A key factor in each of the two cases above was that the No votes themselves were a freak, an aberration, outside the trend of public opinion. They were not representative of the sustained view of public opinion. (They still counted, of course, which was why they set the political agenda, but it was realistic and reasonable to suppose that they might change.)

No-one can say that the Irish vote last week was a surprise. The referendum results in France and the Netherlands in 2005 presaged an era of suspicion by voters of official policy on Europe everywhere. The determination by the heads of government that the Lisbon treaty should avoid referendums where possible is testament to that.

Is there the prospect that a second referendum might reverse the outcome of the first? Of course it is possible, but one cannot view the prospect with the same confidence that might have been justified seven years ago. Public opinion is different now, and the European project has got to recognise that.

Even if the necessary guarantees can be given to the Irish voters and a second referendum held and won as a result, there are still good reasons to doubt that this is the best outcome.

First, it reinforces the idea that referendums do not really count. Opponents of the EU love to argue that a Yes means yes, and a No means yes a year later. That is a very damaging argument to make and it is not wise for supporters of the EU to add fuel to it. The fact that the second referendum result was in fact a Yes gets lost in the face of the suggestion that the second referendum should never have been held.

Secondly, it might lead to the idea that the process set in train by the Laeken Declaration has now been completed. The European Council meeting in December 2001 set out at Laeken a series of concerns about the functioning of the EU that needed to be addressed. These included some concerns about the effectiveness of the way the EU acted, but also some concerns about its relationship with the citizens.

The mess that has been made of the ratification first of the constitutional treaty and now of Lisbon only emphasises the scale of the problem. It is bigger than can be fixed by a set of treaty amendments, particularly this set of treaty amendments.

Now, there are some advantages that would flow from a second Irish referendum that produced a Yes vote – we would get the Lisbon treaty, for a start – but no-one should imagine that the problems of the EU would be magically solved as a result. We might get a short-term fix, merely postponing the difficult and important issues into the long term.

A second Irish referendum, therefore, is not a neat and tidy solution to the problem facing the EU. There are still democratic challenges ahead waiting to be solved.

Richard Laming is Director of Federal Union.