There must be a silver lining

For the best views, John Ruskin advised looking up: "Passing then to skies, note that there is this great peculiarity about sky subject, as distinguished from earth subject - that the clouds, not being much liable to man's interference, are always beautifully arranged."

Well, yes. Vapour trails are seldom an improvement. And I think we can guess Ruskin's opinion on the sort of synchronised multicoloured sky-writing you find at air shows. But however little the skies may be liable to actual human interference, they are - just for that reason, perhaps - very liable to human fantasy.

People have always seen things and creatures in the shapes of clouds: whales, for instance. Or clouds can seem semi-animate themselves, wandering lonely and massed like an invasion force. Clouds are a second, upper-tier landscape, the solid furniture of heaven. Clouds are the sky's expressive facial features, which make the day smile, frown and sulk - the metaphor is so basic that a person who didn't feel the sky as moody would seem to have something wrong with them.

Clouds are the visible beyond, too. Not only, for most of history, beyond reach, but beyond measure. They are outside human scale. We know that clouds are very big, and we notice especially big ones. But (even with our flying experience) we hardly ever refer this size to ourselves, or anything else on earth. We don't imagine how large a cloud would be if we met it on the ground. The clouds are in a world of their own, out of everything but sight, a pure spectacle.

So in that way clouds are already like a painted image. And they are handy things for painters. If a picture can have clouds in, it has a supremely amenable resource, something that can be freely moved up and down and from side to side, increased, cancelled, made darker and lighter, pretty much at will. That is the paradox of the painted cloud. What in reality is the most human-free bit of nature is the element most at the artist's disposal. That's the thing about clouds generally: very much ours and not ours.

Constable's Clouds has just opened at the Walker Art Gallery in Liverpool. It's obviously a good idea. Constable is the great British weatherman. Constable's country would be nothing without his sky. He knew that well: "It will be difficult to name a class of landscape in which the sky is not the key note, the standard of scale and the chief organ of sentiment."

The title may suggest, specifically, the series of small studies just of clouds made in 1821 and 1822. A good few of those are here. But clouds are everywhere in Constable's art, often almost half the view, and the show includes many telling examples of what he called his "skying", some of them famous works - Golding Constable's Kitchen Garden, The Leaping Horse, The Vale of Dedham, Salisbury Cathedral from the Meadowes. It becomes, in effect, a small Constable survey exhibition.

Now I have to admit that, as to the art of John Constable, I really don't like it. You readers who write in crossly when you read that kind of remark, I do not say that deliberately to annoy you. I do not hate the art of painting, nor the English countryside. I am not motivated by self-loathing for my own nationality. No, but the painting of John Constable, especially at its most Constabular, with its flecked, spattered and flickering surfaces, just seems to me turbid and turgid, clogged and agitated with stirred up energy, a fussy-busy-blustery-messy affair suggesting the worst excesses of tachisme. The paint effects have no relation to anything you might feel about what's depicted - rather an extraneous conjuring up and application of feeling to the scene. Of course, that isn't true of all his pictures. The small sketches often seem loving and graceful. I kind of thought that was true of the little oil-on-paper cloud studies, too.

What were they for, those hour-long sketches, made on Hampstead Heath in summer and autumn? It's common to call them scientific. Serious cloud study was certainly in the air. Luke Howard's "Essay on the Modification of Clouds" of 1803 countered the traditional view that clouds were sheer mutable formlessness. Howard introduced the four basic cloud types: stratus, cumulus, cirrus and nimbus, plus their compounds. Earlier cloud painting - with some exceptions - had general effects of cloudiness without registering particular species. As an artist, Constable was not alone in taking a more methodical interest in the subject.

The Cloud Studies have no literal scientific purpose. But they do not have overt artistic pretensions either (of the sort Constable would have recognised, at least). They are not much composed. They are works of self-informing observation. They show not vistas, but quite tight sections of the sky. The viewpoint often seems closer than ground level. The angle of vision is sometimes straight upward. Consequently, for the modern eye, they become extremely artistic, being (a) quasi-abstract, being fields of almost pure painting, and (b) quasi-surreal, with that pointed decontextualisation of things that you find in scientific illustration, which surrealism borrowed so effectively.

Some of them are certainly beautiful. Constable often allows the orange-brown of the paper to show through a lot, adding a warm earthy tone to the whites, greys and blues, which makes for richer colouring, while hardly being true to nature (clouds do sometimes go brown, I know, but not in that way). And in these studies, as in the full vista landscapes, I don't think the cloud painting is very interested in clouds - or only on Constable's terms. I think his cloud painting is actually anti-cloud.

For instance, clouds plainly feel like many kinds of stuff - from the most clear-cut and friable solid to the most evanescent gas, through a range of fibres, fluids and viscosities. But what clouds never feel like is paint. Constable's clouds usually feel like one kind of thing, and that is paint - specifically, like a wall of paint. He gives no sense of cloud substance nor of cloud distances. There is no drama of entities, and no feeling of their great aboveness. The clouds in Constable have no lives of their own, nor any infinities to lose yourself in.

What you get is the sky as expressive backdrop, a screen of moods. Constable has one big idea about clouds, and it is the most human-centred one possible. It assumes the sky is always looking and making faces at us. It is (as he said) the scene's "organ of sentiment", the thing you do the feelings with. It's an organ with all its stops out. His skies are packed and never less than heavily, blatantly moody. And, of course, the face is always really Constable's own, the skies blustering like the frantic brushwork.

But I must not say the work in Constable's Clouds is never true. When the scale is small, the sky's action is done in simple, decisive strokes that wonderfully render the operations of wind and water, render them as powers, not resources of pictorial expression. See, for instance, the sudden downpour of A Rain Storm over the Sea. He surrenders to it. What's good about this sky is that it's not Constable's.

Constable's Clouds: Walker Art Gallery, Liverpool, to 16 July