← All articles

Mikhail Polyakov

How Roman aqueducts made water go uphill: the inverted siphon at Aspendos, the Archimedes screw and water wheels

Level: BeginnerReading time: 7 min

Topics: History, Science

A hilly valley with a stone water channel, two stone towers in the middle of the valley and a tank on the opposite slope

Short answer: Roman aqueducts mostly carried water downhill, under the force of gravity. Where a valley lay in the way, they sometimes built an inverted siphon: a closed pipe that went down into the valley and back up on the other side. The water in it rose because the outlet stood slightly lower than the inlet, and the head of water from above pushed it upward. This is how the aqueduct at Aspendos in what is now southern Turkey was built, for example. At the same time, according to a researcher interviewed by Live Science, Roman engineers avoided siphons where possible and preferred to build a bridge. And to lift water from a river onto a field or pump it out of a mine, the Romans had machines: the Archimedes screw, water wheels with buckets and a hand pump associated with the name of Ctesibius.

Where the question comes from

The Live Science article came out on October 11, 2026, in a section where the editors answer readers' questions. The author is Owen Jarus. Aqueducts, he writes, are among the most prominent Roman structures that have survived to this day. Water flowed along them gradually downhill, toward cities. But some aqueducts, like the one at Aspendos, had sections where the water had to rise.

The inverted siphon at Aspendos

The aqueduct at Aspendos was built in the second century AD. Scholars reconstructed how it worked, including with a computer model. It was described in detail by Paul Kessener, a researcher at Radboud University in the Netherlands, in a 2000 article in the Journal of Roman Archaeology. The arrangement, as Live Science retells it, was as follows:

  1. On one side of the valley stood a header tank, which water reached through an ordinary channel.
  2. On the other side of the valley stood a receiving tank, slightly lower than the header tank.
  3. Between them, in the valley itself, two towers were built.
  4. The water dropped down a steep slope, and the resulting pressure lifted it up the first tower. Then it dropped steeply down the other side, and that head of water allowed it to climb the second tower. Finally, the water reached the receiving tank and continued on to the city.

Why this works I will explain in my own words; this is school physics, not a retelling of the article. In a closed pipe completely filled with water, the water tends to settle at the same level at both ends, as in communicating vessels. If the outlet is lower than the inlet, the water will flow even if along the way the pipe goes down into a valley and up again. If the outlet is raised to the level of the inlet, the flow will stop. The price of this solution is pressure: for every 10 meters of depth, the water presses on the pipe walls about one atmosphere harder. The deeper the valley, the stronger the pipe has to be at its lowest point.

The main danger: air in the pipe

It was important to keep air pockets from forming in the pipes. Giovanni De Feo, a professor of industrial engineering at the University of Salerno in Italy, explained to Live Science how this was handled. The header and receiving tanks were made relatively large. When the flow entered a wide chamber, it slowed down, and air bubbles had time to rise to the surface and escape through the open top. This reduced the risk of air pockets.

Why an air pocket is dangerous is also my explanation. A bubble of air at the high point of a pipe takes up part of its cross-section and gets in the way of the flow. If many such bubbles build up, the water flows worse or stops.

Why the Romans preferred a bridge

Charles Ortloff, a research associate in the Department of Anthropology at the University of Chicago with a background in both engineering and history, told Live Science that the Romans tried to avoid siphons. When they came to a valley, Roman engineers, in his words, preferred to build a bridge with a channel on top instead of a siphon. An example of this solution is the aqueduct at Segovia in Spain, whose bridge survives to this day.

Why, Ortloff does not explain in the Live Science account. A possible reason follows from the physics above; this is my assumption: in a deep valley, siphon pipes are under high pressure, while a bridge carries an open channel where the water also presses on the walls, but less: the pressure depends on the depth of the water, and in a channel it is small.

How water was lifted without a siphon

A siphon carried water across a valley but did not lift it above its source. To raise water from a river onto a field or out of a mine, machines were needed. Live Science names three.

The Archimedes screw. It was sometimes used to lift water from rivers onto fields. Its invention is attributed to the Greek Archimedes, who lived in the third century BC. But, as Live Science writes, historians Stephanie Dalley and John Peter Oleson noted in a 2003 paper that the screw may have appeared earlier. According to Ortloff, a screw turned by animals could lift water from river level to higher areas of land. Live Science quotes a description from the website of the Kotsanas Museum of Ancient Greek Technology: a wooden shaft onto which thin, flexible willow or wicker branches were wound one on top of another, producing an endless screw. The screw was placed inside a wooden pipe and set in the water at an incline of 30 degrees. When the screw was turned, it lifted the water.

The water wheel. Such wheels were also used in Roman mines. According to a 2024 scientific paper as retold by Live Science, it is a large wheel, resembling a carriage wheel but bigger, with buckets fastened to it that scoop up water.

Ctesibius' pump. It was operated by hand, by swinging a wooden lever. According to the same paper, the pump had two cylinders with pistons connected to a central chamber, and the water came out through a nozzle attached to that chamber.

An experiment you can repeat at home

This is my suggestion, not part of the article. The principle of the inverted siphon can be seen in the kitchen.

  • What you need: two unbreakable plastic containers, for example cut-down bottles, a clear flexible tube about a meter long, water.
  • How to do it: do everything in a sink or bathtub. Put the container with water on a stable, wide support, for example an upturned saucepan, and the second, empty container on the bottom of the sink next to it. Fill the tube completely with water under the tap, pinch both ends with your fingers, put one end into the upper container and the other into the second one. Lower the middle of the tube down to the bottom of the sink: this is the "valley." Release the ends: the water will go down into the "valley," rise over the rim of the lower container and flow into it.
  • How to check you understood the principle: lift the lower container by hand until the water level in it is level with the one in the upper container. The flow will stop, although the tube is the same: the outlet is no longer lower than the inlet.
  • Safety: clean water only, no chemicals, unbreakable containers, do not fill the tube by mouth but under the tap; children should do it together with an adult.

Summary

Roman aqueducts relied mostly on gravity. Where a valley had to be crossed, an inverted siphon was sometimes built, as at Aspendos: the water went down and up through a closed pipe because the outlet was slightly lower than the inlet, and large tanks at the ends let the air out. But according to Charles Ortloff, the Romans preferred bridges, as at Segovia. And machines helped lift water above its source, from rivers and mines: the Archimedes screw, water wheels and a hand pump.

I work on AI agents and automation, and here I sometimes write about science and the history of technology. If you want to look at my projects or discuss your own task, visit my portfolio.

Sources