Topographie et restitution du chantier de l’aqueduc d’Arles (Bouches-du-Rhône)

Abstract

The Roman aqueduct of Arles is made up of three sections. The first two were fed by springs on either side of the Alpilles range, one around thirty kilometres to the north-east in the alluvial plain of the Durance, the other to the south of the massif at the foot of Les Baux, at a distance of fifteen kilometres from Arles. These two sections converged in the town of Fontvieille above the deep depression of the Vallée des Baux which separates the Alpilles from La Crau. The third successively crosses this plain and that of the Rhône until its end point in Arles. These three sections are of unequal length. The first measures 38 km from the first point where it is identified, the second 5.60 km from the village of Paradou where the channels bringing water from the two springs converged, and the third 12 km from the basin at this convergence point.They are not contemporary. In the 1990s, excavations established the chronology. The eastern branch dates from the time of Augustus. At the beginning of the 2nd century AD, it was assigned to supply the Barbegal mills while the other branch was extended into the Durance valley. Its 34 m level difference corresponds to an average slope of 0.68 m/km, significantly greater than that of the Nîmes aqueduct which has a level difference of 12.27 m over an equivalent distance, i.e. an average slope estimated at 0.248 m/km. However, these figures do not take into account adaptations to the topography. Differential GPS was used in order carry out georeferencing and hence to achieve accurate levelling. In two cases, the levelling of a section made an essential contribution to the dating of the aqueduct and the evaluation of its flow. At Paradou, it confirms the Augustan attribution by establishing the contemporaneity of the canal with the Aurelian Way, which dates back thousands of years. The slope of the channel was calculated to optimize water flow. It was greatest upstream of the crossing structure, then gradually diminished to prevent overflow caused by excessively rapid slowing. On the basis of the profile of the aqueduct’s route in La Crau, its maximum daily flow can be estimated at 300 l/s or 26,000 m3.In the second part, the altimetry of the canals is used to describe the topographical component of the specifications given to the teams that shared the construction operations, to assess the significance of the route, the skills of the contractors, the technical constraints and the financial restrictions. Since the available data is exclusively archaeological, this section begins by assessing the contribution of written sources and the debates they have prompted. Before the construction of an aqueduct channel, a specialist would determine the precise route, positioning crossing points on the basis of the altimetric factors. The route would then be divided into sections and the construction work allocated to contractors. When the slope was too steep, a drop was created. A bridge would be built or a tunnel dug to avoid the excessive fall loss entailed in going up a valley or around an interfluve. The cost of these operations would affect the choice of route by forcing the project manager to compromise relative to an ideal direct route. Hence, when a topographical feature separated two points from the planned route, the contractors might choose to bypass it rather than opening a deep trench or digging a tunnel. However, bypassing an obstacle would lengthen the channel, reduce the slope and flow speed, and diminish the flow rate. Where this happened, it is possible to observe the effect on the deposits left by the carbonated water: their volume and position are evidence of reduced flows caused by a decrease in the slope or by a counter-slope.The route of the eastern branch of the aqueduct was direct and its slope regular. Between the catchment and the basin where it converged with this eastern branch, the northern Alpilles branch transferred the water captured in the Durance valley to the south. In the northern foothills, the channel trench tunnelled through a height line separating the catchment in the Mollégès marsh from the village of Saint-Rémy. To bypass the western part of the range while losing as little slope as possible, the aqueduct used bridges to enter and cross the valleys cut into the slopes, and crossed the interfluves via tunnels with a cumulative length of 950 m.In the terminal section, which starts at an altitude of 24 m NGF (general levelling of France), the pipe had to maintain sufficient altitude to supply the Roman colony. It successively crossed the Vallon des Arcs on a bridge more than 300 m long and 4 to 5 m high. The deep depression of the Vallée des Baux was crossed by a bridge more than 1 km long comprising around sixty arches more than twenty metres high. Water crossed the subhorizontal plain of La Crau in a deep trench. The aqueduct then descended into the alluvial plain of the Rhône and arrived at Arles in channel supported on arches. It entered a tunnel at an altitude of 16.56 m NGF, 6 m below the highest point of the city.In the Alpilles, the reduced rate of flow in the terminal section crossing the range was caused by the difficulty of levelling an optimal route in a wooded and hilly environment. In La Crau, where this operation did not present any particular difficulty, the shallowness of the slope is explained by a desire to limit the cost of earthworks. During the Augustan period, the colony was satisfied with tapping into springs in the south of the Alpilles, which was enough to supply water to most of the city. At the beginning of the 2nd century, tapping into new sources north of the Alpilles offered the possibility of bringing water to the highest point of the city. But this would have entailed increasing the height of the structures in the final section, at a cost that must have seemed excessive. The most likely explanation for tapping into new sources lies either in an increase in the city’s needs or a drop in the flow of those captured by the southern branch.Financial conditions explain why different and sometimes more appropriate choices were not chosen. Overall, the main conclusion that emerges from this study is the primacy of the desire to secure the city’s water supply. By tapping springs fed by the Durance aquifer, the people of Arles were able to adapt to a change in the Mediterranean climate characterised by increased episodes of drought. They thus anticipated the use of Alpine water resources, which is a specific feature of current water management in Provence and distinguishes this territory from other French regions.