Between 1760 and 1830 England dug a network of canals. Did the towns they reached boom, and do those towns still lead today? I dated the network, followed England's largest towns from 1801 to 2021, and checked whether the canal towns were already on the rise before the first spade went in.
In 1761 the Duke of Bridgewater's canal opened to carry coal into Manchester. Within a few decades England was laced with them. The busiest years came in the 1790s, the decade remembered as canal mania. Then, from 1830, the railways arrived and the canal age was over.
Canals were built to move heavy, cheap goods: coal, stone, clay, grain. The towns at their ends and along their banks grew fast. Or did they? Towns that sat on coalfields and rivers were already growing, and canal promoters wanted traffic, so they dug towards places that already had it. That makes it hard to tell what the canals did and what they simply followed.
I tried to answer three questions with open data: did canal towns grow faster in the canal age; does their lead still show today, in size and in income; and were they already rising before the canals came?
The map shows 3,884 km of canal from OpenStreetMap. I gave 77% of it, by length, an opening year from Wikipedia's list of UK canals. Press play to watch the network grow decade by decade. Lines turn red in the decade they open. Grey lines have no date in my sources.
The dots are England's largest towns in the historical tables, sized by population. Red dots are canal towns: within 2 km of a canal open by 1830. Switch the year to see them swell. Tap any dot or line for its details.
The busiest decade is the 1790s, with 586 km of dated canal finished, and 44 of the canals in the list completed in that decade alone. Of the dated length, 84% belongs to canals finished between 1760 and 1830. By length the network on today's map is overwhelmingly a creation of the canal age.
One warning about the dates. The list gives one year per canal, normally the year the whole line was complete. Long canals opened in pieces. The Leeds and Liverpool Canal is dated here 1816, but parts of it carried boats decades earlier. So the map, if anything, shows the network arriving late.
To compare like with like I start from the 29 largest towns in England in 1801, leaving out London. 14 of them were canal towns by 1830, on my definition, and 15 were not. The other group is mostly old cathedral and county towns and seaports: 6 of the 15 lie within 5 km of the sea, against 1 of the 14 canal towns.
Between 1801 and 1861 the canal towns grew by a median of 4.0 times. The other towns grew by at most 2.7 times. I have to say "at most" because the 1861 table lists only the 30 largest towns: 4 of the other group and 2 canal towns fell out of it, so all I know is that they had fewer than 40,400 people. For comparison, England as a whole grew about 2.0 times between 1801 and 1851.
The fastest risers include canal towns such as Stoke-on-Trent and Sheffield, but the very fastest was Sunderland, a coal port on the Wear with no canal, and two others in the top six, Oldham and Bolton, are cotton towns that my data count as "other" because the canal branches that reached them are not dated in my sources. The gap between the groups is real but the groups overlap a lot. Of the canal towns, 12 of 14 stayed in the top 30 in 1861; of the others, 11 of 15. With groups this small that difference could easily be chance (Fisher's exact test p = 0.65).
The pattern holds in the North and in the Midlands, and among both the bigger and the smaller towns of 1801. It does not hold in the South and East, where only 2 towns count as canal towns. Most of the canal towns are in the North and Midlands (6 northern), which is also where coal, cotton and metal were. Region and canals are tangled together.
Another way in is to ask which towns broke into the rankings after 1801. 18 towns appear in the 1861, 1881 or 1900 tables that were not among the 30 largest in 1801. 9 of them are canal towns, including Preston, Rochdale, Huddersfield, Burnley and St Helens. The rest include Middlesbrough, Gateshead, South Shields and seaside Brighton: 4 of the 9 non-canal newcomers are coastal. Half and half is about the same mix as the 1801 group (48% canal towns), so the newcomers do not single out canals.
This is the crucial test, and the data give a fairly clear answer. The same Wikipedia article ranks the largest towns in 1377, 1523 and 1662, and gives estimates for 1750. Of the 14 canal towns of 1801, only 3 were among the 30 largest towns of 1662. Of the 15 other towns, 11 were. That difference is unlikely to be chance (Fisher's exact test p = 0.009).
So the canal towns were mostly new towns: places that had risen from nowhere in the century before 1801. Manchester, Birmingham, Liverpool, Leeds and Sheffield are all absent from the 1662 table, yet by 1750, eleven years before the Bridgewater Canal opened, they already had between 12,000 and 24,000 people each.
The canals were dug to towns that were already booming. That is what canal promoters would do, and it means the canal premium is not a clean measure of what canals caused.
That does not mean canals did nothing. Cheap coal and cheap carriage surely helped the new towns keep growing. But the data cannot separate the two, and I would not claim that the 4.0 against 2.7 gap is the effect of canals.
Two centuries on, the canal towns of 1801 have grown by a median of 10.2 times to their 2021 built-up areas, against 6.9 times for the others. Allowing for size in 1801, the coast and the North in a simple regression, canal towns grew about 1.38 times as much, but the uncertainty is wide (95% interval 0.90 to 2.12). Size in 1801 still ranks size today well: the rank correlation is 0.69.
Income is another story. In the ONS estimates for the financial year ending 2023, mean household income in the canal towns' built-up areas has a median of £45,596, against £47,006 for the others. Allowing for the same factors, canal towns are about 6.9% poorer (95% interval 0.4% to 12.9% poorer). Being in the North matters more: about 9.9% lower.
The same holds across all 884 built-up areas in England and Wales with at least 10,000 people in 2021. 272 of them (31%) have a canal of 1830 or earlier within 1 km of their edge. Within the same region, canal places are about 54% bigger. Allowing for region, size and the coast, their household income is about 2.7% lower (95% interval 0.6% to 4.7% lower). Without allowing for size, the gap within regions is only 1.2%, and could be zero (95% interval 3.2% lower to 0.8% higher). In the canal age, though, canal places in the North and Midlands sat on coal and made things; today those places are both bigger and slightly poorer than their neighbours.
A small difference, then, and pointing the wrong way for anyone hoping that canals left a lasting gift. The likeliest reading is that the canals tied places to the industries of their age, coal, textiles and heavy goods, and those industries have since shrunk. Size outlasted the boom; wealth did not.
Lancashire's mill towns are the classic canal towns. The Leeds and Liverpool Canal runs through Wigan, Chorley, Blackburn, Burnley, Nelson and Colne on its way over the Pennines, and the Bridgewater Canal still links Runcorn, Sale and Altrincham with Manchester.
Blackburn grew more than tenfold between 1801 and 1900, from 11,980 to 129,200 people. Burnley was not even in the 1801 or 1861 tables, yet by 1900 it had 97,000 people. Wigan never appears in the tables at all. Today all three rank in the bottom quarter of the 884 places for household income. Of the 30 built-up areas of 10,000 or more along the Leeds and Liverpool Canal, 26 are below the median for England and Wales. Their median income, £42,457, is 4.4% below the North West's median of £44,432.
Railways. The Liverpool and Manchester Railway opened in 1830, and over the next decades railways reached almost every town. The period from 1801 to 1861 is therefore half canal age and half railway age. I have no railway data here, so I cannot separate the two, and I have not tried to.
Only the biggest towns. The historical tables list only the top 30 or 40 towns in each year. I could not find open town populations for 1851 or for smaller towns, so the canal age comparison rests on 29 towns, and the 1861 figures stand in for 1851. Wikidata does hold 1801 to 1851 figures for 343 places, but 78% of them are parishes in and around Cambridgeshire, so I left them out.
Dates and matches. 875 km of canal on the map, 77% dated by length, leaves gaps: some branches (at Stockport, Bolton and Oldham, for instance) are undated, so those towns count as "other". Canals that were abandoned and filled in survive on OpenStreetMap only in pieces, or not at all: Derby and Bradford had canals into their centres, but their nearest dated canal on the map is more than 5 km away, so they count as "other" too. These errors all push canal towns into the other group, which, if anything, shrinks the gaps I report. The Shropshire Union is dated by the list to 1835, though the old Chester Canal along it is older; counting any canal on today's map instead puts Chester among the canal towns and changes little (canal towns 3.9×, others at most 2.8×). With a 1 km rather than 2 km threshold it is 3.9× against at most 2.9×.
Boundaries. An 1801 town, an 1861 borough and a 2021 built-up area are different shapes. Growth to 2021 compares unlike units and is a rough guide only.
Network. 8,356 OpenStreetMap waterway=canal ways in the UK (spillways, mill races, irrigation channels and waterway relations removed), 3,884 km in all. I matched each way's name to the 193 canals in Wikipedia's List of canals in the United Kingdom: exact name, then a list name contained in the OSM name, then a short table of spelling variants. OSM's "Grand Union Canal" is the 1929 merger of older canals, so I split it by position into the Grand Junction, Warwick and Napton, Warwick and Birmingham, Leicestershire and Northamptonshire Union and old Grand Union canals, each with its own year. Unnamed pieces take the year of a dated canal within 300 m. Wikidata canal items fill a few remaining dates.
Towns. Every town in the ranking tables of Wikipedia's List of towns and cities in England by historical population, placed at its OS Open Names point and matched to its Census 2021 built-up area (BUA22). 2021 residents are summed from LSOAs; income is the population-weighted mean of the ONS MSOA estimates of mean total household income, financial year ending 2023, over the MSOAs that best fit the area.
Tests. Medians by group; Fisher's exact test for table membership; least-squares regressions of log growth and log income on a canal flag, log 1801 population, a coastal flag (within 5 km of the Natural Earth coastline) and a North flag, with robust standard errors. For all built-up areas, regressions with region fixed effects. These are associations, not causes.