The Victorian seaside is usually told as a railway story: the line arrives, the day trippers pour off the excursion trains, and a fishing village becomes Blackpool. I followed 68 resorts through every census from 1801 to 1911 to see whether they boomed after the trains came, or were already growing.
In 1801 Weston-super-Mare was a village of 138 people. By 1911 it held 23,235. Blackpool went from 473 to 58,371. Stories like these are why the railway gets the credit for the British seaside: the cheap excursion ticket, the Wakes Weeks, the mill towns emptying onto the sands.
But Brighton was already a fashionable town of 46,661 people when its London line opened in 1841, and Margate and Scarborough had been bathing places since the eighteenth century, reached by road and by boat. So which was it? Did the railway make the resorts, or did it simply follow the money to towns that were growing anyway?
Canal Mania, my earlier study, asked the same kind of question about canals and found the canal towns were already rising before the first spade went in. Here I use a sharper tool. For each resort I find the year its railway opened, then line up its census growth decade by decade before and after that year. If the railway invented the seaside, growth should jump when the line arrives and not before.
Wikipedia's list of seaside resorts in the United Kingdom has 205 places on it (I left out 13 entries that are beaches, peninsulas or stretches of coast). For 177 of them I could date the railway: the earliest opening year of any station within 3 km, from Wikidata, checked against the resort's own Wikipedia article. 28 never had a station that close.
Press play to watch the network reach the coast. Each dot turns solid in the year its first station opened; its colour shows the decade. The bigger dots are the 68 resorts with census figures, sized by their 1911 population. Tap any dot for its railway date, station and census figures.
The rush came early. The first line in my data touched the sea in 1830 and 51 resorts got a station in the 1840s, the decade of the railway mania. Half of them had one by 1861. The last stragglers waited until the 1900s, among them Lyme Regis and Dornoch.
Who got the railway first? Resorts close to the big industrial towns did. Taking the ten largest English towns of 1861 (London, Liverpool, Manchester, Birmingham, Leeds, Sheffield, Bristol, Plymouth, Newcastle and Bradford), resorts in England and Wales within 50 km of one got their line in a median year of 1849. Those more than 150 km away waited until 1878. The rank correlation between distance and arrival year is 0.23 (159 resorts, p = 0.003): clear, but far from perfect.
The cleanest test is to look at the years when no resort in my census sample had a railway at all. The first lines here opened in the late 1830s, so the three decades from 1801 to 1831 are pure pre-railway time. For 52 resorts I have all three decades.
Their typical growth was 1.63% a year. For comparison I drew a random sample of 60 towns at least 25 km from the sea and found census series for 20 of them: their typical growth was 1.43% a year, and the difference is well within chance (Mann-Whitney p = 0.34). A third group, 280 inland villages and small towns with census figures in Wikidata, grew 1.09% a year.
So the resorts were already growing as fast as ordinary market and industrial towns, and much faster than the villages, before any train ran. Some were racing: Weston-super-Mare grew 7.5% a year, Brighton 5.7% and Torquay 4.8%. Those are boom-town rates, decades before the line.
The railway did not find empty beaches. It found towns that were already growing.
Now the main test. For each resort I count decades relative to the railway: the census decade in which the line opened is decade 0, the one before is minus 1, the one after is plus 1, and so on. Then I average growth at each step. The same is done for the inland towns, relative to their own railway.
The resorts' line is flat and high before the railway: around 1.6% a year on average. In the decade the line opens it jumps, and it stays a little higher afterwards (2.0% on average). The inland towns also spike in their railway decade, but from a much smaller set of places, and afterwards they settle back below their earlier pace.
Averages like these mix different resorts and different decades. So I also fitted a regression with a separate baseline for every resort and every census decade, which compares each resort with itself and allows for the national ups and downs. Against the decade just before the railway, resort growth was 1.33 points a year higher in the decade it arrived (95% interval 0.24 to 2.42) and 1.18 points higher in the decade after (-0.06 to 2.43). The decades before the railway show no build-up: their estimates sit between -0.04 and +0.29.
Is that bump special to the seaside? Not on this evidence. When I put the resorts and inland towns in one model, the inland towns grew 0.27 points a year faster after their railway than before, and the resorts' extra gain on top of that was 0.08 points (95% interval -1.20 to 1.36). In other words, I cannot tell the resorts' railway effect apart from the inland towns'. Against the inland villages, which were losing people in the later nineteenth century, the resorts do look different (an extra 1.53 points a year), but that comparison says more about rural decline than about the seaside.
A simpler check: for each place, compare average growth over up to three decades before the railway with up to three decades after. 26 of 43 resorts grew faster afterwards; the typical change was +0.68 points a year. For the inland towns it was 6 of 15, with a typical change of -0.21. That difference leans towards the resorts but is not conclusive with so few inland towns (Mann-Whitney p = 0.07).
The two ends of the chart tell the real story. At the top are the places the railway did transform: Frinton-on-Sea, Eastbourne, Southend-on-Sea, Seaford, Skegness, Cromer and Hunstanton, most of them growing slowly, or not at all, before the line. At the bottom are the resorts that had boomed before it: Weston-super-Mare grew 8.5% a year in the decades before its railway and 2.0% after; Torquay, Brighton, Clevedon and Ramsgate tell the same story. For them the railway arrived after the boom, not before it.
Over the whole period the resorts did grow more than inland towns, though by less than the famous cases suggest. Because parish boundaries moved, I measure long-run growth by chaining the same-boundary decades from 1801 to 1851 and from 1871 to 1911 (the nine decades every series in England and Wales can pair). Over those nine decades the typical resort multiplied its population 3.9 times (40 resorts with all nine); the typical inland town 3.0 times (15 towns). With so few inland towns that gap could be chance (Mann-Whitney p = 0.42).
Because small places can grow faster, I also paired each resort with the inland town closest to it in 1801 population. In 27 of 40 pairs the resort grew more (sign test p = 0.04), and the typical resort grew 1.35 times as much as its partner.
Type a resort's name to see its census figures and the year its railway arrived.
The story of the railway resort is real for some places. Six cases show the range. In each small chart the line is the census population (on a log scale, so equal slopes mean equal rates of growth) and the red mark is the railway.
I also split the resorts by their size in 1801. The 22 that had fewer than 1,000 people, places like Frinton, Skegness, Weston and Llandudno, look like the railway's creations. The 31 established towns, like Brighton, Margate, Scarborough and Great Yarmouth, look like towns it found. But measured by the change in growth around the railway, the two groups behave much the same: 61% of the small places with enough decades grew faster after the line, against 60% of the established ones.
No, not on these figures. The resorts were growing as fast as any inland town before the first train ran, and some of the famous ones were growing faster than almost anywhere. The railway brought a clear burst in the decade it arrived, worth about 1.3 points of growth a year, and the resorts kept on growing for the rest of the century, outgrowing inland towns of the same starting size in 27 of 40 pairs.
What the railway did was make the seaside bigger, sooner, and for many more people. It reached the resorts nearest the industrial towns first, and the burst of the arrival decade is clear. But inland towns that got lines show a lift I cannot tell apart from the resorts', so the rails alone do not explain the seaside's edge. That edge was already there before the trains, and it lasted: the sea, the holiday, and the money of the towns at the other end of the line.
Resorts. Every place on Wikipedia's List of seaside resorts in the United Kingdom (218 entries; 205 are places), placed with the coordinates in its article (7 from OS Open Names). Northern Ireland's 4 resorts are on the map but have no census series here.
Railway year. The earliest Wikidata opening date (P1619) of any station within 3 km. Wikidata sometimes holds the date of a rebuilt station, so I compared it with the resort's own Wikipedia article and used the article's year where it was earlier (7 resorts, all quoted in data/resorts.csv), and corrected 5 inland towns from their station articles. A station 3 km away is not the same as a line into the town centre, and opening years do not say how good the service was.
Census figures. Total population of each resort's census parish or township, 1801 to 1911, from the "Total Population" pages of A Vision of Britain through Time (University of Portsmouth), which reproduce the published census reports. I read 770 figures from those pages with an automated web reader and recorded each with its page link in data/vob_transcriptions.tsv. All 5 figures I could check against numbers quoted on Wikipedia matched exactly. I found usable series for 68 of the 201 resorts in Great Britain, 50 with both 1801 and 1911; many others are suburbs, or were split between parishes, or were created as parishes too late. The sample leans towards England and towards resorts that were separate parishes.
Growth. Each decade's growth uses the figure for ten years earlier restated on the same boundary where the census printed one. The English and Welsh parish tables have no 1861 figure, so I do not use 1851 to 1871 for those places rather than risk a boundary change looking like growth. Parishes are not towns: a resort can be a small part of a large parish (Perranporth in Perranzabuloe) or one of several (Hastings, left out).
Comparisons. Inland towns: a random sample (seed 1801) of 60 OS Open Names towns at least 25 km from the sea; I found census series for 20. Inland villages: 288 Wikidata places with census figures, mostly in Cambridgeshire, Huntingdonshire, Dorset and Somerset, so a rough check only. None of this proves cause: railways were built to places expected to pay, which is exactly the problem the before-and-after design tries to handle.