How Did Roman Aqueducts Work? Gravity, Patience and Very Small Slopes

The whole system depends on a gradient steep enough to keep water moving and shallow enough not to waste height, because height is the resource: once it is spent the water cannot be lifted again. Roman aqueducts typically fell somewhere between about twenty centimetres and a few metres per kilometre, and some sections were far gentler, with a stretch of the aqueduct at Nimes in France falling by about seven centimetres per kilometre, a slope of roughly one in fourteen thousand that had to be maintained across rough ground. Surveying was done with a chorobates, a long levelling table with a water channel and plumb lines, and a dioptra for angles, and the accuracy achieved with those instruments over those distances is the part engineers still find remarkable. Where the ground fell away the channel was carried on arches or a bridge, where it rose the route went through a tunnel or around the obstacle, and the great majority of every aqueduct, typically more than eighty percent, ran underground in a covered masonry channel, which is why most Roman aqueducts are invisible.

A complete aqueduct is a chain of components, each solving a specific problem:

• •A spring house at the source, capturing water and excluding debris
• •The specus, the channel itself, usually rectangular, lined with waterproof mortar containing crushed ceramic, covered to keep out sun and contamination and fitted with access shafts for cleaning
• •Settling tanks at intervals, where the flow slows so that sediment drops out and can be removed
• •Bridges and arcades where a valley had to be crossed, which were built only when the alternative routes were worse, since arcades are expensive and vulnerable
• •Inverted siphons for valleys too deep to bridge, using lead or stone pressure pipes running down one side and up the other, which work because water rises to nearly the height it fell from, and which required pipes able to withstand substantial pressure
• •A castellum divisorium at the city end, a distribution tank splitting the flow between public fountains, baths and private customers, with the public supply given priority by the geometry of the outlets
• •Lead and ceramic pipes carrying water through the streets, with taps and fittings, feeding fountains that ran continuously because the system had no valves to stop it

The common assumption is drinking water, and the volumes involved make clear that it was not only that. Rome was supplied by eleven aqueducts delivering an enormous daily flow, and the majority went to public baths, fountains and the flushing of sewers rather than to household consumption. The baths were a civic institution requiring continuous water, heated and unheated, at a scale that dwarfed domestic needs, and the constant overflow from fountains washed the streets and carried waste into the drains. Private connections existed and were taxed, granted by imperial permission, and illegal tapping was a persistent problem, described in detail by Frontinus, the water commissioner who wrote a treatise on the system around 100 CE and complained at length about officials colluding with users to divert supply. That document is the single best source on how the system was administered and shows a bureaucracy with inspectors, maintenance gangs and legal powers over land near the channels.

Roman pipes were frequently lead, and the idea that lead poisoning contributed to the empire's decline has circulated for a century. The evidence does not support it strongly. Water in the system flowed continuously rather than standing, and the hard water deposited a thick layer of calcium carbonate inside the pipes, which insulated the water from the metal, a scale so substantial that archaeologists use it to estimate how long a pipe was in service. Lead exposure from cooking vessels and from grape syrup boiled in lead pots was probably a larger contributor. What did end the aqueducts was maintenance: the channels require constant cleaning and repair, and when the administrative capacity disappeared, so did the supply, which is why the population of Rome collapsed from around a million to a few tens of thousands and settled near the river. Several aqueducts were restored in the Renaissance and one, the Aqua Virgo, still feeds the Trevi Fountain, making it arguably the longest-serving piece of water infrastructure anywhere.

A Roman aqueduct moves water entirely by gravity down a continuous very gentle slope, sometimes as little as seven centimetres per kilometre, surveyed with a levelling table and plumb lines. More than eighty percent of the length typically ran underground, with arches used only to cross valleys and inverted siphons where a bridge was impractical. Most of the water supplied baths, fountains and sewer flushing rather than households. Continuous flow and limescale largely prevented lead contamination.