Thirteen sittings, one of them long.

My grandson put this into my hands in March and asked what I made of it, which is how a good deal of my reading gets chosen these days. Twenty-one days, thirteen sittings, and one of those ran to three and a half hours because I wanted it finished before I lost track of who was who. A seven, and most of the seven is the first eighty pages and the physics.

The opening section is the best thing in the book by a distance. A physicist is beaten to death in front of a crowd, his daughter watches, and the writing there is plain and unhurried and declines to tell you how to feel about any of it. I would have read four hundred pages of that quite happily. When the book leaves the 1960s it becomes a different sort of object altogether and a lesser one, and I spent much of the remainder waiting to be taken back up the mountain to the observatory.

On the science I can be of more use than usual. The central problem is real and genuinely unsolvable in the general case, which is a fine thing to build a novel on: three masses, no closed solution, and a climate that lurches between stable eras and chaotic ones. That much is honest, and the chaotic era passages are the only place where the book's imagination and its arithmetic agree with each other. I checked two of the orbital claims against a textbook I still own, and they held up rather better than they needed to.

The proton business is where I got off. Unfolding a particle into eleven dimensions, etching circuits onto the surface and folding it back down into something that can sit inside an accelerator and lie to it — I accept that a novel is allowed that licence. My objection is structural rather than scientific. From that point on, one member carries the entire plot; there is no redundancy anywhere in the arrangement, and remove it and nothing at all stands up. An engineer is trained to distrust exactly that.

The game sections defeated me. By my rough count a quarter of the length is spent inside a virtual world, where costumed historical figures explain a problem the reader worked out several chapters earlier. Two of those are a lecture with a listener sitting in the room for the sake of politeness. I skimmed the back half of that material and lost nothing, and I can say so with confidence because I went back afterwards to check what I had missed. There was nothing there.

A word for the woman at the centre of it, since nobody I have talked to about this book wants to discuss her. Ye Wenjie is the only person here who is built rather than assembled out of a description, and the decision she takes at the radio telescope is the one genuinely enormous moment in the novel. It lands about a third of the way through and everything after it is consequence. Had the book stayed with her I would be arguing for a nine. Instead it leaves her for a man I could not describe to you now, a fortnight later, beyond his job title and his temper.

I have no opinion on the prose. It reached me through a translator and I've no way of knowing what belongs to whom. The translator's footnotes, though, are the most useful I have met in a novel, thirty-odd of them, and they carry the history that the book itself assumes you already have.

I will admit that for two chapters I believed the sophon was a person. Nobody was there to correct me, and by the time I understood I was too far in to want to go back and re-lay the foundations.

If you want the feeling this book is reaching for, Rhodes on the Manhattan Project gets there with real men and no help from physics that hasn't been done yet. That is an unfair comparison and I am making it anyway.

The second volume is on the shelf. My grandson tells me it is better and the third better still, and I have heard that particular promise before. I will get to it in the winter.