When Marina Warner began writing about myth, in the mid-1970s, 'myth critic' was a term of abuse. It meant that one was probably an unthinking admirer of Carl Jung, and was given to unsubstantiated generalisations about primordial narrative patterns and about archetypal images such as the 'great mother' and the 'wise old man'. This nonchalant approach was in the process of being replaced by aggressive new methods derived from structural linguistics. A leading influence was Roland Barthes, author of the recently translated Mythologies. The title, of course, was meant to be provocative; the point was that when Barthes used the word 'mythology' he really meant 'ideology'. Thus, instead of musing upon rites of passage and quests for the Grail, one should be rigorously exposing the way advertisements, magazine covers and sporting events deluded their consumers, persuading them that what was artificial was perfectly natural, that the way things were was the way they had always been.

Early on schooled in structuralism, which had taught her to see everything from table manners to religious rituals as the phonemes of a cultural grammar, Warner knew Barthes's Mythologies thoroughly. But, without going over to the other extreme, the Jungian game of 'spot the archetype', she avoided the stance of abrasive confidence with which 'semiological' analysis , interrogating the 'production' of meanings by 'sign-systems', emptied myths of all mystery. Indeed, we can trace her development as a successful overcoming of the anxiety of Barthes's influence. In her early work she invokes ...