The idea of dishing the Tories by Lib-Lab tactical voting has been around since 1983, when Labour and the Liberal-SDP Alliance split the non-Conservative vote and Margaret Thatcher won a majority of 144 on a mere 42 per cent of the vote. As the Conservatives went on to win the 1987 and 1992 elections on a similar minority, the attractions of tactical voting ceased to be the obsession of Liberal dreamers and gained ground on the left.
Tactical voting had worked for the Liberals at by-elections as early as the 1960s, and became a regular feature of spectacular Conservative by-election losses. At general elections, however, tactical voting was feeble. The Conservatives recovered almost all their by-election losses. The British Election Study estimated that a mere 3 per cent of voters cast anti-Conservative tactical votes in the three elections of 1983, 1987 and 1992. In the Nuffield study, John Curtice and Michael Steed calculated that tactical voting cost the Conservatives a maximum of eight seats in 1987 (one going to Labour) and 12 in 1992.
But 1997 was different. "The scale and impact of tactical voting," Curtice and Steed concluded, "was unprecedented." According to their estimates, the Conservatives lost 15-21 seats to Labour and 10-14 seats to the Liberal Democrats solely through tactical voting. It turned the Conservatives' inevitable heavy defeat into a humiliating rout. Labour's projected overall majority was inflated from about 140 to 177 and the Liberal Democrat group of MPs from about 34 to 46. These were cautious estimates, moreover, because they excluded the tactical voting that had begun to build up in 1992 and earlier.
Will tactical voting recur - or even spread further - this time round? Sceptics argue that the special circumstances of 1997 have disappeared. Labour and Liberal Democrat supporters were willing to abandon, temporarily, a lifetime's loyalty when the prize was kicking out a much-loathed Conservative government. But, with a competent if uninspiring new Labour government coasting to re-election, loyalists no longer have an incentive to switch parties. Tactical voting could easily slip into reverse, enabling the Conservatives to recapture most of the seats - listed on our map, which folds out from the centre pages - that they unexpectedly lost last time.
It could; but there are good reasons for believing it will not. First, anti-Conservatism remains the single strongest emotion among ordinary voters. New Labour's shine has undoubtedly faded, but poll after poll reveals even deeper disillusion with the Conservatives - deeper, in many respects, than in 1997. Compared with four years ago, the Conservatives lag further behind Labour on most issues, are even less trusted to manage the economy, and are led by someone who commands far less respect than John Major did. The predominant motive at the election among all but unswerving Tory loyalists will be to keep the Conservatives out at any cost.
Second, the primary reason for the growth of tactical voting in 1997 was not the strength of anti-Conservatism (it was pretty intense during the Thatcher years, after all), but the new convergence between Labour and the Liberal Democrats. New Labour's modernisation "project", combined with Paddy Ashdown's commitment to support Labour in a hung parliament, persuaded voters on the centre and left that the two parties had moved much closer together: switching between the two parties required a much shorter ideological leap than ever before.
Relations may have cooled in Westminster, and local Labour and Liberal Democrat activists may be at each other's throats in Liverpool, Sheffield and other big cities. But in Scotland and Wales they are in coalition government, and in the shires, suburbs and small towns of Middle England, where most of the marginal seats are, Labour and the Liberal Democrats remain natural anti-Conservative allies in the voters' eyes.
History suggests that tactical voting rarely goes into reverse. The Conservatives have regularly won back seats originally lost to the Liberals (usually at by-elections) through a squeeze of the Labour vote. But they have regained them on a general swing back to the Conservatives, not because Labour sympathisers ceased to vote tactically.
By-elections since 1997 underline the point. In Romsey, tactical voting at the general election had helped establish the Liberal Democrats as the clear challenger to the Conservatives. In the by-election, the Conservative vote slipped from 46 per cent to 42 per cent, a poor performance, but not enough to lose the seat if the large majority of Labour supporters had not lined up behind the Liberal Democrats, reducing their own party's vote share from 19 to 4 per cent. The outcome was almost unprecedented: the loss of a safe Conservative seat in the middle of a Labour government term of office.
Other by-elections since 1997 suggest that a significant proportion of Liberal Democrats are continuing to vote Labour in Con-Lab marginals. Scotland aside, three of the Liberal Democrats' four poorest performances have been in Con-Lab marginals: Uxbridge (down 5.3 per cent), Beckenham (up only 0.3 per cent) and Eddisbury (up only 0.6 per cent).
Far from subsiding, anti-Conservative tactical voting may actually increase this time. A substantial pool of Liberal Democrat and Labour supporters did not vote tactically in 1997, despite living in constituencies where it made sense to do so. Even in 1997, four out of five Labour supporters in Con-Lib Dem marginals and three out of four Liberal Democrats in Con-Lab marginals stuck to their party when it had no hope of winning.
Why should people vote tactically in 2001 if they were not tempted in 1997? One answer is that, after four years of Tony Blair in office, Labour supporters, while no less determined to keep the Conservatives out, will have lost some of their ardour for the party. Likewise Lib Dems will not be as enthusiastic now Charles Kennedy has taken over from Ashdown.
Another reason is that constituency boundaries had been radically revised in 1997 and many voters did not know the tactical situation in their constituency. This time, the local parties will incessantly remind them of the 1997 result. Moreover, both Labour and the Liberal Democrats will focus ruthlessly on the marginal seats they are defending or hoping to win. No formal pact will be needed for the Labour Party to concentrate all its efforts in South Dorset, which it failed to wrest from the Conservatives by 77 votes in 1997, leaving the Liberal Democrats to concentrate on Mid Dorset & Poole North (where they were 681 votes short) and West Dorset (1,840 votes short).
The map opposite lists the Conservative seats that would fall, in order of probability, simply through further tactical voting. It assumes that the parties' shares of the national vote remain the same as in 1997, which is what the current polls, uncannily close to those of March-April 1997, suggest. The absence of Scotland and Wales reflects the absence of Conservative seats in those countries.
In 1997, tactical voting produced a 3 per cent swing between Labour and the Liberal Democrats to whichever party was the main challenger in Conservative marginals. A repeat in the 2001 election would deliver an additional 13 Conservative seats to Labour and five to seven to the Liberal Democrats. This would reduce the Conservative parliamentary strength to below 150, bring the number of Liberal Democrat MPs to more than 50, and catapult Labour's majority in the Commons to beyond 200.
The actual impact of tactical voting will vary between constituencies. In 1997, the Liberal Democrats mobilised a larger tactical vote where they could build on recent local government successes: this could mean surprise wins in Orpington and Cheadle. But if the Conservatives pick up most of the vote that went to the Referendum and UK Independence parties in 1997, they could stop the Liberal Democrats in Totnes (on paper a more marginal seat) and Labour in South Dorset, Hexham and Bury St Edmunds. In all four seats, the combined vote for the anti-euro parties exceeded 5 per cent in 1997.
Why don't the Conservatives vote tactically to deprive Labour of seats? Simply because, even if they were minded to, there are virtually no constituencies where Labour is vulnerable to tactical voting. A 3 per cent swing from the Conservatives to whichever other party was best positioned to defeat Labour would not deprive Labour of a single seat. In this, as in so many other ways, it will be sitting pretty in June.
Ivor Crewe is vice-chancellor and professor of government at the University of Essex