Books, I fancy, may be conveniently divided into three classes: —
(1) Books to read, such as Cicero’s Letters, Suetonius, Vasari’s Lives of the Painters, the Autobiography of Benvenuto Cellini, Sir John Mandeville, Marco Polo, St. Simon’s Memoirs, Mommsen, and (till we get a better one) Grote’s History of Greece.
(2) Books to re-read, such as Plato and Keats: in the sphere of poetry, the masters not the minstrels; in the sphere of philosophy, the seers not the savants.
(3) Books not to read at all, such as Thomson’s Seasons, Rogers’s Italy, Paley’s Evidences, all the Fathers except St. Augustine, all John Stuart Mill except the essay on Liberty, all Voltaire’s plays without any exception, Butler’s Analogy, Grant’s Aristotle, Hume’s England, Lewes’s History of Philosophy, all argumentative books and all books that try to prove anything.
The third class is by far the most important. To tell people what to read is, as a rule, either useless or harmful; for, the appreciation of literature is a question of temperament not of teaching; to Parnassus there is no primer and nothing that one can learn is ever worth learning. But to tell people what not to read is a very different matter, and I venture to recommend it as a mission to the University Extension Scheme.
Oscar Wilde, “To Read or Not to Read” (1886)