It sat on my shelf for forty-one years. My fault, not the book's.

Sylvia bought me this in a shop in York in 1978 and it went onto the shelf and stayed there through my entire working life. I read it properly two winters ago, forty pages at a time, over about eleven weeks. She would have been pleased I got to it and extremely irritated at how long that took.

What I didn't expect is that it's arranged the way a decent engineering history is arranged. Each chapter opens with a problem the previous chapter failed to solve. How does one put depth onto a flat surface? How do you make a figure stand rather than float? What is to be done about the fact that a painted room has one light in it and a real room has several? A solution arrives, the solution creates a new difficulty, the difficulty gets a name and a generation of men to work on it. I learned more from that sequence than from the six or seven gallery visits I have made in my life.

He opens by saying there is no such thing as Art, only artists. I took that for a trick when I read it and I now think it is the method: he declines to grade, and asks instead what the man was attempting and whether he brought it off. That is precisely how one is taught to look at a failed connection detail. Not is it ugly. What was it for, and did it do it.

Two complaints and they are real ones. The women are effectively absent, and later editions have dealt with this by adding a name here and there rather than by rethinking the shape, and it shows; an addition is not a redesign. The second is that the final chapters, the ones covering the last century, are noticeably weaker. He is out of sympathy with the material and it comes through as courtesy, which is worse to read than an argument would have been. He is also, and I say this without malice, a man who believes the Italians settled most of it and that everyone since has been arguing about the finish; you can feel the centre of gravity of the whole book sitting somewhere around 1500 and never really shifting.

My edition is old enough that the early plates are black and white, which for the chapters about colour is close to useless. I ended up with the book in one hand and a tablet in the other, looking the same painting up in colour, which isn't how anyone intended it to be read and worked perfectly well.

Eight. If you have never read anything about painting, start here, and leave the last four chapters until you have.