The House of Commons has again opted against providing the two-thirds support needed to hold an early general election. However, it will be hard to avoid one for much longer. And although the outcome remains far from clear at this stage, it’s interesting to consider where the election is likely to be won and lost.

Thinking about different swings in different areas is a relatively recent concept. During the postwar two-party phase from 1945 to the 1970s, during which the Liberal vote share was sometimes even lower than in 2015 and 2017, swings were incredibly uniform. The national vote shares, or the first few seat results on election night, basically told the full story.

Indeed the uniformity of swing was so reliable that in 1970, the BBC conducted an exit poll (the first in Britain) in a single seat – Gravesend in Kent. This was chosen because of its representative demographics and bellwether voting history. Much like in 2015, the exit poll showed that contrary to all expectations, Ted Heath’s Conservatives had won locally on a 5% swing from Labour that, true to form, matched the national swing.

These days you can’t do that. It’s not quite 650 entirely separate elections, but there are clearly multiple dynamics at play, and therefore multiple battlegrounds.

The most obvious battlegrounds, barring something earth shattering, will be the Conservative-Labour marginals. I say “battlegrounds” in the plural, because this subset of seats has itself subdivided. In 2017, the overall swing of 2% to Labour masked some dramatic shifts in both directions, chiefly along Brexit lines, with Labour dramatically gaining seats like Kensington and Canterbury but losing seats like Mansfield and Stoke South.