably decorated the open pediments at the front and the back (fig. 11).26

Anejos de AEspA LXXVII

Fig. 8. Votive model of a temple, showing struts holding the corners of the roof (from Capua, Staccioli 1965).

Fig. 9. Votive model of a temple, showing open space in the cella walls under the roof (from Satricum, Staccioli 1965).

CONCLUSION Digital reconstructions in 3D offer great opportunities for research. Temples, being constructed, destroyed and rebuilt, transform according to the development of increasingly complex and technically demanding chaînes opératoires. They reveal building practices, craft communities and the organization of building sites. Reconstruction and contextualisation is particularly   Staccioli 1965; Klein 1998; Wright 2009.   Lulof 2010: 130-132. 26   Lulof 1999: 183-187; Winter 2006: 45-48; Lulof 2006: 235-243. 24 25

Fig. 10. Revetment plaque of the Latin temple of Satricum, showing nail holes (Photo Archive Soprintendenza dell’Etruria meridionale, Museo nazionale di Villa Giulia, Rome).

Anejos de AEspA LXXVII

Networks and Workshops. Construction of temples...

339

Fig. 11. 3D Model of the façade of the Latin temple of Satricum.

useful to investigate when one realizes that the use of religious architecture to express élite identities or power relationships tends to be most prominent during the formative stages of a state or soci­ ety, in this case the rise of Rome. As the social relationships of power changed through time, so did apparently, the physical appearance of the monumental buildings. Analysis of the building process of the monuments, with the application of petrographical techniques and 3D modeling, may explain how different technologies played a part in the construction and the transformation of temples at the dawn of the Roman Republic. ACKNOWLEDGEMENTS I thank Janet DeLaine for the opportunity to present this paper at the 5th international workshop on the Archaeology of Roman Construction, held in Oxford 11-12 April 2015. I also thank the peer-reviewer for the valuable comments. The presentation and article have been prepared during my fellowship at the Netherlands Institute for Advanced Study, as a member of the Research group Biographies of Buildings. All drawings and models shown in this article have been made by the 4D Research Lab, University of Amsterdam. REFERENCES Audouze, F. 2002: “Leroi-Gourhan, a philosopher of technique and evolution”, Journal of Archaeological Research, 10.4, pp. 277-306. Belelli Marchesini, B. 2011: “Il tetto di Caprifico a confronto: Veii”, in Conti, A. (ed.), Tetti di terracotta: La decorazione architettonica fittile tra Etruria e

Lazio in età arcaica. Atti delle Giornate di Studio (Sapienza, Università di Roma, 25 marzo e 25 ottobre 2010), pp. 278-288, Officina Etruscologia 5. Officina Edizioni, Roma. Belelli Marchesini B. and Ten Kortenaar, S. 2011: “Veio. Considerazioni sulle lastre di rivestimento con fregi figurati”, in Conti, A. (ed.), Tetti di terracotta: La decorazione architettonica fittile tra Etruria e Lazio in età arcaica. Atti delle Giornate di Studio (Sapienza, Università di Roma, 25 marzo e 25 ottobre 2010), pp. 107-115, Officina Etruscologia 5. Officina Edizioni, Roma. Bianchini, M. 2010: Le tecniche edilizie nel mondo antico. Dedalo, Roma. Carlucci, C. 2006: “Osservazioni sulle associazioni e sulla distribuzione delle antefisse di II fase appartenenti ai sistemi decorativi etrusco-laziali”, in ­Edlund-Berry, I. E. M., Greco, G. and Kenfield, J. (eds.), Deliciae Fictiles III. Architectural Terracottas in Ancient Italy: New Discoveries and Interpretations. Proceedings of the International Conference Held at the American Academy in Rome, November 7-8, 2002, pp. 2-21. Oxbow Books, Oxford. Carlucci, C. 2011a: “Roma. Novità dal Colle Palatino”, in Conti, A. (ed.), Tetti di terracotta: La decorazione architettonica fittile tra Etruria e Lazio in età arcaica. Atti delle Giornate di Studio (Sapienza, Università di Roma, 25 marzo e 25 ottobre 2010), pp. 202-215, Officina Etruscologia 5. Officina Edizioni, Roma. Carlucci, C. 2011b: “Il sistema decorativo del tempio delle Stimmate di Velletri tra Veio Portonaccio e Roma Palatino”, in Conti, A. (ed.), Tetti di terracotta: La decorazione architettonica fittile tra Etruria e Lazio in età arcaica. Atti delle Giornate di Studio (Sapienza, Università di Roma, 25 marzo e 25 ottobre 2010), pp. 251-259, Officina Etruscologia 5. Officina Edizioni, Roma.

340

Patricia S. Lulof

Ceccarelli, L. 2011: “Santuario costiero di Ardea: le terrecotte del tempio arcaico”, in Conti, A. (ed.), Tetti di terracotta: La decorazione architettonica fittile tra Etruria e Lazio in età arcaica. Atti delle Giornate di Studio (Sapienza, Università di Roma, 25 marzo e 25 ottobre 2010), pp. 220-237, Officina Etruscologia 5. Officina Edizioni, Roma. Ceccarelli, L. 2012: “Ethnicity and the identity of the Latins. The evidence from sanctuaries between the sixth and the fourth centuries BC”, in Cifani, G. and Stoddart, S. (eds.), Landscape, Ethnicity and Identity in the Archaic Mediterranean Area, pp. 108119. Oxbow Books, Oxford. Cifani, G. 2010: “I grandi cantieri della Roma arcaica: Aspetti tecnici ed organizzativi”, in Camporeale, S., Dessales, H. and Pizzo, A. (eds.), Arqueología de la construccíon II. Los procesos contructivos en el mundo romano: Italia y provincias orientales (Certosa di Pontignano, Siena 13-15 de noviembre de 2008), pp. 35-49, Anejos de Archivo Español de Arqueología 64. CSIC, Madrid-Mérida. Cifani, G. 2014: “Aspects of urbanism and political ideology in archaic Rome”, in Robinson, E. (ed.), Current Approaches to the Archaeology of First Millennium B.C. Italian Urbanism, pp. 15-28. Journal of Roman Archaeology Suppl. 97. Journal of Roman Archaeology, Portsmouth, R. I. Colonna, G. 2005: “Tra architettura e urbanistica. A proposito del tempio di Mater Matuta a Satricum”, in Mols, S. and Moormann, E. (a cura di), Omni pede stare. Saggi architettonici e circumvesuviani in memoriam Jos de Waele, pp. 111-117. Electa, Napoli. Colonna, G. 2008: “L’officina veiente: Vulca e gli altri maestri di statuaria arcaica in terracotta”, in Torelli, M. and Moretti Sgubini, A. M. (eds.), Etruschi. Le antiche metropoli del Lazio. Catalogo della mostra (Palazzo delle Esposizioni, Roma, 21 ottobre 2008-6 dicembre 2009), pp. 53-63. Electa, Napoli. De Waele, J. 1981: “I templi della Mater Matuta di Satricum”, Mededelingen van het Nederlands Instituut te Rome, 43, pp. 7-68. Dobres, M.-A. 2010: “Technology’s links and chaînes: the processual unfolding of technique and technician”, in Preucel, R. W. and Mrozowski, S. A. (eds.),  Contemporary Archaeology in Theory: The New Pragmatism, pp. 124-146. Wiley-Blackwell, ­Chichester-Malden, Mass. Frischer, B. 2008: “From digital illustration to digital heuristics”, in Frischer, B. and Dakouri-Hild, A. (eds.), Beyond Illustration: 2D and 3D Digital Technologies as Tools for Discovery in Archaeology, pp. v-xxii, BAR International Series 1805. Archaeopress, Oxford.

Anejos de AEspA LXXVII

Gatti, S. 2011: “Praeneste. Le terrecotte di “Prima Fase”; problemi e contesti”, in Conti, A. (eds.), Tetti di terracotta: La decorazione architettonica fittile tra Etruria e Lazio in età arcaica. Atti delle Giornate di Studio (Sapienza Università di Roma, 25 marzo e 25 ottobre 2010), pp. 237-250, Officina Etruscologia 5. Officina Edizioni, Roma. Hopkins, J. N. 2012: “The Capitoline temple and the effects of monumentality on Roman temple design­”, in Thomas, M. and Meyers, G. (eds.), The Ideology and Innovation of Monumental Architecture in ­Etruria and Early Rome, pp. 111-127. University of Texas Press, Austin. Klein, N. L. 1998: “Evidence for west Greek Influence on Mainland Greek Roof Construction and the Creation of the Truss in the Archaic Period”, Hesperia, 67.4, pp. 335-374. Knoop, R. R. and Lulof, P. S. 2007: “L’architettura templare” in Gnade, M. (ed.), Satricum. Trenta anni di scavi olandesi. Catalogo della mostra (Le Ferriere 2007-2008), pp. 32-42. Amsterdams Archeologisch Centrum, Amsterdam. Leroy-Gourhan, A. 1945: Milieu et techniques. Albin Michel, Paris. Lulof, P. S. 1996: The Ridge-Pole Statues from the Late­ msterdam. Archaic Temple at Satricum. Thesis, A Lulof, P. S. 1999: “The image of Perseus in archaic roof-decoration in central Italy”, in Docter, R. F. and Moormann, E. M. (eds.), Proceedings XVth Inter­ national Congress of Classical Archaeology (Amsterdam 12-17 July, 1998), pp. 183-187. Allard Pierson Museum, Amsterdam. Lulof, P. S. 2000: “Archaic terracotta acroteria representing Athena and Heracles: Manifestations of Power in Central Italy”, Journal Roman Archaeol­ ogy, 13, pp. 207-219. Lulof, P. S. 2006: “Roofs from the South. Campanian Architectural Terracottas in Satricum”, in EdlundBerry, I. E. M., Greco, G. and Kenfield, J. (eds.) Deliciae Fictiles III. Architectural Terracottas in Ancient Italy: New Discoveries and Interpretations. Proceedings of the International Conference Held at the American Academy in Rome, November 7-8, 2002, pp. 235-243. Oxbow Books, Oxford. Lulof, P. S. 2007a: Architectural Terracottas in the Allard Pierson Museum Amsterdam, Collections ­ of  the Allard Pierson Museum 2. Allard Pierson Museum, Amsterdam. Lulof, P. S. 2007b: “L’Amazzone dell’Esquilino. Una nuova ricostruzione”, Bullettino della Commissione Archeologica Comunale di Roma, 98, pp. 1-25. Lulof, P. S. 2010: “Manufacture and reconstruction”, in Palombi, D. (ed.). Il tempio arcaico di Caprifico di

Anejos de AEspA LXXVII

Networks and Workshops. Construction of temples...

Torrecchia (Cisterna di Latina). I materiali e il contesto, pp. 79-111. Quasar, Roma. Lulof, P. S. 2011a: “The late archaic miracle. Roof decoration in central Italy between 510 and 450 BC”, in Lulof, P. S. and Rescigno, C. (eds.), Deliciae Fictiles IV. Images of Gods, Monsters and Heroes. Proceedings of the Fourth International Conference on Architectural Terracottas from Ancient Italy (Rome, Syracuse, 21-25 October, 2009), pp. 23-32. Oxbow Books, Oxford. Lulof, P. S. 2011b: “Il tempio di Caprifico a confronto. L’immagine ritrovata”, in Conti, A. (ed.), Tetti di terracotta: La decorazione architettonica fittile tra Etruria e Lazio in età arcaica. Atti delle Giornate di Studio (Sapienza, Università di Roma, 25 marzo e 25 ottobre 2010), pp. 267-277, Officina Etruscologia 5. Officina Edizioni, Roma. Lulof, P. S. 2012: “Un miracolo d’immagini. Il tempio cosiddetto tardoarcaico di Satricum”, in Marroni, E. (ed.), Sacra Nominis Latini. Un trentennio  di scoperte nei santuari di area latina tra l’età arcaica e la tarda repubblica. Atti del Convegno (Roma, 19-21 febbraio 2009), pp. 427-440. Loffredo, Napoli. Lulof, P. S. 2014a: “Reconstructing Rome’s golden building age. Temples from the last Tarquin to the Roman Republic (530-480 BC)”, in Robinson, E. (ed.), Current Approaches to the Archaeology of first Millennium B.C. Italian Urbanism, Journal of Roman Archaeology Suppl. 97, pp. 113-125. Journal of Roman Archaeology, Portsmouth, R. I. Lulof, P. S. 2014b: “Archeologia dell’architettura an­ tica. Ricostruire edifici templari da contesti dell’Italia centrale e processi di modellazione 4D”, in L’Archeologia olandese in Italia, Forma Urbis, 19.9 (settembre 2014), pp. 28-29. Lulof, P. S. 2016: “Le terrecotte architettoniche del tempio di Iuno Sospita”, in L’archeologia del sacro e l’archeologia del culto. Sabratha, Ebla, Ardea, Lanuvio. Atti del Convegno (Roma, Accademia Nazionale dei Lincei, ottobre 2013), pp. 221-239. Lulof, P. S., Opgenhaffen, L. and Sepers, M. S. 2013: “The art of reconstruction. Documenting the process of 3d modeling: some preliminary results”, in Addison, A. C., Guidi, G., De Luca, L. and Pescarin, S. (eds.), Proceedings of the 2013 Digital Heritage International Congress (DigitalHeritage), Volume 1, pp. 333-337. IEEE, New York. Minke, G. 2012: Building with Earth. Design and Technology of a Sustainable Architecture. Birkhäuser, Basel-Berlin-Boston. Mura Sommella, A. M 2010: “Esquilino e Campidoglio. Elementi della decorazione architettonica nella

341

Roma dei Tarquini”, Annali della Fondazione per il Museo Claudio Faina, 17, pp. 87-112. Mura Sommella, A. M 2011: “Roma. Le lastre di rivestimento con sfilate di guerrieri e di divinità nel tempio arcaico del Foro Boario”, in Conti, A. (ed.), Tetti di terracotta: La decorazione architettonica fittile tra Etruria e Lazio in età arcaica. Atti delle Giornate di Studio (Sapienza, Università di Roma, 25 marzo e 25 ottobre 2010), pp. 187-202, Officina Etruscologia 5. Officina Edizioni, Roma. Nijboer, A. J. 1998: From Household Production to Workshops. Archaeological Evidence for Economic Transformations, Pre-Monetary Exchange and Urban­ ization in Central Italy from 800 to 400 B.C. Rijks­ universiteit, Groningen. Pairault Massa, F.-H. 1992: Iconologia e politica nell’Italia antica: Roma, Lazio, Etruria del vii al i secolo a.C. Longanesi, Milano. Staccioli, R. A. 1965: Modelli di edifici etrusco-italici. I modelli votivi. Sansoni, Firenze. Strazzulla, M. J. 2011: “Gli altorilievi tardo arcaici tra Roma e Lazio”, in Lulof, P. S. and Rescigno, C. (eds.), Deliciae Fictiles IV. Images of Gods, Monsters and Heroes. Proceedings of the Fourth International Conference on Architectural Terracottas from ­Ancient Italy (Rome, Syracuse, 21-25 October 2009), pp. 3244. Oxbow Books, Oxford. Torelli, M. 1997: Il rango, il rito e l’immagine. Alle origini della rappresentazione storica romana. Electa, Milano. Winter, N. A. 2006: “The origin of the recessed gable in Etruscan architecture”, in Edlund-Berry, I. E. M., Greco, G. and Kenfield, J. (eds.), Deliciae Fictiles III. Architectural Terracottas in Ancient Italy: New Discoveries and Interpretations. Proceedings of the International Conference Held at the American Academy in Rome, November 7-8, 2002, pp. 45-48. Oxbow Books, Oxford. Winter, N. A. 2009: Symbols of Wealth and Power: Architectural Terracotta Decoration in Etruria and Central Italy, 640-510 B.C., Memoirs of the American Academy in Rome Suppl. 9. The University of Michigan Press, Ann Arbor, Mich. Winter, N. A., Iliopoulos, I. and Ammermann, A. J. 2009: “New light on the production of decorated roofs of the 6th c. B.C. at sites in and around Rome”, Journal of Roman Archaeology, 22, pp. 7-23. Winter, N. A. 2010: “The Caprifico roof in its wider context”, in Palombi, D. (ed.) Il tempio arcaico di Caprifico di Torrecchia (Cisterna di Latina). I materiali e il contesto, pp. 113-122. Quasar, Roma. Winter, N. A. 2011a: “Testimonianze di tetti dei sistemi decorativi Veio-Roma-Velletri e ­Roma-Caprifico”,

342

Patricia S. Lulof

in Conti, A. (ed.), Tetti di terracotta: La decora­ zione architettonica fittile tra Etruria e Lazio in età arcaica. Atti delle Giornate di Studio (Sapienza, Università di Roma, 25 marzo e 25 ottobre 2010), pp. 217-219, Officina Etruscologia 5. Officina Edizioni, Roma. Winter, N. A. 2011b: “Analisi petrografiche di campi­ oni da tetti dei sistemi decorativi V ­ eio-Roma-Velletri e Roma-Caprifico”, in Conti, A. (ed.), Tetti di terracotta: La decorazione architettonica fittile tra Etruria e Lazio in età arcaica. Atti delle Giornate di Studio (Sapienza, Università di Roma, 25 marzo e 25 ottobre 2010), pp. 289-295, Officina Etruscologia 5. Officina Edizioni, Roma.

Anejos de AEspA LXXVII

Winter, N. A. 2011c: “Considerazioni conclusive”, in Conti, A. (ed.), Tetti di terracotta: La decorazione architettonica fittile tra Etruria e Lazio in età arcaica. Atti delle Giornate di Studio (Sapienza, Università di Roma, 25 marzo e 25 ottobre 2010), pp. 296302, Officina Etruscologia 5. Officina Edizioni, Roma. Winter, N. A. 2013: “The phenomenon of terracotta: architectural terracottas”, in Mackintosh Turfa, J. (ed.) The Etruscan World, pp. 903-912. Routledge, London-New York. Wright, G. R. H. 2009: Ancient Building Technology. Volume 3: Construction, Technology and Change in History 12.1. Brill, Leiden-Boston.

Arqueología de la construcción, V. Man-made materials, engineering and infrastructure International Workshop, Oxford, April 11-12, 2015

CONCLUSIONS LYNNE C. LANCASTER Ohio University

ABSTRACT: This paper provides an overview of themes represented in both the papers and the posters presented during the conference along with some of the issues raised during the discussions. Topics include archaeometry and material provenances, production and distribution models for bricks and tiles, lifting and moving of materials, the organization of the construction site, responses to mishaps during construction, and the economics of construction. The intention is to synthesize the information presented and to highlight some of the common topics, goals, and methods. It ends by stressing the importance of putting the results of construction analyses into a broad cultural context in order to highlight their relevance to those outside our discipline. KEYWORDS: Mudbrick, Pisé, Tegula, Brick, Tubulus, Pozzolana, Mortar, Brick Stamp, Crane, Transport, Vault, Formwork, Drawing, Truss, Trade. RESUMEN: Estas conclusiones ofrecen una visión general de los temas tratados en las ponencias y los pósteres presentados durante el seminario, así como algunas de las cuestiones planteadas durante los debates. Los temas del congreso incluyen arqueometría, procedencias de materiales, producción y modelos de distribución del ladrillo, procesos de levantamiento y colocación de los materiales, organización de la obra, respuestas a imprevistos durante la ejecución de los trabajos y la economía de la construcción. La intención es la de sintetizar la información presentada y resaltar algunos de los temas comunes, objetivos y métodos. Se destaca, finalmente, la importancia de poner en un contexto cultural amplio los resultados de los análisis sobre la construcción con el objetivo de evidenciar su importancia fuera de nuestra disciplina. PALABRAS CLAVE: adobe, tapial, tegula, ladrillo, tubulus, puzolana, mortero, sello de ladrillo, grúa, transporte, bóveda, encofrado, dibujo, trabajo.

This Fifth International Workshop on the Arch­ aeology of Roman Construction has taken the theme of “Man-made materials, engineering, and infrastructure”. It has been organized somewhat differently from past workshops in that posters were included in order to maximize the number of participants. The result was the densest of the workshops to date, with discussions revolving around twenty talks and thirteen posters over the two days. Some of the posters are published elsewhere;1 however, given that both types of contributions were 1   Some of the papers based on the posters will be published in the journal Arqueología de la Arquitectura, 2016.

central to the discussions that took place, I include all thirty-three contributions in these concluding remarks. In addition to the three main themes in the conference title, a number of sub-themes grew organ­ ically from the discussions so that we had a nice cross-fertilization of topics. The methodologies represented were also wide-ranging from archaeometric studies and stratigraphical excavations to what philologists might call “close reading” of the evidence to actual philological approaches to the work site. In what follows, I have organized my comments according to the sub-themes that developed over the course of the two days.

344

Lynne C. Lancaster

TERMINOLOGY AND CHRONOLOGY First things first – the importance of using correct and precise terminology was emphasized in two of the papers. A. Guus Gazenbeek and Tim Clerbaut examined a peculiar type of tile that is often designated as a tegula mammata in excavation reports. These tiles, however, have projections made of small lumps of clay that are much smaller than true tegulae mammatae, which have large projections at each corner to create hot air spaces along the walls of baths. More significantly, there are often only one or two of the small lumps on these tiles; therefore, they clearly had some other purpose than creating heated walls. Gazenbeek and Clerbaut propose that the lumps were spacers added to guarantee an even flow of heat around the tile during the firing process.2 They rightly point out that these tiles with the small lumps should not be called tegulae mammatae because that term has implications that go beyond a simple description of form – it implies a particular type of bath heating system and is often taken as a diagnostic element to identify the presence of a bath building when the superstructure no longer survives. Discussion during the conference revealed that these tiles are found throughout the west in Britain, Gaul, Germany and Italy. Elizabeth Fentress and Ben Russell also emphasized the importance of terminology in distinguishing between mudbrick and pisé de terre, pointing out that they are different in both material and technique. Mud bricks consist primarily of clay and straw and are individually formed and set out to dry whereas pisé de terre consists of much less clay and more sand and gravel that is rammed between wooden formwork. Distinguishing between the two is important because they imply very different site organization and construction schedule. They stressed the importance of making such distinctions even when the walls are s­ everely deter­ iorated, which is possible with a simple analysis of material content of the remains. In addition to problems of terminology, some issues of dating also arose and provoked questions regarding the socio-political impetus for the introduction of innovations. For example, Fentress and Russell challenged the traditional notion that pisé de terre was a Punic technique that was later adopted in Roman construction. They point 2   For discussions of these types of tiles, see also Brodribb 1987: 60-62; Bouet 1999: 22-28.

Anejos de AEspA LXXVII

to potential examples from Iron Age Italy that predate the examples in North Africa and propose that, in fact, pisé de terre may well have been introduced to North Africa from Italy in place of the more common mud brick. We await the results of their continuing investigations into the use of pisé in Iron Age and Republican Italy. Dating issues also arose regarding the introduction of concrete employing hydraulic mortar containing volcanic ash (pozzolanic mortar). Jacopo Bonetto presented recent excavations at the city walls of Aquileia (181 BC) where scanning electron micro­ scopy (SEM) determined that mortar used in concrete uncovered in the foundation trench contained volcanic ash, probably from central Italy. This is remarkable to find at such an early date and so far away from the volcanic zone of central Italy. One question was whether the ash was transported over land or by sea. Another was whether the concrete could have been a later addition within the foundation trench. The question becomes particularly intriguing when viewed in light of Marcello Mogetta’s poster on the early development of opus incertum in Pompeii and his recently published article on early concrete where he argued that pozzolanic mortar developed later than generally assumed, around the mid second century BC.3 Thus we are left in a quandary as to the early chronological development of pozzolanic concrete, which in turn hampers our ability to understand the innovation in its broader historical context. ARCHAEOMETRY AND MATERIAL PROVENANCE The use of archaeometric analysis of mortars proved useful to determine material provenance. Marie Jackson presented an overview of some of her most recent work on lime mortar employing volcanic ash including the mortar samples from the Roman Maritime Concrete Study (ROMACONS), which were used to follow the long dis3  Mogetta 2015. Bonetto also cited a study identifying pozzolanic mortar in the late third/early second-century BC brick walls of Ravenna (Costa et al. 2000). The researchers found small granules of pozzolana and the presence of augite in the lime mortar between the bricks and note that augite is commonly found in pozzolanic material “soprattutto dell’area laziale”. However, the presence of augite is not sufficient to identify provenance and the amount of pozzolanic material in the mortar is not made clear. The report raises intriguing questions, but further archaeometric work is necessary to clarify these results.

Anejos de AEspA LXXVII

tance trade in volcanic ash for building harbor facilities. In the harbor samples, the results of trace element analysis together with petrographical analysis corresponded with the signatures of products from the Bay of Naples, thus suggesting that Campania was a central distribution point for the trade in volcanic ash; however, not all of the samples were so clearly defined, which also demonstrates the limitations faced in determining provenance of volcanic materials.4 At a more region­al scale, Roberto Bugini and Luisa Folli (poster) dealt with the provenance of the lime for making mortar at Milan, which does not lie in a zone with limestone outcrops. They determined that the presence of magnesite indicated a dolomitic limestone was used (as opposed to more common calcitic limestone), which allowed them to narrow the source to the dolomitic outcrops along the eastern shore of Lake Maggiore. Ed Peveler’s poster demonstrated the use of SEM to identify the origins of the clay used for tile found at Dorchester on Thames, just outside Oxford. He found that some tegulae were transported as far as 50 km to this rural area. Finally, Jennifer Wehby­ Murgatroyd’s study of the mortar from a  group of Hadrianic buildings at Ostia, which have also been studied from a constructional pers­pective by Janet DeLaine,5 demonstrated that the types of volcanic ash employed fell into distinct groups. Further such analyses may allow for distinguishing work groups based both on material choices as well as building techniques.

Conclusions

345

Brick and tile provide valuable evidence for identifying production modes and the movement of materials because they come in various forms, the material can be traced, and they often bear stamps or graffiti that allow for further categorization. They are particularly useful for tracing movements of materials and itinerant workers and for identifying particular workshops whether private, municipal, military, or imperial. In this conference, we have seen how ceramic building

materials (CBM) from Britain, Spain, Italy, and the Levant ranging in date from the sixth century BC to the fourth century AD have been analyzed to provide a broad view of production methods and the methodologies used to study them. Patricia Lulof presented the evidence for the identification of workshops m ­ anufacturing the terr­ acotta roofs of temples in central Italy around the turn of the sixth to the fifth century BC. She employed methods ranging from microscopic pollen and petrographic analyses to iconographic studies to details of tile joining methods to 3D reconstructions of whole structures. She then related these results to the broader socio-political circumstances in central Italy and showed how there was a clear change that reflected the shifting political situation as Rome ousted its last king and became a Republic in 509 BC. In his presentation on the brick and tile supply to London, Ian Betts mentioned one particular workshop from the late first century AD that has been identified via a combination of fabric analysis and the use of a type of roller stamp, which was employed almost exclusively in Roman Britain.6 The products were distributed as far as 105 km from the tilery (likely located near Chichester in Sussex). Betts pointed out that it specialized primarily in bath tiles and has suggested that it was probably involved in the design and construction of the baths themselves.7 Both of these examples from Lulof and Betts demonstrate the ways in which detailed analysis and the combination of methodologies can provide insight into the relationship between design and construction as well as into political changes and the effects they have on the regional economy. The presentations on tile production and distribution also highlighted the various modes of production in different areas and the way in which those changed over time. For example, Phil Mills pointed to evidence for different production modes in different parts of the Empire. Tile production in the Levant tended to be the reserve of specialist centers whereas in Bulgaria and western Turkey there is more evidence for tile production on farms (demonstrated by the hoof prints of farm animals). Mills also noted that clear changes

4   One question that arose in discussion was whether there was any sort of central database where one can find the comparanda on trace element data for the major volcanic systems in the Mediterranean – not really. However, a good start would be the bibliography of Jackson’s contribution to the recent ROMACONS publication: Brandon et al. 2014: 141-187. 5   DeLaine 2002: 41-101.

6   The roller stamps in Britain have been documented by Betts and his colleagues of the Relief-Patterned Tile Research Group: Betts et al. 1997. The one known group of reliefpatterned tiles outside of Britain occurs in Germany: Baatz 1988. 7  Betts et al. 1997: 52. See also Lancaster 2012.

BRICK AND TILE PRODUCTION, ORGANIZATION, AND DISTRIBUTION

346

Anejos de AEspA LXXVII

Lynne C. Lancaster

took place in Britain from the second century, when evidence points to itinerant tile makers, to the third and fourth centuries, when the evidence indicates a move toward specialized tile making centers. Betts emphasized this point as he traced the changes of brick and tile supply in London from the early period when the city was supplied from private producers (like the Sussex workshop) and from procuratorial workshops, the tiles of which were stamped PP.BRI.LON. The procur­ atorial bricks and tiles are linked to a massive public building that began during the last quarter of the first century in London. By the mid second century many of the workshops that had supplied  London were replaced with regional production centers, and reused tile became more common. During the fourth century, stone roof tiles largely replaced the ceramic ones. As with the procuratorial tiles produced for London, official entities were often invested in tile production, both as part of the process of urbanization in the west and as part of the competition for status among the cities of the east. Mills provided a telling example of the way in which tiles could stand as a proxy for prestige. When the Greek city of Tarentum was conferred the status of Roman colony in 89 BC, likely with an influx of new settlers, the charter stated that to be a member of the town council one had to own a dwelling with at least 1500 roof tiles. Presumably the roof tile in this case was used as a measure of both size and wealth to ensure that members were properly invested in their new community. The concern on the part of cities for the supply of tiles is evident from the municipal stamps found in ­cities of both east and west. Mills cited examples from Sagalassus and Corinth. Roof tiles from Corinth are particularly interesting because two stamps were used together: one for the city and a separate one for the worker responsible for the tile.8 Lourdes Roldán and Macarena Bustamante (poster) documented stamps bearing the name of the city Carteia along with bricks bearing its patron god, Hercules. All these municipal stamps raise questions regarding the relationship between the cities and the producers. Did the city contract with private producers to make bricks? Who owned the land from which the clay was taken? Did the city use its own municipal properties and facilities and simply contract for the labor or did the producers provide the clay and facilities?   Broneer 1932: 136-139.

8

Roldán and Bustamante also noted imperial stamps in southern Baetica at Baelo Claudia. In discussions, Bustamante indicated that similar imperial stamps have been found recently at ­Tamuda, site of a Roman military base in northern Morocco. Preliminary analysis of the fabric suggests that they were manufactured locally.9 Given the military nature of the site, the question of the relationship between imperial and military production arises. Both military and imperial production is known elsewhere in northern Morocco near Tangiers.10 Did the military use imperial properties as a source of clay? Were private tile makers contracted to make the tiles on imperial properties? It has long been assumed that the stamped bricks found at Baelo Claudia were shipped across the strait from Morocco.11 We hope that further analysis of bricks from Mor­ occo and Baetica will begin to answer some of these questions. The involvement in tile production of the emperor (or his properties) and the military brings up another question – what was the role of the state in promoting technology transfer within the empire? The military is often assumed to have been the harbinger of advanced technologies in the provinces absorbed into the Empire. However, this is not always the case. For example, the imperial stamps in Morocco and Baetica occur on unusual forms of bath tiles that are only found on either side of the Strait of Gibraltar.12 Thus in spite of the imperial stamps, the t­ ypology is distinctly regional. Craig Harvey (poster) presented a similar phenomenon is his discussion of a particular type of box-tile (tubulus) found only in bath buildings in Nabatea. Unlike the typical box-tiles made with slab construction, these were  made by throwing a cylinder on a wheel and then hand-pressing the wet clay into a roughly shaped rectangle. The method was employed by the early first century AD and continued to be used even in military contexts after the official annexation in 106 AD. Both of these examples from Spain/Morocco and from Nabatea suggest that the military adopted local innovations – either through imitation or though using local producers – rather than imposing external methods of production.   Bernal Casasola et al. 2013.   Gliozzo and Camporeale 2008; Gliozzo et al. 2011. 11   Étienne and Mayet 1971. 12   For the tiles, see Camporeale 2008.  9 10

Anejos de AEspA LXXVII

LIFTING AND TRANSPORTING BUILDING MATERIALS Moving to the building site itself, we heard two presentations that dealt with machines and the expertise required to build and operate them. In both cases, the nature of the evidence was philological, which provided a different and welcome perspective to the conversations. Pauline Ducret examined an excerpt from Cicero’s second speech against Verres in which a very expensive “machina” was employed in the restoration of the columns of the Temple of Castor and Pollux. Ducret explains that this machina was contracted and paid for separately from the other workers (fabri) on the job, who were paid much lower wages. The high price (albeit much too high according to Cicero) paid for the machina and the team to run it raises questions about what specialized know­ ledge would be necessary for such a crew. Duncan Keenan-Jones, Ian Ruffel and Euan McGookin’s discussion of Hero’s automaton and its relationship to lifting machines provided some points of similarity. They compared the terms used by Hero to describe the working parts of the automaton to those used by both Hero and Vitruvius to describe machines used for lifting and transporting stone, which were often the same or closely related­. This raised the issue of theory vs. practice – how much theoretical background did the crane operators have or need to have? We can imagine that the crane workers would have decided on the size of the rope, the number of pulleys to be used in a pulley block, and the diameter of the treadmill, capstan or windless arms necessary to produce the appropriate mechanical advantage. Thus they had to have some sense of the basic theoretical principles at work. Perhaps like the catapult engineers of both the Greek and Roman armies, they had some degree of specialized knowledge that raised their standing (and their earnings) higher than others on the building site. For projects employing large stones, lifting technology can also be a major factor in organizing the building site. Burkhard Emme and Arzu Öztürk presented evidence from the construction of the Lower Agora at Pergamon (mid second century BC) suggesting that measures were taken to minimize the use of lifting machines. For example, the drums at the bottom of the columns were taller than those at the top and only the upper ones had cuttings for lewis irons. In addition, the entablature blocks of the upper floor had

Conclusions

347

channels facing inward for pouring lead into the dowel holes thus suggesting that scaffolding was not used along the outer facade of the structure. They propose that the builders probably used the road rising behind the complex as a type of ramp to deliver stones to the upper level of the site in order to minimize the amount of lifting required. The use of inclined planes in lieu of mechanical lifting devices was also the theme of the paper presented by Giangiacomo Martines, Cinzia Conti and Matthias Bruno, who re-examined how the enormous blocks of the Columns of Trajan and Marcus Aurelius could have been put into place. My own earlier attempt at reconstructing a plaus­ ible method for building Trajan’s Column relied on Hero’s description of lifting towers and pulleys along with comparanda from Domenico Fontana’s description of moving the Vatican obelisk for Pope Sixtus V.13 Alternatively, they suggest that using an inclined plane instead of complicated lifting machines would have reduced the dangerous task of lifting the blocks, especially the higher blocks (the capital at the Column of Marcus Aurelius weighed almost 80 tons). As Mark Wilson Jones asked during the discussions, if we are confronted with two different proposals, how do we decide between them? As someone respons­ ible for one of the proposals, I prefer to let those with more objective perspectives wrestle with this question, but from a methodological perspective it is an important one. Unfortunately, we have no ancient source comparable to Fontana, who so meticulously recorded his process of moving the obelisk. Along with lifting large stones comes the task of transporting materials from quarry to building site. For the lime used at Milan, Bugini and Folli (poster) suggested that the dolomitic limestone taken from the pre-Alps at Lake Maggiore would have been calcined at the quarry site so that the lightweight quick lime could have been more easily transported via river to Milan and other sites in the Po valley that lack immediate access to limestone. Paolo Barresi (poster) gathered archival material for the transport of the large columns used to rebuild San Paolo fuori le Mura after it burned in 1823. The columns, which also came from the area around Lake Maggiore, were transported along rivers until they reached Venice 13   Lancaster 1999; see also the National Geographic video at http://video.nationalgeographic.com/video/magazine/150 315-ngm-building-trajans-column.

348

Lynne C. Lancaster

where they were loaded onto sea going vessels for the long trip down the Adriatic coast of Italy and through the treacherous Strait of Messina and back up to Rome, a trip taking four months. Clearly the use of downstream waterways determined the route to the sea despite the fact that they led to the Adriatic coast. This harks back to Bonetto’s presentation on the central Italian volcanic ash found in the foundations of the city walls at Aquileia – if Barresi’s research is any indication, transport by sea seems more likely than the shorter overland route. Patrizio Pensabene and Javier Domingo (poster) took up the story by dealing with the transport of large columns from the Tiber into the urban fabric of Rome. Once large columns shafts began to be employed (after the Claudius’s construction of Portus), a major limiting factor was the narrow width of many of the streets, given the vast numbers of draft animals to haul them and the wide turning radius required by the long shafts. For access to southern parts of the city the largest blocks were probably off loaded near San Paolo fuori le Mura and brought in through the Via Ostiense as described by Ammianus Marcellinus (17.4.14-15) for the obelisk that Constantius II put up in the Circus Maximus. They point out that transport of large columns would have been largely limited to major roads and processional routes. This brought to mind Keenan-Jones’s quote from Hero on the spectacle of the automaton and reminds us of the spectacle that these large stones would have provided to the people of Rome – a triumphal procession of a different sort. ORGANIZATION OF THE WORKSITE This brings us to the worksite itself. The first task involved preparing the foundations. Depending on the nature of the site, the builder could be faced with complex pre-existing conditions. Caterina Previato demonstrated the approaches taken in the wetlands of northern Italy. The best-known method was simply driving wooden piles into the wet soil, but Previato also illustrated some lesserknown techniques, such as building wooden slabs and sleeper beams, burying terracotta amphoras, and alternating layers of fine soil and coarse gravels. She points to the Greek origins of many of the techniques and suggests that they could have been introduced by Greek workers in northern Italy, evidence of which Bonetto spoke in his pa-

Anejos de AEspA LXXVII

per on the walls at Aquiliea where fired bricks in Greek modules (pentadoron) were used. Alejandra Albuerne (poster) presented complexities encountered by the builders of the Basilica of Maxent­ ius, who were faced with pre-existing structures on the site. The structures, which were not originally intended to support the massive loads of the Basilica, were incorporated into the foundations under the south nave whereas the north nave was built on virgin soil. Evidently, the earlier structures proved inadequate and led to the collapse of the south portion of the structure, probably due to an unrecorded seismic event. Benjamin Clément focused on the remains of mud brick structures at Lyon where excavations have revealed not only the mud brick walls but also the pits on and around the site from which the clay was taken. Extraction from the site itself would at first seem a logical and efficient practice, but as Clément pointed out, the clay had to be moulded into brick and set out to dry for a month and half. This required a vast open area to lay out the bricks – where to put them? As mud brick was the  most commonly used material in domestic arch­ itecture at Lyon for three centuries, this would have been a critical issue. Clément’s point regarding the mud brick highlighted one of the advantages of pisé mentioned by Fentress and Russell: it did not require the vast areas to process the material. As some of the pits were on public land and on other properties, Clément also raised the issue of who controlled where the clay could be mined. Clément’s paper was complemented by other contributions dealing with the organization of the work site. Arnaud Coutelas and David Hourcade presented the results of their team’s work in the substructures of the Baths of Longeas at Chassenon in France. “Close reading” of the vaulted substructures revealed the red painted lines that were used to mark significant points in the construction process. Similar red lines, marks, and letters have been recognized in imperial structures in Rome in recent years,14 but Coutelas and Hourcade showed that the practice was clearly not limited to the capital. They even found the imprints of workers’ clothing in the wall mortar. A careful study of the concrete vaults also revealed the impressions of formwork boards 14   Trajan’s Baths: Volpe and Rossi 2012. Hadrian’s Villa: Attoui 2008; Domus Augustana: personal observation Ulrike Wulf-Rheidt.

Anejos de AEspA LXXVII

that left grain imprints and workers’ marks inscribed on the boards, which ­allowed for the ident­ ification of a reused formwork board. This is unusual evidence verifying the practice of reusing boards, which has long been assumed but rarely proven. Antonio Dell’Acqua added to our understanding of what can  be  gleaned from such detailed observation at the site of the Capitolium at Brescia. The excavators found a graffito representing what is probably a scale used by the workers along with the full scale drawing of the moulding around the main door. Traces of modular grid lines for laying out the complex were also discovered. Similarly, Josef Soucek (poster) showed how the ground plan of the bath at Pollena Trocchia on the north slopes of Vesuvius was laid out with inscribed lines. Elsewhere at the Brescia Capitolium site, Dell’Acqua showed remains of an iron workshop producing tools, inclu­ding two chisels along with flakes from the stone carving occurring in the area. As with the reuse of formwork boards at the Longeas baths, the presence of ironworkers on sites requiring stone carvers is generally assumed, but finding direct evidence for it is less common. DEALING WITH MISHAPS Even with the best-laid plans, things go awry on building sites, and a number of the contributions dealt with such problems. In the paper of Carla Amici and the poster of Oliva Rodríguez and her colleagues, we saw ingenious responses to  reinforcing columns of highly valued colored stone. Amici presented examples of iron bars that had been inserted into holes drilled along the axis of the columns made of giallo antico, porta santa, africano, and most impressively red granite (much harder than marble). The holes, almost 4 cm in diameter and over a meter long in one case, were created using a crown drill, which even left the tell tale remains of the circular groove of the drill cylinder at the bottom of the cutting. The bars were apparently intended to reinforce column shafts that were weakened by faults in the stone, and Amici found them in at least four different sites (Rome, Privernum, Arcinazzo, Carthage). At the Traianeum at Italica, Rodríguez demonstrated similar evidence of reinforcing columns with iron bars, but a different approach was used there. In columns of porta santa and cipollino, circular holes were drilled at an angle

Conclusions

349

­ erpendicular to the grain of the marble rather p than along the axis of the column. The iron bars (sometimes preserved in the holes) were concealed with a patch on the surface. This technique of ­using iron bars as reinforcement, as opposed to repair, has rarely been documented. Moreover, the use of a crown drill on such hard materials demonstrates the skill attained both in crafting the tool and putting it to use.15 In spite of the ­effort required, the use of these techniques was apparently more time and cost effective than obtaining new columns. Two Hadrianic buildings appear to have under­ gone changes due to unforeseen circumstances: the Mausoleum of Hadrian and the Pantheon. Paolo Vitti presented his detective work at the Mausoleum of Hadrian where he was able to identify the uneven settling of the foundations during the construction process. Most interesting, however, was Vitti’s discovery of the remains of a “structural assessment” undertaken by the builders to determine the location and extent of the cracking in the massive foundation platform. The builders had carefully excavated a tunnel in the concrete to follow the crack to its source. Through analysis of the structural changes in the cracked zone, Vitti went on to show how they dealt with the problem. Dorothee Heinzelmann and Michael Heinzelmann presented new research on the bronze truss of the Pantheon portico that may suggest a change of plan in that project as well. The truss was drawn by Borromini before its removal by Pope Urban VIII, so they were able to compare this information with a detailed laser scan made of the front of the intermediate block. Their reconstruction of the truss revealed some unusual design features, which when compared to the structure itself suggest that there was a change in design of the roof at some point. In the discussion afterwards, this was taken up by Mark Wilson Jones, who noted that the truss’s unusual configuration could have resulted from the lowering of the porch roof, which he has argued was originally designed to be taller.16 The Pantheon never ceases to bring new surprises, and this work on its bronze truss takes a step further in drawing out its mysteries and revealing the remarkable nature of its construction. 15   The crown drill was used in medical contexts to saw bone from the fourth century BC: Jackson 2005: 104-108. 16  Wilson Jones has long advocated that the porch was lowered from the original design: Wilson Jones et al. 1987.

350

Anejos de AEspA LXXVII

Lynne C. Lancaster

ECONOMICS OF CONSTRUCTION The economics of construction have become increasingly important in the study of the Roman economy, particularly thanks to the work of our conference host, Janet DeLaine, who introduced the methodology of cost estimation in her study of the Baths of Caracalla.17 Appropriately, we have seen the fruits of her pioneering work in some of the contributions here. Before costs can be estimated, one has to have some idea of the quantities of materials and how they are obtained. Maura Medri and her group (poster) presented the results of their analysis of the brickwork of the Aurelian Walls. In the past there has not been agreement on the degree to which reused bricks were employed, so their project takes a more systematic and quantitative approach to the question in order to establish a basis for further work. Likewise, Mirella Serlorenzi and her group (poster) have begun work at the Domus Tiberiana where they are documenting wall materials and techniques used there in order to develop a better understanding of how the supply of materials affect­ed the organization of the building site in the heart of the city. One of the more unusual contributions was Christina Triantafillou’s poster on the quantification of draft animals required to move materials in and out of the building site. In addition to estimating numbers of animals, she also included estimates of the amount of excrement they produced and compared the results to the potential areas that could be fertilized with their waste. She then proposed that there was likely a sort of symbiotic economic relationship between urban projects employing the animals and rural activities that benefitted from the fertilizer they produced. The study was ingenious in terms of turning a problem into an asset. Dominik Maschek took us into new territory by presenting the results of his quantification of labor required on a project from the Roman Repub­lic, the sanctuary at Tusculum (second half of second century-mid first century BC), which is a welcome addition to previous analyses of imperial projects. His calculations revealed the seasonal fluctuations of unskilled labor required as the project proceeded. As Triantafillou did with her analysis of waste production, Maschek relates his results to the surrounding agricultural activity and postulates a sharing of the labor force b ­ etween   DeLaine 1997.

17

agricultural and constructional activities. Using figures cited in ancient literary sources, he then compared the costs of using slave labor to using non-slave labor, noting that slave labor would have cost three to four times as much as unskilled non-slave labor. He thus challenged the notion that building projects during the late Republic were necessarily the result of large gangs of slaves. The contributions of Maschek and Triantafillou raised new questions regarding the economic relationship between local agricultural production and large construction projects. FINAL THOUGHTS With this conference we have come full circle – Janet DeLaine gave the concluding remarks at the end of the first conference in 2007 and now she is our host for this fifth conference in 2015. So, this seems an opportune moment to reiterate a question she posed in her concluding remarks eight years ago “What are we trying to achieve?”. We gather on these occasions as a group of people intensely interested in materials and the construction process, but I have discovered through my teaching that many do not share our enthus­ iasms. For example, after returning home from this conference, I finished my semester, and at the end I received a nice note from one of the students who had enjoyed my Roman archaeology course. It began: “I was worried when I first signed up that we would just be looking at columns and bricks, and I was worried that it was going to be really boring...” I immediately reflected on all the excitement over columns and bricks I had witnessed at this conference. However, no matter how excited I become over columns and bricks, my students look at me blankly at worst and quizzically at best – they are waiting for me to tell them why they should care about columns and bricks. Nevertheless, this conference reassured me that we are getting past our own enthusiasms and dealing with broader questions. We ended with Patricia Lulof ’s excellent present­ ation on the workshops responsible for the terracotta roofs of the sixth and fifth centuries BC. She presented the material in such a way that the connection between the details and the social and political relevance made for a riveting story, in spite of the fact that most of the attendees were not specialists in this early period. As the humanities in universities in all of our countries

Anejos de AEspA LXXVII

are fighting for survival, we must take care to put our work into the broader context. If we want to continue to receive funding both at universities and in cultural heritage organizations, we must continually remind the wider world why studying the minutia of building construction is a worthy and relevant endeavor. We are involved in one small area of research that is part of a larger quest to understand how human intellect and ingenuity led to cultures that dominated the Mediterranean and beyond. Architectural expression is one of the most visible and enduring means of recording cultural heritage, which is one reason that some groups are now so intent on destroying it.18 Our work helps record and recount powerful stories of the people who contributed to the creation of such impressive cultures, so we must strive to tell those stories when we have the opport­unity. REFERENCES Attoui, R. 2008: “Segni di cantiere nella ‘Palestra’ di Villa Adriana a Tivoli”, in Camporeale, S., Dessales, H. and Pizzo, A. (eds.), Arqueología de la construcción I. Los procesos constructivos en el mundo romano: Italia y provincias occidentales (Mérida, Instituto de Arqueología, 25-26 de octubre de 2007), pp. 33-48, Anejos de Archivo Español de Arqueología 50. CSIC, Madrid-Mérida. Baatz, D. 1988: “Verkleidungsziegel mit Rollstempelmustern aus Südhessen”, Saalburg-Jahrbuch, 44, pp. 65-83. Bernal Casasola, D., Bustamante, M., Díaz, J. J. and Raissouni, B. 2013: “Sellos latericios impe­ riales en el castellum de Tamuda”, Boletín de la ­SECAH, 4, pp. 15-17. Betts, I. M., Black, E. W. and Gower, J. L. 1997: A Corpus of Relief-Patterned Tiles in Roman Britain, Journal of Roman Pottery Studies 7. Oxbow Books, Oxford. Bouet, A. 1999: Les matériaux de construction en terre cuite dans les thermes de la Gaule Narbonnaise, Scripta antiqua 1. Ausonius, Talence. 18   As I write these remarks in May 2015, the militant group ISIS (Islamic State of Iraq and Syria) has just taken over Palmyra after having destroyed antiquities at the sites of Hatra, Nineveh, and Nimrud earlier in the year. Postscript (February 2016): In August 2015, Kahled al-Asaad, archaeologist and head of antiquities at Palmyra, was beheaded for trying to protect the site and the Temple of Bel was destroyed with explosives.

Conclusions

351

Brandon, C. J., Hohlfelder, R. L., Jackson, M. D. and Oleson, J. P. 2014: Building for Eternity. The History and Technology of Roman Concrete Engin­ eering in the Sea. Oxbow Books, Oxford. Brodribb, G. 1987: Roman Brick and Tile. Sutton, Gloucester. Broneer, O. 1932: The Odeum, Corinth vol. 10. Harvard University Press, Cambridge, Mass. Camporeale, S. 2008: “La tipologia dei laterizi”, in Akerraz, A. and Papi, E. (eds.), Sidi Ali ben Ahmed – Thamusida, vol. 1. I contesti, pp. 179-197. Quasar, Roma. Costa, U., Gotti, E. and Tognon, G. 2000: “Nota tecnica: malte prelevate da mura antiche dallo scavo della Banca Popolare di Ravenna”, in Quilici, L. and Quilici Gigli, S. (eds.), Fortificazioni antiche in Italia. Età repubblicana, pp. 25-28, Atlante tematico di topografia antica 9. “L’Erma” di Bretschneider, Roma. DeLaine, J. 1997: The Baths of Caracalla. A Study in the Design, Construction, and Economics of LargeScale Building Projects in Imperial Rome, Journal of Roman Archaeology Suppl. 25. Journal of Roman Archaeology, Portsmouth, R. I. DeLaine, J. 2002: “Building activity in Ostia in the second century AD”, in Bruun, C. and Zevi, A. G. (eds.), Ostia e Portus nelle loro relazioni con Roma. Atti del Convegno dell’Institutum Romanum Finland­ iae (Roma, 3-4 dicembre 1999), pp. 41-101, Acta Instituti Romani Finlandiae 27. Institutum Romanum Finlandiae, Roma. Étienne, R. and Mayet, F. 1971: “Briques de Belo, relations entre la Maurétanie Tingitane et la Bétique au bas-empire”, Mélanges de la Casa de Valázquez, 7, pp. 59-74. Gliozzo, E. and Camporeale, S. 2008: “I laterizi”, in Gliozzo, E., Turbanti Memmi, I., Akerraz, A. and Papi, E. (eds.), Sidi Ali ben Ahmed – Thamusida, vol. 2. L’archeometria, pp. 148-183. Quasar, Rome. Gliozzo, E., Damiani, D., Camporeale, S., ­Memmi, I. and Papi, E. 2011: “Building materials from Thamusida (Rabat, Morocco). A diachronic local production from the Roman to the Islamic period”, Journal of Archaeological Science, 38.5, pp. 10261036. Jackson, R. 2005: “Holding on to health? Bone surgery and instrumentation in the Roman empire”, in King, H. (ed.), Health in Antiquity, pp. 97-119. Routledge, London-New York. Lancaster, L. C. 1999: “Building Trajan’s Column”, American Journal of Archaeology, 103, pp. 419-439. Lancaster, L. C. 2012: “A new vaulting technique for early baths in Sussex. The anatomy of a Romano-

352

Lynne C. Lancaster

British invention”, Journal of Roman Archaeology, 25.1, pp. 419-440. Mogetta, M. 2015: “A new date for concrete in Rome”, Journal of Roman Studies, 105, pp. 1-40. Volpe, R. and Rossi, F. M. 2012: “Nuovi dati sull’esedra sud-ovest delle Terme di Traiano sul Colle Oppio. Percorsi, iscrizioni dipinte e tempi di ­costruzione”, in Camporeale, S., Dessales, H. e Piz-

Anejos de AEspA LXXVII

zo, A. (eds.), Arqueología de la construcción III. La economía de las obras (Paris, 10-11 de diciembre de 2009), pp. 69-81, Anejos de Archivo Español de Arqueología 69. CSIC, Madrid-Mérida. Wilson Jones, M., Davies, P. and Hemsoll, D. 1987: “The Pantheon: triumph of Rome or triumph of compromise?”, Art History, 10, pp. 133‑153.

ARCHIVO ESPAÑOL DE ARQUEOLOGÍA (AEspA) NORMAS PARA LA PRESENTACIÓN DE MANUSCRITOS Dirección Redacción de la Revista: calle Albasanz 26-28, E-28037 Madrid; Teléfono: +34 91 6022300; Fax: +34 913045710; correo electrónico: [email protected] Contenido Archivo Español de Arqueología es una revista científica de periodicidad anual que publica trabajos de Arqueología, con atención a sus fuentes materiales, literarias, epigráficas o numismáticas. Tiene como campo de interés las culturas del ámbito mediterráneo y europeo desde la Protohistoria a la Alta Edad Media, flexiblemente abierto a realidades culturales próximas y tiempos fronterizos. Se divide en dos secciones: Artículos, dentro de los que tendrán cabida tanto reflexiones de carácter general sobre temas concretos como contribuciones más breves sobre novedades en la investigación arqueológica; y Recensiones. Además, edita la serie Anejos de Archivo Español de Arqueología, que publica de forma monográfica libros concernientes a las materias mencionadas. Los trabajos serán originales e inéditos y no estarán aprobados para su edición en otra publicación o revista. Formulario de autoría Al enviar el artículo, los autores deben incluir una declaración específica de que el artículo no se ha sometido a presentación para su evaluación y publicación en otras revistas simultáneamente o con anterioridad. En el momento en que el artículo sea aceptado, al enviar el texto y figuras definitivas, deberán rellenar un formulario específico donde constarán las condiciones de copyright de las publicaciones del CSIC. Normas editoriales 1. El texto estará precedido de una hoja con el título del trabajo y los datos del autor o autores (nombre y apellidos, institución, dirección postal, teléfono, correo electrónico, situación académica) y fecha de entrega. Cada original deberá venir acompañado por la traducción del Titulo al inglés, acompañado de un Resumen y Palabras Claves en español, con los respectivos Summary y Key Words en inglés. De no estar escrito el texto en español, los breves resúmenes y palabras clave vendrán traducidos al español e inglés. Las palabras clave no deben incluir los términos empleados en el título, pues ambos se publican siempre conjuntamente. 2. Se entregará una copia impresa y completa, incluyendo toda la parte gráfica. Se adjuntará asimismo una versión en soporte informático, preferentemente en MS Word para Windows o Mac y en PDF, con imágenes incluidas. 3. El texto no deberá exceder las 11  000 palabras. Sólo en casos excepcionales se admitirán textos más extensos. Los márgenes del trabajo serán los habituales (superior e inferior de 2 cm; izquierdo y derecho de 2,5 cm). El tipo de letra empleado será Times New Roman de 12 puntos a un espacio, con la caja de texto justificada. Aparecerá la paginación correlativa en el ángulo inferior derecho. Se empleará a comienzo de párrafo el sangrado estándar (1,25). Salvo la separación lógica entre diferentes apartados, no se dejarán líneas en blanco entre párrafos. En ningún caso se utilizarán negritas. 4. Se cuidará la exacta ordenación jerárquica de los distintos epígrafes, numerándolos indistintamente mediante guarismos romanos y árabes, e incluso sin numeración. 5. Cuando se empleen citas textuales en el texto o en notas a pie de página se entrecomillarán, evitando la letra cursiva. Dicha letra se acepta para topónimos o nombres en latín. En estos casos, se preferirán las grafías con v en lugar de u, tanto para mayúsculas como para minúsculas (conventus mejor que conuentus). 6 Por lo que se refiere al sistema de cita, deberá emplearse el sistema «americano» de citas en el texto, con nombre de autor en minúscula y no se pondrá coma entre autor y año (apellido o apellidos del autor año: páginas). Si los autores son dos se incluirá la conjunción y entre ambos. Si los autores fueran más de dos se indicará el apellido del primero seguido por la locución et alii. Se incluirá una bibliografía completa al final del trabajo. En la bibliografía final, los títulos de monografías irán en cursiva, mientras que en los artículos el título se colocará entrecomillado. Los nombres de los autores, ordenados alfabéticamente por apellidos, en la bibliografía final irán en letra redonda, seguidos por el año de publicación entre paréntesis y dos puntos. Si los autores son dos, irán unidos por la conjunción “y”. Si son varios los autores, sus nombres vendrán separados por comas, introduciendo la conjunción “y” entre los dos últimos. En el caso de que un mismo autor tenga varias obras, la ordenación se hará por la fecha de publicación, de la más antigua a la más reciente. Si en el mismo año coinciden dos o más obras de un mismo autor o autores, serán distinguidas con letras minúsculas (a, b, c...). En el caso de las monografías se indicará el lugar de edición tal y como aparece citado en la edición original (p. e. London, en lugar de Londres), separado del título de la obra por una coma. En el caso de artículos o contribuciones a obras conjuntas, se indicarán al final las páginas correspondientes, también separadas por comas. Los nombres de revistas se incluirán sin abreviar. Las referencia a las consultas realizadas en línea (Internet), deberán indicar la dirección Web y entre paréntesis la fecha en la que se ha realizado la consulta. Las notas a pie de página, siempre en letra Times New Roman de 10 puntos, se emplearán únicamente para aclaraciones o referencias generales. Ejemplos de citas en la bibliografía final: Monografías: Arce, J. 1982: El último siglo de la España romana: 284-409, Madrid.

Artículos en revistas: García y Bellido, A. 1976: «El ejército romano en Hispania», Archivo Español de Arqueología 49, 59-101. Contribuciones a congresos y obras conjuntas: Noguera Celdrán, J. M. 2000: «Una aproximación a los programas decorativos de las villae béticas. El conjunto escultórico de El Ruedo (Almedinilla, Córdoba)», P. León y T. Nogales (coords.), Actas III Reunión sobre Escultura Romana en Hispania, Madrid, 111-147 Trabajos dentro de una serie monográfica: Alföldy, G. 1973: Flamines Provinciae Hispaniae Citerioris, Anejos Archivo Español de Arqueología VI, Madrid. 6. Toda la documentación gráfica se considerará como Figura (ya sea fotografía, mapa, plano, tabla o cuadro), ordenándola correlativamente. Se debe indicar en el texto el lugar ideal donde se desea que se incluya, con la referencia (Fig. 1), y así sucesivamente. Asimismo debe incluirse un listado de figuras con los pies correspondientes a cada una al final del artículo. El formato de caja de la Revista es de 15 x 21 cm; el de la columna, de 7,1 x 21 cm. La documentación gráfica debe ser de calidad, de modo que su reducción no impida identificar correctamente las leyendas o desdibuje los contornos de la figura. Los dibujos no vendrán enmarcados para poder ganar espacio al ampliarlos. Toda la documentación gráfica se publica en blanco y negro; sin embargo, si se enviara a color, puede salir así en la versión digital. Los dibujos, planos y cualquier tipo de registro (como las monedas o recipientes cerámicos) irán acompañados de escala gráfica, y las fotografías potestativamente. Todo ello debe de prepararse para su publicación ajustada a la caja y de modo que se reduzcan a una escala entera (1/2, 1/3… 1/2000, 1/20000, etc.). En cualquier caso, se puede sugerir el tamaño de publicación de cada figura (a caja, a columna, a 10 cm de anchura, etc.). Las Figuras se deben enviar en soporte digital, preferentemente en fichero de imagen TIFF o JPEG con al menos 300 DPI y con resolución para un tamaño de 16 x 10 cm. No se aceptan dibujos en formato DWG o similar y se debe procurar no enviarlos en CAD a no ser que presenten formatos adecuados para su publicación en imprenta. Aceptación Todos los textos son seleccionados por el Consejo de Redacción según su interés científico y su adaptación a las normas de edición, por riguroso orden de llegada a la Redacción de la Revista, y posteriormente informados por el sistema de doble ciego, según las normas de publicación del CSIC, por al menos dos evaluadores externos al CSIC y a la institución o entidad a la que pertenezca el autor y, tras ello, aceptados definitivamente por el Consejo de Redacción. Correcciones y texto definitivo 1. Una vez aceptado, el Consejo de Redacción podrá sugerir correcciones del original previo (incluso su reducción significativa) y de la parte gráfica, de acuerdo con las normas de edición y las correspondientes evaluaciones. El Consejo de Redacción se compromete a comunicar la aceptación o no del original en un plazo máximo de seis meses. 2. El texto definitivo se deberá entregar cuidadosamente corregido y homologado con las normas de edición de Archivo Español de Arqueología para evitar cambios en las primeras pruebas. El texto, incluyendo resúmenes, palabras clave, bibliografía y pies de figuras, se entregará en CD, así como la parte gráfica digitalizada, acompañado de una copia impresa que incluya las figuras sugiriendo el tamaño al que deben reproducirse las mismas. El texto definitivo se podrá enviar también por correo electrónico. 3. Los autores podrán corregir primeras pruebas, aunque no se admitirá ningún cambio sustancial en el texto. DOI El DOI (Digital Object Identifier) es una secuencia alfanumérica estandarizada que se utiliza para identificar un documento de forma unívoca con el objeto de identificar su localización en Internet. La revista Archivo Español de Arqueología asignará a todos sus artículos un DOI que posibilitará la correcta localización del mismo, así como la indización en las bases de datos de CrossRef. de todas las referencias bibliográficas comprendidas en el volumen de Archivo Español de Arqueología. Varia 1. Entrega de volúmenes: los evaluadores recibirán gratuitamente un ejemplar del volumen en el que hayan intervenido; los autores, el volumen correspondiente y el PDF de su artículo. 2. Devolución de originales: los originales no se devolverán salvo expresa petición del autor. 3. Derechos: la publicación de artículos en las revistas del CSIC no da derecho a remuneración alguna; los derechos de edición son del CSIC. El autor se hará responsable de los derechos de propiedad intelectual del texto y de las figuras. 4. Los originales de la revista Archivo Español de Arqueología, publicados en papel y en versión electrónica, son propiedad del Consejo Superior de Investigaciones Científicas, siendo necesario citar la procedencia en cualquier reproducción parcial o total. Es necesario su permiso para efectuar cualquier reproducción.

ANEJOS DE ARCHIVO ESPAÑOL DE ARQUEOLOGÍA (ÚLTIMOS TÍTULOS PUBLICADOS)

ISSN 0561-3663 LIII R. Ayerbe, T. Barrientos y F. Palma (eds.): El foro de Avgvsta Emerita. Génesis y evolución de sus recintos monumentales VII. Instituto de Arqueología de Mérida. Mérida, 2009. 868 págs. + figs. en texto. – ISBN: 978-84-00-08934-4.

LIV L. Caballero: Las iglesias asturianas de Pravia y Tuñon. Arqueología de la arquitectura. Instituto de Historia. Madrid, 2010. 232 págs. + figs. en el texto. – ISBN: 978-84-0009128-6.

LV T. Tortosa, S. Celestino (eds.) y R. Cazorla (coord.): Debate en torno a la religiosidad protohistórica. Instituto de Arqueología de Mérida. Mérida, 2010. 309 págs. + figs. en texto. – ISBN: 978-84-00-09177-4.

LVI A. Pizzo: Las técnicas constructivas de la arquitectura pública de Augusta Emerita. Instituto de Arqueología de Mérida. Mérida, 2010. 614 págs. + figs. en texto. – ISBN: 978-84-00-09181-1.

LVII G. Camporeale, H. Dessales y A. Pizzo (eds): Arqueología de la construcción II. Los procesos constructivos en el LVIII LIX LX LXI LXII LXIII LXIV LXV LXVI LXVII LXVIII LXIX LXX

LXXI LXXII LXXIII LXXIV LXXV LXXVI

mundo romano: Italia y provincias orientales. Instituto de Arqueología de Mérida. Mérida, 2010. 646 págs. + figs. en texto. – ISBN: 978-84-00-09279-5. M. P. García Bellido, L. Callegarin y A. Jiménez Díez (eds.): Barter, money and coinage in the Ancient Mediterranean (10th-1st centuries BC). Instituto de Historia. Madrid, 2011. 396 págs. + figs. en texto. – ISBN: 978-84-00-09326-6. V. Mayoral y S. Celestino (eds.): Tecnología de información geográfica y análisis arqueológico del territorio. Actas del V Simposio Internacional de Arqueología de Mérida. Instituto de Arqueología de Mérida. Mérida, 2011. 832 págs. + figs. en texto (ed. electrónica) – e-ISBN: 978-84-00-09407-2. J. A. Remolá y J. Acero (eds.): La gestión de los residuos urbanos en Hispania. Xavier Dupré Raventós (1956-2006), in memoriam. Instituto de Arqueología de Mérida. Mérida, 2011. 418 págs. + figs. en texto – ISBN: 978-84-00- 09345-7. L. Caballero, P. Mateos y T. Cordero (eds.): Visigodos y omeyas. El territorio. Instituto de Arqueología de Mérida. Mérida, 2012. 384 págs. + figs. en texto. – ISBN: 978-84-00-09457-7. J. Jiménez Ávila (ed.): Sidereum Ana II. El río Guadiana en el Bronce Final. Instituto de Arqueología de Mérida. Mérida, 2012. 572 págs. + 365 figs. y tablas en el texto. – ISBN: 978-84-00-09434-8. L. Caballero Zoreda, P. Mateos Cruz y C. García de Castro Valdés (eds.): Asturias entre visigodos y mozárabes. (Visigodos y Omeyas, VI – Madrid, 2010). Instituto de Arqueología de Mérida. Mérida, 2012. 488 págs. – ISBN: 978-84-00-09471-3. S. Camporeale, H. Dessales y A. Pizzo (eds.): Arqueología de la construcción III. Los procesos constructivos en el mundo romano: la economía de las obras. Instituto de Arqueología de Mérida. Madrid-Mérida, 2012. 399 págs. + figs. en el texto. – ISBN: 978-84-00-09500-0. M. Bustamente Álvarez: La terra sigillata hispánica en Augusta emerita. Estudio tipocronológico a partir de los vertederos del suburbio norte. Consejo Superior de Investigaciones Científicas, Instituto de Arqueología de Mérida. Mérida, 2013. 538 págs. + figs. en el texto + CD. – ISBN: 978-84-00-09673-1. T. Cordero Ruiz: El territorio emeritense durante la Antigüedad tardía (siglos iv-viii). Génesis y evolución del mundo rural lusitano. Consejo Superior de Investigaciones Científicas, Instituto de Arqueología. Mérida, 2013. 290 págs. + figs. en el texto. – ISBN: 978-84-00-09743-1. B. Soler Huertas, P. Mateos Cruz, J. M. Noguera Celdrán y J. Ruiz de Arbulo Bayona (eds.): Las sedes de los ordines decvrionvm en Hispania. Análisis arquitectónico y modelo tipológico. Consejo Superior de Investigaciones Científicas, Instituto de Arqueología. Mérida, 2013. 368 págs. + figs. en el texto.– ISBN: 978-84-00-09771-4. M. Pérez Ruiz: Al amparo de los lares. El culto doméstico en las provincias romanas Bética y Tarraconense. Consejo Superior de Investigaciones Científicas. Madrid, 2013. 520 págs. + figs. en el texto + CD. – ISBN: 978-84-00-09790-5. J. Bonetto, S. Camporeale y A. Pizzo (eds.): Arqueología de la construcción IV. Las canteras en el mundo antiguo: sistemas de explotación y procesos productivos. Consejo Superior de Investigaciones Científicas, Instituto de Arqueología. Mérida, 2014. 444 págs. + figs. en el texto. – ISBN: 978-84-00-09832-2 E. Salas Tovar (coord. científico), R. Mataloto, V. Mayoral Herrera y C. Roque (eds.): La gestación de los paisajes rurales entre la protohistoria y el periodo romano. Formas de asentamiento y procesos de implantación. Consejo Superior de Investigaciones Científicas, Instituto de Arqueología. Mérida, 2014. 258 págs. + figs. en el texto (ed. electrónica) – e-ISBN: 978-84-00-09814-8. M. Bustamante y D. Bernal (ed.): Artifices idóneos: artesanos, talleres y manufacturas en Hipania. Consejo Superior de Investigaciones Científicas, Instituto de Arqueología. Mérida, 2014. 488 págs. + figs. en el texto.– ISBN: 978-84-00-09843-8. T. Tortosa (ed.): Diálogo de identidades bajo el prisma de las manifestaciones religiosas en el ámbito mediterráneo (s. iii a.C.- s. i d. C). Consejo Superior de Investigaciones Científicas, Instituto de Arqueología. Mérida, 2014. 316 págs. + figs. en el texto. – ISBN: 978-84-00-09855-1. C. J. Morán Sánchez y A. Pizzo: Fernando Rodríguez. Dibujos de arquitectura y antigüedades romanas. Consejo Superior de Investigaciones Científicas, Instituto de Arqueología. Mérida, 2015. 218 págs. + figs. en el texto + CD. – ISBN: 978-84-00-09929-9. M. Á. Utrero Agudo: Iglesias altomedievales en Asturias: arqueología y arquitectura. Consejo Superior de Investigaciones Científicas. Madrid, 2016. 380 págs. + figs. en el texto.– ISBN: 978-84-00-10071-1. V. Mayoral Herrera: La revalorización de zonas arqueológicas mediante el empleo de técnicas no destructivas. ­Consejo Superior de Investigaciones Científicas, Instituto de Arqueología. Mérida, 2016. 288 págs. + figs. en el texto. – ISBN: 978-84-00-10112-1. A. Corrales Álvarez: La arquitectura doméstica de Augusta Emerita. Consejo Superior de Investigaciones Científicas. Madrid, 2016. 328 págs. + figs. en el texto + CD. – ISBN: 978-84-00-10138-1.

ARCHIVO ESPAÑOL DE

ARQVEOLOGÍA SUSCRIPCIÓN Y PEDIDOS / SUBSCRIPTIONS AND ORDERS DATOS DEL PETICIONARIO / CUSTOMER DETAILS: Nombre y Apellidos / Full Name: ������������������������������������������������������������������������������������� Razón social / Institution, Company name: ����������������������������������������������������������������������������� NIF-CIF / Tax number: ________________ Dirección / Address: ���������������������������������������������������������� CP / Postal Code: ____________ Localidad / City: ________________________ Provincia / Province: ��������������������������� País-Estado / Country-State: __________________________________________ Teléfono / Telephone: ��������������������������� Fax: _________________________ E-mail: ______________________________ Fecha de la solicitud / Order date: ___ / ___ / _______ Suscripción / Subscription: Precios de suscripción 2016 / Annual subscription Price 2016: ■  España / Spain: 68 euros ■  Extranjero / Other countries: 105 euros

Precios de suscripción 2017 / Annual subscription Price 2017: ■  España / Spain: 68 euros ■  Extranjero / Other countries: 105 euros

Números sueltos / Single issues: Cantidad / Amount

Año / Year

REVISTA / JOURNAL

Precios de número suelto 2016 / Single issue Price 2016: ■  España / Spain: 74 euros ■  Extranjero / Other countries: 116 euros

Volumen / Volume

Fascículo / Issue

Precios de número suelto 2017 / Single issue Price 2017: ■  España / Spain: 74 euros ■  Extranjero / Other countries: 116 euros

A estos precios se les añadirá el 4% (21% en soporte electrónico) de IVA. Solamente para residentes en la Unión Europea / 4% VAT will be charged to these prices, 21% VAT for electronic format. Additional for UE residents. Forma de Pago / Payment method:   Factura pro forma / Pro forma invoice -  Transferencia bancaria / Bank transfer -  Cheque / Check -  Tarjeta de crédito / Credit card: Visa - Master Card - Eurocard - 4B   Reembolso (solamente para números sueltos, en España / single issues only, in Spain) Distribución y venta / Distribution and sales: Editorial CSIC / Press CSIC C/ Vitruvio, 8. 28006 Madrid Tel.: +34 915 612 833, 915 681 619/620/640 Fax: + 34 915 629 634 e-mail: [email protected] editorial.csic.es

Firma / Signature ____________________________________________________________

ANEJOS

AEspA

LXXVII

ISBN 978-84-00-10142-8

Man-made materials, engineering and infrastructure

(eds.)

anejos de

aespA LXXVII

ARQUEOLOGíA DE LA CONSTRUCCIÓN V 5th International Workshop on the Archaeology of Roman Construction Man-made materials, engineering and infrastructure

th

ARQUEOLOGíA DE LA CONSTRUCCIÓN V

5 International Workshop on the Archaeology of Roman Construction

2016

Janet DeLaine Stefano Camporeale Antonio Pizzo

ARCHIVO ESPAÑOL DE

9 788400 101428

Cubierta _ARQUE_V_vol_77.indd 1

CSIC

ARQVEOLOGÍA

21/12/16 11:25