Engineering Rome

Water Delivery Infrastructure in Ancient Rome

By Cameron Stuart

*All photos produced by the author unless otherwise stated


Visiting Rome allows one to see countless breathtaking creations of the ancient world. From the Colosseum to the Pantheon, people today are still amazed by what Romans were able to make nearly 2,000 years ago. When thinking of the water supply in Rome, people may only consider the Tiber River and the aqueduct arcades visible across Europe. These are both iconic sights, yet they only make up a small piece of the complex water infrastructure of ancient Rome.

Figure 1: Sections of the Aqua Claudia and Aqua Anio Novus still stand today

Dug through the hills, waterproof tunnels continued water flow between the arcades. Collection systems allowed for sediment settling and distribution to various sources with differing pressure. Public baths, toilets, and fountains ran constantly, and waste was carried off through a sewer system. In this piece, we will cover each piece of the aqueducts and distribution system to fully understand the complexity of ancient Rome’s water supply.

History and Background

The Tiber River

Rome’s geography is defined by the great Tiber River, as well as its history. The twin children Romulus and Remus were said to have been left to drown in the Tiber but were saved by a wolf and founded their own city in 753 BC on the banks of the river [1]. From the founding of Rome—named after Romulus, after he killed his brother and became the first king—the Tiber was an important source of water along with wells and cisterns [1, 2]. For over 400 years the Romans acquired water by these more traditional means, although they quickly became polluted by waste which produced the need for alternate sources of water [3].

Figure 2: The Tiber River at dusk in the present day

Today, the river is still an iconic part of the Roman experience. In the summer you can find markets along the lower banks or take a food and drink cruise through past famous neighborhoods. You can cross and observe it from any of the myriad of bridges along its length. These days you can do so much in and around the river aside from drink from it. For centuries, the river has been a critical part of Rome from water source to tourist attraction.

The Cloaca Maxima

Also preceding the aqueducts was a lesser known but still impressive component of water management. The Cloaca Maxima was constructed in the 6th century BC as an open sewer, and later covered in the 2nd century BC [2] which is what we see today.

Figure 3: the Cloaca Maxima outlet today

In Figure 3 the large arch which signifies the outlet of the Cloaca Maxima. It now lies along the bank of the Tiber and fills partially with water when the river is high. Looking closely, inside the arch, you can see a line along the back wall between whiter bricks on the bottom and darker on the top. This is the line remaining from the previous high water. 

Despite nearly 2,500 years and several regimes between its construction and the present day, the Cloaca Maxima still functions in some places and the outlet in Figure 3 still allows wastewater to flow into the river. Like the Tiber River this piece of ancient Rome’s initial water infrastructure has changed over time while remaining in use and an important landmark.


Over roughly 500 years since the construction of the first aqueduct, they were constructed across the entire Roman empire, totaling around 1,600 empire-wide [4].

Figure 4: Locations of known aqueducts from the Roman Empire (from 4)

Above, Figure 4 shows the spread of aqueducts that sustained the Roman Empire at its peak. While the specifics of this discussion will cover the aqueducts that fed the city of Rome, it is important to understand how much the Roman Empire relied on water to sustain itself as it grew across the world.

Figure 5: A map of the 11 aqueducts from their sources to the city of Rome (from 5)

The city of Rome was the recipient of 11 aqueducts which spanned over 420 km in total [3], supplying its nearly one million residents [2]. The first of which was the aforementioned Aqua Appia constructed in 312 BC and the last being the Aqua Alexandrina constructed in 226 AD [2], both of which can be seen in Figure 5.

At its peak, it is estimated that this system could deliver around 144 L of water per resident per day to the city [6]. In comparison, Americans consume roughly 300 L of water per resident per day [7]. This estimate for ancient Rome includes all water on a per capita basis, meaning each individual person used far less than we do today. Most of this water was diverted for public uses anyways and there were far less people pulling directly from the source for their own private use.

Aqueduct Design and Hydrology

As mentioned earlier, the aqueducts served the purpose of delivering water from the mountains to the city. This meant traversing distances of anywhere from 16 to 90 km with only slight elevation changes relative to the distance [3]. Each aqueduct consisted of a string of arcades, tunnels, and siphons to achieve a steady downward slope from the source to the city.

Figure 6: An example of an aqueduct using each component across varying terrain (from 4)

A glance at Figure 6 above quickly shows how each of these components were used depending on the terrain. Imagine the complex planning required to construct an aqueduct like this example across the long, winding terrain such as that traversed by the Aqua Claudia (see Figure 5). While some only imagine the impressive arcades as aqueducts, the entire aqueduct consisted of this system of interconnected parts which relied solely on gravity to deliver water to the city, where it was then distributed to a variety of destinations.


The aqueduct arcades are what one first imagines upon hearing the word “aqueduct” as they are the main visual representation of said infrastructure. These arcades consist of a series of arches atop which runs a covered channel for the water to flow (see Figure 1). Because the slope of the water channel needed to be gradual and constant, arcades were built to raise the surface on which the channel rested to meet the elevation of the outflow of a tunnel, or the spring itself. See Figure 6 for a visual representation of how the arcades maintained the continuity of the aqueduct.

Figure 7: A look at one of the channels of the Aqua Claudia, now exposed

When visiting these in person, it is nearly impossible to tell that they gradually slope downward towards Rome at less than a 1% grade [3]. It’s impressive to imagine how this was measured and executed two millennia ago. The arcades still standing serve as a testament to the ancient Romans’ expertise with construction materials and techniques.

Unfortunately, these beautiful structures were the most troublesome components of aqueducts. Although Romans were known for their construction prowess, this does not mean their technology was equivalent and without error. At times, a pillar that happened to sit over softer ground would sink slightly, thus slowing or stopping the flow of water altogether. To resume flow a footer would need to be installed to serve as a foundation to support the pillar on unstable ground. An example of one of these footers can be seen below in Figure 8.

Figure 8: A concrete footer added after construction to support pillars above soft ground

Additionally, the exposed arcades were very easy targets for either theft or attack. We will discuss the distribution system later, but Romans were very particular about how the water was allocated. Some private residents could pay for a direct source, but most of the supply was for public works such as baths. As in any society some people attempted to skirt this and tapped the aqueducts by diverting water from the arcade and creating their own personal supply without paying [2].

More dire was also the prospect of enemy armies destroying parts of the arcades to cut off water from the city to lay siege to it, as the Barbarians did in the 5th century [8]. Considering this and general wear and tear, the arcades required relatively constant oversight. Looking back at Figure 7 you can see the right arch in the image is supported by smaller, more uniform bricks that stand out compared to the large stones the arcade was initially constructed from. This is just another example of later work required to maintain these structures over the centuries.


The underground work which went into the aqueducts usually remains unknown to the public who admire the grand arches that remain all over Europe. However, the importance of the tunnels cannot be understated. In fact, most of the length of the aqueducts were underground, with only about 20% of the aqueducts traveling above the surface on arcades [9]. Although they were carved through the ground, the tunnel walls in Figure 9 below are clearly smooth. This is because Romans coated the walls with a special concrete called “opus signinum”. It was like other forms of Roman concrete except that it could harden underwater and remained waterproof [10]. This made it perfect for lining the aqueduct tunnels to prevent water seeping out through the ground.

Figure 9: A photo taken from within an aqueduct tunnel

In Figure 9 you can see the interior of one of these tunnels. In the background is a person, standing at full height. This is not a coincidence, as the aqueducts were designed to accommodate people so they could perform maintenance rather than designing for a specific volumetric flow rate. Dimensions varied slightly, but one could expect an aqueduct channel to measure around a meter wide and two meters tall on average [2, 11]. In Figure 10 below you can see an example of a maintenance shaft which would have been used to both dig down to the correct elevation to dig the tunnel as well as service it when necessary.

Figure 10: A shaft leading from the aqueduct tunnel back to the surface

The attentiveness to maintenance was crucial to ensure smooth operations because Rome has hard water sources, meaning there is a high presence of limestone in the water [4]. This is generally a harmless occurrence, but this led to a buildup of solids along the sides of the aqueduct channels. Therefore, easily accessibly tunnels and human-sized flow areas were necessary, as every year several millimeters of limestone could accumulate within these channels [4]. To get a sense of how much this truly is, one can examine the aqueduct in present day Nimes, France, which accumulated almost half a meter of limestone over the course of 200 years [2]. Limestone buildups in aqueducts was obviously undesirable for flow conditions, yet it had one unintended benefit which will be discussed later.


Like arcades, siphons were a method of keeping the water flowing above ground over lower terrain. The difference with siphons was that siphons were used when the drop in elevation between two points where tunnels were used was too great to build arcades. Siphons were instead used, or perhaps it would be more accurate to refer to them as “inverted siphons” because they dip down initially and form a “U” shape. This component of the aqueducts is less known and preserved, yet perhaps had the most complicated technicalities to consider.  

Siphons consisted of a header tank on the uphill side and a receiving tank on the other. These tanks are simply where water was collected to enter or leave the siphon and were connected to regular flow channels on either end. The connection between these tanks was a series of pipes which either crossed the ground or a small arcade of their own, as seen in Figure 11 below.

Figure 11: A closeup look at a siphon spanning a valley (from 12)

It is known that a flowing fluid if diverted down will be able to return to its initial height, which the Romans also knew. However, they also knew that this was not realistic and due to friction within the pipes there was significant head loss between one end of the siphon and the other [12]. This was remediated by simply lowering the receiving end a substantial amount—this is noted as the hydraulic gradient in Figure 11 and is the amount of head loss between each end of the siphon. The longer the siphon, the more head loss would occur as the water would experience friction with the pipe for a longer duration. This is not a desirable occurrence as it requires a drop in water level, but it was often the only way to continue the flow across a large valley where building an arcade was not safe or feasible.

Figure 12: The remains of a siphon header where the pipes originated (from 12)

Unlike the other parts of the aqueduct, siphons channeled fast flowing water as it traveled steeply downhill. The arcades and tunnels had rather slow water with their gradual slope, which helped to prevent damage to the infrastructure. The siphon pipes could not avoid dealing with high speeds and therefore high pressure which they needed to accommodate. The pipes therefore had to be built different and it was after the header tank where water entered pipes to traverse the valley, seen in Figure 12. These pipes were roughly 25 cm in diameter and often experienced pressures of up to 12 atmospheres, with later experiments finding they failed around 18 atmospheres [12].

In Figure 11 you can see two other labeled components, the geniculus and the venter. The venter refers to the small bridge pictured which was built to reduce some of the elevation drop of the water, therefore reducing the pressure in the pipes and the head loss [12]. The geniculus is the bend of the pipe which experienced the most forces due to the water changing directions after falling downhill.

Romans had a fair understanding of how pressure impacted these pipes, as the venter reduced some of this stress as well as practices they had when performing maintenance. Their engineers knew to turn off the water supply slowly to avoid water hammer, which would cause a shockwave up through the siphon and damage the pipes and header tank [12]. Similarly, they would turn the water on slowly to prevent higher than normal pressure from being exerted on the geniculus which could cause cracks and leaking. Overall, the siphons are perhaps one of the least known components of the aqueducts yet demonstrate some of the Romans’ greatest understanding of hydrology.



Once water reached the city it was collected, underwent a very basic treatment process, and was piped out to its various destinations. These collection points were called castella and showcase another incredible feat of ancient Roman engineering. Water entered the city from the aqueducts at a high elevation which would allow them to continue to use gravity to move it throughout the city [8].

Figure 13: A look at the inside of the Nimes castellum (from 13)

The castella had pipes set up to ensure some sources would receive constant flow, such as the outflow holes visible in Figure 13. These would likely have sent water to public fountains and baths as the Romans prioritized public sources. You can also see the other pipes along the rim of the castellum base which would have taken water to other sources like private homes.

Additionally, castella served as a very basic form of water treatment—primary settling. The basic concept of primary settling applies the same today as it did 2,000 years ago. Water was allowed to flow slower than the general particle settling velocity such that suspended particles would precipitate out before entering the pipes.

When leaving a castellum, devices called calices were installed to manipulate the flow rate [2]. Romans at the time did not have a full grasp of flow rates as they measured them in a unit called quinaria which was effectively a unit of area [2]. Regardless, the concept the Romans employed was the relative difference in flow rates so they could deliver different amounts to a given source [13]. Various sources will disagree on the approximate flow rate of water ultimately entering the city from the 11 aqueducts due to lack of accurate measurements or documentation of the calices.


The ancient Romans were known for their expertise at building and left us with countless sites to wonder at. Among these are the aqueducts, both the arcades that remain as a visual reminder and the many other components which go overlooked by the public. Through a complex series of meticulously engineered works, Romans brough water from the mountain springs into their homes and public fountains and baths with a constant and steady supply. Their understanding of hydrology was not limitless but far surpasses what one may expect given the time they lived. Hopefully a brief glimpse into a critical piece of ancient Rome’s infrastructure allows you to realize what made their civilization so powerful and long-lasting. Their society was fueled by the constant presence of water across a vast empire.


[1] Editors. (2009, October 14). Ancient Rome. Retrieved September 25, 2022, from

[2] Deming, D. (2019). The aqueducts and water supply of Ancient Rome. Groundwater, 58(1), 152–161.

[3] Hansen, R. D. (n.d.). Water and wastewater systems in Imperial Rome. Retrieved September 25, 2022, from

[4] Kessener, P. M. (2021). Roman water transport: Pressure Lines. Water, 14(1), 28.

[5] Los Angeles Times. (n.d.). Interactive graphic: The aqueducts of Rome. Los Angeles Times. Retrieved September 25, 2022, from

[6] Water in Ancient Rome. (1901). Scientific American, 84(22), 346–346.

[7] Environmental Protection Agency. (2022, May 11). Statistics and Facts. EPA. Retrieved September 25, 2022, from,in%20the%20United%20States%20in%202015%29.%20More%20items

[8] Public Broadcasting Service. (2000, February 22). Watering Ancient Rome. PBS. Retrieved September 25, 2022, from

[9] Roman Aqueducts —Marvels of Engineering. Jehovah’s Witnesses. (2014, November). Retrieved September 26, 2022, from

[10] Šimunić Buršić, M. I. (2021). Opus signinum – roman concrete without Pulvis Puteolanis: Example of the substructures of Diocletian‘s Palace. 12th International Conference on Structural Analysis of Historical Constructions.

[11] Baiocchi, V., Alimonti, C., Bonanotte, G., & Molnar, G. (2020). Geomatic measurement of “New aniene” and “Claudia” roman aqueducts for flows estimation. IOP Conference Series: Materials Science and Engineering, 949(1), 012078.

[12] Hodge, A. T. (1985). Siphons in Roman Aqueducts. Scientific American, 256(6), 114–119.

[13] Ortloff, C. R. (2018). The Pont du Garde Aqueduct and Castellum: Insight into Roman hydraulic engineering practice. Journal of Archaeological Science: Reports, 20, 808–817.

Cameron Stuart

Add comment

Follow us

Don't be shy, get in touch. We love meeting interesting people and making new friends.