Water for Paris' Left Bank

II-ii Water for the Left Bank

Sources from the South

The Seine was much different in pre-historical times than it is today. Torrential rainfalls once had swelled it to a width of five kilometres and gave it a strength that sharpened its curve towards the north, but when the climate became less humid it gradually began to recede towards the form it has today. The Seine had two arms at one point of this evolution: one followed today's river, and another slower and shallower branch, the remains of its sharper curve, left the first to the south of the Bastille, went around today's 3rd and 4th arrondissements, and returned near the palace of Chaillot. The land between its two arms was still a swamp, but it had bared a few islands in its most southerly one. It on the largest of these that the Parisii tribe lived until 52 BC, and it was on the Left Bank that their conquerors, the Romans, built their village of Lutèce.

The river was more than an abundant fresh water supply for both the Parisii and the first Romans. Lutèce had begun as a country village for retired soldiers, but the wide reputation of its cleanliness and peaceful situation in the hills soon made it into a town. Its popularity drew nobler ranks and larger edifices, and by the early third century it had become a small city. As well as theatres and arenas, baths were an essential point in Roman culture, and these were fed by naturally flowing water from higher lands. Romans had known about tapping sources 260 years before their arrival to the Parisian basin, and its first baths and residences were fed by the springs from nearby hills.

Paris' first major aqueduct was between the Thermes de Cluny area and sources near the town of Wissous around 12 kilometres to its south. Traces of this construction unearthed over the years allowed a reconstitution of its path, but the date of its construction is uncertain: some historians think that the aqueduct dates from the Emperor Juilen's rule of Lutèce from 355 AD, but others think that it was built at an earlier period, around 300 AD, when the city saw a rapid expansion under the rule of Constance Chlore. The vestiges of the Thermes de Cluny (Roman baths in central Paris), on the other hand, date from around 215. A possible theory is that the aqueduct of Arcueil was built for the Baths in the early 3rd century, damaged or destroyed by Barbarian invasions (from the south) towards the end of the same period, and rebuilt during the time one of Paris' before-mentioned rulers.

Aqueducts from that time weren't made in the same way as roman roads, that is to say arrow-straight: From the altitude of the basins used to collect the source water, the covered stone trough followed the contours of the land whenever possible to maintain a steady grade until its destination, and it crossed bridges only where a valley crossing was inevitable. Paris' Roman aqueduct measured over 16 kilometres in length, and crossed a bridge over the Bièvre river valley to the north of the town of Arcueil. The name of the latter town, because it held the aqueduct's most remarkable feature, later became the "dit" of the entire man-made waterway.

The trough itself was ingenious for the time. It was a thick "U" of sand-based cement, but the inside was smoothed with a layer of mortar. This in turn was waterproofed with a thin layer of fine mortar mixed with crushed tile. The trough was then mortared shut with slabs of rough cement; the completed conduit, completely closed on all sides, thus formed a "pipe" that maintained the purity of the water it carried. Sometimes aboveground, but more often under, the Roman aqueduct followed natural land formations northwards at a steady grade averaging 1.59 metres per kilometre.

The exact path of the Roman aqueduct was uncertain until the mid-19th century. Today we know that, within Paris, it followed today's rue de la Tombe Issoire, Faubourg Saint-Jacques, and Saint-Jacques until below the Musée de Cluny (old Roman Thermes), where it branched out to feed different thermal installations in several properties. The derivations towards the river are still a mystery today, because even if they had been uncovered in short lengths in many places, central Paris' many centuries of construction and reconstruction has eliminated both the aqueduct and the buildings it served.

Marie de Medicis

Historians think that the first aqueduct was abandoned after the fall of the Roman Empire. This could have been for several reasons: the lack of knowledge as to the aqueduct's exact path which made it impossible to maintain, or possibly the late third-century Barbarian invasions from the south which forced Parisians to abandon the Rive Gauche to centre their town on the islands and the Rive Droite. The Left Bank's lack of water may even have accentuated the population's permanent move to the north: wells were the only source of inland water then, and these were easier to dig in the softer ground to the other side of the river. In any case, with its hard ground and un-walled roman ruins, the Left Bank remained practically barren until the early eighth century.

The left bank eventually became a centre for universities and ecclesiastic institutions, and these in developing had attracted immigration to the land around them. Both its population and industry were in a rapid expansion by the time of Henri IV, who was the first to plan an eventual renovation of the Roman aqueduct. Unfortunately, the king's version of the project ended with his assassination by Les Innocents in 1610, so it was continued by his wife, Marie de Medicis, but more to the aim of serving the Palais de Luxembourg being built under her orders then.
The new version would follow the same path as the old, but completely underground, and it would tap sources near the town of Rungis, about three kilometres north of Wissous. On the plans drawn by the architect Jacques Desbrosses, the same as the Palace de Luxembourg, the total length of the conduit from its sources to a "chateau d'eau" below the Observatoire would be 12,956 metres, accessible from the surface through with 30 stone "regards" spaced every 300 "toises" (384.71 metres). These regards were small stone houses with stairways down to the aqueduct; a well at the stair bottom collected arriving sourcewater from a mini-waterfall and relayed its overflow to the next regard, a system which 'filtered' water of its solid content as well as 'freshening' it by an extra exposure to air.
The aqueduct itself is an inverted 'U' of masonry that lies just below ground level in most places. It averages 2.2 metres wide and 2.7 metres high in its outer dimensions; its inner channel is of roughly the same form, with a width of around one metre and a height of 1.4 metres, but its floor has a central groove of forty centimetres in depth and width. The aqueduct between each regard had removable stone slabs in its ceiling, numbering 130 in all for the whole project- these were covered under several metres of earth, but their position was marked on the surface by a stone marker, or "borne."

The first stone of the Rungis main collector regard (dit: prise des eaux) was posed in a ceremony, attended by Louis XIII, on the 17th of July 1613. Its contractor, Jean Coin, died in 1619, and a certain Jean Gobelins (one of the founding brothers of the Royal Tapestry manufacture) took his place until the end of the work. The aqueduct's plans seemed to have changed somewhat in the meantime, because upon its completion it had only 28 regards, but 258 "borne" maintenance accesses. The chateau d'eau designed by Jean Gobelins had three reservoirs that divided water between the Royal palaces, the City of Paris, and the contractor.

Water filled the chateau d'eau for the first time on the 18th of May 1624. The royal family was served from that day, but Paris had to wait until the completion of 14 new fountains on its Left and Right bank. This was done in 1628, but in the meantime the nobility and religious institutions had already claimed most of Rungis's water. In all, the people of Paris had a right to less than ten percent of the aqueduct's output, even if their tax product had paid for the totality of its construction.

In all, the aqueduct gave disappointing results, as its water was not near enough to provide for a growing city. New sources almost doubled its output from 1651, but this soon began to drop. The fact that the aqueduct lie over old stone mines was still unknown then, but before long the conduits began to suffer fissures and cracks that let its water drain to the caverns below. By 1671 its output was down to two thirds of that at its time of construction.
The Administration des Carrières conducted consolidations to the underlying mines from 1777, but by then the conduit's degradation was too far advanced to return it’s to its original state. Two civil engineers, Mary and Lefort, made one last effort to recuperate the aqueduct in 1845 in building a secondary reservoir to collect the night-time flow for daytime distribution. This solution was only temporary, as the aqueduct was completely abandoned after other solutions for the Left Bank's water problems appeared in later years.


Belegrand was an engineer with a leading role in Napoleon III's renovation of the Capital. His work, primarily improving its sanitary conditions, led him to begin studying the possibility of a new aqueduct system from 1854.

Pre-annexation Paris covered a circle of land that began to rise only towards its exterior. Its sources of water, in diminishing importance, were the Canal de l'Ourcq's "Aqueduc de Ceinture" that fed fountains, the pumps along the river Seine, the Grenelle artesian well, Aqueduct d'Arcueil, and les sources du Nord. The first was the largest and next-to-cleanest but served the Right Bank population only, the last two were the purest but their production was of little importance, Grenelle's water was also pure but not very fit for consumption, and the polluted Seine water was used only as a last resort.
The Capital's Aqueduc de Ceinture gave an abundant amount of water in proportion to its population. The conditions in the suburbs were far below this, for they still depended for the most part on water brought to reservoirs by horse and wagon - and per person their share was only one quarter of that of a typical Parisian. These situations became one from Paris' 1860 annexation of the suburbs, which of course lowered the average Parisian per person water share. In reality this problem was irrelevant, because Belgrand considered almost all of the Capital's water as undrinkable.

Paris' former suburbs were not only far from its centre; they were also on higher ground. The bassin de La Villette (aqueduc de Ceinture source) was 27 metres above the summer level of the Seine, which was high for the Capital's centre, but too low to carry water to the growing populations of Montrouge, Belleville, and Montmartre. If Belgrand was to provide the whole of "new Paris" with clean water through aqueducts, he had to find sources at higher altitudes. Belleville's high population made it a priority, so its aqueduct was the first to begin in 1863 from the river Dhuis to a reservoir in Menilmontant. For the time being, from the same year, the 17th, 18th, 19th, and 20th arrondissements had to make do with a new river steam pump at the pont d'Austerlitz.

The Left Bank remained as it was until the aqueduct and reservoir "de la Vanne" began in 1868. Its water came from two small tributaries to the river Vanne over 173 kilometres to the south of today's Paris, and was collected in a reservoir, dit: "de Montsouris," next to the park of the same name.
This aqeduct, though much longer than those built by the Romans and Marie de Medicis, shared the same bridge over the Bièvre valley. The viaduct's first versions measured eight metres at its highest point, but this was too low to carry the sources from much higher altitudes. With Belgrand's project, the bridge gained another 35 metres - a series of colossal arches, built with stone from the mines of Arcueil, were added to the viaduct of Marie de Medicis. As for the original Roman bridge, only two of its arches remain from the two renovations.

The Montsouris reservoir, built at the same time as its viaduct d'Arcueil, was an equally colossal construction. Its proposed site to the west of the Parc Montsouris suffered from the same problem as much of Paris' southern districts - its location over abandoned stone mines. The Inspection des Carrières (formerly the "Administration des Carrières") were called to consolidate the underground tunnels that would carry the prodigious weight of the reservoir and its 300,000,000 litres of water.
The reservoir covered a rectangle of 36,640 square metres. Its walls and 1,860 columns were repeated in the underground; the latter continued upwards thirty metres to the surface with wells filled with mortared masonry. Any fontis were opened to the surface through additional wells, and after the construction of supporting walls and arches in the cavern below, were filled with cement. The strength of the underground consolidation was slightly exaggerated in regards with the weight they would carry, but this just served to provide extra security against an eventual collapse. This work began in September 1868, but stopped in March the following year, first because of budgetary problems, then the wars of 1870-71. It began again from June 1872 with the construction of the reservoir itself.
At its time of its completion, the 12th of August 1874, this reservoir was the largest covered construction of its type in the world. Its two levels supplied all of the lower quarters of Paris on both its Left and Right banks as well as a secondary reservoir in Passy, and permitted the installation of running water in apartments as high as 50 metres above sea level. It officially entered service on the 12th of August 1874.

The Montsouris reservoir was too low to serve the upper floors of the apartment buildings that later appeared around it. Water for these came from higher Menilmontant, as did the water for the higher buildings of the Pantheon and Place d'Italie quarters.

The new System

Belgrand had also developed a modern sewer system for Paris, and the walls of its underground tunnels also carried the city's water system. There were two types of water from then: one was the purer "usage privet" water from the upper level of the Menilmontant reservoir and the Montsouris reservoir, and the second, was a "usage public" system for impure water fed by the canal de l'Ourcq, the Seine and Marne rivers, the artesian wells, and the old aqueduct d'Arcueil. The second system, raised by pumps and piped to smaller secondary reservoirs throughout the capital, was used for decorative fountains, watering parks, and the cleaning of streets and sewers.

The expansion of Paris' 19th century water system continued even after the death of Belgrand in 1878. A drought and cholera epidemic in 1881 and 1884 underlined the need for a greater water supply for the continuously growing Parisian population, and motivated new aqueducts: The "Vigne and Verneuil" aqueduct brought the river Avre to a reservoir to south-western town of Montretout, near Saint-Cloud, from 1893, and the "Loing et Lunain" aqueduct added its water to the Montsouris reservoir from 1898. Paris' last regions to be served were its two highest hilltops: twin towers by the Belleville cemetery distributed water from the Menilmontant reservoir, and the upper level of single water tower next to the church of Saint-Pierre pumped its water from the newer Montretout reservoir. After the creation of riverwater filtration plants in Ivry and Saint-Maur to fill reservoirs in case of drought, Paris' water conditions were well assured at the end of the 19th century.

As for the old Arcueil aqueduct, it had been cut at the fortifications since the 1860 annexation. Its water was at first derivated to a "usage public" reservoir near the Pantheon for the cleaning of sewers, but found a more noble use when the Parc Montsouris claimed it for its fountains and gardens from 1904.