The Rule That Cannot Always Be Kept

Every painting you have looked at in a museum has been through a conservator's hands, usually several times, and some of what you are seeing is their decisions rather than the artist's. This is not a scandal. It is the only reason the object still exists.

Modern conservation runs on three commitments, taught together and holding together badly.

Minimal intervention. Do the least that stabilises the object and makes it legible, on the reasoning that every addition is a decision somebody later may disagree with, so fewer decisions is the safer position. This one is straightforwardly good advice.

Distinguishability. Your work should be tellable apart from the original, so that a later viewer or conservator can see which marks are whose. This one is contested in practice and we will come back to it.

Reversibility. Whatever you do should be undoable, so a future conservator with better methods and better information can take it off and start again.

The third is the one that fails, and the failure is not occasional.

Removing a discoloured varnish is not reversible; the varnish is gone. Consolidating flaking paint means introducing an adhesive into the layer structure, and it stays. Lining a canvas, attaching a weakened original to a new support, cannot be undone without a new intervention at least as invasive. Cleaning a stone surface removes material. A great deal of the work that actually keeps objects alive is permanent.

So the honest version of the principle is: prefer reversibility where you can have it, and where you cannot, record everything.

Which is why the condition report is not paperwork. If a decision cannot be undone, the only thing standing between a future scholar and total ignorance of what was there is the documentation made before anybody touched the surface. Photographs under several kinds of light, pigment analysis, cross-sections, a written account of the damage and the reasoning. That file is the object's only remaining past.

Now the contested commitment, because it is a real split and not a technicality.

You have a painting with a loss: an area where the paint is simply gone. Two defensible things to do.

Fill it with a neutral tone, or with fine hatching that reads as unified from across the room and announces itself at arm's length. The viewer is told the truth and the picture is interrupted. Italian practice has tended this way, and the hatching technique has a name and a literature.

Or reconstruct the missing passage convincingly, from the surrounding evidence and the artist's habits elsewhere, so the picture reads as whole. The viewer gets the image and cannot tell which parts are original.

Both are honest, in the sense that both are documented. They produce different objects. One is a damaged painting that tells you it is damaged; the other is a painting that looks complete and whose completeness is partly a modern hypothesis.

Which means the first question to ask about any restored work is not how good the restoration is. It is which school treated it, and where the file is.