The Ethics of Repair

A conservator's most frequent decision is not how to repair something. It is whether to.

This is counter-intuitive to anyone outside the field, where the assumption runs the other way: an object is damaged, a specialist repairs it, the damage is undone. The profession has spent a century establishing that this description is wrong at every step, and the reasoning repays attention, because it turns on a question that extends well beyond museums — what exactly are we preserving when we preserve a thing?

The problem, stated concretely

Consider an excavation at which painted figures are recovered from dry soil after two thousand years. Beneath the surface they are vividly coloured. Within minutes of exposure to air, the pigment layer begins to curl away from the surface it sits on, because the lacquer binding it to the ceramic has been stabilised by constant humidity and dries catastrophically once that humidity is removed.

The archaeologists face an immediate choice. Excavate, and the colour is lost during the act of discovery. Leave the figures buried, and the information they carry remains inaccessible, possibly for generations, and possibly for ever if the site is disturbed by other means.

The eventual solution was neither: a consolidant applied within moments of exposure, developed over years of collaboration, which holds the lacquer in place long enough for controlled removal. That worked. But notice the shape of the problem, because it recurs constantly in less dramatic forms. Every act of examination is an intervention. There is no neutral position from which we simply look.

Four principles, and why each is a compromise

The framework the discipline has settled on rests on four commitments, and it is more interesting when one sees what each concedes.

Minimum intervention. Do as little as possible consistent with stability. Behind this lies an admission: we do not trust our own judgement about what a future generation will want, because previous generations' confident interventions are precisely what we now spend our time undoing. A structure cleaned in the 1930s with copper tools and acid — because the specialists of the day believed the original surface had been white and the accumulated patina was mere dirt — lost material that cannot be replaced. They were not careless. They were certain.

Reversibility. Any material we add should be removable without harming what was there. In its strong form this is unattainable — nothing is perfectly reversible, and solvents that remove a modern adhesive usually attack something else — so the profession has largely retreated to a weaker and more honest formulation: retreatability. Whatever we do must not prevent the next person from doing something better.

Distinguishability. A repair should be detectable on close examination, even if invisible at a normal viewing distance. This is the principle the public most often objects to, and the objection is reasonable: why leave a visible scar? The answer is that a seamless repair converts an object into a document that lies. A scholar examining it a century hence, without our records, would read our guesses as evidence.

Documentation. Everything done is recorded — materials, methods, reasoning, and what was rejected. This is the quiet one, and arguably the most important, because it is the only principle that acknowledges we will turn out to have been wrong.

Where the principles collide

The framework is coherent until two goods conflict, which is most of the time.

A seventeenth-century warship recovered from a harbour is the standard case. Its timbers, waterlogged for three centuries, would have collapsed on drying, so they were impregnated over many years with a polymer that replaced the water and supported the cell structure. The treatment succeeded and the ship stands. Decades later it emerged that sulphur compounds absorbed from the polluted harbour were reacting with oxygen and iron from corroded fastenings to form acid within the wood — a process the original treatment had not anticipated and could not have, because the chemistry was not understood at the time.

Nobody involved made an error by the standards available to them. The object was saved by a method that created a slower problem, and the current work consists of managing a condition produced by the rescue.

The second familiar collision is access against survival. A decorated cave closed to visitors within fifteen years of its discovery, because the breath, warmth and carried spores of a few hundred people a day were doing in decades what twenty thousand years of stable darkness had not. A facsimile was built nearby, and visitors now see a copy. Some regard this as a defeat; the alternative was to permit the destruction of the original in order to look at it, which is not obviously preferable.

A different tradition

It is worth setting against all of this a philosophy that begins from an entirely different premise.

In the Japanese practice of mending broken ceramics with lacquer mixed with powdered gold, the repair is not concealed but emphasised — traced in bright metal along every fracture. The object is not returned to a prior state. It is understood as having a history, of which the breakage is a genuine part, and the mend records rather than erases it. The result is frequently valued above the unbroken original.

There is a parallel in architecture. Certain timber traditions maintain buildings by periodic replacement of members according to established practice, so that a structure of great antiquity may contain little original material while being, by the standards of that tradition, entirely authentic. What persists is the form, the technique and the continuity of the practice rather than the substance.

This forced an explicit reckoning in international conservation thinking during the 1990s, and the resulting position is now widely accepted: authenticity cannot be assessed by a single universal standard. For some objects it resides in the original material. For others it resides in form, in use, or in the unbroken transmission of a craft. A framework that recognises only the first is not neutral. It encodes one tradition's assumptions and treats them as physics.

What we are actually preserving

Which returns us to the initial question.

We are not preserving objects, exactly. An object is a physical arrangement that will, without exception, degrade. Given long enough, everything in every collection will be gone, and the profession's honest description of its own work is not preservation but the deliberate slowing of loss.

What we are preserving is the capacity of someone not yet born to ask a question we have not thought of. That is why documentation matters more than skill, why distinguishability matters more than appearance, and why minimum intervention is not timidity but humility with a method attached.

The conservator's discipline, in the end, is to act decisively while assuming they are probably wrong — which is an uncomfortable way to work, and one that a great many other professions would benefit from adopting.