Arles / Barbegal (France)

The city of Arles, ARELATE, set up as a roman colony for veterans of Caesar's 6th legion on a low hill on the left bank of the Rhône near the river mouth, grew out to be one of the most important cities in the Western Roman empire, known as "Gallula Roma", the Rome of Gaul. Such an important city needed an aqueduct, and the most suitable area to provide the city with water was a 35 km long EW trending limestone massif to the NE, known as the Alpilles, with a maximum elevation of 492m. In roman times, the range was densely wooded and the water that drained into the limestone's reemerged in springs at the foot of the hills.

Springs from both the North and South sides of the range were used to feed an aqueduct system of 62 km long with a capacity of 45.000 m3 of water per day. The aqueduct system of Arles is complicated since it did not only provide the city with water; about half the water provided by the aqueducts was used to power a major factory, a series of 16 watermills that provided the city with flour. This mill complex, the Mill of Barbegal, was built on the steep Southern dip slope of a limestone ridge at the South side of the Alpilles. This ridge, the Chaînon de la Pène, is separated from a satellite massif of the Alpilles known as the Défend de Sousteyran by a narrow valley, which is crossed by a double aqueduct bridge after which the valley is named (Vallon des Arcs numbers 2 on the map below). The aqueducts cross the valley at it highest elevation.

Map of the aqueducts of Arles (aqueducts in green)

The Eastern aqueduct bridge in the Vallon des Arcs transects the Chaînon de la Pène to end in the mill complex of Barbegal, as explained below.
The Western aqueduct bridge ends in a rock cut channel that makes a sharp curve to the West and runs along the North side of the Chaînon de la Pène until it bend South where it crossed the Vallée des Baux over an 800m long and up to 20m high aqueduct bridge to the edge of the Crau plateau in the South. The aqueduct channel followed the Northern rim of this plateau in the direction of Arles. Finally, it crossed the 2 km wide depression between the Crau plateau and the hill of Saint-Pierre-des-Mouleyrès at Arles by an aqueduct bridge close to the 18th century Craponne aqueduct bridge of Pont-de-Crau.

In Arles, the aqueduct entered the city on the East side at 17m altitude, and split into a branch towards the forum, and one that ran along the Westside of the amphitheatre. The castellum aquae must have been somewhere in this area. Lead pipes that were found in the bed of the Rhône in the 19th century prove that the important suburb on the right bank of the river (now the suburb of Trinquetaille) was supplied with fresh water through an inverted siphon below the river.

Until recently, it was not clear how the water was distributed between the Barbegal Mills and the city. Two aqueducts channels meet just North of the double bridge in the Vallon des Arcs. The so-called Eygalières Aqueduct comes from the North side of the Alpilles massif, from at least as far as Mas Crema near Mollégès at 55m altitude, but possibly from further SE in the foothills of the Alpilles. This aqueduct follows the Northern and Western flanks of the Alpilles through St Rémy and St étienne to Fontvieille, and then the West and South side of the massif of Défend de Sousteyran towards the Vallon des Arcs.
The second, Eastern aqueduct branch, known as the the Caparon Aqueduct derives from several sources on the South side of the Alpilles and runs over 11km to the Vallon des Arcs. Although no remains have yet been found at potential sources, the following ones are the most likely candidates. The furthest sources must lie in the valley of Entreconque since ruins of the aqueduct were found near Mas de Flandrin at the Southern exit of the valley. Here, the sources at les Arsacs just North of the Rochers d'Entreconque and South of Mas de la Dame were probably tapped. The other sources may have been at Manville near the present golf course, and at the Fontaine d'Arcoule further West, probably with a small dam. The last two branches seem to have joined in la Burlande in a distribution basin that has been excavated there. The Caparon aqueduct then passed through Paradou along the South flanc of les Défend de Sousteyran to the Vallon des Arcs. Work by Guendon (2005) and Leveau (2005) on the distribution basin found at the site where the two main branches meet has shown that the distribution of water towards the city and the mill has changed with time.

Changes in the distribution of water of the Eygalières and Caparon aqueducts with time. After Guendon (2005)

The original setup of the aqueduct seems to have been the Caparon branch and a branch from the West near Fontvieille, possibly from the Auge valley. This was built in the middle of the First Century to provide the city of Arles with water. The two branches joined in the distribution basin and crossed the Vallon des Arcs on a single bridge built of large blocks of limestone, the present Western bridge. After 50-75 years, in the beginning of the Second Century, the setup was changed. Apparently, there was a problem with the Caparon branch in that it either did not provide enough water all year round, or water of poor quality. This may be due to the quality of the sources and the fact that this aqueduct has a very gently slope.

At this point, some brilliant mind suggested the option to use the water of the Southern branch for industrial means by feeding a water mill complex. As far was we know, this was a unique idea; no other water powered factory on this scale is known in the entire Roman Empire, and we have to wait more than a thousand years before similar large industrial complexes were again attempted. The two branches of the aqueduct were separated; a new bridge was built next to the existing one on the East side, the limestone ridge was transected and the water mill complex of Barbegal was built. One may wonder why it was necessary to build an expensive, second aqueduct bridge over the valley where one large bridge would have been sufficient; in Rome, after all, up to three aqueduct channels are carried by a single bridge. It may be that the difference in height between both channels was too small, and that raising the last section of one of the aqueducts sufficiently for a combined bridge was no option. Another possibility is that the water of the Caparon aqueduct was supplemented with sediment-laden and polluted water from the valleys around les Baux for use in the mills, and that it was feared that this could contaminate the water supply of Arles.
Approximately 60 years later, more adaptations to the complex were apparently needed. Part of the water of the Eygalières Aqueduct was fed to the mills as well, either to increase capacity or to compensate for falling input from the Eastern branch. Finally, some of the water of the Northern aqueduct was diverted out of the distribution basin to the West for some unknown purpose.

Cross-section of the Roman Barbegal mill (S of France, E of Arles)

A look at the map shows that the Arles aqueduct system is a giant structure that drained much of the available spring water of the region to the city and the Barbegal mill. It even enbraces the territory of another independent town, Glanum, at the North side of the Alpilles, which had its own aqueduct system. The furthest branches of the Eygalières aqueduct are also close to the Durance river and the city of Cavaillon, the Roman town of Cabellio, which had its own aqueduct derived from the Fontaine de Vaucluse.

Visible remains

[The photo's of quite a few visible remains are presented below]

Eygalières aqueduct
There seem to be several sites on the North side of the Alpilles where minor remains of the aqueduct channel are visible. The best remains, however, are on the South side along the road that runs along the North side of the Vallon des Arcs. Each of the narrow valleys in the Défend de Sousteyran has minor aqueduct bridges in petit appareil, some with a single arch, others with culverts. Sections of the aqueduct conduit, in concrete poured into formwork, are also visible here.

Caparon aqueduct
The first and most Easterly part of this aqueduct was found in the Quartier Flandrin of Mausanne at the Southern end of the Entreconque gorge. However, no remains are presently visible here or in the source area. However, a section was uncovered in 1988 and can be seen in la Burlande (Paradou) near the local museum.

The conduit 69 to 61 cm wide and 55 cm deep built of U-shaped limestone blocks, originally covered with flat slabs of limestone, some of which are preserved on site. The holes for fixing the slabs are still visible. The Via Aurelia locally crossed the channel, and the imposts of the arch of the bridge are still in place. The bridge was 7,8 m wide.
A small section of the aqueduct is also preserved in a holiday home in the Résidence Garrigue in Paradou.

Part of the vault of the conduit is preserved just before the Caparon aqueduct reaches the (now covered) distribution basin where it meets the Eygalières aqueduct.
Vallon des Arcs
The double aqueduct in the Vallon des Arcs is the most spectacular structure of the Arles aqueducts. Two parallel aqueducts up to 7m high cross the Vallon des Arcs over 317m from the (now covered) distribution tank in the North to the top of the Pène ridge in the South. The two bridges are 70 cm apart, bus locally the gap has been filled to increase stability.

Vallon des Arcs and trace of the aqueducts. After Leveau (2005) (aqueducts in green)

The structure can be seen on both sides of the D82 road. Parking is possible East of the ruins along the road. The structures starts North of the road on the North side of a drainage channel.

The Western bridge, also known as bridge A (A for Arles) has a Northern retaining wall of 52m long, a 251m long stretch of 36 arches and a Southern retaining wall of 14 m. The spacing between arches is of 25 roman feet (7.3 m), their diameter between 3 and 4.2 m. The aqueduct bridge was originally built from giant single limestone blocks. The piers are 2,4 by 1,8 to 2m and built in "grand appareil" from blocks with dimensions 1.50 x 0.80 x 0.55m. The use of monolithic blocks is typical for early, first century building techniques. The voussoirs are also single blocks of 7 roman feet (2.10m) long. Bridge A was completely reconstructed some time after bridge B was built, probably because of settlement of the foundations in the marshy valley and major damage to the suprastructure. The foundations and imposts of the arches built of large blocks were kept on the North and South sides, but not between piers 9 and 35 (numbering downstream). The reconstructed piers and imposts are concrete with large facing blocks, mounted by concrete piers, spandrels and arches faced with courses of small limestone blocks, known in France as "petit appareil". These alternate with leveling courses of red tiles (opus vittatum). This solution was obviously much cheaper than rebuilding with monolithic blocks in the original style. The use of concrete also reflects the more modern approach to aqueduct building after the first century AD.
Some piers have buttresses on the Western side. The specus is 90 x 120cm. Sinter deposits are about 15 cm thick and reach 20cm high, indicating a relatively low water level in this conduit.

The Eastern bridge, also known as bridge B (for Barbegal) was built 50-75 years after the Western bridge, probably during the reign of Trajan. It has a Northern retaining wall of 69m, a Southern wall of 52m, and a 196m long section of 28 arches. It is completely built in concrete with "petit appareil" facing, alternating with leveling courses of red tiles, although the base and imposts of the piers are in concrete faced with large blocks. The specus is 80 x 100 cm. Sinter deposits in the conduit are visible up to 1 m high and are up to 20 cm thick at the base. The slope of bridge B is much steeper than that of bridge A; this is well visible at the Southern end where both speci are clearly at different level. The water level in the Eastern mill conduit, however, was lower than that in the Western Arles conduit and both had approximately the same debit of 20.000-23.000 m3 per day.

The lower part of specus both aqueducts is well preserved in many places along the bridges and lined with opus signinum, with the well known quarter circles of opus signinum sealing the corners.
Barbegal mill complex
The ruins of the water mill are presently rather overgrown. It consisted of two rows of eight mills in series, separated by a wide staircase. The site is 61 m long and 18,6 m high, with a mean slope of 17 degrees. The water fell on overshot waterwheels 2,1m in diametre and 70 cm wide over a total height of 20m and into basins several of which are well preserved on the East side of the complex, covered with thick sinter deposits. After passing the mills, the water was probably used for irrigation. The rotation of the waterhwheels was transferred to vertical shafts in the mill buildings, each driving a conical basalt millstone of 90 cm diameter. The mills were built in the beginning of the second century and operated till late antiquity. It has been estimated that the mills could produce several tons of flour per day.
Many published models of the Barbegal mills are erroneously drawn with a steep slope of 30 degrees, based on an error in the original publication of the excavations by Benoit (1940); he meant 30% (17 degrees), not 30 degrees.

Chaînon de la Pène
The conduit of bridge A runs into an excavation in the limestone of the Chaînon de la Pène and turns abruptly West to run along the North side of the Chaînon where it can be followed over several hundred metres. The conduit is partly buried, and runs on a wall for a short stretch.

Baux Valley bridge
On the South side of the Canal de la Vallée des Baux on the grounds of the castle of Barbegal are the ruins of a tall pillar, the last visible remains of an aqueduct bridge 800m long and up to 20m high that carried the specus over the marshy valley of les Baux. This valley was a lake in the 17th century, but probably dry in antiquity.

After following the West side ofthe Crau plateau, the aqueduct crossed the depression of Pont de Crau close to the 18th century Craponne aqueduct bridge of Pont-de-Crau; this is NOT a Roman aqueduct!
Remains of the Roman aqueduct are again visible in the Boulevard Emile-Combes in Arles below the ramparts. The aqueduct has been cut by this Boulevard and occurs on both sides of the road, in the outcrop below the walls on the West side, and better even in the outer wall of the cemetery. A think deposit of sinter is visible in the specus.
A branch of the aqueduct conduit is again visible along the West side of the amphitheatre, where the conduit lies in an excavation in the bedrock.

The Roman aqueduct in Arles (aqueduct in green) including the branch towards Trinquetaille crossing the river Rhone by means of an inverted siphon

Cees Passchier
The photographs for this article were made in August 2006


Item Info
Length Eygalières / Caparon 51 / 11 km
Cross-section at Vallon des Arcs 0,90 m x 1,20 m
Volume (Arles) 22.500 m3/day
Fall of the Eygalières branch 0,07 %
Period Claudian - Trajan
  • two coupled aqueducts
  • one intended for 16 industrial grain mills

Recommended literature :
  • The building of the Roman aqueducts: financial and technological problems. The case of the Arles aqueduct. Raffard et al., in Jansen 2000
  • E. Blanchet, Aqueducs romains. Essai de recherches détaillé du tracé de l'aqueduc nord des Alpilles, 1999.
  • E. Blanchet, Aqueducs romains. Trajet de Barbegal à Arles, 2001
  • Dépôts carbonatés et fonctionnement des aqueducs romains: le bassin amont du vallon des Arcs sur l'aqueduc d'Arles (Bouches-du-Rhône), p. 87-96. Jean-Louis GUENDON avec la collaboration de Philippe LEVEAU
  • Le pont de Barbegal au vallon des Arcs à Fontvieille (Bouches-du-Rhône): étude archéologique de la dérivation de l'aqueduc d'Arles, p. 97-105. Philippe LEVEAU and Robert THERNOT
  • Arles: Gallula Roma - das Rom Galliens. Meike Droste, Zabern, Mainz.
  • Arles antique. Guides archéologiques de la France. Heijmans, M, Rouquette; J-M, Sintès, C. Monum.
  • Report of the Wright Paleohydrological Institute
  • Les aqueducs d'Arles : présentation générale. Blanc, Montmajour
Recommended website :
How to visit : Vallon des Arcs and the Barbegal mills: from Arles NE with the D17, a few km before Fontvieille to the right (D82). From Fontvieille with the D31 to the South (see maps).
HOME More literature on more aqueducts Last modified: February, 2007

Roman amphitheatre Arles

Roman forum in Arles

The aqueducts of Arles

Vallon des Arcs

Distribution basin

Bridge of the Eygalières

Bridge of the Eygalières

Culvert of the Eygalières

Inside the conduit

Crossing with Via Aurelia

Imposts of the road bridge

Inside holiday home

Vault of the Caparon aqueduct

Duo aqueduct in the centre

Vallon des Arcs

Bridges A and B

Arch of the older bridge A

Bridge in opus vittatum

Restored arches of bridge A

Opus signinum

Chaînon de la Pène

Cutting and bend

Ruined bridge

Bend in the conduit


The Barbegal mill site

Chamber for a waterwheel

Near the Castle of Barbegal

The Craponne aqueduct

Map of Roman Arles

Conduit in Arles

Conduit in Arles

Conduit near the cemetery

Near the amphitheatre

Roman bath building