But the Shakespeare chapter also turns heavy guns on (and has some scholarly fun with) people who think that Shakespeare's Bottom alludes to the weaver's posterior: "In Shakespeare's time, your arse could be your bum, butt, cheeks, croup, prat, rump, seat, stern, tail, toute, backside, buttocks, hurdies, fundament, or sitting-place; but before about 1794 it was not your bottom." Fowler says that it is obvious that the name refers to "the spool a weaver's thread is wound on". But can he possibly be right?

Milton was very careful about names, and also about not using names. In Paradise Lost, the fallen angels have lost their old heavenly names, so Satan does not know how to address them. Speaking to the one who is only later to be named Beelzebub, Fowler observes, "Satan must resort to circumlocution: ‘Fallen Cherub'. In hell as in the House of Commons, circumlocutory naming is the rule."

The arrival of realistic or naturalistic novels in the 18th century takes us fully into the world of fiction where, as Walter Scott put it, the hero should have "an uncontaminated name, bearing with its sound little of good or evil, excepting what the reader shall be pleased to affix to it". The modern author may have some feeling, perhaps quite unclear to himself, about why the name suits, but the reader, if he cares at all, will have to feel it for himself. He will get no help from the author.

Henry James cared intensely about the names of his characters, and took long lists of possible names from newspapers. He did not like over-meaningful names such as Trollope's doctors, Mr Rerechild and Mr Fillgraves, but like Dickens, he always wanted — as Dickens himself said — "the name that conveyed the outward show and inward mystery of a character". That is the kind of name most novelists would like to find for their books.

Fowler makes this point thoroughly, with many examples of what novelists have said about the matter, but from here onwards the book will perhaps be found disappointing by some readers. Fowler is above all a scholar,  and he never goes far beyond discovering identifiable sources for names.

One good observation he makes on a "modernist naming", as he puts it, is about Conrad's Nostromo. We come to admire Nostromo as a good, reliable man, he says, so when Nostromo commits a terrible crime "how are we to distance ourselves from it and disown him? After all, he is ‘our man'."

But Fowler is not much drawn to guesswork. His book stops short for the most part of trying to understand why any modern novelists have actually decided on the names they have given their characters, and why those names often sound just right.

There is scope, one might think, for a book that does that. But unless we come to understand the workings of the human brain much better, would it not turn out to be a book consisting almost entirely of speculation and fantasy?