Ever since Gordon Brown became Prime Minister at the end of June he has had a lead in the polls over David Cameron's Conservatives. So speculation is rife that he will call an autumn election.
I have already pointed out a couple of flaws with this. He would need to do so in September, on the basis of three months opinion polls, and two of those months would be July and August, when most people's minds are not really on politics. He is too cautious for that. He remembers 1970, when Harold Wilson, having trailed in the polls for some time, called an election on the basis of a few months in the lead, and lost.
Secondly, there is nothing to be gained from an election. If he called an election after four months in power and won, the Blairites would claim the victor as theirs: Tony\'s Fourth Term. And if he lost, he would be the shortest serving Prime Minister since Canning, almost 200 years ago. Neither result would be ideal for a man of such monumental self-belief.
But a further factor has occurred to me. What if the polls are right, and Brown's lead is maintained? What if Labour gains votes in the election, but he still loses?
This seems entirely credible to me. I have always expected that under Gordon Brown's leadership Labour would be more popular with its traditional voters, but somewhat less popular with the voters Blair called 'Middle England'. Tony Blair is the only Labour leader alive who ever successfully connected with that constituency.
At first glance it seems ridiculous to suppose that Labour could gain votes and lose seats. This has never happened. A party (Labour) has won an election (Feb. 1974) with fewer votes and more seats before, but no party, as far as I am aware, has gained votes while losing seats. It would be an aberration.
Except, that it wouldn't. It was the 2005 election - and the two preceeding elections at which Blair also lead Labour - that were aberrations. 2005 was a particular doozy. Labour won 36% of the vote and had a two point lead over the Conservatives but won a 66 seat overall majority and and around 150 seats more than the Conservatives. By contrast, in 1992 the Conservatives had 43% of the vote and an eight point lead over Labour, but won a majority of just 21. The last time the aggregate vote distribution was anything similar to 2005 was October 1974, when Labour won an overal majority of three.
All three of Tony Blair's victories were achieved with much exaggerated Commons majorities not so much because of how many people voted Labour, but where they did. Labour made major gains in the marginal seats while losing ground in its heartlands. Turnout dropped markedly in safe Labour areas such as Wales, Scotland, and northern industrial towns and cities.
Brown could win those votes back and achieve nothing other than ever larger majorities in the Welsh valleys and central Liverpool and Newcastle.
My own guess - and this is very hard to project, if voters in one area are not behaving as voters in another area do - is that Labour could win exactly the same number of aggregate votes as in 2005, but lose 50 or 60 seats, if we returned to pre-Blair voting patterns of piling huge Labour majorities in traditional Labour areas. The votes that Blair won over - floating voters in marginal constituencies - were simply worth much more than the ones he alienated on northern council estates.
So if Brown reconnects Labour with its roots, he could win more votes but no more seats. And if he seems less English, and less middle class, he could lose votes where they really count.