"Imagine Machiavelli's The Prince," writes Joseph Epstein in The Wall Street Journal, "written not by an Italian theorist, but by a true prince. Imagine, further, that he also lets you in on his desires, his fears, his aesthetic, his sensuality, his feelings about death - in a manner at once haute and intimate, and in a prose any emperor would be pleased to possess..." The work he refers to, however, was written not by "a true prince," but by a Franco-Belgian intellectual adopting the persona of one.

Marguerite Yourcenar was, perhaps, an accidental novelist. "In our time," she wrote, "the novel devours all other forms; one is almost forced to use it as the medium of expression. This study of the destiny of a man called Hadrian would have been cast in the form of a tragedy in the Seventeenth Century, or of an essay, perhaps, in the period of the Renaissance." Memoirs of Hadrian, however, is far from being a conventional 20th Century novel. There is, for example, no dialogue. She first conceived of the work as a series of dialogues, but found that it did not work, that "Hadrian's voice was drowned out by all the others...," that she was not succeeding in her attempt "to reconstruct that world as seen and heard by one man." She kept on writing and revising for 27 years until she did succeed.

Memoirs of Hadrian is, perhaps, the supreme triumph of historical viewpoint and historical voice in all of 20th Century fiction. Had she written it in Greek (as she was capable of doing) and contrived to produce a manuscript, the consensus of scholarly opinion today might well be in favour of the authenticity of Hadrian's autobiography. One could by no means make such a claim for any comparable work, not even Robert Graves's Claudius novels: superbly researched though they are, they are stylistically far more "of their time" than Yourcenar's masterpiece, and include insights that could only ever have been the product of a 20th Century consciousness.

Yourcenar has left us not only the work itself, but also her "workings," an intimation of the process by which she brought it into being. Her two essays, "Reflections on the Composition of Memoirs of Hadrian" (reprinted with the book itself) and "Tone and Language in the Historical Novel" (reproduced in a volume of her essays entitled That Mighty Sculptor, Time, translated by Walter Kaiser), offer practical advice to writers of historical fiction, and indeed of all fiction. She reminds us that we simply do not know what the spoken language of the ancient world would have sounded like, and she therefore does not attempt to reproduce it. Instead she writes the novel as a letter, without dialogue, taking as her model those formal addresses that have survived from the Roman world. From this is born her rendition of the Oratio Togata which, even in the labour of love that is Grace Frick's English translation, can never be as close to the original Latin in English as it is in Yourcenar's French.

"Someone will say that Corneille in Cinna, Racine in Britannicus and Shakespeare in Julius Caesar managed it all rather well. But that is, of course, because they possessed genius. It is also - and perhaps even more to the point - because they were not concerned with tonal authenticity."

We may say what Yourcenar herself dared not say - that she, too, possessed genius - but she clearly was concerned with tonal authenticity, and this concern, above all, shapes the work. "No other document," writes Joan Acocella in The New Yorker, "takes us so deeply into the pre-Christian mind."