My first reaction was that the mysterious glowing balls were some kind of unusual meteorological phenomena, perhaps akin to ball lightning: an atmospheric electrical discharge triggered by the odd characteristics of the nanotubes, one not previously observed since the bridge was the first large-scale deployment of carbon-fibre composites. This hypothesis was strengthened by the observation that the lightning balls only appeared on clear, dry nights; when there was rain or any kind of dampness in the air, or any low cloud, there were no reports.

The reports of the appearance and disappearance of the lights was convincingly consistent. Eyewitnesses agreed that the balls would appear suddenly, first quite dim and faintly shimmering, then in a moment brightening to a dazzling shade of actinide blue, or a yellow as intense as the sun. After a second or two, the balls would accelerate, usually straight up and always away from any observer, and then disappearing from sight in a few seconds.

The pattern was repeated in all except for one case, which brightened as usual, then flickered several times before falling towards the sea. Before it hit the waves, the construction worker swore that it exploded in a haze of fine particles that dispersed in the wind almost immediately.

I fully admit that the causes of the lightning balls on the bridge rapidly became an all-consuming mystery for me. In the weeks after I decided to go and look for myself, I spent an unconscionable amount of time investigating the phenomena, much to the detriment of both my professional and personal life.

I read up on ball lightning, which confirmed that this phenomenon was poorly understood, with no scientific agreement on the mechanism that created the fiery spheres. There was one difference, though: the balls that appeared on the bridge always seemed to be about the same size - a sphere about the size of a soccer ball or basketball - rather than the hugely variable sizes reported in the literature.

There were some patterns to the appearances of the spheres, whatever they were. They always arrived close to the cables, always after dark and only when the weather was fine. An appearance never occurred more than once a night, occasionally on successive evenings, also sometimes there would be a gap of several days before any reoccurrence.

There was one key pattern which, frankly, I did not at first believe. After a few weeks, the evidence became undeniable: on nights when the spheres did not appear, accidents happened. On some nights, people got hurt and a couple of them died: one falling from a section of the roadbed which was in the process of being installed, and the other crushed by a falling crate.

By this stage, the rate of progress on the construction had ground to a complete standstill. There was an extensive review by the Heath and Safety Executive, an institution often irreverently known as Elfin Safety. Any such review, even for the most trivial of site accidents, was bound to be disruptive and, at that time, it was almost impossible to make any progress because of the endless investigations and reports.

The HSE made recommendations, of course, as they were bound to do. My company started employing more people, building larger teams, so that nobody had to do anything without several others around him. At first, it seemed to work. The rate of accidents dropped, and progress picked up again, although it never as good as it was at the beginning. It seemed that the superstitions remained, and that everybody was always looking over their shoulders both literally and metaphorically.

The incidents were never far away. I heard several stories of how a serious accident had only just been averted by the attention of a co-worker, whose quick-thinking heroics managed to prevent an injury caused by the sudden distraction of the victim. There was something about the bridge itself which seemed to be making the people working on it prone to uncharacteristic lapses in attention.

Part 4 Part 6