By 1723, the small river Kennet which joined the Thames near Reading had been made navigable upstream to Newbury. Five years later, the Bristol Avon was canalised from Hanham Mills – its tidal limit – as far as Bath. Strangely, although these developments raised the possibility of establishing a navigable inland route between Bristol and London, it was not until the 1780s that plans to connect the two waterways by canal began to develop in earnest.

On 29 July 1788, a newly formed canal committee under chairman Charles Dundas, 1st Baron Amesbury, commissioned engineers James Barnes, Samuel Simcock and Samuel Weston to survey a route linking the canalised Avon with the Kennet Navigation. The trio proposed a circuitous connection which took in Bradford-on-Avon, Melksham, Laycock, Chippenham, Calne, Calston, Marlborough and Hungerford, climbing 157ft 6in over 26 locks on its way before plunging into a 3442 yard tunnel through Calstone Hill. Concerns about the scheme’s expense and the adequacy of the water supply encouraged Dundas’s committee-men to seek a new surveyor.

They chose John Rennie who explored a number of alternative routes. In 1794, Parliament authorised a line from Bath to Newbury by way of Bradford-on-Avon, Trowbridge, Devizes, Great Bedwyn and Hungerford. By crossing the River Avon at Avoncliff and then re-crossing it near Monkton Coombe, Rennie not only cut a corner but also kept the canal on a level course between Bradford Lock (no 14) and Bath Top Lock (no 13) and avoided having to bridge the River Frome and Midford Brook. Nevertheless, his short-cut necessitated building a pair of sizeable aqueducts.

He conceived them on a resplendent scale, in a style that shows all his design-flair. At 150 yards (137.2m) long, the Dundas Aqueduct near Monkton Coombe has a hemispherical central arch with a span of 65 feet (19.8m) and elliptical side arches of 20 feet (6.1m) span. Framing them are revetments – wing-walls – which slope somewhat, buttressing the aqueduct on each side; pairs of vast pilasters flanking the central arch and single pilasters bordering each revetment. A wide cornice crowns a Doric frieze which runs the length of the aqueduct, extending over each pilaster. Made up of alternating triglyphs – tablets showing triple vertical grooves – and metopes – plain rectangular spaces- the frieze lends the Aqueduct a neo-classical flourish, befitting its proximity to the City of Bath.

An early drawing indicates that it is built on inverted arches – a foundation used upon ground that has little weight-bearing capacity and does not allow for deep excavation, and where the load borne by the structure –with an aqueduct it will be great – is concentrated over its walls. Building inverted arches required considerable skill and was apt to be costly, not least because of the need to install cofferdams – enclosures on the river bed which could be pumped out and kept (relatively) dry –for the labourers to work in. Unsurprisingly perhaps, although contractor James McIlquham undertook to build the aqueduct above waterline level, no-one tendered to build its foundations and the work had to be done by direct labour.

At 110 yards (100.6m) long, the aqueduct at Avoncliff is smaller but equally characterful. Its central arch spans 60 feet (18.3m) and the side arches are both 34 feet (10.4m) across. The cornice has a block-bracket design which extends along the curving abutment walls. Like the haunches of the arches, they are built in contrasting courses of smooth ashlar masonry and rougher rusticated stone blocks. Cutwaters rising to deck-height serve as buttresses while the balusters along the approaches enhance the structure’s elegance.

Avoncliff’s central arch sags – not greatly, but perceptibly nonetheless. Most probably the trouble resulted either from an over-hasty contractor’s removing the centring on which the aqueduct was erected before the mortar had fully hardened or that the porous limestone, set in place amid frosty conditions, became friable at an early stage.

Recognising that difficulties with crumbling masonry caused expense and delay, in 1803 Rennie urged the canal proprietors to build in brick rather than stone. They dismissed his suggestion out of hand – not for aesthetic reasons, but because they wished to preserve friendly relations with the local quarry-owners who were likely to make much use of the canal when it opened.

The Kennet and Avon Canal prospered until the Great Western Railway opened between London and Bristol, when its traffic ebbed away. Trying to compete with the railway, the canal company cut their tolls throughout the 1840s. It was not a sustainable strategy and in 1852, they sold the canal to the GWR.

Although the law required the railway company to keep it in a navigable condition, their approach to canal maintenance was extremely lax. Winter ice-breaking, for example, ceased in 1857 and when the masonry of the aqueducts fractured, it was patched with blue bricks. With ever diminishing traffic, in the mid-twentieth century the British Transport Commission pressed for the canal’s abandonment – a move which met with powerful opposition. In 1951 Lord Methuen had the Dundas Aqueduct designated as a Scheduled Ancient Monument – the first canal structure to enjoy such protection. Soon after, Kennet and Avon enthusiasts formed themselves into an Association – later a Trust – and began to restore the Canal.

In recent years, the Canal and River Trust has overseen repairs to both aqueducts, sourcing stone that is somewhat harder and of better quality than its 1790s counterpart, if much the same colour, and employing techniques with which Rennie and McIlquham would have been familiar. Numerous mason’s marks – geometric symbols identifying the stone’s source – are visible on each structure’s fabric. Above the Dundas Aqueduct’s central arch are two bronze plaques; one, facing south-west commemorates Charles Dundas, first chairman of the Canal Company; the other, facing north-east, celebrates John Thomas of Bristol, a grocer by occupation who, in a late-life career change, took responsibility for supervising the canal’s construction-in-progress after resident engineer Dudley Clark was dismissed for disregarding Rennie’s instructions about economical work-practice. Traces of the tram roads used to bring stone down from the quarries survive along the wooded paths around Murhill and Conkwell.