T E R R A C o fl^ p iro a tó W D e v o t io n in R oM ^ fes J an De Beenh o uw er Publications o f the Gallo-Roman Colofon De deputatie: Herman Reynders, gouverneur Mare Vandeput, Ludwig Vandenhove, Igor Philtjens, Frank Smeets, Jean-Paul Peuskens, Inge Moors, gedeputeerden Renata Camps, provinciegriffier Cover Image © Gallo-Roman Museum Print Drukkerij Paesen, Opglabbeek Paper Artie volume white gogr/rn2 ISBN 9789074605694 Atuatuca / Publications o f the Gallo-Roman Museum Tongeren, Under the supervision of Guido Creemers Kielenstraat 15 B-3700 Tongeren Tel 003212 670330 e-mail: grm@limburg.be http://www.galloromeinsmuseiim .be All rights, including translation, reserved in all countries. No part of the text or illustrations may be reproduced without written permission of the publisher Any reproduction by any means including photocopy, photographing microfilming, taping* recording or otherwise is an offence liable to be pun­ ished by law Tongeren, 2014 T e r r a c o t t a F i g u r i n e s a n d D e v o t io n in Ro m a n T o n g e r e n J a n D e Be e n h o u w e r Publications o f the Gallo-Roman Museum Tongeren, 2014 PR EFA C E Dear reader am delighted to introduce the fifth issue in the scholarly A t u a t u c a series from the GalloRoman Museum. The focus this time is the Roman terracotta figurines in the Museum’s own collection. I The reason for devoting an entire issue to these simple figurines is clear: despite their simplicity, they are hugely fascinating. Nonetheless, it has taken rather a long time for them to be studied in depth. The largest and most interesting find of such figurines was made back in the early 20thcentury. They were then super­ ficially interpreted by various researchers, including Renard, Van de Weerd, Dheedene, Paquay, Mertens and Marien. After that, it was a question of waiting for Jan De Beenhouwer to be bitten by the ‘figurine bug’, first as an ordinary student and later as a PhD candidate. When he present­ ed his substantial doctoral thesis in 2005, his startling conclusions attracted a good deal of media interest. The Gallo-Roman Museum asked Jan some time ago to supplement the findings from his PhD research with new insights about Roman Tongeren, especially in relation to Roman religion. The result is the new A t u a t u c a series publication that lies before you. Thanks to Jan De Beenhouwer we finally have a fully documented inventory of our fine collection of terracotta figurines. And thanks to his interdisciplinary research we are now in a position to present these new insights and conclusions to the public. This publication represents the first step in the process - communicating this information to academics, researchers and museum staff. At a later date we will of course incorporate the results of this research into our new muse­ um presentation so that it can reach the public at large. Bringing the results of scholarly research to a wider audience is one o f the core tasks of the GalloRoman Museum. I am delighted that this publication is yet another step in this constant and ongoing process. Happy reading! Igor Philtjens Deputy for Culture Atu a tu ca 5 - Co n ten t Preface.......5 Introduction......8 Terracotta figurines as an expression of local devotion in the Greco-Roman tradition.......10 The reproduction process of moulded terracotta figurines....... 12 Terminology The reproduction process Shrinkage Moulding errors and retouches Production areas with special mould characteristics and assembly technique The research method....... 20 The classification model: generation, series, and tree of descent. Dating. The collections.......24 The Contexts.......30 The importance of context in the dating and meaning of thefinds Roman Tongeren The discovery of thefigurines: details on thefinds The north temple The ‘Thien Schuur’ find ‘Broekberg’ ‘A an Paspoel’ ‘Kielenstraat’ ‘Sint-Truidersteenweg’ ‘de Schaetzengaarde’ ‘Elfde Novemberwal’ The southwestern cemetery The isolated burial group of the ‘Armand Meesenlaan’ Research into the trees of descent.......50 Tree of descents: Portuna, Minerva,Juno, and Venus Tree of descent 32: Portuna Tree of descent 45: Minerva Tree of descent 69: woman with dog and woman withfruit (Vindex and VR coroplasts) Tree of descent 75: Portuna andJuno capite velato (Vindex) Tree of descent 84: Mercurius Tree of descent 89: horseman with hooded cloak Tree of descent 101: bust of a woman Tree of descent 105: bust of a boy Conclusion Painting.......74 Slip Glaze Background painting Accent painting Unintended traces ofglaze and colouring slip Coroplasts.......80 -6- Vius and (At)ticus Servandus, 155/160-180/210 AD Victorfrom Cologne, 180-200/210 AD Vindex, 140/150-180 AD Provenance....... 86 Stylistic analysis: the ‘Thien Schuur’ example A thorough observation Chemical analysis Petrographic analysis of the paste Conclusion The fu n ctio n o f terracotta figurin es and th eir m eaning fo r the u se r .. ...98 Function of terracottas and user behaviour based onfindspot analysis Settlement finds of the civitas Tungrorum Finds relating to cult places Terracotta figurines from burial grounds Some Conclusions The significance of clayfigurines Conclusion C atalogu e....... 128 Cybele Diana Fortuna Juno Minerva Venus Venus and Amor Matrona Woman withfruit Apollo Horseman Mercurius Couple Bust of a boy Bust of a woman Bull Dog Goat Animal Hen Rooster Pigeon Handmade and wheel-turnedfigurines Moulded unguentaria Other moulded terracottafigurines P hotograph cred its....... 225 R eferences....... 229 -7- IN T R O D U C T IO N allo-Roman terracotta figurines were made in large batches. As serial products, they have less of an individual character than stone sculptures. They do not bear an inscription G explaining who they depict or the occasion for which they were made. They are also small and fragile. All this makes it difficult for anyone looking at these objects today to gauge con­ sumers’ intentions when they purchased a particular figurine and the meaning they invested in it. This book seeks to look at these figurines through different eyes. Firstly, through the eyes of researchers who analyse figurines and endeavour to place them in the context of today1s surviv­ ing terracotta production from Antiquity. But also through the eyes of those who made the fig­ urines and the buyers who gave them a particular purpose. Underpinning this work is a PhD study which I defended at the KU Leuven in 2005 and which explored Belgian terracottas within a broader Northwest European context. Guido Creemers sug­ gested applying the results of that study to the rich collection of terracotta figurines held at the Gallo-Roman Museum in Tongeren. This has given me an opportunity to publish the research methodology for mould-made pottery that I developed for my doctoral research. Because the fig­ urines were imported during the Roman period from present-day Germany and France, their sig­ nificance transcends the local. We therefore opted to publish in an international language so that this work can also be valorised in the international arena and be used for research purposes in other countries. The Tongeren collection is special in several respects. The majority of the figurines come from an exceptional hoard discovered in the early twentieth century. Many have been preserved virtually intact and still bear traces of the original paint. In addition, the Gallo-Roman Museum holds fig­ urines that were excavated in the city and in nearby Roman cemeteries during the nineteenth and twentieth centuries. In order to present a comprehensive picture, however, it did not make sense to confine the cata­ logue to the Gallo-Roman Museum collection. Guido Creemers and I therefore decided to incor­ porate the entire repertoire of figurines found in Tongeren. This meant including finds from excavations carried out by the Flemish Government and the city of Tongeren. The Tongeren figurines are described in all their many facets: their iconography, the clay paste used, the painting, hallmarks, assembly characteristics and the place occupied by the figurines within serial production as a whole. The result is a nuanced picture of their origin and chronology. -8- An evaluation of the fmdspots has allowed us to form a picture of the people who consumed these terracotta statuettes. We have chosen to do this within a framework that will have been meaning­ ful to these consumers, namely the administrative boundary of the civitas Tungrorum. Gathering together the finds from the civitas has made it possible to compare the capital and the hinterland. The entire repertoire of finds is also compared with known inscriptions from this same region. I would like to thank all those who helped put this publication together. My special gratitude goes to Guido Creemers who took the initiative for this book and who was responsible for the final draft. He was a constant source of encouragement. I wish to thank the staff at the Gallo-Roman Museum - Linda Bogaert, Guido Schalenbourg, Igor Van den Vonder and Oreane Vandenreyt - for providing logistical support. The Museum’s director, Carmen Willems, and the Limburg Provincial Council of Deputies also gave assistance with this publication. Geert Vynckier, Alain Vanderhoeven, Kristien Borgers and Jan Box were always happy to answer my questions about the excavations that they had worked on in Tongeren. Lu Ruyters, Ghislain de Schaetzen for interme­ diating concerning some photographs, furhtermore André van Doorselaer (KU Leuven), Raymond Brulet & Fabienne Vilvorder (UC Louvain). Annette Visser of Wellington, New Zealand translated the preface and chapters 10 and n. Restauratieatelier Restaura provided the photos of many figurines. Also deserving of thanks are all the museum and excavation depot staff at home and abroad who gave me access to their collection, in particular the Römisch Germanisches Museum in Cologne (F. Naumann-Steckner), the Musée National d’Histoire et d’A rt in Luxembourg (J. Krier and F. Dövener), the Rheinisches Landesmuseum Trier (S. Faust), the Institut National de Recherche Archéologiques Préventives in Autun (S. Alix and A. Delor-Ahii), the Universität zu Köln (C. Höpken), the Rijksmuseum van Oudheden in Leiden (S. Rutten) and the Musée Curtius in Liège (J.-L. Schütz). My final word of thanks goes to Marken, who was always ready to exchange views with me and who cast a critical eye over the many versions of the text. - 9 - T e r r a c o t t a f ig u r in e s AS A N E X P R E S S IO N OF LO CAL D EV O TIO N IN T H E G r e c o -R o m a n T R A D IT IO N The Greco-Roman world knew a rich figurative in the first century AD. Figurines of animals, tradition. The famous sculptures from the Classical and Hellenistic period have been admired and imitated widely throughout gods and human beings were mass-produced. In the second century, major production centres were situated on the upper course of the Loire River valley, in particular in the Allier history. Less known however, are the small terracotta figurines that were massmanufactured in the production centres of valley, as well as along the Rhine and Moselle. From these centres the terracotta figurines Greece, Asia Minor, Egypt, and southern Italy from the Archaic period onward. These were traded throughout the northwestern provinces of the Empire. They have been figurines are often found in sanctuaries where they were used as votive offerings. The Celtic world did not have a tradition in offering terracotta figurines. During the pre-Roman discovered in excavations from southern France to the Rhine, and from Great Britain to period, gods were rarely depicted in nonperishable materials.' Most widely demon­ strated is the custom of consecrating objects Made of fired clay paste, the figurines are usually smaller than 25 centimetres.3 The majority of the Gallo-Roman figurines were in holy places, such as the weapons of the defeated enemy, which fulfilled vows to the formed using moulds. Hand-built and wheelturned figurines are rare. Switzerland and Austria (fig.i). gods.2 Under the influence of Greco-Roman culture, the production of terracotta figurines in the Gaulish and Germanic provinces commenced 1 1 1 Gallo-Roman terracotta figurines are often expressions of devotion. Most of the themes depicted are embedded in the Greco-Roman figurative tradition. Although none of the Haffner 1995,26; Watson 2007,6-12: general discussion about ‘aniconism’ within the Celtic world Derks 1995,123. Lange 1994,212-213 Serie 22 No. 1, Taf.3,22, Abb. 2: one o f the largest figurines is a Venus that was found in Cologne, measuring 38,5 cm. Fig. 1. The location o f Tongeren and the main production places in the Rhineland and Central Gaul. preserved figurines bear inscriptions that allow us to unambiguously identify the Romanized society, the local pantheon acquired a new status stimulated largely by the representation, many gods and goddesses are traditionally recognized by their specific attributes. Mercurius, often accompanied by a institutional and social changes within the boundaries o f the new territories, the billy goat and a rooster, holds a herald’s staff (caduceus) and purse. Apollo is represented holding a lyre and plectrum, whereas Minerva can be recognized by her helmet and shield. Behind some of these representations ancient native gods may be hidden, the character of which resembles that of their Roman counterparts. These old gods were not carried on in an unaltered form. Influenced by the new ‘ dvitates.“ In addition to figurines of gods, human beings were also depicted: worshippers who made an offering to their god, but also military men with their horse or even caricatures. Furthermore depictions of animals and fruit were sold. Dogs, horses, pigeons, hens and roosters were particularly popular. Many of those have been recovered from sanctuaries and graves and were an expression of devotion. Van Andringa 2002,133-158. -11 - T h e r e p r o d u c t io n PR O C ESS OF M O U LD ED TER R A C O TTA F IG U R IN E S Terminology generations of a series are derived. In French Several specific terms are used in terracotta research, most of which refer to the typical reproduction process of mould-made literature, special attention is paid to the figurines. In this work we seek to include technical terms that are consistent with the scope of existing terminologies.1The meaning rarely found and, even then, it is difficult to determine with certainty whether it is a first model. of these terms is adjusted to the implemented research model.2 In addition, this research model also produced new terms beyond the scope of some currently prevalent terms in the literature on figurines. Barbotine (DE: Tonschlick; FR: barbotine; NL: barbotine): watered-down day paste, mainly used to attach the various separate parts of a piece of work. C oroplast (DE: Tonbildner, Koroplast; FR: coroplathe, coroplaste; NL: coroplast): producer of terracotta figurines. The word is derived from the Greek Kopo'trX.acrrrio'. archetype or prototype. Both terms are current and used interchangeably.3 A model is only G eneration (DE: Generation; FR: génération; NL: generatie): a range of figurines, entirely or partly derived by moulding from a common model, the common parts of which have the same measurements. When one o f these figurines is used as a model, a new generation begins. Details and internal proportions remain largely unchanged in relation to the previous generation. Each next generation is scaled down, because of the shrinkage caused by the water evaporating from the clay paste during the drying and firing processes. G laze (DE: Glasur; FR: glaçure; NL: glazuur): glassy layer on the surface of pottery. The First model , A rchetype , P rototype (DE: Archetyp, Erste Ausformung; FR: archetype, primary constituent is silicon dioxide. Potassium, lead or sodium oxide were added to lower the melting point. Depending on the prototype; NL: archetype, prototype, eerste model): first model from which the various components, the melting point varies from 900 to 1200 °C.4 If the glaze is *applied 2 3 * NichoILs 1952; Bemont et al. 1993,297-298; Muller 1997; Champion 1981, for more general terms; Vorschläge zur systematischen Beschreibung von Keramik. Suggestions for the systematic recording o f pottery, Kunst und Altertum am Rhein, 124,1986. De Beenhouwer 2007. Bemont et al. 1993,297; Muller 1997,450-451: according to Muller, the archetype is the idea the artisan has in minH, while manufacturing the prototype. The archetype is the source o f inspiration. The prototype marks the start o f the mechanical moulding. Champion 1981,40. -12 - intentionally, it usually covers a large part of the surface. G r o u p (DE: Grappe; FR: groupe; NL: groep): a range of figurines based on the relationship of separate characteristics that aren’t due to a mechanical relationship. Examples are a common hairstyle or a special painted design. A group often goes beyond the limits of theme and type. For example, representations of Minerva and Fortuna possibly belong to the same group because of common iconographie features or clay paste characteristics. M e c h a n ic a l r e l a t io n (DE: mechanische two parts. The separate parts are called h a l f ­ m o u l d s (DE: Halbform, Formhalfte; FR: valve; NL: maldeel, vormhelft). Complex themes require several p a r t ia l m o u l d s (DE: Teilform, Teilhohlform; FR: pièce, moule partiel; NL: partiële mal, deelmal). (DE: Tonperlen, Tonpunkte; FR: nodosités; NL: malknobb eitjes): typical rounded clay knobs on the surface of figurines, M o u l d b liste r s as a result of the use of plaster moulds. They are the imprints of small cavities, caused by the presence of air bubbles on the surface of the mould, which form during the drying process of the plaster mix. Verwandtschaft; FR: parenté mécanique; NL: mechanische verwantschap, malverwant- M o u l d in g schap)5: relationship by moulding. Mechanical relationship can only be attributed to the moulage; NL: moulage, afdruk): product of the act of moulding. In French literature, the term reproduction process. This term enables the classification of mass-produced figurines into ‘surmoulage’ is used commonly to indicate each moulding after the first generation.6Since it is very difficult to determine for certain series and trees of descent. M o d e l , P a t r ix (DE: Vorlage, Modell, Patrize; FR: (DE: Abformung; FR: objet moulé, whether an object belongs to the first generation, one should be careful using the modèle; NL: model, patrijs): object from which a mould has been derived. Only models that were made in durable material, for instance term ‘surmoulage’. clay paste consolidated by firing, are preserved. Models made in perishable materials such as mix prepared by the potter, as a raw material, to make pottery. The main element is natural or purified clay or a combination of several clay unbaked clay paste or wax, are not found on archaeological sites. An engraved line that follows the dividing line of the various partial moulds is a typical characteristic of the model. M ou ld (DE: Abformen; FR: mouler; NL: afvormen, mouleren): act of reproduction by means of a mould in order to obtain represen­ tations of the model. (DE: Form, Hohlform, Model, Matrize; FR: moule, matrice; NL: mal, matrijs): M o u l d , M a t r ix solid form in which the paste is cast or pressed. The raw material mostly used for the production of figurines in antiquity was clay or plaster. The model is the object used to produce the mould. The majority of Gallo-Roman figurines are made in a mould consisting of 5 6 Pa s t e , c l a y p a s t e (DE: Ton; FR: pâte; NL: pasta): types. Depending on the quality of the clay and the desired result, substances called temper, can be added: sand, crashed pottery and organic ingredients like chopped straw. The terms ‘paste’ or ‘clay paste’ are preferable to the term ‘clay to prevent confusion with the chemical or mineralogical meaning of the word. S eries (DE: Serie; FR: série; NL: serie): a range of figurines, similar in form, that were derived from a common model. Differences in size within the series occur because of various successive moulding generations. If touching up the model changes important details, or if the internal proportions are changed dramatically when moulding, it is advisable that the researcher branches off a new series, thus NichoUs 1952. Bemont et al. 1993,297; Muller 1997,454-456: the author distinguishes “surmoulage interne” in the original workshop and “surmoulage exteme” in another workshop. -13 - increasing the number of comparative points within the research of the tree of descent. Relationships between the series are guaranteed by their position in the tree of descent. S l ip (DE: Überzug, Engobe; FR: engobe; NL: engobe): watered down clay paste that was applied on pottery in a thin layer before firing. The layer could have been intended to improve the porosity of the surface, but it could also have been applied simply for decoration. Depending on the firing atmosphere in the kiln and the composition of the slip, the colour of the end product varies. For example, a slip rich in hematite results in either a red or black colour, depending on the amount of oxygen during firing. The slip layer can be applied by submerging. Sometimes, a thin slip-like layer is formed naturally when drying the piece of attributes. In some cases, the relationship only refers to the front, the back or parts of the body such as the head. The complex sequence o f generations within the series is expressed in the descent scheme. Figurines of several series related by moulding, the mould-related parts of which have the same size, belong to the same generation. T ype (DE: Typ; FR: type; NL: type): a range of representations within the same theme, with a well-considered choice o f attributes or posture. A different choice of attributes forms a new type. V a r ia n t (DE: Variante; FR: variante; NL: variant): a range of representations within the same type, with common details such as the position and form of the attributes. work, in the process of which the finest clay parts surface by the migrating water.7 The reproduction process (DE: Statuette, Terrakotte, Tonfigur; FR: statuette en terre T e r r a c o t t a fig u r in e / s t a t u e t t e The manufacturing o f figurines implies a meticulous process including various steps.8 The coroplast starts creating a first model. From this model one or more first-generation cuite, terre cuite, figurine; NL: terracottabeeldje): figurine manufactured from clay paste. In French literature, a distinction is moulds are made. These moulds are then used made between ‘statuette’, the height of which is smaller than or equal to half the natural size of the subject (usually smaller than 80 for the production o f first-generation figurines. The drying and firing of the clay paste involves a shrinking process, both of the centimetres), and ‘figurine’, the height of clay mould and the figurine. This process which is lower than or equal to 25 centimetres.8* causes a decreasing size o f the figurine compared to the model. If one of the firstT h e m e (DE: Leitform, Theme; FR: thème; NL: generation figurines is used as a new model for thema): a range of representations depicting the production of a second-generation mould, the same subject, which can be elaborated in the new mouldings again are smaller. This different typologies. Examples are ‘Mercurius’ and ‘the horse’. T ree of d escen t, d escen t sch em e (DE: Stammbaum, Stammfolge, Dendrogramm; FR: arbre généalogique, schème de descen­ dance, dendrogram; NL: stam, dendrogram) arrangement of partly mould related series, classified according to their origin. In a tree of descent, the related series deviate because of a slight change in the position or change of 7 8 5 Muller 1997,440: “engobe naturel, pseudo-engobe”. Bémont et aL 1993,297. Jastrow 1941; Nicholls 1952. -14- action can be repeated several times, each time involving a considerable loss of height. In that way, the industrial process results in a series of mould-related figurines, derived directly or indirectly from the same model (Fig. 2). Starting from the specific principles of the reproduction process, it is theoretically possible to reconstruct the mechanical sequence of mouldings through a comparative study into the form and sizes of the terracotta figurines. It is therefore important to gain insight into the processes that affect form and size. W ith regard to form, we mean unintentional moulding errors, intentional retouches, and adding new details. W ith respect to size, we mean factors that affect shrinkage. the model while making a mould, thus decreasing the number of figurines produced from the model and increasing the frequency of the succession of generations. A fired model is more durable and can be used to manufacture several moulds o f the same generation. When working with clay moulds, both mould and figurine are equally responsible for the shrinkage. Using plaster moulds, the shrinkage of the mould is negligible. In that case the loss of height can chiefly be ascribed to the shrinkage of the moulding. When interpreting measurements of the figurines, it is important to keep in mind that the difference in size between several generations will be much smaller using plaster moulds and that successive generations will succeed each other more rapidly. Shrinkage The study of shrinkage allows us to determine the mechanical order of specific generations of figurines. The shrinkage is measured in terms o f percentage by comparing digital images. To correctly interpret the results, it is necessary to understand the factors that cause and affect shrinkage. Model, mould, and moulding are the chief agents in shrinkage. Depending on the Various experiments give an indication of the degree of shrinkage. Measurements conducted with modem mouldings in Attic clay of an original Attic figurine, fired in a kiln at an estimated temperature o f 850 to 900 °C, indicate that the shrinkage of the clay mould and the statuette each are responsible for half of the shrinkage.“ The total shrinkage in height amounted to 14.4 % in this experiment. The shrinkage in width diverged slightly and amounted to 12,8 %, measured at shoulder production method, they take on a varying portion of the total amount of shrinkage.1012In height. Other tests with clay paste from Toulon-surAllier assume a shrinkage of 20 % for each general, the coroplast used a figurine, whether dried or fired, as the model for a mould of a new generation. It has been demonstrated generation when using clay moulds.13 According to these modern experiments, the expected reduction per generation can be experimentally that the difference in height calculated. Depending on the composition of the paste, the height decreases by 14 up to 20 % between a dried and a fired mould-formed terracotta figurine is minor.“ Most o f the shrinkage occurs during the drying of the piece of work, not during the firing. In this respect it is an advantage if the clay model is not completely dry and preserves its plasticity, when producing plaster moulds. The loss of height will be less. The main disadvantage using unfired models is the inevitable loss of when using clay moulds and by 7 up to 10 % using plaster moulds, or even less when the model has not been fired or dried completely. Nevertheless, these figures have to be used with caution, as the precise operating procedure of the antique coroplast remains unknown. Attempts to classify series in equal generations are not feasible because it is 10 Van Boekel 1987,227-230: Advantages and disadvantages o f the use o f clay and plaster; Rose 2006,14-16. 11 12 0 Nicholls 1952,220 nt. 21, Pi. 44(bi). Nicholls 1952,220 nt. 23. Lahanier 8r Revel 1993,257: “...le coefficient de rétraction de cette terre de Toulon-sur-Allier est voisin de 10 %". The firing temperature, preparation o f the paste, and the measurements o f the firing tests were not described. -15- Fig. 2. Clay mould for the production o f Venus fi­ gurines signed by the coro­ plast Pistillus and frag­ ments o f figurines made w ith that mould. Autun, ‘Porte d’Arroux* (INRAP). Height o f the m ould 15,4 cm. impossible to determine the precise percentage of shrinkage for each generation. Consequently, throughout the literature varying figures have been proposed with respect to the shrinkage percentage per generation.14In reality, the evolution of a series shows no regular steps of 10 or 20 %. The development is gradual. Differences in size of only 2 or 3 % are measured frequently. This is undoubtedly caused by the complexity of the evolution of a series. For example, it is possible that not one but several models, deriving from a common first model, were used to produce figurines o f the same generation. These models were not necessarily developed in the fill gaps in the series and thus change the order. Therefore, it is more accurate and objective to express the relation between two comparable figurines by means of a reduction percentage.15*18 Finally, it is important to bear in mind that the reduction percentage differs depending on whether one measures the height or width of the figurine. Modern experiments have indicated that the vertical measurements are particularly important when determining shrinkage, since horizontal measurements are subject to distortion and give less significant variation15. same workshop: a different compound of clay paste could cause variations in the shrinkage. Variations in wall thickness of the figurine and the non-firing of the model also affect the changing dimensions. In the complex evolution of the series, these small differences accumulate. The number o f deviations increases from generation to generation. Moulding errors and retouches Not every detail on the figurines originated in the same way. Within one generation up to three levels of detail can be distinguished, which reflect three steps in the production process, in particular the creation o f the Consequently, the succession of generations cannot be expressed in precise, regular steps. When trying to represent the complex model, the mould and the figurine. The overall evolution of the series, one has to deal with creation of the model. In the next step, some details can be introduced by carving the these often small and irregular differences. The antique coroplast was plagued by the problem of the sizable loss in height as recorded in various modern experiments. Theoretically, with shrinkage of 10 % each generation, a figurine of 20 cm in the first generation would be reduced to 13 cm by the fifth generation. We often see that the coroplast tried to compensate the loss of height by stretching the torso or the legs, or heightening the plinth. Consequently, the number of generations cannot be determined with certainty. To use terms as first or second form of the figurine and most of the details can be traced back to the first step, the mould. In the third and final phase, the figurine itself can be retouched. This was mainly done to trim moulding errors. In research focused on finding relationships between series, it is important to distinguish between the characteristics that are inherent to the model, the mould or the figurine. In practice, this is not easy, as the evolution of the detail cannot be limited to one generation. Adding and reworking details is a continuous generation would in most cases be an over­ simplification. Moreover, we have to keep in mind that the number of generations depends process. Usually, the researcher has to proceed according to the relative sharpness of the detail. Strongly blurred or vaguely discernible details date from an early production stage, while sharply cut details are from a later stage. on the completeness of a series. New finds can A difference in sharpness can often be H 15 18 Rey-Delqué 1985,9-10: “environ 10 Rouvier-Jeanlin 1986,16: “il faut compter environ 10 % de déperdition de tadle par surmoulage”; Lange 1990,44: “ca. 10-20 %” ; Kassab Tezgôr et Abd El Fattah 1997,357: “une différence de taille supérieure à 6 %”; Rose 2006,16: “ca. 20%”. Jastrow 1941,2 f£; Nicholls 1952,220 nt. 21: according to Nicholls, the shrinkage caused by differences in clay paste or wall thickness is insufficient to affect the classification in generations. However, he does not take into account the complexity o f the production process, accumulation o f errors, and possible compensations b y the coroplast. Nicholls 1952,224-225 nt. 47. discerned between the front and back halves of the figurine. This could be attributed to the combination o f two half moulds from different production stages. Another reason could be that the figurines were usually placed frontally, the back being considered less important. Consequently, it was less frequently touched up than the front. In addition to the difference in sharpness, there are other methods for discovering the relative moulds were drawn from the same model, this could result in figurines of the same generation, which showed slightly different details. Nicholls sketched this process for Greek moulded terracotta production in the Archaic period.17According to Nicholls, some details were not fully elaborated on the model. Fig- 3- Mouldblisters on the back o f Fortuna TO 21 from the Tongeren ‘Thien Schuur’ find. age of the details. Undoubtedly, the most important is the comparative research of the various specimens that are part of a series. Details common to various generations can be traced to an early phase. Details that change during the succession of generations or new details that were added can be attributed to the generation in which they first appear. To be able to sketch the mechanical succession of the generations, it is important to know which details are inherent to the model and which details were added by retouching the mould or For the head, this may involve the hair, brows, eye definition, exact outer edges of the lips, the figurine. The use o f plaster moulds provides a special advantage here. The surface and ears. The missing details were cut into the mould. Depending on the depth of detail, this o f figurines that were moulded in plaster moulds shows characteristic mould blisters could happen in various drying stages of the that often lodge in hair curls or pleats (Fig. 3). The presence of these mould blisters proves that the details on the figurine or in the mould were not secondarily retouched but that they are inherent to the model. Retouches applied on the figurine are difficult to discern from those applied in the mould. They are obvious when applied to correct moulding errors or to restore defaults, due to the assembly of the different parts of the piece of work. Most apparent is the restoration of details that were blurred during the smoothing of the joins. These retouches are inherent to the current figurine and are not clay. The last retouches were sometimes etched with a fine metal point in the dried clay to make sure the detail appeared on the figurine with the intended sharpness. As such, manufacturing the mould is as important as manufacturing the model. Nicholls assumes that the first model and mould of the first generation were manufactured at the same time and by the same artisan. When various moulds were fabricated from the same model, this operating procedure would, in spite of similar measurements, result in a slight variation of the detail. The development of a series of Minerva produced in Roman Cologne, demonstrates the use of parallel moulds in the Rhineland (Fig. 4). passed on from the previous generations. Figurines originate from the same mould if the dimensions are identical, and particularly, if Production areas with special mould characteristics and assembly technique they show characteristic mould errors. Again, A simple observation of the surface of the mould blisters can be of help here in addition figurine and characteristics of the joins can to damaged or blurred details. When different17 lead to important findings with respect to the 17 Nicholls 1952,221. -17- production technique.18 In most cases, the surface, joins, and creation of the plinth immediately give a definite answer about Rhineland or Central Gaulish origins. Secondly, we distinguish between details that mould blisters on the surface of the figurines were present before firing and originate from the original design, or from a manipulation of the mould, and details that were applied after firing The position of the vent, modelling of the interior hollow, and the finish of the lower already present on the model and were not retouched in the mould or on the moulded figurine. Details that, on first sight, seem to be secondarily etched after the moulding, can not only indicates the use of plaster moulds. They are also important for the description of the detail of the mould. Their presence in folds o f clothing shows that these details were rim o f the plinth can provide important indeed be confidently ascribed to the mould in t>iat way.20 Generally speaking, Gallo-Roman additional information. figurines were only rarely retouched after the Fig. 4. Variation o f details within one generation of Minerva statuettes produced in Cologne. From left to right TO 38, TO 39, TO 42, TO 43. Many moulds and half-moulds have been found in Central Gaulish production centres moulding. An exception is the reworking of along the Allier. In the Rhineland, some clay moulds have been found in Trier, Xanten, Frankfurt-Heddernheim, Bad Bertrich, and was heavily smoothened over a broad surface. Blurred details were often re-etched or cut. The smoothing of the seat or the plinth was Karden. However, it is uncommon to find also done after the assembly of the different moulds in this area around the Rhine and Moselle, because the production there was mainly characterized by the use of plaster moulds. Although the plaster moulds parts of the figurine. These are secondary characteristics that do not concern the mould. The broad smoothing out of the surface of the themselves have not been preserved, one can recognize their use on the surface of the figurine that was moulded in it. Plaster moulds leave typical traces. The most characteristic are the small, rounded clay knobs called mould blisters (fig.3). They are the impression of air bubbles that arise while mixing water and plaster and that are marked off in the skin of the mould.19The presence of * 19 “ the join. Especially in Rhineland products, it seat, the plinth or the join is generally characteristic of Rhineland products (Fig. 5). Repeatedly, the lower rim of the plinth was smoothed on the outside, but often also on the inside. Equally on the inside, the corners of the plinth were often cut out. The interior of the figurines also has to be examined at the position of the joins. Figurines manufactured Van Boekel 1987,216-231. Rüger 1980,18-19; Van Boekel 1987,230; Van Boekel 1996, 6-7; Rose 2006,14-15. Catalogue TO 63. in plaster moulds were assembled, while the clay paste was still plastic. The parts remained in the half-moulds, while joining them." This is visible from a shift of the interior join in assembling the parts, with the redundant paste pushed aside on the inside. Until now, this phenomenon has been observed in Rhineland specimens only (Fig. 6). modelling of the lids or the pupil o f the eye. Depending on the freshness of the Fig-5- Broad smoothing of the join. Fortuna TO 17 from the Tongeren ‘Thien Schuur’ find. mould, these charac­ teristics may or may not be visible in individual moul­ dings. It is not always obvious whether cer­ A shift of the interior join cannot occur when the parts of the figurines had already dried before assembly, as was common in Central Gaul.“ When dried, the separate halves of the figurines have to be carefully assembled using thin slip or harbotine. This requires more precision in manufacturing the various parts tain details are in­ herent to the mould or added later. For instance the ears of animals were not always integrated into the mould. In because the joins have to fit almost perfectly. Fig. 6. Shift o f the interior join visible at the fracture area o f the mane. Horse TO 55 from Tongeren 'Aan Paspoel'. In addition to the observation regarding the surface and joins, it is important to verify which parts of the figurine were integrated in the main mould and which parts were formed apart. The plinth of Central Gaulish figurines was usually formed separately. This closed the interior space of the body at the bottom. A vent through the wall prevented the figurine many cases, they were formed separately and from cracking during firing. In the Allier area, the legs and plinth were usually formed separately in mammals, the ear cavity was sculpted out. This is often apparent from the cracks and modelling traces around the ear. In that case, the ears are not whereas they were integrated into the mould in the workshops around the Rhine and characteristic of the mould and the various specimens of the same series may differ Moselle. In spite of this general tendency, slightly.24 exceptions have been recorded (Fig. 7I.23 Therefore, caution is called for when determining the production Fig. 7. Horse originating from a Rhineland workshop found in Taviers w ith legs, tail and plinth integrated in one mould (left). Height u,4 cm. Horse from Elewijt pro­ duced in the Allier area (right). Head and trunk are made in one mould. The legs, tail and plinth were formed seperately and attached to the body using harbotine. Height 16 cm. region. It is wise to take into account several criteria. As a rule, the general iconographic characteristics that define a series are inherent to the mould from which the figurines originate. This is also true for small details such as the “ “ •3 ■» De Beenhouwer 1990,11-13. Rouvier 1972,25; Jeanlin 1993,96-102 De Beenhouwer 2005,364, no. 3605 (204) and no.728 (728): a vent is pierced in some Rhineland specimens; De Beenhouwer 2005; the plinth is integrated into the mould in some central Gaulish series 267,268,585,592,828, 594, and 709; De Beenhouwer 2005: mammals showing atypical assembly series 805,807, and 971. De Beenhouwer 2005, Series 797: The mould for the production o f bulls by Sacrillos does contain the horn, but not the ear. -19- Resea r c h m eth od The classification model: generation, reduction percentage differs for the horizontal series, and tree of descent After specifying which factors affect the form and measurements of the moulded products during the reproduction process, a and vertical measurements, as the coroplast himself could compensate for the loss in height reconstruction of the mechanical order of the form cannot be converted into identical moulded figurines can be carried out. Based on uniformity, the examined objects are classified in corresponding classification units. The basic products. Sometimes the successive changes in details are so dramatic that the uniformity compared to the original model is unit is the series. A series can be defined as a number of figurines, similar in form, that were compromised in the end. For practical reasons, it is therefore advisable that the researcher derived from the same model by moulding. Differences in size within the series occur because of the use of various successive separates a new series if important details were generations of moulds. Based on the reduction in scale, the series is classified in generations. The basic definitional criteria of a series are the similarity in form and the common model. We have already mentioned various factors that affect the uniformity in mouldings, more specifically, the moulding errors and successive retouches. Another agent is the gradual change of the internal proportions, such as the relation of the length of the torso and the length of the legs, during the reproduction process. The 1 by artificially lengthening the legs and narrowing the loins. Therefore, the similarity in altered retouching the model, or if the internal proportions were dramatically altered*. This offers the advantage of an improved series definition, more accurate descriptions, arid more precise chronologies. On the other hand, it means that the series in turn has to be inscribed in a larger tree of descent. The tree of descent can be defined as the classification unit that represents the mutual relationships between the various series, each including several generations. To succeed in relative dating of these relationships, it is necessary to determine in which generation the branching Nicholls 1952,225: sometimes, the working o f a mould can be so radical that the result leads to the creation o f a new series. - 20 - Fig. 8. Detail o f Tree o f descent 104. Bust o f a woman wearing a necklace and earrings. On the left TO 75 from Tongeren and on the right a specimen o f the cemetery ‘La Spetz’ in Arlon. The reduction percentage is nine per cent. off of new series occurs. Therefore, the complex succession of generations within the series has to be integrated into the examination of the tree series evolved faster in the workshops that used plaster moulds than in the workshops that used clay moulds.3The only way to achieve absolute of descent. Figurines belonging to several series related by moulding, of which the mould- dating is by deriving data from archaeological contexts. The series dating has more evidential related parts are of similar size, belong to the same generation of the tree of descent. To value if similar dates for the series are gathered in various archaeological contexts. In its turn, a well-dated series is an important chronological guarantee the objectivity of the generational research, both the reduction percentage and the guide in the evolution of the tree of descent. two compared objects in the scheme of the tree of descent have to be stated (Fig. 8). The tree of descent reproduces a complicated web of relations. A relationship between two A great advantage of this classification system is that it displays the succession in time, in other words the relative chronology. It shows that one generation is younger or older than another or that one series is older or younger than another. It does not express how much time passed before the new generation or series came into being.2 We may assume that the ■ f series in the descent tree only mechanical relationship and the the descent. To find out where workshop or production area was expresses a direction of the original situated and which workshops imitated the original production, external data have to be connected, concerning find-circumstances, coroplasts, production technique, and composition of the paste. Nicholls 1952,226: in connection w ith the chronology, it is clear that the derived production can lengthen the age o f a series considerably. There is no measuring instrument to determine whether a series, consisting o f several generations, had been in use during four years, 40 years, or longer. The dating o f the contexts can be helpful, bearing in mind that it only offers a terminus ante quem for the start o f the production, Rose 2006, 15: whereas clay moulds slowly wear away, a plaster mould only produces around 30 well-defined impressions. After around 100 impressions, it is worn out. - 21 - It is dear that moulded pottery can provide a wealth of data that is inherent to the reproduction process. The tree-of-descent schemes that are the result of this research grow along with the current state of research. It is an illusion to think that they give a complete picture of the production in antiquity. The rdationships we can establish are limited by the archaeological finds that come to light, which in reality is only a fraction of the real potter’s inscription cannot be used as an argument to date the potter. As research makes progress, it is advisable to record and preserve each dating level. If, tomorrow, we have better arguments to date a certain coroplast more accurately, then only the level of the dating from the inscription will be adjusted for the series involved. The archaeological level obviously remains unchanged. production size. Dating Assigning chronology to a series of figurines is a complex process. As several criteria are included in the research, it is necessary to distinguish between the various dating criteria. Evidence gained from an archaeological context is more conclusive than iconographical interpretations. It is therefore important to distinguish between various levels and keep them separated in order to avoid circular reasoning. For example, a series dated by a g . g. Ten steps to date a lie s o f figurines. Various steps have to be carried out to date a series (Fig.g). First, all figurines are dated by the chronology of the site, the find-spot and the context in which they were found. A few figurines are inscribed with the names of the ruling consuls: these figurines carry an accurate dating in a particular year. The archaeological and consular dating of the figurines that constitute a series forms a solid foundation for dating these series (step 1). Taking into account the internal and external mechanical relationships, these archaeologically dated DATING OF TH E TREE OF DESCENT DATING OF THE SERIES Archaeological dating (site,find-spot, context) and C onsular dating (inscriptio n) 2 * Dating from context and tree of descent Dating from context 3 Dating updated by tree of descent 6 Coroplast (inscription) Dating updated by inscription Dating updated by inscription Dating updated by tree of descent and inscription 7 * -------- GE NE RA L OATING WITHOUT USIN G ICONOGRAPHY 10 Iconography Dating updated by iconography - 22 - Dating updated by iconography series help to sketch the evolution of the tree of descent (step 2). The precise position of the series within the evolution of the tree of descent allows us to first correct the dating of the series (step 3). The result is a dating of the series from archaeological contexts, corrected by the evolution of the tree of descent. This general date is the starting point for the dating of the coroplasts (step 4), which takes us to the second dating level. Coroplasts are the figurine makers, whose names are known to us from inscriptions on figurines or moulds. At this level, external arguments can be introduced. Next to with the information about the coroplasts (step 6). Since the evolution of the tree of descent is a mixture of signed and unsigned series, and since external criteria were also considered when deciding on the dating of the coroplast, in a further step the series has to be updated again with the corrected data from the tree of descent (step 7). In the end, the result of all these steps is a general dating of the series without iconographical determinations. This general dating is the basis of the third and final stage: the statuettes, a coroplast could have manufac­ iconographical level (step 8). At this level, we examine which iconographical motifs are tured other products that can be attributed to the dating o f his activities. Additionally, various mutual relationships between the chronologically relevant. In rare cases, external criteria are included, such as well-documented hairstyles or important arguments from the coroplasts, that become visible in the tree of study of stone sculptures. Finally, these icono­ graphie dating criteria are applied to the dating descent, are considered in the determination of the final dating. The result can be applied to a series for which archaeological contexts are unable to provide accurate dating (step 5). Next, of the series (step 9) and the tree of descent (step 10). In doing so, the researcher can always weigh the nature of the chronological determinations the tree of descent, in turn, has to be updated and assess each level separately. - 23 - The c o l l e c t io n s The current collection of the Tongeren GalloRoman Museum has come a long way since the founding of the Société scientifique et littéraire du Limbourg in 1851.12The fifty Roman objects acquired by the society during its first year were later supplemented by many new acquisitions. In 1854, a part of the city hall was set up for the collection.1 However, the enthusiasm had waned 30 years later. After 15 years o f inactivity, the novice François Huybrigts managed to breathe new life into the society in 1895.3 Under his impetus, a museum was founded in 1906 in a wing of the city hall. During WO I some of the objects disappeared. Until 1915, the archaeological fund of the society curated Venus TO 50. Unfortunately it was stolen on 8 January of that year from the museum in the Tongeren city hall.4In 1922, the collection moved to the museum in the former Dominican cloister that the city had acquired.56At that time, 1 2 3 4 s 6 7 8 various collections were being housed there.® On the one hand, there was the archaeological collection of the city of Tongeren, including the Christiaens-Vanderijst collection, acquired in 1921. Later, objects from large infrastructure works in 1934-1935 on the sewer system were added.7 On the other hand, the museum housed the collection of the then Limburgsch Geschied- en Oudheidkundig Genootschap. It included the collection of Oscar Schaetzen that had been donated to the society in 1908, in addition to the fruits of the annual excavations of Fr. Huybrigts, commissioned by the society between 1906 and 1913.8 All these collections later passed into the collection of the current museum. Oscar Schaetzen, Tongeren The knighted Oscar Schaetzen, born in Tongeren in 1836, became a member of the Paquay 1934, 33-34: the society was founded in Tongeren on 22 December 1851 and later converted into a society without commercial interests on 1 August 1924. In 1927, the SSLL was given a Dutch name: Het Limburgsch Geschied­ en Oudheidkundig Gezelschap, later Genootschap. Baillien 1951, XXI-XXIL Baillien 1951, XXIX; Paquay in nécrologie 1924. Paquay 1935,75 nt. 5. Baillien 1951, XXXIV. Paquay, 1935,68-69. Paquay 1935,74. Paquay 1934,33 and 34, w ith bibliography. - 2 4 - Fig. 10. Oscar Schaetzen. Société Scientifique et Littéraire du Limbourg in i860 (Fig io).s In 1869, he became a member of the steering committee, and from 1897 to 1907 he was deputy chairman. He also built a private collection. Shortly after his death in 1907, his collection of antiquities was donated to the fund of the Société. This fund later passed into the current museum collections of the Gallo-Roman Museum. In the report on the donation, two terracotta figurines were mentioned: “Parmi les objets en terre cuite, signalons deux figurines en terre blanche.. One of both figurines was specified as “une statuette, en terre blanche, d’une déesse”. The goddess in question is Diana TO 2." The other statue was not specified and could not be identified. François Huybrigts, Tongeren François Huybrigts was bom in Diepenbeek, Belgium in 1845 (Fig. 11). He later moved to Tongeren.“ He settled there as “hoofdconduc- went up in flames. The remaining fragments were collected into 18 crates. After his death in 1925, his wife gave it all to Baron Philippe de Schaetzen. The content was described as heaps o f partly calcified shards and formless objects.14 Jan Paquay published in 1934 the Fig. 11. François Huybrigts. teur van Bruggen en Wegen”, official in charge of bridges and roads. Together with the knighted Oscar Schaetzen, he was the driving force behind the activities of the Société Scientifique et Littéraire du Limbourg in 1895. He also became the secretary of the society from 1901 until his death in 1925. In 1906, he set up a museum on the ground floor of the city hall of Tongeren. Until the beginning of World War I, the society funded the annual excavations, led by Huybrigts. He also built a valuable personal collection. From 1885, he collected thousands of objects from Tongeren and its surroundings.*10 *3 It would become the largest and most important private collection Tongeren has ever known. When his house burned down in the violent war night of 18 to 19 August 1914* the collection » Bulletin de la Société Scientifique et Littéraire du Limbourg 25,1907,413: summary o f his public career; Baillien 1951, 10 11 i XIX: born in Tongeren in 1836, died there in 1907. Bulletin de la Société Scientifique et Littéraire du Limbourg, 1907,131. Inventory o f Oscar Schaetzen, preserved in the Gallo-Roman Museum o f Tongeren, inv. SC124. Baillien 1951, XXI; Bulletin de la Société Scientifique et Littéraire du Limbourg 39,154 e.v. nécrologie 1924. 0 “ Anon. 1958. Tongeren Romeins Trefpunt, 49-50. Anon. 1958. Tongeren Romeins Trefpunt, 1958,66-70. -25- report for the War Damage Tribunal, made up on 7 June 1920 describing the destroyed collection.'5However, this report is unreliable. None of the objects from the southwestern burial site mentioned in the data on the grave collections in this report correspond with data published in 1905 by Lucien Renard.15 16 When assessing the damage, the financial interests were probably more important than the scientific interests. We may assume that Huybrigts’s notes were also lost in the fire, so Fig. 12. Philippe de Schaetzen. well-documented findings. This is not always the case for the personal collection of Huybrigts. Without questioning his sincerity, it seems possible that Huybrigts supplied his collection not only with the finds of his own excavations, but occasionally bought objects that were told to be of Tongeren origin. This could explain the presence of a Mediterranean figurine, an object from the antique market, in his Tongeren collection.17 Philippe de Schaetzen, Tongeren Baron Philippe de Schaetzen was born in Tongeren on 30 August 1903 (Fig. 12). He was the son of Baron Franz de Schaetzen.18He was raised with an interest in his native city’s past. His grandfather, the knighted Oscar Schaetzen was, as already mentioned, interested in Roman Tongeren as deputy chairman of the Société Scientifique et Littéraire du Limbourg. His parents also owned a mansion on ‘Stationstraat’ in Tongeren, next to the house of François Huybrigts, who kept a valuable archaeological collection. Both houses were set afire by the advancing German troops on 18 an exact description of the objects and their relationship was no longer possible. For that reason, we only assigned value to the rather concise notes of L. Renard from 1905 with respect to the graves, published before the August 1914. Consequently, Huybrigts’ collection was largely destroyed. Philippe de Schaetzen came into the possession of some of the remaining pieces of the collection and acquired 18 crates with heavily damaged remnants. This material formed the basis for destruction of the collection, when financial interests did not play a part. his archaeological collection.192 0He moved into villa ‘Van der Heyden’ along the ‘Sint The journals o f the annual excavations, executed by Huybrigts and his foreman Truidersteenweg’. Its garden would prove to be Timmermans on assignment for the Société, have been published on a regular basis. There is no doubt about the authenticity of these 15 16 17 fertile ground for excavations. He inherited more property along the ‘Romeinse Kassei’ from his mother.” A large part of his collection comes from this site, where excavations were Paquay 1934. Renard 1905. According to L. Renard, a clay figurine w ith the representation o f a couple was found in a grave on Lot D19C, Renard 1905,289. This lot borders the ‘Sint-Truidersteenweg’ near the Roman road to Cassel. Among others the grave held a dark colour coated vessel decorated w ith the motto “LVDE”. According to this vessel, the grave is dated to the second half o f the third century or the first h alf o f the fourth century, Kiinzl 1997,129: general dating between 255 and 355 AD. The figurine is unfamiliar to the Gallo-Roman workshops in many aspects, such as the proportion o f the heads compared to the body, the fabric, the flat unworked backside and the large vent in the middle o f the back. * Baillien 1973? with bibliography o f the works o f Schaetzen by J. Smeesters. 19 Anon. 1958. Tongeren Romeins Trefpunt, 66. 20 House number 46, from 1963 changed into house number 36. - 26 - carried out starting in 1936.212He later equipped the villa he built on this site with a special property of the archaeological city fund.29 Today it is part of the collection of the Gallo- room to show his collection. In 1934 the Société Scientifique et Littéraire du Limbourg appointed him curator of the small museum in the ‘Predikherenstraat’ in Roman Museum. Tongeren, and in 1945 he also became chairman of the society. In this capacity, he convinced the Limburg provincial Christiaens31 was closely associated with the city’s archaeological past (Fig. 13). Later, as a structural engineer, he was professionally government to build a new museum, and devoted himself to the development of a fullyfledged Gallo-Roman Museum in Tongeren, associated with the city’s groundwork. It is thus not surprising that many of the city’s Roman antiquities ended up in his possession. for which the first credit was granted in 1948.“ He remained curator of the museum until 1968 Because of his valuable collection and close association with the Roman past of Tongeren Christiaens took part in the archaeological Mathieu Christiaens, Tongeren Through his parent’s collection30, Mathieu and died on 22 September 1973. In the period between 1930 and 1950, he brought together an important collection of policy as committee member of the Société Scientifique et Littéraire du Limbourg. antiquities, mainly from his own research in Fig. 13. Mathieu Christiaens. Tongeren. The terra sigillata collection alone included around 5000 shards.23 His research concentrated on ‘Broek’ along the ‘Hasseltsesteenweg’ and ‘Aan Paspoel’.24 Christiaens-Vanderyst, Tongeren The Christiaens25 family owned properties at the height o f the Roman southwestern cemetery at ‘Aan Paspoel’, some parcels leaning against the second-century city wall.26 During the exploitation o f some clay extraction sites, grave finds were collected that formed the basis of the ChristiaensVanderyst collection.27 Additionally, finds from other places were also integrated into the collection.28 In 1921, the collection was acquired by the City of Tongeren and became 21 Inventory book o f P. de Schaetzen, No. 4201 to No. 5200. 22 23 24 Baillien 1951, XXXVIII. Baron de Schaetzen & Vanderhoeven 1964,2. Baron de Schaetzen & Vanderhoeven 1964,8: summary o f the various sites w ith cadastral lots; Vanvmckenroye 1984, » Tongeren city archive, civil records: Jan Hendrik Christiaens, bom in Tongeren in 1838 and married to Maria Joseph “ Vanderiist in 1862. ■ ..................... 1 . , De Schaetzen & Vanderhoeven 1956; Vanvinckenroye 1984,14 Afb. 3: near the Romemse Kassei, Komnksemsteenweg, 17 •* Lindenstraat, and the second-century city wall. Paquay 1934, 61; Anon. 1958,44. Paqay 1934, 62: Kerkhofweg excavation, 1894. « I Paqay 1935, 69. See Christiaens-Vanderijst collection _ , . Tongeren city archive, civil records: Jean Joseph Mathieu Christiaens, born in Tongeren m 1865 and married to Mane Anne Christina Peeters in 1889. - 2 7 - Around 1900, either in the garden or at the height of an outbuilding of his residence adjoining a street named ‘Thien Schuur’, Christiaens came across a remarkable deposit consisting o f at least sixty-four terracotta figurines. Apart from this noteworthy collection of terracotta figurines from one site in the Roman city, the collection also included bronzes, coins, pottery, and glass, mainly from the southwestern cemetery.32 On 3 September 1953, the ‘Mathieu Christiaens-Peeters’ collection was acquired by the Belgian Province of Limburg and housed in the Tongeren Gallo-Roman Museum.33At the time, it still numbered fifty-one terracotta figurines. The other figurines remained in private hands, distributed over various collections. Robert Christiaens, Tongeren Mathieu Christiaens donated one Fortuna figurine that belongs to the Thien Schuur find, probably TO 38, to his brother Robert.34 Thien Schuur find.37The current depository of these figurines is unknown. Gilkens, Hasselt One Minerva figurine from the Thien Schuur find was given to Mr. Gilkens from Hasselt.38 This is probably TO 43. In 1967, this Minerva figurine ended up again in the Tongeren museum as a donation from the Begijnhof museum collection from Hasselt.39 Huberts, Liège Shortly after their discovery, four figurines from the Thien S churn find were given to A. Hubens.40 One of the figurines represents a Fortuna; we don’t know much about the other three figurines. The current repository of the objects is unknown. Since we assume that the donation only included double specimens, no iconographical information was lost. It is important for the interpretation of the Thien Schuur find to realise that the number of double specimens is larger than what we can derive from the number of preserved or Georges Meyers, Tongeren Baron Georges Meyers was a committee member of the Société Scientifique et Littéraire du Limbourg.35Shortly after the discovery of the Thien Schuur terracotta find in Tongeren, published specimens. Bust TO 68 and Fortuna TO 14 probably belong three figurines from this collection were given to Meyers. Lucien Renard published them in Weerd in 1932. It is therefore plausible that two of the four double specimens belong to the donation to Hubens. 1905 as property of G. Meyers, without mentioning the site.36Only in 1932 Hubert Van to this donation. They were portrayed by Paquay in 1913, but not described by Van de de Weerd made clear that they belonged to the 32 33 34 De Schaetzen & Vanderhoeven 1956. De Schaetzen & M. Vanderhoeven 1956; Anon. 1958,50 Van de Weerd 1932, 279, nt. 4: een Fortuna aan wijlen Rob. Christiaens, broeder van den eigenaar der hier besproken verzameling”. 35 Anon. 1958,43. 36 Renard 1905. 37 Van de Weerd 1932,279, nt. 4. 38 Van de Weerd 1932,279, nt. 4: “een Minerva werd geschonken aan den h. Gilkens van Hasselt”. 39 Inventory book from the Begijnhof collection, preserved in the Gallo-Roman Museum o f Tongeren. 40 Van de Weerd 1932, 279, nt. 4: “Vier dubbelen, waartusschen een Fortuna behooren toe aan den heer A. Hubens, apotheker, te Luik”. - 2 8 - C o lle c tio n N u m b e r in th e T o n g e re n catalo gu e Société Scientifique et Littéraire du Limbourg 6,92 Oscar Schaetzen, Tongeren 2 François Huybrigts, Tongeren 79.85.87,90,94 Philippe de Schaetzen, Tongeren 7.49.54,55.77.78 ,97 Christiaens - Vanderyst, Tongeren 95 Mathieu Christiaens, Tongeren 3, 4, 8,9,11,12,13,14,15,16,17,18,19,20,21, 22,25, 26, 28,30,31,32,34,36,37,39, 40,41, 42, 44,45, 46, 47, 48,51,52,53,58,59, 60, 61, 63, 64, 65, 66, 67, 68, 69,70, 71,72 ,73 ,74 , 7 5 . 7 6 , 86 Robert Christiaens, Tongeren 38 Georges Meyers, Tongeren 5) 10, 29 Gilkens, Hasselt 43 Hubens, Liège 68,14 Table 1. Historie collections acquired b y the Gallo-Roman Muséum Tongeren and the catalogue numbers. - 29- The contexts The importance of context in the dating information is acquired from the archaeo­ and meaning of the finds logical context or an individual archaeological feature. Examples are a grave, a certain strati­ In each study o f archaeological finds, the interaction with field research is essential. The excavation provides information on the origin and spread of a series, its chronological evolution and on the function of the figurines. graphic layer in a temple, or a waste pit in a residential area. A consistent distinction between the site, the find location and the context provides a solid The value of a chronological or functional interpretation of an archaeological object often depends on the quality of the excavation and its report. The more detailed the information, the more valuable the inter­ pretation will be, depending on whether the object is known to be found in a site in general, or in a certain location on the site, or in a specific context on that location. Differen­ tiating between these three levels is para­ mount. The site has the widest chronological limits and usually includes several areas that would have experienced their own functional and chronological evolution. For example, a certain area of the site can be interpreted as a sanctuary, another as a quarter for artisans, a residence, or a burial site. Each of these specific locations can have a more limited chronological definition than the site in its entirety. The most accurate and detailed -30- basis for chronological criteria and enables the appropriate assessment of the specific values used to date an individual figurine. The better we know the site in its chronological and spatial development, the more valuable the information it provides on the finds. Site, location, and context are firstly spatial terms. In contrast to this spatial aspect, the chronological and functional data of a site or location are not necessarily constant data. It often happens that a certain area was reorganized in the course of the development of a settlement and that its function changed. For example, a potter’s quarter could be reutilized later as a cemetery. Scientifically precise interpretations require emphasis not only on the site, but also on the individual contexts. Without a good knowledge of the contexts the researcher runs the risk of erroneous interpretations for both the chrono­ logy and meaning of the find. Research into the specific contexts of figurines seeks to provide answers to questions concerning their use and meaning. Although much has been written about terracotta objects, and on the other hand, the age of the objects at the time of the deposition has to be indicated. These are two different aspects that must remain separated as much as possible. figurines in the last decades, work on their particular find contexts has been widely Both criteria are connected. The time of the deposition provides a terminus ante quem for the age of the objects, whereas the youngest object from the context provides a terminus post quem for the time of the deposition. We ignored. This is most notably due to resear­ chers placing most of their energies on the typological and stylistic characteristics of tried to enter both dating aspects, with the first figure group representing the age of the oldest objects from a particular context, and figurines. Nevertheless, the context provides a solid foundation for contemporary interpreta­ with the second figure group indicating the upper limit for the deposition of the objects. Entering AD 50/110-130/140 for a context tions of terracotta use in a particular region. The figurines recovered from systematic means that at least a part of the objects occur both their occurrence and use. in a period between 50 and 110, while their deposition is situated between AD 110 and Placing the context in the wider framework of 130/140. When the research on the figurines provide excavations enable a better understanding of site, offers the advantage of a more accurate characterization of the historical context, dates that contradict the accepted date of a context, or that are supplemental, this is entered after the description under the particularly in terms o f the level of Romanization, urban development and heading interpretation. A corrected dating of the context was entered functionality of the place. i f the evaluation of the research into the terracotta series could supplement the dating The dating of a site or a specific location is of the context. This dating is the result of the rarely absolute. It is an illusion to think that the start or end of the period of occupation of research into the series and is not used when dating the series. It does, however, provide a site can be determined up to the calendar year. This was taken into account in entering supplemental information on the context. the dating of sites and locations: the first group of figures indicates the start date and Roman Tongeren: historical sketch the second group of figures, the end date. A slash indicates uncertainty. In that way, a dating of 90/100-200/220 AD indicates a start antiquity as Atuatuca Tungrorum,' dates back to around AD 10 and frames the administrative between 90 and 100 AD and an end between location along the Roman road from Boulogne to Cologne was selected as the principal town of the civitas Tungrorum. In the beginning, the military took responsibility for the organi­ the find location, and generally within the 200 and 220 AD. At the level of the context, the breadth of variation of the first group of figures that indicates the start date is usually broader. Dating closed contexts always has a double character. On the one hand, one must try to represent the time of the deposition of the The foundation of Roman Tongeren, known in organization of Gaul in the Augustan era.1 A zation and designed a regular street grid. Rather quickly, from the late Augustan era, the first inhabitants from the surrounding area settled in stable-houses, divided in a living and stabling section, uniting people and Raepsaet-Charlier 1995,43; Vanderhoeven 2004,481: the name is derived from various records, including Aduaga Tungrorum in the Itinerarium Antonini, Atuaca on the Tabula Peutingeriana, and Atouatoukon according to Ptolemaeus. Vanderhoeven, Martens & Vyndrier 2001,71; Vanderhoeven 2007,311,315, and 324. - 31 - livestock. They were built according to the native tradition, though respecting the original street grid.3Around the middle of the first century other loam and wooden houses had replaced most o f these traditional dwellings, with various rooms, organized around a central courtyard. The early building history shows the fast adaptation of the native inhabitants to the Roman lifestyle. Romanization was a process of articulation of both Roman and native values.4 In this respect, the transition from Roman military to native civilian population is demonstrated by the shift of the import of kitchenware from Mediterranean areas to the use of regional kitchenware and in the dietary pattern.5 occasion is therefore associated with the Batavian revolt in 69/70 AD,“ Traces of fire are found throughout the city and provide a good dating criterion. After the fire in the Flavian period, various districts were rebuilt in loam and wood. Little by little, stone was also used as a building material. This evolution continues at various speeds, depending on the district and the function of the buildings." In buildings with artisanal activities, the inhabitants of Tongeren kept to wood construction the longest. Stone foundations were gradually introduced for residential dwellings. An aqueduct that entered the city from the In the northern part of the town, which also has the highest elevation, a monumental north guaranteed the water supply. Its construction is dated to the first half of the first century.67A part of its route within the city presumably was a district with a public character to the south of the temple, which is was identified in excavations along the ‘Elisabethwal’.' evidenced by the dense spread of monumental architectural elements and fragments of Extensive cemeteries gradually stretched along the large exit roads to Bavay and Cassel in the southwest, and in the direction of Nijmegen and Cologne in the north and east. They remained in use from the first century to the first decade of the fifth century.8 temple was built in the second century. There sculptures." The construction of the medieval wall in the thirteenth century badly disrupted the zone, hampering the interpretation of the remnants of the buildings. The central area of the town at the northern slope of the Jeker valley is better documented. The excavations of the last few decades have The first civilian settlement that developed at the beginning o f the first century was destroyed during the transition from the Julian-Claudian to the Flavian dynasty.9*This 3 3 5, ‘ 7 ’ 5 ” ” “ ’’ contributed greatly to our knowledge of the housing plans in this part of the urban area.13* Some buildings had a decidedly residential character. Others combined a residential Vanderhoeven 2007,315-317 Fig. 4: farmsteads o f the Alphen-Ekeren type. Vanderhoeven, Martens & Vynckier 2001. See also Derks 1998,70-71. Vanderhoeven, Martens & Vynckier 2001,65; Vanderhoeven 2007,317-319: in the early period, consumption o f pigs largely dominated, which is indicative o f Mediterranean influence. In the second phase, the share o f cattle gradually increases. Vanderhoeven 2007,313. Vanderhoeven 2007,321,323 Fig.11. Lesenne 1975,82; Vanderhoeven et al. 1992,387-402,395; Vanderhoeven 2007,309-311. Vanvinckenroye 1985,40; Vanderhoeven et al. 1992,388-389; Vanderhoeven 2007,315. Raepsaet-Charlier 8t Vanderhoeven 2004: however, the precise circumstances and date are unknown. Vanderhoeven 2007,324-325: only after the second large fire, dated after 152 AD, they began to use more durable materials for the construction o f houses. However in the Flavian period, stone was already used in foundations and parts o f the superstructure in residential quarters. Mertens 1977; Vanderhoeven 2007,310 Fig. 1. Vanderhoeven 2007,327-331: summary o f the housing plans. Until now, around five per cent o f the city area has been excavated. - 3 2 - function with artisanal activities. Intensive economic situation caused a decline in wealth, seasonal crafts such as working horn and glue production were practised in several but in the fourth century, the city experienced a revival. During the first decades of the fourth century a new city wall was built that defined a residential town houses.14 These crafts demonstrate the interdependence of town and country. The lower southern area near the river Jeker is covered by a thick colluvium, and has hardly been examined yet.15 Probably artisanal activities and possibly a small dockland can be expected in the marshy zone around the Jeker. smaller city area.21 It ran around the highest and most easily defendable part of the city. Its course crosses the ruins of various deserted districts. The building stones of the destroyed buildings were systematically recycled for the construction of this new enclosure.22 In the middle of the fourth century, Tongeren was the seat of Bishop Servatius. This fixes the An approximately 4500 meters long city wall was constructed in the second half of the presence of a Christian community in this period in Tongeren.23 The construction of a second century AD.16 There may be a relationship between the construction of this basilica in the fourth century attests these wall and the incursions of the Chauci around AD 170/175.17A second major fire that left traces across a large part of the city was dated after AD 152 and probably testifies to these riots.18 An altar dedicated to Jupiter and the genius of the municipium of the Tungri, dating from the second half of the second or the beginning of the third century, testifies to the city status as municipium Tungrorum.'9There is a possibility that the acquisition of this status is related to the reorganization of public space, such as the construction of the long city wall and the temple terrace in the northern area of the city. events.24 However, the revival was short-lived. A gradual shift in the position of power from Tongeren to Maastricht led to the depopu­ lation of the city after the first decade of the fifth century, caused by the gradual de­ population of the hinterland and the declining importance of the road transport in favour of the rivers.25 The discovery of the figurines: details on the find locations The distribution pattern of the terracotta finds includes various find locations inside and outside the second-century city wall (Fig. 14)Intra muros, many finds were made near the In the second half of the third century a third large fire ravaged the city, which has been northern temple complex, although we cannot associated with the incursions of the Franks.20 prove a direct relationship. The most In this period the general political and remarkable find is that of the ‘Thien Schuur’ “ Vanderhoeven 2007,332-333: refuse pits were found w ith horns o f around 500 cattle near a rich town house along the Elisabethwal. This presupposes the presence o f a workshop for working horn and the proximity o f a tanneryIn the open space behind a house w ith a central courtyard in the Hondstraat, a large number o f pits were found w ith remains o f long bones that testify to systematic extraction o f marrow and marrow oil and the production ot » 16 v 18 ■» bone grease and bone glue. Vanderhoeven 2007,311. Vanderhoeven et al. 2007a, 185-187. Vanderhoeven 2007,309,313. Vanderhoeven 2007,313,324. §2 I , ___ Vanvinckenroye 1994; Raepsaet-Charlier 1995; Raepsaet-Charlier & Vanderhoeven 2004,53. 57; Vanderhoeven 2007, « 311Vanvinckenroye 1985, 63; Vanderhoeven 2007,324. ” ** | Vanderhoeven 2007,309,324. Raepsaet-Charlier 8i Vanderhoeven 2004, 63; Vanderhoeven 2007,325. Raepsaet-Charlier & Vanderhoeven 2004,63; Vanderhoeven 2007,482. “ “5 Arts et al. 2009. Raepsaet-Charlier 2004. - 33 - Kg. 14. The find locations o f figurines in second century Tongeren. some distance to the southeast o f the The north temple26 sanctuaiy. One fragment was found at the north side of the sanctuary, while on the The sanctuary was built on an artificially elevated site against the inside of the northern the city wall. The place is called ‘Ilzerbron’, which ‘Broekberg’ there were various finds. A second concentration is located in the western periphery at ‘A an Paspoel’, where presumably adjacent Roman refuse dump of refers to the wells with ferrous water that originate in the vicinity.27As the proximity of artisanal activities took place. A third major intra muros concentration was in the water was necessary to perform the rituals, it should be noticed that the presence of springs ‘Kielenstraat’, where excavations unmis­ in or near a sanctuary not always implies a sacred or curative significance of the spring.28 takably demonstrated its artisanal character at the time contemporary to the figurines. The finds extra muros are almost exclusively From 1964 to rg67, excavations revealed two construction phases.25 The temple was built according to the Gallo-Roman plan on an concentrated on the southwestern cemetery. Remarkably, these finds are clustered in a very lrmrted zone of this extensive Roman burial site. Until now, no clay figurine was found on elevated terrace and surrounded by a porticus. A number of outbuildings lay at the north side. According to Mertens, when renovating, the cemeteries at the north and east of the city, apart from an isolated burial group at the northeast of the city along the ‘A rmand Meesenlaan’. the original Gallo-Roman temple was adapted to the classical architecture by incorporating a promos within the front ambulatory. There are indications that the area was frequented from the middle o f the first century until the end of the Roman occupation in Tongeren. Mertens dates the oldest features 27 28 So far m Roman Tongeren one temple has been identified w ith certainty. We w ill refer to it as the north tem pi* because o f its location against the northern city wall. Mertens 1967,106, nt. 19: Mertens associates the temple with a text by Plinius that mentions a therapeutic well with ferrous water in the Civitas Tungrorum; Plinius, Nat. Hist. X X X12. Derks 1998,207. Mertens 1967,103 Afb. 2 and 105 Afb. 4. Reconstruction in Vanvinckenroye 1975,70 Afb. 35. Referred to as a classicized Gallo-Roman temple in Derks 1998,148. - 34- to the third quarter of the first century. The archaeological items whose dating extends to Flavian period is mainly represented by finds from the infilling of the terrace on which the temple was built.30 The temple, of which the the middle of the second century.33 Furthermore, the city wall that was constructed in the second half of the second centuiy cuts these buildings.34 If the terrace and the Gallo-Roman temple were built in the middle of the second centuiy or shortly after foundations were excavated, dates from the second century. The fact that the sanctuary was integrated within the area defined by the much smaller fourth century city wall may indicate the survival of the temple in the late Roman period. that, the adaptation to the classical building style has to be dated to an even later period. Indications of the gods that were worshipped The precise dating of the temple still poses problems. According to Mertens, the terrace defined by the porticus, on which the first in the sanctuary are limited. A stone pedestal o f a Mercurius statue surfaced during groundwork on the temple court.35 It was podium temple was built, was constructed at the end of the first or the beginning of the found together with fragments of two sculptures that once crowned columns, showing Jupiter on a horse defeating two second century. The adaptation to the classical building style would have taken place in the second half of the second century.3* It can be argued that the large-scale construc­ serpent-footed giants (Fig. is).36 This monumental representation of the bearded horseman god with a lightning bolt or wheel tion of the terrace and porticus took place in in his hand, vanquishing a giant with a snake’s the second half of the second century. This is demonstrated in the research o f W. Vanvinckenroye in 1979 and 1980 on the tail, is a creation of provincial religious art in Gaul and Germania. The theme is based on the Graeco-Roman triumphal iconography of the ‘Broekberg’, the adjacent area west o f the Gigantomachy, but is new as a way of representing Jupiter. Some authors interpret it temple terrace. This place served as a dump in the Roman period. Both the abundantly present terra sigillata and common pottery determined the end of the refuse dump to as alluding to the emperor’s victories, while for others it signifies a cult in which Roman and Celtic elements are blended.37 between 135/140 and 150/160 AD.32The levelling package that covered the Roman dumping The layers over at least 5 metres was cut by the terrace wall of the sanctuary and stretched out behind that wall, in the terrace itself. Therefore the construction of the terrace must be dated to the middle of the second century AD or later. A similar picture is provided by plinth of a terracotta figurine representing Juno (TO27) was found during the 1965 excavation of the northern temple as a stray find. An amateur collected the object in the soil thrown up from the trenches, by the northern outhouses near the city wall.383 9The the infilling layers that covered the older buildings on the slope during levelling of the figurine was later sold to the Gallo-Roman museum and published under the general entry ‘Bilzersteenweg’.33 The fragment is area north of the temple. These also include signed by the Cologne coroplast Servandus 30 31 32 33 34 35 36 37 38 39 Mertens 1967,104-105. Mertens 1967,103-105 Afb. 4, in particular p.105. The City o f Tongeren conducted new excavations on the temple site in 2012 (Peter Cosyns). Vanvinckenroye 1989,9-11, Afb. 1C , 55-57. Mertens 1967,104. Vanderhoeven et al. 2007a, 85-87. Cahen-Delhaye & Smeesters 1981. Noelke 1981,490. Bauchhenss 8r Noelke 1981; Van Adringa 2002,190-191; Kemkes en W illburger 2004 31-33. Zie voor de civitas Tungrorum: Raepsaet-Charlier 2003,53 en Raepsaet-Charlier 2007,444. Found by P. Jadoulle, info. J. Box. Despriet 1974,53-60 Fig. 1. - 35- and therefore dated in the second half of the second century.40 Chronologically, it is contemporary with the functioning of the sanctuary. The place is located southeast of the north temple. A street that was part of the Roman urban grid, ran in its vicinity.43 In 1966, immediately to the south of the find location, opposite the ‘Hemelingenstraat’, a massive foundation of a major building was found.43 According to Mertens, it is possible that this was part of a second temple. New excavations in 1994 immediately to the west of the figurine Fig. 15. Jupiter on a horse defeating two serpent-foo­ ted giants from the Tongeren north temple. deposit confirmed the official character of the zone, when the remnants of a construction with apsides were discovered on the corner of the ‘Hemelingenstraat’ and ‘Pliniuswal’.44 There may be a connection with the foundation found earlier. Judging from the size and permanency of these constructions the sector must have had a rich residential or public function. As on the site of the northern temple, a fragment of a Jupiter column was found here in 1934.45 The interpretation of the archaeological remains in this zone is gravely hampered by The ‘Thien Schuur’find major disturbances in the post-Roman era. Because of the construction of the thirteenth century city rampart and its demolition in the The triangle between the current ‘Hemelingenstraat’ (formerly ‘Hasseltsestraat’), ‘De Tieckenstraat’ (formerly ‘Thien Schuur’), and nineteenth century, a large part of the Roman layer in this part of the town has disappeared. ‘Pliniuswal’ is the location of a substantial find Excavations in 1994 on the corner o f the of well-preserved terracotta figurines at the ‘Hemelingenstraat’ and the ‘Pliniuswal’ revealed that the ancient occupation layers from the second, third, and fourth century have not been preserved.46 turn of the nineteenth century (Table 2). Until now it was referred to as the find of the ‘Hasseltsestraat Tienden Schuur’ or ‘Hasseltse­ straat depot’.41* Adjoining the former ‘Thien Schuur’ street, the find is further on referred to as the ‘Thien Schuur find’. “ 41 See the chapter on the coroplasts. Van Boekel 1987, 679; Schauerte 1985,342, De Beenhouwer 2005,271. 41 43 44 This was established during public works on the Tongeren sewer system in 1934 and 1935: Paquay 1935,21: “een weg m de nch tin g der huidige Hasseltsche straat. Bij den ingang der straat is de w eg waar te nem en m idden de straat, verder wijkt hij a f in de richting van ‘thuis Mathieu Christiaens, nr. 47 (C 92tg), voorbij de kromming der straat.” Mertens 1977,146-147 Afb. tan d 3; Mertens 1984,46 Fig. 6. Vanderhoeven & Vynckier 1994,75-83, Fig. 1 and 2. 45 46 Paquay 1935,74: Hasseltsche straat tegenover huis nr. 3, C 948d; Vanvinckenroye 1985,114. Vanderhoeven en Vynckier 1994,75. - 3 6 - Fig. 16. Map showing the position o f the north tem­ ple, the major building with apsis (in brown, left) along the ‘Hasseltsestraat* and the terracotta finds in this area. Around 1900, Mathieu Christiaens found a large number of well-preserved terracotta figurines on his property near the ‘Bilzerpoort’. The distribution of a part of the find across various private collections, and the publish and depicted three specimens of the collection of baron Meyers in 1905.47 Apparently, Christiaens had given away various specimens to this prominent resident often incomplete, sometimes even contra­ dictory reporting, has obscured the picture of Tongeren almost immediately after the find. It is Jan Paquay who, in 1913, was permitted to publish an illustration with a selection of 38 until now, even concerning the number of figurines. Collecting the dispersed short specimens from the Mathieu Christiaens collection.48 entries of the find in the literature made it possible to reconstruct the facts to come to a better understanding of the size, location, and nature of the find. In the first few years after the find, little publicity was given to the discoveiy. Lucien Renard, who could not appreciate the silence that surrounded such an important discoveiy, by-passed Mathieu Christiaens’ refusal to 47 48 49 50 51 Extensive publication at last followed in 1932 and 1933 by Hubert Van de Weerd. At that time the collection still numbered 55 figurines.49In the years before, Christiaens had given away at least nine double specimens.50Altogether, the original find therefore numbered at least 64 specimens.51 In total, 52 figurines of the original find are preserved in the Gallo-Roman Museum of TO 10,5, and 29: see also the Meyers collection. Paquay 1913,94-95,2 photos between 88-89. Van de Weerd 1932; Van de Weerd 1933. According to Van de Weerd, the collection numbered 54 specimens at that moment, but he describes 55. Van de Weerd 1932,279, nt. 4; also see the collections o f R. Christiaens, G. Meyers, A. Hubens, and Gilkens. See the collections o f M. Christiaens-Peters, R. Christiaens, G. Meyers, A. Hubens, and Gilkens. - 37 - Fig. 17. Part o f the Thien Schuur find first published by Jan Paquay in 1913. Tongeren.5 1*53 Only a photo remains of nine missing figurines.53 Of three other missing figurines, no image is preserved. Van de Weerd mentioned one of these lost fragments.54 The two other missing figurines belonged to the Huberts collection. Because they were double specimens, it can be assumed that no iconographical information is lost. 51 a “ !! 56 The first reference to the find by Renard in 1905, shortly after the discovery, neither mentions the find location nor the owner.55 Renard, who had not seen the entire find at that time, did not obtain permission to study the figurines. From the 1932 publication by Van de Weerd we do learn that the figurines of the baron Meyers collection he describes, were also part of the find of Christiaens.56 W hen the province o f Limburg acquired the collection in 1953, there were only 51 figurines left. The numbering in the inventoiy book o f the ‘Oud Fonds’ o f the Gallo-Roman Museum precisely follows the order o f the publication o f Van de Weerd: inventory numbers 2001 to 2051. Since the inventoiy, one Minerva figurine has disappeared: inventory number 2019 is m issing (presumably TO 37). Two o f the figurines, once given away by Christiaens, eventually ended up in the museum (TO 12 and 43). This concerns three figurines from the Meyers collection, nos. 10,5,29, and nos. 59,60,16, and 37 and two figurines that were possibly included in the donation to A. Hubens, No. 68 and 14. Van de Weerd 1932,290: this concerns the same head as o f TO 28 w ith traces o f brown painting. Renard 1905,285-286: “... récemment encore, une trouvaille, sans égale peut-être dans nos contrées, y a été faite. Il s’agirait de la découverte d’une ‘boutique de figurines en terre cuite’ dont on aurait retiré, indépendamment de quelques statuettes, une suite curieuse de petits bustes. Il est regrettable que le propriétaire actuel de ces objets n’ait pas cru devoir en autoriser, dans un intérêt scientifique, la publication. On doit donc se borner aujourd’hui à mentionner cette trouvaille sans plus de détails.” Van de Weerd 1932,279, nt. 4. -38- J. Paquay is the first to mention the name Christiaens in 1909. This author also provides the first indication of the location near the ‘Bilzerpoort’.57 In 1934, Paquay gives an accurate cadastral description o f the find location that matches the division o f the cadastre around the turn of the century.585 9In 1902, these parcels were in the possession of the Christiaens family.55 They included both the garden and an outhouse that were adjacent to the then ‘Thien Schuur’ street.60It is clear that Mathieu Christiaens became the owner of a collection of terracotta figurines that he himself had found around 1900 in his own rather a description of the condition of the collection in 1932. Although it is nowhere explicitly said that the objects were found in a closed context, various indications sustain this proposition. On the one hand, there is the large concentration of similar objects and, on the other hand, there is the large profundity in which the objects were found. In this area of the city, on account of mediaeval groundwork and recent sand extraction, the layers from the second, third, and fourth century have not been preserved. Therefore, the original depth of the pit must have been even more than 2 metres. garden or outhouse adjacent to the ‘Thien Schuur’ street. ‘Broekberg’ For many years after the discovery, nothing ‘Broekberg’ is the name of the slope that marks the steep boundary of the city plateau at the was mentioned about the circumstances of the find that could explain its nature. About 30 years after the find, Van de Weerd provided a concise description: “The figurines were found at a depth of about 2 metres. They lay disorderly as if they had been emptied hastily from a basket or barrel. Some figurines were intact, but most of them were broken. Of northeast side of the Roman town. From the Roman wall and the lower ‘Broek’ it climbs into the city, along the current Hasseltsesteenweg. The find is located inside the Roman city wall at the edge of the north temple complex. Although the archaeological observations are sparse and the comprehension is still incom­ some, there was only debris. Other objects that could have shed light on the date or of the plete, in the early Roman period parts of the place were used for sand extraction. Mainly after the Batavian uprising a dump was built nature o f the find were not discovered or neglected.”616 2These data are of a general nature. on the slope. A special find that surfaced here in 1934 is a series of wooden writing tables.61 The first two sentences show that Van de Weerd interviewed eyewitnesses about the circumstances of the find. What follows is 57 58 59 60 61 62 Hoping to enlarge his personal collection, baron P. de Schaetzen intensively searched the Paquay 1909,51: “Récemment encore, M. Christiaens, ingénieur-architecte, a trouvé a Tongres, aux abords de la Porte de Bilsen, section, C 921E, plusieurs statuettes de divinités provenant du lararium d’une maison privé.” Paquay 1934, 66-67 and Fig. 7: “54 statuetten afkomstig van perceel C 921e en f ”. Department o f spatial planning o f the City o f Tongeren: cadastral data associated with Jan Jozef Mathijs Christiaens-Peeters, structural engineer in Tongeren. The residence o f Christiaens bordered the ‘Hasseltsetraat’ (then lot C92ig), while his garden and an outhouse bordered the ‘Thien Schuur’ street (then lots C92ie and f ).. There is no connection w ith the tith barn as Van de Weerd was led to suspect, indicating ‘Tiendenschuur’ as the find location o f the figurines. See Van de Weerd 1932,279: “...in de Hasseltsche straat, op de plaats genaamd Tiendenschuur”. Old map sheets show that ‘Thien Schuur’, was the name o f the street adjacent to the garden o f Christiaens and is now called ‘De Tieckenstraat’. This street was named after the tithe barn, which, at least until 1864, remained preserved opposite the street. For the location o f the tith bam , see BaiUien 1978,181: “daar waar het huis van de notaris Hougaerts (1977) staat”; Baillien 1995,131. Van de Weerd 1932,280: “De terracotta’s werden op een diepte van ongeveer 2 m gevonden en lagen ordeloos dooreen alsof ze haastig u it een mand o f vat waren uitgestort. Enige beeldjes waren gaaf, de meeste waren gebroken, van enkele kwamen slechts brokstukken te voorschijn. Andere voorwerpen, die licht hadden kunnen werpen op den datum o f de natuur der vondst werden niet ontdekt o f werden veronachtzaamd”. Breuer & Van De Weerd 1935,495-496 Pl. XLI; Paquay 1935,74; Van De Weerd 1944,132,352 Pi. XXVII; Vanvinckenroye 1985,114. -39- area between 1934 and 1949.63In addition to the figurine fragments mentioned below, he found hundreds of decorated terra-sigillata sherds and stamps. According to de Schaetzen, the place was run as a sand quarry from the middle of the first century AD, after which the pits were filled up again with debris mixed with pottery (Table 2).64 Terracotta fragments were found on locations at both sides o f the ‘Hasseltsesteenweg’, named ‘Broek I f and ‘Broek I lf after the records of de Schaetzen. Based on archaeo­ logical observations on a parcel656bordering same Fortuna ‘Kielenstraat’.69 was discovered in the ‘Aan Paspoel’ The toponym ‘A an Paspoel’ covers an area in the southwest of the city, near the southwestern cemetery, both inside and outside the second century city wall. Between 1936 and 1944, Baron P. de Schaetzen excavated two lots, situated between the large second century wall and the smaller fourth century enclosure.70 So the find locations were ‘Broek Ilf, W. Vanvinckenroye reached the conclusion that the place was used as a dump until between 135/140 and 150/160 AD.“ The enclosed inside the city area in the second half of the second and in the third century, thus excluding the occurrence of graves during that period. supply of waste came to a sudden end, which may be related to the construction of the terrace of the north temple. Though the context o f the figurines is unknown, we can try to picture the nature of the place using the objects collected by de In 1935, a fragment of a Minerva figurine was Schaetzen.71 According to his records, the finds count a number of intact vessels and found during work on the sewer system in the ‘Hasseltsesteenweg’ opposite the ‘Keverstraat’ (TO93). The museum inventory book mentions numerous sherds of decorated terra sigillata. Additionally, many bronze objects were found, “Hasseltsesteenweg opposite the Keverstraat, left trench”. This position corresponds to the street rim that edges the find location ‘Broek including a Mercurius figurine and coins. Since the area is inside the second century enclosure, it is not likely that the objects come If, examined by de Schaetzen in the same period.67 from a funerary context. Apart from the ‘Broek Ilf is located between the Hasseltse­ steenweg and the temple terrace.68The back of a Fortuna figurine and fragments of a bust were found here (TO 7 and 77). The front of the * 64 65 66 luxury goods, remains of artisanal activity were recovered. Pieces of carved bone and a large number of polished bone needles, in addition to partly polished specimens, indicate the vicinity of the workshop of a bone craftsman. Baron de Schaetzen & Vanderhoeven 1964,8: mention o f the precise lots; Schaetzen & Vanderhoeven 1953/1954,7: “...emplacement qui, en partie, a été exploité comme carrière de sable et remblayé déjà à l’époque romaine, nous a fourni 780 tessons ornés...” Anon. 1958, 68; Vanvinckenroye 1989,11. lots A526b2 and C 2 , bordering ‘Broek HT 68 69 See the north temple. The Roman soil formed a humous package o f successive layers holding both building waste and household waste. It contained burned loam from walls, tiles, stone waste and charcoal, next to large quantities o f animal bones, oyster shells, mussels, and especially many glass and pottery sherds. Then Lot A537. In 1936, during excavations on the site, denoted by de Schaetzen as ‘Broek IT, along the Hasseltsesteenweg on lot A537U2, a fragment o f a terracotta mask was found. See De Beenhouwer 2005, No. 813. Then Lots A5250 and q. Archive o f the Gallo-Roman Museum o f Tongeren. See ‘Kielenstraat*, stray find 1935. 71 Then Lots 0225a and 223d. Baron de Schaetzen & Vanderhoeven 1964, 8; Schaetzen & Vanderhoeven 1953/1954,7: part o f the site was systematically examined; 250 decorated terra-sigillata sherds were found. Anon. 1958,67-68. 67 -40- The inventory book of the P. de Schaetzen collection gives four fragments of figurines as coming from “Mon terrain 46, Chaussée romaine Tongres” and a fifth fragment “Mon terrain aan Paspoel, Chaussée romaine Tongres” (TO 49,54, 55,78).72Combining both locations leads us to the conclusion that this is the same place owned by P. de Schaetzen, on the street named ‘Romeinse Kassei’ in the area ‘Aan Paspoel’.73 The so-called “Chaussée romaine” does not refer to the Roman BavayCologne road but to the current street named ‘Romeinse Kassei’. Five fragments of figurines were found here between 1938 and 1944 (Table 2). They are representations of Venus, Apollo, a horseman, and a bull. The fifth fragment, probably part of an animal, has gone missing. These terracotta finds are dated after the middle of the second century by the series to which they belong.74 ‘Kielenstraat’ Between 1986 and 1995, the bitants settled in traditional stable-houses. One of these houses is distinguished by the absence of the stable section and the presence of two wooden cellars. A ritual deposit was hidden in one of the postholes. According to Vanderhoeven, these distinguishing features indicate a social hierarchy within the group of inhabitants.76 In the third phase, during the Claudian period, the street was gravelled for the first time and the traditional stable-houses were replaced by a large timber construction with rooms arranged around a central court, influenced by the Mediterranean architecture. Under the reign of Nero, a number of renovations took place, after which the courtyard-house burned down by a large fire, linked with the Batavian revolt around AD 69/70. The design of the residence and the presence o f wall painting refer to the well being and increasing Romanization of the native inhabitants. Flemish The fourth phase covers the Flavian period and Archaeological Service, systematically exa­ mined the site on the corner of the ‘Kielen­ straat’ and ‘Predikherenstraat’.75 The excava­ the first half of the second century. After the fire of 69/70, again timber houses were built, although elsewhere in the city in this period tion brought to light a street oriented north by northwest, with occupation features on both the use of stone was gradually introduced. In the fifth phase, in the middle and third quarter sides from the last decade of the first century of the second century, continuing the plan of the previous timber constructions, stone pillars were added, using recycled building BC to the second half of the third century AD. Six occupation phases cover three main periods, marked by the three large fires that ravaged the city. In the pre-Flavian period, the place went through an evolution, as found in several areas throughout Tongeren. The oldest structures mainly consist of pits and ditches, following the orientation of the street grid. They bear witness to the military presence in the first decade BC. In the second phase, during the late Augustan and Tiberian reign, native inha­ stones, and stone foundations were planted under the wooden foundation beams. Around AD 170/175, a fire raged on both sides of the street and destroyed the last wooden buildings. This catastrophe is probably related to the incursions of the Chauci, the burnt layer of which appeared here for the first time in Tongeren.77The division between the fourth and fifth phase is rather vague. At an architectural level, there seems to be continuation. Artisans among which were metalworkers, operated 72 Inventory o f P. de Schaetzen, preserved in the Gallo-Roman Museum o f Tongeren. ” Also see the collection o f P. de Schaetzen. 74 Series 30 after 150 AD, series 35 after 150/180 AD, and series 36 after 180 AD. ” Vanderhoeven et al. 1987; Vanderhoeven et al. 1991; Vanderhoeven et al. 1992,90 Fig. 1: localization o f the excavation pits 19 and 20); Vanderhoeven, Martens & Vynckier 2001, 61; Vanderhoeven 2007,311,314-3*7 K g. 3-4 and 313 Fig. 15. 74 Vanderhoeven 2007,317. 77 Vanderhoeven et al. 1991,113. - 4 1 - here in both occupation phases. The remains of their activity mainly concern a group of ten kilns stretched along the street, a few pits with bronze slag and a water basin spread with clay.78 The function of six square pits with a depth of three to four metres, arranged around a wooden cellar is still unclear; they may be cesspits. Large quantities o f cattle offal indicate a butcher’s enterprise.79From an urban point of view, the district held on to wood and earth construction in the period between the large fires of the first and the second centmy. The consumer behaviour is marked by the relative scarcity of terra sigillata and glass.80 group of craftsmen.83A fragment of a goat (TO 81), was found in a square pit that contained burned loam and roof-tile fragments. The pit belongs to the phases 4 and 5, dated between AD 69/70 and 170/175. The bust of a wreathed boy (TO 62) was found in an infilling layer of a pit, at the edge of a stone construction from the last quarter o f the second and third century, near a building from the second h a l f o f the second centuiy. The rectangular pit was 1.5 metres long and 90 centimetres wide and its remaining depth was 1 metre. Above the pit, the soil had already been dug off two metres to construct a post-mediaeval basement. At the time, the walking level of the During the sixth and last phase the first stone buildings appear. They were constructed at the end of the second century and remained in use stone-construction phase was at least 1.25 metres higher than the top level of the remains into the third century. Presumably, they were also destroyed by fire. This can be observed by the walls were almost vertical. The tip of two heavy posts in the corners at the edge of one fire waste in a few third-century pits and wells. The floor level of these houses has not been short side did not reach the bottom level of the pit. They were presumably set up after the pit preserved. Even the foundations were removed in the course the fourth century. After the third century, the place was left fallow for a had already been partly filled up. The dating of the series that includes the bust demonstrates long time. After the construction of the smaller fourth-century city wall, it was no longer part of the urban area. In the twentieth century, figurines were found in this area at various occasions. Two fragments of the front of a figurine were found in 1935 during work on the Tongeren sewer system (T07).8‘The back of the same figurine is found on the ‘Broekberg’.82The dating of the series that includes the Fortuna figurine demonstrates that it belonged to faze 6, dated between AD 69/70 and 150. Six other figurines were found during the excavations conducted by the Flemish Archaeological Service (IAP), five of which in contexts that mainly belong to the fourth and the sixth phase, when the place was nm by a* of the pit. The bottom o f the pit was level and that it belonged to phase 4, dated between AD 69/70 and 150. A figurine depicting Fortuna (TO 23) was found in an occupation layer, just above the floor level of a timber construction from phases 4 and 5. There was a small oven in the same room. The layer was dated to the last quarter of the first century until before the fire around AD 170/175. Furthermore a handmade bird (TO 89) was found in an infilling layer above some subsided pits. The layer contained many roof-tile fragments. It is dated to the end of the second centuiy. A wheel-turned bird (TO88) was found in an occupation layer from phases 4 and 5, dated between AD 70 and vo/vs- And finally a figurine of a hen (TO 83) was found as a stray find. The dating of the series to which it belongs suggests that the figurine belonged to phase 4, dated between AD 69/70 and 150. 75 “ Vanderhoeven et al. 1991,111,115 Fig. 9 No. 8; Vanderhoeven et al. 1987,133 Fig. 8,13. Vanderhoeven etal. 1987,137 Fig. 8,134; Vanderhoeven etal. 1991,121. Vanderhoeven et al. 1991, 121. ” Found on the 25tho f October 1935, ‘Kielenstraat’ 16, at a depth o f 2 metres. See ‘Broek III’. * Vanderhoeven et al. 1991,121; letter o f G. Vynckier 15/2/99. -42- ‘Sint-Truidersteenweg’ Excavations in 2000 and 2001 revealed the and fourth-century wall.91 A small-scale excavation in 2004, affected the middle of a remains of timber constructions that were block, following the orthogonal plan of the covered by a north to south oriented street.84 street grid.92 Pre-Flavian remains were not To the west of the street and in the same found. They may have been destroyed as a orientation, stood a building with stone result of later construction activities. The foundations. The place is located in the occupation is recorded from the end of the southwestern area of the Roman town, within first until the third century. An isolated sherd the second-century city wall.85 from the fourth century is a silent witness to Refuse pits were found throughout the place. In the space between the stone construction the human activity in the late Roman period. In an occupation phase from the end of the and the street, two cesspits were excavated; first and the first half of the second century, they probably succeeded each other in time.86 five refuse pits were found.93 Three of them In the infilling layers above the subsided fill of were completely excavated. They had been one of these pits, a coarse pot covered with used simultaneously.94 The research into the glass paste on the bottom indicates nearby glass production.87 On a lower level, a Mercurius figurine (TO 56) was found.88 It was carefully hidden under a large fragment of a animal and plant remains in the refuse pits provides a good picture of the nature of the occupation in this period. Next to bones of pigs, sheep, and cattle, a large volume of field roof tile or tegula in one of the upper layers of the cesspit. The layers on top included objects and grass plants is considered the waste from burnt hay.95This implies the presence of cattle that cover the entire second century.89 in the vicinity. In the Flavian period and the However, it is not probable that the figurine first half of the second century, this part of was part of these layers. The density of the ground around the figurine being much looser, this led part of the context to collapse before the section could be drawn. From the density of the soil and the special position of the figurine beneath the tile, it seams the figurine was buried separately at a later time90. ‘de Schaetzengaarde’ The place is located in the southwestern part of the Roman town, both within the second ® * 89 25 ” 18 89 90 91 92 93 94 95 96 town apparently had a rural aspect. A fragment of a figurine, depicting a horse or cattle (TO 82), was also found in one of these pits. The representation clearly matched the interest of the local inhabitants. The plinth of the quadruped was found in an oblong refuse pit.96The fill consisted of two layers of sandy loam of equal thickness. In addition to bones from cattle, pig, sheep and fish remains, the bottom layer contained a lot of botanical remains, in particular barley, spelt, wheat, Excavations by the Tongeren Archaeological Service and the Flemish Archaeological Service (IAP) Vanderhoeven, Van Rechem & Vynckier 2003,75-76; Van Rechem & Vynckier 2006,57-60. Vanderhoeven, Van Rechem & Vynckier 2003,75. Vanderhoeven, Van Rechem & Vynckier 2003,76: according to the authors, the dating entered here in the fourth century has to he revised; Van Rechem & Vynckier 2006,60. Sint-Truidersteenweg trench 001, trace 337, layer 391: Cadastral lot D215S Information G. Vynckier. Information G. Vynckier; Van Rechem & Vynckier 2006, 60: “mogelijk gaat het om een intentionele depositie uit de fase van het dichten van de put”. Tongeren dept. 1, section I)252t. Excavation conducted b y the Flemish Archaeological Service (VIOE), in collaboration w ith the Tongeren Gallo-Roman Museum. Vanderhoeven et al. 2007b, 159-161, Fig. 2-3: location plans. Vanderhoeven et al. 2007b, 162 fig .4 ,165-166. Vanderhoeven et al. 2007b, 172. Vanderhoeven et al. 2007b, 174,176-177. Vanderhoeven et al. 2007b, 166. A pit measuring 80 x 180 centimetres and w ith a level bottom at a depth o f 40 centimetres. - 43 - pulses and nuts, together with field weed and meadow plants.37 Apart from the figurine the upper layer held a piece of Egyptian blue pigment, as another exceptional find. Stratigraphically, the pit is older than the stone baths and held a lot of ceramic finds, mostly tableware, kitchenware, and storage pots, dating from the second half of the first and the first half of the second century. In the second half of the second century and in the third century, the view of the area changed. In that period monumental baths were built on this location.38 Old finds from the wider surroundings indicate that the ‘de Schaetzengaarde’ zone in this period belonged to a monumental and luxurious district, possibly with a public function.33Around and outside the ‘Sint-Truiderpoort’, remains of side the smaller fourth-century rampart.'" It was intersected by a Roman road running north-south which was resurfaced seven times in all from the early first century to the early third century.'03 Initially, wooden buildings were erected along the road. In the second century, two stone buildings with a stone cellar were constructed on either side of the road. A small oven, possibly for food prepa­ ration, a chalk pit and some hearths attest to the artisanal activities that took place there. The western building had at least three small rooms on the street frontage, which the excavators interpreted as shops. The stone dwellings underwent different building phases, but were abandoned in the course of the third century and demolished. In the fourth century the site lay outside the walled section of the city. sculptures and monumental architecture have regularly been discovered since the nineteenth century, among which an inscription to Behind the western building with its ‘shops’ lay open ground. Three of the pits found here Fortuna, fragments of Jupiter and Juno statues, a bust of a genius, parts of pillars and capitals, and an itinerary.“0On the location of yielded special finds: a stone statuary group of Juno and Jupiter,“4 terracotta fragments of a Mercurius figurine (TO 57), and a fragment of the baths a gold ring with the portrait of emperor Commodus was found (AD 180-193).“' Clearly, the 2004 excavation demonstrates that the rather rural character of the occupation in the Flavian period and first half of the second century differs fundamentally from that of the environment of the later bathing facilities. ‘Elfde Novemberwal’ a glazed flask in the shape of a monkey (TO 92). The stone sculpture group can be dated to between the last part of the second to the first half of the third century“5 Judging by the pottery the pit, in which the Mercurius figurine was found, was dug in the second half of the second century or later.“6 The monkey was found in a pit with an irregular shape.'07Its stratigraphic position in the topmost The area excavated in 2008 was located insidp excavation plane, suggests that the feature the large second-century enclosure but out­* could date to the third century. Vanderhoeven et al. 2007b, 171-172,175-177 Vanderhoeven et al. 2007b, 160-165. Vanderhoeven et al. 2007b, 178. Vanderhoeven et al. 2007b, 178 nt. 31-33. Vanderhoeven et al. 2007b, 177; Sas & Creemers 2002,121-123 Excavation NV Anicius TO 08 AN, situated at Tongeren lots, 7th division, sections A 363ms and 363I5. NV Anicius is a partnership project between De Lijn, the Flemish Community and the City o f Tongeren Borgers et al. 2009. 6 Find number 041, pit 3, plane 2, feature 8. Borgers 2009,34-35 and Appendix 5: identified by T. Panhuysen. Pit 03, plane 03, feature 116. The oval pit had a width o f 3.6 m. It had an irregular cross-section and a stratified fill. A Mercurius figurme was found in layer 116.2, next to a fragment o f a Dragendorff 27 bowl in terra sigillata, a fragment . “ ” se pot 311(1013 dun-walled colour-coated beaker related to Brunsting 6b (IIB-IIIA). The layer below, 116 3 yielded among other things a wall fragment o f a pot with a lid seating, whose rim profile matches Gose 506 toe). Pit 03, plane 01, feature 011. The brown-grey sandy fill contained charcoal, mortar, chalk and brick. The ceramics were m ainly made up o f kitchenware, alongside storage vases such as amphora and dolia. Also found were a large number oitegulae fragments, iron nails and two chunks o f lead. A mortarium in terra sigillata dates all o f these rinds to after the middle o f the second century. -44- The southwestern cemetery The cemetery stretched out along the south­ western exit road. Leaving the Roman town, the road subdivided in a north-western branch to Cassel via Tienen and a southwestern branch to Bavay. The necropolis that developed alongside these branches was used during the entire Roman period. Since the nineteenth century, the area was intensively investigated in search for antiqui­ cemetery were discovered on two adjacent lots, located between the current ‘Romeinse Kassei’ and the Roman road to Bavay (Table 2).“ The lots directly border to the west the area that was examined by Vanvinckenroye. Between 1898 and 1906, during the excavation campaigns of F. Huybrigts and the Société Scientifique et Littéraire du Limbourg, four ties. The section of the burial site embedded between the Roman roads to Bavay and Cassel figurines and two unguentaria were discovered. Most o f them ended up in the private collection o f Huybrigts and were heavily damaged by the fire that ravaged his collection is undoubtedly the best documented. In excavation campaigns of the Tongeren museum between 1972 and 1981, W. Van- in 1914.“ Exactly one hundred years later, in 1997 and 1998, the place, threatened by allocation, was re-examined by the Flemish vinckenroye was able to survey 291 graves.“8In archaeological service.“3 Traces from at least two old excavations were recorded. They were small exploration pits from the nineteenth earlier excavations, dozens of graves had already been discovered in the same zone.“9A gravel path ran straight through the mapped area and parallel to it a fencing ditch. Nearly all graves from the middle Roman period (70-275 AD) are located to the west of the path. It may have been the boundary with a residential zone to the east that, from the Flavian period on developed along the Roman roads. Only two pre-Flavian graves were discovered on the century and small-scale trenches from the beginning of the twentieth century that were presumably related to the activities of Huybrigts and the Société Scientifique et Littéraire du Limbourg. In spite of these disturbances, another twenty-two inhumation graves and one hundred and eighty cremation entire area.“ They were located to the east of the fence, in the zone that was used for graves surfaced in the new campaign. Again three o f the graves contained terracotta figurines. In the entire zone, no graves from habitation from the Flavian period on. In the late Roman period (275-4.ro AD), this the third and fourth century were recorded. In general, the burials date from the middle of residential zone was also in use as burial site. In spite of the large number of graves that were the first to the end of the second century. excavated, fragments of figurines didn’t occur or passed unnoticed in these campaigns. In 1898 Huybrigts excavated a grave that held a All terracotta figurines and figurative unguent dog figurine (TO 79), an unguent bottle with a human shape (TO 90), and two coins, one of Trajan (AD 98-U7) and one of Hadrian (AD 117- flasks that were found on the southwestern10 *4 138).“4According to L. Renard, it was a child’s 9 8 108 Vanvinckenroye 1984: w ith general excavation plan attached. 109 Vanvinckenroye 1984,14-15. Between 1931 and 1934, baron P. de Schaetzen found some late Roman inhumation graves. The first systematic research took place in 1957 and 1958, when ten graves were examined by the Belgian Archaeological Service, and forty-four graves b y the excavation service o f the Tongeren museum (Roosens & Mertens 1970); Vanvinckenroye 1963. Thirty-five o f these graves were found on the lots bordering those that produced the terracotta finds by Huybrigts and the Société Scientifique et Littéraire du Limbourg (Vanvinckenroye 1963,19:23 graves excavated on lot B680 and 12 graves on lot B69b3). They date from the middle o f the first to the end o f the second century. 110 Vanvinckenroye 1984, 62,144-145,223, nos. 95 and 289: this concerns the ‘Brandschüttungsgràber’, one o f which was covered by the m ound o f a tumulus, together with the stake. m Lots B67 and B68. m See Huybrigts collection. 1,3 Lots B67f, g, and d , Vynckier and Vanderhoeven 1999. 114 Lot B68c, Grave 5/12/1898. The grave was excavated on 5 December 1898 on Lot B68c. Its inventory belonged to the F. Huybrigts collection. Renard 1905,291 and nt. 1; Paquay 1934,45 No. 103 and 102: unreliable description o f the grave (see Huybrigts collection). - 45“ grave. The idea is probably supported by the fact that the anthropomorphic bottle was not functional. The neck of the vase is not completely pierced. According to Renard, that is why it served as a toy. This conjecture is unfounded. The burial is dated between AD 117 and 200, the coin of Hadrian counting as contained two vessels made of green glass, of which a small unguent bottle, two worn-off coins, and a flat metal box enclosing a small stone. The metal box that produced a clear terminus post quern, and the part o f the cemetery in which it was found, offering a terminus ante quern. period of use of this part of the cemetery, dated between 50 and 200 AD. The series to which the Fortuna figurine belongs allows for a more On the same day in January 1905 Huybrigts discovered two adjacent graves that held terracotta figurines.11516 *In the first one he found a figurine of a pigeon (TO 87). According to L. Renard, besides the pigeon, the grave contained fourteen different vessels, four silver fibulae, and an Antoninus Pius coin (i38-i6i)."s Because of this coin and the dating sound was interpreted to be a rattle. No cinerary run was mentioned. The dating could only generally be determined according to the precise dating of between 140 and 180 AD. The second grave was a cremation burial.”9 According to the report of the society from 1907, in addition to the pigeon-shaped bottle (TO 91), the following were found: a terracotta lamp, a black beaker, three coins, a fragment o f a fibula, and a handful of cremation remains. The concise description o f the objects does not allow a chronological determination of the context. The part of the of the part of the cemetery in which it was found, the grave must have been dug between AD 138/161 and 200. cemeteiy in which it was found, is dated between 50 and 200 AD. In addition to a hen figurine (TO 85), the second grave contained another nine ceramic During the vessels and four large silverfibulae.“7Renard is more precise on two objects: “deux belles Archaeological Service in 1997 and 1998, again three figurines were found."0 A Minerva (TO poteries en terre rouge décorées sur les bords de feuilles de lotus en relief”. This could refer to the forms in terra sigillata Ritterling 12 or Curie i f Drag. 35 or Drag. 36. In spite of this 35) was found just beneath the plough layer. There was no sign of a burial pit, or of other information, it is the particular area of the 50 and 200 AD. The series to which the cemeteiy in which it was found that offers a general dating between 50 and 200 AD. Minerva figurine belongs is dated after the middle of the second century. On behalf of the Société Scientifique et Littéraire du Limbourg, in October 1906, Huybrigts excavated two graves holding a Fortuna figurine and a zoomorphic unguent flask.”8In addition to the Fortuna (TO 6), the first grave “s campaigns o f the Flemish grave goods. The part of the cemetery in which it was found, offers a general dating between A Cybele (TO 1) was found in a cremation burial from the second century. The series to which the figurine belonged can be dated to after 130/140 AD. The hen (TO 84) was found in a cremation burial. The bird was placed on top of the Lot B67C, found on 20/1/1905. 116 Two bronzed vessels were stamped “ ...MANI”, and a terra sigillata plate was stamped “XII”, Renard 1905,298 nt. 5; Bulletin de la Société Scientifique et Littéraire du Limbourg 25,1907,228-229; Paquay 1934,41 No. 61: unreliable description o f the grave (see Huybrigts collection). Renard 1905,299; Paquay 1934,41 No. 60: unreliable description o f the grave (see Huybrigts collection). Bulletin de la Société Scientifique et Littéraire du Limbourg 25,1907,226-227. The grave was found at a depth o f 1.3 metres. Lot B68i, grave 23/10/1906. ” Bulletin de la Société Scientifique et Littéraire du Limbourg 25,1907,228. The grave was found at a depth o f between 70 and 80 centimetres. Lot B68i, grave 27/10/1906. Lot B 67 f, g, and d. -46- cremation remains, and found together with a painted, coarse jar of Tongeren origin, named ‘honingpot’.“ This type of jar was in use from the end of the first century and throughout the second century.“2 The grave did not hold any further gifts. The isolated burial group of the ‘Armand Meesenlaan’ In 1982, constructing a house in the ‘Armand Meesenlaan’, two cremation graves were partly northeastern cemetery of Tongeren.*12 *62812 4 3 9Upon further examination of the building excavations of the nearby construction sites, no graves were found. Previously, the remains of a building complex from the Roman period were located to the west of ‘Hazelereik’, about 110 m north of the graves.“5A second, isolated collection of three graves from the Flavian period was excavated in the ‘Merkenstraat’, at around 120 m disrupted.“3With a mutual distance of about 2 metres, two burials and a pit took shape in the northwest o f the graves of the ‘A rmand Meesenlaan’ and at a distance of around 100 m profile o f the building excavation.“4 from the remains of the Roman buildings. Fortunately, the Tongeren amateurs J. Box and Chronologically, there is a connection between P. Jadoulle, examined and sketched the graves. both isolated burial groups. The relationship Initially J. Box kept the objects in his private between the graves and the remains of the collection and restored them.“5 The Gallo- buildings has not yet been demonstrated. Roman Museum acquired the grave goods in 2007. Size, characteristics and dating of both graves are very similar. All objects can be dated Flavian. One of the graves held a terracotta dog (TO 80).“6 The broken fragments of the dog figurine lay at the edge of the cremation remains, concentrated at the northwestern side of the grave. The grave also contained luxurious ceramics and a bronze fibula.“7 This small, isolated collection of graves, south of the diverticulum from Tongeren to Berg, does not seem to be related to the larger, m m 123 124 “s 126 1x7 128 129 Information G. Vynclder. Vanvinckenroye 1991,77 type 357. ‘A rm and Meesenlaan’ 37. Tongeren 6de afd. Sect. C, lot 292g and partly bordering lot 292f. The trapezoidal p it was 2.2 m wide at the top and 1.4 m wide at the bottom at the height o f the level bottom. Its depth was 1.2 m. The dark grey loamy and humous infilling contained fragments o f roof tiles, carbon, and cremation rest, but no ceramic fragments. J. Box is preparing a publication o f the collection. The rectangular burial pit was 60 cm wide. The bottom was 40 cm under the ground level. O f the actual grave, only a layer 12 cm thick was preserved under the plough layer. It held a fragm ent o f a bowl in an orange-coated white fabric, a cup Drag. 27 and a plate Drag. 18, both in South Gaulish terra sigillata and w ith the stamp OF CALyi. In addition, two objects were not salvaged because o f their poor condition: a fragment o f a bronze fihula and fragments o f a brown jar. Creemers et al. 1991,24-25: B1-B3, B11 and B12; Oswald 1983,55: CA LW S o f La Graufesenque: Nero-Domitian, especially Vespasian. The northeastern cemetery almost reaches the ‘Sint Antoniusstraat’. J. Box examined the building excavations between the ‘Sint Antoniusstraat’ and ‘Hazelereik’, but without results. The exception is a very rich grave north o f the ‘Merkenstraat’, salvaged by J. Box and currently preserved in the Tongeren Gallo-Roman Museum. It m ay have been a tumulus grave. Creemers et al. 1991,24-25. - 47- C o n te x t C a ta lo g u e n u m b e r North temple 27 Thien Schuur’ find 3. 4,5. 8, 9.10, u , 12,13,14,15,16,17,18,19, 20,21, 22, 25,26, 28, 29,30,31,32,34,36,37,38,39, 40, 41, 42, 43, 44,45, 46, 47, 48,51,52,53,58,59, 60, 61, 63, 64, 65, 66, 67, 68, 69,70,71, 72, 73,74, 75, 76, 86,96 Broekberg’ 7. 7 7 ,9 3 Aan Paspoel’ 4 9 , 5 4 , 5 5 , 7 8 ,9 7 Kielenstraat’ 7, 23, 62, 81, 83, 88, 89 Sint-Truidersteenweg’ 56 de Schaetzengaarde’ 82 Elfde Novemberwal’ 57 , 9 i Soutwestem cemetery 1, 6 , 3 5 , 7 9 , 84, 85, 87,90,92, Armand Meesenlaan’ 80 Table 2. Different find locations and correspon­ ding catalogue numbers o f the figurines. - 4 8 - R e s e a r c h in t o t h e T R EES OF D E SC E N T Revealing the evolution of the series, including the relative chronology o f the successive production generations, can best be achieved by comparing the forms and dimensions of the image-processing program, we have the ability to slide the images with a different transparency level on top of each other in layers, making the similarities and differences clearly observable.3The reduction in scale, due to the moulding process, can be calculated individual figurines. In the past this was done by systematically photographing the figurines in front of a grid.1 This way, changes could be with great precision. The results are repre­ estimated. Comparing the dimensions proved to be useful, not only for figurines. The study of sented in a visual form in the ‘tree of descent’. A selection of major descent schemes in which the decoration of mass produced relief lamps also indicated the benefit of such research.2The method consisted of placing traced drawings finds from Tongeren take centre stage gives us a good idea of the possibilities of such a study. of the lamps based on rubbings or photos on top of each other. Because of the timeconsuming nature of this activity it was never applied on a large scale. The result of image comparison, expressed in a In the current research environment, the development of new technology and new media to preserve and process images opens the door to a new research phase, in particular the digital image comparison. By using an 2 reduction percentage, is represented as a web of relations between the figurines. In this web, all individual figurines are connected, each with a different story that contributes to the entire descent scheme. Workshop finds, assembly characteristics and data from the chemical and petrographic analyses lead to the identification of the workshops. The painting • on the figurines can provide additional De Beenhouwer 1986; De Beenhouwer 1991b, 75 fig. 12; De Beenhouwer 1996b. Bemont & Bonnet 1984,142. The accuracy o f the results depends on the care with which the photos were taken. Ideally, photos should be used in which the recording areas form a vertical projection with respect to the camera. To prevent distortion o f the perspective, there has to be sufficient distance between the object and the camera. Therefore, the results o f the comparisons with visual material from the literature depend on the accuracy o f the mentioned dimensions and quality o f the recordings. - 5 0 - arguments to the origin. If the names of shrinkage of the seated specimens DB 3663 and manufacturers are preserved on the figurines or the moulds, the descent scheme provides an interesting perspective on their correlation. The rise and disappearance of certain themes, attributes, and styles uncover the iconographic DB 3664 is 5 and 10 per cent, respectively. The youngest Fortuna, DB 3597, is 20 per cent smaller than DB 3601. Minerva TO 33 has the evolution. Finally, the tree of descent allows gaining new insights about the production method. The following summary, in which series from Tongeren play a key role, depicts several tree of descent evolutions. The numbering of the specimens preceded by DB refers to the numbering in De Beenhouwer 2005, where more tree o f descent relationships are described.4 In order to distinguish them from same dimensions as Fortuna TO 5 and DB 3598. The common parts of the folds falling down are identical. The sitting Juno DB 4106 is 5 per cent smaller than Fortuna DB 3664 but has slightly higher shoulders; also, the upper part of the body has shifted more to the right. The falling folds are almost identical. Juno TO 29 and TO 31 are 3 per cent smaller than DB 4106. The Venus series is related to the Juno series through the head. The front of the head of Venus DB 3153 almost equals that of Juno DB 4108. The face of Venus TO 44 is rounder. The head of Venus DB the Tongeren specimens, the catalogue number of the current catalogue is preceded by TO. 508 has the same proportions as that of Juno TO 29 and TO 31, but is 4 per cent smaller. It is 12 per cent smaller than that of Venus DB 3153. Tree of descent 5: Fortuna, Minerva, Juno, Chronology: Four dates determine the chronology of the tree of descent. The oldest and Venus The relationship between seven series that belong to four different themes provides a good picture of the complexity of the terracotta industry. In addition to the mechanical relationship of the moulding process, the touching up that refreshes each new gene­ ration is executed in the same style, which reveals that a single workshop manufactured the entire tree of descent. The products from this workshop are well represented in Tongeren. They were found not only in the town centre but also on the burial sites. Specimens from the Tongeren find of the ‘Thien Schuur’ are spread over the entire figurine, Fortuna DB 3601, was found in a grave in Bingen and is dated between 100/125 and 150 AD. Venus DB 3153 from the Hasengasse depot find in Bingen is situated halfway in the evolution of the tree of descent. This depot was closed around 150 AD, but in particular held finds from the second quarter of the second century. The plinth of Minerva DB m o from Braives comes from a pit with terra sigillata from the second quarter of the second centuiy. The youngest specimen, Fortuna DB 3597, was found in a zone of the Cologne workshop at the Rudolfplatz, dated in the second half of the second century. This demonstrates that the series was still in production after the middle descent scheme. This demonstrates that the figurines from this depot were distributed over a length of time. of the second centuiy. In general, the evolution of the tree of descent can be dated between 125 Evolution: Compared to the largest specimens of the standing Fortuna, TO 4 and DB 3601, the Workshop: Using the workshop finds and results of petrographical research and * 5 and 160/175 AD.S The figurines and series to w hich they belong have numbers that refer to De Beenhouwer 2005: the number o f the series follows the catalogue number between brackets. For example, “734 (839)” signifies “Cat. No. 734 belonging to series 839”. Because o f new data in connection w ith the relationship o f the find o f Fortuna DB 3597 (Hôpken 2005, No. 31-001), the upper lim it o f the dating differs from that in De Beenhouwer 2005,849. - 5 1 - Fig. 18. Tree o f descent 5: Fortuna, Minerva, Juno and Venus. DB 208 (274) Tongeren TO 33 DB 3153 (30) Bingen - 5 2 - DB 3728 (274) F rankfurt-Heddemhei m ■Jjpy* éÊÊâ&k DB 3596 (201) Köln am Rudolfplatz DB 3597 (201) Köln am Rudolfplatz -2 DB 298 (473) Tongeren TO 29 DB 4106 (473) B ingen- B ingerbrück 300 (473) Tongeren TO 30 -6 DB 3201 (30) Cologne ‘am Rudolfplatz’ DB 130(30) Tongeren TO 44 “ 53 - chemical analysis as a guide, almost the entire development of the tree of descent can be situated in the Cologne workshop Am Rudolfplatz. Three specimens o f the Venus series were found there (DB 3201, DB 3202, and DB 3203). Moreover, TO 45 can be attributed petrographically to the Cologne workshop. The typical development of the engraved detail, especially the haircut of one of the youngest pieces, DB 508, is very close to that of the largest Venus figurine from Bingen, BE 3153. they appear slightly later, the evolution of the sitting Fortuna series runs almost parallel to that of the standing Fortuna. The standing Minerva also evolves in a parallel way. This lends weight to the idea that these series were created simultaneously from the same archetype. A striking phenomenon is the reworking of the standing and sitting figures evolving in a similar way. The finely worked retouches in linear style with parallel grooves and herringbone motif link all figurines in this Therefore we can assume that the same workshop is responsible for the Venus series, from the largest to the smallest specimen. Two specimens from the standing Fortuna series descent scheme. The entire production can therefore be attributed to the same Cologne workshop. were also found in the Cologne workshop Am The development o f the Juno series is in Rudolfplatz (DB 3596). TO 4 can be attributed to a Cologne workshop, both by chemical and petrographic analysis. Petrographic analysis of the standing Minerva made it possible to contrast to this logical and uninterrupted evolution. They appear halfway along the attribute three specimens to Cologne (DB 208, DB 209, and DB 1120). Finally, pigments and traces o f painting also point to the same workshop for Fortuna DB 3597 and DB 3598, and Minerva DB 1120. Pigment: A characteristic painting technique connects the different series to a single workshop. A Reddish-brown to black linear painting, often in the form of a diamond pattern, can be found on the plinth of the sitting Juno DB 4107, the standing Minerva DB 1120, the standing Fortuna DB 3598 and Venus DB 3153. On this last specimen, also anatomical details are indicated. The left side of the plinth of Fortuna DB 3598 is decorated with two crossed lines. Iconography: This tree of descent is exceptional because it bridges both the sitting and standing figures o f the themes Fortuna, Minerva, Juno, and Venus.6 The connection between the standing and sitting series of Fortuna and Minerva is remarkable. Although development o f the tree of descent and iconographically form a mix of different series: they derive the body from the sitting Fortuna and the head from the Venus series (see production technique). Because the head was originally created for the Venus series, it sets in slightly earlier than the Juno series. However, these older mouldings of Venus were not preserved. The similar reworking of the haircut of one of the youngest Venus figurines (DB 508) and the oldest preserved moulding (DB 3153) reveal the same hand. Treated by the same craftsman, these two figurines cannot be far apart chronologically. Production technique: An important conclusion about the production technique is that not only entire figurines underwent a development. Separate parts of the body such as the torso or the head often evolved independently. A striking case in point is the relationship between the seated and standing Fortuna series. The common upper part of the body, including the head, is clear proof o f the. mechanical relationship. The lower part of the body was reworked. Another example is the See also Schoppa i960: attributes the series o f Juno and Fortuna to the same middle Rhine production centre; Van Boekel 1987,317-320, Fig. 49-51: attributes the series o f Juno, Fortuna and Minerva to one workshop, also in the middle Rhine area; see also the chapter on the origin. - 54- adding of new attributes to the Minerva series, such as the helmet and shield, replacing the horn and rudder of Fortuna. The coroplast presumably made these adjustments already during the creation of the first model. Other adjustments such as the changing of the head occur in the course of the succession of the series. A good example of this process offers the head of the Venus series that was adopted by Juno. At the right, the round hairpiece of Juno DB 4108 runs broadly over the shoulder. At the left, it ends at the beginning of the neck. The effect is unbalanced and unnatural. The only explanation must be that the head originally belonged to the Venus series. The position of Venus’ hand, grasping the right danger of dating based on the hairdo of members of the imperial house.8 Starting with the oldest and ending with the youngest specimen of the tree of descent, the reduction percentage involves 22 per cent. The large number of differences in height and adjustments within the evolution of the tree of descent demonstrates how complex the production process was. This clearly is no linear process whose evolution can be expres­ sed in regular reduction percentages. Undoubtedly, several moulds of the same series were used simultaneously, each with their own small deviations. These deviations accumu­ plait explains the absence of the hairpiece at the right side of Juno’s neck. lated in the course of the evolution. Generally, an evolution of the tree of descent only gives us a relative chronology. It is Method: The typical hairdo of the Venus and difficult to get an idea of the real duration of the evolution of the production. In addition to Juno series also appears among the central Gaul busts.7 ~A high rounded hairpiece composed of several sections above each other surmounts the head. The sections are divided into small blocks. At the back, the hair is brought together into a large bun in the middle of the back of the head. Only Venus is depicted with plaits falling freely on the the archaeological dating in the scheme of descent, clear stylistic characteristics can be important. In the case of tree of descent 5, the relationship in style of the retouches suggests that the entire development took place within the workshop of one coroplast and, conse­ quently, also within one generation. shoulders, along both sides of the neck. The high degree of detail of the central Gaul busts reveals that the bottom section above the forehead is a free hair strip that does not Tree of descent 32: Fortuna belong to the hairpiece. The bun at the back of analysis of the clay paste revealed that it was not made in Cologne, but in another the head consists of rolled-up braids. Haircut and bun resemble that of Marciana, Trajan’s sister, o f whom images were made during the reigns of Trajan and Hadrian.89The Rhineland Venus series has its origin in the Hadrian period, but the majority of the production is to be placed later. Consequently, the striking similarity to an imperial haircut can only be used as terminus post quem. In the field of methodology, this is a good example of the 7 8 9 A Fortuna from Tongeren heads the develop­ ment of tree of descent 32. The petrographical Rhineland workshop. The precise area of production is unknown. Specimens of terra­ cotta figurines belonging to this tree of descent were found in Trier, Cologne, and Frankfurt, each known for their production of terracotta figurines. The sudden change in style in the younger figurines leads us to suspect that several workshops enter into this line of production. Rouvier-Jeanlin 1972,279 nr. 747; De Beenhouwer 2005, series 611 and 612. Wegner 1956,77 Taf. 35. Schauerte 1985,132 “Beginn dieser Variante um 100”; Lange 1990 (Salzburg), 159 Nr. 160: “Ende i.-Anfang 2. Jh. n. Chr. (Toupetfrisur)”; Lange 1994,203, series 2: “spätflavisch-trajanischer Toupetfrisur”. - 55 - Fig. 19. Tree o f descent 32: Fortuna. DB 166(204) Tongeren TO 7 DB 3607 (206) Wederath DB 3605 (204) Kretz-Miesenheim DB 181 (205) Martelange - 5 6 - Evolution: Fortuna TO 7 is the largest specimen. DB 3605 and DB 3606 are 3 % smaller but belong to the same generation. With the exception of the constriction under the torso, all folds of the clothing remain practically identical. DB 3606 has a different head and the body is stretched at the right arm and shoulder. The horn is also raised and leans closer against the body. The bundle of folds that falls free from the left arm is shortened and reduced. DB 3607 is 6 % smaller than DB 3605. The proportions have slightly changed: the figurine is smaller and the upper part of the body straighter, the lower legs are shorter and the horn is pressed closer to the upper part of the body. The coroplast tried to compensate the loss of height caused by the successive mouldings by stretching the figurine. Compared to DB 3606, the loss of height of DB 3629 amounts to 17 %. The posture, the general development of the clothing, and the position of the rudder and horn were maintained, but the details become blurred and the forms simplified. The head is broadened and the right arm and shoulder are raised again. The neck is almost missing. The diadem is round and the horn less pointed. The century, the transition to series 206 and 931 at the beginning of the second century, and the final phase with series 205 in the first half of the second century. Fortuna TO 7 stands at the beginning of the development of the tree of descent and dates from the end of the first century. Workshop: The tree of descent originated and evolved in the Rhineland. No workshop finds were preserved. Specimens were discovered in Trier (Series 204), Cologne (Series 93r), and Frankfurt (Series 205). Petrographic analysis attributes TO 7 to a workshop different from Cologne, possibly Frankfurt or Trier. Iconography: The opening of the horn of Fortuna TO 7 points upwards. The fruit that bulges out of the horn of DB 3607 are highly blurred at the backside of the figurine. Finally the cornucopia evolves into a horn with the opening at the front in series 205. The plinth changes from a highly constricted cylinder with a sharp top and bottom edge into an only slightly concave plinth with a softly rounded top and bottom edge. plinth is slightly concave. Chronology: DB 3607 from the Wederath burial site was found in a grave dug between 90 and 125 AD. DB 3629 from Frankfurt-Heddernheim was found in a grave from the burial site of the Feldbergstrasse that was dated 103/111-150 AD. According to the image research, the evolution starts with series 204, changes into series 206 and 931, and finally ends in the simplified Production technique: The Fortuna figurines from tree of descent 32 show an evolution from faithful, well-detailed forms (Series 204) to increasing schématisation (Series 205), with simplified and rounded forms. This process can be explained by the increasing smoothing that can be expected when using successive moulding. The right arm and shoulder are design of series 205. The archaeological dating seems to confirm this evolution with a first systematically raised, the neckline disappears, and the rudder is no longer placed diagonally but right against the leg. The evolution from a development of series 204 at the end of the first horn with the opening pointing up to a horn - 5 7 - with the opening at the front may be inter­ preted as a growing trend of attention on the front of the figurine, making details at the back less important. At a technical level, tree of descent 32 illustra­ tes that coroplasts tried to compensate the loss of height by stretching the figurine, which may lead to adjustments in the proportions. Replacing or entirely reworking the horn and head is another technique that emerges in the evolution of tree of descent 32. The reworking can become so substantial that the relationship between series 204 and 205 can not be proved without the link of series 931. has his origins in a Trier workshop since red firing paste was frequently used in Trier. Pigment: The painting and colour preserved on Minerva DB 4944 (292) refer to production in a Cologne workshop. The figurine has a deep brown diamond pattern at the front and sides of the base. The same motif was applied at the front of the plinth of DB 3761 (293). Iconography: The specimens of series 293 have a Gorgon head (gorgoneion) on the torso in a frame of small circles. In Rhineland terracotta, the origin of the Gorgon head with the typical frame of circles is dated around 90 AD.*11 The specimens of the younger series 292 do not have a gorgoneion and the folds of the garment cross each other diagonally on the chest.“ Tree of descent 45: Minerva Tree of descent 45 connects two Minerva series, both of which specimen have been found in Tongeren. Tough diverse in detail, the Tongeren specimens are closely related. They were all manufactured simultaneously in Cologne workshops with various sets of mould. Their diversity illustrates the extent of the production. The Cologne origin is not yet certain for al the specimen belonging to this tree of descent. Probably Trier also took part in an early stage of the production. Evolution: DB 3755 (292) is 7 % smaller than DB 3762 (293). Production technique: The specimens of both Minerva series were manufactured using a twopart plaster mould. A moulding from a previous generation was often used as a model for a new generation mould. If moulded from a worn out mould, the blurred detail of this new model was then reworked. The typical little mould blisters in the freshly engraved folds, as found in TO 38, show that the retouching was not applied in the mould, but indeed on the model. Methodological consideration: The evolution of series 293 sketches the complex development Chronology: Tree o f descent 45 is dated between 90/100 and 150 AD.“ of several production generations.13 It is remarkable that the six specimens from Tongeren are halfway along the evolution of Workshop: DB 3758 was found in the am the series with shrinkage of 11 up to 14 per cent compared to figurine DB 3762 (293) from Trier Rudolfplatz workshop in Cologne. The figurines TO 38, TO 42 and TO 43 were petrographically attributed to Cologne. All specimens were Altbachtal. They more or less belong to the same production generation, but show manufactured in white-firing paste, except for DB 200 from Martelange, which is one of the oldest specimens. It is possible that the series remarkable differences in proportions and detailing. Using parallel moulds, each of which start off a new production fine, the differences “ “ “ 11 See catalogue: dated contexts o f series 26 and 27 De Beenhouwer 2005,1149, m otif 62. De Beenhouwer 2005,1149, m otif 49. See catalogue: internal mechanical relationships o f series 27. - 5 8 - Fig. 20. Tree o f descent 45: Minerva. DB 3759 (293) Cologne DB 200 (293) Martelange -14 DB 196(293) Tongeren TO 40 -19 DB 3757 (293) Heerlen -13 DB 3762 (293) Trier ‘ AltbachtaP DB 199 (293) Tongeren TO 43 DB 3755 (292) Bingen ‘ Hasengasse’ in proportions and detail increase in each new generation. According to recent experiments, shrinkage makes jumps of 7 to 10 per cent every generation when using plaster moulds. However, the situation is much more complex. For series 293, the entire shrinkage of the specimens under discussion adds up to 19 per cent in total, which theoretically corresponds with two or even three production generations. In reality irregular intervals of 9, 2, 2,1, and 5 per cent, respectively are measured. We can also observe at least five adaptations of the Medusa head. These observations demonstrate that it is not advisable to define generations in regular jumps. Small differences in height are also significant. The main reason is found in the use of parallel moulds, successive reworking, trading moulds and copying from - 59- other workshops. Moreover, we should not forget that the schemes only depict a small part of the production discovered in excavations, and that the techniques used in modern experiments and antique production methods may differ. Tree of descent 6g: woman with dog and woman with fruit (Vindex and VR coroplasts) The seated woman with fruit, TO 52, stands at the beginning o f the development of the iconography o f the Rhineland matrona terracotta production. This specimen has a quently, the dating of series 444 directs the ar­ chaeological dating o f the tree of descent between 140 and 260 AD. Inscription: DB 3997 (411) was signed in relief at the back by Vindex, who produced terracotta figurines between 140/150 and 180 AD. The figurine is at the beginning of the tree of descent. At the back of the chair of DB 4050 (444) the letters VR were signed in relief. This coroplast, who also worked in Cologne, stands at the end of the evolution of the tree of descent and is therefore younger than Vindex. The final stage of this descent scheme can be situated after the active period of Vindex, but before the remarkable high-quality execution and is richly painted. It is related to a design by the coroplast Vindex from Cologne. The tree of workshops am Rudolfplatz in disappeared around 200/210 AD. descent also represents the chronological relationship between Vindex and another Workshop: The tree of descent originated and evolved in the Rhineland, more specifically in coroplast who used the initials VR. Cologne. The oldest specimen, DB 3997 (411), was signed by the coroplast Vindex from Evolution: The height of TO 52 is 2 per cent smaller than that of DB 3997 (411). Both specimens have similar width since the sides of Cologne. According to petrographic analysis, TO 52 was attributed to Cologne. the chair in DB 3997 are almost completely missing in frontal view. The folds on the lower legs, the fruit in the lap and the position of the Pigment: TO 304 (442) is decorated with a dark lunula, the hands and the feet match perfectly. The plinth of DB 3997 is higher than TO 52. The right side of the plinth and the side of the chair. head and back are different. TO 52 has a thicker profile than DB 3997. DB 4047 (443) is 3 per cent smaller in comparison with DB 304 (442). The contours, proportions, and most folds are similar. DB 4048 (444) is 3 per cent smaller than DB 4047 (443) and 7 per cent smaller than TO 52. The contours and general proportions are similar, but the folds are slightly reworked. DB 4045 (442) is 9 per cent smaller than DB 3041 (442). Cologne brown, diamond-shaped painting on the front of the plinth and a Saint Andrew’s cross on the The outline of the chair’s contours and of the plinth is preserved both on the left and right side. The V-neck was highlighted by a dark brown colour. DB 4048 (444) shows traces of dark paint on the necklace and in the folds of the garment. Both the motifs and the dark brown pigment were used in Cologne workshops.14 Iconography: The slight difference in height between the seated woman with the dog DB 3997 and Matrona TO 52 being inconclusive, Chronology: DB 4049 (444) from FrankfurtHeddemheim can be dated between 75/79-260 AD, based on the general occupation period of the site. DB 1177 (444) from a villa in Ambresin was dated between 140 and 275 AD. Conse­ “ the fine elaboration of the cloak folds of the seated woman with dog by Vindex leads us to conclude that DB 3997 is closest to the original model. The hair is arranged in rows of superimposed braids that drop down from a De Beenhouwer 2005,1,159-1,167, decoration techniques 5,14,16, and 17. " 60 - Fig. 21. Tree o f descent 69: woman with dog and Matrona (Vindex and VR coroplasts). DB 3997(411) Cologne DB 4047 (443) DB 304 (442) Tongeren TO 52 DB 4048 (444) Worms 4045 (442) Koln DB 4050 (444) Cologne centre parting and tied up into a bun at the back of the head. This hairdo, which is characteristic of Vindex, is dated between De Beenhouwer 2005,1138, m otif 21. 140/150 and 170/180 AD.,SIn all other series, fruit replaces the dog on the lap and the woman wears the Ubian bonnet, a semi-spherical headgear fastened with a pin beside the left cheek16. This is a second important mono­ graphic dating criterion. In terracotta figu­ rines, the bonnet is generally dated to the period between 140/150 and 250 AD.17Therefore, at the earliest, the tree of descent commenced shortly before this date with the specimen by Vindex DB3997 (411), which is not yet equipped with the Ubian bonnet. All series depict a smooth chair, the armrests articulated in front. The rim of the cloak falls down on the chest, where it is clasped. Between the legs, the folds fall down in the shape of a zigzag or figure eight.18 In the course of the It is possible that the introduction o f the matronal triad in stone sculpture around 160 AD marks also the introduction of Ubian clothing in the religious iconography.19 In current research, it can be argued that the Ubian triad dates from after 160 AD and that there is no certainly about the iconography of the matronae before that time. Production technique: All specimens were manufactured from a two-part mould. The surface shows small mould blisters. The joints are all broadly smoothed. This is indicative of the use of plaster moulds and wet assembly. evolution to the series, the folds on the chest become simplified, more particular in series 443; the cloak folds on the upper arms run Tree of descent 75: Fortuna and Juno capite velato (Vindex) concentrically and are arch-shaped, with a point of contact in the middle of the chest. This trend The tree o f descent relates three series depicting women, the head covered with their to simplification can be dated after 160 AD. cloak: a woman with cornucopia and rudder, a woman showing a round object in the right Vindex’s design of a dog on a lap filled with fruit stands at the beginning of the develop­ hand and a woman holding a patera. The scheme of descent leads to gain a clear insight ment of the tree of descent. Almost immedia­ tely, the dog disappears and only the fruit in the lap remains. At the same time, the Ubian into a special procedure: the coroplast used an bonnet is added. In that way, shortly after the creation of Vindex’s first design, the icono­ graphy of the Ubian matrona is standardized in terracotta. 16 " “ " armless model without attributes for the production of various themes. Evolution: Fortuna TO 6 is the largest specimen. Juno DB 4120 (477) is 5 per cent smaller. From the head to the plinth all proportions correspond, The Ubian clothing was worn by native women in the area o f the Ubians, from the first to the third century (Horn 1987,36-37: grave monument from the Claudius-Nero period; W ild 1968 214: usually second h alf o f the second and first h alf o f the third century; Van Boekel 1989 (Scheveningseweg), 19. De Beenhouwer 2005,1144, m otif 39. De Beenhouwer 2005,1151,1155, m otifs 56 and 71. In the context o f religion, the oldest depictions o f matrons in Ubian clothing in stone sculpture can be traced to the foundation o f the temple for the Matronae Aufaniae in Bonn (Rüger 1987,110-113,22-24,28; Horn 1987,41-42). The circumstances o f the find ensure a dating after the middle o f the second centuiy for the building inscription (Rüger 1987,10). Epigraphically, the building inscription and therefore the foundation o f the temple for the Aufaniae has to be placed in AD 161, because it still mentions Antoninus Pius, but also already Lucius Verus (Rüger 1987,22 nt. 30). According to Rüger, the foundation o f the sanctuary for the Aufaniae in Bonn in AD 161 and the sudden consecration fever in the following years are related to the war o f Marcus Aurelius against the-Parthians (Rüger 1987,10-13: all other datable consecrations to the Rhineland Matronae would be o f a later date. The youngest altars date from 250/260, shortly before the trouble w ith the limes (Rüger 1987,13; see inscriptions: consular dating). Epigraphic and monographic sources don’t provide indications o f the existence o f the typical Matronae in Ubian clothing in the Rhineland before AD 160. According to Horn, the cult sculpture in the temple in Bonn m ust have provided the monographic standard for the uniform ity o f the reliefs on the many altars that were consecrated in subsequent years (Horn 1987,41-42; Derks 1998,126). This concerns the triad. Opinions differ on the phase before AD 160. According to Rüger and Horn, before that date, the cult is in a pre-anthropomorphic phase (Rüger 1987,28; Horn 1987,49-53 tree cult). According to Derks, the depiction is inherently anthropomorphic from a linguistic point o f view (Derks 1998,126). Additionally, according to Derks, the two cult buildings from the first-centuiy phase o f the Pesch sanctuary m ust also have included cult sculptures (Derks 1998,126). - 62 “ Fig. 22. Tree o f descent 75: Fortuna and Juno capite velato (Vindex). DB 3595 (200) Worms DB 193 (306) Tongeren TO 6 DB 3636 (306) Melick - 6 3 - but DB 4120 (477) is considerably broader, especially at the height of the right arm and shoulder. These adaptations at the right side and the change of attributes, show that the same armless model formed the basis of both the Fortuna and Juno figurines and that the arms and attributes were applied separately. Chronology and inscription: Fortuna TO 6 was found in a sector of the southwestern cemetery that was used between 50 and 200 AD. Vindex’s workshop in Cologne stands at the origin of the tree of descent: the back of the plinth of Fortuna DB 3636 (306) from Melick is signed by this coroplast, who worked between 140/150 and 180 AD (Fig. 38). The start of Vindex’s workshop provides a terminus post quern for the tree of descent. Comparing the three series we can see that the various creations are derived from an armless model without attributes that is the starting point of this tree of descent. This technique had not yet been proven for the Rhineland, but was already known from a series of figurines originating from Central Gaul. Clay mould parts of a figurine without arms and attributes are found in the Toulon-sur-Allier workshop.” They were used to manufacture Fortuna and Minerva figurines, whose style is strongly related to that of the Rhineland tree of Descent 75. The Rhineland as well as the Allier workshops were probably exposed to common external influences, which chronological connection.” suggest' a Tree of descent 84: Mercurius Workshop: TO 193 (306) is petrographically attributed to Cologne and DB 3636 (306) was Two terracotta figurines depicting Mercurius were discovered in Tongeren. One of them (TO signed by the coroplast Vindex from Cologne, 56) belongs to the older figurines of tree of what allows us to conclude that series 306 was produced in Cologne. Since the other series are a variation of the same armless model to which descent 84 which were manufactured in Cologne. One of the younger series, represen­ various elements were added, they could be attributed to the same workshop. Pigment: DB 3636 (306) is decorated with multi­ colour painting. At the back of the plinth, a ted by a find from Martelange, was manufac­ tured in the Moselle area (DB 332). During the production process, the Treverian copyist applied changes to compensate the shrinkage caused by moulding. grey-brown cross was painted. At the left side, the coroplast applied a cross and a vertical edging on the plinth.” Light red paint was preserved on the dress, hand, rudder and base. Development: DB 333 (947) and TO 56 have similar dimensions. DB 4179 (526) also has the DB 3637 (306) bears a faded brown painting on the plinth: at the front a star shape executed in in height but, nonetheless, the figurine is not at the head of the tree of descent. The double lines and cross patterns at the sides are preserved.” This kind of painting points to Cologne. proportions clearly changed. The upper part of the body is 2 per cent thinner than that of DB Production technique: All specimens were manufactured from two plaster mould parts. The surface shows small mould blisters and the joints were broadly smoothed.*2 3 “ ** 23 same height, but is thinner, especially at the height of the animals. DB 332 (949) is the largest 333 (947) and the torso is 10 per cent thinner in profile. The back is straighter and the plinth was raised by 1 cm. The wings of the winged heat (petasus) are also raised. Of all specimens belonging to tree of descent 84, the breast of Mercurius DB 332 (949) has the thinnest profile. De Beenhouwer 2005,1160 and 1163, decoration techniques 6 and 14. De Beenhouwer 2005,1164, decoration techniques 5 and 13. Rouvier-Jeanlin 1972» 188 no. 428 and 429; De Beenhouwer 2005,868-869,970, tree o f descent 26. Concerning the influence o f stone sculpture, the resemblance to a statue o f a Vestal Virgin o f the Museo Nazionale delle Terme is illustrative. Bieber 1977,198, Pi. 140 Fig. 820: dated about AD200. -64- Fig. 23. Tree o f descent 84: Mercurius. DB 1198 (948) Tongeren TO 56 DB 4197 (947) Frankfurt-Heddemheim DB 333 (947) Hoeselt DB 4179 (526) Cologne DB 332 (949) Martelange - 6 5 - Chronology: DB 332 (949) from Martelange was found in a tumulus containing amongst others central and eastern Gaul terra sigillata from the second century. DB 4182 (526) from Dieburg is part of the contents of a grave from the first half of the third century AD. The combined data from Martelange and Dieburg suggest a date between 125/200 and 250 AD. Inscription: The upper edge at the back of the rock on which Mercurius DB 333 (947) from Hoeselt is seated bears the incomplete inscription AI.TICVS.FIICIT/C.C.A.A. (Fig. 35) Rhineland terracotta this motif is mainly found in depictions of Mercurius, Bacchus and Apollo manufactured by Servandus.26 The motif can also be found in some Rhineland imitations of central Gaul nutrices. All are dated to the second half of the second century or later. Production technique: The takeover of a Cologne product by a workshop in the Moselle area involved several changes to the original figurine. The shrinkage associated with copying figurines was clearly a problem. The figurine from Martelange was stretched and On the same spot the inscription VTVS FE CA can be found on Mercurius TO 56 and TO 57 from Tongeren (Fig. 34). No other work of artificially enlarged by extending the plinth and raising the wings on Mercurius’s hat. • either of these coroplasts is known. All series were manufactured from two half Workshop: The tree of descent originated and further developed in the Rhineland. The inscription C.C.A.A. on figurine DB 333 (947) is an abbreviation for ‘Colonia Claudia Ara Agrippinensium’, Roman Cologne. The letters CA, found on TO 56 and TO 57, are a part of the same abbreviation. DB 4179 (526) was found in Cologne, but a connection with a specific workshop cannot be proved. DB 334 (948) and DB 333 (947) are petrographically attributed to Cologne. The younger specimen DB 332 (949) from Martelange is attributed petrographical­ ly to the Treverian area. Pigment: The painting on one of the specimens o f Series 948 refers to Cologne: TO 56 is embellished with dark brown paint, which accentuates the wings of the winged head (petasus) and the deeper cavities of the heralds staff (caduceus).24 The same dark brown paint was preserved on the rooster’s right eye, the right side of the bottom edge of the base, and two armrings around the right upper arm.25 Iconography: All figurines show a tongue­ shaped cloak tip on the bare shoulder. In 24 25 26 moulds with an integrated rock-shaped plinth. Specimens of series 526, 947, and 948 show mould blisters, as a result of the use of a plaster mould. Method: When determining the successive generations, comparing the heights o f the specimens is not enough. This would wrongly position the younger specimen from Martelange at the head of the descent scheme. Only full image comparison can lead to correct interpretation. Tree of descent 89: horseman with hooded cloak The tree of descent 89 shows that analysis of the site and the context of the find have major implications for the interpretation o f the figurines. Although some inscriptions on horsemen are related to the horseraces, other figurines are linked to the military world. Development: DB 4274 (566) is 6 per cent smaller than TO 55 and 5 per cent smaller than DB 4276 (567). The head of the horse of DB 4276 (567) is reworked in a clumsy way: the muzzle became pointed, the eye is engraved lower and the De Beenhouwer 2005,1160, decoration technique 16. De Beenhouwer 2005,1159,1,161, decoration techniques 2,3, and 4. De Beenhouwer 2005,1152, m otif 59. - 6 6 - Fig. 24. Tree o f descent 89: horseman with hooded cloak. DB 786 (566) Tongeren TO 55 DB 4274 (566) Bonn DB 4279 (569) Cologne DB 4276 (567) Niederursel manes are no longer of equal length. This shows that an older generation must have preceded DB leg there is an inscription NTVS (?). 4276 and TO 55 although they are at present at Workshop: The production originated and the head of the tree of descent. DB 4279 (569) is 10 per cent smaller than DB 4274 (566). developed in the Rhineland. TO 55 is attributed to Cologne, both by chemical and petrographic analysis. Chronology: DB 4277 (567) from Altenstadt was found in a grave that was dug at the end of the second century or later. DB 4279 (569) from Cologne was recovered from a grave in the ‘Neusser Strasse’, dated by a coin after 180 AD. The grave also contained a figurine by Servandus and one bearing the initials TM. These finds from two different series lead to the conclusion that this tree of descent developed at the end of the second century. Iconography: The man wears a hooded cloak that stretches onto the hip and completely covers the clothing underneath. Hooded cloaks occur in the iconography of Gallo-Roman terracotta figurines after the middle of the second century, mainly in the last quarter o f the century.28 Inscription: DB 4274 (566) has the letters ROX engraved on a palm above the left front leg and the letters TAN/S above the left hind leg under Production technique: The mould knobs on the surface point to the use of a plaster mould. Horseman and horse are integrated into one mould that consists of two half moulds. The joint is broadly smoothed. Sometimes, the legs an ivy leaf.27Above the piercing of the left front and muzzle of the horse are pierced. 17 18 According to some authors, the name o f coroplast Roxtanus can be reconstructed: see Fremersdorf 1939,11 Taf. 7,7; Lange 1994,291,292 series 261-262. Another possibility is that the name o f the triumphant horseman or horse is indicated: see Van Boekel 1987,699. De Beenhouwer 2005,1147-1148, m otif 46. - 6 7 - Function: The figurines from Bonn and Tongeren show perforations that made it possible to add reins and separately formed terracotta wheels by means of a wooden axle. This way, the unit could be pulled. Nonethe­ less, these are not just toys. The palm, as well as both names that were engraved before firing in Tree of descent 101: bust of a woman The tree o f descent illustrates the close connection of the Thien Schuur find with the Cologne production centre. No less than three o f the four series related to Cologne are represented in the find. Where the archaeo­ logical dating of the specimens of the tree of descent fails, the iconography fills this gap. A the horse’s flank demonstrate that the Bonn figurine depicts a victory scene after a horse race. The central figures whose names are mentioned are the winning horse and the triumphing rider. Similar scenes are known quem for some figurines represented in the tree of descent (Fig. 25). from mosaics and ancient medals.29It is clear that this figurine was made as a reminder of an important contest.3“ Another horseman Development: TO 72 is the largest specimen occurring in the tree of descent. TO 73 is -five figurine completed with the name of the horse was found in a military camp of the Roman fleet on the Alteburg near Cologne.31 Not all of per cent smaller. DB 4411 (636) is five per cent smaller than TO 73. The head of TO 74 has nearly the same dimensions as that of DB 4411 the representations of horsemen are necessarily related to the horse races. In fact only two of them bear an inscription. The subj ect seems to have been very popular among (636), but the eyes are larger. nice sequence o f haircuts that refers to imperial hair fashion provides a terminus post Chronology: Based on the active period of the soldiers. It clearly reflected their daily activities. In addition to the Alteburg figurine, Cologne workshops ‘am Rudolfplatz’ DB 4416 (632) and DB 4410 (636) are dated between 70/90 and 200/210 AD. two figurines representing a horseman were found near the military camps of ZugmantelKastell (DB 4275) and Altenstadt (DB 4277). To Workshop: The figurines represented in the tree of descent originated and developed in the the owners they were of great consequence, so special they even took them to the grave (Altenstadt (DB 4277) and Cologne (DB 4279). The horseman from the Cologne burial was accompanied by a terracotta figurine representing the god Bacchus with a panther (DB 4213) and a third figurine depicting the goddess Venus standing in a small shrine (aedicula) (DB 3184). A profound analysis of the site and the archaeological context is of crucial importance in order to achieve a correct interpretation of the depiction since its meaning is connected with environment of the owner. ” 30 * 31 the social Rhineland. DB 4416 (632) and DB 4410 (636) were found in the Cologne workshops ‘am Rudolfplatz’. In addition, petrographic analysis made it possible to assign the figurines TO 73 and TO 74 to Cologne. Consequently, three of the four series can be attributed to this workshop. Pigment: TO 74 shows traces of unintended green lead glaze on the surface. The bottom edge of the base is painted dark brown. The same colour is used to fill the rectangular frame on the upper edge of the plinth.32 This type of painting is often used in the Cologne workshops ‘am Rudolfplatz’. Charioteers w ith victory palm and mention o f names, see Goethert 2007,344-349. Sometimes, terracotta figurines were also manufactured on the occasion o f special celebrations, at w hich the date was engraved: see Hopken 2004 and chapter coroplasts, Servandus. Lange 1994,292 Series 262 No. 1 and commentary o f Series 261: interpretation o f the inscription as the horse’s names. This figurine shows no perforations and does not belong to Tree o f Descent 89. De Beenhouwer 2005,1163,1166, decoration m otifs 14 and 18. - 68- DB 4410(636) DB 367 (631) Tongeren TO 72 4411 (636) Münstermaifeld Fig. 25. Tree o f descent 101 bust o f a woman. 366(632) Tongeren TO 73 368 (637) Tongeren TO 74 4416(632) Cologne ‘am Rudolfplatz’ Iconography: In series 631 and 636 concentrically large hairpiece, divided into radial segments. A arched folds cover the lower end of the chest.33 hair roll runs around the back of the head. That This motif occurs seldom and is mainly found of Trajan’s sister Matidia inspires this hairdo, on terracotta figurines from the first half of the as we know it from a bust preserved in the second century. It creates a strong iconographie Palazzo dei Conservatori in Rome that dates connection with specimens of tree of descent 105 that can be attributed to the same workshop. The hairstyle presents a nice sequence. In series 631 and 632 the hair above the forehead is parted in the middle and surmounted by a 33 34 from the Hadrianic period.34*Therefore, the beginning of Hadrian’s reign in 117 AD is a terminus post quem for the dating of these figurines. In series 636 and 637 the hair is constructed with braids that coil around the De Beenhouwer 2005,1157, m otif 79. Fittschen and Zanker 1983,9 No. 8, Taf. 10: Diva Matidia: Hadrian; Wegner 1956,82 Taf. 37; De Beenhouwer 1991 (statuettes), 74, De Beenhouwer 2005,1137, m otif 19. - 69 - head. Above a band of hair on the forehead, various rows of braids were wound up around the head into a ‘nest of braids’, in which the The proportions of the eyes, hair, mouth, and two V-shaped folds in the neck are similar, but the plinth and width at the bottom of the chest skull remains visible. The haircut represents the hair fashion of a series of anonymous private portraits from Hadrian’s period.35 The comparison with the early-Hadrian hair fashion again provides a terminus post quem around 120 AD. Both hairstyles originated in are different. TO 66 has nearly the same dimension as TO 65. DB 4458 (665) is 5 % smaller than TO 62. The contours of the torso and plinth match. The position o f the diagonal the Hadrian period. According the tree of descent, the hairdo with the radial segments is rim of the cloak and the orientation of the diagonal folds are also similar. The head is changed and, on the inside of the V-shaped neckline, DB 4458 (665) has one fold more on the left. older. Production technique: Mould blisters on the surface indicate the use of a plaster mould. The mould also includes the plinth. The support rib on the back is integrated into the mould part of the back. Chronology: The current state of research indicates a dating in the first half of the second century, based on bust DB 4458 (665). It was found on the southern burial site of St. Matthias in Trier, in a grave from the first half of the second century.36Two other specimens of the same series were found in Trier Altbachtal Tree of descent 105: bust of a boy and come from the Nischenkeller, which was in use in the second half of the second and the Once again the oldest series of tree of descent 105 illustrate the relationship of the Thien Schuur find with the Cologne workshops. An beginning of the third century.37Consequently, it is possible that this younger series was analysis of the technical characteristics shows that, although no mechanical mould relation­ produced after the middle of the second century AD. ship can be proved, an example from central Gaul must have inspired the Cologne manufacturers. The youngest series of the tree of descent were produced in the Treverian area, none of which have reached Tongeren (Fig. 26). Workshop: Although finds from a workshop lack, there is enough evidence to attribute series 671, 672, 673, and 674 to a Cologne production centre. TO 61, TO 62, TO 63, TO 64, Development: TO 61 and DB 4472 (671) are the and TO 66 are attributed to Cologne by means of petrographic analysis. Moreover, according largest specimens of tree of descent 105. TO 62 is seven per cent smaller than TO 61. While the folds on the breast are very similar, the facial to chemical analysis, the clay paste of TO 64 can be attributed to Cologne. Series 665 was not examined, but the distribution of the features are retouched. The eyes are large and simply outlined by a groove. The nose is much broader and flattened, and the mouth is less figurines suggests a production in the Treverian area. expressive. The head and top of the torso of TO 62 are two per cent smaller than those of TO 65. Iconography: A low rectangular frame is modelled between the torso and the front of the !s Fittschen & Zanker 1983,61-62 No. 83, Taf. 104-105: frühhadrianisch; De Beenhouwer 1991b, 74; De Beenhouwer 2005,1138, m otif 20. 11 Anon. 1983,185-186 Grab N42, Abb.123 right (K. Goethert-Polaschek): dating: first half o f second century; Lange 1994,285-286 Series 236 No. 2; von Gonzenbach 1995,357 Taf. 177,3. Gose 1972, Abb. 387,44 and Abb. 386,16; De Beenhouwer 2005, nos. 4459 (665) and 4460 (665). 37 - 7 0 - Fig. 26. Tree of descent 105 bust of a boy. DB 349 (674) Tongeren TO 66 DB 811 (672) Tongeren TO 62 DB 352 (671) Tongeren TO 61 DB 4458 (665) Trier -71 - plinth in series 665, 671, and 672. On the same spot, in series 673 and 674, concentric arched ridges cover the transition between the torso and the front of the plinth. This motif relates tree of descent 105 to specimens of tree of descent 101 that can be attributed to the same workshop. reason for the holes does not seem to be a technical one. Apparently the solution should be looked for in the source of inspiration for the first model. This can be deduced from the form development of the openings in the tree of descent. The back of bust TO 61 that is closest to the first model is completely cut open, except for the support at the height of the spine. In the younger specimens the gaps Iconographically the figurines are strongly related to a central Gaul bust of a boy wearing a laurel wreath found in Vichy.38 The inclined become smaller or disappear completely. A possible explanation is that a central Gaul position of the head, the face with the round cheeks, the hair locks and the execution of the wreath are very similar. In spite of the strong specimen served as an example for the first model of the tree of descent. Busts from Central Gaul were manufactured according to a process similarity, the general proportions and detail of the torso are different.39 Therefore, a Central Gaul origin of the tree of descent cannot be proved based on mould relationship. Perhaps a very different from that of their counterparts in the Rhineland. The mould for the backside Central Gaul example inspired the Cologne coroplasts. completely open. The bust was mounted on a separately formed plinth, using a hand-shaped Production technique: The back was cut open in the specimens of series 671, 673, and 674. In a number of cases, the cavities are smaller or are missing. At first sight, the reason for the gaps is obscure. The creation of a vent is redundant, because the plinth of the figurines is open at the bottom, which allows the air to expand during the firing process. Another explanation could be that through these openings the joints on the inside could have been finished. In TO 61, the openings are large enough to work the inside, but in all younger specimens, they are too small or even missing completely. The only included the back of the head and shoulders. The back of the torso remained support rib at the height of the spine to prop the head and shoulders. The Tongeren figurine may have imitated this technical process, but according to the Rhineland tradition, the head, torso, and plinth were completely formed in one two-piece mould. The tree o f descent illustrates that he reason for cutting the back grew dim and the cuts became smaller or disappeared. Although we cannot prove a direct mould relationship, it appears to be a figurine from a central Gaulish workshop that served as model for the oldest specimens of this tree of descent. 38 Tudot i860,40,41 Fig. LXII, 55 Fig. LXXVI, LXXVIII, and pi. 48 (this last drawing is mirrored in comparison with the other three). 39 The head o f the Vichy specimen is 30 per cent larger than that o f 352 (671) and the torso is 20 % larger. - 7 2 - Conclusion In general, the analysis of the various descent schemes reveals that the terracotta figurines imported into Tongeren came mainly from Cologne. The evolution of descent scheme 5 shows that the Thien Schuur figurines were made over a longer period rather than at a single time. Some Rhineland figurines bear a strong resemblance to Central Gaulish series. They may have been inspired independently of one another by a common stone sculpture model, such as with the Juno and Fortuna in descent scheme 75. A further possibility is that there was a direct link and that workshops in the different regions copied one another, as was the case with the busts in Rhineland descent scheme 105, which were inspired by a Central Gaulish example. There were also exchanges between Rhineland workshops, as is clearly evidenced in the evolution of Minerva and Mercury figurines in descent schemes 45 and 84. The style of the representation is therefore not a useful criterion for determining provenance. One exception are the stylistic attributes o f the retouches added when figurines were copied. These allow us to identify individual modellers, as is clearly the coroplasts to produce a range of themes using the same basic model. This was the case for the Juno and Fortuna figurines in descent scheme 75. An existing figurine was sometimes radically modified by replacing the attributes and even the head, as can be seen in descent scheme 5, which links the figures of Fortuna, Minerva, Venus and Juno, or in the Matronae and seated women with dog in descent scheme 69. For successful representations the coroplast used different moulds of the same figure alongside one another, as shown by the many versions of figurines in descent scheme 45. The coroplast sought to compensate for the systematic shrinkage of the figurines with each new generation by lengthening the body or raising the base. This can be seen in the younger depictions of Mercurius in descent scheme 84 and of Fortuna in descent scheme 32. Thus it is not enough to compare only the heights in order to establish the sequence of the series and generations. The descent schemes also provide useful information about the meaning of the figurines. An analysis of the find contexts for the horseman in descent scheme 89 has case for descent scheme 5. revealed that the interpretation cannot be viewed separately from the cultural and social Working with a model without arms and attributes was a technique that enabled context of the owners and their personal intentions. -73- Pa i n t i n g Today most figurines show the colour of the clay paste they were moulded with. Painting Many Trier figurines were made of red firing clay paste, containing iron oxide, and covered that often added new details or embellished the looks o f the figurine has mostly disappeared. This is mainly caused by by a thin layer of fine white slip. This was common practice in the Trier workshops, where fine white firing clay paste was scarce. influences of the soil in which the figurine was buried for several ages. Sometimes the surface Most of the Tongeren figurines were made in of the object was already damaged before. This is for instance the case with burial gifts that were burned on a funeral pyre. As a result figurines that show remains of painting are rather exceptional. In contrast with this remarks, the collection of Tongeren largely contributes to the knowledge of painting techniques. Although we do not know the circumstances in which the ‘Thien Schuur’ terracotta find was buried, the abundant painting residues prove that the soil conditions favoured their preservation. No less than 29 of the 52 preserved figurines of this find show traces of painting or a slip layer. Slip A clear white surface is the ideal background for painting rich in contrast, which most potters aimed for. One of the ways to achieve this aim was to apply an engobe or slip layer by submerging the figurine in a highly diluted clay paste of a pure white firing quality. -74- Cologne workshops that disposed o f large amounts of fine white firing clay. Although the Cologne potters seemed to have no need of an additional slip layer on their white figurines, closer investigation of the Tongeren collection shows that in some cases, Cologne fabrics where also finished with a fine slip layer. We cannot always clearly prove the presence of an engobe, certainly not when it has nearly the same colour as the body of the figurine. Poor adhesion o f the layer may help us to distinguish the engobe. For example, a thin peeling layer is visible at the height of the breast of Mercurius TO 56, manufactured in Cologne. Apparently, a thin white engobe was also applied to some white firing figurines from Cologne. The engobe is recognizable at the point where the colour of the core slightly differs. A good example is Juno TO 28, whose thin white slip layer hides the slightly yellow body. Another good example is a group of figurines made in Cologne whose bodies Fig. 27. A heavily worn off, white slip layer covering the dull orange body o f bust TO 73 from the Tongeren 'Thien Schuur' find. discoloured into dull orange and which are covered with a very thin white layer. It concerns an iconographically varied group including the standing Fortuna TO 4, the seated Fortuna TO 25, the bust of a boy TO 64, and the woman’s bust TO 73 (Fig. 27). The finishing slip on the dull orange figurines is worn heavily; the layer is very thin and has disappeared in places. Slip was mainly used to improve the surface, not only to apply painting: of all examples finished with white slip, only Mercurius TO 56 and Juno TO 28 show traces of brown painting to accentuate Fig. 28. Shiny yellowishbrown engobe covering the white body o f the couple TO 58 from the Tongeren 'Thien Schuur' find. Dark brown armrings are painted on the upper arm. details. A white surface was clearly the norm, but not intended in all cases. For example, the couple TO 58 is completely covered with a faintly shiny yellowish-brown engobe, onto which the details of the mouth, the hands, and two armrings were painted in a darker brown paint. The intended effect may have been to convey the patina of the much more expensive bronze figurines (Fig. 28). The figurative perfume bottles TO 90 and TO 92, manufactured in Central Gaul, prove that slip was not only applied in the Rhineland. They both belonged to the Huybrigts collection and were severely damaged by fire Glaze Although very rare on terracotta figurines,1 glaze was frequently applied to the surface of figurative flasks, often in the form o f an animal.2 One of the zoomorphic flasks from Tongeren is covered with a yellowish glaze. The flask, in the shape of a monkey with her young, was imported from Central Gaul (Fig. 29). A similar yellow to yellow-brown glaze Fig. 29. Detail o f glaze applied on the unguent flask in the shape o f a mon­ key TO 91 found in Tongeren ‘Elfde Novemberwal’. in World War I. The peeling grey surface layer of the central Gaulish pigeon bottle clearly is a fired white slip layer. The colour of this layer and the colour of the core underneath it have been affected by the secondary burning in 1914. The second bottle, depicting a sitting person, is covered with a dark reddish brown slip over the entire body, which now shows crackles because of the fire. A Venus from Brive-la-Gaillarde bears traces o f a yellow glaze: De Beenhouwer 2005,453, no. 3058; an example from the Rhineland is a Venus from Trier w ith the remains o f a green glaze on a preparatory slip. See Schauerte 1985,131 no. 37; De Beenhouwer 1991b, 86. Déchelette 1904,322-324; Jeanlin-Rouvier 1994; Corrocher 1994. - 75 - features on a flask representing a hare, found in the settlement of Braives, also in the civitas Tungrorum (Fig. 30).3 Fig. 30. Glaze on a fragment o f an unguent flask showing the pad o f a hare found in Braives. generally dated to the first century AD, from the Tiberian period and with a peak in the mid-first century.® This matches the general date for glazed ceramics from Central Gaul, which began in about 20 AD, reached a peak between 40 and 80 AD and vanished at the beginning of the second century.9 This early date means that the flask in Tongeren remained in use for a long time, for it was found in a context that arose after the middle of the second century. Figurative flasks are found most often in a funerary context.10Like the hare-shaped flask of Braives, the Tongeren The application of a glaze layer that reduced the porosity of the wall made these flasks suitable for holding liquids. Because of their small capacity and sumptuous execution, it is often assumed that they contained cosmetic products.4Although the Mediterranean region flask was found in the dwelling area, which demonstrates that glazed unguentaria were used not only in the grave cult, but also in domestic situations. Background painting had a long tradition of figurative ceramics, animal-shaped flasks were not made anywhere on such a large scale as in Central Gaul.5Traces In many cases larger parts of the figurine such as the skin or the hairdo were coloured. This is same way using two-part moulds. These glazed figurative flasks from Central Gaul are yellow pigment residue is found more often in Rhineland terracotta.” most apparent for the dark pigment residue in of glazed pottery production have been the hair of several figurines: TO 2, TO r5, TO 22, identified in various production centres.6 TO 2r, TO rr, TO 6r, TO 63, TO 65, TO 66, TO 72, Workshops where figurative flasks were made TO 74, and TO 75. The shiny yellowish layer on often produced also terracotta figurines.7*The the fur of the goat TO 81 and on the face of a reason is obvious; they were moulded in the bust TO 74 is another example. Similar shiny ’ 4 ' ‘ 7 1 ' “ Brulet 1990,76 no. 19,77 fig. 34. Corrocher 1983: ‘récipients à essences parfumées’; Hochuli-Gysel 1998: ‘balsamarium’, ‘Salbgefasse’, ‘Verpackung von Toilettenessenzen1. Van Boekel 1987,776 ‘unguent flask’, ‘unguentaria’; EckaTdt 1999, 67: ‘contained perfumed oils and ointments’. Hochuli-Gysel 1998,72. Workshops identified on the basis o f the presence o f kiln finds, misfires and half-finished products in HochuliGysel 1998,65. Groups defined on the basis o f macroscopic characteristics o f the paste and glaze in Hochuli-Gysel 1998, 66-67. A Central Gaulish chemical group is described in Walton 2004,164-165: the glaze was made in Central Gaul o f PbO alone applied to non-calcareous clays, in contrast to the Italic technique that applied Pb0 -Si0 2 mixtures to calcareous clays. Corrocher 1994,120: Lezoux, Vichy, Saint-Rémy-en-Rollat, Gannat, Yzeure, Thiel-sur-Acolin, Saint-Pourçain-surBesbre and Bourbon-Lancy. They were replaced b y glass flasks after the first century. See Van Boekel 1987,777. They occurred in Oberwiterthur between 30 and 60/70 AD. See Hochuli-Gysel 1998,74. A n important ensemble is the well-known grave from Colchester from the middle o f the first century. One o f the ten animal-shaped flasks from the grave also represented a monkey suckling her young. The grave also contained other vessels w ith a lead glaze. See Van Boekel 1993,246-250; Eckardt 1999. Vilvorder 2010,292. See also Corrocher 1983,32; Desbat 1986,33,37. Van Boekel 1987,777: Netherlands; Van Boekel 1993,244-245: Luxembourg; Hochuli-Gysel 1998,75: Switzerland. Bachmann and Ott 1980,129-130, Probe 9: yellowish glazy leaded pigment; Behrens 1918 (Legionskastell Mainz), 43- 44: “In den Vertiefungen liegt hier und da ein gelber Überzug, von dem es zweifelhaft ist, ob er der Rest der ursprünglichen Bemalung ist oder sich im Lauf der Zeit erst angesetzt hat”; see also De Beenhouwer 2005, Fortuna Series 209 and cock Series 851. - 7 6 - Fig. 31. Faint grey paint borders the eyes. The indication o f the round iris and the arched eyebrows is still visible, especially on the left eye. Faint dark pigment remains in the grooves o f the hair. Bust TO72 from the Tongeren 'Thien Schuur' find. Accent painting The prevalent way of decorating terracotta figurines, especially in the Rhineland work­ shops, undoubtedly is accent painting. Using colour, the artist emphasized certain details, or added new elements to the composition. In the specimens from Tongeren, the colour of the painting is mostly yellowish brown to brow­ nish black. Most painting emphasizes small detail that was already present in the mould of the figurine. In that way attributes or anato­ mical features such as eyes, mouth, and hands are stressed (Fig. 31). Sometimes the painting also adds new elements such as bracelets, a hair ribbon, or a decorative pattern on the front of a plinth or the sides of a chair. Twenty-three figurines from the Tongeren collection show remains of such painting, most of which are urban finds. relief of the figurines were added in painting to the upper arms of the couple TO 58 and the right arm of Mercurius TO 56. Other examples are known, mainly from the Cologne workshops and, less frequently, from the Treverian area.“ Since the double bracelets are represented in the same way on the upper arm of both men and woman, they were worn in Only one figurine, a hen TO 84, with accent the same way without distinction between genders (Fig. 28). Other jewellery was preferable to women. A diadem was painted painting was found in a grave. The remaining pigment is burnt, and where it has on Fortuna TO 20 and Juno TO 28 and a hair ribbon was added on Fortuna TO 8. A similar disappeared, a gray “smoky” trace is left. We assume that this poor preservation is connected with the grave ceremony, by which hair ribbon was used in a parallel of Fortuna TO 25 and TO 26.‘3 Some o f the painted decoration is not the figurine came into contact with the funeral pyre. In any case, the Tongeren urban contexts figurative. It shows a remarkable lozenge­ shaped pattern of diagonally crossing lines. appear to have been a more favourable The pattern is usually painted in a frame that outlines the sides of the chair or on the plinth. Five figurines found in Tongeren, show this environment for the preservation o f the pigments. The “sheltered” burial of Mercurius TO 56 underneath a roof tile and the great depth o f the Thien Schuur deposit must type of pattern on the plinth (Fig. 32). Fig. 32. Yellowish brown lozenge-shaped pattern on the plinth o f Matrona TO 51 from the Tongeren ‘Thien Schuur’ find. certainly have played a part in the conser­ vation of the painting. Nineteen examples showing accent painting are from the Thien Schuur find. The artist sometimes added new detail, such as jewellery, to the design of the moulding. Double bracelets that were not indicated in the The colour o f the preserved Cologne painting is mostly dark brown to black: a standing woman with child from Kerkhove, a Venus from Vechten (Van Boekel 1987, 517-518 No. 95), a standing woman with a ju g in the right hand and a child on the left arm by the Cologne coroplast Vindex from Frankfurt-Heddernheim (Rüger 1980, 63 No. 101). Exceptionally, red bracelets are also found on a standing Apollo b y the Cologne coroplast Servandus with red painting from Arentsburg (Van Boekel 1987,254-255 No. 2). The colour o f a Venus from the Mosel area is red (Schauerte 1985,168 No. 210 Taf. 22,1-3). Sitting Fortuna (Rüger 1980,42 No. 16). The same m otif with paint was applied on a sitting Fortuna (Behrens 1920, 56 No. 5 Taf.21,9) and a sitting woman w ith dog from Bingen (Behrens 1920,57 No. 10a Taf. 21,8). “ 7 7 - One of them, Juno TO 27, was signed by the Cologne coroplast Servandus. The lozenge pattern was used chiefly in the Cologne pottery centre, but some examples prove that plinth one of the glaze splashes landed on the brown pigment of the underlying painting. It is clear that the painting on this figurine was applied before firing. it was also adopted in Treverian workshops. In these workshops a preference for red paint was evident.14*Sometimes a St Andrews cross was also applied to the side of the chair, as in matrona TO 52, which was manufactured in Cologne. Outlining without the filling lozenge pattern is also found. Most series with outlined areas were attributed to Cologne, but the motif has been occasionally adopted in Trier.'s In contrast with the yellow brown to black colour that was mostly used in Cologne, the outlining in Trier is reddish. To conclude, we may argue that the painted details mainly point to the Cologne workshops. The same motifs were often adopted in the Moselle area, but the colour is not identical. Why did pigments of painting on Cologne products survive the centuries? The reason must be searched for in the used technique. Pigments applied before firing (ante cocturam) are supposed to be much more durable than the ones applied afterwards (post cocturam). For some time the question concerning Unintended traces of glaze and colouring slip Not all remains of glaze or colouring slip, nowadays found on the figurines, were intended by the potter. A special example of unintended glaze is displayed by bust TO 74. During firing in a Cologne workshop the head of this figurine became stuck to a glazed object stacked higher up in the oven.18A large stain of green lead glaze dripped from the glazed object onto the head of the bust. In order to remove the figurine from the kiln the potter had to snap the head off the glazed object. On top, where the stain is thickest, a round fractured surface with thick brown resinous edges is visible. At the bottom of the stain, where the glaze becomes thinner, it gradually turns yellow. Little splashes of green glaze also splattered on the plinth and on the brown painting that outlines the bottom edge of the plinth. The underlying accent painting was whether the figurine was painted before or after the firing remained unanswered.16 The solution can be found in Tongeren. Thanks to applied before firing. This incident shows us that the figurines in Cologne were not necessarily fired separately, as they were in one of the busts from the Thien Schuur find (TO 74). In the Cologne workshop were it was Mainz, where a kiln full of misfired figurines manufactured, the accent painting was partly covered unintentionally with glaze during the firing process.17 On the bottom edge of the 14 “ was excavated.19 They could be fired together with other ceramics, in this case glazed pottery. The unintended lead glaze and the fractured surface on the head of the bust did Examples are a sitting Minerva from Frankfurt-Heddernheim (Rüger 1980,49 No. 54) and a sitting woman with dog from Dhronecken with red painting (Schauerte 1985,261 No. 609 Taf. 72,5-6). The figurine o f a gladiator w ith painted delineation o f the base was manufactured in Trier (Trierer Zeitschrift 12, 1937,281IIL18) “ " Bachmann 8t Ott 1980,129; Van Boekel 1987,220; Rose 2005,18: the same issue applies to the masks. See 8.4 uninteded glaze. “ De Beenhouwer 1991b, 72-73. For the use o f green glaze in Cologne, see also Mathea-Foertsch 1999; Hoepken 2005, 62, Farbtafel 1, Warenprobe 3. Witteyer 1996. " - 7 8 - Fig-33- Dark brown accent painting on the upper rim o f the plinth and an unintentional plash o f a reddish brown slip on the back o f Diana TO 3 from the Tongeren ‘Thien Schuur’ find. not prevent the potter from selling the figurine, Apparently the finished article didn’t have to fulfil high standards to meet the needs of the market. Some figurines show traces of unintended colouring slip (Fig. 33). Numerous splashes of a reddish brown slip on the surface of Diana TO 3, Fortuna TO 13 and Minerva TO 38 are a case in point. The reddish colour of the splashes differs from the brown accent painting on the plinth and the cloak of the Diana figurine. These drops do not contribute to the decoration and were therefore not intended. During the drying process, these figurines were unintentionally sprinkled when the potter was performing other activities. Once again this indicates the production of figurines was not a separate industry, but integrated in the fabrication of other ceramic products. - 7 9 - C O R O PLA STS Several of the figurines that were found in Tongeren bear a signature o f the manu­ facturer. These always involve coroplasts who were active in workshops in Roman Cologne. Vius and (At)ticus, 125/200-250 AD Mercurius TO 56 from Tongeren bears the inscription VTVS 11 (C)CA(A) at the back, on the edge of the rock he is seated on. There are no separators between the words. The name VIVS is blurry. The first two letters are clearly legible, but the last two are less clear. The bottom of the V and the top of the S are preserved. Only two vertical grooves of the next word are preserved. After that the letters Fig. 34. A t the top (left and right) the inscription on Mercurius TO 56 from the ‘Sint-Truidersteenweg’. A t the bottom (left and right) the inscription on Mercurius TO 57 from the ‘Elfde NovemberwaT. The figurines are part o f the same series. The inscription is identical Blurred parts o f one inscription can be completed by the other. -80- CA follow. This obviously is a part of the abbreviation CCAA that refers to Roman Cologne, known as Colonia Claudia Ara Agrippinensium. The coroplast didn’t write the full abbreviation. Fortunately, a second Mercurius form the same series, TO 57, bears an identical inscription. The blurred parts of one inscrip­ tion can be completed by the other (Fig. 34). A Mercurius figurine from Hoeselt, DB 333 (947), which is related to the same tree of descent but that belongs to another series, is completed with the regularly formed letters AI.TICVS.FIICIT on the top edge of the chair­ shaped rock and below in diagonal letters Fig. 35. Photographic compilation showing the inscription on the Mercurius figurine from Hoeselt. C.C.A.A in relief (Fig. 35).* Remarkably, the Servandus, 155/160-180/210 AD separators come in the shape of a small, well­ shaped little circle between the words and The plinth of Juno figurine TO 27 that was found near the northern temple in Tongeren letters of the abbreviation. In spite of the word separation after the two first letters, Atticus carries the name Servandus. This famous Cologne coroplast made a major contribution could be a possible reading. We have no further knowledge of a Cologne coroplast to terracotta production in the Rhineland. He signed a large number of terracotta figurines.12 known by that name. His name can also be found on a number of relief lamps, made according to the same moulding technique as the figurines.3 The The relationship between the inscriptions inscriptions not only mention his name, but also a precise location of his workshop and a from Tongeren and Hoeselt remains unclear. Both figurines belong to the same tree of descent, but not to the same series. The inscriptions differ fundamentally and are from two different coroplasts. Their production can be dated between 125/200 and 250 AD. According to the petrographic data, the Mercurius figurines from Hoeselt and Wervik should be attributed to a Cologne workshop. 1 2 J very precise date. Due to this, Servandus’s oeuvre is one of the best described in the entire north-western European terracotta industry in the Roman period. In total, forty-six figurines with his signature are preserved, through which no less than twenty-five series can be attributed directly to Servandus. His reper­ toire includes Venus, Fortuna, Minerva, woman seated with dog, woman seated with The text has been interpreted in various ways in the past (Bamps 1898,71-72: wrong reading “I'IDHVIS”; Schuermans 1898,401: Mercurius from “substructions” from Hoeselt. Schuermans reads the stamp: “k SICVS.FIICIT”; Renard 1905,286-289, No. 1 pi. A: Renard reads “...LICVS.FIICIT” and cites two names (p. 287, note 1) that end on ICVS: VRBICVS (referring to Tudot i860,66) and LVBRICVS (referring to Tudot i860, pi. 64); De Maeyer 1940,107: reading...LICVS FIICIT; Dheedene 1959,82 No. 458: ffilLICVS.FIICIT; Despriet 1975 (Wervik), 29-37 Fig. 1: reading as Renard; Van Boekel 1987,395 sub i: reading AI.TICVS.FIICIT (ATTICVS ?) on the top edge o f the rock and CCP. (CCAA?) below; Lange 1994,271 Series 196 No. 7: [...]ICVS FIICIT/CCA[A] or CCP[.]; Van Boekel 1996, 23: on the edge o f the rock ACCEPTVS.FIICIT and under the rock [AC]CEP[TVS], [A]CCE[PTVS] or C.C.A.[A]; Hdpken 2005,51: [...]ICVS; CIL XIII, 10,015, No. 119: ...ICVS.FIICIT. Schauerte 1985, 60-65; Lange 1994,149-151; De Beenhouwer 2005,1074: Victoria No. 4,144 (500) and figure removing a thorn No. 4,251 (547) do not mention Servandus’ full name, but the initial S. Binsfeld 1964,30 Taf. 10 Fig. 6-14; Hoepken 2005,48: signatures SE, SER, or SERV. - 81 - Fig. 36. Part o f the name o f Servandus and the abbreviation CCAA on TO 27 from the Tongeren north temple. from Kobern-Gondorf mentions the consulate of Macrinus and Celsus in 164 AD, more specifically 13 September of that year. The figurine belongs to a series that is attributed to Servandus.7 The inscription on a figurine of a god accompanied by an animal from Arentsburg has the youngest consular dating yet found on a statuette signed by Servandus. In addition to the name Servandus, the consulate o f P. Coelius Apollinaris and, possibly, joint consul Senecio is mentioned. This refers to the year 169 AD.8*10The dates mentioned on the figurines presumably represent important days. According to Hopken, these figurines were made to mark certain celebrations, related to the calendar of the Colonia or the province of Germania Inferior. Engraving the date on the figurine, the coroplast turned it into a souvenir.8 The inscriptions enable us to demonstrate a fruit and Ubian hat, Cybele, Juno, Diana, Mercurius, Bacchus, and Mars.4 With chronological relationship between Servandus and two other Cologne coroplasts, Acceptus and Lucius. A Mars figurine that was found in certainty, his workshop was established in Roman Cologne. Cologne is mentioned 26 Cologne bears two signatures that were times with the letters CCAA, the abbreviation of Colonia Claudia Ara Agrippinensium or Roman Cologne (Fig. 36). applied at different occasions." At the back, we find a blurred signature by Servandus, which was already present in the mould. At the front, a sharp inscription is applied with Acceptus’ name. A comparison of the freshness of the The location o f the workshop, ad forum inscriptions shows that Servandus created the hordiarium, is also mentioned four times. We cannot yet locate this toponym in Cologne, but model, and that Acceptus copied Servandus work, adding his name without wiping off the the distribution of the finds clearly shows the relationship with the Roman potters’ district near the Rudolfplatz.5 Some of Servandus’ shallow signature of Servandus. This could imply Servandus’ workshop at that time had ceased to exist. The workshop of Acceptus is figurines carry the name of the consuls from also located in Cologne, according to the that year, making very precise dating possible.6 inscriptions, but in his later professional life, For example, the inscription on a Minerva he probably migrated to Trier. A bust carrying * 5 ‘ 7 ’ * 10 De Beenhouwer 2005: Venus series 135,146, and 135, Fortuna series 246,247, and 259, Minerva Series 280, sitting woman with dog Series 368, sitting woman w ith Ubian hat cat. No. 4,098, Cybele Series 465, Juno Series 478, Diana series 484 and 486, Mercurius Series 525, Bacchus Series 532 and 534, and Mars Series 552. De Beenhouwer 2005, series 146,153,246,280, and 486; Hopken 2005,48-49: Fundstelle 30, Hopken 2004,40; Hopken 2005,49. Lange 1994, Series 95; De Beenhouwer 2005, Series 280. Van Boekel 1987, 609 No. 149; Lange 1994,278 Series 214 No. 1. HiSpken 2004. Lange 1994,267-268 Series 186, No. 1, Abb.11; Van Boekel 1996,22 Fig. 43; De Beenhouwer 2005,4,253 (552). -82- his name, and found in Trier Altbachtal, bears witness of this event.“ It is made of red-baking clay paste, a common practice in Trier. Lucius is another coroplast that imitated the work of Servandus. This is illustrated by a series that depicts Venus holding a mirror. In addition to a specimen signed by Servandus, the series also includes two specimens signed by Lucius." The image comparison o f the figurines, demonstrates that the work of between 160 and 180 AD. The final phase is less well defined. The continuation of Acceptus’ workshop in Trier, presumably after the decline of the Cologne ceramic centre, and the dating of the Oudenburg figurine in the first half of the third century, support a proceeding of Servandus’ activities, at least until the nineties of the second century. A plinth with the inscription of Servandus Lucius is younger than the one of Servandus.*13 that was found in Cologne in 2002 The Venus series may have been traded as late demonstrates that not all of Servandus’s as the beginning of the third century14, which creations were original designs.192 0It belongs to was the final phase of the workshops near the a series of busts of a woman with earrings, of Rudolfplatz in Cologne.15 The only which a specimen was found before in archaeologically dated specimen is found in Cologne, and which is also represented by two Oudenburg in a context of the first half of the specimens in the Thien Schuur find in third century AD.16 Because the Oudenburg Tongeren (TO 50 and TO 51).” The original figurine isn’t signed, we cannot be certain whether it was made by Servandus himself or design probably dated from shortly before the middle of the second century.21 Until now, the by Lucius who worked later. We can consider Acceptus and Lucius successors of Servandus, entire production has been anonymous. Just the plinth being preserved, it is not possible to chronologically situate the moulding of or at least younger contemporaries. We know that Lucius was active in Cologne in a different location, “ad gantunas novas”. Thus, the same series was manufactured in two different Servandus within the development of the tree of descent of this bust. Cologne workshops. Victor from Cologne, 180-200/210 AD Relying on the archaeological dating of the series signed by Servandus, the beginning of the workshop has to be situated not long before 160 AD.17 An earlier dating shortly before the middle of the second century is no longer valid.18 His main activity takes place u “ 13 H 15 16 17 18 19 20 11 The bust o f a bald boy (TO 69) found in Tongeren has the letter ‘V’ in relief on a disc­ shaped medallion between plinth and torso. This letter is the initial of Victor, a coroplast whose full name appears twice on a separately formed medallion: once on the bust of a De Beenhouwer 2005,1067. Lange 1994, Series 20. De Beenhouwer 2005, Series 135. Lange 1994,210-211 Series 20; De Beenhouwer 2005, Series 135. Höpken 2005,43: “...Aufgabe der Töpfereien im frühen 3. Jahrhundert. Für das 3. Jahrhundert is in Köln bisher eine Werkstatt überliefert, die jedoch nicht über die Jahrhundertmitte hinaus arbeitete.” De Beenhouwer 2005, No. 832 (135). De Beenhouwer 2005: the Fortuna Series 259 was dated between 150 and 260 CE, Juno Series 465 between 170 and 230 CE, Venus Series 135 between 200 and 250 CE, and the development o f the Tree o f Descent 87 was dated between i 6o/i 8o-192/20o (?). Van Boekel 1987,210 nt. 91,894 nt. 682: early dating based on the Hofstade dating by S. J. De Laet; De Beenhouwer 1993,235 nt. 13: idem; De Beenhouwer 1996: Hofstade s reinterpretation o f the site; see also De Beenhouwer 2005, 861-862,961: Tree o f Descent 15. Höpken 2004,39. De Beenhouwer 1991a, 409 111. 5,3; Lange 1994,280-282 series 221/222 No. 1, 111. 82. De Beenhouwer 2005,916-917,1032-1033: Tree o f Descent 104. - 8 3 - Fig. 37. Impress o f the coroplast Victor on a disc­ shaped medallion on TO 69 from the Tongeren ‘Thien Schuur’ find. petrographic research, the figurine from Tongeren was attributed unequivocally to a Cologne workshop. According to the development of the tree o f descent, the specimen dates from the end of the second or the beginning of the third century AD, which immediately guides our dating o f the workshop.25 A coroplast named Victor was also active in woman from Bonn“ and once on a Fortuna Xanten, possibly around the middle of the second century. Two clay half-moulds, found figurine from Andernach (Fig. 37).2 23 in the ‘Kriemhildstrasse’, testify to this fact.262 7 At the back, we read the inscription VICTOR F. The produced themes are Venus and hen.22It is The production technique of the bust of the woman from Bonn clearly took its inspiration most unlikely that Victor from Cologne who put his name on the medallion of the figurines from central Gaul.24It has a separately formed plinth and medallion, and the half mould of the back only includes the head and shoulders. We therefore cannot ignore the possibility that is the same as the Victor who engraved his name on the clay half-moulds from Xanten, in view of the later dating of the coroplast from Cologne.282 9 the artisan, coming from a central Gaul workshop, migrated to the Rhineland. However, the bust from Tongeren is elaborated Vindex, 140/150-180 AD in is Though no inscriptions of Vindex are found in integrated in the mould of the bust, the back­ side being completely incorporated in the half-mould of the back, and the medallion Tongeren, tree of descent 69 shows a clear relationship between Matrona TO 52 and the work of this coroplast. Vindex offered a being incorporated in the half-mould of the front. Apparently, Victor adapted his production technique to the Rhineland considerable choice. Next to moulded lamps, he signed no less than eight different figurine series.23 His repertoire includes Venus with process. Based on chemical analysis and Priapus, Fortuna, woman standing with child, Rhineland tradition. The plinth 22 23 24 15 26 Van Boekel 1987,215 Fig. 9; De Beenhouwer 2005, 695, No. 4360 (638). Schauerte 1985,203 No. 335. Van Boekel 1987,212-216, 638. De Beenhouwer, 918-919,1035: Tree o f Descent 106. Hinz 1963,395-403, Abb.2 floor plan, Abb.5 moulds and Taf. 48: In i960 and 1961, research was conducted on the ‘Deckers’ lot, in w hich settlement traces were found from the first to the third century in a stratigraphical package two meters thick. This included an artisanal quarter with bread ovens, a potter’s kiln, and wells. In addition to the moulds that were signed b y Victor, a mould o f ‘firmalamp’ Loeschcke Type IX was found, the m ould o f a little Luna head, for the production o f relief appliques and a half-mould o f a standing Jupiter w ith lightning. If one assumes that the moulds belong together and are related to the potter’s kiln, they belong in the second century. Haupt dates the finds in the middle o f the second century (Haupt 1979,203). 27 De Beenhouwer 2005, series 30 and 868. We are still not sure whether the fragment o f the signed half-mould from Xanten actually belongs to this Venus series. The contours and internal proportions are similar, but the loss in height between the mould and the largest preserved specimen o f the series is 17 per cent. Moreover, the other specimens o f the series were produced using plaster moulds. 28 According to Schauerte, the indications are too few to relate the production to the potter’s kiln that was found here. In spite o f the mould finds, the entire coroplastic in Xanten would be uncertain. If terracotta were manufactured here, they would be o f secondary concern, according to Schauerte, and m ay be regarded as a workshop that depended on Cologne (Schauerte 1985,55~56).Van Boekel concludes that the statuettes and moulds probably came from one modeller (Van Boekel 1987,212-216 fig. 7 and 8, 638; Lange 1994,119). 29 Schauerte 1985,57-60: list with wrong notation under K and wrong identification o f Mercurius under E, G, and H; Lange 1994,148-149; De Beenhouwer 2005, series 177,306,317,411,531, and 983. - 8 4 - woman seated with dog, Bacchus and Sucellus. In addition to his name, various figurines also bear the abbreviation CCAA, have worked in that period. Other trees of descent confirm this later dating.38 which provides certainty about the location of his workshop in Cologne.30Like his colleagues Alfius and Lucius, Vindex also mentions the address of his workshop “ad gantunas novas”.31 Fig. 38. Fortuna found in Melick DB 3636 signed at the back o f the plinth by Vindex. The figurine is part o f the same series as the unmarked Fortuna TO 6. Traditionally, the ‘Firmalampen’ o f the Loeschcke-X type with a broad open channel that were signed by Vindex were used to date the workshop.32Lehner dated the lamps to the end of the first and beginning of the second century AD. According to S. Loeschcke, the type with the broad channel only occurs after the first century.33Initially, a dating in the first third of the second century was therefore accepted for Vindex.34According to H. Lange, Vindex worked in the second quarter of the second century. As evidence, he mainly offers the typical hairdo o f some figurines by Vindex.35 Descent scheme 6g offers strong arguments for a date in the beginning of the second half of the second century. A seated woman with a dog on her lap, signed by Vindex, stands at the roots of the Ubian matrona.36 Possibly, he also designed the mould-related matrona TO 52. After all, it was customary for coroplasts to create several themes from a new design, adapting the Fig- 39- Detail o f the back o f the plinth o f Fortuna DB 3636 from Melick. The faint inscription shows part o f the name VINDEX and the lines o f a St Andrews cross on the burned surface. attributes and haircuts corresponding to the needs of the market.37The introduction of the Ubian clothing of the matrons in the religious iconography is related to a cult figure that was consecrated in Bonn in r6r AD. Vindex must 30 31 3J 33 34 35 36 37 38 De Beenhouwer 2005, series 177,306,411, and 983; Höpken 2005,50-51: Fundstelle 28. De Beenhouwer 2005, series 177 and 983. Lehner 1903,191-193 Fig. 2 No. 2 and 3: drawing o f the Trier lamp and signature with CCAA addition; Binsfeld 1964, 29-30, 111. 6 No. 1: drawing o f signed lamp preserved in the Cologne RGM museum. Loeschcke 1919,495. Schauerte 1985, 60; Van Boekel 1987,208, 876, and 895. Lange 1994,148-149: the haircuts o f series 145 and 160 are said to not have originated from the Flavian-Trajan hairpiece, but to be inspired by the late-Hadrian to early-Antonius haircut o f Sabina, Hadrian’s wife; see also Wegner 1956, 84. See also De Beenhouwer 1991 (statuettes), 74-76: comparative study o f the series o f Vindex and the matrons with Ubian hat. De Beenhouwer 2005,844-1050: trees o f descent 35,38,46, 67, and 87. De Beenhouwer 2005,1138: trees o f descent 15 and 50. The hairdo, m otif 21 that is applied twice in the signed series o f Vindex is dated in the tree o f descent between 140/150 and 170/180. -85- Pr o v e n a n c e To determine the provenance of the figurines, there is a broad range of research methods available, ranging from a simple observation of the figurine to extensive scientific analyses. Each method yields further clues. The inter­ Juno, caught the attention of Schoppa.3A com­ parison of its style with other examples, most notably from Bingen and Heddernheim but also from Tongeren, led him to conclude that there was a stylistically uniform group of ter­ racotta figurines, manufactured in the same connection between the mould-made fig­ urines, revealed through the study of the workshop. Based on the find distribution, this series and the tree of descent, increases the range of application. hypothetical workshop had to be sought in the Middle Rhine area. In Schoppa’s view, two fig­ urines from Tongeren belonged to this group. One is the Tongeren Juno TO 31, which is part Stylistic analysis: the Thien Schuur example In the first study of the figurines from the Thien Schuur find in 1932, Van de Weerd expressed his belief that the figurines came from a single workshop in the Lower Rhine area.1He established a link with products from Cologne, but was unable to substantiate his assertion because it was based on literature alone. He was aware that the matter could only of the same series as the Juno of Hofheim DB 4107. He also placed the seated woman with fruit TO 53 in this group. The 1980s saw the compilation of major stud­ ies of terracotta figurines. Various authors adopted Schoppa’s ideas.4As a result, a total of 25 figurines from Tongeren were attributed to the hypothetical production centre. Twentytwo of them, spread across 13 series, were part be resolved by comparing the paste of the Cologne figurines with that of the hoard.2 Twenty years later, a number of terracotta fig­ urines were found during an archaeological of the Thien Schuur terracotta find (Table 3). Although there was consensus on most attri­ butions, some series were placed in the Middle Rhine group by one author, but were viewed by exploration of the canabae of the castellum of Hofheim. One of them, a representation of others as originating from a Cologne work­ shop. Attributions on the basis of stylistic fea- 1 2 3 4 Van de Weerd 1932,295-299. Van de Weerd 1932,297. Schoppa i960. Schauerte 1985; Van Boekel 1987; see also critical note on the find distribution in Rüger 1980,28. -86- Tongeren them e series Scho pp a Schau erte 1985 Van Boekel 1987 1960 chem ical petrographical an alysis an alysis paste g ro u p 3 D iana 4 Fortuna M iddle Rhine 7 Fortuna M iddle Rhine 8 Fortuna M iddle Rhine uncertain paste 26 9 Fortuna M iddle Rhine uncertain paste 26 10 Fortuna C ologn e 11 Fortuna C o lo gn e C ologn e paste 24 12 Fortuna M iddle Rhine 13 Fortuna C ologn e paste 23 14 Fortuna M iddle Rhine 17 Fortuna C o lo gn e M iddle Rhine M idd le Rhine C o lo gn e C o lo gn e paste 20 C o lo gn e paste 24 (TO 4) C ologn e paste 20 (TO 9, TO 12) paste 21 (TO 13) C ologn e M iddle Rhine C ologn e paste 22 (TO 25), paste 24 (TO 26) 19 Ju n o C ologn e M idd le Rhine 20 Ju n o M iddle Rhine M iddle Rhine M iddle Rhine M iddle Rhine 21 '' Ju n o M iddle Rhine C o lo gn e paste 25 C ologn e paste 23 (TO 3 2 ) 23 M inerva 25 M inerva 26 M inerva 27 M inerva C o lo g n e C ologn e paste 20 (TO 3 8 ,4 2 ), paste 21 (TO 43) 28 Venus M idd le Rhine M iddle Rhine C o lo gn e paste 20 (TO 45) uncertain paste 26 (TO 48) or C o lo gn e 29 Venus M idd le Rhine M iddle Rhine 32 M atrona C o lo gn e C o lo gn e 33 M atrona C o lo g n e C o lo gn e 34 w o m an w ith fru it M idd le Rhine M iddle Rhine 38 co u ple 39 bust o f a boy 40 bust o f a boy 42 bust o f a boy M idd le Rhine C ologn e C o lo gn e C ologn e paste 20 C o lo gn e paste 21 C o lo gn e paste 22 C o lo gn e paste 20 C o lo gn e (TO 64) 43 bust o f a boy 44 bust o f a boy 45 bust o f a b o y C o lo gn e Trier 46 bust o f a boy 47 bust o f a w o m an 48 bust o f a w o m an paste 20 (TO 63), paste 24 (TO 64) C ologn e paste 24 C o lo gn e paste 21 (TO 67) C ologn e paste 20 Treveri paste 34 49 bust o f a w om an C o lo gn e paste 21 50 bust o f a w om an C o lo gn e paste 23 51 bust o f a w o m an 61 rooster C o lo gn e paste 20 Fran kfu rt Table 3. Stylistic analysis: the ‘Thien Schuur’ example. - 8 7 - tures are obviously open to interpretation; they are not a sound basis for resolving issues of provenance. This has already been pointed out by the present author in a 1991 study com­ paring the figurines of the ‘Thien Schuur’ with finds from Cologne, Trier and Heddernheim, this time not on the basis of stylistic features, but of observations of the paste,5 technical aspects of manufacture and decorative tech­ nique.6 The classification of the paste, based mainly on the presence of crystalline inclu­ evolution o f descent scheme 5, which also includes the Juno figurines that - together with the Fortunas - were the point of departure for Schoppa’s stylistic group.9 Thus the similari­ ties that Schoppa had observed were the result of the typical reproduction technique for terra­ cottas, whereby similar characteristics were transferred through the mechanical process of moulding, which transcends the individual workshop. This brief survey of provenance issues for the Thien Schuur figurines makes it sions, yielded a homogeneous picture for the series under study, with the exception of TO 70. A comparison with products from the mould-made terracottas. Cologne and Trier workshops led to the con­ clusion that TO 70 was produced in Trier and the other figurines in Cologne. The study It should be stressed that from these series, only the specimens from the Thien Schuur clear that mere stylistic analysis is not a useful criterion for establishing the provenance of therefore also raised legitimate doubts about the Middle Rhine group, as defined by Schoppa and his followers on the basis of sty­ listic features alone. Lange confirmed this find were considered. It can not be exlcuded that more than one workshop is involved in doubts with his catalogue of Cologne terracot­ tas.7He attributed a total of 32 series from the Neerharen (DB 142) that belongs to the same series as TO 46-48 and that corresponds to the Thien Schuur find to a Cologne workshop. He believed that seven of these series could also have been manufactured in other workshops besides Cologne. chemical composition of a workshop situated around Frankfurt. the production of a series, as suggests the result of the chemical analysis of a Venus from A thorough observation More recent the paste of the figurines of the ‘Thien Schuur’ was analysed petrographically as part of a comprehensive study of Northwest European terracottas.8 In addition, seven fig­ urines were selected for chemical analysis. The results confirmed the earlier hypothesis from Assembly characteristics - afirst step The first key research method is observation of the production technique. During the Roman period there were two major production regions for terracotta figurines in Northwest Europe, each of which developed its own tradi­ 1991. According to the chemical composition of the clay, To 70 was produced in a Trier work­ shop, while the other six figurines were made in Cologne (Table 3). One of them, Fortuna TO tion in terms of both production technique and themes for representation.10 One was Central Gaul, where production centred on the Allier river and further north around Autun. 4, belonged to the hypothetical Middle Rhine group as described by Schoppa and others. The other was the Rhineland, including the Mosel region, where Cologne and Trier played a key role (Fig. 1). This Fortuna series is an important link in the * 7 * ’ ” W ith a closed find like the Thien Schuur hoard in Tongeren, we can assume that the soil acted upon the figurines in the same way and that differences in colour and hardness are relevant. De Beenhouwer 1991 a; De Beenhouwer 1991b Lange 1994. De Beenhouwer 2005. See descent scheme 5 and footnote 6. See the chapter on the reproduction process. - 8 8 - One of the main differences is that clay moulds were used in Central Gaul, while the Rhineland mainly used plaster moulds. Air bubbles in plaster moulds often left traces on the external surface of the figurines in the form of small blisters (Fig.3). Another important characteristic is how the two halves of the figurines were put together. With the Central Gaulish figurines, the lightly dried halves were stuck together with a watered-down clay paste or barbotine. This required a deft moulding as the halves had to fit together perfectly. The seams then required little or no touching up. In the Rhineland tra­ On the basis of mould characteristics and assembly technique, it turns out that only four of all the terracotta figurines from Roman Tongeren were manufactured in Central Gaul; the others are of Rhineland origin. This is not surprising. If we include the wider region, we find that the civitas Tungrorum was mainly supplied by Rhineland terracotta workshops, whereas the neighbouring civitas Nerviorum was chiefly supplied by workshops in Central Gaul.“ The few Central Gaulish imports in Tongeren are all animal figurines, which were primarily used as grave goods.112 dition the halves were joined while the paste The manufacturer’s hallmark Some figurines bear the name of the maker or was still fully plastic. The halves were pressed together while still in the mould, and super­ the place of origin. It is this hallmark that pro­ vides the most certain identification when it fluous paste was forced out along the seams. This meant that a broad strip along the seam comes to provenance. Examples are known in Tongeren from different Cologne coroplasts: Servandus, Victor and Vius (Fig. 34 - 37). Other on the outside had to be smoothed and touched up (Fig. 5). On the inside only a small band at the base had to be touched up. The rest of the interior was inaccessible and was left as Tongeren figurines can be associated indirect­ ly with the work of two other Cologne work­ shops, namely that of Vindex and (At)ticus. it was. To this day the expelled paste at the seam on the inside of the figurines betrays the These workshops are discussed in the chapter on coroplasts. assembly method used by the coroplasts of antiquity (Fig. 6). A third major difference is the number of parts from which figurines were composed. Whereas the bases of busts and standing fig­ ures in Central Gaul were usually made sepa­ rately, they tended to be integrated into the mould for the body in the Rhineland figurines. The same applies to mammals. The Rhineland coroplasts sought to simplify the production method as much as possible by integrating the separate components such as legs, tail and base into the body mould, even though the fig­ urine lost some of its elegance in the process. 11 “ Series distribution Another method for finding out more about provenance is to examine the distribution of findspots for figurines belonging to the same series. If we look for example at the distribu­ tion of the Fortunas in series 4, we can see at a glance just how many figurines have been found in Cologne and, more specifically, in the workshops ‘am Rudolfplatz’. Two examples come from Nijmegen, where terracottas were produced on a small scale and only for the local market. It therefore is likely that all examples in this series were made in Cologne. A greater difficulty is presented by the De Beenhouwer 2005, 805-808. Grave goods: dog TO 79 (AD 98/117-138/200), dog TO 80 (Flavian) and pigeon TO 87 (AD 138-161/200). Settlement find: quadruped TO 82 (AD 70-150). - 8 9 - Fortunas in series 7, where the parallels for the ones in Tongeren come from FrankfurtHeddemheim, Cologne and Xanten. The three findspots are known as production centres for terracotta figurines and it is impossible on the basis of the find distribution alone to find out which of the three centres produced the figu­ rines that were imported into Tongeren. It can’t be taken for granted that all examples in a series were by definition made in a single workshop. This is demonstrated by Minerva series 27, where alongside the white-firing examples one Minerva was made in red-firing paste. A different material could point to a dif­ ferent workshop. Decorative technique Observations of the paste, but also decorative techniques, can link figurines to a particular workshop or production centre. The glazes, slips and painted motifs have been discussed at length in the chapter on painting. Some slips are typical of a particular workshop. Different Tongeren figurines are covered with a thin, subtle layer of slip. Of these, the group with an orange core and a thin white slip layer can clearly be attributed to Cologne, as demonstrated by the chemical analysis of TO 4 and TO 64. Several white-firing figurines were Cologne, but to a lesser extent also in Trier.14 Although a brown hue was generally used in Cologne and redder hues were preferred in Trier, determining the provenance on the basis of these individual motifs sometimes remains uncertain. There is greater certainty where dif­ ferent motifs were combined into original decors, as with the Mercurius series 37 that bears a Cologne hallmark. Here, besides an outline of the base, both anatomical details and attributes are marked in a brown hue, and new details that were not present in the mould, such as arm bands, are also painted on. The combination o f a St Andrews cross, lozenge-shaped painting, outlining o f the grounds of the seat and accenting of anatomi­ cal details and attributes is also found in the Ubian matronae TO 52, all in the brown hue typical of Cologne. Chemical analysis Analysing the chemical composition is a tried and tested method for determining the prove­ nance of pottery. A 100 mg sample is required in order to obtain a representative picture of the relationship between the major elements by means of X-ray fluorescence spectrometry (XRF). Because of its destructive nature, this method is applied selectively. It has already also finished off with a thin layer of slip. Because the slip is the same colour as the core, it is often difficult to detect. We know that this proved useful in the past for analysing the provenance of terracotta figurines. As part of a technique was also used in Cologne because one of the figurines, Mercurius TO 56, was research project on Roman ceramics from the provinces north of the Alps, Schneider1516exam­ signed by a Cologne coroplast. Decorative technique helps to identify a workshop. For example, the St Andrews cross that is used as a ground m otif has so far only been found on ined 13 terracotta figurines from NidaHeddemheim.*6He identified a uniform group Cologne products.13 Other motifs such as the lozenge pattern and the outline around the sides of the plinth or seat occur mainly in a 14 15 16 of white-firing figurines with a high Ti02 con­ tent, which differed from the production in Cologne and Trier and was closely linked to the ceramics from Franlcfurt-Nied. In addition, an important terracotta find from Xanten could De Beenhouwer 2005,1160 m otif 6. De Beenhouwer 2005,1163-1164, m otifs 5 and 14. Arbeitsgruppe Archäometrie o f the Freie Universität Berlin. Schneider 1988; Schneider 1990. -go- be ascribed to Cologne and Trier workshops on the basis of a chem­ ical analysis of the paste.17 Table 4. Chemical compostion. The major elements are expressed in percentages and the trace elements in ppm. Eleven figurines from Tongeren were selected for an analysis of their chemical composition.18The analysis was carried out by Dr J. Naud.19*2 1 The samples taken from Rhineland figurines were com­ pared with the analysis results of the Arbeitsgruppe Archäometrie of the Freie Universität Berlin.“ The samples from Central Gaulish figurines were compared with published analysis results from both the Laboratoire de Recherche des Musees de France” and the Arbeitsgruppe Archäometrie." In this way it was possible to test our data against existing chemical groups. A Mercurius figurine, DB 4198, from the group with the high Ti02 content was reanalysed in order to optimise the comparison of the results (Table 4).*1 17 18 19 10 21 “ 13 Liesen 2003. De Beenhouwer 2005,1081-1095. *9 samples o f terracottas from Belgian find contexts were examined. The project was partially funded b y the Departement Archéologie o f the KU Leuven (M. Lodewijckx) and the Centre de Recherches d’Archéologie Nationale o f the Université Catholique de Louvain (R. Brulet). Laboratoire de Géologie et Minéralogie o f the Université Catholique de Louvain (Département de Géologie et de Géographie). Schneider 1988,322-323 Tab.2. and info. G. Schneider; Liesen 2003. Camuset-Le Porzou 1985; Rey-Delqué 1.985; Jeanlin and Lahanier in Rabeisen and Vertet 1986,208. Raselli-Nydegger 1998, 84 Tab. 10. Rüger 1980,114 Nr. 308; Schneider 1988,322-323 Tab. 2 Analysis Nr. 4766. The dual analysis from Naud’s labora­ tory has a slightly lower Ti0 2 and K20 content than the analysis from Schneider’s laboratory. It is important to bear this m argin in mind when interpreting the other analysis results. 91 Fig. 40. Scatter diagram Ti0 2/K20 o f the Rhineland examples listed in table 4. Also included is a Venus figurine made by Servandus found in Xanten (Liesen 2003,320 Nr. 63 Abb. 3,3: sample J562), joining the red group from Cologne. Titanium/potassium 3.00 2,50