‘If only one could write! After that, perhaps one could think,’ Gaston Bachelard writes in The Flame of a Candle, published in 1961, a year before his death. He is picturing himself at his desk, waiting in vain for the ability to write to return, for the solitude of the blank page to end. There were too many times, he says, when thinking he was thinking (‘croyant penser’), he was dreaming. Is that so bad? Hadn’t he spent much of his life, if not dreaming, then defending the power of dreams? This is just what Steven Connor suggests in his new book: ‘Bachelard’s work was a dreamwork, both in the sense that it performs work on dreams and in the sense it gives of a self-amusing dream of what intellectual work could be.’ But then what about the writing and the thinking? Connor has some good commentary on those activities too.

Bachelard was born in 1884 in Bar-sur-Aube, a town in eastern France. He taught science in a school there, served in the First World War, and got a doctorate in letters at the Sorbonne. He hadn’t left his earlier field entirely, though, because he became a philosopher of science. After that he became … There is no simple continuation of this sentence, a fact which is a large part of the attraction of his career and of Connor’s book.

After publishing, among other works, The Inductive Value of Relativity (1929) and The New Spirit in Science (1934), Bachelard turned to The Dialectic of Duration (1936). He then began work on what Connor calls his ‘sequence of books on the elemental imagination’, the result of ‘his ten-year sabbatical among poets’. The elements involved are fire, water, air and earth, and the books are called The Psychoanalysis of Fire (1938), Water and Dreams (1942), Air and Dreams (1943), Earth and Reveries of Will (1948) and Earth and Reveries of Repose (1948). Later books include The Poetics of Space (1957) and The Poetics of Reverie (1960).

For Bachelard, the real in the age of relativity is not a ‘given’, or not only a given. It has to be ‘taken’ by someone or some culture. ‘So the giving,’ as Connor puts it, ‘depends on the taking, rather than vice versa.’ Bachelard plays with Descartes’s dictum, which becomes ‘it can be thought therefore it exists.’ Connor suggests that ‘it’ does a little more than this in Bachelard’s practice: ‘To think was always to break with what was given or immediate in experience. Thinking was what freed you from a life lived unreflectively from day to day, and so in a sense was a refusal of mere life.’

We can see how this inference, if pursued too hard, could get out of hand, and in a way, this is what Bachelard wants. The mode is not realism but liberation of the mind. In his book on relativity, he writes:

The newness of relativity is not static in nature. It is not things which surprise us, but the mind which constructs its own surprise and submits itself to the play of questions. Relativity is more than a definitive renewal in the way in which the physical phenomenon is thought, it is a method of progressive discovery.

This tone reflects what Connor calls Bachelard’s ‘magical confidence’, or more critically, ‘the spellbound gloaming of magical thinking’. Connor also finds ‘a hint of the Gothic’ in Bachelard’s prose, citing a passage from The Formation of the Scientific Mind (1938): ‘Even in a clear mind there are dark areas, caverns still haunted by shades, and traces of the old remain in our new ways of thinking. The 18th century still lives secretly within us and may – alas – return.’

This metaphor may have sounded more mild and innocent in 1938 than it does now. And in a later text, Rational Materialism (1953), the Gothic shades and magical thinking have turned into history:

Through chemistry and nuclear physics, man is granted unexpected means of power, positive means which surpass all the philosophical reveries of power … It seems that there too, in the psychological dimension, the will-to-power knows a chain reaction. The more one has, the more one wants … While the will-to-power was naive, while it was philosophical, while it was Nietzschean, it was only effective – for good or ill – at the level of the individual … But once man effectively appropriates the power of matter … he becomes a veritable magician, a positive demon.

In a 1967 article, Leo Bersani called Bachelard ‘the Freud of the French new critics’, meaning literary critics; the identification works well, especially for the books resulting from what Connor calls Bachelard’s sabbatical. Gilles Hieronimus says The Poetics of Space is ‘the most popular and most cited of Bachelard’s works’. Bachelard didn’t abandon science but he did embark on a journey into the imagination and never came back for long. He was looking, in Bersani’s words, for ‘an acausal psychology’. It’s true that phenomenology gradually took over from psychoanalysis, but the quest was the same: to live the unlived (‘vivre l’invécu’) by hanging out with poetry as if it were a composite friend, or ‘as if the reader were the ghost of the writer’. But then we have to get beyond the obvious.

Rilke no doubt loved locks that close doors. But who doesn’t love this kind of lock? Psychoanalytic literature is full of this theme … Poetry overflows psychoanalysis from all directions … The poet lives in a waking dream, and above all his dream remains in the world, faced with the objects of the world.

In The Psychoanalysis of Fire, Bachelard gets the idea of a ‘complex’ to do a lot of work. If you are the victim of a Prometheus complex, for example, you don’t have to steal fire; you just have to want to ‘know as much as our fathers, more than our fathers, as much as our teachers, more than our teachers … The Prometheus complex is the Oedipus complex of intellectual life.’ And more important, you have to begin to understand that ‘dreams are stronger than experience.’

Bachelard knows he is on dangerous interpretative ground. With something like the wry humour of J.L. Austin talking about happiness rather than truth, Bachelard says ‘words, which are made for singing and seducing, rarely meet up with thought.’ And commenting on the blue flower of Novalis, he shows how easy (and entertaining) it can be for interpreters to steal the show. Can this miscoloured plant have anything to do with mythological fire? ‘Just go to the bottom of the unconscious; find there, with the poet, the primitive dream and you will clearly see the truth: the little blue flower is red.’ You might also note, if you were in the right mood, that ‘red’ can be a homonym of ‘read’.

Bachelard has some eloquent pages on what he calls the ‘Charon complex’, broadly associating water with death and recalling Baudelaire’s poem, where Death is a ship’s captain and not a young man. The poet tells him it’s time to leave: ‘Ô Mort, vieux capitaine, il est temps! levons l’ancre!’ There is an Ophelia complex too. A quotation from Hamlet is taken as an invitation to imagine a ‘creature born to die in water’, ‘the woman who knows only how to weep for her sorrows’ and whose condition brings out in Laertes ‘everything that is woman in him’.

And, mermaid-like, awhile they bore her up; Which time she chanted snatches of old tunes; As one incapable of her own distress, Or like a creature native and indu’d Unto that element: but long it could not be Till that her garments, heavy with their drink, Pull’d the poor wretch from her melodious lay To muddy death.

A drowning mermaid, clothes that drink. Could Shakespeare have been reading Bachelard? We should note that Ophelia is ‘incapable of her own distress’ for two reasons: she is mad and she is going home. Who are we in this context? Bachelard likes to play with pronouns. He starts a sentence with ‘I’ and ends it with ‘one’. In other cases, he is ‘we’, and he says ‘you’ quite a lot. There are plenty of ‘theys’, ‘shes’ and ‘hes’ too. In its way, this confusion is a sort of theory. The writer is several persons and so is the reader. When Bachelard says that our ‘intuitions of fire are epistemological obstacles all the more difficult to reverse because they are psychologically clear,’ he is inviting us to go beyond what we think we know. That is, how to counter boring intuitions with interesting ones. But who is to say which is which? I suppose the answer depends on how we feel once we have accepted the invitation.

Northrop Frye thought ‘myths’ would have been a better word for what Bachelard called complexes. Freud’s friend Ludwig Binswanger said of Bachelard’s ‘psychology’ that ‘we would call it an anthropology’ and mildly complained that what is missing from his work is ‘an anthropological, and even more, an ontological basis’. I wonder if a looser definition might make real room for Bachelard. We could take anthropology in the sense evoked by Claude Lévi-Strauss or James Frazer: a study that hopes to understand a culture through its immaterial life, the stories it tells and the pictures it likes. Except that there is no single culture here, no indigenous world or classical country. There is whatever the imagination can learn about or attribute to the imagination. Or to alter the credit slightly, what a structuralist might find if she looked behind the scenery. Connor has a good line about ‘turning glamour back into grammar’.

We can’t fail to see the potential vagueness of the project, the vulnerability created by constant use of random examples (Connor says Bachelard’s formulations are ‘childishly easy to puncture’). But Bachelard has two devices in particular that help us to see what else he is doing. One is to be modern by sounding as old-fashioned as possible. We get this effect every time he uses the words ‘rêve’ and ‘rêverie’, and he has a marvellous excuse for being out of date: ‘Dreams don’t modernise themselves as quickly as our actions do.’ The practice also responds to Connor’s sense of Bachelard’s life being ‘never fully contemporary with itself’. But then he takes these old words as proof that we need to think again – with the same words. In Bachelard’s usage, ‘rêve’ and ‘rêverie’ carry programmes that we don’t meet in their ordinary translation, ‘dream’ and ‘daydream’. ‘Taken on its own,’ Bachelard writes, ‘reverie is a psychic instance we confuse too often with a dream.’ In relation to reverie, the OED gives us ‘a moment or period of being lost, esp. pleasantly, in one’s thoughts’, but Bachelard has in mind something more dynamic, almost a total contradiction of the term, like a drowning mermaid.

The other device is a distinction between metaphor and image, where metaphor gets to play the straight guy:

If metaphors are often only displacements of thought, driven by the will to say better, to say differently, the real image, when it is living its first life in the imagination, leaves the real world for the imagined, imaginary world.

This thought seems only to say that the imaginary is imagined, but Bachelard’s point is that we should start rather than end there. He says an image can be ‘a giver of being’, ‘une donatrice d’être’. This is to suggest that the human mind and its images, in dreams as in the theory of relativity, are creators of each other.

There is a form of liberation in this idea. But there are many other, less attractive possibilities in it too, and this is where we need to return to Connor’s sense of gloaming. The magic in Bachelard’s work is often wonderful, but it may be even more remarkable for what it doesn’t include, doesn’t invite to the eclectic seminar. He began his sequence on the elements in 1938 and ended it in 1948 – or perhaps never ended it. Connor suggests the work done at this time represents ‘a gradual movement away from historicity’ and a new focus on ‘cosmic archetypes’. There is also a movement away from history itself, and we may be grateful for that.

The working images Bachelard praises are in themselves wonderful performances: ‘the poetic act has no past’; ‘the soul announces its presence in a poetic image’; ‘not-knowing is not ignorance but a difficult act of going beyond knowledge’; ‘images do not go well with quiet ideas, or above all with definitive ones.’ But taken together they begin to seem like a persistent rejection of a rational, intended world, or like a desperate plea on behalf of all those who can’t bear to think of the modern ‘power of matter’. We are not to talk of the Nietzschean magicians who became demons.