What are we to make of it? Some critics, drawing on its unusual aspect ratio, have suggested a musical score: arpeggios, clefs, glissandos and quavers, etc. Schönberg's Suite, Op. 29, for the unusual combination of two clarinets, bass clarinet, violin, viola, cello and piano, comes to my mind. When I described Suite as a cacophony, a musical friend of mine was indignant. 'There are melodies in it - continually shifting melodies!' Not uncoincidentally, perhaps, Arnold Schönberg was an expressionist composer, whereas Jackson Pollock was an expressionist artist. (Cue for another essay, although I'm sure it's been written already). Even so, when I look at Summertime, Rachmaninov does not leap readily to mind. And while we're musically musing, George Gershwin composed Summertime, his aria for Porgy and Bess, in 1935 - perhaps there is some connection? No matter: the painting's title Summertime, I argue here, is surely misconceived.

Other critics have suggested a series of dancing figures, with associated rhythm and movement. The rhythmical property of Summertime I found intriguing because, glanced at perfunctorily, one sees only a formless, amorphous mess. Amorphousness and randomicity might describe other works by Pollock; Convergence, perhaps. Summertime, though, has something different about it; there is a sense of progression, say, through time or space, as one studies it from left to right. There is a central band of black and grey lines; but some cordoned-off areas, filled by patches of yellow, blue and burgundy, appear at regular intervals. Indeed, a definite sense of order emerges from disorder. And this would be my choice of title: Order from Disorder. Rather than music and dancing, therefore, I propose another interpretation of this work; one from the viewpoint of the scientist.

Order from disorder is a widespread natural phenomenon which the general public accept without realising just how weird it is. In the air around us the gas molecules normally whiz about in all directions: this is an incomprehensible anarchy, a disco in which all dancers jiggle away independently, strutting their own funky stuff. Yet, turn on the weather report, and we're shown something very strange indeed: enormous masses of air rotate in a depression or a hurricane or a tornado. Inconceivably large numbers of molecules now participate in a highly organised large-scale structure; a waltz in which all dancers rotate in synchrony around the ballroom. Now, when humans come together in large numbers, they must make agreements. We have committees, local authorities, national governments, treaties, laws, international protocols and many other things. But gas molecules - nonsentient, nonliving particles - require none of these. How on earth do these ordered, coherent structures arise in nature? Have mindless molecules somehow colluded? Well, the first requirement is an energy gradient.

If a fluid is sandwiched between two surfaces, one hotter than the other, thus creating a thermal gradient, then, in certain circumstances, something rather curious happens: as sketched below, there are cycling convection currents - these are known as Bérnard cells. Fluid heats up at the hot surface, rises, cools at the cold surface, then falls. The view from above, which I also provide, can show some remarkable structures, just one of which is pentagonal. Now obviously we can argue that colder pockets are more dense, and so fall, displacing hotter pockets upwards. But we also know that fluids are composed of molecules; and that molecules are individual actors. Yet, without any traffic lights or even roads, they somehow organise themselves into these elaborate, orderly structures: the molecular motions are no longer anarchical, but seemingly cooperative. As it happens, if the conditions are not quite right, there is a free-for-all in which upward-moving molecules battle against downward-moving molecules. If there are two crowds of people, one of them moving from left to right, the other from right to left, we know how stupid it would be to have a free-for-all: we separate the people into two streams. When we do this, everyone moves so much more quickly. And when Bérnard cells form, the transfer of energy (from hot to cold) is so much the faster. This is an important clue. We are often told that 'nature abhors a vacuum'. A far less-cited maxim is this: 'nature abhors an energy gradient'. Whenever there's an energy gradient, nature will form some ordered structure that degrades or mitigates or dissipates that gradient.