I was once - in a very previous life - involved in making submissions to past boundary reviews and from that experience think that it is important to understand why fair boundaries cannot be created under first-past-the-post.

It is all to do with political geography/sociology - and there is a political history to the current boundary bias. The law states - not unreasonably - that the criterion of ensuring even electorates is the most important, but that it is important to consider natural communities. We know - with some important exceptions - that the more urban a constituency, the more likely it is to elect a non-conservative - in most of the country - a Labour MP. The shift in population from the major cities to the suburbs has resulted in a reduction in electorates in major cities and former industrial areas, while it has increased suburban and county electorates. This should, according to much conventional wisdom, favour the Tories, as the big cities concern safe Labour seats and the counties are regarded as traditionally Tory. But this isn't quite the case, and this gets to the heart of the reasons the current system contains much pro-Labour bias.

Thirteen years ago after the 1992 election many newspapers took this conventional wisdom and argued that new boundaries would produce a greater pro-Tory bias. This didn't happen for two reasons. Firstly and most importantly, the increase in seats allocated to the counties made it possible for the creation of seats in some counties that were purely urban, thus enhancing Labour's chance in them. Bedfordshire is a good example - if the 1997 election had been held on pre-1992 boundaries, Labour would have won the old Luton North seat easily, the old Luton North seat narrowly (by something under 2,000) and missed out in the old North Bedfordshire seat (by under 2,000). In fact in 1997 under new boundaries the reduction in the size of seats enabled Labour to win Luton North and South by large margins, and the new marginal of Bedford comfortably. Secondly, the concerted Labour response to the commissions interim proposals, exploited the provision that boundaries respect natural communities to argue against the creation of seats that split urban centres into divisions and included rural hinterlands, and instead argue for purely urban and purely rural seats in their place. Some seats, like the Swindon North and Swindon South, were instances of the exceptional cases where Labour was unsucessful.

This logic is intensified in the current boundary review. Let's look at Warwickshire, which had five seats in 2005 and will have six next time. In 2005 we have three Labour seats in the county - North Warwickshire, which must count as a relatively safe Labour seats, Nuneaton which is a Labour marginal (majority of 2,000) and Warwick and Leamington (which was almost a dead heat). Of the two Tory seats Rugby & Kenilworth is a marginal, with solid Stratford-on-Avon. Next time the Tories will have two ultra-safe seats, but Warwick & Leamington will lose its rural hinterland while Rugby will lose Kenilworth and the conservative-voting areas to its south. Warwick & Leamington would have been held by over 4,000 on its new boundaries in 2005, Rugby by around 3,000, the Nuneaton majority would have doubled to 4,000, as a result of losses of rural wards.

These kinds of shift also have a less politically obvious, sociological effect. Every seat becomes more homogeneous socially - there are more purely inner-city, purely urban, purely suburban and purely rural seats. Thus parliamentary boundaries re-inforce the fragmentation of the electorate - broadly speaking this has concentrated the Tory vote in rural seats and diluted its impact to varying degrees in the other three categories of seats. Thus, the fact that the Labour vote is more thinly spread becomes an advantage even on a relatively low share of the total vote.

Correting this under FPTP could be done, but only at a cost. It would involve effective national, or regional allocation of the seats - rather than by local government area. It would also mean abolishing the provision for the respect of natural communities.

The two principal costs are firstly, that this legal framework is similar to that under which congressional districts are drawn in the United States, or National Assembly constituencies are drawn in France. Both are highly vulnerable to naked political gerrymandering because of the lack of legal control over the drawing of the boundaries. Second, boundaries would result that would bear no relation to the patterns of settlement and peoples' patterns of identification - thus weakening the links between individual and constituency and thus individual and constituency MP; undermining one key argument for FPTP.