(Revised and extended.)
In the UK voting results, the top three parties’ voting percentages appear to be 36-29-23. So, let’s repeat the exercise from just before the election, based then on polling averages: how (un)common is such a close vote among the top three under first-past-the-post elections?
Of 210 elections in my database (20 jurisdictions, the most recent 6-17 elections in each jurisdiction as of 2006), how many saw the first party with less than 37% of the vote and the third with more than 22%?
Just four: Nova Scotia 1998 and 2003, Ontario 1975, and the last (2005) UK election. Quebec 2007 would add a fifth (and I am not aware of any others since 2006–till now.)
A related question is: how often has the third party been over 22% of the vote, yet under 10% of seats (as the LibDems almost certainly will be in the final result)? The answer is eight out of twenty three, but only one of those in which the first party was under 37%. That case was UK 2005. And among the larger set of severely under-represented third parties despite 22% (or more) of the vote, we also find the 1983 and 1987 UK elections.
Normally, that is, when a third party has as large a share of votes as the LibDems have won in the last two elections, that party has sufficient regional support to win a substantial share of seats. The UK, on the other hand, is unusual among FPTP systems in featuring a persistent third party that has relatively little in the way of regional strongholds. It has much of southwest England and a good presence in Scotland, but otherwise continues to run no better than second place in most of the country. And second just does not cut it under FPTP.