School of Art History, Cinema, Classics and Archaeology The University of Melbourne

Transcription

1 THE CONSTRUCTIONS OF MARCUS AGRIPPA IN THE WEST Geoffrey Mottershead Submitted in total fulfilment of the requirements of the degree of Doctor of Philosophy March 2005 School of Art History, Cinema, Classics and Archaeology The University of Melbourne

2 Dedicated to HILARY PATRICIA ANNE

3 ABSTRACT i Marcus Agrippa was the chief supporter of Octavian, the heir of Julius Caesar, in his rise to prominence as the first Roman Emperor Augustus. He also played a central role in the Augustan establishment of the new order of Empire, which replaced the late Republic. Agrippa's land and sea victories were crucial for the success of Octavian, but it will be argued that his constructions were important instruments of change in this pivotal historical period. Consequently, all Agrippan works are investigated, whether for war or peace, and whether known from material remains or other evidence. Agrippan constructions in the West (Gaul and the Iberian Peninsula) are described in detail, and the others are described in more general terms. Previously, Agrippan constructions have either been included in biographies of Agrippa, and treated generally, or have been studied as particular works with detailed description, but little explanation. Also, constructions in towns with material remains or inscriptions have been extensively studied, but important works outside towns with fewer remains have been largely overlooked. Consequently, previous writers have represented Agrippa as a builder of monuments in towns and there has been little understanding of the nature or purpose of the totality of his works, and no proper account of them. The comprehensive analysis in this thesis shows that the principal Agrippan construction in the West was the road network, radiating from Lugdunum (Lyon) to the Atlantic, the English Channel and the Rhine. This constituted the first physical framework of Roman control outside the more settled south of Gaul. Agrippa's planning and part construction of a monumental town at Augusta Emerita (Mérida), and his probable foundation of Caesaraugusta (Zaragoza) were part and parcel of Roman expansion by colonisation into new parts of the Iberian Peninsula. These major works in Gaul and Hispania set a pattern of development for centuries to come and were, therefore, influential over a long period. This was not the case in the East, which was already Hellenised and under Roman hegemony. In the case of Rome, Agrippan works were subsumed in the building programme of Augustus and later works. Consequently, the Agrippan constructions in the West were, ultimately, more important than those elsewhere. The general review of Agrippa's works contained in this thesis leads, therefore, to conclusions which represent a new viewpoint on Agrippan constructions in the West. He was certainly a builder of monuments in towns in the region, as previously known, but it has been shown here that his constructions in the West were instruments of important regional change in a pivotal historical period. There is also a new and more detailed description of the works in the West. In addition, constructions elsewhere are explained and described generally in a new integrated study of all the works.

4 ii This is to certify that: (i) the thesis comprises only my original work towards the PhD; (ii) due acknowledgement has been made in the text to all material used; (iii) the thesis is less than 100,000 words in length, exclusive of tables, maps, bibliographies and appendices. GEOFFREY MOTTERSHEAD

5 iii PREFACE As a civil and structural engineer, who later studied Classical Archaeology, ancient structures were already of particular interest to the author. The idea of writing a thesis about the constructions of Agrippa arose from reading a comment about Agrippa in Reinhold's biography. He said: 'If Augustus was the architect of the Roman Empire, Marcus Agrippa was his superintendent of construction' 1. Reinhold was referring to Agrippa's support of Octavian, the heir of Julius Caesar, in his rise to supreme power in the Roman world as Augustus, the first Roman Emperor, and to his continuing help in the Augustan shaping of the new Imperial Order replacing that of the Republic. He had in mind all the actions of Agrippa, including his land and sea victories, his general support, and, also, his important physical constructions. Reinhold's description of Agrippa as 'superintendent of construction' prompted consideration of a study of his actual physical constructions. In this, the constructions would be explained as part and parcel of Agrippa's activities. Consequently, all the constructions would be included, whether for war or peace, and whether they are identified as Agrippan from their materials remains, or from other evidence. The best possible description of the works was to be provided. It soon became clear that it would not be possible to explain all the Agrippan constructions and to also describe them in detail. A concentration on one particular region was necessary. Agrippan works at Rome were primarily intended to influence the citizens in favour of Octavian, and to introduce the New Order of Augustus. Those in Italy formed no cohesive corpus, and, in the East, the Agrippan works were not important agents of change. The remaining region of Agrippan construction was Gaul and the Iberian Peninsula, where the constructions were instruments of substantial change. It was, therefore, possible to develop hypotheses about their relationship to important historical development. Also, although there were substantial material remains of Agrippan works in the West, it was possible to describe the constructions there in detail in the space available. It was, therefore, decided to focus on the Agrippan constructions in the West. Yet it was not possible to fully account for these without also considering those elsewhere as part of the whole corpus of Agrippan works. These other constructions, were, therefore, to be not only explained, but, also, to be described to the extent that this cast light on those in the West. 1 Reinhold M., Marcus Agrippa, Rome, 1965.

6 iv ACKNOWLEDGEMENTS I am grateful for the help given by Trinidad Nogales Basarrate of the Museo Nacionàl de Arte Romano at Mérida, who generously provided me with a great deal of information and gave me publications of which I was unaware. At Lyon, Armand Desbat kindly took me on a tour of his investigations at the 'Temple of Cybele' site and of other monuments. He also presented me with several publications which brought me up-to-date on archaeological investigations at Lyon. I would like to thank my supervisors, Professor Frank Sear and Professor Ron Ridley for their patient guidance through the many difficulties of research and writing, and for their encouragement throughout the entire period of investigation. Finally, any errors or omissions in the final text are mine rather than those of Vyvyian Ferguson-Sharp who prepared it.

7 ` Vol. 1 Text TABLE OF CONTENTS PAGE THE INVESTIGATION 1 GAUL Figs. A and B 9, 10 Introduction 11 Chapter1: Agrippa's road system 14 " 2: Lugdunum (Lyon) 39 " 3: The Temple of Valetudo at Glanum 62 " 4: Nemausus (Nîmes) 84 " 5: Other Agrippan works in Gaul 98 Agrippan Works in Gaul : Conclusions 107 HISPANIA Figs. C and D 109, 110 Introduction 111 Chapter 6: North-West Hispania 114 " 7: Agrippa at Augusta Emerita (Mérida) 128 " 8: The theatre at Augusta Emerita : Agrippan works 137 " 9: Possible Agrippan works at Augusta Emerita 162 Augusta Emerita : Conclusions 180 Chapter 10: Agrippan constructions at urban centres other than Augusta Emerita 182 Agrippan Works in Hispania : Conclusions 202 OTHER REGIONS 207 Chapter 11: Agrippan works other than in the West 207 AGRIPPAN CONSTRUCTIONS IN THE WEST : CONCLUSIONS 218 ABBREVIATIONS 223 BIBLIOGRAPHY 225 (1) Texts 225 (2) Modern Works 227 (3) Additional Items 246 Vol. 2 Illustrations of the chapters ILLUSTRATIONS Chapter 1: Figs " 2: " " 3: " " 4: " " 5: " " 6: " " 7: " " 8: " " 9: " " 10: " " 11: " SOURCES OF ILLUSTRATIONS

8

9 - 1 - THE INVESTIGATION HISTORICAL INTRODUCTION The historical introduction is not fully documented by reference to footnotes since it is intended to provide only a general background for the detailed study of the Agrippan constructions. These are documented by reference to texts and modern works. Agrippa (Marcus Vipsanius Agrippa) was probably born in 63 B.C. 1 and, therefore, in the same year as Octavius (the first Emperor Augustus). Julius Caesar sent both of them to Apollonia on the Adriatic coast of Epirus in B.C. for education and military training 2. Agrippa and the future Augustus formed a close association which was to be maintained until the untimely death of Agrippa in 12 B.C., some quarter of a century before the end of the principate of Augustus in A.D. 14. The relationship between Agrippa and Octavius was governed by their origins and natural talents. As the heir of Julius Caesar, Octavius was to become Gaius Julius Caesar Octavianus, generally referred to as Octavian. Octavian could advance his position as the heir to Caesar, and because he was from an appropriate social class 3. Agrippa, whose background was humble 4, could advance with Octavian by using his talents as a military commander, organiser of practical affairs, and technician. On the assassination of Caesar in 44 B.C., Octavian, at the age of 18, travelled to Rome and the protracted struggle to take up his inheritance began. The 'Second Triumvirate' of Antony, Octavian and Lepidus (43 B.C.), the triumph of the Caesarian party at Philippi in 42 B.C., and the subsequent wars of Perusia need not be described, since there is no indication of any Agrippan works of this period 5. Agrippa became praetor urbanus at Rome in 40 B.C., where he celebrated the Cassius Dio, dated Agrippa's death to 12 B.C. and Pliny, HN, 7.46, described him as 51 years of age at his death. Also, Reinhold, M., Marcus Agrippa, Rome, 1965, p.1, and Roddaz, J.M., Marcus Agrippa, Rome, 1984, pp Reinhold, pp.12-15, Roddaz, Marcus Agrippa, pp Suetonius, Augustus, 1-7; Reinhold, pp.1-11, Roddaz, Marcus Agrippa, pp Reinhold, pp Roddaz, Marcus Agrippa, pp Cary, M., and Scullard, H.H., A History of Rome, (Third Edition), London, 1975, pp , for a general account of the period.

10 - 2 - Ludi Apollinares 6. At the Peace of Brundisium in 40 B.C., Antony received control of the East, leaving Octavian Italy, Gaul, Spain and Illyricum, with Lepidus in charge of Africa. Antony's wife, Fulvia, had died, and Antony took Octavian's sister, Octavia, as his wife, creating a dynastic alliance 7. After the elimination of Salvidienus Rufus 8, Agrippa was Octavian's right hand man in war. Shortly after the Peace of Brundisium, Agrippa was sent to Gaul to take control there for Octavian. He won a brilliant victory over the Aquitani and crossed the Rhine, as the first Roman general to do so since Caesar. It is argued here that he planned and built the first form of a strategic road system radiating from Lyon (Lugdunum), to the Atlantic Ocean, the English Channel and the Rhine, and connected to southern Gaul. His Temple of Valetudo at Glanum in southern Gaul is also from this period, as are possible works at Arles (Arelate) and at Fréjus (Forum Iulii). Agrippa was recalled to Rome after about two years to take up his first consulship 9, and to break the sea power of Sextus Pompey, the surviving son of Pompey the Great, who had seized Sicily and was threatening the corn supply of Rome with his fleet. Octavian had suffered a severe defeat in a naval action against him 10. Agrippa constructed a new port in the Bay of Naples, where he built a fleet and trained it, defeating Sextus Pompey at Naulochos and Mylae in 36 B.C. Octavian could now guarantee the corn supply of Rome and control the seas round Italy. Agrippa campaigned in Illyria and Pannonia in 35 and 34 B.C., but there is no evidence of any Agrippan works associated with his actions there 11. On his return to Rome, Agrippa, as aedile, embarked on an enormous programme of works in 34 and 33 B.C. to improve the water supply and other infrastructure of the city, which had fallen into disrepair during the civil wars. These constructions greatly improved the living conditions of the people and were an important cause of Octavian becoming more popular than Cassius Dio, 48.20,2 Reinhold, p.22 and Note 6, Roddaz, Marcus Agrippa, p.62. Cary and Scullard, pp.291, 2. Salvidienus Rufus, who had been with Octavius and Agrippa at Apollonia, was Governor of Transalpine Gaul for Octavian, but offered his legions to Antony who divulged this information to Octavian at the Peace of Brundisium, Reinhold, p.24, Note 26. Ehrenberg, V., and Jones, A.H.M., Documents Illustrating the Reigns of Augustus and Tiberius, Oxford, 1976, p.33, also Reinhold, p.28 and Note 1. Cassius Dio, , and Cary and Scullard, pp.292, 3. Reinhold, p.45, Roddaz, Marcus Agrippa, pp

11 - 3 - Antony, and in Octavian's success in persuading the Senate to declare war against Cleopatra, and, therefore, also against Antony 12. As admiral, Agrippa played a major role in securing the victory for Octavian over Antony and Cleopatra at Actium in 31 B.C. He became consul for the second and third (and last) time in 28 and 27 B.C. respectively 13. The part played by Agrippa in the re-organisation of the State in 27 B.C. cannot be gauged from the evidence, but it must be assumed that, as consul, he had some influence 14. Agrippa created a new quarter of Rome in the Campus Martius in B.C. Buildings there marked the naval victories of Augustus (and Agrippa), and the connection of Augustus to Caesar and the gods. Agrippa also built voting places for the people, thus acknowledging their political rights, and created baths and other facilities for their recreation. These works were intended to symbolise the new constitutional arrangements of 27 B.C. Agrippa's new aqueduct, the Aqua Virgo, which served the Campus Martius, was dedicated in 19 B.C. 15. Agrippa withdrew from Rome in 23 B.C. under circumstances related to the succession and the marriage of Julia, the daughter of Augustus, to Marcellus, the son of Octavia. Agrippa's powers in the East and any constructions there are discussed below 16. On his return from the East, Agrippa married Julia, the widow of Marcellus 17, who gave birth in 20 B.C. to Gaius 18, and, later, to Lucius, and Agrippa Postumus, although none became the princeps. Agrippa was sent to Gaul in late 20 B.C. to restore order, and left it in the early part of 19 B.C. The only strong indication of his construction in Gaul is a praetorium at Lugdunum, built in about 20 B.C., although this cannot be attributed to him positively. Agrippa may also have drawn up a large scale plan of monumental development for the town, including its Cassius Dio, , also below Chapter 11, p.207. Ehrenberg & Jones, p.35. The set speech (incomplete) put into the mouth of Agrippa by Cassius Dio, , together with the longer reply of Maecenas, , provided no reliable information on Agrippa's contribution to the first settlement of 27 B.C. Cassius Dio, for the aqueduct and below Chapter 11, Notes 3, 5, 6 and 8-11, pp.208, 9 for the works. See below Chapter 11, Note 20, p.212. Cassius Dio, , also Roddaz, Marcus Agrippa, pp Cassius Dio

12 - 4 - aqueducts, and may have started construction on some works in addition to the praetorium 19. Augustus had set in train the construction of an Augusteum at Nîmes (Nemausus) in 26/25 B.C. and Agrippa may have advanced work there in 19 B.C. There is no definite evidence of Agrippan works elsewhere at Nemausus, despite the existence of inscriptions there which might indicate this 20. It is argued here that any Agrippan road works in Gaul in 19 B.C. were cosmetic. The pacification of N.W. Hispania by Augustus in B.C. was incomplete and the region was finally subdued by Agrippa in a difficult campaign in 19 B.C., which called for substantial Agrippan military works. There is no evidence of Agrippan civil works in the region. Agrippa provided the theatre at Mérida (Augusta Emerita), founded by Carisius in 25 B.C. This became the capital of the province of Lusitania, possibly inaugurated by Augustus during his visit to the West in B.C. It is argued here that Agrippa was the 'father' of the monumental town with its bridges and aqueducts, and that he started to construct other works in addition to the theatre 21. It is also suggested that Agrippa almost certainly founded the first Roman town at Zaragoza (Caesaraugusta), and that he may have established a new naval base at the Punic city of Gades (modern Cádiz) 22. It is possible that Agrippa was responsible for other constructions in the Iberian Peninsula, but there is no definite evidence of these and it is unlikely that they were important. Agrippa returned to Rome from Spain in the first half of 18 B.C., and received the tribunicia potestas for the first time when Augustus was having his renewed 23. He was now almost equal to Augustus. Agrippa may have built the first form of the theatre at Ostia before he left for the East in 17 B.C., where he remained until 13 B.C. Agrippa built the large concert hall, or Odeion, in the Agora of Athens in B.C. and, possibly, other works at the city. He also built at Antioch, the Roman capital of Syria, certainly after Actium and in B.C., and, possibly, See below Chapter 2 for the Agrippan works at Lugdunum. See below Chapter 4 for the Agrippan works at Nemausus. Cassius Dio, for the campaigns of Augustus in 26/25 B.C. and for the foundation of Augusta Emerita, for Agrippa's movement to Spain and his campaign, and for the visit of Augusts to the West. Also below, Chapters 6 for the campaign, 7 for Agrippa at Augusta Emerita, 8 for the theatre and 9 for possible Agrippan works at the town. See below Chapter 10, pp.187, 8 for the naval base and pp for Caesaraugusta. Cassius Dio,

13 - 5 - also in his earlier visit of B.C. Agrippa may have founded colonies at Patras, in the Gulf of Corinth, at Alexandria Troas, near Troy, and at Beirut in the Lebanon, but the dates of his possible actions there are uncertain and may be as early as the period shortly after Actium. Agrippa campaigned in Pannonia in B.C. and died in 12 B.C. while returning to Rome 24. THE SUBJECT OF INVESTIGATION The constructions referred to in the historical introduction included civil works such as roads, buildings of various kinds, port facilities, and also those for military campaigns. Agrippa built ships with special mechanical devices 25, but these are not included in this study. Some of the constructions, and particularly military ones, were built in timber or earth and little or nothing of these has remained. Building in stone has often survived, and sometimes substantially, and there has, therefore, been a previous focus on investigation of these. Yet all Agrippan works for which there is reliable evidence, whether from inscriptions, texts, or other valid sources are to be included here. Furthermore, both the completed entity and the planning from which this resulted are to be investigated. For military constructions, and some civil works, the decision to build, and the construction were virtually simultaneous. Some other works, which were part of an Agrippan town plan, were built later, and, in some cases, after his death. Here, therefore, Agrippa is investigated as the 'father' of general development, the planner of something more specific and the builder of a particular construction. All constructions are to be explained, but only those in the West are to be described in detail. CRITERIA FOR INCLUSION OF CONSTRUCTIONS The known Agrippan works are of two types. The first is a specific construction which is known to be Agrippan because of an inscription or a text. In the West, only the Theatre at Mérida in Spain and the Temple of Valetudo at Glanum have inscriptions naming Agrippa 26. The Agrippan road network in Gaul is identified as See below Chapter 11, p , for the Agrippan constructions and possible foundations after his return to Italy in 18 B.C. until his death in 12 B.C., also Roddaz, Marcus Agrippa, Chapter 3, 'Prince de l'orient', pp for his activities in the East. Cassius Dio, 49, 12 and Appian, and , described the ships built by Agrippa to defeat Sextus Pompey. See below Chapters 8, pp , and 3, p.70, for the inscriptions at the theatre and the temple respectively.

14 - 6 - Agrippan from a text of Strabo 27. The second type of known Agrippan works is identified by Agrippan actions recorded in texts or other reliable evidence, which, necessarily, included construction, or by imputation from specific known works. For example, Agrippa would have built road stations along his roads in Gaul, and have constructed substantial military works in N.W. Spain for his large campaign of 19 B.C. 28. This second type of known Agrippan works can be located precisely and described in detail only if material remains are both found and identified as Agrippan. There are, also, constructions which may be Agrippan. Those with material remains may be Agrippan if they are of the Agrippan period and there are indications from other reliable sources. For example, at Mérida, the theatre is Agrippan, there is Claudian epigraphy indicating Agrippan action at the town, and historical indication of Agrippan urban planning 29. Some works there may, therefore, be his in addition to the theatre. Texts, coins and the general history of a place or region can indicate possible Agrippan action, such as his possible foundation of Caesaraugusta (Zaragoza) 30. THE OBJECT OF INVESTIGATION All known or possible Agrippan constructions in the West, as defined in the 'subject of investigation' and in the 'criteria for inclusion of constructions', are to be identified, dated, and explained historically, and particularly as supporting the aims of Octavian and Augustus. Their importance is to be assessed at the time of construction and subsequently. The known works of Agrippa in the West are to be described as fully as possible. The possible Agrippan works in the West are to be investigated like the others there, but their description is to be less comprehensive. There is also to be an assessment of the degree of probability of them being Agrippan works. Agrippan works, other than in the West, whether known, or possible, are to be investigated only to cast further light on those in the West, and particularly, to indicate the specific features which distinguish these. There is, therefore, to be only a general description of works outside the West Strabo, See below Chapters 1 and 6 respectively for the roads and the campaign. See below Chapters 7(a) and (b) for the history and (c) for the Claudian epigraphy. See below Chapter 10, section C, pp

15 - 7 - METHOD OF INVESTIGATION Formal investigation of the remains of the Agrippan works in the West was the natural province of the local archaeologists, or of authorised bodies from elsewhere. Investigation of the material evidence here was, therefore, to be by personal inspection without physical disturbance of the works, and, wherever possible, with advice from local archaeologists or historians. Visits were made to all the sites in France and Spain where there were significant remains of major Agrippan construction or a distinct possibility of Agrippan works. The principal material remains of known Agrippan construction in the West are at the Theatre of Mérida in S.W. Spain. These were inspected in detail and contact was established with local archaeologists. The town, its bridges, and aqueducts with their dams, were examined for possible Agrippan works. The remains of the Agrippan Temple of Valetudo at Glanum, near St. Rémy-de-Provence, were also inspected in detail. At Lyon, the remains of a praetorium, probably of Agrippa, were inspected with the archaeologist in charge of the excavations, together with sites at and near the town. At Nîmes an inspection was made of the sacred fountain area and other parts of the town. The vestiges of the Agrippan works at the Campus Martius at Rome were also examined, together with a possible Agrippan warehouse near the Roman forum. The published literature on the known Agrippan works, which is mainly restricted to examination of those with surviving remains, was examined in detail, consolidated and subjected to critical analysis. This literature included the biographies of Agrippa by Reinhold and Roddaz, referred to above 31, and a large number of papers and other publications on particular Agrippan works. It was also necessary to examine a wider range of information and comment in order to identify the Agrippan works of both war and peace with few or no material remains. This entailed research of the historical, epigraphic and numismatic record (particularly of the provincial coinage, as distinct from the Imperial 32 ) available in libraries in Australia, Rome, London, Madrid and elsewhere. Particular care was taken to investigate all aspects, including those of engineering, which have not generally been accounted for by the archaeologists and historians who have written nearly all the literature on the Agrippan works See above Note 1, p1. Burnett, A., Amandry, M., and Ripolles, P.P., Roman Provincial Coinage, Vol. 1, London, 1992, was particularly valuable.

16 PRESENTATION The study constitutes a virtual biography of Agrippa restricted to his planning and construction in the West, and this indicates a chronological treatment. Yet, Agrippa was in Gaul twice and built at some places during both visits. Also, some major works call for treatment in dedicated chapters and others are conveniently investigated with others which may be from a different period. The presentation is, therefore, in accordance with the works to be investigated while maintaining the chronology as far as is practicable. The first five chapters are reserved for constructions in Gaul from B.C. and 19 B.C., with some chapters referring to both periods. The general chronology is set out in the historical introduction. These chapters are followed by conclusions on constructions in Gaul. Chapters six to ten inclusive are set aside for constructions in Spain in 19/18 B.C., and are followed by a conclusion. The eleventh, and final, chapter contains information and discussion on the works outside the West. This is followed by a general conclusion on Agrippan constructions in the West in which some reference is made to the earlier conclusions on Gaul and Spain, but which is intended to present general findings. All the descriptions of the constructions and their technical details are contained in the chapters. Illustrations of the constructions are of particular importance. These should not be compromised by accommodation to the text, and it should be possible to look at them while reading the text. These objectives can be achieved most fully by having the text in one volume and the illustrations in another. Four illustrations in the text volume showing the regions of Gaul and Hispania are indicated by letters, while all the others illustrating the chapters are numbered.

17 - 9 -

18 - 10 -

19 GAUL INTRODUCTION The word Gaul is used here to describe the Celtic heartland called Gallia by the Romans. The part of Gaul considered here is Gallia Transalpina, or 'Gaul over the Alps', as distinct from Gallia Cisalpina or 'Gaul this side of the Alps' when described in relation to Rome. Fig A 1 shows one interpretation of this region when Caesar began his conquest of Belgica and Celtica, sometimes referred to as Gallia Comata, or long-haired Gaul. The Caesarian settlement did not impose any new Roman administrative divisions. Fig. B 2 shows the division of Gaul in the Empire. The Provincia (with minor changes) was now Gallia Narbonensis, and the old Gallia Comata was split into the three provinces of Aquitania (now extending almost to the Loire), Lugdunensis, and Belgica. The first step towards forming these provinces was taken by Augustus during his visit to Gaul in 27 B.C 3. There is no evidence of Agrippan construction across the Rhine, although he crossed this river 4. Consequently, the area of Agrippan works in Gaul is that bounded by the region of the four Gauls, shown on Fig. B. This encompassed the whole of modern France, Belgium and Luxemburg, and parts of The Netherlands, Germany and Switzerland. There is no clear indication about the dating of Agrippa's arrival in Gaul following the Peace of Brundisium, which was probably in September 40 B.C. 5, at which Octavian discovered the treachery of his governor in Gaul, Salvidienus Rufus 6. Agrippa had been appointed praetor urbanus at Rome 7 in the first quarter of 40 B.C. 8 and would have been eligible to take over a province in Based on Drinkwater, J.F., Roman Gaul. The Three Provinces 58B.C.-A.D.260, London, 1983, Map 1. Based on Drinkwater, Map 2, with excision of Germania and some other features. Cassius Dio, 53.22, also below Note 134, p.34. Cassius Dio, Roddaz, Marcus Agrippa, p.67 and Note 207. Reinhold, Marcus Agrippa, p.24, Note 26, also Note 8, p.2. Cassius Dio, Reinhold, p.21, Note 3.

20 B.C. 9. Yet, Octavian needed a new governor for Gaul before that time. He would, therefore, probably have sent Agrippa to Gaul shortly after the Peace of Brundisium. Roddaz suggested that Agrippa left Rome in October 40 B.C. 10, and this dating is accepted here. The date of Agrippa's departure from Gaul is also unknown, but it was probably in the early part of 37 B.C. 11. Agrippa was, therefore, in Gaul for more than two years. Agrippa was sent to Gaul to restore order there in B.C. After he had done this, he travelled to Spain to put down a rebellion there 12. He probably arrived in Gaul in late 20 B.C. 13, and was almost certainly in Spain before the summer of 19 B.C. 14. In this case, he was in Gaul for about six months. Kleiner has stated that Agrippa was in Gaul in 27 B.C., but without providing any evidence for this 15, as pointed out by Roddaz, who noted the absence of any text or document 16. This hypothesis cannot, therefore, be accepted. Also, the circumstantial evidence indicates otherwise. Augustus was in the West in 27 B.C. 17, and Agrippa dedicated the precinct of the Saepta in the Campus Martius at Rome in that year 18. It is unlikely that Agrippa undertook a journey to Gaul before Augustus went there, but, even if he did so, there is no indication of any Agrippan action, or of constructions resulting from this. Agrippa was, therefore, in Gaul twice, with an interval of some 18 years between the two visits. When he was first there, Octavian had only recently acquired the region. Agrippa's main task was to consolidate Octavian's position there by putting down disturbances and setting up a new system of control. Little is known of his military Ibid, p.25, Note 27 and Roddaz, Marcus Agrippa, p.67. Roddaz, Marcus Agrippa, pp based on Appian 5.65, 274-6; 5.73, 318 and Cassius Dio, Cassius Dio interpreted by Reinhold, pp.26, 27, and Roddaz, Marcus Agrippa, p.68 Cassius Dio, Reinhold, p.88 and Note 70 endorsed by Roddaz, Marcus Agrippa, p.383 and Note 3, and p.636. Roddaz, Marcus Agrippa, pp.405 and 638. Agrippa probably got to N.W. Hispania before the summer of 19 B.C. to allow time for his extensive military campaign before the winter. Kleiner, F.S., 'Gallia Graeca, Gallia Romana and the introduction of classical sculpture in Gaul', AJA, 77, 1973, pp , p.384. Roddaz, Marcus Agrippa, p.383, Note 1. Cassius Dio, Ibid,

21 activities, except that he was victorious over the Aquitani in late 38 B.C. 19 and that he was recalled to Italy by Octavian just as he had crossed the Rhine into German territory 20. It is argued here that the Agrippan road network radiating from Lugdunum to the Atlantic, the English Channel and the Rhine described by Strabo 21 was largely completed in its first form by Agrippa between late 40 B.C. and early 37 B.C. The small Temple of Valetudo at Glanum, near St. Rémy-en-Provence, was provided by Agrippa during this period 22 and there are some indications of Agrippan construction at Arelate (Arles) and Forum Iulii 23. When Agrippa was in Gaul for the second time, in B.C., Augustus had already set in hand the process of reorganisation in the old Gallia Comata and Agrippa's task was to put down disturbances and advance the Augustan initiatives in the short period of some six months when he was in the region. There is evidence of substantial Agrippan construction only at Lugdunum 24. The works in Gaul, which can be attributed to Agrippa with a greater or lesser degree of certainty, cannot always be dated readily. In some cases, there are substantial material remains, but, elsewhere, a complete lack of these, and there is no relationship between the extent of survival and the importance of the construction. At some locations, there are Agrippan works from both his visits. The treatment adopted is designed to accommodate all these factors while still providing the best possible description of the remains and explaining and evaluating all of them. Chapter 1 is devoted to the Agrippan road network because this is regarded as his most important. The system radiated from Lugdunum and Chapter 2, describing and explaining all his works there, follows naturally after it. Chapter 3 is set aside for the Temple of Valetudo at Glanum, since this has the most substantial material remains of any Agrippan work in Gaul and is of great architectural interest, although it is a small building. The possible Agrippan works at Nemausus are considered in Chapter 4. All the other Agrippan works in Gaul, whether known to be his or possibly so, are investigated in the last and fifth chapter. This is followed by a conclusion on Agrippan works in Gaul Appian, Cassius Dio, Strabo, See below Chapter 3, pp See below Chapter 5, Section (b), PP See below Chapter 2, Section (b), the praetorium, Note 58, p.50.

22 AGRIPPA'S ROAD SYSTEM INTRODUCTION A passage from Strabo referred specifically to Agrippan roads in Gaul. This described a road network by the routes which it followed: Lugdunum is in the centre of the country an acropolis, as it were, not only because the rivers meet there, but also because it is near all parts of the country. And it was on this account, also, that Agrippa began at Lugdunum when he cut his roads that which passes through the Cemmenus Mountains as far as the Santoni and Aquitania, and that which leads to the Rhenus, and, a third, that which leads to the ocean (the one that runs by the Bellovaci and the Ambiani); and, a fourth, that which leads to Narbonitis and the Massilian seaboard. And there is also, again, in the Poeninus itself (if you leave on the left Lugdunum and the country that lies above it), a bye-road which, after you cross the Rhodanus or Lake Lemenna, leads into the plains of the Helvetii; and thence there is a pass through the Jura Mountains over to the country of the Sequani and also to that of the Lingones; moreover, the thoroughfares through these countries branch off both ways both towards the Rhenus and towards the ocean 1. The passage from Strabo is the point of departure for all modern research into the Agrippan road system 2, since there are no other known texts, inscriptions, or coins which mention it. Furthermore, it is not possible to distinguish Agrippan road material from later construction, since his work has been repaired, upgraded, abandoned and rebuilt over many centuries. There are no dates in Strabo's text, but Agrippa was in Gaul twice. Consequently, Strabo did not say when Agrippa planned his system or state the extent of his construction in B.C., or in 19 B.C. Uncertainties in Strabo s text have resulted in different views about how it should be regarded. Goudineau saw the passage as little more than a list of things which could be done to provide a road network, and suggested that the actual roads, completed 1 2 Strabo, Geography , as translated by H.L. Jones, in the Loeb Classical Library, The Geography of Strabo 2, Harvard, 1969, pp Roddaz, Marcus Agrippa, p.390.

23 long after the Agrippan period did not necessarily correspond to the description 3. Roddaz thought that the text described a road system planned by Agrippa in B.C., but built over a long period 4. Drinkwater was more specific, since he considered that the system described by Strabo was envisaged from Octavian's acquisition of Gaul in 40 B.C., and was completed to a high standard before 27 B.C. 5. Some writers, including Grenier 6 and Chevallier 7, have seen Strabo's text as describing first action by Agrippa in 19 B.C., although Chevallier 8 has now cast some doubt on this dating. Here, it is argued that the text of Strabo is essentially reliable and that it described a real road network planned by Agrippa and substantially completed by him in B.C. by improving the existing roads. This first system was upgraded over many decades and any Agrippan action in 19 B.C. was cosmetic. It might be argued that Strabo's text is unreliable because of the generally defective understanding of geography, and of cartography in the ancient world 9. Strabo s description of Gaul in his Geography left much to be desired, since all the coastlines faced north, the Pyrenees ran N-S, and all the distances were incorrect 10. Woolf suggested that the choice of Lugdunum as the centre of the Agrippan road system, and the general plan of the system itself, resulted partly from the inadequate Roman knowledge of Gallic geography 11. This comment was probably based on Strabo s statement that 'Lugdunum is in the centre of the country, when it is really in S.E. Gallia Comata. Geographical errors may have led to over-confidence in Roman Goudineau, C. & Rebourg, A., Eds., Les villes augustéennes de Gaule, (Actes du colloque international d'autun, 6-8 Juin, 1985), Autun 1991, Introduction, pp.11, 12. Roddaz, Marcus Agrippa, p.73 and pp and Note 29, p.390. Drinkwater, J.F., Roman Gaul, London, 1983, pp.121 and 125. Grenier, A., Manuel d'archéologie gallo-romaine 2, L'archéologie du sol, les routes, Paris, 1934, pp Chevallier, R., Roman Roads, London, 1976, p.160. This reference is not used further in the thesis. Chevallier, R., Les voies romaines, Paris, 1997, pp.210 and 211. Dilke, O.A.W., Greek and Roman Maps, Baltimore, 1998, pp for an account of geographical concepts and their representation, and pp.43, 44, 47 and pp for Strabo. As pointed out by Goudineau, CAH 10, pp.466, 7. Woolf, G. Becoming Roman, Cambridge, 2000, p.89 and Note 23.

24 achievements 12, but Strabo's text gave practical information on places reached or traversed without reference to points of the compass or distances. Furthermore, the choice of Lugdunum as the road hub was based on strategic factors 13. Also, it is not possible to doubt Strabo's text because there was insufficient information which he could use. There was the map of the world on the Porticus Vipsania in Rome near the Campus Martius, erected by Agrippa s sister after his death, and still incomplete in 6 B.C. 14. The portico and the map are lost, but Pliny mentioned both, and, also, said that the map was derived from the lost Commentaries of Agrippa 15. Roddaz concluded that these were taken into account in making the map 16. Agrippa s Commentaries were probably placed in the Archives at Rome after Agrippa died in 12 B.C. 17. Strabo, writing his Geography between 9 and 5 B.C., and revising it in A.D , could have used them 19. It is, also, possible that the unknown 'chorographer', or geographer, of Strabo was Agrippa 20. The lost Autobiography of Agrippa contained a passage on Agrippa's Portus Julius in the Bay of Naples 21, and, therefore, probably described the Agrippan roads in Gaul. The evidence, therefore, indicates that Strabo's account of Agrippa's road network is reliable, even if Strabo, the partly 'armchair geographer' 22 had no direct knowledge of it. Furthermore, an Agrippan road network can be expected since he was in Gaul twice and could have built this first essential instrument of control for Octavian or Augustus 23. If Strabo had not mentioned the Agrippan road system it would probably have been imputed Moynihan, R., Geographical Mythology and Roman Imperial Ideology, Winkes, R., Ed., The Age of Augustus, Louvain-la-Neuve, 1985, pp , at p.152, suggested that Augustus believed that he had nearly conquered the world by his penetration into Germany. Drinkwater, J.F., 'Lugdunum : Natural Capital of Gaul?', Britannia, 6, 1975, pp , also see below, p.40. Cassius Dio, , and Roddaz, Marcus Agrippa, pp and Notes Pliny, Roddaz, Marcus Agrippa, pp and Notes Ibid, p.389 and Note 27. Dilke, p.62. Roddaz, Marcus Agrippa, pp and Notes Dilke, pp.43, 4 for the 'lost chorographer'. Roddaz, Marcus Agrippa, pp and Notes 6-19 referring to Pliny, 7.148, and , also below Chapter 11, p.211, for the port. Dilke, p.62. See above pp.11, 12, for Agrippa's two visits to Gaul.

25 EXTENT OF AGRIPPA'S ROAD NETWORK Strabo mentioned four Agrippan routes radiating from Lugdunum, which can be identified by reference to Fig The first of these, to the West, reached the Santoni, marked on Fig. 1.1 as the Santones. The second route to the Rhine was poorly described, since it could have reached this river at any point between the territory of the Raurici and the estuary on the North Sea. Strabo's route to Ocean went through the territories of the Bellovaci and the Ambiani marked on Fig. 1.1 to reach the English Channel. The fourth route must have gone down the valley of the Rhône. Strabo also referred to a 'bye road', described here as a by-road, after noting the four Agrippan routes, although he did not say that this was part of Agrippa's road network. Strabo said that the by-road crossed the Rhodanus (the Rhône), or Lake Lemenna (Lake Geneva) into the plains of the Helvetii, going through a pass in the Jura Mountains into the territories of the Sequani and the Lingones. As can be seen from Fig. 1.1 and Fig , the Rhône passes through Lake Geneva. Consequently, the road crossing the river or the lake could have come from Lugdunum or from over the Alps via the Great St. Bernard Pass. Strabo also said, And there is also, again in the Poeninus itself (if you leave on the left Lugdunum and the country that lies above it) a bye-road. Strabo s mention of Lugdunum and the Poeninus itself (the Alps) in the same sentence does not properly define the road location. This was recognised in the Loeb edition by a note on the translation. Strabo s brevity is again confusing. He suddenly shifts his standpoint from Lugdunum to the Poeninus (the Alps). He has in mind two roads: (1) the road which ran through the Poeninus to the Rhodanus (the Rhône) at the eastern end of Lake Geneva, crossed the river, circled around the lake, crossed again at the other end and then followed the Rhodanus to Lugdunum; and (2) a road that branched off the same at some point north of the lake, into the plains of the Helvetii' 26. Jullian 27 thought that the passage referred to two roads from Lugdunum to Italy, one over each of the two St. Bernard Passes shown on Fig Based on Brogan, O., Roman Gaul, London, 1953, Fig. 1, p.19. Based on Chevallier, Les voies romaines, Fig. 146, p.234, and Grenier, 2, Figs. 10 and 11, pp.40, 41. Strabo, Geography, translated by H.L. Jones, The Geography of Strabo, 2, Harvard, 1969, Note 2, pp Jullian, C., Histoire de la Gaule, 4, Paris, 1929, p.84, Note 7.

26 Grenier discerned only one road from Lugdunum to Italy via Lake Geneva and the Great St. Bernard Pass 28, and Roddaz followed this 29. None of these interpretations explain Strabo's passage 'If you leave on the left Lugdunum and the country that lies above it'. This phrase seems to indicate that the traveller from Italy has already passed the fork in the road leading to the Little St. Bernard Pass after departing from Augusta Praetoria (Aosta), and is about to cross the Great St. Bernard Pass towards Octodurus. In this case, there was one by-road from over the Alps and it crossed the Rhône to the east of Lake Geneva. It is most unlikely that Agrippa would have built such a road. The territory of the Helvetii between Augusta Raurica and Lake Geneva was strategically important for Roman security, since it lay between the Alpine tribes (including the Helvetii) and Southern Gaul. It also gave access to the Great St. Bernard Pass and Italy. After the frustration of the Helvetii in their plan to migrate to Western Gaul and their defeat by Caesar, they were obliged to retain their lands 30 and Colonia Julia Equestris (Nyon) was established by Caesar 31. Colonia Raurica (modern Augst) was founded by L. Munatius Plancus with Lugdunum in 43 B.C. 32. The Roman position at Nyon guarded passage over the Swiss Plateau to Lugdunum and the Alps. That at Augst guarded the Rhine crossing, and, also, blocked the natural route between there and Vesontio via the Belfort Gap, or the pass between the Jura Mountains and the Vosges to the north 33. Consequently, the Swiss Plateau was guarded before Agrippa first came to Gaul, although there may have been no continuous road between Nyon and Augst 34. Agrippa, therefore, was more likely to provide for security by new road construction reaching a position on the Rhîne below Augst. There is also no reason to think that Agrippa used a road through the Great St. Bernard Pass for his supplies or troop movements. In B.C., he recruited most of his forces for action in Gaul Grenier, 2, pp Roddaz, Marcus Agrippa, p.390 and Note 28. Caesar, BG, Wells, C.M., The German Policy of Augustus, Oxford, 1972, pp.35-36, Notes 1 and 2, indicating the scholarship on the foundation, for which there seems to be no primary text. As indicated by CIL, , on the tomb of Plancus. The Belfort Gap is marked on Fig As argued by Wells, p.37. Note 63.

27 in that territory 35, and it is most unlikely that forces would be diverted from Italy. Furthermore, the two St. Bernard Passes could not be used without let or hindrance until Augustus freed them from local control in B.C. Even in 19 B.C. it is not certain that the Great St. Bernard Pass could accommodate wagons 36. Both Jullian 37 and Drinkwater 38, considered that Agrippa planned and built nothing east of the Jura Mountains. Grenier stated that the by-road was Agrippan 39 and Roddaz followed him 40, but neither gave any justification for their positions. Chevallier was silent on the matter. Attribution of the by-road to Agrippa was perhaps argued because Strabo mentioned it immediately after the four Agrippan roads, yet this is probably explained by Strabo's account of the clearing of the Alpine roads by Augustus earlier in his book Having described Agrippa's roads, he was reminding the reader of the achievement of Augustus and noting that, as a result of both these actions, it was possible to travel directly from Italy to Gaul, and from there to Ocean and the Rhine. Agrippa did not say that the 'thoroughfares were the Agrippan roads, but some writers have assumed this 42. INVESTIGATION OF THE AGRIPPAN SYSTEM OF FOUR ROADS Investigation is confined to the roads themselves and limited to identifying their locations without reference to bridges or other works which were part of them. Road stations and other buildings for road use and security built by Agrippa are considered in Chapter 5. Roads at and near Lugdunum are examined in Chapter 2. As noted above, Strabo's text mentioned no dates. Agrippa could have devised his system in or 19 B.C. and have started work on any particular road at different times. The Roddaz, Marcus Agrippa, p.70 and Note 217. Strabo, and Hyde, W.W., Roman Alpine Routes, Philadelphia, 1936 (facsimile Ann Arbor, 1983), Wells, p.37 and Chevallier, p.273. Jullian, 4, p.84, Note 7. Drinkwater, Roman Gaul, p.125. Grenièr, 2, p.39. Roddaz, Marcus Agrippa, p.390, Note 28. Strabo, 4.6 and 7. See below Notes , p.30.

28 reasons for his actions in the two periods were also different for particular roads. It is, therefore, necessary to investigate each road separately, and, also, to consider the whole network. A first section is set aside for examination of each of the four roads in which the location is considered in relation to historical and strategic factors. A second section is reserved for investigation of the factors influencing planning and construction of the whole network. It is then possible to draw conclusions from the findings of the two sections. 1. INVESTIGATION OF EACH OF THE FOUR ROADS The location of a first Agrippan road must be deduced rather than stated because of the lack of material evidence of its existence. If a later Roman road followed its trace, there would be changes in alignment over the centuries. If the first trace was abandoned for another, these changes could be substantial. Consequently, the location assessed here is approximate. It can be shown only on a small scale map, and indicated by the places traversed or reached by it. Drinkwater marked the whole network on a small scale map 43 and Chevallier described the locations in his text, illustrating only that to the South 44. Here, a road location diagram is provided for each of the roads. The locations of Agrippa's first roads are deduced from the written and the material evidence. The Antonine Itinerary consists of a text describing journeys of the Antonine period and naming stopping places and the distances between them 45. There is no information on when the road was built. The Peutinger Table is a painted map of the 12 th or 13 th century based, by several removes, on a road map of the 4 th century. This contains information from the early and late Empire, but it cannot be used to date any of the roads in Gaul 46. These written sources indicate the existence of a Roman road later than the Agrippan period, but they do not describe all the Roman roads, and can, therefore, be used only suggest the possible existence of some Drinkwater, Roman Gaul, Map 7, p.238. Chevallier, pp.210, 211 and Fig Dilke, pp.125-8, Chevallier, pp Miller, K., Die Peutingersche Tafel, Stuttgart, 1962, Levi, A. & M., Itineraria Picta, Rome, 1967, Dilke, p.114 and Chevallier, pp

29 roads of Agrippa 47. The material evidence is from the remains of the Roman roads. Sometimes these can be readily identified from site inspection, but the line of the road can be discerned most easily from the air, by noting changes in relief or vegetation, the land division and the locations of bridges and buildings of various periods. The general chronology of roads in a particular zone can be built up by combining local historical information with the physical evidence 48. The written and material evidence of the Roman roads is not always sufficient to deduce the location of the first Agrippan roads, and, particularly, where there are alternative routes. In these cases, it is necessary to consider reasons for Agrippa to adopt one route rather than another. (a) THE ROAD FROM LUGDUNUM TO THE WEST: Strabo said that the road passed through the Cemmenus Mountains 'as far as the Santoni and Aquitania. The Cemmenus Mountains, or the Cévennes, lie to the west, and mainly south, of Lyon. The chief city of the Santoni was at Mediolanum (modern Saintes). The general route can be identified from Fig. 1.1 as going west from Lugdunum over the tributaries of the Loire, skirting the higher part of the Averni (the Auvergne Plateau), and reaching the territory marked 'Santones'. Fig shows the principal known Roman roads in the area. The shortest route was directly from Feurs to Clermont-Ferrand, but the Peutinger Table shows the Roman road going through Vichy, and the Antonine Itinerary indicates no westerly road from Lugdunum. The direct road between Feurs and Clermont-Ferrand, rose to some 1,500 metres, whereas that through Vichy reached only about 500 metres, but the road gradients on the direct route were not a serious problem for military forces with pack animals. The direct road was discerned some 80 years ago by Besnier, and then confirmed by Grenier 50. The Peutinger Table would not have noted it if the road was re-routed through Vichy later for easier passage of vehicles. The short route has been identified as that of Agrippa by a recent study in which the historical and Chevallie r, Fig. 121, p.200 and Fig. 122, p.201, provided maps showing, respectively, the roads according to the Peutinger Table and the Antonine Itinerary. These are used here. Chevallier, Chapters 5-7, pp , and, particularly, Chapter 7, 'Les voies et les données gétopographiques', pp Based on Drinkwater, Roman Gaul, Map 8, p.289 and Chevallier, Fig. 121, p.200. Besnier, M, Le point de départ des grandes routes de la Gaule Romaine, BAC, 1923, pp.73-96, pp, 88, 89, and Grenier, 2, pp.35, 36.

30 physical characteristics of the transport zone were taken into account 51. The Agrippan road went from Clermont-Ferrand to Limoges, where there is evidence of it 52. It has been identified at Saintes for the first 600 metres after it crossed the River Charente 53. There is, therefore, no dispute about the location of the first Agrippan road to the West from Lugdunum to Mediolanum. The Santoni were allies of Rome, as noted by Caesar 54, and as can be seen from the Romanisation of the aristocratic Gaulish family of a man called Epotsorovidus. A son of this family received Roman citizenship, taking Caesar s nomen Julius, and his grandson Caius Julius Rufus, a priest of the Cult of Rome and Augustus at Lugdunum, paid for the amphitheatre there. He also built the surviving triumphal arch at Mediolanum in A.D. 18/19 55 over the existing Agrippan road from Limoges. Archaeological investigations indicate the existence of an agglomeration with a pottery in production by about 40 B.C. The town grid was set out to suit the monumentalisation of the early 1 st century A.D. and the topography of the town plateau. It had no relation to the line of the Agrippan road which was governed by the crossing of the Charente River 56. There was, therefore, no planning connection between the Agrippan road and any monument, except the arch of A.D. 18/19 which was built over it. A road to Mediolanum reached the Charente River with access by this to the Atlantic Ocean, and also brought Roman forces closer to the valley of the Loire to the north. Agrippa's road to the English Channel through the territories of the Bellovaci and the Ambiani and the road to the west provided the bare bones of a strategic road system for controlling N.W. Gaul, from which Agrippa had little to fear Denimal, P., 'La voie Aquitanique d'agrippa de lyon à Saintes'. Thèse de l'universitè de Paris, 4, Chevallier, p.210, Note 47. Maurin, L., Les villes augustéennes de l'aquitaine occidentale : Bordeaux, Périgueux, Saintes, Goudineau, C., & Rebourg, A., Eds., Les Villes augustéennes de Gaule, pp.45-59, Fig. 1, p.48 and p.51. Caesar, BG, 3.11; King, A., Roman Gaul, London, 1990, pp.66,67 and Note 3, p.218 referring to CIL , also Woolf, Becoming Roman, p.40. Maurin, pp.46-48, 54-56, for the chronology of development and Woolf, p.118, for the town grid.

31 Strabo's text stated that the road went as far as the 'Santoni and Aquitania'. The old Aquitania of Caesar was almost entirely south of the Garonne, shown on Fig. 1.1, whereas the province of Aquitania, created by Augustus in 27 B.C., extended almost as far north as the Loire 57. Strabo mentioned both 58. Since Mediolanum was in the province of Aquitania, a road reaching the town automatically reached this Aquitania and it is curious that Strabo mentioned the province. He may have been referring to the old Aquitania, which had long been a trouble spot before the conquest of Caesar. The Aquitani were overcome by Crassus acting for Caesar only after a lengthy struggle following a Roman approach from the south via Toulouse, Carcasonne and Narbonne 59. The Cantabri of northern Spain helped the Aquitani 60. Caesar himself went to Aquitania in 51/50 B.C. and campaigned there, noting that Crassus had conquered it to a certain extent 61. Agrippa campaigned in Aquitania in 38 B.C. and seems to have won a brilliant victory, but there is no other information from the ancient texts 62. It is, therefore, not known whether he approached Aquitania from the south, like Crassus, or from the north through Saintes, or through some other region. The last major campaign in Aquitania seems to have been that of Messala Corvinus in 29/28 B.C. 63. It is not certain that the region was entirely quiet after this, but it is unlikely that there was any significant assistance from the Cantabri of northern Spain, after Agrippa s pacification of Cantabria in 19 B.C. 64. Aquitania, therefore, remained a trouble spot for the Romans at least until 28 B.C., and, since Agrippa campaigned there himself he would surely regard it as a strategically important region. Goudineau noted the existence of problems in S.W. Gaul and N.E. Gaul and suggested that Agrippa's roads to the West and to N.E. Gaul built to counter these problems, probably originated from Agrippa s first visit to Gaul, referring specifically to the old Aquitania and to Strabo's description of the Agrippan roads 65. By extending the road beyond Mediolanum to the old Aquitania, Agrippa could have kept his forces in The two Aquitanias are shown on Figs. A and B, pp. 9, 10. Strabo, 4.1. Caesar, BG, Ibid, Ibid, 4.6. Cassius Dio, , Appian, Roddaz, Marcus Agrippa, p.70 and Note 219. Cassius Dio, and below Chapter 6. Goudineau, CAH 10, 1966, pp.489, 490.

32 Gallia Comata for rapid response to unrest in N.W. Gaul and on the Rhine, and have reduced his forces in southern Gaul. Extension of the road on flatter ground beyond Saintes for an extra distance of some 170 km was a small task in relation to building the road the 600 km from Lyon to Saintes over the Auvergne Plateau. There is, therefore, a strong indication that Agrippa built his road in B.C. extended beyond Mediolanum along the east side of the Gironde to the Garonne near present Bordeaux. Maurin argued that there was a pre-roman road from Saintes to a long-established trading place at Bordeaux, and that Agrippa, improving this, transferred the Biturges and the Vivisci to the area (indicated on Fig. 1.1) in B.C. In fact, Maurin considered that the road to the Bordeaux area passed through Saintes only because the Charente could be crossed there conveniently 66. Yet the Garonne could be reached through Limoges and Perigueux. It is also unlikely that Agrippa built a road to transfer people. He more probably made them build the road themselves to serve his strategic advantage. Apart from the indications of Goudineau and Maurin, the received opinion is that the Agrippan road to the West ended at Saintes. Yet the evidence indicates that it extended beyond Saintes to the Garonne. (b) THE ROAD TO THE RHINE: Strabo neither named the place where the road reached the Rhine, nor the regions through which is passed. Fig shows the course of the Rhine from Augst to its estuary, and the principal physical features which would have influenced Agrippa's road location. It was argued above that Agrippa's road passed west of the Jura Mountains because the invasion routes over the Swiss Plateau and the Belfort Gap had been guarded by the colonies at Augst at Nyon, established before he came to Gaul. A road could be built easily up the valley of the Saône, marked on Fig. 1.1, and most of that of the Moselle. It was not practicable to use the Moselle valley north of Trier, since the river was convoluted and confined by steep cliffs. A road to Mainz, Mogontiacum, left the Rhine below this point unguarded. A road to Cologne or to Bonn over the Eifel reached the plains north of the Ardennes, the northern margin of which formed a natural barrier against movement to the south Maurin, p.47 and Notes Based on Wightman, E.M., Military arrangements, native settlements and related developments in early Roman Gaul, Helinium, 17, 1977, pp , Fig. 1, p.109 (part).

33 from across the Rhine over the plains, or from the plains themselves. Agrippa's road to Ocean passed through the territories of the Bellovaci and the Ambiani to reach the English Channel, as indicated on Fig An Agrippan road to the Cologne area and a road to the Channel could be connected by a road along the northern margin of the Ardennes to form a triangle of road control. There is evidence of such a W-E road being in partial operation by about 20 B.C 68. An Agrippan road to the Cologne area would, therefore, have been the most effective as a counter to invasion from across the Rhine north of Augst. The threat of invasion from across the Rhine, and of insurrection in N.E. Gaul, was a primary concern for Agrippa in B.C. and he would, therefore, have built his road to the Rhine at that time. A road reaching the general area of Cologne ended in the territory of the Ubii, who had lived on the east, or right-hand side of the Rhine before the conquest of Gaul, and had co-operated with Caesar in his crossing of the River 69. The Ubii moved over the river and occupied the area near and to the south of Cologne, including the territory of the Eburones, shown on Fig. 1.1, laid waste by Caesar after his defeat of their leader Ambiorix in 53 B.C. 70. Strabo said that the Ubii were transferred across the Rhine to the left bank by Agrippa at their own request 71, but he gave no dates. Texts of Tacitus indicate that the transfer of the Ubii occurred over a long period 72. Agrippa crossed the Rhine in the later part of his visit to Gaul in B.C. 73, and probably in winter/spring 38/37 B.C. 74. Oak piles driven for a wharf at Cologne have been dated to 38 B.C. 75, and the dating is reliable 76. Agrippa was probably responsible for building the wharf 77. There is no record of Agrippa crossing the Rhine in 20/19 B.C Wightman, E.M., Gallia Belgica, London, 1985, p.50, noted the road as a supply route, but it was also a defensive road because of its location. Caesar, BG Ibid, 6.43, 44. Strabo, Tacitus, Annales and Germania 28. Cassius Dio, Reinhold, pp.26, 27 and Note 34 and Roddaz, Marcus Agrippa, p.71. Doppelfeld, O., Das römische Köln. I. Ubier-Oppidum und Colonia Agrippinensium, ANRW 2.4, 1975, pp , p.718. Renfrew, C.; Bahn, P., Archaeology, London, 1991, pp , for the technology of dating timber by dendrochronology. As assumed by Doppelfeld, p.719.

34 The evidence, therefore, indicates that Agrippa was active in the Cologne area in 38 B.C., and that the transfer of the Ubii occurred at this time, although probably as a formalisation of an earlier process. Dating of the transfer to 38 B.C. is widely supported 78. The relationship between Rome and the Ubii was important from the conquest onwards and the Ara Ubiorum at Cologne, in existence by A.D. 9 79, was the cult centre for the new province of Germania. Colonia Ara Agrippinensis founded at Cologne in A.D. 50 was a major Roman centre. Gechter provided an analysis of the special relationship between Rome and the Ubii 80. The location of the Oppidum Ubiorum at Cologne does not automatically indicate that the first Agrippan road ended there, since it is not possible to trace the development of Cologne from a transfer of the Ubii in 38 B.C. to the foundation of the colonia 90 years later 81, except for the altar being in use by A.D. 9, as noted above. The earliest Roman occupation of Cologne has been dated to 5 B.C. from the remains of a Roman camp, and no pre-roman pottery has been found. At Bonn (some 25 kms up the Rhine from Cologne), the pottery indicates a lowland oppidum from 30 B.C. Cologne, on the other hand, was in an open area unsuitable for an oppidum 82. Gechter suggested that the nobility of the Ubii had been destroyed before the conquest 83. It is possible, therefore, that the Ubii initially settled in various locations, with Bonn amongst the earliest of these, and that the Oppidum Ubiorum at Cologne was created later by the Romans as the Ubian cultural centre. Certainly, the Oppidum Ubiorum was laid out to a grid pattern by Roman surveyors 84. Bonn was better sited than Cologne to control movement from across the Rhine into Gaul immediately north of the Ardennes, since it is closer to these uplands. There is, therefore, an archaeological and a strategic case for Agrippa ending his road at Bonn rather than at Cologne, whether Agrippa started to build his road in B.C. or in 19 B.C. The 38 B.C. dating of the wharf timbers at Roddaz, Marcus Agrippa, pp.71, 72 and Notes and pp and Notes 6-16 provided a summary of the literature. La Baume, P., Colonia Agrippinensis, Chicago, 1969, pp Gechter, M., Early military installations and Ubian settlements in the Lower Rhine. Blagg, T. & Millet, M., Eds., The Early Roman Empire in the West, Oxford, 1990, pp , at p.102. King, p.157; Wilson, J.R.A. Gechter, p.100. Ibid, p.99. Doppelfield, p.718.

35 Cologne is explained by Agrippa's crossing of the Rhine 85. A W-E road reaching Cologne by about 20 B.C., as suggested by Wightman 86 is not pertinent to discussion on an Agrippan road of B.C. A termination of the road at Bonn was suggested by Jullian some 80 years ago based on his interpretation of the remains of roads ending at both Bonn and Cologne 87. Since that time there has been general agreement that the road ended at Cologne, with the notable exception of Chevallier, who indicated that the road did not go to the Cologne district, but to Mainz 88. As noted, this termination would have been ineffective as a strategic road. The general belief that Cologne was the destination of the road to the Rhine is probably to be explained by the presence of the colony and the Ara Ubiorum, together with the road going there being marked on the Peutinger Table and the Antonine Itinerary. Yet all these indicators are from the Empire and not from the Agrippan period, and the evidence from that time points strongly to the road finishing at Bonn. Fig shows a Roman road through Mayen to Andernach, which reached Bonn by going down the Rhine and, also, a branch off the road over the Eifel through Zulpich to Bonn. Jullian favoured the road through Zulpich as the Agrippan road 90. Wightman 91 and Drinkwater 92 also preferred the road over the Eifel, but going to Cologne. Grenier 93 adopted the Andernach road and was followed by Roddaz 94. All these opinions were based on different interpretations of inadequate information, and there is no detailed study of the Agrippan road in the region corresponding to that of his road to the West 95. A road over the Eifel would have had military advantage since it commanded high ground and could be used to Cassius Dio, See above Note 68, p.25. Jullian, S., p.88 and Note 5. Chevallier, p.210. Based on Wightman, Gallia Belgica, general map (part). Jullian, 5, p.88 and Note 5. Wightman, 'Military arrangements ', Fig. 1, p.109. Drinkwater, Roman Gaul, Map 7, p.238. Grenier, 2, p.37. Roddaz, Marcus Agrippa, p.392 and Note 38, referring to Grenier. The Thesis of Denimal above, Note 51, p.22.

36 deploy forces to the west or the east. This route must, therefore, be preferred to that through Mayen. Wightman has indicated a re-routing of the first Agrippan road at and near Trier resulting from Agrippa's visit to Gaul in 19 B.C., based on the dating of the first Roman bridge at Trier to B.C. from dendrochronological investigation. Wightman argued that Agrippa's first road approached Trier from the south on the west of the Moselle in order to avoid the ford there, which had steep cliffs on its north side. When the bridge was built and the cliffs were cut back, the road was deviated to cross the river 96. Her hypothesis is to be accepted. In this case the Agrippan road of B.C. was probably re-routed elsewhere in the general region of Trier. Although there is still controversy and some uncertainty about the route of the first Agrippan road north of Trier, its general location between Lyon and Trier has been established. The Agrippan road, shown on Fig , went through Chalon, Dijon, Langres, Toul and Metz, following the valleys of the Saône and the Moselle, except at Dijon, where it was west of the Saône, and, also, where it went over the Langres Plateau. Identification of the route has been based on the evidence of the Peutinger Table over the whole distance, the Antonine Itinerary from Langres to Trier, the remains of Roman roads and the natural routing indicated by the topography 98. As noted above, construction of the entire road from Lyon to Bonn would have been a primary concern for Agrippa in B.C. (c) THE ROAD TO THE OCEAN (THE ENGLISH CHANNEL): Fig. 1.6 shows the principal Roman roads between Lugdunum and the English Channel. Strabo's text described the road to Ocean as passing through the territories of the Bellovaci and the Ambiani, shown on Fig. 1.1, whose respective capitals were Beauvais and Amiens. The only road shown on Fig. 1.6 which reached both is that through Chalon, Sens, and Paris. Any road reaching Boulogne from Langres directly would not Wightman, E.M., Roman Trier and the Treveri, London, 1970, p.36 and Gallia Belgica, pp.49, 50 and Note 77, p.342. Based on Wightman, 'Military arrangements ', Fig. 1, p.109, extended to include Lyon. Drinkwater, Roman Gaul, Map 7, p.238; Roddaz, Marcus Agrippa, p.390; Chevallier, p.210 and Notes 58, 59.

37 pass through Beauvais. Fig , indicates that the route through Paris was indirect. It was easier to reach Beauvais from Sens through Meaux and Senlis, and such a road existed 100. Paris was not an important Roman centre even in the Empire 101. Thévenot's detailed study published in , but still authoritative 103, identified a pre-roman road connecting the Aedui of central Gaul to their allies, the Senones and the Bellovaci, and to the English Channel, which passed through Chalon, Sens, Meaux, Senlis, Beauvais and Amiens. Thévenot indicated that the first Agrippan road followed the general line of this road, perhaps upgraded after the conquest 104. Such improvement is to be expected, since trade increased between S.E. Britain and Gaul over the short channel crossing at Boulogne after Caesar visited Britain 105. The Aedui were the first, and the principal, allies of Rome in Gallia Comata 106. It would, therefore have been natural for Agrippa to adopt the road to Ocean controlled by the Aedui, when he came to Gaul in B.C. Yet, Thévenot dated the first Agrippan road to 19 B.C., relying partly on Grenier who argued that all the Agrippan roads should be dated from 19 B.C. onwards 107, and, partly on the archaeology of Bibracte, the old capital of the Aedui marked as Mt Beuvray on Fig. 1.6, and of Autun, the new Roman centre to its East. Thévenot thought that Bibracte was abandoned in the period B.C. and that Autun was founded at that time with the Agrippan road passing though it 108. It is now known that Bibracte and Autun continued in use together. There is uncertainty about the foundation date of Autun 109, and, consequently, no clear indication of the Agrippan road going through Autun in 19 B.C. Also, a road dating there does not Based on Wightman, Gallia Belgica, general map (part). As indicated by Wightman on her general map of Gallia Belgica and also Poux, M., and Robin, S., 'Les origines de Lutèce', Gallia, 57, 2000, pp , Fig. 22, p.218. Woolf, pp.101, 2, and Notes 62, 63. Thévenot, E., 'Les voies romaines de la cité des Eduens', Latomus, 98, Chevallier, pp.210, 1 and Note 55. Thévenot, pp Caesar, BG, 5, 20, 21. There is also archaeological evidence of a decrease in the trade between S.W. Britain and Brittany from about that time; Wiseman, A. & P., Julius Caesar The Battle for Gaul, Boston, 1980, p.88. BG, 1.11; Grenier, 2, pp Thévenot, pp Woolf, p.9 and Notes 28, 29; Rebourg, A., 'Les origines d'autun : l'archéologie et les textes', Les villes augustéennes de Gaule, pp ; Gruel, K. and Vitali, D., 'L'oppidum de Bibracte', Gallia 55, 1998, pp and Rebourg, A., L'urbanisme d'augustudunum (Autun, Saône-et-loire), Gallia 55, 1998, pp

38 indicate dating elsewhere. Chevallier has suggested a dating of B.C. of the whole road from the general archaeology of its formation with typical ditches 110. The route through Chalon, Autun, Sens, Meaux, Senlis, Beauvais and Amiens is preferred by modern writers 111, although Drinkwater adopted one through Langres, Reims, Soissons and Amiens 112. This is probably explained by his reliance on earlier writers who preferred a road through Langres. Jullian adopted the route through Langres, Reims and Soissons which went through Amiens but avoided Beauvais, relying on the prominent remains of the road. He suggested that Strabo had made a mistake in saying that the road went through Beauvais 113. He was supported by Grenier 114. Thévenot convincingly rebutted this hypothesis of a 'mistake' by Strabo and contested Jullian's adoption of a well-defined route of the Empire as a first Agrippan road 115. The hypothesis of an Agrippan road from Langres arose from Strabo's remarks on the by-road and the thoroughfares by which it was possible to reach from this to the Rhine and the English Channel. A by-road through Besançon and Mirebau, shown on Fig. 1.6, joined Agrippa's road to the Rhine south of Langres, and would, therefore, have connected to a road to the Channel from there. Yet the road from Langres through Soissons is probably post-agrippan while that through St. Quentin and Arras is probably pre-agrippan 116. The Agrippan road, therefore, went to Boulogne through Chalon, Sens, Meaux, Senlis, Beauvais and Amiens along an existing commercial road controlled by an ally of Rome. This gave direct access to N.E. Gaul and the short Channel crossing to counter potential unrest in that region and co-operation in this by British tribes. Adoption of the road and any necessary improvement were called for in B.C. in Chevallier, p.211. Roddaz, Marcus Agrippa, p.391 and Notes 35, 36, Wightman, Gallia Belgica, p.50 (through Paris between Sens and Beauvais) and Chevallier, pp.210, 211. Drinkwater, Roman Gaul, Map 7, p.238. Jullian, Grenier 2, p.37. Thévenot, pp See above, p.19, for Strabo's comments on the by-road and the explanation for these, also Chevallier, p.211 and Notes 56, 57 referring principally to Leman, P., 'La voie de Léman à l'ocean', la branche orientale. État de la question et propositions nouvelles', Caesarodunum, 10, 1975, pp for the dating of the two roads from Langres.

39 conjunction with the construction of the road to the Rhine. A later road, connecting the road to Ocean and the Rhine 117, would seal off the northern plains along the northern margin of the Ardennes. (d) THE ROAD TO THE SOUTH: Fig shows Roman roads connecting Lugdunum to the South. Lugdunum and Arelate were both Roman colonies and Vienna was almost certainly a colony before Agrippa first came to Gaul 119. It is, therefore, almost certain that there was a Roman road from Arelate to Lugdunum by 39 B.C. Drinkwater argued on this basis that Strabo's road to the South was not part of the Agrippan system 120. Yet, this could not function without the link, which was a funnel for all the through traffic. Whether Agrippa was obliged to carry out any works on the road link in B.C. or 19 B.C. depended on its condition and the rate of build-up of the traffic on the other three Agrippan roads, and nothing is known about these factors. The road on the left, or east, side of the Rhône has been identified as the 'Agrippan road' of the Empire. Two routes between Lugdunum and Vienna are from the Claudian period when bridges were built connecting Lugdunum on the west bank of the Rhône to the east via the island of Bellecour 121. The first Roman road between Lugdunum and Vienna was on the right or west side, and it crossed the Rhône at Vienna to continue on this side to the south 122. The road on the right bank south of Vienna was convoluted and subject to flooding, being confined to a strip of land between the river and high land to the west 123. It connected settlements on the river, whereas the 'Agrippan road' on the left bank, meeting the river only at Vienna, Valentia and Avennio, was set out to a good line by Roman surveyors. There are no See above Note 68, p.25. Based on Rivet, Fig.38, p.278 and Fig. 42, p.302. Arelate was founded by T. Claudius Nero on behalf of Caesar in 46 B.C., Suetonius, Tiberius 4.1. Lugdunum was founded in 43 B.C. by Munatius Plancus, CIL Rivet, pp.305-6, discussed the evidence for a Roman colony at Vienna from which Roman citizens were expelled and for whom the colony at Lugdunum was created (Cassius Dio, 46.50). Drinkwater, Roman Gaul, p.125. Roddaz, Marcus Agrippa, p.372 and Note 41; Chevallier, pp.211-2, also see below Figs. 2.1 and 2.2. Chevallier, p.212, and Note 61. Ibid, p.212, for a description of the road on the right bank.

40 surviving Roman road markers on the road east of the Rhône earlier than 3 B.C At Orange (Arausio), the line of the road entering the town from the north through the Triumphal Arch continues along the cardo maximus, indicating that the town grid was set out from the existing road. There is no definite dating for the foundation of the town, but it was, almost certainly, before the second visit of Agrippa to Gaul in 19 B.C There is, therefore, insufficient evidence to indicate whether Agrippa was responsible for any road construction south of Lugdunum, but it is probably safe to conclude that he did little or nothing in this region. His immediate concern was to provide a strategic network in Gallia Comata, where there were no Roman roads. Furthermore, the colonies, which used the road south of Lugdunum, would have been expected to upgrade it to carry the essentially commercial traffic, which built up because of the road works to the north. 2. INVESTIGATION OF THE AGRIPPAN NETWORK Fig shows the locations of the Agrippan roads as deduced in section 1 where it was also argued that construction was put in hand in B.C., except for the road to the South, where any Agrippan construction cannot be dated. In this section, the object of enquiry is identification of factors which influenced the development of the Agrippan network as a whole. Investigation is divided into two parts. In the first of these, consideration is given to what Agrippa would have sought to achieve. In the second, the practical limitations on Agrippan achievements are examined by reference to technical factors. He could build only a certain amount of road of a particular standard in a specific period of time with the forces available to him. He was, therefore, constrained by elements, some of which he could not control Rivet, p.79 and Note 45, p.82. The town was mentioned in Livy, 67, Strabo, and Pliny, Cassius Dio, 49.34, said that mutinous soldiers were sent to Gaul after the defeat of Sextus Pompeius, in 36 B.C., and, perhaps, to Orange and Béziers, as suggested by Rivet, pp.150 and 172. Goudineau, CAH 10, 1996, p.477, dated the foundation between 40 B.C. and 28 B.C. Bellet, M.E., Orange Antique. Guides Archéologiques de La France, Paris, 1991, p.14, noted that the earliest dateable evidence of Roman occupation at Orange is from 15 B.C., but the principal Roman monuments there are Augustan or later, and earlier works may have been disturbed when these were built. Rivet, p.79, noted the remains of a structure under the Triumphal Arch dedicated to Tiberius. Based on Drinkwater, Roman Gaul, Map 7, p.238, and his place names, pp

41 (a) HISTORICAL AND STRATEGIC FACTORS: The Caesarian policy in Gallia Comata following the conquest was essentially one of laissez-faire. Provided that tribute was paid, and there were no serious disorders, there would be little Roman interference in Gaulish life and government. Roman interests in Gallia Comata were safeguarded by the legions in encampments, or at Gaulish sites, together with alliances between Rome and local chieftains 127. When Octavian had the West allocated to him at the Treaty of Brundisium in 40 B.C, he did not want to direct substantial resources there while he was consolidating his position in Italy, and in the Roman world generally. On the other hand, he wanted to take definite possession of the territory which had been conquered by his 'father', and was a counter to Antony's control of the East. His policy would have been to introduce some measures of additional control to ensure that the territory remained quiet. The most powerful control measure was effective road communication reaching potential trouble spots for rapid transmission of intelligence about unrest and movement of military forces in response. This would enable the size of the army to be reduced. A road system radiating from Lugdunum to the Atlantic, the English Channel, and the Rhine could be used by forces stationed at Lugdunum or in Gallia Comata, leaving southern Gaul unaffected. The roads, forts and garrisons near or on them would be built by the army. Drinkwater considered that Agrippa built well engineered roads, starting in 39 B.C., and that this programme was completed by 27 B.C He also discerned military changes on Agrippa's arrival with larger forces being replaced by smaller more mobile units using the roads to reach trouble spots from garrison forts on the roads 129. It is unlikely that such a large network could have been built to a high standard by 27 B.C. Also, completion of the network by this date was of little use to Agrippa, who was probably asked by Octavian to institute a control system before he left Gaul. Agrippa had to reorganise the military forces and build the roads at the same time, otherwise he would not achieve his objectives of control and recasting the military forces, which went hand in hand. Roddaz argued that Agrippa was sent to Gaul in October 40 B.C. 130, and he was probably recalled in early 37 B.C 131. It is possible that Agrippa was meant to stay in Gaul longer and was Drinkwater, Roman Gaul, pp for a brief assessment of the Caesarian settlement. Ibid, pp.21, 93, Ibid, pp.122, 3. Roddaz, Marcus Agrippa, pp.60-69, also above Note 10, p.12. See above Note 11, p.12.

42 recalled to deal with Sextus Pompey 132 earlier than needed for him to become consul in 37 B.C Certainly, Agrippa was compelled to act on the roads and the forces in a few years, and had no option but to provide a road system of a lower standard than that described by Drinkwater in order to complete his network before he left the region. Augustus organised the first census in Gaul and set in train the creation of the three provinces of Aquitania, Lugdunensis and Belgica in 27 B.C., and not when he was again in the West in B.C There was no need for new roads to be built to serve the capitals of the provinces 135, which probably became effective administrative units many years later. Nor is there any reason to think that Agrippa did anything to the roads in 19 B.C. because of the initiative of 27 B.C., even if he had sufficient time to do so (he was in Gaul for only about six months, put down disturbances in N.E. Gaul and was active at Lugdunum 136. Drinkwater emphasised the connection between roads and the census, since the latter called for a cadastre with details of boundaries, ownership, etc., and roads provided fixed points in the landscape 137. Yet Agrippa's road network formed only a narrow ribbon of land and did not exist in N.W. Gaul. It is, therefore, difficult to discern a strong connection between census and the roads in Gallia Comata as a whole. A census could proceed irrespective of road works. Consequently, a census ordered in 27 B.C. would have had no significant influence on road construction, or have led to Agrippan road action in 19 B.C. Chevallier suggested that Agrippa may have been asked to undertake cadastration in 19 B.C. 138, but any such action did not necessarily call for road construction See above Note 10, p.2, for Sextus Pompey. Ehrenburg and Jones, p.33, for the consulate. Cassius Dio, 53.22, for the ordering of the census and the initial step to create the three provinces in 27 B.C., as supported by Goudineau, CAH 10, 1996, p.487, and Drinkwater, Roman Gaul, p.20. Cassius Dio, 54.23, describing the visit of Augustus to the West in B.C. did not indicate establishment of the provinces. The first capital of Aquitania at Saintes (Drinkwater, Roman Gaul, p.96, and Note 21, p.115, Goudineau, CAH 10, 1996, p.488) was surely on a road by 27 B.C., even if Agrippa had not built his road to it in B.C. Reims, the first capital of Belgica (Strabo, 4.3.5, Drinkwater, Roman Gaul, p.38, Goudineau, CAH 10, 1996, p.488) was the long-established capital of the Remi. Lugdunum, the first and continuing capital of Lugdunensis was on a Roman road before Agrippa visited Gaul, as already noted. See above Note 14, p.12', Cassius Dio, 54.11, and below Chapter 2. Drinkwater, Roman Gaul, p.125. Chevallier, p.211 and Note 57.

43 It can, therefore, be concluded that the actions of Augustus in 27 B.C. had little or no effect on road planning or construction. Consequently, they would have led to no major Agrippan road works in 19 B.C. The impetus for building a first Roman road network was the acquisition of Gaul by Octavian in 40 B.C. The first form of this was completed by Agrippa in B.C. It remained to form the trunk routes of the much larger and higher standard Roman road network of the Empire. The Agrippan roads were, therefore, gradually upgraded over a long period, which started when Agrippa left Gaul in 37 B.C. The Agrippan visit of 19 B.C. was a short event during which little could have been built. (b) TECHNICAL FACTORS: Agrippa's roads were to reach his desired destinations as directly as possible, be open to traffic all year, and have the capacity to carry his military forces and military intelligence. The existing roads would have been investigated by Agrippa, and he would then have assessed how he could best cobble together an acceptable system from these. Where Agrippa was obliged to build ex-novo the standard which he adopted would have been the minimum to achieve his objectives. i. Locations of the pre-agrippan Roads: As pointed out by Renardet 139, Agrippa s roads were planned by a single authority to reach specific destinations as directly as possible. Pre-Roman through routes, on the other hand, had evolved over a long period, and were formed in some cases by connecting local roads radiating from local centres. The Roman roads would surely disregard them and by-pass the places on them. Yet this finding must be qualified in the case of the Agrippan roads. Through roads or tracks along the Moselle, Saône and Rhône arose because of the rivers and movement along these. The settlement pattern was therefore of the ribbon rather than the radial type, and was what Agrippa wanted for his road locations. In the case of the road to the West, both pre-roman and Agrippan road locations were influenced by the same topography of natural routes through an upland area. As noted above, Agrippa's road to Ocean followed the trace of an existing through route. Also, Renardet had in mind the fundamental difference between pre-roman and Roman road planning, but Agrippa was 139 Renardet, E., Vie et croyances des gaulois avant la conquête romaine, Paris, 1975, p.100.

44 obliged to compromise on road location to some extent to achieve his aims. Furthermore, as pointed out by Jullian 140, the Agrippan roads followed the ancient routes of the people dictated by the topography. There is no need to take into account the extensive use of rivers in Gaul referred to by Strabo 141, and which can be envisaged from the penetration of these into the interior, as shown on Fig Rivers could not be used along the routes to the West and the English Channel, and, as already noted, roads existed along both sides of the Rhône, which could, in any case, not always be used for transport 142. Although Agrippa could have used the general locations of the existing roads, it was necessary to adjust the actual alignments. Fig shows a typical pre-roman road, Fig , a secondary Roman road, and Fig , the 'Agrippan road' of the Empire between Amiens and Beauvais. It was not feasible for Agrippa to aim at alignments like that on Fig. 1.12, because this would have entailed ex novo works which he could not complete. Yet he could not tolerate the sinuous pre-roman roads over the whole of his network, since this would have greatly increased the total road length. He probably wanted to create alignments like those shown on Fig In some places he could provide this within the existing road formation. Elsewhere he would be obliged to build ex novo. This new work could not be extensive if he was to provide his system quickly, nor could his re-alignments be drastic and frequent. ii Standards of pre -Agrippan Roads : The integrity of the roads under the loading of military traffic is not in doubt since they had been developed over a long period from tracks to carry vehicles and consisted of many layers of consolidated materials 146. Caesar had no Jullian, 4. p.85. Strabo, Ibid, 4.1, 14, noted that it was difficult to sail up the Rhône. Chevallier, Fig. 2, p.27. Ibid, Fig. 102a, p.160. Ibid, Fig. 100, p.159. Renardet, pp

45 difficulty in using the roads for his conquest 147. Also, the Helvetii intended to use wagons in their migration to the west coast of Gaul 148. This is not surprising since the cart, or wagon, had been in general use in Gaul for centuries before Caesar's conquest, having appeared in Europe and nearby areas in the 2 nd millennium. In the Iron Age, there was widespread use of carts, as is known from the cart burials of the Hallstatt culture 149. The site of a cart burial can be seen on Fig The Romans borrowed technology of wagon building from the Celts and the names of the different types of wagons in Gaul, which were well-adapted to the economy and terrain 150. Agrippa would not have encountered serious difficulties in using the existing roads throughout the year since there were many pre-roman bridges 151, indicating that the roads would not generally be subject to flooding. Also, frequent river crossings were necessary only on the roads to the West and to the English Channel, since the other two generally followed rivers rather than crossed them, except at Vienna. The pre- Roman roads were probably not suitable for rapid communication by light vehicles and horsemen carrying intelligence, since the road surfaces had not been designed for this purpose. Resurfacing could be done easily since the pre-roman roads were surfaced with small stones and gravel, which could be harrowed and reinforced by smaller gravel. It can, therefore, be concluded that Agrippa could have formed a viable strategic road system on the routes described by Strabo, and have substantially completed this while he was in Gaul in B.C. He could have done this only by relying principally on the existing roads, which he would have improved, and by adding some minor ex-novo work. Agrippa's first system could only have been built to a modest standard, but nothing further was needed for a strategic network Chevallier, p.25 and Notes 9-11, for Caesar's use of the roads and general comments on the pre-roman road works. Caesar, BG, 1.3. Pigott, S., Daniel, G. and McBurney, C., Eds., France Before the Romans, London, 1974, pp.183-5, René, J., Vix et ses trésors, Paris, 1979, pp and Fig p.89, Powell, T.G.E., The Celts, London, 1985, Fig. 9, p.30 and Fig. 12, p.35, showing the evolution of carts, and Augouze, F.K. and Büchsenschütz, O., Towns, Villages and Countryside of Celtic Europe, London, 1992, pp and 24, for the general development of road transport. Renardet, p.101; Coulon, G., Les gallo-romaines, Paris, 1985, p.25; Chevallier, p.28. Chevallier, Note 10, p.26, provided a list of bridges mentioned by Caesar in Bellum Gallicum.

46 CONCLUSIONS The evidence indicates strongly that Agrippa created the first form of his road network in B.C for strategic reasons, and mainly by using and improving existing roads. This system remained to form the framework of the much larger Roman road system of the Empire, and was upgraded over a long period 152. There were no significant road alterations as a result of the initiatives of Augustus in 27 B.C. in creating the three provinces and taking the first census, and any road works by Agrippa in 19 B.C. were cosmetic. There is no case for Agrippa starting his road planning and construction in 19 B.C. Agrippa's road network provided the first framework of Roman control, and places on it were, in some cases, the first locations of Roman influence. Agrippa built his road to the West to reach the trouble spot of the old Aquitania and did not end it at Saintes, as generally understood. Also, his road reached the Rhine at Bonn and not at Cologne, as is usually argued. It is unlikely that Agrippa was responsible for road construction south of Lugdunum. It is now generally agreed that Agrippa planned and started to build his road system in B.C., rather than in 19 B.C., but no consideration seems to have been given to his use of the existing roads to cobble together a first viable network. Yet, this was simply an extension of Julius Caesar's use of the roads, and what should be expected from Agrippa as a Roman general using existing resources to the greatest possible extent. The generally accepted hypothesis of Agrippa starting to build a high standard road system ex novo rests on the powerful modern image of the 'Roman road' built to a high standard, straight, and elevated above the landscape, as seen on Fig Yet, there were many categories of Roman roads 153, one of which is shown on Fig. 1.11, and the standard depended on the local needs of the time. The high standard roads of the Empire, as shown on Fig. 1.12, served the needs of that time. The key to understanding the Agrippan road network is to set aside modern ideas of the high standard Roman roads of the Empire and to assess Agrippa's reactions to the problem faced by him Rivet, p.79 and Note 45, p.82, noted Augustan markers on the road to the South. Roddaz, Marcus Agrippa, p.390 and Note 29, described Claudian markers on the Agrippan roads and Chevallier, p.210 noted 25 markers on the Agrippan road to the Rhine dating from the Claudian period to Constantine 2 (A.D ). Chevallier, pp.32-35, describes the various classes of Roman roads.

47 LUGDUNUM (LYON) It was argued in Chapter 1 that Agrippa built the first form of his road network radiating from Lugdunum in B.C. and that he did little, if anything, to it in 19 B.C. The constructions to be investigated in this chapter are his road works of B.C. at and near Lugdunum, excluded from discussion in Chapter 1, and all his other works at Lugdunum of B.C. and 19 B.C. 1. THE FRAMEWORK OF AGRIPPAN ACTIVITY The general location of Lugdunum was indicated in the chapter on the roads. Fig shows the topography and the courses of the rivers Saône and Rhône in the Roman period. Fig shows a model of the city of Lugdunum in the Empire. Fig indicates the modern city with solid ground and extensive development between the Saône and the Rhône from the foot of the Croix-Rousse, including the former Island of Bellecour. Consequently, the ancient writers referred to the confluence of the Saône and Rhône, as shown on Figs. 2.1 and 2.2, and, the Roman works at river level are largely lost. The main public buildings of Lugdunum were on the Hill of Fourvière and the plateau of La Sarra, with the circus on the Trion below the plateau, and there was extensive development on the Island of Bellecour, which was connected by bridges to both sides of the rivers. The Altar of Rome and Augustus and the amphitheatre were on the Croix-Rousse across the Saône and joined to the town by a bridge. As noted in Chapter 1, the location of Lugdunum made it a key centre of Roman security. Also, the Plateau of La Sarra, with its Hill of Fourvière, could be defended, and it overlooked a crossing of the Saône north of a complicated river confluence. Yet, since rainfall was the only source of water on the plateau, and the ground was permeable 4, it was not naturally suitable for largescale permanent human habitation until cisterns and wells were in general use. It is, therefore, not surprising that there is no evidence of permanent native occupation Audin, A., Lyon miroìr de Rome, Paris, 1979, Fig. p.31 Photograph of the maquette in the Musée de la Civilisation gallo-romaine Lyon, Burdy, J., Guide des aqueducs romains de Lyon, Lyon, 1999, p.5. Desbat, A., 'Nouvelles recherches à l'emplacement du prétendu Sanctuaire Lyonnais de Cybèle', Gallia, 55, 1998, pp , Fig. 2, p.239. Burdy, J., Guide des aqueducs romains de Lyon, p.7.

48 before Roman colonisation 5, but only of military encampments by Roman armies 6. There is evidence of Roman cisterns and wells, and four aqueducts served the Roman town 7. In fact, Lugdunum became a Roman colony, and the hub of Agrippa's road network, because its strategic location outweighed the water shortage. The fact that there had never been an oppidum on the plateau made the area a neutral zone where native leaders could meet, and, therefore, appropriate for the construction of the Altar of Rome and Augustus 8, although this was not at the town itself. The Roman colony at Lugdunum was founded by Munatius Plancus in 43 B.C. for Romans who had been expelled from Vienne, and, also, to establish a strategic military centre 9, but the foundation text said nothing specific about its location. A text of Strabo has been seen by some to suggest that it was at or near river level, but the general consensus now favours the plateau 10. Also, a Roman colony would not have been founded below dominating positions, especially when old encampments on one of these could be used for an urbs quadrata. Italian merchants, expelled from Vienne in 62/61 B.C., may have settled at the confluence of the Rhône and the Saône 11, but their works, now lost, were almost certainly near the river and are disregarded here. There is, therefore, no possibility of an Agrippan first Roman settlement on the plateau. He could only add to what Plancus had done there Walker, S., Ed., Récentes recherches en archéologie gallo-romaine et paléochrétriene sur Lyon et sa région, BAR-1S, 108, 1981, Oxford, pp.29-54, also Burnouf, J. et al, Lyon avant Lugdunum. Un habitat du premièr Age du Fer à Vaise, Goudineau, C. & Mandy, B. Eds., Aux Origines de Lyon, pp See below Notes 22 and 24, pp.43, 4, for the archaeological information on the camps. See below Note 95, p.57, for the cisterns and p.52, for the aqueducts. Drinkwater, J.F., Lugdunum : Natural Capital of Gaul?, Britannia 6, 1975, pp Cassius Dio, 46.50, for the foundation text. CIL 10, 6087, on the tomb of Plancus recorded the foundation (and that of Raurica), also, Goudineau, C., 'Les textes antiques sur la fondation et la topigraphie de Lugdunum', Aux Origines de Lyon, pp Strabo, , said that Lugdunum was sited under a hill near the confluence of the rivers. He described the Altar of Rome and Augustus as rising in front of it at the confluence. Goudineau, Les textes antiques, p.29, pointed out that the text indicated the town as only near the confluence, whereas the altar was at it, therefore proving two different sites. He also suggested that Strabo described the town as generally below the hill of Fourvière. Cassius Dio, 37, 47-48, mentioned expulsion of settlers from Vienne in 62/61 B.C., but, in his foundation text, (referring to 43 B.C.) he gave no indication when settlers at the confluence of the Rhône and the Saône had gone there. Audin, pp.38, 39 and Drinkwater, Lugdunum, pp supported the idea of a settlement in 62/61 B.C., but neither provided any clear evidence. Goudineau, Les textes antiques ', p.24, argued against it on the basis that settlement would not have occurred there before the conquest, and that, if this had happened, a colony would have been created before 43 B.C.

49 Plancus probably formed an urbs quadrata on the plateau. By 39 B.C., Lugdunum had a garrison on the hill and an agglomeration of settlements at various sites. The population was probably numbered in hundreds rather than thousands and no significant construction in permanent materials can be envisaged. In B.C. Agrippa was interested in Lugdunum for strategic purposes. It is, therefore, unlikely that he would have been concerned to expand the town, or to start any monumental construction. He would have focussed on the military and road functions. In 19 B.C. Lugdunum, the hub of Agrippa s road system for some two decades, must have become a commercial centre. The principal mint of the Empire was established at Lugdunum in about 15 B.C. 12. This dating is not to be confused with that of the first issues of the Altar series, in 10 B.C. 13. The 10 B.C. dating has, in turn, led to the dating of the Altar to Rome and Augustus to 10 B.C. and the dedication by Augustus himself 14. Yet, dating to 12B.C. with the dedication by Drusus seems to be well established 15, and is accepted here. The new Mint and the Altar indicate political development of Lugdunum, reinforcing the previous commercial expansion, and, therefore, also a corresponding programme of monumental development. The need for urban development has been accepted, but this has been dated to B.C., following the visit of Augustus to the West 16. Yet, it is possible that this was undertaken earlier, since, in 27 B.C., Augustus set in hand the creation of the three provinces, with Lugdunensis as the most important of these, and Lugdunum as its capital. Also, the problem of water may have become serious before 15 B.C. and, in this case, it would have been necessary to consider the construction of aqueducts in addition to expansion on of the town itself. There is, therefore, reason to think that Agrippa in Gaul in 19 B.C. with his experience in urban expansion and water supply at Rome 17, could have undertaken Burnett et al, p.150. RIC, p.28. Ibid, p.27. Gros, P., L Augusteum de Nîmes, RANarb, 17, 1984, pp , p.129, and Woolf, Becoming Roman, p.216 and Notes 50 and 51 referring to several studies on the altar. Tranoy, L. and Ayala, G., Les pentes de la Croix-Rousse à Lyon dans l antiquitie. État des connaissances, Gallia, 51, 1992, pp , p.188. See below Chapter 11, pp.207, 8.

50 planning, and, possibly, construction for the expansion of Lugdunum, and that he may have been asked to do so by Augustus. Yet, these initiatives could have been taken by Augustus in B.C., and there is insufficient evidence to decide who was responsible, either partly or wholly, for the expansion. It can only by said that Agrippa could have acted rather than Augustus, as is generally understood. 2. AGRIPPAN PLANNING AND CONSTRUCTION Investigation in this section is in three parts. The first is concerned with Agrippan roads at and near Lugdunum constructed in B.C. Agrippan planning and construction at the urban centre in B.C. and 19 B.C. is examined in the second. In a third part, Agrippan planning and construction of aqueducts serving the urban centre is investigated. (a) AGRIPPAN ROADS: According to Strabo, Lugdunum is in the centre of the country an acropolis as it were, not only because the rivers meet there, but, also, because it is near all parts of the country. And it was on this account, also, that Agrippa began at Lugdunum when he cut his roads 18. It has already been argued here that Agrippa knew Lugdunum to be nowhere near the centre of the country and that he built his roads from there for strategic reasons. The acropolis was a disadvantage for a transport node which, ideally, is at the same level as the roads to be connected so that through traffic need not negotiate a hill. The meeting of the rivers refers to the confluence of the Saône and the Rhône. This made Lugdunum a strategic position but not necessarily a major river trading station. Vienne and Chalon-sur-Saône, shown on Fig. 1.6, were both more important than Lugdunum in the pre-roman era 19, although Lugdunum could have been a base for trade up the Rhône to Lake Geneva. It seems unlikely that the pre-roman river trade had changed significantly by 39 B.C. after foundation in 43 B.C. Even if Lugdunum had become important for river trade, Agrippa would have chosen it as the hub of the roads because it was a Roman colony with a garrison at a strategic position in an existing system of military security connecting Vienne, the Roman positions on the Swiss plateau and Raurica. He would have avoided Vienne, which had expelled Romans. It was also on the east of the Rhône when Strabo, Cunliffe, B., Greeks, Romans and Barbarians, London, 1988, pp.49-55, 86, 87.

51 Agrippa's roads to the West, the English Channel and the Rhine were all to the west of the Rhône-Saône system. In short, selection of Lugdunum was almost inevitable unless Agrippa founded a new colony. This would not have been done when Octavian wished to contain potential problems in Gallia Comata. It has been argued in Chapter 1 that Agrippa formed his initial road network of B.C. as far as possible from existing roads. A natural crossing of the Saône at or near Lugdunum, marked on Fig. 2.1, indicates a pre-roman W-E road, but this would not have passed over the uninhabited plateau. A road must have been made by the Romans before B.C. to connect Vienne, Lugdunum and the Roman positions in the Swiss Plateau. This would have crossed the Rhône at Vienne to reach Lugdunum on its west side, since the crossing there was simpler than at Lugdunum. At Lugdunum the road would then have ascended to the garrison on the plateau, descended, and have crossed the Saône to go up the Rhône to Lake Geneva. Agrippa could form his road to the West by striking off from this Roman town in the direction of Feurs 20. He probably had to build little, if any, new roads once he was clear of Lugdunum itself since the route was defined by the topography. For his road to the South Agrippa had only to follow, and perhaps upgrade, the Roman road from Lugdunum to Vienne for his first construction, although, as seen on Fig. 2.2, the later form of the road crossed over the river at Lugdunum via the Island of Bellecour 21. The region of Lugdunum, with its plateau, formed a cul-de-sac for S-N land passage 22. Consequently, the main pre-roman road between Vienne and Chalon would have by-passed Lugdunum, leaving it to the east. Fig shows the position of the rivers. A route leaving the west bank of the Rhône near Givors and going up the valley of the Garon could reach the valley of the Brévenne and then go down this to Anse. Agrippa would, therefore, have used the existing road southward from Chalon as far as Anse, but would have been obliged to cut a new road from this point to reach Lugdunum, some 17 km further south. At Lugdunum, it connected to the existing Roman road to the south and the pre-roman road to the west Marked on Fig See above, Chapter 1, Note 121, p.31, and Fig Mandy, B., Sandoz, S. et al, Les fossés du plateau de La Sarra, Aux Origines de Lyon, pp.37-94, at p.92. Based on Besnier, M.M., 'Le point de départ des grandes routes ', Fig. 2, p.87.

52 All the Agrippan roads reaching Lugdunum would have been brought together most conveniently at the Trion site, below the top of the plateau, marked on Fig The road from Vienne to Lugdunum would have passed through the Trion, as can be seen from the contours on Fig. 2.1, between St. Irénée and the Trion. The road from the north (the combined road to the English Channel and the Rhine) reached the Trion by climbing the contours on the west of the plateau. The road to the west would automatically pass through the Trion. With a junction at the Trion, through traffic did not have to ascend the slope of the plateau, but there was a connection for traffic to the urban centre at the top. It is clear from Fig. 2.2 that the interchange facility at the Trion was still used in the Empire, although other roads provided more direct access to the top of the plateau, or avoided it altogether by passing around it. (b) AGRIPPAN PLANNING AND CONSTRUCTION AT THE URBAN CENTRE: Works of the period B.C. Agrippa must have built a construction camp at or near the Trion, since he formed his roads from Lugdunum. He must also have provided for his military movements on completion of the roads by building horse changing facilities, stores for animal feed, and perhaps overnight accommodation at the Trion to reinforce facilities on the plateau. Supplies from the south via the Rhône would have been received at river port facilities, and a road connection from these must have been built. The decision to form his strategic road network at Lugdunum increased the military importance of the town and Agrippa almost certainly planned, and built, some new permanent military works on the plateau. There seems to be no evidence of early Roman buildings at the Trion site. Ditches discovered on the top of the main plateau of La Sarra contain traces of fortifications or encampments, the full extent of which is, as yet, unknown. Fig shows the ditch at the Rue Henry Le chatelier site, two ditches at the Clos du Verbe-Incarné site, and the fourth ditch at the Hôpital Sainte-Croix site at the S.W. side of the hill of Fourvière. Dating of the finds in these ditches indicates transitory use from the Caesarian period, and perhaps earlier. The Verbe-Incarné ditches were probably not used after 60 B.C., but the other two were used as late as about 40 B.C. All ditches contained evidence of military use and the Hôpital Sainte-Croix ditch also had animal 24 Mandy, B., Monin, M., and Krauzz, S., L hôpital, Sainte-Croix à Lyon un quatrième fossé, Gallia, 47, 1990, pp , Figs. 1, p.80.

53 bones of a type which indicated civil as well as military food consumption 25. The ditches, therefore, provided evidence of occupation of the plateau shortly after the foundation period, but it is not possible to know whether this resulted from the activities of Agrippa in B.C. Figs. 2.6a and 2.6b 26, respectively, show the Verbe-Incarné site located on Fig. 2.5 with nine divisions, and the cross-hatched part of this to a larger scale. Fig. 2.6b indicates a permanent building 30m by 20m in size, of which the first construction shown in bold black lines has been dated to about 40 B.C. There is, therefore, evidence of a new grid superseding that of the ditches and dating from the time of foundation, with a permanent building aligned to it. The grid was probably based on the original Roman road over the plateau, the alignment of which is perpetuated by the present Rue Roger Radisson. It is not possible to say whether the building was built by Agrippa or Plancus 27. If it was Agrippan it could have been a first praetorium. Permanent construction dated from 40 to 15 B.C. has been discovered at the higher (western) end of the Rue des Farges site shown on Fig , but it is not possible to suggest Agrippan works at this site which was nowhere near the garrison on the top of the plateau. The baths of the Claudian period on the lower (eastern) end are discussed when investigating the aqueducts 29. Very early construction at the Sanctuary of Cybele site immediately above the level of the theatre 30 marked on Fig. 2.7 cannot be related to Agrippa in B.C. There is, therefore, no definite archaeological evidence of any Agrippan works at Lugdunum from B.C., but it is possible that the earliest material remains at the Verbe-Incarné site are partly Agrippan. It is also clear that Agrippa must have built his roads, the facilities needed for their construction and operation, and that he almost Ibid, pp.95, Desbat, A.; Genin, M. et al, La chronologie des premières trames urbaines à Lyon, Aux Origines de Lyon, pp , Fig. 61, p.100. Ibid, pp and Figs. 61b and c, p.100, also Mandy, B., Le quartier antique du Verbe-Incarné, Les Dossiers d histoire et d archéologie, 78, 1983, pp for further details. Desbat, Genin et al, Fig. 57, 'La chronologie ', Fig. 57, p.96. See below, p.57. See below, p.48, for the phasing of construction.

54 certainly built military works on the plateau, as well as facilities for recovering his river borne supplies and the road connection from these. Works of 19 B.C. Fig showing the traditional idea of development on the main plateau in the Augustan period was presented by Desbat and Mandy to indicate what should be questioned in the light of more recent archaeological investigations 32. The long, straight street running N-S along the eastern side of the plateau was first identified as the cardo, with a T-junction at the decumanus at right angles to it. An astronomical calculation indicated that the sun rose on the line of the decumanus on 11 October 43 B.C. 33. Audin later discerned the cardo continuing north of the intersection, as seen on Fig. 2.8 and revised the date to 9 October 43 B.C. He also related the annual return of migrating crows to Lyon in October to the augury of foundation 34. The location of the street intersection and the foundation date must be questioned in the light of the early grid at the Verbe-Incarné site mentioned above. Also a foundation of Lugdunum earlier in 43 B.C., is indicated by the local coinage and the texts 35. Furthermore, the intersection would surely have been on the plateau in a position that could be defended, and where buildings could be on level ground, rather than on the eastern slope of the plateau. The theatre, shown on Fig. 2.8, built on the plateau slope, has traditionally been regarded as Augustan, with the Odeum immediately to its south, marked on Figs. 2.2 and 2.3, dated later. The first works at the theatre were dated to B.C. by Audin, following the investigations of Wuilleumier 36, indicating a construction by Augustus, who was in the West at that time. This dating has been questioned in a Desbat, A. and Mandy, B., Le développement de Lyon à l époque augustéenne : l'apport des fouilles récentes, Les villes augustéennes de Gaule, pp.79-97, Fig. p.80. This figure is based on Audin, Lyon miroir de Rome, Fig. p.90. Ibid, p.97. Wuilleumier, P., Fouilles de Fourvière à Lyon, Supplément à Gallia 4, 1951, pp for the identification and the calculation. Audin, pp Burnett et al, p.150, indicated coinage of Antony struck at Lugdunum in 43 B.C. following coinage of Plancus. Also, Goudineau, C., 'Les textes antiques ', pp.26, 27, noted that the correspondence between Cicero and Plancus about the foundation of Lugdunum, Ad fam 10, 22 and 24, was from the spring of 43 B.C. Audin, pp.84-87, Wuilleumier, 'Fouilles de Fourvière '.

55 new study of the archaeology 37. Agrippa was in the Iberian Peninsula in B.C., and his theatre at Mérida in Spain was dedicated in B.C. 38. It is, therefore, possible that Agrippa was also responsible for providing the theatre at Lugdunum when he was in Gaul in 19 B.C. Fig shows Mandy s interpretation of the first phase of the works. According to this, the final cavea diameter of 108.5m was allowed for from the beginning. It would, therefore, have been necessary to complete all the earthworks and most of the foundations and cavea substructures, together with the entrances, and most of the circulation system. Part of the cavea seating and the first form of all the other buildings would have been included. Previous interpretations of the construction phasing postulated an initial cavea diameter of 89m and it is not certain which concept is correct 40. Even with the lesser diameter, the first phase was a large structure, and also very much larger than the small Agrippan Temple of Valetudo at Glanum 41. The theatre at Lugdunum is, therefore, potentially a major addition to the Agrippan corpus of work in Gaul. Fig. 2.8 shows an Augustan forum, capitolium, and, possible Imperial palace on the Hill of Fourvière. As can be seen from Fig. 2.3, this site is occupied by the Basilique Notre Dame de Fourvière, and little has remained of the Roman construction, none of which can be dated 42. The name Fourvière is a corruption of Forum Vetus, indicating the first forum, but this could have been at the Verbe Incarné site where the early grid has been discerned, or, possibly, in the area of the esplanade of the theatre marked on Fig. 2.8, where there was a building pre-dating the theatre and referred to as the 'praetorium of Plancus 43. The location, and dating, of the early forums are, therefore, uncertain and Agrippan planning or construction of a forum cannot be established. It can be said only that any Agrippan urban planning of 19 B.C. would have made provision for a monumental forum Mandy, B., Hernandez, E., and Mar, R., 'Plaidoyer pour de nouvelles recherches', Dossiers d'archèologie, 134, January 1989, pp.30-35, pp.30, 31. See below, Chapter 8, pp for the inscription on the theatre at Mérida. Mandy, Hernandez and Mar, unnumbered, Fig. p.34. Professor F.B. Sear has analysed the staging suggested by Mandy and previous interpretations. He concluded that no alternative can be eliminated pending further investigations. The writer acknowledges this information with thanks. See below Chapter 3 for the Temple at Valetudo. Audin, p.89, for the 140 x 75 metres cryptoporticus and some other fragments. Ibid, pp and 88.

56 There is, therefore, no indication of any Agrippan construction of 19 B.C., other than that of the theatre, from the traditional understanding of the Augustan town, as indicated on Fig Yet this emerges from later investigations. The site of the socalled Sanctuary of Cybele is marked C on Fig The western part is almost at the general level of the top of the plateau, and the site slopes down over some 14 metres to the level at the top of the theatre marked D. Desbat has established three main phases of construction on the Cybele site. The first one, at the lower level, contained private buildings from the early days of the colony, or, perhaps, earlier still. The second phase saw the addition of a large structure extending up the hill to occupy much of the site. This building existed from approximately 20 B.C. until A.D. 10. It was then replaced by a yet larger complex known as the 'Sanctuary of Cybele. A final phase, dating from approximately A.D. 160 may have included construction of a reservoir for the Gier Aqueduct 45. Archaeological investigations continue 46. The various phases of construction have been deduced by Desbat from numerous excavations which revealed the process of building and the chronology of the building phases from the datings of finds, mainly ceramic, in the same stratigraphy as the constructions 47. The findings of the phases and the chronology are accepted here as the basis of discussion. The layout of the second phase building is shown on Fig in solid black, with the Sanctuary of Cybele as a background and the curve of the cavea of the theatre indicated to the east. Figs. 2.12a and 2.12b 49 illustrate the principal features of the construction. The three figures indicate a large house with an atrium and a peristyle. Construction of the Sanctuary of Cybele involved destruction of the second phase building above its floor level, but the foundations were partly undisturbed since the new building was on a different axis and had a different layout Based on Desbat, A., Nouvelles recherches à l emplacement du prétendu Sanctuaire Lyonnaise de Cybèle, Gallia 55, 1998, pp , Fig. 3, p.240. Ibid, pp.237, 256, 263 for main indications of the phasing. Desbat, A., Rapport sur les fouilles effectuées en 1998, 1999 et 2000 dans l édifice dit Sanctuaire de Cybèle, 2001, Lyon. Desbat, A., Nouvelles recherches, pp and Figs and Ibid, Fig. 19, p.257. Ibid, Figs. 20 and 21, p.258.

57 The prime position, large size, and lavish facilities of the house, reflected the status of its occupier 50. Desbat discovered evidence of opus tessellatum, hitherto not noted in buildings of this period in Gaul 51. The layout was typical of the late Republican house of Italy 52, but it was remarkable for its symmetry. This feature is stressed by Vitruvius as an essential element of houses 53, but is seldom seen so clearly as here. Such symmetry indicates the praetorium of the typical Roman camp. Desbat compared its layout with that of the praetorium at Oberaden in Germany (near the River Lippe) and there are certainly similarities, including the large atrium. He also mentioned the buildings at la Butte Saint-Antoine and la Plate-Forme at Fréjus interpreted as the residence of the Commander of the Fleet, but the layouts of these are less well understood 54. It is possible that the location, large size, facilities and strict symmetry of the building indicate the seat of the local governor. This idea is reinforced by the large atrium. Desbat considered that the building of the second phase was the praetorium of Agrippa on the basis of its dating, location, size, facilities and general features, although he acknowledged that there is no actual record which can confirm such an attribution 55. Desbat relied partly on Roddaz who stated that Agrippa was in Gaul between 20 and 19 or 18 B.C. 56. This suggested a lengthy stay. In fact, as stated by Roddaz elsewhere, Agrippa was in Gaul for only about six month 57. He had less time to build a praetorium than suggested by Desbat, but this does not weaken his hypothesis since Agrippa may have started to build it and have been called away unexpectedly. It could also be explained as part of possible Agrippan development As prescribed by Vitruvius, Desbat, Nouvelles recherches, p.234. Sear, F., Roman Architecture, London, 1989, pp.32, 33 and for a general description of the Roman house of the period and some examples of it at Pompeii. Vitruvius, De architectura, Desbat, Nouvelles recherches, p.256 and Fig. 4, p.260; also Février, P-A., Forum Iulii (Fréjus), Paris, 1963, Fig. 36, p.51, for the layout of the Plate-Forme, and Février Fouilles à la citadelle mérridionale de Forum Iulii (Fréjus, Var) en 1955, Gallia, 14, 1956, pp.35-53, Fig. 3, p.39, also below Chapter 5, p.105 and Fig Desbat, Nouvelles recherches, p.256. Ibid, p.256, Note 9, referring to Roddaz, Marcus Agrippa, p.384. See above Note 14, p.12.

58 of Lugdunum in 19 B.C., mentioned above. The building can be regarded as possibly Agrippan 58. As noted above, the next building phase involved virtual destruction of the praetorium above floor level, and the insertion of the large reservoir of the Gier, perhaps built in about A.D caused further extensive damage below the floor level. The parts of the praetorium which have survived are, therefore, most of its foundations, and fragments of floors with a large space punched out for the cistern. Desbat suggested that the atrium, which he measured as 12 m by 16 m, was covered by a roof since he found no trace of an impluvium to receive rainwater through the opening in the roof (the compluvium), nor any system to discharge water from the impluvium 60. It is difficult to understand how there could have been sufficient light without an opening in the roof, and the absence of an impluvium must be regarded as unusual for a building of this period 61. Uncertainties, therefore, remain about the absence of an impluvium and the roofing in the western part of the praetorium. As pointed out by Desbat, the oecus shown on Fig was almost certainly a dining room 62. Its 12 x 6m dimension indicating a length twice its width conforms to the precepts of Vitruvius 63. An atrium measuring 12m wide and 16m long would agree almost exactly with one laid out to the third Vitruvian method 64. The peristyle is unusually long in relation to its width 65. The layout of the building, therefore, corresponded in part to Vitruvian requirements. Since these also stipulated the vertical dimensions in relation to the horizontal ones it might be possible to form at least some idea of the elevations of the structure in more detailed studies. All the The criteria for Agrippan attribution are listed above, pp.5-6. See above Note 45, p.48. Desbat, 'Nouvelles recherches ', p.252 and Note 10. Sear, pp.32, 33 and Fig. 2b, p.13, commented on the possible early type of atrium (3 rd and 2 nd century B.C.) without the impluvium. Desbat, Nouvelles recherches, p.252. Vitruvius, De Architectura, Ibid, described three methods. The first two were based on the length as the measure. The width was either three-fifths of this or two-thirds. With a length of 16m, the width would be 9.6m or 10.67m. The third method used the width as the base and derived the length as the diagonal of the square formed by the width. With the width of 12m, this would give a length of 17m. Ibid, 6.3.7, prescribed a peristyle lying across the house to be only one-third longer than it is wide.

59 materials of construction for the praetorium above floor level were removed before building started on the next phase. Fragments of terrazzo flooring have been found, together with tiles, perhaps under a mosaic, and a section of opus tessellatum, in the balnea section. As can be seen from Fig. 2.12a, the floors were supported on soil built up above natural ground except for the western portico of the peristyle where there was a cryptoporticus with beams above this. The foundations of the walls were carried down into the natural ground. The remaining sub-floor walls and the foundations were in rubble masonry of very high quality using stones and mortar. Thickness varied from 6.60m to 6.80m 66. The locations of the praetorium and the theatre close to each other on the same hillside can be interpreted as an indication that they were planned, and, perhaps, built, together, and possibly by Agrippa. The different orientation of the theatre can be explained by necessary adjustments to suit the shape of the hillside. Nevertheless, it is also possible to argue that the two buildings were not planned or built together, since, as noted above, there is no certain dating of the theatre. As noted above, Agrippa provided the theatre at Mérida and may have provided that at Lugdunum. The conjunction of the two buildings can, therefore, be noted as a further possible indication of Agrippan responsibility for both structures. It can be concluded from investigations into Agrippan planning and construction at the urban centre that, in B.C., Agrippa built something on the plateau, at the Trion site, and at river level. None of these works can be identified as his from the material remains, but such works were needed because he had decided to base his road system on Lugdunum. The town became important because of this decision, and, in 19 B.C., Agrippa may have been responsible for planning and starting construction of a new monumental city, although the praetorium at the 'Temple of Cybele' site is the only building which has so far been identified as his, or probably so. Agrippa would have included a forum and a theatre in such urban planning. The theatre may, therefore, have been planned and perhaps partly built by him, like that at Mérida, but a forum of that period has not yet been identified, and no Agrippan construction of such a building can, therefore, be suggested from present evidence. 66 Desbat, Nouvelles recherches, pp , and Fig. 22, p.259.

60 (c) AGRIPPAN ACTION ON AQUEDUCTS: This is assessed as part of possible Agrippan urban planning, and also because of a local suggestion of Agrippan aqueduct construction. The local suggestion is considered after an independent examination. As can be seen from Fig , the four aqueducts of Lugdunum formed a very large system, of which some parts are yet to be found. There is, also, uncertainty about the datings of the various aqueducts. Consequently, investigation here can be neither comprehensive nor authoritative. It is based on the published literature and some local inspection, and is focussed on Agrippan actions. The following table has been drawn up to indicate the principal features of each of the aqueducts which are to be discussed. All the figures are from Burdy, except those in brackets, which are from Hodge 68. Differences of length result from uncertainties about the course, and difference in capacity from varying calculations based on uncertain evidence about the water channel features. NAME LENGTH (KM) CAPACITY (M³ PER DAY) LEVEL OF TOWN ENTRY (M) PART OF PLATEAU SERVED TRADITIONAL DATING Mont d'or 26 (28) 6,000 (10,000) 260 All except 30 m 10 B.C. to A.D. 10?? Yzeron 27 or 40 (25) 8,000 (13,000) 268 All except 22 m 10 B.C. to A.D. 10? Brévenne 70 (60) 10,000 (28,000) 284 All except 6 m Beginning of 1 st Century A.D. Gier 86 (75) 15,000 (25,000) 300 All Mid 1 st Century A.D. to early 2 nd Century A.D Bromwich, The Roman Remains of Northern and Eastern France, London, 2003, Fig. 95, p.417. Burdy, 'Guide des aqueducs romains de Lyon, Lyon, 1999, pp.13, 33, 71 and 78, Hodge, A.T., Roman aqueducts and water supply, 2 nd Ed., London, 2002, p.347.

61 All four aqueducts had siphons bringing the water across the lower land around the plateau which they served, and there were other siphons elsewhere, together with some cascades 69. Tunnels and viaducts were numerous on the Gier aqueduct and less frequent on the Brévenne. There is less evidence of these on the Yzeron, and no definite indication of tunnels on the Mont d'or. The traditional chronology is based on the work of de Montauzan published some 90 years ago 70. Burdy noted that the absolute and relative datings of de Montauzan, had become almost traditional, but that they were not beyond dispute 71. Wilson suggested that the chronology remained as elusive as ever, relying on Jeancolas 72. Jeancolas considered that the chronology of the construction of the aqueducts of Lyon was very uncertain 73. His comments on the particular aqueducts are taken up as these are discussed. Hodge was concerned with description and explanation of aqueducts rather than with the thorny problem of their dating. He only suggested in passing that the Gier aqueduct which, alone, could serve all parts of Lugdunum, was the last to be built 74. The traditional chronology is, therefore, doubted in the modern scholarship, but there is no consensus on what should replace it. Here, the object is to assess Agrippan action on the aqueducts by examining the material and historical evidence in a first section, and the aqueduct functions in a second. (i) The Material and Historical Evidence The Gier aqueduct is traditionally dated from the mid 1 st century to the early 2 nd century A.D. Inscriptions and (possibly) marking on lead pipes indicate A siphon was built where it was more convenient (or necessary) to avoid the construction of a viaduct with an open channel. Pipes were laid at or near ground level and the water flowed because of the difference in level over the pipe length. Cascades had 'drops' between channels on acceptable gradient to allow steep descents, Hodge, Roman aqueducts, pp De Montauzan, G., Les aqueducs antiques de Lyon, Paris, Burdy, J., Some Directions of Future Research for the Aqueducts of Lugdunum (Lyon)', Hodge, A.T., Ed., Future Currents in Aqueduct Studies, Leeds, 1991, pp.29-44, at p.43. Wilson, R.J.A., Tot aquarum tam multis necessariis molibus Recent studies on aqueducts and water supply, JRA, 9, 1996, pp.5-29, p.15 and Note 70. Jeancolas, L., Présentation des aqueducs antiques, de Lyon (problèmes anciens observations nouvelles), Boucher, J., Ed., Journées d études sur les aqueducs romains, Lyon, 1983, pp , pp and 185. Hodge, Roman aqueducts, p.174.

62 Claudian and Hadrianic work at some places on the aqueduct 75. The aqueduct has survived in several places over its length 76 and the remains indicate a common design and construction method, including standardisation of dimension 77. Fig shows a typical viaduct construction with opus reticulatum over a concrete core and brick. The same construction appears on the siphon structures supporting the pipes. The tunnels of the Gier are either cut in the rock and unlined, or lined with concrete with some brickwork, but there is no opus reticulatum. This was used in the tunnel inspection shafts 79. The opus reticulatum, typical of Rome and central Italy, is rare in Gaul, and, therefore, difficult to date. That of the aqueduct is probably of the mid 1 st century A.D. 80 Bricks have been noted at Lyon in coursework of the Augustan period, and in arches of the Tiberian principate 81, although Adam stated that bricks in courses in Gaul started as late as the early 2 nd century A.D. 82. The material evidence, therefore, indicates that the tunnels, which have no opus reticulatum, may be Augustan, but that everything else with both brick and opus reticulatum in the same structure is later and probably from the Claudian period, or after this. The indications of Claudian connection to Lugdunum and his substantial work on aqueducts at Rome are further evidence for his probable provision of the Gier aqueduct Desbat, Nouvelles recherches,pp.273, 4 and Note 16, Burdy, Guide des aqueducs romains de Lyon, p,39, Wilson, p.15 and Note 61, Leveau, P., Research on Roman Aqueducts in the Last Ten Years', Future Currents in Aqueduct Studies, pp , p.150, Burdy, 'Some Directions of Future Research ', p.43, and Jeancolas, p.183. Burdy, Guide des aqueducs romains de Lyon, described the remains. These included 11 tunnels, 10 viaducts and four siphons. Burdy, 'Some Directions of Future Research', pp Burdy, Guide des aqueducs romains de Lyon, photograph, p.59. Ibid, pp.16, 23, 28, 37, 43, 60 and 61. Opus reticulatum is first noted at the Theatre of Pompey at Rome dedicated in 55 B.C., Sear, p.76. It was used for parts of the aqueduct of Forum Iulii (Fréjus in S.E. France), which is probably post-augustan, Wilson, pp.18 and 19. Rivet, Gallia Narbonensis, p.230. Adam, J.P., Roman Building, Bloomington and Indianapolis, 1994, p.132, considered that the opus reticulatum of the Gier was probably from the mid 1 st century A.D. Desbat, A., 'Note sur l'appplication des constructions à arases de briques dans la région Lyonnaise', Gallia, pp.49, 1992, pp Adam, p.143. Claudius was born at Lugdunum, Suetonius, Claudius 2. At Rome, Claudius built the Aqua Claudia and the Aqua Anio Novus and repaired part of Agrippa's Aqua Virgo. Aicher, P.J., Guide to the Aqueducts of Ancient Rome, Wauconda, Illinois, 1995, pp.39-40, pp

63 The surviving Gier aqueduct is, therefore, predominantly Claudian. Claudius may have brought out an architectus from Italy to supervise the work in opus reticulatum 84, and the standard dimensions and uniform construction methods indicate a military project 85. Yet, it is possible that there was an earlier aqueduct on the general Gier route. This is indicated by unexplained remains along another trace not far from the principal remains 86, and by the possibility of the tunnels being earlier than the other works, as noted above. These tunnels could have been retained with some repairs, since they were largely cut in rock, while all the other parts were replaced. At Rome, Augustus had to virtually rebuild the entire aqueduct system above ground level not long after the death of Agrippa in 12 B.C., despite Agrippa's enormous aqueduct works of B.C. 87. It can also be argued that Lugdunum would have had an aqueduct capable of supplying the town centre before the Claudian period. There is, therefore, some material, historical and comparative evidence for an Augustan Gier aqueduct. In this case, Agrippa, in Gaul in 19 B.C., could have been responsible for its planning, and, conceivably, for setting in hand the preliminary works of survey. The Brévenne aqueduct has several remains over an identified course, but these are less substantial and frequent than those of the Gier 88. No opus reticulatum has been found, but there is brick in coursework which has not been observed on the Yzeron or Mont d'or aqueducts. Consequently, the Brévenne has traditionally been dated later than these. The early 1 st century A.D. dating on the table has been disputed by Jeancolas, who indicated a dating from the Trajanic period, or later 89, but the remains are probably insufficient to reconstruct the history of the aqueduct, which may have been repaired or renewed. The Yzeron aqueduct has not yet been identified over the whole of its course. No tunnels or viaducts have been found. The only substantial As suggested by Wilson, pp.18, 19 and Jeancolas, p.193. Although Février, P.A., Armée et aqueducs, Journées d etudes sur les aqueducs romains, pp did not indicate the work. Burdy, 'Some Directions of Future Research ', p.43 and Jeancolas, pp The aqueduct works of Agrippa in B.C. are noted below in Chapter 11. The comprehensive overhaul of Augustus is known from CIL , also Evans, H.B., Water Distribution in Ancient Rome, Ann Arbor, 1994, pp.49, 138. Burdy, Guide des aqueducs romains de Lyon, pp Jeancolas, pp

64 remaining structure, a double siphon 90, has been dated to the 2 nd century A.D. 91 rather than to the 10 B.C.-1O A.D. dating of the tabulation. There seems to be insufficient material evidence to date the aqueduct as a whole. The Mont d'or aqueduct trace has been recognised over much of its length. It had three or four bridges and two siphons, but little remains of these 92. There seems to have been no driven tunnels, but 'cut and cover' of shallow depth with removable stones covering the channel as substitutes for inspection shafts. Fig shows the rudimentary design and the irregular placing typical of a hurried construction, also indicated by the avoidance of tunnels. No brick or opus reticulatum has been observed on the Mont d'or aqueduct, and the work cannot, therefore, be dated from the materials used. Nevertheless, it is inferior to that on the other aqueducts, and indicates a construction earlier than these. It is possible, therefore, that there was an Augustan Gier aqueduct, and that this was largely replaced by the one which has survived. The earlier aqueduct, perhaps shorter than the later one, could have been planned and partly built in the Agrippan period. There is insufficient material evidence from the Brévenne or Yzeron aqueducts to form a clear idea of the initial works, their repair or replacement. The Mont d'or aqueduct, also, cannot be dated, but the material remains are congruent with work of the Agrippan period, and indications that its construction was hurried point towards it being an early rather than a late aqueduct. (ii) The Aqueduct Functions As can be seen from the table on p.52, the Mont d'or aqueduct could deliver water to a level of 260 m. This was some 30 m below the top of the plateau, but more than 85 m above river level. As can be seen from the 10 m contours on Fig. 2.1, this aqueduct, and the rivers, could supply a large area. The Yzeron aqueduct increased water supply level by only 8 m and did not reach the theatre. The Brévenne aqueduct supply was markedly higher, at 284 m, and could serve the theatre, but only the Gier could supply the praetorium of Agrippa and the top of the plateau. The most satisfactory course of action for water supply would have been to build the Mont d'or Hodge, Roman aqueducts, Fig. 111 and pp.157, 8, described and illustrated the double siphon function, also Burdy, Guide des aqueducs romains de Lyon, photograph, p.74. Jeancolas, p.183. Burdy, Guide des aqueducs, pp.82-89, Jeancolas, pp Hodge, Roman aqueducts, Fig. 48, p.96.

65 and the Gier aqueducts together, and the other two later, since the Mont d'or aqueduct could be built quickly and cheaply, and a simultaneous start on the Gier would have ensured the earliest possible supply to the top of the plateau, and, therefore, to all of it. Such a double action would have imposed the least restraint on urban development because of water shortage, and have increased the status of the town. On the other hand, it was expensive. The construction of the Gier could, therefore, have been deferred until after the Mont d'or had been built and cisterns and wells had become inadequate to service the higher levels. Roman towns regularly depended on both local and distant water supplies, and some never had an aqueduct, relying on cisterns and wells, like some earlier towns 94. More than 50 cisterns from m³ capacity have been discovered on the top of the plateau and many more must have existed 95. The rainfall of Lyon is now 85 cm per year and well distributed over the seasons 96. If this was the same in the Roman period, it would have kept cisterns and wells replenished. Water could also be carried up from the river by the aquarii or water carriers of whom there was no shortage 97. The first public baths, discovered at Lyon at the Rue des Farges site, shown on Fig. 2.7, was at an elevation of 250 m, and could have been fed by the Mont d'or aqueduct, yet it was not built until the time of Claudius, or perhaps Nero 98. This dating indicates that water was used before that period only for domestic and general use. Consumption depended very largely on the ease with which water could be got, with demand increasing sharply once an aqueduct was installed 99. An absolute minimum of 25 litres per day per person has been calculated for undeveloped parts of the world today, and 200 litres at Lugdunum when fully developed, and with all the aqueducts 100. Uncertainties, therefore, remain about the possible population on the Hodge, Roman aqueducts, pp.48, 49, also Crouch, D.P., Water Management in Ancient Greek Cities, Oxford, 1993, for a general account of the use of rainwater in the Greek and Hellenistic East before the Roman period. Burdy, Guide des aqueducs romains de Lyon, p.7. Pearce, E.A. and Smith, C.G., The World Weather Guide, Oxford, 1997, p.357. Juvenal Satires implied that aquarii were the lowest of the slaves and, therefore, that plenty were available. Desbat, A., Les Fouilles de la Rue des Fanges à Lyon, , Lyon, 1984, pp Hodge, Roman aqueducts, pp noted the increase in demand at Pompeii and the influence on the design of houses and creation of gardens. Ibid, p.464, Note 4, using UNESCO figures for the 25 litres and quoting de Montauzan, p.340 ff for the 200 litres.

66 top of the plateau without aqueducts, but it was in the hundreds, and perhaps thousands 101. It would have been feasible to develop a town centre on the top of the plateau with some permanent housing there. The functional analysis indicates that the Mont d'or aqueduct was built first. The Gier aqueduct was not necessarily built only after all the other routes had been used, as stated by Hodge 102. It is more likely that it was built after the Mont d'or aqueduct, but not long delayed, since it alone could supply all the plateau. The Brévenne and the Yzeron aqueducts would have been built to supplement the other two. When the material evidence and functions of the aqueducts are considered together, it can be said only that Agrippa was more probably associated with the Mont d'or aqueduct than with the others. The case for Agrippa planning and building aqueducts in 19 B.C., as argued above, remains unaltered. He may, therefore have been the 'father' of the aqueducts, instigating investigations into all water sources and the prospective aqueduct routes. The decision on the priorities of the various routes would have been taken after this initial survey. It is, therefore, unlikely that Agrippa was responsible for any aqueduct construction. If investigations into aqueducts had been made before 19 B.C. and a first route had been selected, Agrippa would have ordered that the Mont d'or aqueduct be built. (iii) Local Suggestion of Agrippan Aqueduct Construction Desbat suggested possible Agrippan action on the Gier aqueduct only in Annexure 3 to his paper on the supposed Sanctuary of Cybele 103. Desbat observed the imprint in concrete of a pipe serving the praetorium. Since he had discovered no impluvium at the building, he concluded that the pipe brought in water from an aqueduct. With the praetorium construction dated to about 20 B.C., and only the Gier reaching the top of the plateau, he argued that there must have been a Gier aqueduct, and that this was The area of the main plateau is some 600,000m². 85cm of rain falling on this would deposit 510,000 m³ of water in a year, or 510,000,000 litres. This is 1,400,000 litres per day averaged over the year. If all this water were intercepted it would supply 56,000 people at 25 litres per day and 7,000 at 200. An interception of one-tenth would supply between 5,600 and 700 people. The extent of interception would depend on land use. Maximum recovery of water for direct use would res trict agriculture to river level with roofs and paved areas trapping water and leading it to cisterns. Water absorbed could be recovered at least partially from wells. Hodge, Roman aqueducts, p.174 and above Note 74, p.53. Desbat, 'Nouvelles recherches ', pp , as annexure to his paper pp

67 probably Agrippan 104. Yet it is also possible that the pipe supplied water from one of the many cisterns on the top of the plateau 105. The pipe does not, therefore, prove the existence of an aqueduct in the period 20 B.C. to A.D. 10, when the praetorium existed, assuming the absence of an impluvium. If there was an impluvium, the pipe could have come from a cistern to bring in extra water. Another argument advanced by Desbat for a first (Augustan) Gier aqueduct being built before the Mont d'or aqueduct is that only the Gier could supply the top of the plateau 106. Desbat reinforced his hypothesis of Augustan construction by pointing out that the tunnel shafts were spaced strictly in accordance with the precepts of Vitruvius, and that this strongly indicated an early construction 107. There is no doubt that this is the case 108. He further argued that the remaining aqueduct is the Augustan one restored, suggesting that the opus reticulatum in the structures was in use in the Republican period, and that the Oolitic limestone used was typical of early works at Lyon. He also referred to the 'Augustan brickwork' 109. Yet, as argued here, opus reticulatum was essentially an Italian form in the Republic and that used in the structures is probably of the mid-1 st century A.D The Oolitic limestone could have been used in the Claudian period, unless the source was exhausted. Also, the brickwork in the Gier aqueduct structures was not necessarily Augustan 111, and there is no doubt that the above ground structures of the Gier, like those shown on Fig. 2.14, were built with the opus reticulatum and brick (penetrating the whole section of the piers) in the same operation. The evidence, therefore, indicates that the structures existing above ground are from the Claudian period, or later, and are not earlier structures which have been repaired. It is possible that the unexplained Ibid, p.274 and Fig. 32, showing the pipe imprint, also pp.237, 256, 263, for the general chronology of the buildings on the Cybele site. Guide des aqueducs romains de Lyon, p.7, and above Note 95, p.57. Desbat, 'Nouvelles recherches ', p.275. Ibid, p.275, referring to Vitruvius, De Architectura, The spacing of the tunnel shafts was noted by Burdy, Guide des aqueducs romains de lyon, p.22 and Hodge, Roman aqueducts, p.102, to be in accordance with Vitruvius, and unusual in this respect. Desbat, pp.274, 5 and Note 17. See above Note 80, p.54. See above Notes 81, 82, p.54.

68 remains along another trace not far from the existing structures 112 indicate a pre-claudian construction. The structures of this would have been kept in use until the new structures had been completed and could be connected to the tunnels which were to remain, possibly with repairs, indicated by some use of brickwork. The Vitruvian spacing of the tunnel shafts also indicates retention of original tunnels. Much, therefore, remains to be said about the aqueducts of Lyon, and particularly about the history of them as utilitarian constructions which were repaired and restored over many centuries. On present evidence, the first aqueduct at Lyon was the Mont d'or, and this is the only one which Agrippa could have started to build if there had already been investigations into alternative routes. Otherwise, he may have been the 'father' of the aqueducts, and, therefore, of the Gier with the others. Any further connection between Agrippa and the Gier aqueduct has not yet been established, but Desbat has pointed the way to a better assessment of this. CONCLUSIONS Munatius Plancus founded a Roman colony at Lugdunum in 43 B.C. on the top of the main plateau. Agrippa, in B.C., selected Lugdunum as the hub of his road system, building about 17 km of new road to connect his common road to the English Channel and the Rhine to it. His road to the Atlantic was based on a pre-roman road in that direction. Agrippa's road to the south used the existing road to Vienne. Agrippa brought his roads together at the Trion below the top of the plateau. He built a construction camp there, and provided facilities for the operation of the roads as through routes. Agrippa almost certainly built on top of the plateau at the military centre and may have constructed the first permanent buildings at the Verbe-Incarné site, including a first praetorium. He improved the port facilities to handle his supplies brought in by river, and built a road connection to these. In 19 B.C., Agrippa may have been responsible for planning a major expansion of Lugdunum, including water supply by aqueducts. In this case, Agrippa would have been the father of all four aqueducts. If plans for aqueducts had been developed by 19 B.C., Agrippa would have ordered the construction of the Mont d'or aqueduct before that of the others. At the town, Agrippa may have been responsible for planning which would have included a forum and a theatre. No particular building 112 See above Note 86, p.55.

69 can be attributed to Agrippa beyond doubt, but the praetorium at the 'Cybele' site was almost certainly his. He may also have provided the theatre at Lugdunum in addition to that at Augusta Emerita. Agrippa, therefore, did much more at Lugdunum than select it as the hub of his road network and build roads meeting there, as is generally understood. It is probable that Agrippa, rather than Plancus, built the first permanent structure on the plateau at the Verbe-Incarné site. Agrippa s choice of Lugdunum for his road junction was the principal catalyst for the town becoming an important commercial centre, the site of an Imperial Mint, and the virtual capital of the old Gallia Comata, with the Altar to Rome and Augustus. Plancus founded Lugdunum as a garrison town, but Agrippa was the real father of the great city of the Empire, which came into existence despite the problems of water supply. Previously, Agrippa s contribution to the development of Lugdunum seems to have been under-emphasised. This is explained largely by an imputation that works at the urban centre resulted from the visit of Augustus to the West in B.C., and that Agrippa had no part in planning or building at Lugdunum in 19 B.C. The discovery of the praetorium at the 'Temple of Cybele' site with its dating to the period when Agrippa was in Gaul, new uncertainties about the dating of works at the theatre, and new ideas (including those of Desbat) about the early history of Lugdunum, are all catalysts for further enquiry into Agrippa's planning and construction at the town, and for its aqueducts.

70 THE TEMPLE OF VALETUDO AT GLANUM INTRODUCTION Roman Glanum 1, located, as shown on Fig was near a slip road on the Via Domitia and on a road over Les Alpilles to the Via Aurelia, as seen on Fig Fig shows 'Les Antiques', with the arch on the slip road and Glanum proper to the south. Hodge has suggested that Glanum may actually have been at or near Saint-Rémy-de-Provence or on the Via Domitia, with the presently named Les Antiques and Glanum as suburbs to the south 5, but there is no archaeological evidence of this. There was continuous human occupation at Glanum from the Neolithic to the late Iron Age, concentrated in the southern part of the site, where there was a cleft in the hillside, marked aven on Fig. 3.3, and a spring, probably an early place of worship, marked 2 on Fig Finds of pottery and coins in this area indicate that people living there traded with Massilia from the 6 th century B.C. 6. Glanum was, therefore, a typical Celtic centre with a sacred spring located near the ancient road from Italy to Spain. It was important both because of its religious significance, and its position 7. Influence from the Mediterranean through the port of Massilia, founded by Phocaeans from Asia Minor in about 600 B.C. 8 was joined by that brought along the road from north Italy. Glanum of the 2 nd century B.C. had Greek/Hellenistic features, including Delian type houses, porticoes, a bouleuterion, temple and other public Hodge, A. T., Ancient Greek France, London, 1998, p.151, suggested that 'Glanon', on which 'Glanum' is based, was a possible Hellenisation of the aboriginal name 'Glanikon. Gros, P., Hercule à Glanum, Gallia, 52, 1995, pp , Fig. 3, p.319. Roth Congès, A., La fortune éphémère de Glanum : du religieux à l économique, Gallia, 54, 1997, pp , Fig. 1, p.161 (part). Ibid, Fig. 6, p.171 (part), with minor additions. Hodge, Ancient Greek France, p.152. PECS, p.356, also Salviat, F., Glanum et les Antiques (Guides Archéologiques de La France, No. 19), Paris, 1990, pp Salviat, pp.17, 18 for the importance of the position and Hodge, Ancient Greek France, pp for the religious aspect. Ebel, C., Transalpine Gaul, London, 1976, p.10, and Cunliffe, B., The Ancient Celts, Oxford, 1997, pp.48 and 49.

71 buildings. The ancient sacred spring was converted into a nymphaeum and gates and ramparts were built 9. These architectural influences cannot be traced back directly to Massilia, since virtually nothing remains there from this period 10. Two destruction levels at Glanum indicate Roman sackings of the last quarter of the 2 nd century B.C., and there is evidence of minor house building during the first half of the 1 st century B.C., including occupation of one house by Sulla 11. With the fall of Massilia in 49 B.C. and the foundation of Arelate in 46 B.C. 12, Glanum became part of the territory of the Arelate 13 as an oppidum Latium with the ius Latii 14. The major Roman development at Glanum followed the visit of Augustus to the West in 27 B.C. 15. Fig shows the layout of the town of the 2 nd century B.C., and Fig at the end of the Roman period. The Roman works recast the public sector. In the sacred zone south of the gates, the Roman works, as shown on Fig. 3.5, included the Temple of Valetudo. Glanum seems to have been abandoned after Germanic invasions in about A.D. 270, and was gradually covered over by alluvium 18. The monuments at Les Antiques on higher ground remained visible and received their modern name. In late Antiquity, the buildings of Glanum proper were largely demolished for stone recovery, with the parts useless elsewhere being dumped in wells and other convenient places 19. Consequently, at Glanum, the building parts were sorted after demolition, resulting in separation from in situ remains. For the Temple of Valetudo, this dislocation was compounded by the presence of the Temple of Hercules on the other side of the Roth Congès, A., Nouvelles fouilles à Glanum ( )', JRA, 5, 1992, pp.39-55, pp.40-46; also Goudineau, PECS, p.356 and Salviat, pp Roth Congès, 'Nouvelles fouilles ', p.44 and Note 33. Ibid, pp Pliny, 3.36, Suetonius, Tiberius 4.1, Rivet, p.197, also below Chapter 5. p.100. Rivet, p.198. Ibid, p.198 and Note 47, p.208, interpreting Pliny Roth Congès, 'Nouvelles fouilles ', pp Roth Congès, 'La fortune éphémère ', Fig. 9, p.175. Ibid, Fig. 10, p.175. Rivet, p.193, and Notes 55 and 56, p.208; also Goudineau, PECS, p.356. Lequemont, M.F., Lambert, N., Roth Congès, A., Un puit livré de la sculpture. Les Dossiers d archéologie, 140, July-August, 1982, pp

72 Fontaine or nymphaeum which contained a convenient basin for the disposal of surplus material. As can be seen from Fig , the searchers for stone could throw unwanted pieces into the basin from the north, east, or south confusing the excavator centuries later. The position of the Temple of Valetudo on Fig. 3.6 is that of the existing reconstruction, which, as argued below, is not established as that of the original building. The dedication of the Temple to Valetudo, the Roman goddess of health and well being, was related religiously to the nearby ancient sacred area with its nymphaeum. The Altar to Glan, shown on Fig. 3.6, was dedicated to the Glanicae, the local form of the triple Celtic goddesses associated with healing springs 21. One inscription at the altar marked Marcus Licinius Verecundus, veteran of the 21 st legion, seeking a safe return after battle from the Glanicae. This legion was created to reinforce the army of the Rhine after the Varian disaster of A.D The construction of the Temple of Valetudo had probably aided syncretism between the pre-roman and Roman religious observances. INVESTIGATION OF THE TEMPLE It must be said, at once, that the evidence of the material remains is insufficient for a complete description to be given. In fact, questions remain about the location of the temple. These lacunae and uncertainties result principally from the disturbances and partial removal of late antiquity, noted above. It is possible that future archaeological investigations may turn up new evidence, but investigation here is based on the published literature and personal observation of the remains without any disturbance of these. A first section is set aside for a discussion on reconstruction. This section is followed by one on description. An attempt is made in a third section to explain why the temple was built. Finally, there is a general conclusion. The question of dating is taken up as appropriate in the various sections and summarised in the conclusion Roth Congès, 'La fortune éphémère ', Fig. 11, p.176, with minor alterations. Green, M., Celtic Goddesses, New York, 1995, pp Salviat, p.30.

73 RECONSTRUCTION OF THE TEMPLE (a) EXCAVATION AND RECONSTRUCTION : Henri Rolland discovered the scattered remains of parts of the temple in the basin of the nymphaeum, and in the area north of this, and the street. Most of the pieces were non-regular blocks which could not be used elsewhere. His paper reconstruction had the temple facing the nymphaeum with a dead drop to it, and the west cella wall on the street line, as shown on Fig Rolland s location and reconstruction of the cella layout depended on in situ material, shown on Fig. 3.7 in solid black, and dotted lines, and on this material being Roman. Rolland located the positions of the four columns by his discernment of Roman reconstruction of the north wall of the Hellenistic basin at its upper levels 25, where he found that a narrow arch connecting the two walls of the basin had collapsed 26. This arch is seen more clearly on Fig , which shows the Temple of Valetudo in its present reconstruction, as on Fig Rolland's presentation of his proposed layout is unconvincing when judged by modern standards, since he provides insufficient textural and photographic information to adequately describe the in situ remains. There is no doubt that he found the in situ remains of some building, but whether these were of a Hellenistic building or a Roman one is open to question. There could be no access to Rolland s temple from the south or the west, and a side entrance from the east must be regarded as very unconventional for a Roman temple. Rolland concluded that the entrance was from the north, and he noted a pedimental block, which may have indicated a doorway there 28. His solution of a prostyle temple facing south with a door in the back wall is more credible than an amphiprostyle layout with colonnades at both ends 29, since Rolland, M. H., Fouilles de Glanum (1951-2), Gallia 11, 1953, pp.5-17, pp.13, 14, for discovery of the nymphaeum; Rolland, Gallia 12, 1954, pp for clearance of the basin of the nymphaeum, discovery of Hellenistic coins, evidence of Roman repairs and discovery of temple remains; Rolland, Un temple de Valetudo à Glanum, RA, 46, July-Dec 1955, pp.27-53, for a description of the temple and a reconstruction of it. Rolland, 'Un temple de Valetudo ', Fig. 5, p.36. Ibid, p.37 and Note 2. Rolland, 'Fouilles de Glanum (1951-2)', p.14, with Fig. 11 showing the springing points of the arch. Salviat, photograph p.31 (part). Rolland, 'Un temple de Valetudo ', p.39, also below Note 92 and Fig Vitruvius, De architectura,

74 only four columns and capitals have been recovered 30. Nevertheless, his layout was unusual for a Roman temple, which is conventionally placed on a podium approached by steps from the front. Rolland stated in his text that there was a space between the columns of one-and-a-third column diameters, based on this relationship at the Temple of Mars Ultor at Rome 31, resulting in a clear spacing of metres based on the 0.44 metres column diameter at Glanum. This is rather narrow, even for such a small temple. Also, to transpose the spacing/diameter relationship at a large building with thick columns to a small one with thin columns is questionable. A clear spacing of about two column diameters, or 0.88 metres, is shown on Fig. 3.7, and this is what might be expected. Measurement from the bar scale on Fig. 3.7 indicates that the temple was nearly 5.0 metres wide, but, in the text, it was described as 3.90 metres in breadth 32, possibly indicating the interior width. With column spacings of 0.88 metres, the width would be 4.40 metres. Rolland made a partial site reconstruction, shown on Fig This was north and east of the position indicated on Fig. 3.7, but on the same orientation, pending its permanent reconstruction directly over the face of the nymphaeum with the wall of the basin wall repaired. He did not provide the full height of the podium, assessed by him as at least 1.90 metres 34, and also set no upper moulding on the podium. Rolland placed one of the pilasters and its capital to form part of the entrance from the pronaos to the cella. This position seems to be appropriate, since both the pilaster and the capital are carved on two sides, and the other two are dressed back for plastering, or to receive a door frame 35. Rolland s reconstruction of the temple, therefore, cannot be accepted entirely. Nevertheless, much of the uncertainty resulted from the lower archaeological standards of his day and his inability to concentrate on a small building when he was excavating most of Glanum. It is remarkable that he provided so much information Rolland, 'Un temple de Valetudo ', pp.41, 42. Ibid, pp.39, 41. Ibid, p.37 and Note 3. Ibid, Fig. 7, p.40. Ibid, p.37. Ibid, p.37, for Rolland s explanation of this placement of the pilaster, also see below, Note 41, p.68.

75 The fact remains that Rolland was a distinguished archaeologist who also both excavated the temple and described it. His unique testimony is, therefore, to be respected. The most outstanding feature of Roland's reconstruction is its location. The typical Roman temple was approached by steps from a public place. Here it is placed directly against the basin of the nymphaeum. Rolland discovered a small pillar altar inscribed Valetudini which he believed to have stood on the pronaos 36. This would have replaced the conventional altar in front of the temple, which could not exist because of the dead drop into the basin of the nymphaeum. Ritual at the pillar altar would have been visible from the sacred spring and from the Altar to Glan and the Glanicae, shown on Fig It is difficult to imagine a position for a temple more suitable for religious syncretism between the Celtic shrine and the temple. Also, it is more likely that a Hellenistic temple would be built in such a position, which respected the local conditions, than a Roman one which would alter them. Agrippa could, therefore, have modelled his temple on the ruins of a Hellenistic building. His action would then have been an act of reparation for Roman destruction. (b) THE EXISTING RECONSTRUCTION: There seems to be no explanation of the existing reconstruction in the published literature, although doubts have been expressed about its validity 37. The two temple locations shown on Figs. 3.6 and 3.7 are clear of each other, but the steps to the present reconstruction occupy part of Rolland s cella location. The break in the wall on the east of the street shown on Figs. 3.6 and 3.7 coincides with a break in the wall discovered by Rolland 38. The southern part of this break marked the point where Rolland s west wall of his cella, shown in solid black, changed to the dotted lines shown on Fig Rolland discerned the cella platform north of this point, where the present reconstruction would have steps up to the temple. It is not possible to resolve this crucial clash in interpretation from the evidence used here. Rolland did not refer to his discovery of the break in the wall in his proposal for the reconstruction, so that it is not possible to relate the two elements as he understood them, or to know whether they were connected in his mind at all Ibid, pp.51, 52, also see below Note 48, p.70, and below Fig Salviat, p.31, and Roddaz, Marcus Agrippa, p.397, raised doubts about the reconstruction without giving reasons. Rolland, 'Fouilles de Glanum ( )', Fig. 1, p.4.

76 The placing of the existing reconstruction is conventionally Roman, since it faces the public road directly, and is approached from this by steps leading up to a podium sitting on a levelled hillside. The conventional W-E orientation agrees with that of the other two Roman temples, marked XXIV and XXV on Fig Agrippa s temple in this position would be a small prostyle building of the treasury type. As can be seen from Figs. 3.6 and 3.8, the building was placed between the line of the street and the cliff face, which was not dressed back. It was not possible to take levels at the site, and the steps have not been rebuilt. Consequently, it was not easy to obtain a clear idea of both the vertical and horizontal geometry of the layout, but there was little room for a pronaos. This was most unusual in a Roman temple, in which the deep pronaos was an important feature derived from the Etruscan period 39. Figs. 3.10, 3.11 and show, respectively, the front, south side, and south-east corner of the reconstruction. This, like Rolland s preliminary reconstruction, seems to have been placed on a prepared ground without any obvious use of in situ foundations. The column bases are set higher than in Rolland s site reconstruction since they allow for the addition of the upper moulded part of the podium. This part appears only at the S.W. column, with the other two remaining columns built up from sundry blocks, including new rubble walling. It is not at all clear that the S.W. column is supported by a correct reconstruction of the parts in their original positions. The tallest of the three rebuilt columns is incomplete. No column capitals remain on site. One pilaster and its capital have been placed at the S.E. corner. This position does not seem to be suitable, since, as already noted 41, both pieces have the uncarved faces dressed back, suggesting plastering or door frames, which might be expected in an opening between the pronaos and the cella, but not at an exterior corner of the building. The inscription of dedication on a string course originally above the architrave has been placed on blocks where the steps would be. The clear column spacing in the centre is twice the column diameter, and one-and-a-quarter diameters at the sides, as far as can be deduced with the northern column missing. These spacings indicate an adaptation of the eustyle column arrangement of Boëthius, A., Etruscan and Early Roman Architecture, Harmondsworth, 1978, pp.35-48, and Photographs by the writer. See above Note 35, p.66.

77 Vitruvius, appropriate for a small temple with a restricted length 42. The reconstruction agrees with Vitruvian precepts connecting column diameter and building width 43. Examination of the existing reconstruction, therefore, indicates that this resulted from the application of general ideas about Roman temple location and Vitruvian concepts to the arrangement of incomplete remains, but that the result is not entirely convincing. The evidence examined above is insufficient to form a clear opinion about the relative merits of the two reconstructions. It is also possible that the temple was located in some other position north of the nymphaeum, or even to its east or south, especially since, as seen on Figs. 3.6 and 3.8, the rock face was cut back at these positions. The Sanctuary of Hercules is shown as a Roman construction on Fig. 3.5, but Gros has argued that it may have existed at a Greek religious site on the Heraclean route from Italy to Spain, with Glanum as a centre of transhumance 44. This was contested by Roth Congès 45, but neither could rely on up-to-date archaeological investigations at the Sanctuary of Hercules. There is, therefore, no reliable information on what buildings may have stood to the south or east of the nymphaeum before Agrippa built his temple. Consequently, the origins of unwanted stone dumped in the basin of the nymphaeum, as noted above, remain uncertain. Rolland discerned the sanctuary of Hercules to the south of the basin, and Agrippa's temple to the north, with no other building in the area His interpretation may be questioned as a result of future investigations. 2. DESCRIPTION OF THE TEMPLE It is not possible to provide a complete description, since not all the remains have been recovered and there is no established location. The order of description is Vitruvius, De architectura, 3.3.6, recommended for the eustyle arrangement a central intercolumnation of three column diameters and side ones of two-and-a-quarter, but this would make the temple too long to fit in the space available if the length was in proper proportion to the width. Ibid, 3.3.7, specified that the column diameter should be the building width divided by The column diameter of 0.44 metres and the temple width would be 5.6 metres on this basis. This is the approximate width of the temple reconstruction. Gros, Hercule à Glanum, pp Roth Congès, 'La fortune éphémère '.

78 from the foundation to the acroteria, except for the inscription identifying the temple as Agrippan and dedicated to Valetudo, which is considered first. (a) INSCRIPTION OF DEDICATION: Rolland found this in the basin of the nymphaeum 46. Fig shows it as now placed. The beginning of the inscription [VAL]ETUDINI. M. AGRIPPA is confirmed by the inscription VALETUDINI on a small altar shown on Fig recovered by Rolland 49 from the basin of the nymphaeum. The inscription must have continued with L.F., indicating him as the son of Lucius, and by his consular rank. Agrippa was consul first in 37 B.C., and was probably designated as such as early as 40 B.C. He was consul for the third and last time in 27 B.C. 50. In B.C. COS DESIG may have appeared on the inscription, and, in 19 B.C., COS III. Rolland suggested completion with COS II because there were two incomplete inscriptions bearing the name Agrippa at Nîmes interpreted to end in this way, and he considered these, like that at Glanum, to have been found at nymphaea associated with healing 51. The two inscriptions at Nîmes cannot be dated beyond question to 19 B.C., and the buildings on which they were placed have not been identified 52. It is, therefore, not possible to either date the temple or to associate it with the healing process 53 because of the inscriptions at Nîmes. (b) FOUNDATIONS AND BASES: As already noted, some questions remain about the construction of the parts below the level of the column bases and how this is represented by Rolland on paper and in the existing reconstruction. Rolland described the podium as having a minimum height Rolland, 'Un temple de Valetudo ', p.45. Photograph by the writer. Rolland, 'Un temple de Valetudo ', p.51 and Fig. 18. Rolland, Gallia 12, p.450. Ehrenberg & Jones, pp.33 and 35. Rolland, 'Un temple de Valetudo ', p.46 and Notes 1 and 2 referring to CIL and See below Chapter 4, pp, See below Note 112, p.80, for an Agrippan medical condition.

79 of 1.9 metres, but he did not state where this occurred, or indicate the maximum height. He said nothing about the upper moulding of the podium nor did he provide a drawing of it. Rolland recovered eight parts of the lower moulding, including all four corner pieces, but these were insufficient to complete the temple perimeter 54. These lower mouldings can be seen on Figs and They are of a well established profile which was in use from an early period, and are quite similar to those at the late 3 rd century B.C. temple at Gabii 55. All four bases of the columns were recovered, and three remain on site, as seen on Figs. 3.9 and Fig shows the S.W. base, and the lack of plinths or base plates. There is no evidence of stucco being used on the bases except, perhaps, for a finish to the limestone. It was used in the fluting of the columns. The profile of the base is shown as No. 2 on Fig The upper torus or convex moulding of the base is of much the same diameter and thickness as the lower one separated from it by the scotia. The almost equal tori are common in late Republican temples, as at the Temple of Vesta at Tibur (Tivoli) from the early 1 st century B.C. 58, although the bases of the columns in the quadriporticus of Agrippa s theatre at Mérida dated to 16/15 B.C. are of the earlier type 59. Nevertheless, the general trend after the Republic was towards a base in which the lower torus is more prominent, as seen on Fig showing the base (and capital) of the Temple of Castor at Rome dedicated in A.D The other bases shown on Fig are No. 1 for the smaller of the two temples near the forum at Glanum, Nos. 3 and 4 for the mausoleum, and No. 5 for the triumphal arch, both buildings at Les Antiques. The chronological order of the bases shown in Fig. 3.16, judged from the size of the upper torus, is 2, followed by 1 with 3, 4 and 5 later, indicating that the Temple of Valetudo was earlier than the forum temple constructed after the visit of Augustus for the West in 27 B.C. The base and the lower part of the column cut out of the one block is also common in Rolland, 'Un temple de Valetudo ', p.37 and Notes 1 and 2. Boëthius, Fig. 46, p.56. Photograph by the writer. Gros, P., 'Les temples gémines de Glanum'. RANarb, 14, 1981, pp , Fig. 47, p.150. Boëthius, Fig. 154, p.163. See below Fig Sear, F.B., Roman Architecture, Fig. 36, p.66 (parts). Ibid, p.67.

80 Republican temples, but it cannot be used as an indication of dating at Glanum, since this has been adopted at the part reconstruction of the smaller of the two forum temples 62, with the profile No. 1 of Fig Fig shows the reconstruction. (c) COLUMNS: The columns, of mean diameter 0.44 metres, are fluted, with 16 grooves in the shape of a meniscus separated by fillets 64, as against the 24 and the 20 prescribed by Vitruvius for the Ionic and Doric Orders respectively 65. The lesser number is aesthetically appropriate for a small temple, and, also, easier for the mason to provide. It is not possible to know the height of the columns, since there is no recorded complete recovery of any one shaft. Vitruvius specified that the columns of the Ionic Order to be used with the Corinthian Capital 66 are to vary in height from eight to ten times the lower column diameter, depending on the column spacing 67. Since the column spacings at Agrippa s temple are unknown, the shaft height, according to this, could vary from 3.70 to 4.40 metres, assuming no column taper. The columns of the smaller of the two temples near the forum temple shown on Fig are barely seven times the lower column diameter in height. It is, therefore, not possible to assess the length of the column shafts at Agrippa's temple, and, consequently, a drawing showing the elevation of the front of the temple cannot be provided. Each side of the pilaster, shown on Fig. 3.12, has four grooves in the shape of a meniscus separated by fillets. Similar pilasters, but with six grooves, are seen at the Maison des Antes located on Figs. 3.4 and 3.5, but this house cannot be dated 68. It has been argued here that the pilasters at the temple with their capitals framed the entrance to the cella, as proposed by Rolland Varène, P., L apport de l ethno-archéologie à la connaissance des techniques antiques de construction : deux examples tirés de la reconstruction partielle d un temple à Glanum, JRA, 6, 1993, pp , for the explanation of the reconstruction. Photograph by the writer. Rolland, 'Un temple de Valetudo ', Fig. 8, p.41 and p.39, Note 6. Vitruvius, De architectura and Ibid, Ibid, Salviat, pp See above Note 35, p.66, and Rolland, 'Un temple de Valetudo ', p.37.

81 (d) CAPITALS: Only one capital remains on the site. This is the pilaster capital shown on Fig All the four column capitals are in the museum at the Hotel de Sade at Saint-Remy-de-Provence, but it was not possible to identify these or photograph them. Figs and represent the two kinds of column capitals. Fig shows the type with the blossom at the top centre, and Fig that with no blossom, but with tendrils or helices separated by acanthus. Rolland, and Picard concluded that both types came from the same workshop 72. This is indicated by the same astragal decorated with bead and reel, no decoration on the abacus, and a common type of acanthus treatment. Rolland noted the generally squat appearance of the capitals 73, as did Picard, who attributed this to the thick abaci, and, particularly, to the arrangement of the acanthus. He considered that a strong horizontal emphasis was created by the lower ring of leaves and the one above this turning over quite close to each other 74. This can be seen on Fig. 3.20, and to a lesser extent on Fig. 3.21, but not on Fig There is some doubt also about the effect of the abacus, since this is not remarkably thick 75. The squat appearance can be assessed further from Fig This shows the Vitruvian type capital on the left with two others of the late Republic. There is little difference in the thickness of the abaci. All three have the same height, and much the same widths at the astragal and the abacus, yet the Vitruvian capital looks narrower than the other two. This is mainly explained by the three rings of the acanthus 77, which have a stronger vertical emphasis from the more definite vertical connection between three rings. In addition, the larger volutes on the Republican capitals place more emphasis on the top of the bell, and the lower part of this seems to be crushed. If the column capitals of the Temple of Valetudo were shown with the other three, and in the same manner, they would look at least as broad as the Photograph by the writer. Rolland, 'Un temple de Valetudo ', Fig. 10, p.43 and Fig. 12, p.45, respectively. Ibid, p.42 and Picard, G-C., Glanum et les origines de l art romano-provençal. Première partie : Architecture, Gallia, 21, 1963, pp , p.113. Rolland, 'Un temple de Valetudo ', p.42. Picard, p.113. It varies from one-fifth to one-sixth of the total height of the capital. Boëthius, Fig. 148, p.158, referring to Morgan, Vitruvius. The Ten Books of Architecture, Figs. 1, 2 and 4, p.105. Vitruvius, De architectura

82 Republican models because the acanthus division is similar to these, and, also, because of the more prominent horizontal basket work. The pilaster capital is higher from astragal to the top of the bell than it is broad at the astragal, whereas all the other five capitals, shown on Figs. 3.20, 3.21 and 3.22, have equal corresponding dimensions. It would, therefore, be recognised as taller than the other three if represented in the same way with them. Nevertheless, it still looks quite broad, and mainly because of the horizontal lines of the basket work. The capitals of the Temple of Valetudo are, therefore, not of some outlandish form like the very squat capitals of the Temple of Athena Alea at Tegea of the mid-4 th century B.C 78. Also, they have at least some general characteristics of those in late Republican Italy and of the Vitruvian model. They can be placed more securely in the history of the development of the Roman Corinthian capital by comparison of them with the capitals of the developed Corinthian Order, represented by the Temple of Castor at Rome, as seen on Fig The capitals at Rome have a height above the astragal some one-third greater than the diameter of the column immediately below the astragal. All the capitals shown on Fig. 3.22, and at Glanum, exceed this diameter in their heights by no more than one-eighth. This is generally in accordance with the Vitruvian precepts 79. The developed Corinthian Capital is, therefore, more slender than those at Glanum. The capital at Rome has the three rings of acanthus lacking at Glanum. Other differences are the decorated abacus and plain astragal at Rome, while at Glanum the reverse is the case. At Glanum the blossom is large and over the bell and the abacus. At Rome it is smaller and confined to the abacus, and the volutes are under the abacus which is cut off at an angle at the corners. At Glanum the volutes, as shown on Fig. 3.20, rise up into the abacus space and the abacus is not well defined at the corners. At Rome the volutes and the helices seem to spring naturally from the vegetation, but this is not the case at Glanum. Here the basket work is prominent, but it is barely evident at Rome. Apart from these differences the general treatment of the vegetation at the Temple of Castor is represented naturally, in detail, and the modelling is subtle. The vegetation is also more like that of the olive than the acanthus. At Glanum, the rendering of the Lawrence, A.W., Greek Architecture, Yale, 1996, p.144 and Fig Vitruvius, De architectura, , specified that the height of the capital including the abacus should equal the column diameter at its base. Some taper of the column from bottom to top can be expected.

83 acanthus is less natural, with less detail, and the modelling is crude. The capitals at Rome balance nature and symmetry in an elegant execution. At Glanum the work is more primitive. This is not to deny the power and originality of the execution. Here the drill was used extensively, while at Rome the chisel was the principal tool. At the Temple of Castor the overlapping leaves of the acanthus form smooth cavities shaped like pears, but, at Glanum, these are more like triangles with sharp corners. In fact, there is no evidence in Figs. 3.19, 3.20 and 3.21 of the reshaping and development of the Classical Orders which were a major feature of the developed Corinthian Order 80. The capitals at Glanum, therefore, have little in common with those of the developed Corinthian Order but are like those of the late Republic in Italy in respect of their general proportions and the main feature. This similarity does not extend to the treatment of the acanthus. In the late Republic, as can be seen from Fig. 3.22, there were large shaggy leaves amply turned over at their ends. The turning point was rounded so that the ends of the leaves formed a festoon. The subdivision of the leaves was rather indefinite. At the Glanum capitals the acanthus is turned over much less, and the festoon effect is not seen. Also, the leaves are much more split up. The treatment is stiff, whereas it is florid in the Republic. The critical element in understanding the capitals and their dating is the treatment of the acanthus. This is a specialised subject and it is necessary to rely on the researchers expert in this particular field. Picard provided the first study on the capitals of the Temple of Valetudo 81 and Gros made reference to them in his study of the two temples at the Forum 82, but Roth Congès has provided an authoritative account of the capitals of the temple in a paper focussed on the treatment of the acanthus in Provence 83. Her treatment was comprehensive, and it is possible here only to indicate the main features of the Sear, Roman Architecture, p.62. Picard, Glanum et les origines, Note 72. Gros, 'Les temples gémines ', p.146. Roth Congès, A., L acanthe dans le décor architectonique protoaugustéen en Provence, RANarb, 11, 1983, pp

84 conclusions. Her classifications are shown on Fig , starting at the top, and from left to right, with the traditional typology of the acanthus treatment in the late Hellenistic period, the second triumvirate, and the end of the Augustan age. The next lines show, firstly, the interaction between leaves generally, illustrating the space formed by these, secondly, the symmetrical type of the Republic and, thirdly, the asymmetrical type of the Empire. The change between these types is dated by her to the proto-augustan period. The third line shows her classification, in this period, of the various symmetrical types. Roth Congès placed the capitals of the Temple of Valetudo principally in category C. This is characterised by a combination of rounded and pointed leaves with distinctive depressions like drops at the ends of the rounded ones. Roth Congès subdivided this category C further into two types appearing at the bottom of Fig The type on the right was seen by her to be the closest to the capitals at the Temple of Valetudo. She traced the origin of this back to Asia Minor in the first quarter of the 3 rd century B.C. at the capitals of the mausoleum of Belevi, near Ephesus, continuing at the Agora of Smyrna, then, in Egypt, on a capital of Heliopolis Magna towards B.C., and, finally, about 200 B.C. at the Gymnasium of Miletus 85. Roth Congès has, therefore, established a connection between the capitals of the Temple of Valetudo and the region of origin of the founders of Massilia, since this was Phocaea on the coast of Asia Minor. The style spread from Asia Minor to Egypt. Perhaps it extended into North Africa and reached Massilia from there if it did not come directly from Asia Minor. She saw some similarities between the treatment of the acanthus at Agrippa s temple, and that at the Arch of Rimini dated 27 B.C., and, also, at the Arch of Aosta, dated 25 B.C. 86. The findings of Roth Congès were based on a detailed examination of capitals from many places, and they are accepted here as the best explanation for the capitals at Agrippa s temple in relation to other Roman capitals. Roth Congès dated the capitals of the Temple of Valetudo to 27 B.C., based on the material evidence, but she was prepared for them to be a decade earlier than this Ibid, Fig. 1, p.105, Fig. 2, p.106, Fig. 15, p.117 and Fig. 17, p.117 corresponding to the four lines and starting at the top. Ibid, p.119, 121, 2 and Note 90. Ibid, p.111, Note 54.

85 archaeologically if she could accept the idea of Agrippa building a temple at Glanum in B.C. 87. (e) THE STRUCTURE ABOVE THE CAPITALS: Rolland considered that he had found parts of the pediment and the cella cornice. Fig shows the architrave, an irregular pedimental block, and two parts of the cornices. There is no reason to doubt the attribution of the architrave, which is a Roman version of the triple layered Classical form. All the other parts must be regarded with some suspicion, since the mouldings on these are very like that on parts recovered from the Tuscan Hellenistic temple near the dromos, and located on Fig as XVII. There is no decoration, and there are no modillions. The absence of the second is not surprising since these have been noted in stone for the first time at the third rebuilding in 36 B.C. of the Regia in the Via Sacra at Rome., although they were made in stucco early in the 1 st century B.C. 90. The fact that modillions appear at the other two temples of Glanum 91 indicates that these are later than the Temple of Valetudo. The large pedimental block is also difficult to explain as part of Agrippa s temple. Rolland noted the sloping underside, and suggested that this indicated a ramp below, forming a possible entrance to his temple concept, but he seemed to be uncertain of his own idea 92. His reticence must be supported, since, the very low angle of the roof, at barely 15, is to be expected only at the broad cella of the Tuscan temple 93. At all other Roman types, the slope would be at least 20. The column assemblies are quite unlike those of the Tuscan temple, which had plinths, and capitals with a height only half the column diameter. The Tuscan Hellenistic temple near the dromos shown on Fig. 3.4, had mouldings like those on the upper parts recovered by Rolland. It is improbable that remains of this could end up some 150 metres further south in the nymphaeum basin. Some Hellenistic temple or other building with a flatter than normal roof must have stood somewhere near the Ibid, p.111 and Notes 52, 53. Based on Rolland, 'Un temple de Valetudo ', Figs. 6 and 8, pp. 38 and 41 (parts). Roth Congès, 'L'acanthe dans le décor ', p.111, Note 54. Sear, Roman Architecture, pp. 63, 64; and Claridge, A., Rome, Oxford, 1998, pp Gros, 'Les temples gémines ', pp , for his examination of the cornices of the two temples in relation to others in Provence, including the modillions and other decoration. Rolland, 'Un temple de Valetudo ', p.39. Vitruvius, De architectura, 4.7.

86 nymphaeum. Gros has suggested a Hellenistic Sanctuary of Hercules 94. The Hellenistic type of roof parts, is, perhaps, to be explained by use or imitation of the parts of this sanctuary, or of a temple sited as shown on Fig (f) THE ROOF AND ITS DECORATION: Rolland made no mention of any part of the roof of the temple, except acroteria, although he recovered other objects from the basin of the nymphaeum, including coins, votive offerings, etc. 95. The absence of the tiles can almost certainly be accounted for by their removal for use elsewhere in late antiquity. The acroteria would probably have been cast aside like all the other decorated work. Rolland recovered two acroteria from the basin of the nymphaeum 96. That from the pediment angle is shown, as recovered on Fig The apex acroterium was restored, as shown on Fig The angle figure, recovered only to a height of 0.65 metres 99 represents a palmette above vegetable volutes, with acanthus tendrils below. As complete it was probably like that shown on Fig The female bust on Fig is Gaulish, as can be seen from the dress, the head covering, and, above all, from the torc or torque. This ornament has been associated with human men or goddesses in the late La Tène period 100. The face is disfigured, but its carving was probably typical of Entremont 101. This carving style can be seen in the general treatment of the stone on Figs and The treatment of the acanthus on Fig is quite like that on the capitals of Agrippa s temple, and particularly as shown on Fig The acroteria from Agrippa s temple seem to be Roman versions of Greek and Hellenistic models as interpreted by the local craftsmen. There is no obvious reason to think from their design or style that they are from a building other than the Temple of Valetudo. One of similar type has been placed on the temple near the Gros, 'Hercule à Glanum', pp See also above Note 44, p.69. Rolland, Gallia 12, pp. 450, 451. Ibid, p.450. Rolland, 'Un temple de Valetudo ', Fig. 13, p.47. Photograph from the museum at the Hotel de Sade, Saint Rémy -en-provence. Rolland, 'Un temple de Valetudo ', p.48, Note 1. Powell, T.G.E., The Celts, London, 1985, pp.69-73, and Figs Salviat, p.105.

87 forum, as seen on Fig Yet, Salviat stated, without explanation, that the apex acroterium shown on Fig was pre-roman, and did not, therefore, belong to what he saw as the Temple of Agrippa 102. If Salviat dated the apex acroterium as pre-roman from the Gaulish bust, this must be questioned. There was no obvious religious difficulty in placing an acroterium showing a Gaulish figure on a temple with an inscription dedicated to Valetudo. If the figure was Valetudo in Gaulish guise, this symbolised the religious syncretism in the same way as would a figure representing a Gaulish goddess. (g) THE CULT FIGURE: Rolland recovered the lower part of the statue of a female figure in the area of the temple cella, and this is shown on Fig There is no indication of the identity since there is no inscription, but the figure is Roman or Hellenistic, and the long dress is typical of either the Roman Valetudo, or the Greek Hygieia 104, the daughter of Asclepius whose cult at Epidaurus was introduced at Rome in the hope of ending a major epidemic there in the 3 rd century B.C A connection arose between the Greek goddess and the Roman Valetudo 106. Both deities were venerated as bringers of health and were related naturally to the Glanicae of the healing spring. It is possible that the cult statue of Hygieia from a Hellenistic temple was found, and was used in the Temple of Valetudo for practical reasons and as an act of piety and reparation for the Roman destructions. DESCRIPTION : CONCLUSIONS It has not been possible to determine the precise location of the Temple of Valetudo, but this was not far from the nymphaeum, and probably to the north of it. The temple had four columns supporting a pediment, but uncertainties remain about its ground plan. The Roman temple has been partly recovered from the remains up to the level of the architrave. The acroteria has also been recovered, but questions remain about Ibid, p.109. Rolland, 'Un temple de Valetudo ', Fig. 16, p.52. The remaining part of the statue is 0.57 metres high and the figure was, therefore, of natural size for quite a small woman. Ibid, p,53 and Note 1. Livy, 10, Rolland, 'Un temple de Valetudo ', pp.27-35, provided a quite comprehensive account of the cult of Valetudo as recorded in coins, inscriptions and the literature, and of the relationships to Asclepius and Hygieia.

88 the roof. The detailed study of the acanthus by Roth Congès indicated that the capitals were dated between 37 and 27 B.C 107. In a more recent study, the Temple of Valetudo has been dated to the period B.C. in the general chronology of the monuments at Glanum 108. This finding is not universally accepted. Salviat, writing in 1999 in the official government guide of Glanum stated that the Temple of Valetudo was built a little after 20 B.C Levau, writing in 1989, assumed a dating of 19 B.C Neither produced any evidence supporting their datings. The dating of the Temple to B.C. is based on the most comprehensive and latest evidence, and it is accepted here. The building was, therefore, built by Agrippa during or shortly after his first visit to Gaul. 3. EXPLANATION OF THE TEMPLE Agrippa had been sent to Gaul in late 40 B.C. to take control there on behalf of Octavian. His mission was strategic and there is no obvious reason why he should have built a temple in a small place like Glanum. The dedication to Valetudo, the Roman goddess of wellbeing, indicates a possible medical reason. Agrippa suffered from a serious foot condition in later life, and undertook a radical treatment by immersing his legs in hot vinegar 111. He also sought a cure from some ailment at the medical centre of Esclapius at Epidaurus in 16/15 B.C Yet, in B.C., Agrippa was not yet 30 and it is unlikely that he built the temple at that time in the hope of a medical remedy. The explanation for the temple must, therefore, be sought from the location of Glanum, and its particular characteristics in the context of Agrippa's aims at that time. As can be seen from Figs. 3.1 and 3.2, the zone containing Glanum, Arelate, and Tarusco was well placed for mobilisation of forces for Agrippan action in Gallia Comata, since there was direct access up the Rhône valley by land and water, and from the west and east along the Via Domitia. The sea port of Arelate was See above Note 87, p.77. Gross, P., Roth Congès, A., Varène, P., Le centre monumental et gallo-romain, Les Dossiers d archeologique, 140, July-August, 1989, pp.24-33, and plan 26,; also Roth Congès, 'Nouvelles fouilles ', pp.50 and p.51. Salviat, p.30. Leveau, P., L alimentation en eau, Les Dossiers d archéologie, 140, July-August 1989, pp , p.62. Pliny IG, , recorded his treatment at Epidaurus.

89 available for supplies from Italy and elsewhere. There had been a Hellenistic town at Glanum and, though sacked by the Romans, it had been developed for housing 113. It was also a religious centre, and, perhaps, a market place. Gros considered that Glanum was a pre-roman centre of transhumance, with a market for food, fodder and pack animals 114. If so, this activity probably continued into the Agrippan period, making Glanum a natural centre for procurement. Glanum was suitable as an Agrippan command post for mobilisation since it would keep forces away from the colony of Arelate and Agrippa could use one of the large houses there as his praetorium. Agrippa needed substantial military forces for his campaigns and his road works. Roddaz suggested that he brought no troops from Italy. He also noted that the eleven legions of Antony in Gaul had been disbanded, and implied that there could be little if any support from Hispania 115. Agrippa would, therefore, have needed to form a new army, recruiting his officer cadre mainly from southern Gaul. There are strong indications, from the mausoleum of Les Antiques shown on Fig , of a military tradition at Glanum going back to the time of Caesar. This well-preserved monument has an inscription reading Sextus, Lucius, and Marius, sons of Caius of the family of the Iulii honour their grandfather 117. The monument has reliefs of battle scenes on all four faces below the level of the arches. One of these shows a member of the family in battle, as noted on a small inscription at the bottom right corner of the relief. The mausoleum may be from the late 30's to the early 20's B.C., but there is little doubt that the grandfather was involved in the Caesarian wars 118. Construction of the triumphal arch in the late Augustan period 119 very close to the military mausoleum (as seen on Fig. 3.28) also indicates the wish to mark a military tradition, since it could have been built at Tarusco, where it would have been seen from the Via Agrippa as well as from the Via Domitia. There were only four other Roth Congès, 'Nouvelles fouilles ', pp Gros, 'Hercule à Glanum', the abstract p.331. Roddaz, Marcus Agrippa, p.71, Note 217. Salviat, photograph, p.57 CIL SEX.L.M.IULEI.C.PARENTIBUS.SUEIS. Gros, P., Le Mausolée des Iulii et le status de Glanum, RA, 1, 1986, pp.65-80, and Salviat, pp Salviat, pp

90 triumphal arches in Gaul 120 and the decision to build one of these at a small oppidum latium seems to suggest its military rather than its civil status. As already noted, a Latin inscription of the late Augustan period at the Altar to Glan was placed by a Roman soldier 121. In addition, there is some evidence of a military contribution from Glanum indicated by a statute of Hercules and seven inscriptions at the Sanctuary of Hercules excavated by Rolland 122. There is, therefore, reason to suggest that Agrippa would have formed his first headquarters in Gaul at Glanum where he could prepare for his operations in Gallia Comata. It was essential for him to achieve this mobilisation before he embarked on his activities, since he could not expect to live off the land as Caesar had done in his conquest. This would have caused resentment, and frustrated his wish to control Gallia Comata and keep it quiet while Octavian was concerned elsewhere. Also, Lugdunum, the hub of his road system, was more than 200 kms from Arelate, and at the extreme S.E. edge of Gallia Comata in which Agrippa was to build his roads to the West, the Rhine and the English Channel. Agrippa would have built up substantial forces and supplies in southern Gaul before undertaking his military operations, military re-organisation, and large scale road works in Gallia Comata. Agrippa's use of Glanum as his mobilisation centre would have stimulated activity there, and the local elite could have advanced themselves by supplying provisions and by contact with the virtual governor of Gaul. Glanum was already a religious centre, but there is no evidence of a Roman temple before 39 B.C. The provision of a Roman temple would have increased the importance of the town, bringing new business, which would continue after Agrippa had left. A temple could function as a treasury, and as a bank, charging a higher than normal rate of interest because of the security of the sacred site and the stone walls 123. There is some indication of a connection between the merchant classes and the Celtic cult of the mother goddesses in Germania in the Ibid, p.58 for triumphal arches at Orange, Avignon, Carpentras and Cavaillon. Salviat, p.30 and above Note 22, p.64. Rolland, 'Fouilles de Glanum ', pp.15 and 17 interpreted the statue and the inscriptions, Figs. 12, 13 and 14, as clear proof of a disproportionately large military contribution from Glanum. Roth Congès, 'La fortune éphémère ', pp disputed this interpretation. Henig, M., Religion in Roman Britain, London, 1984, pp

91 Empire 124, which may have applied to Glanum. Construction of a temple would have advanced the cause of Agrippa by marking Roman reparation for the sackings of Glanum and by relating the Roman religion to that of the indigenous soldiers to be recruited. This religious connection would have helped him in Gallia Comata as a whole, since the Glanicae were local variations of the goddesses worshipped throughout that region. To build the temple at Glanum was, therefore, in the interests of the Roman power and the local elite. It is most unlikely that the Temple of Valetudo was votive, or promised by Agrippa before a military campaign 125. This would have been out of keeping with Agrippa s consistent avoidance of public honours, and of less use as a rallying point for recruitment than an immediate construction. The hypothesis of a 'mobilisation' building explains why it was built at least ten years before other Roman public buildings at Glanum, construction of which resulted from the visit of Augustus to the West in 27 B.C. CONCLUSIONS The Temple of Valetudo was built by Agrippa early in the period B.C. as a 'mobilisation' building intended to advance Agrippa's strategic aims in Gaul. It is possible that it was a partial reconstruction of a Hellenistic temple sacked by the Romans. The building was near the sacred spring of the local Glanicae, but its precise location, layout and structural form have not yet been determined. The temple is of great architectural interest as an early Roman building in southern Gaul, and, particularly, for the study of the Corinthian capital in the region. The design of the capitals of the temple can be traced back to Greece and Asia Minor and the founders of Massilia. The carving of the capitals provides a good example of local interpretation. The religious aspects of the Temple of Valetudo have been mentioned here only as they relate to its location, layout, or the reasons for its construction. Rolland investigated the introduction of the cult of Asclepius into Italy and the connection which was formed between his daughter Hygieia, and the Roman Valetudo. He also made some comments about the cult of Valetudo at Glanum. Juno, the Bona Dea, Demeter, and perhaps Hygieia, Tyche, Cybele and Isis all appear at Glanum with the Glanicae 126. Glanum, therefore, offers a rich field for investigation of feminine deities in southern France Webster, G., The British Celts and their gods under Rome, London, 1986, p.111. Orlin, E.M., Temples, Religions and Politics in the Roman Republic, Leiden, 1997, p.4, for votive temples. Salviat, pp

92 NEMAUSUS (NÎMES) The general location of Nîmes is shown on Fig with the course of its aqueduct, which included the Pont du Gard viaduct. Fig marks the principal Roman urban constructions identified to date, and the circuit of the walls. Agrippa was generally regarded as the builder of the aqueduct until quite recently, and as responsible for the Maison Carrée until some two decades ago. The prominence of the two sons of Agrippa, Gaius and Lucius, has been seen as evidence of Agrippan activity there, yet, this is in doubt. It is, therefore, necessary to clear the ground for investigation by first examining the evidence of the two buildings and of Gaius and Lucius. THE AQUEDUCT AND THE MAISON CARRÉE The aqueduct was attributed to Agrippa by Espérandieu in , and this view was generally supported until quite recently. Hodge perpetuated the idea, although only as a passing comment on an engineering problem 4. Wilson, focussing on dating, pointed out that the Agrippan attribution was based on an Agrippan Maison Carrée and Agrippa, the builder of aqueducts at Rome, being in Gaul in 19 B.C. 5. The Agrippan attribution must now be abandoned 6. The reasons are set out concisely in the abridged version in English 7. These include: the existence in the aqueduct structure of ceramic fragments which could not be earlier than about A.D. 40, the tunnel for the aqueduct which was driven under the existing town wall (started in 16/15 B.C.), and the conversion into a channel of the stream from La Fontaine between A.D. 20 and Nemausus had ample water resources from La Fontaine and the groundwater under the town and the aqueduct was built for irrigation and Rivet, A.L.F., Gallia Narbonensis, London, 1988, Fig. 18, p.166 (modified). Ibid, Fig. 17, p.164. Espérandieu, E., Le Pont du Gard et l'aqueduc de Nîmes, Paris, Hodge, A. T., Roman Aqueducts and Water Supply, 2 nd Ed., London, 2002, p.184 Wilson, 'Tot aquarum tam multis...', p.13. Fabre-G., Luc Fiches-J. & Louis Paillet-J., L'Aqueduc de Nîmes et le Pont du Gard Archéologie, géosystème et histoire, Nîmes, 1991, p.14. Fabre-G., Luc Fiches-J. & Louis Paillet-J., 'Interdisciplinary research on the aqueduct of Nîmes and the Pont du Gard', JRA, 4, 1991, pp Ibid, p.72.

93 extended to Nemausus for its prestige 9. Agrippa could not, therefore, have been responsible for any works at Nemausus associated with the aqueduct, as has sometimes been stated 10 or implied 11. The Maison Carrée, thought for many years to be Agrippan and dated to 16 B.C., was re-dated to A.D. 2-3 in a new study by Amy and Gros published in In this, the inscription of dedication in bronze letters was interpreted as naming Gaius and Lucius, and the second line of this was dated to A.D. 4-5, or after the deaths of the two. A dating of A.D. 2-3 for the buildings seems to have been accepted in the literature, but the dating of the inscription has been disputed 12. Amy and Gros reported the discovery of an earlier building to the west of, and partly under, the Maison Carrée, with finds of pottery and coins indicating construction of the late Republic 13. More recent investigations have uncovered quite extensive development in the area which might be dated to the early 1 st century B.C. 14. THE PROMINENCE OF GAIUS AND LUCIUS Gaius was patron of Nemausus 15 and, as already noted, the Maison Carrée was dedicated to Gaius and Lucius. Both these sons of Agrippa were, therefore, important at the town. Roddaz assumed that this prominence indicated the prominence of Agrippa even though he knew that Agrippa did not build the Maison Carrée, and was not convinced that he built the aqueduct. He also assumed that Agrippa and Augustus were patrons of the town, and that Gaius became patron since his father had been patron 16. There is no proof that either Augustus or Agrippa was patron of Nemausus, although this was probably the case. Also, it Ibid, pp.72, 73. Balty, J.C., COL(onia) NEM(ausus). Notes 'd'archéologie et d'histoire augustéennes', RBPhil, 1960, pp.59-73, pp Roddaz, Marcus Agrippa, pp Amy, R., and Gros, P., 'La Maison Carrée de Nîmes', Gallia, Supplément 38, 1979, for the datings to A.D. 2-3 and A.D. 4-5, and Christol, M. and Goudineau, C., 'Nîmes et les Volques Arécomiques au 1 er siècle avant J-C', Gallia, 45, 1987, pp , p.92 and Note 19 for the dispute about the dating of the inscription. Amy & Gros, pp See below Note 33, p.88. CIL Roddaz, Marcus Agrippa, pp , for his general comments on Agrippa (and Augustus) at Nemausus.

94 does not necessarily follow that Gaius received the status of patronus because his father had been patronus, since, as the grandson of Augustus, he could have been the patron, even if his father had not been Agrippa. Roddaz suggested that the dedication of the Maison Carrée to Gaius and Lucius indicated the regard of the town for the family from which it has received benefits 17. Yet the Maison Carrée was, perhaps, dedicated to the 'young princes' Gaius and Lucius to demonstrate the greater importance of the temple of the Imperial Cult than the ancient sacred fountain of Nemausus 18. There is, therefore, no automatic connection between the patronus of Gaius and the dedication of the Maison Carrée to Gaius and Lucius, and any Agrippan works at Nemausus. The other imputed connection between Agrippa and Nemausus through Gaius and Lucius arises from a text of Suetonius 19. This described Tiberius in the East (in 6 B.C. to A.D. 2) wearing Greek garb and growing daily more despised until the people of Nemausus were encouraged to overthrow his statues and busts. Suetonius also referred to rivalry between Gaius and Tiberius. This text has been interpreted as indicating support at Nemausus for the clan of Agrippa represented by his sons Gaius and Lucius, and the corresponding unpopularity of the clan of Livia represented by her son Tiberius 20. The destruction of the images has also been accounted for by the fact that Tiberius was the son of Tiberius Claudius Nero, and the assumption that the latter had granted colonial status to Nemausus in 45/44 B.C. In this case, the people of Nemausus were objecting to the disgrace brought upon the town by the scandalous conduct of the founder's son 21. It is argued below that Agrippa may have granted the colonial status to Nemausus in B.C. rather than Tiberius Claudius Nero in B.C. 22. In this case the destruction of the images of Tiberius was a reaction in favour of the clan of Agrippa. Nevertheless, since the case for Agrippa taking this Ibid, p.400. Gros, P., 'Nouveau paysage urbain et cultes dynatiques : remarques sur l'idéologie de la ville augustéenne à partir des centres monumentaux d'athènes, Thasos, Arles et Nîmes', Goudineau, C. & Rebourg, A., Eds., Les Villes augustéennes de Gaule, pp , at p.139 and Note 62. Suetonius, Tiberius, 13. Gros, P., 'L'augusteum de Nîmes', RANarb, 17, 1984, pp , p.134 and Note 90. Christol, M. and Goudineau, C., 'Nîmes et les Volques Arécomiques au 1 er siècle avont J-C', p.92 referring to the grant of colonial status to Nemausus by Tiberius Claudius Nero in B.C., as argued by them. See below Notes 31-36, p.88, 89.

95 action in B.C, cannot be proved, the interpretation of the text of Suetonius is problematic. INVESTIGATION OF AGRIPPA AT NEMAUSUS Agrippa has traditionally been seen to have acted at Nemausus only in 19 B.C. when he was assumed to have set in train the construction of the aqueduct and the Maison Carrée. His Temple of Valetudo at Glanum was dated to 19 B.C. until very recently, but it is now generally dated to B.C. 23. There is, therefore, reason now to consider Agrippan construction at Nemausus in B.C. This is investigated in a first section, followed by a second section where Agrippan construction of 19 B.C. is investigated. Two inscriptions bearing the name of Agrippa from unknown buildings at Nemausus cannot be dated. These are, therefore, examined in a third section. (1) AGRIPPA AT NEMAUSUS B.C. Nemausus was on the southern edge of the high country to the north, and on the ancient W-E route which became the Via Domitia. Settlements on Mont Cavalier and at the sacred fountain of Nemausus, both shown on Fig. 4.2, had developed into a typical agglomeration of the district penetrated by Greek and Hellenistic influence from Massilia before Roman control was established in about 120 B.C. 24. The site then probably became a changing station on the Via Domitia 25. The triumph of Caesar over Pompey, who had favoured the Massilians in their struggle with the local Volques Arécomiques of Nemausus, must have resulted in substantial independence for Nemausus and other centres of the Volcae 26. Caesar's general grant of the Latin right to communities in Gallia Transalpina 27 was reflected in the creation of a Latin colony at Nemausus, generally attributed to Tiberius Claudius Nero in B.C. This interpretation relies on extrapolation from a text of Suetonius noting his See above Chapter 3, pp Py, M., 'Recherches sur Nîmes préromaine habitats et sépultures', Gallia Supplément 41, 1981, pp , for an account of development from the Bronze Age to the beginning of the 1 st century B.C., also Célie, M., et al, 'Enceintes et développement urbain : Nîmes, antique des origines au I er S, ap. J-C', JRA, 5, 1994, pp , pp Rivet, p.43. Goudineau, C., 'Le Statut de Nîmes et des Volques Arécomiques', RANarb, 9, 1976, pp Sherwin-White, A.N., The Roman Citizenship, Oxford, 1937, p.176, and Brunt, P.A., Italian Manpower, Oxford, 1971, p.239.

96 foundation of Roman colonies at Narbo and Arelate at that time 28. Yet coins with the legend NEM COL indicating the change of status, have been dated no earlier than about 40 B.C. 29, and as late as 20 B.C. 30. This evidence points towards Agrippa in Gaul in B.C., being responsible for issuing the coins rather than Tiberius Claudius Nero in 46/45 B.C., and the coins themselves neither name nor show any person. In this case, Agrippa may have been responsible for creating the Latin colony of Nemausus rather than Tiberius Claudius Nero. Agrippa also had something to gain from such an action. Nemausus had 32 places subject to it extending over a wide area 31. As such an important centre on the Via Domitia, Nemausus would have been able to provide him with supplies. He could also have drawn recruits for his military operations from the town and the surrounding areas for which it was the centre as a substantial urban site. As can be seen from Fig , there is no doubt about the extent of the town of the 3 rd century B.C. The development after this period and before the first visit of Agrippa in B.C. is less certain, but the area shown hatched on Fig. 4.3 was at least partly built over by this time 33. There is no archaeological evidence of a deductio by Tiberius Claudius Nero in 45/44 B.C. and the largest urban expansion seems to have occurred between 100 and 75 B.C. 34. There is also no evidence of any urban change in B.C., but this should not be expected, since Agrippa was recruiting forces for his operations in Gaul 35 and was not discharging veterans, or encouraging other settlement. The general evidence of urban development is, therefore, more in agreement with Agrippa at Nemausus than with Tiberius Claudius Nero Suetonius Tiberius 4.1, as interpreted by Christol and Goudineau, 'Nîmes et les Volques Arécomiques ', p.92 and Notes 16-18, also Burnett et al, p Burnett et al, pp.152, 3. Christol and Goudineau, pp.95 and 96 and the lengthy Note 38 described several interpretations of the dating for the first NEM COL coins. Strabo, , and Pliny, 3.37, described the 32 places subject to Nemausus, also Rivet, pp , and Christol and Goudineau, pp.98, 9. After Célie et al, Fig. 1, p.384. Célie et al, pp The analysis refers to Benoit, J. 'Nîmes : Études sur l'urbanis me antique. Problèmes de méthode et resultats', Bulletin de l'ecole antique de Nîmes, 16 (Nîmes, 1981), pp Ibid, pp.388, 9, referred to major agricultural, irrigation and defensive changes in the period B.C. and suggested that the Christol-Goudineau hypothesis of action by Tiberius Claudius Nero should be examined in the light of this evidence. Roddaz, Marcus Agrippa, p.71, Note 217, also see above, Chapter 3, p.81.

97 Furthermore, since the town was already well developed, and capable of providing its own needs, Agrippa would have obtained the support of the local elite most effectively by making the town a Latin colony, and, therefore, conferring Roman citizenship on the leading people 36. There is, therefore, a better case for Agrippa creating a colonia Latina at Nemausus in B.C. than for Tiberius Claudius Nero doing this in B.C. and an argument for Agrippa doing this for his mobilisation. At Glanum, Agrippa could obtain support by building a new temple at a religious site, and there was no possibility of a colonial status, since Glanum was controlled by Arelate 37. At Nemausus, the town capable of building its facilities would be made co-operative by the grant of increased status. There is, therefore, little reason to think that Agrippa built anything of significance at Nemausus in B.C. Also, there seems to be no evidence of Roman construction at that time. The few public buildings which have been identified indicate a local interpretation of Hellenistic building ideas. This is seen at a large portico which has been excavated immediately S.W. of the 'Temple of Diana'. Construction of this began at the beginning of the 1 st century B.C. and continued until B.C. 38. (2) AGRIPPA AT NEMAUSUS 19 B.C. Agrippa has traditionally been seen as important at Nemausus because of coins showing him with Augustus. Fig conveniently illustrates the principal features of the 'crocodile' coins of Nemausus. The crocodile has been universally recognised as representing Egypt conquered, and there is no doubt that these coins are post-actium. The obverse has the heads of Augustus and Agrippa with various crowns and the legends IMP and DIVI.F. On the reverse, a crocodile is chained to a palm, and there is a wreath above. The legend COL NEM is split by the palm 40. The status of the colony remained that of a colonia Latina, although the COL NEM replaced the earlier NEM COL. The crocodile coins were struck until A.D. 14 in Sherwin-White, p.177 and Note 2. See above Chapter 3, Notes 13 and 14, p.63. Guillet, E., et al, 'Un monument à portique tardo hellénistique près de la source de La Fontaine à Nîmes (Gard)', Documents d'archéologie Méridionale, 1992, 15, pp Seaby, H.A., Roman Coins and Their Values, London, 1954, Figure p.33, (enlarged). RIC, pp.51, 52.

98 three main groups 41, but the date of the first strikings has not yet been established beyond doubt. The most commonly accepted date is about 27 B.C. 42, but 20 B.C. 43, and B.C. 44 have also been suggested. The 'crocodile' coins have been found in abundance over most of the Roman West 45. It is, therefore, possible that the coins were designed to function as propaganda in the West, and not to reflect the circumstances of Nemausus. First issues in 27 B.C. coincided with the presence of Augustus in Gaul and his first census there, together with the initiation of the three provinces of Northern Gaul 46. Inscriptions dated 25 B.C. marked an initiative by Augustus at the sacred fountain 47. First issues in 16/15 B.C. coincided with the visit of Augustus to Gaul 48, and an inscription of 16/15 B.C. marking the start of works on the town walls by him 49. Agrippa was in Gaul neither in 27 B.C. nor in 16/15 B.C, and could not, therefore, build anything, or strike coins. Coins first struck in 19 B.C. would indicate striking by Agrippa and current action by him, since Augustus is not known to have been in Gaul at that time. Evidence of a first striking of 19 B.C. rather than in 27 or 16/15 B.C. is, therefore, needed to impute a dominant role for Agrippa from the coins. Yet the weight of the numismatic opinion indicates a first striking in about 27 B.C., and the dating of 20 B.C. is based only on hoard-evidence 50. Also, a first striking in 27 B.C. is congruent with the coins themselves, with Augustus and Agrippa shown naturally as the victors in the aftermath of Actium and the operations in Egypt. Any strikings significantly later than this are harder to explain, although this has been attempted 51. Furthermore, the IMP on the obverse is perhaps to be Ibid, p.27 and Burnett et al, pp Burnett et al, p.153, Roddaz, Marcus Agrippa, pp.595-6, Grant FITA., pp RIC, p.27 and Note 5. Christol and Goudineau, pp As printed out first by Grant, FITA., pp Cassius Dio, See below Note 60, p.92. Cassius Dio, See below Note 54, p.91. RIC, p.27. Christol and Goudineau, p.102, explained the joint appearance of Augustus and Agrippa on coins first struck in 16/15 B.C. as indicating their joint powers of the tribunicia potestas.

99 expected more in about 27 B.C. near the dawning of the Augustan principate than in 16/15 B.C. The 'crocodile' coins are, therefore, to be explained as commemorations of Actium for general distribution in the Roman West. It is possible that there was some settlement of veterans at Nemausus, but there is no automatic indication of Agrippan activity or construction at the town, as is commonly supposed. The traditional view on Augustan development at Nemausus is that Augustus and Agrippa were responsible for a radical re-organisation of a town which had limited development outside the area of the sacred fountain of Nemausus 52. Roddaz considered that Augustus and Agrippa re-organised Nemausus, although he was reticent about the extent of the changes 53. As can be seen from Fig. 4.3, there was substantial pre-roman urban development south and east of the sacred fountain, and into the zone where the Maison Carrée was built in the first years of the Christian Era. There is no doubt that Augustus initiated works at his Augusteum at the sacred fountain, and, also, at the town walls and gates 54, and the Maison Carrée was a Roman initiative. Nevertheless, it is unlikely that Augustus or Agrippa would have initiated or paid for works needed for the general development of a prosperous town. Also, Garmy has indicated the uncertainty about the general urban development in the Augustan period 55. The general location of the sacred fountain of Nemausus is marked on Fig Fig shows the layout of the Roman Augusteum. Fig provides an impression of its remains looking from the east, with the Temple of Diana in the background. These two figures show the 18 th century remodelling of the Roman Balty, COL(onia) NEM(ausus), p.62 ff. Roddaz, Marcus Agrippa, p.399. See below Note 60, p.92, for the inscriptions recording the initiative of Augustus at the Augusteum, and CIL , for the inscription on the Porta Augusta dated 25 B.C. noting the initiative of Augustus on the town walls and gates. Garmy, P., 'Quelques réflexions sur la romainité augustéenne Nîmoise à la lumière des fouilles recentes', Les Villes augustéennes de Gaule, pp.41-44, also Célie et al, pp Gros, P., 'L'Augusteum de Nîmes', RANarb, 17, 1984, pp , Fig p.126 based on Naumann, R., Der Quellbezirk von Nîmes, Berlin and Leipzig, 1937, Plate 2. Roth-Congès, A., and Gros, P., 'Le Sanctuaire des eaux à Nîmes', Revue Archéologique du Centre de La France, 22, 1983, pp , Fig. p.137, based on Naumann, R., Der Quellbezirk von Nîmes, Plate 26.

100 works 58. The theatre, marked on Fig. 4.5, was partly visible before the 18 th century remodelling, but new steps up to the Tour Magne obscured it and it has only been partly uncovered since that time 59. As can be seen from Fig. 4.6, the main concentration of the remains are at the nymphaeum and the 'Temple of Diana', with little left of the triple portico, and only the foundations of the South Temple. Two identical inscriptions dated 25 B.C. and indicating dedication to Augustus were found near the sacred fountain, cut to suit the shape of the two exedrae which remain to the south of the basin 60. These inscriptions almost certainly indicated the start of the works 61, resulting from the visit of Augustus to the West in 27 B.C. The only parts of the complex which can be dated confidently to the Augustan period from the material remains are portions of the nymphaeum and its central platform, excluding the altar itself, which has not survived. This dating is based on dating of mouldings and decoration of the platform and of the bases and capitals of the free-standing columns at its corners, shown on Fig The capitals of these columns have been dated to 10 B.C., or somewhat later. All the other moulded or decorated parts date from about B.C. Agrippa, therefore, could have advanced progress on works at the nymphaeum in 19 B.C. 62. It is possible that the whole scheme was conceived from the beginning, but there seems to be no information on the dating of the porticoes, since virtually nothing remains of these, although the South Temple on the southern portico is generally regarded as post-augustan 63. Gros argued that the original layout of the 'Temple of Diana', which contains work from many periods, is Augustan 64, but this cannot be established beyond doubt, and this dating could be as late as A.D. 14. There is, therefore, no indication that Agrippa may have advanced construction other than at the nymphaeum, which can be dated to the general period of his second visit to Gaul Rivet, p.168 and Note 22, p.177, for a brief account of the remodelling. Ibid, p.165 and Note 23, p.177, also Notes CIL 12, 3148, 9, can be dated to 25 B.C. from known datings of the ninth consulate and the tenth designated consulate of Augustus marked on them. Gros, L'Augusteum, p.129. Roth, Congès and Gros, 'Le sanctuare', pp.132, 138, 140, 142, who provided illustrations of the frieze of the platform, p.139 and the capitals, p.146. Gros, L'Augusteum, p.128 and Notes Ibid, p.128 and Notes

101 Some kind of pre-roman altar, dedicated to the deity of the sacred fountain, must have existed at or near the basin, although no trace of such a structure has been discovered. The remaining features from the ancient world are confined to the Roman inscriptions at the two exedrae, parts of the steps there, and the square base near these shown on Fig The masonry of this base has mortared joints indicating Roman work as distinct from local techniques using dry joints. Yet, it is devoid of features showing what stood on it, and it cannot be dated within the Roman period. Grenier suggested that the square base indicated a square Celtic temple of the type seen in parts of Gaul, but it is now thought that there was probably a central altar with statues 65 or columns at the corners. In this case, the small base with the altar and the columns was a smaller version of the platform of the nymphaeum with its altar to Rome and Augustus 66. Such a combination would have symbolised the absorption of the cult of Nemausus by the cult of Rome and Augustus, with both the small and the large base as part of the Roman works started in 25 B.C. Another explanation is that the base, the altar and the statues (or columns) were built by Agrippa in B.C. when he constructed the Temple of Valetudo at Glanum near the sacred spring there 67. An Agrippan base with its altar would have been retained as part of the new Roman works started in 25 B.C., and, perhaps used as a model for the platform of the nymphaeum. The first alternative of both bases being built together provides the simplest explanation of both the square constructions, and is adopted here. Also, it has been argued above that Agrippa acted politically in B.C., rather than by works. Construction of the small base and the altar by Agrippa in 19 B.C. can be ruled out since Agrippa would have advanced the work started by Augustus in 25 B.C. rather than build something new. In the Iberian Peninsula in B.C., Agrippa was responsible for providing the theatre at Augusta Emerita 68. He could have built a theatre at Nemausus when in Gaul in 19 B.C., introducing a Roman cultural feature at a religious complex. As already noted, there is virtually no exposed material at the theatre and investigations Ibid, p.128 and Note 52 convincingly ruled out the idea of a square Celtic temple, with a square cella and a verandah, as advanced by Grenier, Manuel d'archéologie gallo-romain, Paris, 1960, p.496. Ibid, p.127 for the dedication of the altar of the nymphaeum to Rome and Augustus. See above Chapter 3, pp.79, 80. See below Chapter 8, pp

102 have not been extensive or sustained. It is known that the orchestra was some 24 m in diameter. There is evidence only of nine rows of seats in four cunei (or divisions) of the cavea (auditorium), and the diameter of the cavea of 38 m, discerned from the remains, may originally have been 85 m 69. This latter dimension indicates quite a large theatre of the conventional Roman type and is compatible with the diameter of the orchestra. Yet the theatre could have been of the Gallo-Roman type where there was no standard design, or fixed relationship between the size of the orchestra and the cavea 70. The only inscriptions discovered at the theatre are of the 2 nd century A.D. 71, perhaps indicating that there was no theatre at the Augusteum until that time. It is not, therefore, possible to suggest that Agrippa built a theatre near the Augusteum. No theatre of the Roman period has yet been discovered in the town. There is, therefore, some evidence that Agrippa may have advanced work at the nymphaeum of the Augusteum put in hand by Augustus in 26 B.C., but no indication of any other Agrippan works of 19 B.C. Agrippa was in Gaul for only a few months 72. He put down disturbances in Gaul 73, and probably built a praetorium at Lyon 74. He would, therefore, have had little time to linger at Nemausus, assuming that he passed through there on his way to put down the serious rebellion in N.W. Spain, which was of great political importance to Augustus 75. The evidence does not, therefore, indicate any significant Agrippan works in 19 B.C., despite previous assumptions of a major Agrippan role. (3) THE INSCRIPTIONS BEARING THE NAME OF AGRIPPA As noted above, these were not investigated as works of B.C., or 19 B.C. since they cannot be dated. The inscriptions CIL and 4, discovered in the 18 th century, have not been assigned to any particular structures by the local Information kindly provide by Professor F. Sear. Bromwich, J., The Roman Remains of Northern and Eastern France, London, 2003, Fig. 20, p.116 and Fig. 42, p.211, indicated the large variations in the relationship between the size of the orchestra and the cavea in the Gallo-Roman theatre. Gros, 'L'Augusteum', pp.130, 131. See above Note 14, p.12. Cassius Dio, See above Chapter 2, pp See below Chapter 6, Notes 3-8, pp

103 archaeologists. It was not possible for the writer to examine the inscriptions, or to obtain photographs of them from the Musée Archéologique at Nîmes which has them in storage. The incomplete CIL : M. AGRIPPA.L.F.C. survives as a single fragment 1.64m long, 0.27m high with a moulding below. The letters are 0.15m high. The inscription was discovered in 1742 in the cellar of a house in the present Rue Thédenat (near the Lycée A. Daudet in the general location of the 'baths' marked on Fig. 4.2 west of the amphitheatre). The incomplete CIL : M. AGRIPP survives in two adjacent fragments making up a length of 1.35m with a height of 0.43m and a moulding above. The letters are 0.20m high. CIL was discovered in 1740 in the heaped up debris of the Roman remains to the south of the sacred fountain of Nemausus 76. CIL can be reconstructed: M(arcus).AGRIPPA.L(ucius).F(ilius).C(os). The final surviving designation could be completed COS III, indicating a date of 19 B.C., following the third Consulate of Agrippa in 27 B.C., or COS DESIG, referring to Agrippa's designation as consul for the first time in 40 B.C. 77. Completion of the inscriptions by COS III DAT 78 or COS III FECIT 79 are, therefore, not necessarily correct. It is conceivable that the M.AGRIPPA was preceded by other letters, as in the case of the Temple of Valetudo at Glanum with its inscription VALETUDINI.M.AGRIPPA 80. It is also possible that the inscription does not indicate any action by Agrippa, but was on a structure where Agrippa was named with other members of the Imperial family. In this case, the inscription was almost certainly later than 19 B.C., but probably before the death of Agrippa in 12 B.C. The much less complete CIL may have been on the same building as CIL , and, therefore, the same as it, but on a larger scale. On the other hand, it could have been on another building, and, perhaps, of a different date. In either case, the same comments for completing the inscription made for CIL apply Germier-Durand, E., and F., and Alimer, A., Inscriptions Antiques de Nîmes, Toulouse, 1893, pp.115, No. 63 (CIL ), and 117, No. 64 (CIL ). Ehrenburg & Jones, pp.33, 35. Germier-Durand & Alimer, pp.115, 117. Reinhold, p.91, Note 81. See above Chapter 3, p.70.

104 The discovery of CIL amongst the ruins of the Roman works at the sacred fountain indicates that the building of which the inscription was a part was probably near the place where the inscription was found. CIL was found in a cellar, and had been placed there after removal from some other location. The first recording of this inscription in 1742, or some two years after the discovery of CIL , may indicate that CIL was taken from the sacred fountain by a collector. Germier-Durand suggested that both inscriptions may have either come from the same building or from different ones, without indicating a preference for either case 81. The house in the present Rue Thédenat, where CIL was discovered, was in the former Rue des Étuves (street of the old steam baths) indicating the remains of the public baths. The ruins of an enormous Roman baths have been found in the area, but cannot be dated. It is known that baths existed at Nemausus in the Tiberian period from the evidence of inscriptions. Since the aqueduct is probably Claudian, these buildings were supplied with water form the sacred fountain 82. Baths could, therefore, have been supplied from this source in the Agrippan period and it is possible that CIL came from an Agrippan baths building. There are so many uncertainties about the inscriptions that it is difficult to form conclusions about the structure or structures to which they were attached. Both were on quite a small building, since the letters do not exceed 20cm in height. Whether they indicate an Agrippan construction or mark him as a member of the Imperial family is unknown. They are almost certainly not from B.C. CONCLUSIONS Agrippa was more probably responsible for Nemausus becoming a colonia Latina in B.C. than Tiberius Claudius Nero in 45/44 B.C. It is unlikely that he built anything at Nemausus at that time. If he did so, the only construction which can be related to him is an altar to Nemausus on the square base on the margin of the sacred basin. In 19 B.C., Agrippa may have advanced construction of the nymphaeum of the Augusteum, which had been initiated by Augustus in B.C., but there is no Germier-Durand, p.117. Information on the baths is contained in Fabre et al, Le Pont du Gard : Water and the Roman Town, Paris, 1992, p.35, and Fabre et al, L'eau à Nîmes Languedoc, 1994, p.79 and Fig. 39, p.36, showing the locations of the five baths and the numerous wells of Nemausus.

105 evidence indicating any other particular work of Agrippa at that time. Also, there is no reason to think that he was responsible for general planning or construction at Nemausus. The two inscriptions bearing the name of Agrippa cannot be dated or related to any structure, and it is possible that they do not indicate construction by him, but were part of something dedicated to the Imperial Cult. The indications of Agrippan construction at Nemausus are, therefore, both uncertain and restricted, pending any reliable interpretation of the two inscriptions referred to in section 3. He can no longer be regarded as an important builder there, as was believed for many years, and until quite recently 83. This belief was based on mistaken attribution to Agrippa of the Pont du Gard viaduct and the Maison Carrée, undue imputation of Agrippan activity from the local prominence of his two sons, Gaius and Lucius, and questionable interpretation of the 'crocodile' coins. Also, it is now widely recognised that there is no clear evidence of a general re-organisation of the urban area of the town by Augustus and Agrippa in co-operation, and in which both had a continuing financial interest 84. Such widespread commitment would be surprising at a Latin rather than at a Roman colony. Furthermore, Nemausus had the resources to develop the town. Rome would surely have taken initiatives only at works which increased its influence, as at the Augusteum and the Maison Carrée, and, to a lesser extent, at the town walls and gates. Nevertheless, it would have expected the local elite to find as much money as possible to carry forward the works to completion. Also, Agrippa, in Gaul only briefly in 19 B.C., would surely have been little involved in Nemausus, and certainly less than in Lugdunum, where his actions in B.C. had been so important for urban development Roddaz, Marcus Agrippa, pp Ibid, p.399, suggested that Augustus and Agrippa were 'sleeping partners' in the re-organisation of the town, therefore implying a continuing financial commitment to work executed by others.

106 OTHER AGRIPPAN WORKS IN GAUL The constructions investigated in this chapter fall into two categories. The first is made up of works which can be imputed to Agrippa from his known activities. These include military works of his known campaigns and structures needed for the operation of his road network, such as road stations. There is no doubt that Agrippa built these campaign and associated road works, but there is little or no physical evidence of them. The second category consists of possible Agrippan works not considered in Chapters 1-4. (a) WORKS IMPUTED FROM KNOWN AGRIPPAN WORKS, OR ACTIVITIES: Agrippa was directly responsible for the military works of campaigns where he was the general in charge. There is no specific information about Agrippan military operations in Gaul. He won a brilliant victory against the Aquitani in 39/38 B.C. 1, but the location of his operations is unknown, and works have not, therefore, been identified. Agrippa was recalled from across the Rhine 2, but nothing is recorded about his operations in the region. Dating of the timbers of a wharf at Cologne to 38 B.C. is the only recorded evidence of his presence 3. In late 20 to early 19 B.C., Agrippa was in Gaul putting down disturbances 4, but there is no further definite information about these pacifications or about any works for them. The general lack of information in the historical texts and the loss of Agrippa's Commentaries present great difficulties for the archaeologist seeking Agrippan military sites in Gaul. It is unlikely that many remains of these will be discovered, except in N.E. Gaul where there were semi-permanent forts and other defences. Yet, the sum total of Agrippa's military works in Gaul was probably quite large since it resulted from large scale operations. It is not possible to speculate on these works, as is done here in Chapter 6 for the works of Agrippa's pacification of Cantabria in N.W. Spain in 19 B.C Appian, Cassius Dio, See above Chapter 1, Note 75, p.25. Cassius Dio,

107 It was argued above that Agrippa used the army to build his road network, and that this construction went hand-in-hand with a re-organisation of the military forces on or near the roads 5. Consequently, there was a large scale activity on developing construction camps, other camps, garrisons and forts in a wholesale re-deployment of large forces. These changes must have had a large immediate effect on the areas traversed by the road, and have also led to some permanent influence on development patterns there, and in wider zones. All these constructions were almost certainly in timber or earth, and have, therefore, perished. Even at the Trion site at Lugdunum there are no recorded remains 6. Nevertheless, their influence of later development was far reaching. It would be necessary to investigate these constructions separately in order to form more definite opinions about the extent of their influence at particular places. It can be assumed that any facilities on existing roads used by Agrippa to form his system were inadequate for rapid transmission of intelligence and the movement of military forces to trouble spots. Agrippa, therefore, had to build road stations for changing horses, overnight accommodation (in addition to encampment), stores, workshops for running repairs, etc. Some of these would have been at forts and garrisons. The Peutinger Tables and the Antonine Itinerary show the locations of some of the road stations of the Empire. There was an average distance between them of some kms 7. While the Agrippan stations may have not always been at these locations, the average distance between them was probably similar, being dictated by the practical aspects of travel. In this case, there would have been between 60 and 90 road stations on the Agrippan network, exclusive of the road south from Lugdunum. There seems to be no archaeological information on road stations in Gaul as early as B.C., but there is evidence of construction from about 30 B.C. to A.D. 240 at the Mansio of the Great St. Bernard Pass. Here there were not only two large buildings in timber on lower walling of stone, but also a stone temple 8. Most of Agrippa's road stations were not so large, but they made up a substantial amount of construction. Furthermore, many of these nodes on the first See above Chapter 1, Noes , p.33, 4. See above Chapter 2, p.44, for the construction camp on the Trion. Chevallier, Les voies romaines, pp Ibid, p.288 and Fig. 185, showing the Mansio.

108 Roman road network must have become the nuclei of later settlement 9. Since the Agrippan constructions associated with the Agrippan roads are lost, while at least the routes of the roads remain, the constructions have been largely overlooked. Yet there were so many of them that, together, they had a major influence on the development pattern in large parts of Gallia Comata, whereas the camps, garrisons and forts affected particular places. (b) WORKS WHICH MAY BE AGRIPPAN: Arelate: Arelate (Arles) has been noted as a port through which Agrippa brought in supplies in B.C. 10. The main reason for examining Arelate is, therefore, to assess possible works by Agrippa at its port, although Agrippa could also have been responsible for construction at the town. As can be seen from Fig. 3.1, Arelate was some 100 km N.W. of Massilia, the site of the ancient Greek colony. It was on the Rhône, and connected to the Mediterranean by the Canal of Marius, built in about 100 B.C. 11. Massilia controlled the canal and a Greek settlement at Arelate until the conflict between Caesar and Pompey in which Massilia supported Pompey 12. Caesar built ships at Arelate for his successful siege of Massilia 13. Arelate was founded as a Roman colony by Tiberius Claudius Nero on behalf of Caesar in 46 B.C. at the same time as he founded the colony of Narbo 14. Arelate was, therefore, one of the oldest Roman colonies in Gaul, the gateway to the Rhône, and destined to overshadow Massilia. Fig shows Arelate in the Empire, with the port area west of the Rhône and the town to the east. The first works at the town were on a grid to which the forum and theatre were aligned, but the amphitheatre, built after the forum and the theatre 16, was Woolf, Becoming Roman, p.88, noted the great importance of the Roman road network in shaping the settlement pattern of Gaul. See above Chapter 3, p.80. Vella, C., Leveau, P., et al, 'Le Canal de Marius et les dynamiques littorales du Gulfe de Fos', Gallia, 56, 1999, pp Rivet, p.190 and Note 2. Caesar, BC, 1.36, stated that twelve ships were built in 30 days. Suetonius, Tiberius, 4.1. Rivet, Fig. 22, p.192. Ibid, p.195 and Note 34, p.207, indicated a possible Flavian dating.

109 orientated differently. The (pontoon) bridge, shown on Fig. 5.1, almost certainly did not exist in the Agrippan period 17, and there is no evidence of a fixed bridge at this time. In this case, transfer across the Rhône was by boat. The town has been extensively investigated for its general history and development in the Roman period 18. Gros has also considered the evidence for an Augustan programme of works there which he related to other Roman works both before and after 19. The forum was built on sloping ground. It was, therefore, necessary to construct a cryptoporticus to provide a level surface for the forum and its columns. Fig shows the general layout of the large structure, 240 m long and 12 m wide. Fig indicates the construction of vaults supported on exterior walls and piers connected by arches. Amy has described the building 22. This utilitarian type of work could be dated to 46 B.C. or later in the absence of decoration other than cornices. There is evidence of a sanctuary of the Imperial Cult in the north galley of the cryptoporticus from the discovery of detached marble fragments. These have been identified as a representation of the clipeus virtutis offered to Augustus in 27 B.C. 23 and a bust which may represent either Gaius, son of Agrippa, or the young Octavian 24. The second interpretation would indicate a visit to Arelate by Octavian when he was in Gaul in 39 B.C. 25 and, also, possible construction of some or all of the cryptoporticus at that time, with the sanctuary added after 27 B.C. A bust depicting Gaius would still not rule out such a construction chronology. The sparse remains of the forum The bridge was sited to provide the most direct route to the amphitheatre from across the river. There seems to be no text mentioning it in the early Empire and the mosaic of it at Ostia in the Piazzale of the Corporations is probably of the 2 nd century A.D. (Ward-Perkins, J.B., Roman Imperial Architecture, Yale, 1983, p.143). Rivet, pp , and Heijmans, M. & Sintes, C., 'L'évolution de la topographie de l'arles antique. Un état de la question, Gallia, 51, 1992, pp Gros, P., 'Un programme augustéen: ' Le centre monumental de la colonie d'arles', Jdi, 102, 1987, Berlin, pp Bromwich, The Roman Remains of Southern France, Fig. 51. Coulon, Les gallo-romains, Paris, 1985, Photograph p.39. Amy, R., 'Les cryptoportiques dans l'architecture romaine', CEFR, 14, 1973, 275 ff. RG 34 and Gros, 'Un programme augustéen ', p.346, Note 33. Gros, 'Un programme augustéen ', pp , for the marble fragments and the sanctuary, and Note 41, p.349 for the bust. Appian, , as interpreted by Roddaz, Marcus Agrippa, p.66, Note 203, for evidence of the visit.

110 above the cryptoporticus have been dated no earlier than 26 B.C. 26. The archaeological evidence is congruent with a cryptoporticus construction of 46 B.C., or later, irrespective of the sanctuary dating. There is, therefore, a possibility of Agrippan works on the cryptoporticus in B.C. Fig shows the principal surviving part of the theatre built on the same grid as the cryptoporticus. This large structure, with a cavea (auditorium) diameter of 102m 28 was built on virtually level ground with the cavea supported on substructures. The project was, therefore, much more extensive than that for a theatre built on a hillside, as at Lugdunum or Nemausus. Such a large theatre in stone is not to be expected at Arelate in B.C. since, even at Rome, the first stone theatre of Pompey, dedicated in 55 B.C., was not joined by others in stone until some 40 years later 29. Also, the remains of the surviving theatre indicate construction predominantly from about B.C., with some earlier work, which cannot be dated confidently as early as B.C. 30. Agrippa could have built a timber theatre at that time with some stone supports which were incorporated in the later work, but this possibility is purely speculative. Nothing is known for certain about the nature, extent, or dating of works in the town built before or with the forum and the theatre, although Gros discerned from the extent of constructions that these were begun on foundation 31. Also, there is little definite information about the location of the walls, as shown on Fig The 'Arch of Constantine' and the Arcus Admirablis, both shown on Fig. 5.1, probably marked the original pomerium, but neither now remain, and the first stone Gros, 'Un programme augustéen ', Fig. 3, showing a cornice from the forum and pp , for the dating of this in relation to other remains at Arelate. Ibid, Fig. 10, p.351. Rivet, p.194, pointing out the 103m diameter of the theatre at Orange and the 102m diameter of the theatre at Arelate. Cassius Dio, 39.38, for the dedication of the theatre of Pompey at Rome in 55 B.C., Cassius Dio, 54.25, for the dedication of the theatre of 'Balbus Minor' there in 13 B.C. Gros, 'Un programme augustéen', pp , also Heilmeyer, W.D., Korinthische Normalkapitelle, Studien der Geschichte der römischen Architecturdekoration, Heidelberg, 1970, p.112, and Tafel 7, 2 and 3. Gros, 'Un programme augustéen', p.342.

111 construction there cannot be dated, although this has been described as 'Octavianic' by Rivet 32. There is, therefore, an indication from the archaeology of possible Agrippan works of B.C. at the cryptoporticus of the forum, and perhaps at a first theatre. Other early works for which Agrippa may have been responsible at the town or for the walls and gates cannot be identified or dated. Nevertheless, it is unlikely that Agrippa built anything in the town, as distinct from at the port, unless it was related to the logistics of his supplies. In 19 B.C. Agrippa could have advanced construction at the existing theatre and the forum, or elsewhere, which had been initiated by Augustus during his visit to the West in 27 B.C., as has been argued for the nymphaeum at Nemausus in Chapter 4. Yet, Agrippa was active in other parts of Gaul during his six months there, as discussed above. There is, therefore, little possibility of Agrippa being responsible for any significant building at the town of Arelate in 19 B.C., even if he visited it and did not go to Hispania on the Via Domitia through Nemausus. Improvements at or near the port dating from B.C. must, therefore, be the focus of investigation of Agrippan works. Agrippa could obtain substantial general supplies from southern Gaul, but much of the specialised military equipment could probably be acquired only from Italy 33. Since Agrippa's military forces in Gaul in B.C. were substantial, and were largely recruited there 34, large quantities of military supplies had, therefore, to be brought in as goods. Fig. 5.1 shows only one wharf, which was on the west bank, but wharves existed on the east bank near the baths, shown on Fig. 5.1, in the Roman period, although the dates of their construction are unknown 35. Agrippa's supplies up the valley of the Rhône to Lugdunum and beyond could be taken by river craft, or by road on the east bank, shown on Fig The Rhône could not always be used for upstream traffic 36. As noted already 37, the pontoon bridge shown on Fig. 5.1 almost certainly did not exist in Rivet, pp.191-3, and Notes 12-21, and Heijman's & Sintes, pp.48, 49. King, Roman Gaul, p.116. Roddaz, Marcus Agrippa, p.70 and Note 217, and above, p.81 and Note 15. Heijmans & Sintes, Fig. 6, p.143 and p.147. Strabo, See above Note 17, p.101.

112 the Agrippan period. Agrippa, therefore, needed a wharf on the east bank for direct trans-shipment on to the road to Lugdunum in order to avoid barging across the river. Such a wharf would have been strictly secured because of the military goods handled there. Agrippa would almost certainly have been obliged to build a new one, even if one already existed. Large shipments brought in by sea had to be safely stored pending further movement in smaller lots. Acryptoporticus within the town walls was an ideal storage place, and, also, needed to develop a level base for the forum. It is, therefore, possible that Agrippa and the town authorities co-operated in building or augmenting the capacity of a cryptoporticus. It can, therefore, be concluded that Agrippa probably built at the port and at the cryptoporticus in B.C., but that he built nothing else at the town at that time, or in 19 B.C. Forum Iulii: Forum Iulii (modern Fréjus) was located not far from the coast of the Mediterranean near the eastern border of Gallia Narbonensis, and some 200 kms east of Arelate. The name Forum Iulii appears first in a letter dated 43 B.C. from Munatius Plancus to Cicero, without description or explanation of the name, indicating it as an established place, but with no reference to a colony 38. It probably did not become one until after Actium 39. It had a port formed from a natural lagoon, which may have been further developed because of the struggle for mastery of the sea between Octavian and Sextus Pompey 40, since it was far from the main base of Pompey in Sicily, and provided shelter at a time when most of the west coast of Italy was without harbours 41. The fleet used by Agrippa to defeat Sextus Pompey was built principally at a new artificial port created by Agrippa in the Bay of Naples using Lakes Lucrinus and Avernus 42. Ships were also constructed in many parts of Cicero, Ad fam, Rivet, p.226, suggested foundation in 35 or 27 B.C. based on various texts of Tacitus and Pliny the Elder. Gascou, J., in 'Quand la Colonie de Fréjus fut elle fondée?', in Latomus 41, 1982, pp , p.144 suggested the period after Actium, and probably from B.C. Gros, 'Un programme augustéen', at p.362, indicated foundation by Octavian in about 31 B.C. See above Note 10, p.2, for Sextus Pompey. Cassius Dio, See below Chapter 11, Note 18, p.211.

113 Italy 43, and Forum Iulii was very close to Italy. Agrippa may, therefore, have built vessels at Forum Iulii. Roddaz considered this possibility, but rejected it on the grounds of insufficient archaeological evidence 44. Yet ships were constructed on timber slipways, and it was necessary only to have timber moorings and a few sheds, also in timber. Such works would have perished. There is some evidence of remains under buildings of the Augustan period in the port area 45, although this does not necessarily indicate shipbuilding. The possibility of Agrippan shipbuilding at Forum Iulii in B.C. is, therefore, based entirely on imputation from the general history. Tacitus said that Italy was guarded by two fleets, one on each seaboard, and based at Misenum and Ravenna. He also described Forum Iulii as defending the near coast of Gaul, and stated that Octavian had sent warships captured at Actium there strongly manned 46. Strabo described Forum Iulii as an Augustan port of size similar to Massilia, and stated that these two places provided the only important harbours on this part of the coast 47. It can, therefore, be presumed that a naval station was built at Forum Iulii after Actium. Fig shows Forum Iulii in the Empire with a port area of 20 hectares connected to the coastline by a canal. The sea has now receded. It can be assumed that the ships from Actium were brought into a small harbour from which the large harbour was formed. This harbour was flanked by two port constructions on natural eminences, at the Butte St. Antoine to the west, and La Plate-forme to the north. The first was probably the base of the commander of the fleet, and the second the residence of the governor of the province 49. The harbour constructions shown on Fig. 5.5 have been dated to the general period B.C. by Février 50. Some traces of foundations and a workshop from an earlier period were found under the port Cassius Dio, 48, Roddaz, Marcus Agrippa, p.74 and Notes See below Note 51, p.106. Tacitus, Annales, 4.5. Strabo, Rivet, Fig. 29, p.227. Ibid, p.228 and Notes 18 and 19. Février, P.A., 'Fouilles à la citadelle méridionale de Forum Iulii (Fréjus, Var) en 1955', Gallia 14, 1956, pp.35-53, with summary of dating on p.53; also 'Fouilles à la plate-forme de Forum Iulii', Gallia 20, 1962, Fasicule 1, pp , p.203.

114 works at the Butte St. Antoine 51. Agrippa, as admiral of the fleet at Actium, would surely have made the decision to send captured ships to Forum Iulii. This indicates his prior knowledge of that port, and strengthens the hypothesis of his building ships there in B.C. As naval victor and the builder of Portus Iulius, Agrippa would have been consulted on the layout, design and facilities of naval bases, and would, therefore, have been involved in the development of that at Forum Iulii. Agrippa was, therefore, the 'father' of the naval base at Forum Iulii, and the planner and builder of it to an extent presently unknown, but which may come to light from further investigations. Nevertheless, it is almost certain that Agrippa would have visited the base under construction, especially since it was almost in Italy. The construction of the naval base must have prompted new constructions at the town itself, but there is no indication that Agrippa was responsible for any works at the town from what is presently understood about these 52. CONCLUSIONS Agrippan works at road stations, forts and garrisons, in association with the construction and operation of his road network, were substantial, and more influential in the development of Gallia Comata than any of his other works, except his roads. Agrippa's military works remain unknown pending possible discovery. Agrippa may have built at Arelate in B.C. to handle his supplies, and at Forum Iulii in the same period for shipbuilding. He certainly planned, and, perhaps, supervised the construction of a naval base at Forum Iulii after Actium. It is possible that Agrippa built elsewhere in Gaul, and that the evidence for this is lost. Yet there is little reason to impute such works since construction of the first form of the Agrippan road network and his other activities would have occupied most of his time in Gaul in B.C., and he had less opportunity to build there in 19 B.C. Any constructions, other than those listed in Chapters 1 to 5, were probably at urban centres such as Vienne, Orange, Narbonne, Beziers, Toulouse, etc. Nevertheless, these were almost certainly not major works Février, Gallia 14, 1956, Fasicule 1, Goudineau, C., (Informations archéologiques), Gallia 33, 1975, p.562 and Goudineau (Informations archéologiques), Gallia, 35, 1997, Fasicule 2, p.498. Rivet, pp , provided a brief account of Roman works at Fréjus.

115 AGRIPPAN WORKS IN GAUL : CONCLUSIONS Agrippa's main tasks in Gaul in B.C. were to establish order, and to institute a framework of strategic control. The only remaining material evidence of his military actions is at Cologne, where timbers of a wharf dated to 38 B.C., indicate that he crossed the Rhine for unknown operations in Germania. Nevertheless, Agrippa must have been responsible for other military constructions in N.E. Gaul, and, also, in Aquitania, where he conducted a military campaign. These works may yet be found. Agrippa's most important construction in B.C. was his road network radiating from Lugdunum to the Atlantic Ocean, the English Channel, the Rhine, and southern Gaul. Agrippa formed this principally from existing roads, since he could not otherwise create an effective system before he left the territory. Agrippa's road construction went hand in hand with a reorganisation of military forces. Larger units were broken down into mobile detachments stationed on or near his road network. This was used to convey intelligence of insurrection and for response by the security forces travelling along it. Agrippa's road network, with its road stations and military installations, formed the framework for the much larger Roman road system and was a major influence on the development of Gallia Comata in the Roman period. The creation of the road network radiating from Lugdunum made this town the strategic centre of control in Gallia Comata, and started the process of its conversion into the virtual capital of that large territory. Agrippa built at Lugdunum to establish his road junction there, and he improved port facilities for his supplies brought in by river. Agrippa probably constructed some buildings at the urbs quadrata established by Munatius Plancus on the plateau of La Sarra when he founded Lugdunum in 43 B.C. Agrippa's construction of the small Temple of Valetudo at Glanum in B.C. was, almost certainly, related to his mobilisation of military forces and supplies in a region not far from the port of Arelate and near the Via Domitia. Agrippa may have improved port facilities at Arelate and have built at the cryptoporticus of the forum to store his supplies brought in by sea. Agrippa probably adapted the existing port of Forum Iulii to build ships for the naval contest with Sextus Pompey, and, after Actium, he certainly selected Forum Iulii as the site of a naval station and was concerned with its planning, and, possibly, with its construction.

116 Agrippa was in Gaul from late 20 B.C. to the first quarter of 19 B.C., initially putting down disturbances in the north-east. None of his military constructions have yet been identified positively, but the first Roman bridge at Trier has been dated to the general period, and road works were associated with it. At Lugdunum, Agrippa was almost certainly responsible for the construction of a praetorium at the Cybele site, and, possibly, for setting in hand a major programme of urban expansion, including first planning and building of aqueducts. Agrippa was, perhaps, responsible for the start of works at a new forum yet to be identified, and at the existing theatre where the first works have not yet been dated. The first of the four aqueducts to be built was the Mont d'or and Agrippa may have been responsible for construction of it if there had already been aqueduct planning by 19 B.C. If there had been no such planning, Agrippa would have been the 'father' of all the aqueducts as a general urban planner. Agrippa has generally been regarded as an important builder at Nemausus in 19 B.C., but this opinion can no longer be supported. Agrippa may have been responsible for advancing works at the Augusteum started by Augustus in 27 B.C., but there is no clear evidence of any other possible Agrippan construction in 19 B.C. Agrippa, therefore, built in Gaul in B.C. rather than in 19 B.C. He was also a builder of strategic works rather than of urban monuments. Furthermore, his road network, the largest and most influential of his constructions, was, essentially, an instrument of Roman expansion into a large territory previously conquered but little organised by Rome. These conclusions represent a new interpretation of Agrippa in Gaul, replacing that in which Agrippa was seen to have been active mainly in 19 B.C., and predominantly at Nemausus. The new interpretation results from a focus on Agrippan constructions, consideration of these as instruments of policy, and, in the case of the Agrippan roads, an engineering analysis of Agrippa's response to an engineering problem.

117