In the late eighth and early seventh centuries BCE, Sennacherib, king of Assyria, brought water to his capital at Nineveh in northern Mesopotamia through a system of canals. At Jerwan, his builders carried one of its channels across a valley on a great limestone aqueduct. The channels had to cross uneven ground without losing the gentle descent that kept water flowing. The stonework still stands, together with inscriptions crediting the king with drawing water from mountains far from the city.
Canals supported agriculture, roads moved armies and merchants, harbours brought grain and other cargo ashore, and drains carried water out of crowded settlements. Building these works was only the beginning. Surveying, skilled labour, available materials, decisions about funding and regular repairs determined whether an installation remained useful. A channel blocked by silt supplied little water; even a well-built road lost much of its value when its bridges or mountain crossings became impassable.
Ancient builders approached these problems according to local conditions. Assyrian kings organised work across extensive territories. Indus communities built with bricks of consistent proportions. In China, engineers diverted a river to protect and irrigate fields, while Roman builders developed concrete structures capable of lasting in seawater. Each technique grew out of particular landscapes, practical requirements and ways of organising work.
How Sennacherib Brought Water to Nineveh
Sennacherib ruled Assyria between 704 and 681 BCE, when Assyrian armies and governors exercised power across much of the Near East. He expanded Nineveh, near present-day Mosul in Iraq, into a major royal centre, building palaces and walls as well as gardens. Its inhabitants and fields needed water, but irrigation also served the king’s ambitions. Gardens planted with species brought from distant lands were a visible demonstration of the territories and resources he commanded, much like the scenes of conquest carved on palace walls.
Nineveh did not receive its water through one eighty-kilometre pipe from a single spring. Over time, builders developed a network of canals and watercourses that drew on sources far beyond the city. The channels followed the contours of hillsides and descended gradually enough for gravity to move the water without excessive loss through their banks. Where the route reached the ravine at Jerwan, the engineers constructed a huge stone crossing so the channel could continue above it.
The 1935 excavation report by Thorkild Jacobsen and Seton Lloyd describes the remains at Jerwan. The excavators uncovered a limestone structure more than 280 metres long, built from large dressed blocks with projecting buttresses for support. Five openings spanned the central ravine. Their pointed shape came from corbelling, in which successive courses of stone extend inward until they can be joined over a gap. This differs from a true arch built with wedge-shaped stones. Columbia University’s archaeological documentation illustrates the remains and explains the distinction.
Several of Jerwan’s stone blocks carried inscriptions with Sennacherib’s royal titles and accomplishments. A visitor could read certain inscriptions, while others were built into the masonry itself. The inscriptions celebrated the king for drawing water from the mountains and carrying it over valleys on a bridge of white stone. They show how the king wished the work to be remembered, while revealing little about those who staffed, administered or repaired the canals.
The royal gardens are visible in another source from Nineveh. A relief from the North Palace of Ashurbanipal, Sennacherib’s grandson, portrays cultivated grounds with water channels and an aqueduct. Made decades after the earliest canals, the relief presents an idealised royal landscape rather than an operating record. It supports the connection between Assyrian gardens and irrigation, without supplying the orchards’ precise yields or the operating procedures of each channel. The canals made Nineveh productive and impressive, but the gardens could flourish only while workers maintained the system and the court controlled the surrounding land.
Dujiangyan: Controlling a River Without Blocking It
On the Chengdu Plain of present-day Sichuan, China, engineers faced another problem in the third century BCE. The Min River came down from the mountains and flooded seasonally, threatening fields and settlements. Diverting part of the river could also irrigate farmland across the plain. The irrigation scheme at Dujiangyan is traditionally attributed to Li Bing, a Qin official, around 256 BCE.
Dujiangyan directs the Min River through a set of connected works without depending on a high dam and reservoir. At the Yuzui, or Fish Mouth, a divider separates the river into inner and outer channels. The Baopingkou, or Bottle-Neck passage, limits the water entering the irrigation network from the inner channel. At the Feishayan, the Flying Sand spillway, surplus water and some sediment return toward the outer river. Together the three works regulate the river: a channel taking in too much floodwater would damage farmland, while sediment trapped in the diversion would gradually obstruct it.
Seasonal floods change the Min River’s flow and sediment load. Dujiangyan’s builders had to manage those fluctuations rather than assume a constant flow. The UNESCO description of Dujiangyan explains the arrangement and identifies the three principal works, while also recording later enlargements and alterations under the Tang, Song, Yuan and Ming dynasties. The functioning system visitors see now contains centuries of additions rather than surviving unchanged from 256 BCE.
Water from Dujiangyan still reaches agricultural land on the Chengdu Plain. Its continued operation owes something to the original diversion scheme and just as much to successive generations who rebuilt banks, cleared channels and adapted the layout. Controlling the flow has required continuous maintenance. The continuity lies in the sustained management of the river, rather than in every ancient structure surviving where it was first placed.
Petra: Water Supply in a Landscape of Sudden Floods
Petra, the Nabataean city in southern Jordan, became a prosperous centre of trade and political power before Rome annexed it in 106 CE. The narrow sandstone valleys made water difficult to manage. Water could be scarce through long dry stretches, yet heavy rain could send dangerous floods through the ravines. The inhabitants needed to collect enough water for the city while diverting sudden flows away from its occupied areas.
Nabataean engineers collected spring water and seasonal runoff, cutting channels into cliff faces and fitting ceramic pipes into distribution routes. Reservoirs and cisterns stored water, while settling basins allowed suspended sediment to fall out before water moved onward. Along the Siq, the narrow gorge leading into Petra, remains of channels can still be seen in the rock. The same entrance had to remain passable during floods, so the water-supply works operated alongside measures to divert destructive stormwater.
Hydraulic engineer Charles Ortloff’s 2005 study of Petra’s water network explains how Petra’s reservoirs, continuous-flow pipelines and settling tanks worked together. Storage helped the city through interruptions in supply, pipes directed water toward inhabited districts, and settling basins removed some of the suspended material. The system also depended on the slope of its channels, their capacity and the pressure that developed within the pipes, all of which affected leakage and flow.
A gravity-fed channel must descend, but an excessively steep route can make water scour its lining, while a shallow gradient encourages sediment to settle and can leave too little flow at the destination. Closed ceramic pipes brought other concerns, particularly pressure and the strength of their joints. Ortloff examines these details in his 2020 hydraulic analysis. Petra’s surviving installations show careful use of its terrain, although historians cannot establish the city’s ancient population or total water supply with precision.
One isolated calculation produces an average of only about 2.7 litres per person per day, even before allowing for water lost from the system or used by animals, workshops and agriculture. That mismatch warns against treating estimates based on different quantities as if they measured the same supply. Petra relied on several sources, pipelines and storage facilities working across seasons. One annual figure cannot describe how that changing network delivered water to its inhabitants.
Roman Aqueducts and the People Who Kept Them Working
Rome began expanding its aqueduct network with the Aqua Appia, finished in 312 BCE, and subsequently drew water from springs and rivers at varying distances and heights. Much of it passed through covered channels or tunnels. The high arcades associated with Roman aqueducts were built where a valley or other obstacle required a channel to remain elevated without interrupting its gentle fall. Most of an aqueduct’s route did not look like the elevated sections familiar today.
Toward the end of the first century CE, Sextus Julius Frontinus became responsible for Rome’s aqueducts. In his surviving On the Water Supply of Rome, he describes the work of an official managing an established water network. He records the individual sources, methods of measuring water, the rights of different recipients and the losses caused by unauthorised connections and neglected conduits. Each problem reduced the supply reaching fountains and other users.
Access to the city’s water was unequal. Public fountains and baths, authorised private households, gardens and imperial properties received supplies under arrangements shaped by wealth and public policy. Distribution tanks fed separate pipes, but inspectors needed to know the network well enough to locate leaks or illicit taps. Such diversions reduced deliveries even if the water entering the system remained unchanged. Frontinus shows how hard Roman officials had to work to regulate the system.
Frontinus’s measurements do not yield a dependable figure for water use per person. Ancient measures of flow cannot always be turned directly into modern litres per person per day, while the total volume entering a city says little about what any one household collected. Aqueducts helped large urban populations, provided that officials maintained the works and enforced access rights and repair responsibilities.
Roman Roads: Surveying, Drainage and the Via Egnatia
Roman roads were made for repeated use by troops, messengers, traders and other travellers, but they did not all share one width or surface. Builders sometimes raised the roadway on an agger, an embankment packed with gravel and stone. A cambered surface shed rain toward ditches. Timber or brushwood could stabilise soft ground, while compacted local gravel might be sufficient elsewhere. The archaeological overview of roads in Roman Britain describes this variation and shows why a picture of uniformly stone-paved Roman highways is inaccurate.
Choosing a route could be as demanding as building the road. A direct line saved distance, but mountains, marshes or difficult rivers could make the shortest route costly or impractical. Surveyors considered the gradient, drainage and settlements to be connected, with bridges and retaining walls often posing bigger challenges than ordinary stretches of roadway. Culverts, small passages beneath the road, allowed water to cross without washing away the raised surface from one side.
The Via Egnatia crossed the Balkans from the Adriatic through Macedonia toward Thrace, linking ports and towns on the route between Italy and the eastern Mediterranean. Construction began in the second century BCE as Rome strengthened its control in Macedonia. Milestones preserve the name of the official Gnaeus Egnatius, after whom the road was named. Work proceeded in stages along different parts of the route.
A milestone surviving near Thessaloniki names Gnaeus Egnatius and records a distance in both Latin and Greek. The inscription records both the official responsible and the measured distance. Strabo, the geographer writing around the turn of the first centuries BCE and CE, also mentions its milestones. The historical study by Ivan Todorov investigates the road in the context of Roman expansion into Thrace. Built with military movement in mind, it also served merchants and civilian travellers.
The road needed continuing repairs once traffic began to wear its surface. Storms filled ditches, vehicles damaged the roadway and bridges or retaining walls deteriorated. The Digest, 43.11 includes a passage attributed to the jurist Ulpian concerning the right to open and repair public roads, including clearing material that had accumulated above the former road level. The passage shows that keeping roads passable was a concern of Roman law. It does not prove that all roadside owners owed a uniform share of repairs, or that every damaged milestone exposed its provincial governor to senatorial punishment; practice varied with period and jurisdiction.
Lothal: Was the Great Brick Basin a Dockyard?
Rivers and coasts carried goods long before Rome developed its ports. Lothal, in present-day Gujarat, India, belonged to an Indus civilisation region connected by inland and maritime exchange. S. R. Rao excavated the settlement during the 1950s and early 1960s. Alongside organised areas for crafts and storage, he found a large brick-lined basin, which he interpreted as a dockyard where boats entered with the tide and could remain afloat as the level changed.
The basin’s long dimension measures about 217 metres in the Indian nomination on UNESCO’s Tentative List. It also describes possible inlet and outlet channels, a nearby platform interpreted as a warehouse, and finds of seals, marine shells and stone anchors. The combination supports a dockyard hypothesis for a settlement with coastal connections. But rivers and creeks in the area have changed and silted up since the Bronze Age; the present shoreline cannot show directly how water reached the basin when Lothal was occupied.
Scholars continue to debate its use. A reservoir or another form of water storage has been proposed, and any tidal-dock reconstruction must account for the former channels, tides and water levels. The brick-lined basin unquestionably exists, but its gates, the movements of boats and the water-management procedure are reconstructed from incomplete evidence. Calling it the world’s first proven tidal dock would elevate one disputed explanation into a certainty. Reconstructing the waterfront that once surrounded the basin remains central to resolving the disagreement.
Lothal was also a place of production and storage. Its seals and craft goods connect it to exchange systems in which reliable weights, measures and handling procedures mattered. The basin’s position alongside these facilities helps explain why movement between land and water mattered, whatever its precise function. The wider geography of that exchange is covered in the Spoken Past article on ancient trade routes.
Building Harbours in the Open Sea: Caesarea and Portus
A port required more than an anchorage. Ships needed shelter, access to quays, room for cargo and onward connections to the places that consumed it. Where a coast offered no protected bay, builders might create one with breakwaters. Herod the Great began the artificial harbour of Sebastos at Caesarea Maritima, on the coast of present-day Israel, around 22 BCE. Its breakwaters pushed into the Mediterranean and formed sheltered basins where vessels could anchor or approach the quay.
At Caesarea the builders employed hydraulic concrete, mixing lime with volcanic material that could harden under marine conditions. Volcanic ash brought from the Bay of Naples was particularly important in Roman underwater building techniques. In the first century BCE, the architect Vitruvius described harbour construction and the use of volcanic powder for foundations in On Architecture, 5.12. His instructions describe the construction methods known to Roman builders; they do not reconstruct exactly how each block at Sebastos was poured or placed.
Parts of Sebastos now lie beneath the sea. Investigators examine the masonry alongside the coastal geology. A 2024 study of coastal development in the southern Levant discusses the use of hydraulic concrete at the site, and a 2025 high-resolution survey studies the harbour entrance and the processes responsible for its alteration. Building protection on this exposed shore demanded major engineering, and waves and sediment continued to change the harbour after construction ended.
A much larger complex helped supply Rome itself. Ostia stood near the Tiber’s mouth, where imported goods could be transferred toward the capital. Emperor Claudius began the new harbour at Portus in the first century CE; Trajan added its hexagonal basin in the early second century. Canals, quays and warehouses linked the harbour’s sea-going traffic with smaller craft carrying cargo upriver. Large merchant vessels generally transferred their cargo before it reached central Rome.
Excavations and geophysical surveys by the British School at Rome’s Portus Project have recorded Portus’s warehouses, port buildings and other installations. A 2022 study by Maria del Carmen Moreno Escobar reconstructs the effect of new canals and the Tiber’s changing course on movement between Portus, Ostia and Rome. These harbours worked as linked basins, waterways and cargo-handling centres. Expansion increased Rome’s capacity to receive supplies, but also brought more silting, heavier traffic and greater organisational demands.
Drains Beneath the Streets: Rome and the Indus Cities
Urban planners also had to remove water after it reached a settlement. Rain washed debris into roads, workshops and households produced wastewater, and low-lying land could remain flooded without drainage. Cities built channels under and beside their streets to address those problems. Ancient drains did not automatically connect every house or treat wastewater before releasing it.
The Cloaca Maxima developed in Rome’s low-lying Forum district across several construction phases. Parts received stone vaults in later periods, and its channels discharged toward the Tiber. Roman writers credited early kings with building the drain, but that literary history cannot date every surviving vault or wall to the sixth century BCE. Domenico Palombi’s 2020 study of the Cloaca Maxima examines its difficult chronology against the changing topography of the ancient city.
Drainage made wet ground more usable, but its health effects cannot be reduced to a proven fall in disease rates. Crowded living conditions alone cannot establish that cholera circulated in every ancient city. Drains could carry contaminated water into rivers, and drinking-water access remained unequal. Their most directly demonstrable function was to move water and waste away from places where they caused immediate problems.
The Indus city of Mohenjo-daro, in modern Pakistan, had an extensive arrangement of bathing spaces, brick-lined drains and house and street channels. Some drains were covered with removable bricks or stones, and settling spaces trapped sediment before it entered the next section. Archaeological documentation assembled by Harappa.com records vertical drains at houses and indications that the channels were cleaned. The household drains fed neighbourhood channels, making wastewater disposal a shared concern.
Many Indus bricks followed a thickness-to-width-to-length ratio of about 1:2:4, although actual dimensions changed between sites and periods. Jonathan Mark Kenoyer’s archaeological photographs and measurements show both the repeated proportions and their variations. Consistent proportions helped masons fit bricks together, without establishing whether a central government prescribed them. The masonry gives us evidence for shared building practices more readily than for the political institutions that sustained them.
Concrete, Labour and the Cost of Maintenance
Every project required workers to obtain and transport stone, timber, clay or volcanic ash, organising labour and resolving shortages. Sennacherib’s inscriptions attribute his waterworks to royal initiative. Roman roads, however, might involve officials, soldiers, contractors and communities. Funding and labour obligations differed: workers might be paid, required to serve or coerced. For most individual projects the full composition of the workforce is no longer recoverable.
Roman concrete made large harbour foundations, vaults and other heavy structures possible, though its performance varied with materials and curing conditions. In concrete exposed to seawater, lime and volcanic ash could react over time and form stable minerals. The 2017 mineralogical study by Marie Jackson and colleagues identified phillipsite and aluminium-bearing tobermorite in ancient marine concrete and investigated their formation. This helps account for the durability of some harbour works, without implying that every Roman concrete structure was chemically identical, earthquake-proof or cheaper than stone.
Materials also shaped a project’s cost and upkeep. A harbour built with imported volcanic ash first needed the shipping network it was intended to improve. A gravel road might be cheaper to extend than a monumental paved approach, although resurfacing could be needed more often. Canals gathered roots, mineral deposits and debris. Keeping them in service required the labour of technicians, administrators and maintenance workers whose names generally went unrecorded.
Written sources sometimes reveal the work of maintaining and administering these systems. Frontinus investigated unauthorised taps and aqueduct repairs; the Digest preserves legal concerns about the condition of public roads; Herodotus described royal stations between Sardis and Susa. In his Histories, 5.52–53, Herodotus lists stopping points and estimates ninety days for a traveller’s journey. He supplies a description of the route and its facilities, not a full budget for maintaining each station or evidence of a fourth-century austerity programme.
The Persian route depended on more than a continuous roadbed. Travellers and couriers needed security, river crossings, stopping places and supplies. Rome faced the same interconnected demands: roads brought cargo inland, while seaports received goods that roads alone could not carry across the Mediterranean. Damage to a bridge, harbour entrance or canal junction could obstruct a much longer journey. Networks worked only when their different sections remained connected.
What Survives When the State Changes?
Political change affected ancient infrastructure in different ways. Some Roman roads continued in use and received repairs after the western imperial court disappeared in the fifth century CE; elsewhere aqueducts were restored, altered or left unused at different times. Dujiangyan survived through repeated changes by later rulers and communities. Local terrain, settlement, labour and political priorities determined which works remained useful long after their original builders were gone.
A surviving monument often represents only part of the system that once supported it. Jerwan’s limestone crossing remains, but much of the connecting canal network has disappeared. Lothal retains its basin without the waterways that once surrounded it, and Petra’s channels cannot reveal every litre delivered to the city. Archaeologists combine walls, inscriptions and historical texts with sediments and the physical properties of water and building materials to reconstruct what was once connected.
Ancient roads, canals, ports and drains determined how communities obtained water, grew food, transported goods and sent messages. Kings and governors often received the credit for their construction. Their continued usefulness depended on unnamed workers repairing roads, clearing canals, tending drains and unloading ships. The networks survived only as long as successive generations could keep them operating.
Discussion
Comments