L’analyse du réseau hydraulique de Bibracte – l’un des oppida les plus importants à la fin de l’indépendance gauloise et l’un des mieux étudiés – permet de dresser un premier panorama des types d’aménagements destinés à capter ou à évacuer l’eau dans le courant du ier s. av. J.‑C. et de mesurer les transformations en matière de gestion de l’eau après la Conquête romaine. Ce site urbain, avec ses 132 installations hydrauliques découvertes entre le xixe s. et 2017, n’a pas d’équivalent pour la même période. Son étude révèle notamment l’existence de structures de stockage collectives de grande capacité, l’attention portée à l’évacuation des eaux pluviales et la récurrence des techniques locales côtoyant des équipements typiquement romains.
The hilltop of Bibracte, located on top of the Mount Beuvray hill, was one of the most important oppida of Gaul. The analysis of its hydraulic networks is a particularly rich field of study and is an original approach to understanding the types of water management at the end of the Iron Age and immediately after the Roman conquest. In these particular situations it is possible to examine how populations adapted to their environment by developing specific techniques to channel and store water as well as to evaluate the impact of Roman technologies on their social practices. So far research has focused to only a limited extent on the control of water in Gallic hilltop settlements during the Iron Age. The main objective of this study, carried out as part of a doctoral thesis (Borau 2010) and centred on a symbolic site in Gaul, was to put into perspective the topography and the water resources with the various types of hydraulic structures discovered on this particular hilltop settlement and to describe their main characteristics. In order to do so, exhaustive research was carried out, compiling the entire documentation collected between the 19th, 20th and 21th centuries, in particular ancient excavations and geophysical analyses but also new archaeological research carried out under the auspices of the European Archaeological Centre of Mont-Beuvray directed by Vincent Guichard (team composed of Miklós Szabó, Sabine Rieckhoff, Daniele Vitali, Daniel Paunier, Thierry Luginbühl, Philippe Barral and Jean-Paul Guillaumet). In 2007 several hydraulic structures were excavated and sedimentary analyses were carried out on the deposits taken from one of the basins. The settlement of Bibracte was chosen for several reasons. First of all, during the Iron Age Bibracte was one of the main oppidum in Gaul. It was the capital of a Gallic tribe, the Aeduans, who very early on established relationships with the Romans prior to the Roman conquest, i.e. from the beginning of the 2nd c. BC. (Caesar, I, 33, 36). Secondly, this hilltop settlement was particularly chosen by Julius Caesar as his winter residence in 52 BC (Caesar, VII, 90). The Gallic city was established on top of a mountain, on irregular and uneven ground: it is characterised by steep slopes and deep valleys linked to natural springs. The mountain peaks have an altitude of 822 m and the main human occupations are distributed at an altitude of 700-800 m. Two ramparts protected the site: an initial 7-km-long external rampart (A) enclosed an area encompassing 200 ha, and a second, 5,2-km-long internal rampart (B) encircled an area encompassing 135 ha. Built according to the principle of the murus gallicus. It was pierced by several gates, the main gate being located in the north. From this entrance a 2-km-long road crossed the site from the north to the south. The site was occupied between the end of the 2nd c. BC and the Tiberian period. Several quarters were identified: in the northern part a craftsmen’s quarter [1]; in the central part a public quarter [2] and a residential area [3]; in the southern part a sacred area [5], and lastly two sectors [4 and 6] in which only the springs are known. The buildings were erected on terraces providing the city with a stepped aspect. The initial buildings were timber houses which were later replaced by stone buildings in the mid-1st c. BC therefore changing the aspect of the city. The postholes associated with the timber buildings were discovered beneath the walls of the stone buildings. However archaeological excavations revealed the coexistence of traditional wattle-and-daub houses of Gallic origin and of stonemasonry architecture of Italic origin. This study made it possible to register 132 hydraulic structures distributed across the oppidum. Water was supplied through three systems. First, it can be noted that most of the wells were concentrated in the craftsmen’s quarter. Second, huge basins were discovered around every excavated spring. Third, big reservoirs are distributed across the whole site. Generally, it can be considered that wells are the most common structure for the supply of water; they are the most wide-spread and are the easiest to deploy. However, at the scale of the oppidum, only twenty structures of this type were discovered, twelve of which were located in the craftsmen's quarter. Secondly, wells were found in the residential area of the Parc aux Chevaux in the house PC11, PC14-15 and at PS0. Two wells were located near the sanctuary. The depths of the wells, which are mainly circular in shape, vary greatly between 2 m and 15.50 m. Most of these were lined with stones, others were simply dug into the ground. In other locations several wells stand out because of their large diameter, ranging between 1 m and 1.60 m. The public or private status of these wells is still not easy to determine but close observation of their location provides some clues. Most of the wells were apparently destined for collective use because they were located in circulation areas, in the streets or squares. Other wells obviously had semi-private functions, installed in private areas but accessible from the street. Although it is still difficult to propose an overall plan, our research suggests the existence of one or more public or semi-public wells per insula, i.e. per group of houses/workshops bounded by streets. The basins associated with the springs have common characteristics. These basins are big, encompassing large surface areas, and therefore could store a large volume of water. They were positioned in contact with or near springs. They have similar construction techniques: either they had a wooden casing (generally these are vertical beams with notches into which fit wooden boards) surrounded by a sealing of raw clay, or they were made of stone and were waterproofed by clay sealing, or they were constructed using a combination of several techniques. Several examples can be cited. In sector 6 several pipes captured various resurgences and fed a basin made of oak during the 1st c. BC. It is still preserved over an area of 16 m2 but it was probably even bigger, maybe as big as 35 m2, thus containing a volume of 35 m3. This basin is the most ancient one built of the site. Unfortunately, nothing is known about its immediate environment. To the west of the oppidum sector 4 (Saint-Pierre fountain) corresponds to another tapping of a source, modified several times. The size of the successive basins tends to diminish over time. Dated to the 1st c. BC, three basins, covering a total surface of 205 m2 were built at the level of the spring. Then, around the turn of this era, a new basin, separated from the spring, was built according to a mixed technique: the stone-built walls mark the edge of a wooden casing with a joint of clay. The capacity of this new basin does not exceed 9 m3. In the course of the 1st c. AD this basin was restored and was used up to the beginning of the 2nd c. AD. In the eastern part of the site, in sector 2 (Pâture du Couvent), the investigations carried out in the 19th c. and the geophysical exploration of the 1990s make it possible to reconstruct a basin built with masonry stone blocks that taps a spring provided with a wooden gate which redirects the spring into a sewer if necessary. Built during the 1st c. BC, such a device made it possible to avoid the flooding of the adjacent building, which is especially important as this is thought to be the public centre of the oppidum. Downstream from this building, in the valley, a double resurgence exists forming a stream which flows in the thalweg. During the 19th c. five successive basins were apparently discovered at the contact point between the stream and the rampart: one basin was located inside the rampart and the four other outside. These large basins measuring 7 m on each side cover a surface of about 50 m2 each. They were built with pieces of wood and clay. This assemblage played several roles: collecting and slowing down the water flow of the stream tumbling down the slope; creating a large water supply; preventing the destruction of the rampart; and providing water for the needs of the craftsmen. The steep slope in this sector explains the creation of a cascade of basins: this was an original solution which testifies to a well-conceived organisation of hydraulic structures. Thanks to the present study it can be assumed that in Bibracte there were specific construction methods used for the hydraulic structures corresponding to local, i.e. Gallic skills. This study moreover demonstrates that these Gallic techniques continued to be used after the Roman conquest: other nearby sites that are part of the Aeduan territory have basins with similar characteristics, for example the sites of Montlay-en-Auxois (Côte-d’Or). Another tapping of a source testifies to the combination of techniques and to the designs for the provision of water in the oppidum when it comes under Roman influence. In sector 3 (Parc aux Chevaux) in the western part a source serving as a small private fountain in a garden was connected to several pipes which cross the house in order to feed a vast pond built in front of this house next to a pathway. Thus a private spring supplied a public basin. Built around the turn of this era, this stone basin was sealed with clay. It encompassed a surface of about 18 m2. The bottom of the basin is located more than 2 m below the level of the ancient street; it is thus semi-subterranean. With a capacity estimated at between 35 and 40 m3, this basin is in fact a real reservoir. As a matter of fact, this second type of hydraulic structure, the reservoir, can be observed on the site and it completes the hydraulic equipment of the oppidum: there are five excavated structures which are rather enigmatic and which can be considered as being reservoirs. Two structures in particular have attracted our attention. First of all the elliptic basin discovered in the centre of the main road of the oppidum, near a crossroad in sector 2. The function of this very original structure remained a mystery for a long time, especially because of the absence of a water inlet system. This basin was 10.50 m long and 3.65 m wide and it was preserved to a height of 1 m. It covered a surface of 30 m2. Dated to the beginning of the 1st c. BC, it was also built of stone (quadrangular and irregular-shaped blocks) and waterproofed with clay (the bottom and the foundation trench). The clay deposits underwent microstratigraphical laboratory analyses in 2007: these revealed the existence of an initial layer containing traces of seaweed deposits and coal deposits linked to a nearby fire, topped with a refill of clay. Water was evacuated from the basin via a square hole cut in the bottom of the basin, sealed by a sluice gate operated vertically from the outside and connected to a system of pipes built from wooden boards. The design of this basin is completely original and it has no parallels, but the model could be Greek, not only because of the mastery of the stereotomy but also because of its elongated shape and its location on the public road, which is reminiscent of Greek cities. Formal comparisons are still difficult because of the lack of study of this type. In the absence of an identified water inlet system, the hypothesis can be advanced that we are dealing with an open basin which was filled by rainwater and which would have served as a water reservoir located on the public road. At the entrance to the oppidum, a big quadrangular structure measuring 160 m2 (20 x 8 m) and 2 m deep, i.e. 320 m3, was excavated which was perfectly connected to the eastern wing of the rampart. The bottom was also covered by a layer of clay. A pipe built from stone was used to evacuate of water. This structure is dated to the 1st c. BC. At least a further three structures were excavated at but these are only known through the excavations carried out in the 19th c. One of these structures was located in the square of the residential area, west of the main road, the second near a crossroad of streets in the sacred area. These structures were dug into the ground and had no cladding. They had large capacities, between 30 m3 and 73 m3. The similarities between these various structures are striking: they had a large ground surface, they were located in natural steep-sided ground, they were close to axes of circulation and they had no water supply system. These structures destined for water storage were completed by a water distribution system: several wooden pipes joined together by iron rings were discovered beneath the streets, but no direct connection with a source or a basin could be established: this is the case for the main road crossing the residential area, the craftsmen’s quarter and the side street along the houses in the northern part of the public district. About 50 metal circles and fragments of wooden pipes were discovered. Their dimensions vary between 8 and 9 cm in diameter but some of them were as long as 15 cm. Pipeline sections made from the bellies of amphoras are also known. Lastly, the management of run-off water and waste water seems to have been the main concern of the urban planners. This is due to the abundance of rainfall in the region. No fewer than 34 gutters were listed in the oppidum, belonging to various phases of occupation and visible in all the quarters. They seem to emerge from the mid-2nd c. BC and were maintained or renovated up until the abandonment of the site in the 1st c. AD. The numerous gutters, with a V profile or sometimes flat-bottomed, are either simply dug, or engineered (in a framework of boards for which nails sometimes still exist, or with slabs of stone). On average they are 0,30-0,80 m wide with a depth of 0,25-1 m. They were constructed alongside the streets, on one or both sides, or around buildings. They were essential for the conservation of the terraces on which buildings were erected. On the other hand, the small proportion of sewers is rather significant: this type of structure is typically Roman and could be identified only in the public quarter at the time of the construction of stone monuments in the mid-1st c. BC. The stone-built sewer was 1,70 m high and 0,75 m wide. It evacuated the water from the basin and drained the entire quarter. As regards the natural water resources, seventeen natural water springs can be listed on this mountain. The analysis of their distribution is particularly interesting because these resurgences can be linked with human activity: it can be seen that most of the springs (n = 10) were located at an altitude above 750 m. Furthermore, these springs could supply two thirds of the oppidum, i.e. the whole site except for the northern part that corresponds to the craftsmen’s quarter at the entrance to the site. It can therefore be assumed that distinct town quarters, in particular the public quarter (2) and the residential quarter (3), were established according to the closeness of springs. The two ramparts enclose the greater part of the water sources, which reflects a deliberate choice made by the town planners. In addition, rainfall analysis proves that this sector of Burgundy is one of northern France with the highest rainfall. It rains on average one day out of two. This explains the importance of the network of water evacuation, which was one of the main problems faced by the inhabitants of Bibracte. This abundance of water requires efficient water management to prevent the foundations of buildings and terraces from being destroyed and to ensure the continued circulation of people in the town. Yet, it is rather paradoxical to observe, in a rainy region, the omnipresence of structures for collecting water, such as vast tapping basins or reservoirs. This suggests the presence of specific practices: these high-capacity storage structures were accessible by the inhabitants in public places. They are distributed across the whole site. This is therefore an example of an organised water storage system throughout the whole oppidum, which was needed not only to meet the daily needs of the population, but also for others purposes such as fire-fighting. Furthermore, this study highlights the existence of numerous very well preserved wooden basins as well as previously unknown local technical traditions. These skills were maintained after the Roman conquest, as is shown by other sites close to Bibracte where they also could be evidenced. To conclude, this study reveals the specific organisation of the water supply network in the oppidum of Bibracte.
