The machine was a Nordtank NTK 600-180/43 rated at 600 kilowatts - a workhorse of the mid-1990s Danish wind industry rather than anything exotic. It was inaugurated on 23 December 1996 and operated by Syddjurs, and it carried the designation 570715000000021503 in the Danish wind turbine register, the sort of number that matters to nobody until something goes wrong. Its rotor measured 43 metres across, mounted on a free-standing steel-tube tower 44.5 metres above the flat farmland outside Hornslet. It stood first in line and northernmost in the Hyacintvej Wind Park, five near-identical turbines strung along an axis running south-southeast to north-northwest. From the air, a tidy row. From the road, unremarkable.

The day began with maintenance. A worn brake mechanism had malfunctioned, so a service team from Vestas was called in, and on the morning of 22 February its engineers checked the brake and repaired it. There was a second item on the record. At the previous routine inspection, the turbine's main gear had been making unusual noises, and an endoscopic inspection - a camera threaded down into the gearbox to see what was chewing itself up in there - had been planned. The procedure was expensive. It had not yet been carried out. With the brake fixed and tested several times, the crew restarted the turbine to bring it back into normal operation. The wind that day was very strong. Airbrakes at the blade tips were deployed to hold the rotor back as it wound up, and once the generator synchronized to the grid, a noise came out of the nacelle. The engineers moved to stop the machine by hand.

Then came a large crashing sound, possibly the main gear giving way, and the turbine began to oscillate hard. The rotor stopped dead - and immediately started turning again. It was not fast at first, but it was no longer controllable by anyone. The tower was evacuated. With the airbrakes failed and a strong wind pushing on 43 metres of unrestrained rotor, the blades accelerated turn by turn, quickly passing every speed the machine had been built to survive. Service personnel called the police, who pushed a security cordon out to 400 metres. And then everybody waited, because waiting was the only thing left to do. For two and a half hours a 600-kilowatt turbine spun itself toward destruction in full view of a small audience standing at a careful distance in a Jutland field.

At about 3:20 pm the blades began to disintegrate. One of them struck the tower roughly halfway up. Steel that had stood for eleven winters folded in the direction of the wind, and the top half of the tower sheared off at the bend and came down, carrying the nacelle and what was left of the rotor with it. The base stayed upright, a stump in the stubble. Debris flew 200 to 500 metres from the site - which is to say, most of the way to the line the police had drawn on nothing better than a good instinct for how far pieces of a disintegrating turbine might travel. No injuries were caused. Not to the Vestas engineers who had cleared the tower, not to the police holding the cordon, not to the farmer at the window.

Structural collapses are almost never filmed. They happen without warning, at night, in weather, in places where nobody happens to be pointing a lens - which is why Hornslet became one of only a handful of such failures captured on film, and why the footage travelled the way it did, first onto Danish television, then onto broadcasts abroad, then online, where it has been replayed ever since by engineering students, wind-power sceptics, wind-power engineers, and people who simply like watching very large machines come apart. The site itself is undramatic now. The turbine was out of service until June 2008 and was eventually replaced by a new machine of the same design, standing where the old one stood at the north end of the same row of five. The record is careful about cause: the crash heard from the nacelle was possibly the gear failing. The inspection that might have settled the question had been planned, priced, and postponed.