Devolution is once again on the political agenda. It is likely to prove a big issue in the next general election. Labour has reaffirmed its commitment to Scottish and Welsh parliaments, while the Liberal Democrats are proposing a federal system of government for the United Kingdom as a whole. John Major, by contrast, has reiterated that the Conservatives remain a Unionist party; and he is likely to take to his soap box, as he did in 1992, to warn of the dangers of a break-up of the country.

Devolution, however, is not an issue solely for Scotland and Wales. It affects the interests of everyone in the kingdom. Devolution for Scotland alone would lead to an imbalance in the constitution. Over-represented in the Commons, and with their own Secretaries of State in the Cabinet, Scotland and Wales would, with legislatures of their own, be in a far stronger position to attract public funds which would otherwise go to other parts of Great Britain.

This would give rise to two constitutional anomalies. The first is the West Lothian question, obsessively posed by Tam Dalyell. Is it right that Scottish and Welsh MPs should be able to continue to vote on domestic matters such as education in England, when English MPs would be unable to vote on education in Scotland and Wales, since responsibility for education would have been devolved to the Scottish and Welsh parliaments.

Second, a basic principle of the welfare state would come under threat. For it is generally held that public funds should be distributed, on the basis of need and not of geography. Before the war, a doctor working in Bootle would receive a lower salary than one working in Surrey. The welfare state was established to rectify such anomalies. But devolution to Scotland and Wales alone would recreate them so that a deprived school in Glasgow would receive more funds than a similarly deprived school in Liverpool simply because Scotland had a parliament and the north-west of England did not.

England, therefore, is the key to devolution. Were a federal system of government to be set up, to include English regional assemblies alongside the Scottish and Welsh parliaments, the constitutional anomalies would disappear.

The argument for devolution in England, moreover, is, in one crucial regard, not dissimilar to that in Scotland. There is indeed a strong case for dispersing power from the centre in every part of the United Kingdom. For we are now by far the most centralised member state of the European Union, and one of the most centralised states in the world. Anything that can relieve the over-concentration of power in Whitehall and diminish the span of control of central government would, therefore, be good in itself.

Other member states of the European Union of a similar size to Britain - France, Germany, Italy and Spain - all have regional layers of government. The Maastricht Treaty, by creating an advisory “committee of the regions”, recognised their constitutional role in building European union. Since many European projects are planned on a regional basis, Britain is likely, in the absence of a regional layer, to suffer in the competition for funds. “Without effective regional groupings,” the Audit Commission declared in 1991, “the UK will lose out.”

The Department for Education, unlike, for example, the Department of Trade and Industry, is not organised administratively on the basis of the eight standard regions. Nevertheless, education could benefit from a regional structure of government. Functions such as education, training and employment would then be administered by one authority rather than being divided among separate Whitehall departments as at present. Vocational education in particular could be co-ordinated with regional employment needs much more effectively than at present. There is a strong case for further education and, in the longer run, higher education, to be administered from regional authorities in England, as they would be in Scotland and Wales from local parliaments.

The trouble with regionalism in England, however, is that, by contrast with Scotland and Wales, there seems hardly any demand for it. In many parts of the country, indeed, the regions are insubstantial ghosts, not living realities. It is hardly surprising that the Labour party is having difficulty in deciding a policy on English regionalism.

Spain after the death of Franco faced a similar problem. There was a strong demand for autonomy in the historic provinces - primarily the Basque country and Catalonia - but little if any demand elsewhere. The solution was found through the principle of rolling devolution. Autonomy was given to the historic provinces, and it was announced that any other region of the country which sought autonomy could also enjoy it. Devolution in the Basque country and Catalonia proved, indeed, so successful, that it was not long before demand increased in the other provinces, so that Spain today is a state of autonomous communities, a quasi-federal state, In Britain, Labour should lay down two principles, first that devolution will be given only to nations or regions which seek it, but second that a region which rejects devolution cannot prevent another region from embracing it.

There is, however, one part of England where there is already a strong latent demand for devolution and that is London. It is absurd that London should be one of the very few, if not the only, great capital of a democratic country to have no authority of its own. Moreover, it would be far easier to implement devolution in London, which already has a unitary layer of local government, than it would in other parts of England where, by contrast with Scotland and Wales, the two-tier system is likely to survive largely intact. To introduce devolution on to a two-tier structure of local government would be, surely, to introduce one layer of government too many.

But it would not be sensible to create another GLC, a body which had few substantive powers, and which came to be squashed between the boroughs and Whitehall. Far more attractive would be a new London regional authority with powers similar to those granted to the Scottish and Welsh parliaments (London, after all, has a larger population than either Scotland or Wales), powers over such matters as economic development, health, transport and the non-national functions of the Metropolitan Police as well as education. If the London authority was elected, as the Scottish and Welsh parliaments are to be, by proportional representation, there would be little danger of extremists coming to control it as happened with the GLC.

The besetting sin in the half-hearted local government and devolution reforms of the past 30 years has been its piecemeal nature. We have sought to reform sub-national government without any clear idea of what it is that we want local government actually to do. We have then put forward proposals for devolved assemblies which would have fitted into the structure of local government like a jellyfish on a bed of nails. Above all, our reforms have been undertaken in isolation from what is happening in the European Union as if we were not a member state but a solitary island separate from the main.

In starting a debate on regionalism, the Labour party has challenged us to seek a new and more rational structure of sub-national government and one more congruent with our membership of European Union. Do we still have the confidence to meet that challenge ?

Vernon Bogdanor is Reader in government at Oxford University and a Fellow of Brasenose College. His book Devolution was published in 1979. He is at present working on a book on the British constitution.