Emerging pentimenti

Why do a painter's abandoned first attempts creep back into view centuries after being covered over?

A painter moves an arm, repositions a hat, closes a window that was open. He paints over the old version and, standing in front of the canvas, cannot see it. Neither can anyone else for a long time. Then, two or three centuries on, a ghostly second arm is there in ordinary gallery light — three fingers of it, faint but unmistakable.

Nothing was uncovered. No one cleaned it back. So what changed? The obvious suspicion is that the covering paint wore away, but that would mean the picture is now missing paint it once had — and pentimenti appear on canvases in excellent condition that have never been aggressively cleaned. Something must have changed within the layer that was doing the hiding.

Ask first what hiding actually requires. A paint layer conceals what is beneath it by scattering light — sending it back out before it can reach the lower layer and return. Scattering happens at boundaries between materials of different optical density, and in paint those boundaries are the surfaces of the pigment particles suspended in the binder.

So how much a layer hides depends not on the pigment alone but on the contrast between the pigment's refractive index and the binder's. Push that to its limit as a thought experiment: if the two were identical, the particles would be optically invisible in the medium, and the layer would be as clear as glass no matter how much pigment it held. Glass fibre in a resin of matched index is exactly that trick.

Now the crucial fact, and the one everything turns on: the refractive index of drying oil is not fixed. Linseed oil polymerises and oxidises for as long as the picture exists, and as it does its index rises — the usual figures quoted in conservation science are around 1.48 for fresh oil climbing toward roughly 1.57 with long ageing. I am recalling those values rather than measuring them, and the endpoint in particular varies with the oil, the driers, and the pigment it is bound to; treat the direction as solid and the exact numbers as approximate.

Now derive the consequence rather than asserting it. To a first approximation scattering grows with the square of the index difference between pigment and binder. Take an earth pigment or a red lake, around 1.6. Fresh, the contrast is 1.60 minus 1.48, and squared that is 0.0144. Aged, it is 1.60 minus 1.57, squared 0.0009 — a fall of roughly sixteenfold. The layer has not thinned by a single micrometre, and it has lost almost all its hiding power.

Run the same arithmetic for lead white, whose index is near 2.0. Fresh: 0.52 squared, 0.27. Aged: 0.43 squared, 0.185. Barely a third lost, from a starting value about twenty times higher. Lead white stays opaque essentially forever.

That single contrast does real work, because it turns the model into a prediction rather than a story. If emergence is driven by index convergence, pentimenti should appear preferentially where the covering passage is low in index — dark earths, blacks, lakes, thin glazes — and should be rare through thick lead-white passages, however old. The observation that would refute it: ghost images showing through heavy lead-white paint as readily as through umber. As far as the technical literature goes, that is not what is seen; the emerging pentimento is characteristically a dark or thinly-covered one.

Two things compound it. Painters covering a change often did so economically, with one or two thin layers rather than solid opaque repainting — so the hiding was marginal to begin with. And a first version laid in with a dense, high-index pigment stays perfectly opaque itself while its cover fades, so the contrast between them actually widens with time.

Why so slow, then — centuries, not decades? Because the oxidative ageing that raises the index is slow, and because the effect only becomes visible once contrast falls below what the eye can resolve against the surrounding paint. Both the cause and the threshold are gradual, which is why a pentimento seems to arrive rather than to switch on.

Two distinctions keep this honest. First, the pentimento that emerges is not the same thing as the pentimento a museum detects. Infrared reflectography and X-radiography find buried changes in pictures where nothing is visible to the eye at all, by using wavelengths the upper paint happens to transmit or by imaging heavy elements through it. Those revelations owe nothing to ageing. The naked-eye emergence discussed here is a slower, accidental, and irreversible version of the same access.

Second, increasing transparency is the main mechanism but not the only one. Thinning of upper layers by past cleaning, and drying cracks that open windows onto what lies below, both contribute — and in a given picture it can be hard to apportion them. The reason to put transparency first is the pattern it predicts and they do not: emergence concentrated under low-index cover, on canvases with no history of harsh treatment.

It is also worth resisting a tempting inversion. The paint is not becoming transparent because it is decaying. It is becoming transparent because it is continuing to cure — the same reaction that made it a solid film in the first place, still running, long after anyone would call the picture wet.

• How infrared and X-ray imaging reach buried layers deliberately, and why each sees a different set of changes.
• Why some pigments turn transparent, darken, or blanch through quite separate chemical routes, and how conservators tell those apart from ordinary ageing.

Every term the collection defines is gathered in the glossary.