Ancient Roman Aqueducts: Gravity Engines of the Empire

The Romans moved water without pumps by making sure it never stopped running downhill. Each aqueduct tapped a spring, river or lake higher than the city it served and carried the water in a channel that kept falling, sometimes by only centimetres per kilometre, until it reached tanks that shared it out across town. Most of that channel was buried; the famous arches, and the pressurized pipes, were simply ways of getting across valleys. Rome eventually drew on eleven aqueducts, from the Aqua Appia of 312 BCE to the Aqua Alexandrina of 226 CE. Piped water itself was older: the Minoans of Bronze Age Crete fed spring water to the palace at Knossos through terracotta pipes. Pliny the Elder, writing in the 70s CE, marvelled at "the mountains that have been pierced, the valleys that have been levelled" to bring water to the city. All of it depended on one unglamorous number: the slope.

How Gentle Was the Slope of a Roman Aqueduct?

Gentler than most people would guess. Roman aqueducts typically fell 1 to 3 metres for every kilometre of channel, according to the hydraulic engineer Hubert Chanson. Pliny set the minimum at a quarter-inch of fall per 100 feet, as do some editions of Vitruvius: 1 in 4,800, or about 21 centimetres per kilometre. The aqueduct of Nîmes, in southern France, lived close to that limit. Its builders, working around the middle of the first century CE, needed water high enough to feed fountains and baths by gravity; the city's own spring sat too low. The springs they chose near Uzès lie about 20 km away in a straight line, but hills block the way, so the channel winds for roughly 50 km. Over that distance its floor drops just 12.3 metres, by the Pont du Gard site's figures (older accounts often say 17). That averages about 25 cm per kilometre (roughly 16 inches per mile), around 1 in 4,000, and the water took a day or more to arrive. The fall was not even. The aqueduct's long profile shows a steeper stretch below Uzès and a nearly flat one around the great bridge; in the woods of Remoulins, just downstream, the drop has been put at about 10 cm per kilometre or less. That flat stretch caused trouble: when water was first let in, it overflowed, and the builders raised the channel walls on the Pont du Gard by about 60 cm. For comparison, the stone canal that feeds the fountains of Machu Picchu falls about 3 percent, more than a hundred times the Nîmes average.

Vitruvius describes the surveyors' levelling bench, the chorobates: a wooden straightedge about 20 Roman feet (6 metres) long, with plumb lines hanging over marked crosspieces. When wind set the lines swinging, the surveyor poured water into a five-foot groove along the top and read the level from that. Steep ground called for the opposite trick: stepped chutes or chains of vertical drop shafts that burned off surplus height, as at Recret on one of Lyon's aqueducts.

Why Did Most of the Water Travel Underground?

Because a buried channel could follow the contours of the land and hold its gentle fall without monumental masonry. One tally of the channels serving ancient Rome, in National Geographic, counts about 500 km (315 miles) of channel, of which roughly 430 km ran underground and only about 58 km stood on arches, under 12 percent of the total. The first aqueduct, the Aqua Appia, stayed below ground for all but 60 of its 11,190 paces, by Frontinus's count. Frontinus guessed at two reasons the earliest lines ran so low: their builders were still unsure of their levelling, or they hid the water underground so that enemies, in an age of frequent wars in Italy, could not easily cut it. Burial also kept the water clean and reduced evaporation, notes a guide published by the Pont du Gard site. Tunnels were dug from vertical shafts, often about 70 metres (230 feet) apart; workers hauled spoil up them in baskets and lowered materials down, and openings along the line later let crews in to clean and repair the channel. Inside, masons applied opus signinum, a waterproof render of lime, sand and crushed tile, though at Nîmes only the walls got it. Leaks through the unlined floor soon forced the builders to wall up the open arches of several bridges. The water also fought back. The springs at Uzès carry a heavy load of dissolved carbonate, which settled out as limescale up to about half a metre thick on each wall near the Pont du Gard. By the time the aqueduct went out of use, in the early sixth century, deposits had helped cut its flow from some 35,000 to 40,000 cubic metres a day to under 10,000.

What Happened When a Valley Got in the Way?

Engineers could carry the channel across on a bridge, or let the water drop through sealed pipes and climb out the other side. The Pont du Gard is the first at its most extreme: three tiers of arches lift the Nîmes channel about 49 metres above the river Gardon, making it the tallest Roman aqueduct bridge still standing. The second, the inverted siphon, relies on water in a closed pipe rising back almost to its starting height. Lyon's 86 km Gier aqueduct had four. At Beaunant, water left a header tank about 314 metres above sea level, plunged into the Yzeron valley, roughly 1.5 km wide and 125 metres deep, crossed the river on a bridge some 290 metres long and rose to a receiving tank at about 306 metres, 2.66 km on. Archaeologists think about 11 lead pipes ran side by side; at the bottom, the water pressed on them at more than ten times atmospheric pressure.

Lead on that scale was costly: the Gier's siphons are estimated to have used 10,000 tonnes or more. The archaeologist A. Trevor Hodge argued that the price of producing and hauling lead is why Romans built bridges wherever they could, a view other specialists still debate.

Who Kept Rome's Water Flowing?

In 97 CE the emperor Nerva appointed Sextus Julius Frontinus curator aquarum, Rome's water commissioner, and Frontinus wrote down how the system worked. His treatise, finished around 98, records each aqueduct's builder, date and length, how much of it runs underground or on arches, and what it was meant to carry. He then checked the books against the water. Gauging the intakes himself, he found about 10,000 more units of flow (quinariae) entering the system than the official records allowed for. He also caught the system's own water-men diverting public water to private users, landowners tapping the conduits, and illicit branch pipes bored under the streets to supply businesses. The upkeep fell to two gangs of slaves: about 240 men paid by the state treasury and 460 paid from the emperor's purse. Frontinus even defended a certain amount of waste. Overflow from the reservoirs, said an imperial order he quoted, kept the city healthy and flushed the sewers, whose great trunk drain, the Cloaca Maxima, emptied into the Tiber. He was not modest about the result. Set these indispensable works beside "the idle Pyramids or the useless, though famous, works of the Greeks," he told his readers, and compare.

How Much Water Did Rome Actually Get?

No one knows exactly, partly because Frontinus's unit, the quinaria, described the size of a pipe rather than how fast water moved through it. Converted into modern units, his total comes to a little over half a million cubic metres a day by the archaeologist Philippe Leveau's reckoning, while some estimates for all eleven aqueducts at their peak exceed a million. Measurements from the channels themselves point lower. In 2015 a team led by Duncan Keenan-Jones of the University of Glasgow used the limestone crust inside the Anio Novus to estimate its flow at about 1.4 cubic metres a second, roughly 120,000 cubic metres a day, well short of traditional estimates for that aqueduct. His verdict: Rome was well supplied, but less lavishly than often assumed. It was still enough for 591 public basins, by Frontinus's count, and perhaps for spectacle: Cassius Dio claims Titus once flooded the Colosseum for a mock sea battle, a story historians still argue over. One line is running yet: Agrippa's Aqua Virgo, opened in 19 BCE and named, Frontinus says, after a girl who pointed soldiers to its springs, still carries water, mostly underground, to the Trevi Fountain. Strip away the arches and a Roman aqueduct is a long, patient ditch: a line surveyed to lose height a few centimetres at a time, roofed over, sealed and cleaned for centuries. The monuments in the photographs are the exceptions. The real achievement was the slope.