The UK does have some PV manufacturing capability.  Sharp have a factory in Wrexham which was meant to produce 20MW in 2004 with plans to double capacity... however, I'm not clear whether this is actually cell manufacture or just module assembly?

I have -20% over 20 years in my head from somewhere... a search turned up this:

The project has identified several faults likely
to impair the output of module in the long run; these faults include delaminating of the encapsulant, damp penetration, hotspots, manufacturing and assembling defects, breakages,
cracks and decay in other components. These problems notwithstanding, the project has shown that TISO 10kW is still a perfectly sound system. The annual average loss of power has been 0.2%, an indicator of good endurance. In other words, installations of this kind might, according to estimates, enjoy a lifetime of approximately 40 years with a steadily efficient output. Link pdf

How Long Do PV Modules Last?

PV modules last a long, long time. How long we honestly don't yet know, as the oldest terrestrial modules are barely 30 years old, and still going strong. In decades-long tests the fully-developed technology of single- and poly-crystal modules has shown to degrade at fairly steady rates of 0.5% to 1% per year. First-generation amorphous modules degraded faster, but there are so many new wrinkles and improvements in amorphous production that we can't draw any blanket generalizations for this module type. The best amorphous products now seem to closely match the degradation of single-crystal products, but there are no long-term data. All full-size modules carry 10- to 20-year warranties, reflecting their manufacturers' faith in the durability of these products. PV technology is closely related to transistor technology. Based on our experience with transistors, which just fade away after 20 years of constant use, most manufacturers have been confidently predicting 20-year or longer lifespans. However, keep in mind that PV modules are only seeing six to eight hours of active use per day, so we may find that lifespans of 60 to 80 years are normal. Cells that were put into the truly nasty environment of space in the late 1960s are still functioning well. Link

I don't think that last comment about extending to 60-80 is right, I bet the solar duty cycle is included in the manufactures figure. Generally 20-25 years seems to be quoted but I don't think that's to total failure though, it's to something like 80% of rated output.  

Pondlife, A number of points:-
Photovoltaic generation has been around a long time. I remember reading a book about it as a boy 45 years ago. Most of the basic patents will have long since expired. However as in all complex industrial processes there are always a host of patents on minor aspects. Photovoltaic production is fairly simple technology compared to semiconductors. Doing it cheaply is the difficult part. It is no more poisonous and energy intensive that many other industries

We don't have to go to Japan, there is plenty of production in Germany which is now installing more capacity than the rest of the world put together, over 850MW in 2005 out of a world installation of over 1480MW from crystalline silicon and another 176MW by other means (the UK managed 4MW). European producers have about 26% of the market and the Americans 12%

We cannot produce tonnes of the stuff? According to this report in 2005 10,000 tonnes of silicon was produced for photovoltaic use (out of 30,000 tonnes of semiconductor silicon) and another 7,000 tonnes was used from stockpiles from previous years. This year production will increase 30% to 13,000 tonnes but there is nothing left in stock. This will reduce the amount of silicon available and increase the price of polysilicon still further. It has a already increased it from $30/kg to $60/kg. Fortunately cells are being made with progressively less silicon and output is expected to rise this year to 1738MW rated capacity.

However it is estimated that soon the production squeeze will have eased and by 2009 photovoltaic use will have overtaken electronics and be using 27,000 tonnes of polysilicon a year to produce 3400MW of rated generation which with other photovoltaic will bring the total for the year to 4100MW. This will bring total world installed rated capacity to about 15.6GW.

Tracking the sun ideally  gives only  a 38% increase over an optimally angled fixed panel at UK latitudes and given that the UK tends to be more cloudy morning and evening it is probably less than that. A multi-section tracker on a roof with sections paced to avoid self shadowing will not give more than a fixed panel covering the whole of same roof area. The only advantage is that the total area of the spaced out trackers will be less than the continuous fixed panels. I doubt the gain is worth the cost of the scanners and the reliability problems they might cause. Interestingly the book that I mentioned above and that has stuck in mind suggested building houses that float in a circular pond so that the whole house could turn to track the sun. I wondered then and I wonder still how the sewerage would connect in such a system.

I strongly suspect Chris is right when he suggests that the 20 year live of photovoltaic panels is a cautious under estimate. I have had my system for over 2 years and have monitored it closely. There has been no drop in output at all and I think I could detect a 1% drop off. Space borne systems have survived many years in a vastly more hostile environment than UK roofs.

You say that gas central heating is a problem in the UK. I don't know of many houses that fell down because of such installations. Gas heating is still a lot cheaper than electricity and likely to remain so for many years. As long as most electricity is generated from fossil fuels (and this is not going to change much in the next 15 years) the price of gas and electricity are going to go up together. Domestic combined heat and power is the way to go if you are going to use fossil fuel.