The best of them.

Five weeks, thirty to forty pages an evening, and I was sorry when it stopped. There's no other book I've read that handles this much technical material without either flattening it or hiding behind it.

What it gets right, and almost nothing else in the genre does: the engineering is treated as the story rather than as an obstacle between the interesting parts. The barrier problem at Oak Ridge, the metallurgy, the sheer industrial scale of separating a substance from itself when the two versions differ by less than one percent in mass — those chapters are given the room they need. Somebody in that project had to specify a valve, and this book knows it.

The opening is the finest thing in it. A man stands at a traffic light in Bloomsbury in 1933, the light changes, and he steps off the kerb having just understood how a chain reaction would work. I have gone back to that page more than once.

And the moral argument is not bolted on at the end, which is the usual arrangement. It runs alongside from the beginning and gets heavier as the machine gets closer to working.

Ten. I have given three copies away. Two came back.