There are lots of definitions of engineering. For our purposes the best seems to be “engineering refers to the practice of organising the design, construction and operation of any artifice which transforms the physical world to meet some recognized need”, G F C Rogers The Nature of Engineering 1983.

Engineering was not in antiquity (and arguably still is not now) ‘applied science’. It is not derived from science; it is an autonomous body of knowledge that sometimes applies science. Some, e.g. Lewis Wolpert, argue that technology precedes science, rather than vice versa; that is certainly true for antiquity. Engineers find solutions to practical problems; knowledge is not an end in itself. 

Solving problems 

Cities could not grow without solving the problems of water and food supply for more inhabitants. Engineering develops with the rise of cities and urban life, because cities throw up practical problems which, if not solved, prevent the further growth of the city. In Greece, major urban waterworks were undertaken by seventh and sixth century tyrants, e.g. Theagenes (Megara), Polykrates (Samos), Peisistratos (Athens), and other prominent leaders, e.g. Solon, and then in the fifth century by Themistokles, Kimon, and Perikles. In Athens’ radical democracy, the water commissioner was a single office, elected, and 4-year tenure of office. Compare generals.


Engineered water systems are major projects, consuming vast quantities of labour and often materials too. They do not happen by accident, but only when social, economic and political conditions are right; they are not maintained, or are destroyed, when different conditions obtain. E.g. a monk removed the tiles from the top surface of Nero’s 40+m high dam at Subiaco (Sabine hills) in AD 1305; it collapsed soon after (Wikander ed. 2000:334). There was a great leap forward in hydraulic engineering in Hellenistic times, with political and economic centralisation under Alexander’s successors. There were major urban projects, in a different league from what had gone before. There were also major advances in theoretical and practical hydraulics with Alexandrian mechanicians e.g. Ktesibios and Philon

Transport & storage 

Solving supply problems usually involves moving water or food, so engineering is often concerned with transportation and storage. Water courses, cisterns, reservoirs, dams, aqueducts, public supply points (= ‘fountains’), priority systems, drains. Roads and bridges. Harbours, canals, warehouses. 

General principles of ancient hydraulic engineering 

Gravity alone provides the force to move large quantities of water (ideally at 1:200 gradient). Force pumps etc. lift relatively tiny quantities, so urban supplies depend initially on local water, then bringing in water from higher altitudes. The source must be constant; streams that dry up do so precisely when the need for water is at its highest. 

The ancients preferred to use fresh, running, spring water, not standing water or rivers. This was probably wise from a health viewpoint. A river flowing through a city was usually polluted (sewers). It was also lower than the city, so lifting would have been required. Settling tanks and even charcoal filters existed to reduce physical pollutants (e.g. leaves, dead rats), but there was no chlorination so no defence from microbiological hazards. 

The relatively small number of known dams (13) and reservoirs are found (with 1 exception) in the arid regions of the Mediterranean (1, Homs, is still working). 90% of an aqueduct runs underground (on average); it is put on arches to cross shallow valleys and to keep water high on the final approach to (ideally) the highest point of the city it feeds; on arches rather than embankments across Rome’s plains for the convenience of people in the Campagna. It is an open system except for inverted siphons. It is designed for water to run constantly from start to finish throughout its length, 24/7. 

Scale of supply 

Approximately 1,000,000m3 water per day was delivered into Rome. For comparison, Welsh Water current usage figures are 2m3/person/week. So there was enough water going into Rome to keep a modern population of c. 3.5 million satisfied.

Public water supply in gallons/person/day 

  50 BC AD100 1823 1936
Rome 198 300   150
London      3 35.5
New York        120


Need for maintenance 

Most European water has a high calcium carbonate content. Calcium carbonate deposits choke the system. Cf. scale in kettle, tartar on teeth, stalactites in caves. Aqueducts were built big enough for maintenance men to get through to hack it off. If thick enough, slabs were polished and used like (Travertine) marble. 

The source 

Ideally a significant spring; sometimes a dam or weir built across a stream to direct water into aqueduct; once (at Side, S Turkey) the natural hydraulic currents were exploited to collect spring water flowing into a river from the opposite bank before the waters mixed. See Hodge Aqueducts pp. 69-70 and photos p.107. Spring waters were collected in a basin or tank, sometimes with additional catchment tunnels (water in aquiferous stratum seeps through walls), then channelled to supply point. Alternatively, rainwater sometimes collected by small rock-cut channels fanning out over the hillside, as at Laurion in Attike. Aqueduct = water conduit. Six types were in common use: 1. Built masonry open channel (= 80-90%); 2. Rock-cut open channel or clay-lined leet; 3. Lead pipes; 4. Terracotta pipes; 5. Stone pipes; 6. Wood pipes. 

The principle of the whole system (including piped sections) is that of an open channel with gravity flow. Some piped sections came under great pressure (hence stone or lead pipes used), but that was a function of their relative depth below the header tank. Pipes are used instead of open channels where the water would otherwise flow over the top of the sides (large pipes), and in built-up areas (small pipes). A fully-piped system is designed to run full, and will have nominal pressure, but not high pressure except in inverted siphons (if any). 

Hydraulic pressure 

The main factor governing pressure in a pipe is the vertical distance between the pipe and the water’s last natural surface level (= the ‘head’). A big head gives a high velocity discharge; a small head gives a low velocity discharge. With a continuous flow system the head can be reduced by building a raised tank or urban water tower. See Hodge Aqueducts pp. 234-8. 

Siphons (inverted) 

These were used on valleys more than 50m deep; the largest known is Beaunant, which had 9 parallel pipes, 123m deep, 2.6km long; it used > 2,000 tons of lead, and had 11-12,000 joints to be soldered. On valleys less than 50m deep a bridge was built and the aqueduct run over on arches (Pont du Gard = largest such, 48.77m high, 275m long), because they were cheaper and easier to build (the stone for the Pont du Gard was quarried 600 metres away). Occasionally inverted siphons were used to cross a river; the aqueduct water was carried in pipes across the river bed. The water in a siphon is under pressure of the head; it is very significant pressure in a deep valley, hence stone or thick lead pipes. The receiving tank is lower than the header tank. Water rises in the pipe up the other side of the valley to find its level; no suction is involved. A venter bridge may be built across the bottom of the valley to ‘lift’ the deepest part and reduce the pressure.

The design stage 

We do not have surviving plans of ancient civil engineering projects, but they did exist. Scale models were made for buildings. Architects competed for projects, and had to include projected budgets. Vitruvius complains about budget overruns in Roman cities, & praises laws in place in Greek ones to prevent excessive extra costs (10.1). 


Greek building was in stone, mudbrick & wood. Roman building was in stone, mudbrick, wood, brick (square, oblong, triangular or curved [for pillars]), and concrete. Roman development of the arch made much stronger and taller bridges and buildings possible. The Roman discovery of hydraulic concrete made major bridge and port development possible (solid underwater foundations).


There is no certain example of a true arched stone bridge before Roman times; the earliest is 142 BC when Pons Aemilius was given an arched superstructure. Some Roman bridges are still standing; five e.g. the Pons Fabricius, are still in use in Rome. Semicircular arches and short spans were used usually. The largest was perhaps that built by Lacer for Trajan c. AD 106 over the Tagus River at Alcantara (Spain): height 175’, length 600’, six arches, two central arch spans of 118’ each. It was still open to traffic in 1956; Kirby, Withington, Darling & Kilgour History of Engineering (1956) 68. Another in Cordoba, Spain, is still being used by traffic in 2004. When necessary, cofferdams were built to fill with rubble and concrete or to exclude water while the piers were being built. A double wall of close piles was driven into the riverbed to make a ‘ring’; the water was lifted out and wickerwork and caulk packed between walls to allow (dry) work on riverbed to commence: either laying stone or driving piles further into bed. The pier foundations of the Pons Aelius go > 16’ down into the bed of the Tiber. Trajan’s bridge over the Danube AD 105, commemorated on his column, was 1,120m (=3,675’) long, 40-50’ wide, had 21 spans of 120’ average width. It had stone piers, and wooden segmental arches. It is sometimes claimed as the earliest evidence for a structural truss, but Cotterell & Kamminga (1992) 115-8 argue that the earliest structural truss is C16 AD, and that this is just parapet. What supports this bridge is wooden arches. It was built (like Trajan’s forum, odeum, and baths) by Apollodoros of Damascus. Other famous Roman and Greco-Roman engineers are Appius Claudius Crassus (Appian way, C4 BC), Agrippa (Augustus), Severus & Celer (Nero), Frontinus (Nerva) and emperor Hadrian himself.


Were an early development: Eupalinos’ tunnel in Samos is C6 BC, dug from both ends and met around the middle; for water supply. The longest tunnel before AD 1876 was Claudius’ tunnel dug to drain the Fucine Lake. 3.5 miles long, falling 28 feet over that distance, it involved at least 7 miles of galleries and shafts – the latter up to 400’ deep – dug through the Apennine mountains. Apparently 30,000 men worked on it for 11 years.



T E Rihll  

Last modified: 11 April 2007