Friedrich, Caspar David: Two Men by the Sea at Moonrise (1817)

Don't analyse it. The plea is well known. And since these articles are dedicated to analysing pictures, the case against it should be considered. There's the fear that analysis will simply succeed. It will explain how the picture works, and to explain a picture (like a joke) is to kill it dead. There's the feeling that analysis is the wrong kind of attention. A picture is to be experienced: thinking about how it works, you'll miss or even become positively unreceptive to its power. There's the claim that analysis is impossible. A successful work of art is not a box of tricks; it is a mystery. You'll never work it out, and any explanation will be a distortion.

Some of these objections are wilfully ignorant and fuzzy-minded. The correct reply to them is what the critic William Empson said about the poems of Dylan Thomas: 'It is quite true that they hit you before you know how, but that is no reason for not wanting to know how.' The last objection, though, has a point. A picture may not be a complete mystery, but there's always more going on than you can grasp. The full effect is too dense, and knock-on, and echoey to be formulated. You can never know entirely 'how'.

In his book Depiction, the art historian Michael Podro puts it wisely: 'Critical description can never properly or adequately correspond to the interest and force of a painting... because what we describe takes on its force for us only in the context of innumerable other recognitions in which it is embedded and which lie beyond the scope of describing.'

The problem is not just that your analysis will miss a lot of things out. The problem is that you'll take a lot of things for granted. What's so good about a picture? You identify its big effect, and analyse it, and say that it's due to factors A, B, and C, and thinks that's the answer. No doubt these factors are crucial. But factors D-Z (unmentioned) are important too. They may be the reasons why you even want to look at the picture in the first place. And if you made a picture that just had A, B and C, it might hold no interest at all.

The moral is not to abandon analysis, but to admit its limits. A total explanation is impossible. A partial explanation is always worth a go. Fine " except that there are pictures where a total explanation does look possible. They are pictures that seem to work by a formula. Their force depends on a fairly simple effect, which can be analysed.

Two Men by the Sea at Moonrise, by Caspar David Friedrich, is an original and piercing image. No one before Friedrich had made landscape paintings with so little to them. What have we? A small picture, 51cm by 66cm, in three horizontal layers. There's a layer of dark rocky shore, details hardly perceptible, and a layer of calm ocean, narrower, lighter, coming to a horizon two fifths up the picture, and layer of dim, cloud-flecked sky, as dark as the shore at the very top, with a layer of brightness just above the horizon, brightest at the picture's centre.

Two small figures, in cloaks and hats, stand on the edge of the shore, in the centre too, slightly apart, their backs turned, looking out to sea, their forms as dark as the shore, a bit darker than the sea, with their heads and shoulders rising above the horizon and sharply silhouetted against the brightening sky.

Description at even this level of specificity soon becomes intolerable. But the idea is not to describe the whole image, just the essential features, so that someone, working from the description alone, could produce a scene of similar mystery and solemnity. Which features are essential may have to be discovered experimentally, but one conjecture is that the image would lose little in black and white.

The two figures are crucial. Imagine them removed. You have just the moonlit ocean to contemplate, its simplicity and symmetry giving it a portentous, supernatural air. And that's one factor. Now put the figures back in. They stand in our way, interrupt our contemplation, and supplant it with their own consciousness. But they don't experience the view on our behalf. Their consciousness is quite closed to us. They commune with the seascape, receiving its feeling and but releasing no feelings. We know only their dark, stiff backs.

They're a stoppage. The emotional force of the view is all concentrated in them. The landscape experience is blocked and trapped in these rival viewers with their unfathomable minds. For that to happen, the two figures must be small, relative to the whole view. They must be central. They must have their backs turned. Their heads must show above the horizon. They must be silhouetted. Their stance must be fixed. It could also work with a solitary figure " but having a pair brings an unfathomable relationship into the equation too.

This is the claim, then, which can be tested. Here is the adequate formula for Friedrich's magic: one or two figures, in the above specifications, stood in front of almost any emptyish but impressive view. (And of course a third, to beg them not to analyse it.)