Bellwethers

Which constituencies go most consistently with the national tide?

‘Bellwether’ is an electoral term derived from sheep farming. A wether is a castrated ram, and it is (or was) the practice of shepherds to place a bell round the neck of one of them so that the sound when the wether moved would indicate to the rest of the flock which way to go. The Oxford English Dictionary records the first verified use in the literal sense, to talk about sheep, in 1430 but it is likely that the metaphorical use for ‘leader’ is also of considerable antiquity. In elections, bellwethers are the constituencies whose movements reflect the way the national flock is going: ‘As goes X, so goes the nation.’

I’ve always been interested in the unusual cases, ‘the curious, the marginal and the bizarre’ as the late David Butler once put it to me. This is probably an inclination that goes beyond my interest in elections, but my love of electoral history is influenced by this - there is no shortage of odd stories as the ballot is stranger than the bullet.

In elections, as in other fields of life, one form of weird behaviour is to be rigorously typical. Most of us have our moments when we wander off from the herd, and most of us like to think we have consistent preferences about important things. Constituencies tend to be like this; loyal to a party except when it has really annoyed the electorate or when there is a powerful local issue or magnetic (or repulsive) personality on the scene. On the eve of the 2024 election, the Electoral Reform Society reported that 247 seats had been held by the same party continuously since 1974, and 111 continuously since 1924. The 2024 election was extremely unusual for the large number of seats changing hands - unprecedented at least since 1918. Many hitherto safe seats - particularly Conservative ones - did fall. An element of stability in representation is the norm, rather than the exception. Most of the constituencies that have changed hands have not done so often. Sometimes they have had an election or two away from their usual party thanks to national landslides (like West Bromwich West in 2019, or Braintree in 1997 and 2001). Sometimes constituencies have a short period of transition as a party’s vote declines locally (like Liverpool Garston in 1974, or Bosworth in 1970). Sometimes boundary changes can switch seats in and out of major party marginal status (like the change from Dewsbury to Dewsbury & Batley in 2024).

Some seats have temporary bellwether status because of national electoral realignments; for instance Oxford (East) was a slightly Tory-inclined bellwether between 1964 and 1987 on its way from being a safe Tory seat to safe Labour. The contemporary rise of the student vote, and the reddening of the academic vote, explain this trajectory.

The class of constituencies that go with the national winner each time are therefore, in their way, strange places. They require analysis and explanation just as much as places where there have been dramatic electoral changes, particularly when they sustain their bellwether status for a long time despite the changing demographic bases of party support and large local movements of population. We should not expect a place which was marginal in the era when class defined electoral allegiance to still be marginal in today’s chaotic environment where age and education are more important. And yet, there are such places. Where are they?

Bellwethers by consistency

Bellwether is a journalistic rather than political science term, but let’s try to quantify and measure the quality of being a bellwether. One way of doing so is to count backwards from the last election and see how many seats have a perfect record of voting for the national winner¹ (the seat is ‘accurate’ for a number of elections). There are 411 constituencies that voted Labour in 2024; let’s call this capacious group bellwethers grade 1. How many count as grade 2, accurate over two elections by virtue of having also voted Conservative in 2019? The answer is complicated by boundary changes for two reasons. One is that sometimes the ‘notional’ result on new boundaries is different from the actual result in the predecessor seat - for instance, the old Beckenham seat was Conservative in 2019 (and for a long time before that) while the new Beckenham & Penge would have been narrowly Labour in that election. This problem can be solved by an editorial decision - we take the actual result in the seat that existed in 2019 as definitive. Beckenham & Penge is a bellwether grade 2.

The other issue is less easy to resolve; boundary reviews make new seats appear and old seats vanish, and there is sometimes no clear relationship between any old seat and the new seat (or where multiple seats can make reasonable claims to represent the continuity). Earley & Woodley was a new seat in 2024, drawing mostly on territory previously included in Reading East and Wokingham, both of which have other successor seats (Reading Central and Wokingham). The figures for consistent bellwethers become fuzzy round the edges, and even fuzzier with each boundary review as we go further back.² But with a bit of contestable matching for a handful of seats, we can say that there are 181 bellwethers grade 2 accurate for both 2019 and 2024. That is a lot; there was a high turnover of seats because a comfortable Conservative majority was succeeded by a Labour landslide, a rare conjunction of events - there are usually small majorities or hung parliaments either before (1979, 1997) or after (1964, Feb 1974, 2010) an election when the government changes.

Let’s go back a bit further. There were quite a few seats which went Conservative in 2019 which had previously been Labour-inclined - the famous ‘Red Wall’ and therefore did not follow the swing of the pendulum before that. Of our 181 bellwethers grade 2, how many make it to grade 3 by voting Conservative in 2017? The answer is 133, the drop of 48 reflecting both the Red Wall and the marginal seats Labour gained in 2017. There was less churn in previous elections and the number of accurate bellwethers only falls slightly - to 127 in 2015 (grade 4) and 117 in 2010 (grade 5).

The accurate constituencies back to 2010 include traditional marginals but also a swathe of historically Conservative seats that Labour won for the first time in 2024 - Aldershot and Aylesbury have perfect records if you only start measuring at 2010, but neither can really be called a bellwether. The 2005 election - where a bellwether needs to have voted Labour - weeds out these cases. There are only 55 grade 6 constituencies, and because Labour seats in 2005 tend to have also been Labour at the elections of 2001 and 1997 there is not much further winnowing - the same 55 are grade 7, and 54 of them³ are also grade 8. Interestingly, the next change of government does not make much difference - if a seat has followed the national trend since 1997, the chances are very much that it also did so in the Thatcher-Major years. There are 46 bellwethers accurate back to 1983 (grade 11) and 38 back to 1979. These seats still include several that are consistently to the right of the median constituency - Stafford for instance - because we have not so far included an election that has been a narrow Labour win. The two elections of 1974 narrow the field considerably - there are only 12 bellwethers back to October 1974 (grade 13) and 10 - listed below - back to February 1974 (grade 14).

Several of these cases are arguable, because continuity over four sets of boundary changes can be weak and it’s not really clear that Loughborough’s accuracy in 1974-79 can really grant both its successor seats (created in 1983) grade 14 status, nor that Sherwood/ Sherwood Forest can appropriate the pre-1983 Newark seat’s continuity. The Mid Derbyshire seat that has existed since 2010 can also claim some of Belper’s aura, particularly as it too has gone with the national winner, but South Derbyshire is traditionally regarded as the successor seat.

Statue ‘Don’t Worry, Son’ by Ray Lonsdale, 2011 standing in Swadlincote, the principal town of the current South Derbyshire constituency. Image from Visit South Derbyshire

How far back can we push it? Most of these seats or their predecessors were Labour in the 1970 election - only two (Dartford and South Derbyshire) voted with the nation then. They voted Labour in 1966 and 1964 (grade 17) - but there the line dies out because they were also Labour in 1959 despite the national Conservative victory. Dartford and South Derbyshire, with 60 years of voting with the nation spread over 17 successive elections, are jointly crowned as Britain’s most reliable bellwethers.

Dartford’s most famous feature. Image from Connect Plus

Scotland and Wales

Because of England’s preponderance within the United Kingdom, and the alternating dominance of Labour and Conservative, the hunt for the best bellwether is largely confined to England. The non-English seats with the longest track record are Vale of Glamorgan/ Barry and Bangor Aberconwy/ Conwy, both of which are grade 12, going back as far as 1979. There are no seats in Scotland that voted Conservative in 2019 and Labour in 2024, so grade 1 - for the 37 Scottish Labour seats in 2024 - is as good as it gets.

But what if we disaggregate the UK and look at which seats follow the Scottish and Welsh⁴ national winners? For Wales it is relatively easy - Labour has won the most seats since 1922 so instead of looking at marginals, one should look at Labour strongholds. There are six current constituencies that go back to 1922, three of which (Neath & Swansea East, Aberavon Maesteg and Pontypridd) extend even further back by voting Coalition Liberal in 1918.⁵

For Scotland, we need seats that were Labour in 2024, SNP in the previous three contests and then Labour before that as far as 1959. This is not an uncommon pattern - most of the Central Belt has voted this way. The real test is 1955, when the Tories won a majority. Only Central Ayrshire, Glasgow West (Scotstoun in the 1950s), and arguably Glasgow South West (if you regard it as a successor to the Craigton seat that existed in the 1950s) fit the bill. Looking further back, 1951 was a tie in Scotland so we’ll allow a bellwether to vote for either side then - the two Glasgow seats are grade 20(S) because their predecessors were Tory in 1950. Central Ayrshire is the champion Scottish bellwether (grade 21(S)), voting for the Scotland winner consistently since its creation in 1950.⁶

Separating out Scotland and Wales raises the question of what makes an English bellwether, as that requires a different electoral history from a British/ UK bellwether. An English bellwether will have voted Conservative in February 1974 and 1964, and been part of the small contingent of Labour gains in October 1974. The best bellwethers for England are therefore the two successors to the pre-1983 Rochester & Chatham constituency - Rochester & Strood and Chatham & Aylesford, both grade 16(E) for going with the English seat winner continuously since 1966.

Bellwethers by accuracy

Hang on, you might say. Could we be missing some usually reliable bellwethers by insisting on a perfect recent record? Should a ‘wrong’ vote in one general election really disqualify? Which is a better bellwether, the seat with a perfect score dating back to 1964 or the seat with a 90 per cent record back to 1918?

If one allows for a single lapse, some other constituencies are of similar long-range accuracy to Dartford and South Derbyshire. Harlow, other than sticking with Labour in 1979, has, with its predecessor Epping, a perfect record back to 1951 (and only two lapses in total since 1931). The seat with Stevenage in it has also voted for the national winner since 1951 other than when Shirley Williams kept Hitchin Labour in 1970. Gravesham/ Gravesend voted Conservative in 2005 but is otherwise accurate back to 1955. It was the constituency chosen for an experimental exit poll for election night 1970, and the poll results - showing a Conservative gain - were accurate despite being treated with some scepticism in the television studio. Neither Basildon & Billericay nor South Basildon & East Thurrock voted Labour in 2024, spoiling a strong bellwether record - the second time Basildon (loosely defined) had got the ‘wrong’ answer since 1951, the other occasion being a Tory hold in Billericay in 1964. We may note in passing that New Town constituencies are consistently among the best bellwethers, despite being socially and historically atypical and massive past population growth - for example, Harlow’s population was 5,571 in 1951 but nearly ten times as many by the time of the next census in 1961.

The table is not exhaustive; it is possible in particular that there are some once-reliable bellwethers that have had a poor record since 1974 and therefore not been picked up. Islington North, for instance, was a perfect bellwether from the 1918 election until it stayed Labour in 1951 - yesterday’s bellwethers are not necessarily today’s, and we often cannot tell when a seat is about to give up its status.

Taking a very long view, the best bellwether is a surprising one. Step forward Buckingham & Bletchley. The seat was created as such only in 2024, and the growth of Milton Keynes since 1967 has meant frequent and radical boundary changes in the area that stretch the continuity of the idea of ‘the same constituency’ to breaking point, but let’s run with it. The seat containing Bletchley has gone with the winner in every election since 1931 except for the two 1974 contests when Robert Maxwell was attempting to regain Buckingham for Labour. It stayed Tory in 1929 but other than that and 1974, Buckingham (including Bletchley) has gone with the winner each time since 1868.

Bletchley Park - eccentric, or typical?

But what does it mean?

I’ve identified the best bellwether constituencies, but I haven’t drawn many conclusions yet from the exercise. The question that occurs to me is - are these places actually typical of the nation? Or is something else going on? The preponderance of New Towns among the high-performing bellwethers suggests the latter. But this is already a long essay, and I’ll come back to the deeper questions in a bit…

TO BE CONTINUED

1. The national winner is defined, for these purposes, as being the party with the most seats in the House of Commons after the election. Different definitions of the national winner are available - the UK-wide popular vote plurality party was not the seat winner in February 1974, 1951 or 1929. The largest party in seats did not provide the Prime Minister in 1931, 1923 or 1918. Different views could be taken of smaller parties in electoral pacts with large ones, such as National Liberals in the 1930s or Labour before 1918.

2. Part of the fun, for the author at least, is in this backward tracing of constituency identities. See https://www.parlconst.org/ for mapping of where is in which constituency since 1885.

3. The difference is South Dorset, which voted Conservative in 1997 but Labour in 2001 and 2005.

4. There are no seats in Northern Ireland that voted Sinn Fein in 2024 and DUP in 2019, and therefore no bellwethers better than grade 1(NI).

5. The predecessor seats were large Glamorgan constituencies but if one is a bit loose with the continuity one can claim these seats as grade 34(W), going with the Welsh winning party back to 1885.

6. Both its predecessors were Conservative in 1945, so the bellwether continuity definitively stops there.