Working water

The Roman architect Vitruvius, who wrote between BC31 and AD14, was clearly familiar with machines like the London one. He says that if engineers need to raise water from a deep well a double iron chain will be set up ... with buckets suspended from it. Thus the turning of the wheel ... will carry the buckets to the top, and as they are borne over the wheel, they will necessarily ... pour the water into a reservoir.

An earlier Greek inventor, Philo of Byzantium (3rd century BC), tells us that the drive shaft was facetted (polygonal, not circular, in cross-section), and that the chain links were shaped to fit the facets. Vitruvius and other Graeco-Roman engineers have provided descriptions of the treadwheels and other devices that were used to power machinery of this type.

Astonishingly, the Gresham Street bucket-chain that was used for the reconstruction is one of just four discovered in London. But not only is it the most complete, it is the most sophisticated - and probably the latest in date, having been constructed in, or shortly after, AD108-9. Another chain was found in a smaller, earlier well on the Gresham Street site itself; a second just to the south by a public bath-house; and a third on the north bank of the Thames near London Bridge. No other Roman bucket-chains have been found outside the Mediterranean.

At Cosa in Italy, archaeologists have found evidence for the water-lifting machinery that supplied public baths. The bucket-chain there is very similar to ours. At Pompeii a pair of bucket-chains was used to feed the Stabian Baths with water from a deep adjacent well. The chains were attached to a treadwheel, and it is thought that it would have taken two slaves an hour to produce up to 800 gallons (3637 litres) of water - around half the likely output of the London machine.

The buckets found at Gresham Street were oak boxes or cubes, open at one end but closed at the other. There were cut-outs and holes for the ironwork but no other apertures. The project engineers, Tony Taylor and Dr Bob Spain, were confronted with a difficult problem. How was the water emptied out of the bucket without splashing over the buckets in front and running back into the well? Neither the ancient writers nor the other archaeological discoveries were of any help on this essential point.

They concluded that there must have been a splash-plate in front of each bucket, and that the bucket would have emptied into a compartment within the drive wheel itself. From there it could be channelled out at the side into a trough or launder.

Having made the basic calculations, the engineers were ready to put their theories to the test. Construction experts McCurdy & Co. were appointed to build a machine that could have been made in the Roman period - yet would be practical and safe enough for people to operate today. In their work they were assisted by the Museum's ancient timber specialist, Damian Goodburn, who has himself built a range of replicas, from boats to houses.

The first stage was to construct a plywood model of the drive wheel and chain. Initially, a seven-sided wheel was tried; but after extensive testing, Tony and Bob found that an eight-sided version was much more efficient at collecting and delivering water. Different shapes and sizes of side aperture were also tested. Once they had worked out the best arrangement of wheel, bucket and link, Peter McCurdy's team could move on to building a plywood model of the complete chain.

From an early stage the calculations had shown that even by modern standards this would be a piece of heavy engineering. The bucket chain would weigh about 500kg (half a ton), and strong forces would be exerted by the team of 'slaves' turning the drive shaft. The builders decided to construct a box-shaped framework from solid oak. Mortise and tenon joints would be used to hold all the main components together, with dove-tail joints for some of the braces.

Although the Roman architect Vitruvius seems to have been most familiar with treadwheels, Tony and Bob decided to use a capstan. This device is known to have been used often by the Romans, but would involve translating horizontal movement into the vertical rotation of the drive shaft. A crown-and-pinion gear set was built from two circular oak plates fitted with meshing oak spindles.

With all design problems resolved, the machine could go into production. It was built in McCurdy's yard near Reading and rigorously tested. Then it was brought to London and assembled in the Museum grounds in just two days.

Francis Grew & Jenny Hall
Department of Early London History and Collections, Museum of London
December 2002

Visit the Museum to see...

The reconstructed water lifting machine is on display in the Rotunda Garden. Some of the original excavated artefacts are also in a special display in the Museum foyer. These displays are for a limited period only. Please check before making a special visit.

On the Museum website...

Introductory fact pack: Working water: Roman technology in action

Fact pack: Roman London: A brief history

More on the excavation at Gresham St

The Museum also recently reconstructed a late Iron Age roundhouse which was based on evidence from a neighbouring excavation at 10 Gresham Street. Two Fact packs describe how the roundhouse was built and the archaeological background.