The safety case against nuclear tends to be dominated by reference to Chernobyl. But an earlier near-disaster is perhaps more germane: Three Mile Island, or "TMI2" as it is known, at Harrisburg in Pennsylvania. The Chernobyl reactor design was generally acknowledged to be flawed, long before the disaster, but TMI2 was a direct ancestor of today's PWR designs.

TMI2 happened because private sector owners cut costs and because managers became complacent and lazy. Humans being what they are, this can happen in any organisation, regardless of ownership. In most commercial organisations what happens when managers take their eye off the ball is that production gets inefficient, profits fall, and bankruptcy and unemployment ensues: the market exacting its fair toll. With nuclear power stations (and fuel storage depots and chemical production plants like Flixborough and Bhopal and various other highly-dangerous activities) what happens is that whole communities suffer; with nuclear, whole regions can be devastated as happened at Chernobyl.

It's important that we don't put too much reliance on the ability of humans to keep future reactors safe throughout their lifetime. (suppose a flu pandemic meant that most of the staff were off sick...). Nuclear reactors *can* be designed so that if you stop looking after them they just stop producing electricity. But most need active cooling systems to remove heat from the core, and when (as at TMI2) the backup cooling systems fail, catastrophic heat build-up, so-called meltdown, can ensue.

Generally, though, the nuclear industry has an excellent safety record; it and its engineers know more about radiation than anyone else so it treats it with due respect. What we as the public need to beware of is falling for the line that because it has a good record it is intrinsically-safe. It isn't; it depends on the competence of engineers to keep it so. Good engineering from the start will ensure that it is safe even when there are no more engineers.

What is 'intrinsic' safety?

Are railways 'intrinsically' safe? Hurtling lumps of metal around at 125mph seems intrisically dangerous to me.

I think most things we have and do have potential (intrinsic?) dangers, and safety depends on good design and engineering and perhaps vigilance. Is nuclear power any different?

Train braking systems are, however, a good example of something designed to be fail-safe (the wagon at Tebay had had that feature over-ridden); a wagon is braked unless it is connected to a locomotive's pressure or vacuum line, and locomotives have a dead-man's handle. If the driver has a heart attack, the train stops; but if the cooling systems on a nuclear reactor fail, you get meltdown.

Good engineering counts for a lot, but relying on vigilance doesn't seem to me to be enough.

It's pressurised air these days, rather than a vacuum. Vacuum brakes are a steam age thing, but the point is much the same.

So are you saying that nuclear plant are not designed like that? That if the Homer Simpson falls asleep the thing will melt down?

-- I did say "pressure or vacuum"...

Re the second point, yes, I'm pretty sure this is the case (but relying to some extent on memory, so could be wrong). A reactor core produces heat which has to be removed by cooling systems (and the heat raises steam to generate electricty). If the heat isn't removed, it gets too hot. The only way to switch the reactor off is to put more neutron-moderating materials inside, which stop or slow down the chain reaction. PWR designs are mostly water-moderated; if pressure inside the reactor vessel falls, it turns to steam which is a much less effective neutron moderator. This is part of what happened at TMI-2, after the initial emergency coolant system didn't work properly because the feed valves were closed for maintenance. Chernobyl was graphite-moderated and gas-cooled; when air got into the coolant circuit, the graphite started burning.

So anyway, because nuclear plant aren't failsafe like train braking systems they use multiple redundancy, which is - imho - a second-best, and one that I'm very uneasy about.

Ah, yes. Three Mile Island. Feel the fear of science (maniacal laughter).

Except that according the the American Nuclear Society "The average radiation dose to people living within 10 miles of the plant was eight millirem, and no more than 100 millirem to any single individual. Eight millirem is about equal to a chest X-ray, and 100 millirem is about a third of the average background level of radiation received by U.S. residents in a year."

Then there was that terrible Windscale incident, which resulted in the destruction of some milk.

Chernobyl aside, there have been a handful of deaths at nuclear plants, often related to fires. I cannot believe that this is any greater than the number of deaths due to work-related accidents at Gas, Coal or Oil-fired power stations. Sooner or later some clumsy oaf will be hit by a revolving wind turbine, but I hope nobody will then start demanding we abandon wind energy.

In fact the comparison with wind and nuclear is instructive: one is now and one was in the 50s and 60s fashionable technologies of dubious economic worth. The difference between them is that nuclear can at least generate a decent amount of reliable energy.