The original mathematical construct Butler swing is defined as the average of the Conservative Party percentage-point gain and Labour Party percentage-point loss between two elections, calculated on the basis of the total number of votes (including those cast for candidates other than Conservative or Labour).
In a contribution to The British General Election of 1945 he wrote "this measurement of 'swing', admittedly imperfect, does give us a broad idea of the movement of opinion from Conservative to Labour" and went on to compare the swings in each area of the country.
This made it easy to predict the final outcomes of general elections when few actual results were known, as the swing in the first constituencies to declare could be applied to every seat.
The swing enabled prediction of outcomes to be made even while safe seats were returning results whose victors were not in doubt.
Election-night television programmes from 1955 have usually featured a device known as the "swingometer" which consisted of a pendulum which could point to the swing nationally and illustrate the outcome.
The continuation of the first-past-the-post system, and the tendency for smaller parties to only run in some constituencies, made it increasingly difficult to use measures of swing to predict results.
The success of the SNP in Scotland and Plaid Cymru in Wales, especially in elections to the Scottish Parliament and Welsh Assembly, has also had an effect.
This has led to swing becoming a measurement of the changes in votes of the two biggest parties in the constituency in question, rather than just Labour and the Conservatives.