Canvas tension and humidity

Why does a painting's canvas sag in damp weather and pull drum-tight in dry?

Anyone who has kept a stretched painting in a house without climate control knows the behaviour. Through a wet week the picture plane goes soft and dimples under a fingertip; by the first dry cold snap it is taut as a drumhead again. Nobody tapped a key or moved a tack.

That ought to bother us a little. The stretcher is wood and the canvas is cloth, and neither is doing anything. So what is it that oscillates?

Begin with the constraint, because it is an unusual one. The moment a canvas is tacked to its stretcher its outer dimensions are fixed by the wood. It cannot get meaningfully longer or shorter; it can only be more or less stretched at that one size. Engineers call this a constant-strain condition, and it has a consequence worth sitting with: whatever the cloth wants to do dimensionally, it is forbidden from doing, so it expresses the want as force instead. A canvas that would like to contract cannot — it pulls harder. One that would like to relax cannot go anywhere — it simply pulls less.

So the question becomes: what makes the cloth's wants change? Here the useful idea is that linen and cotton are hygroscopic. Cellulose is studded with hydroxyl groups that hydrogen-bond water out of the air, and the fibre holds a moisture content set by the surrounding relative humidity — roughly four or five percent by weight in a dry room, perhaps eight or nine at ordinary indoor humidity, more than twice that in a damp one. Notice what kind of quantity that is. It is not a one-way accumulation. Leave the canvas in a room long enough and its moisture content stops changing: uptake and loss come into balance at a value the room dictates. The relationship between humidity and that settled value has a name, the sorption isotherm, and it is the whole engine of this puzzle.

Which suggests a reframing. The canvas is not sagging because of damp and tightening because of dryness, as though humidity were pushing it around. The canvas is continuously travelling toward a moisture equilibrium with whatever room it is in, and its tension is simply what we can see of where that equilibrium currently sits. Change the room and you have moved the target; the cloth then walks toward the new one, and the tension follows.

Why does more water mean less pull? Because water in cellulose acts as a plasticiser. It wedges between the polymer chains, breaks some of the hydrogen bonds that were holding them rigidly to one another, and lets them slide. The fibre — and, more importantly, the animal-glue size brushed on to seal it — becomes markedly less stiff. Now recall the constraint: the strain is held fixed by the stretcher, so if stiffness falls, the stress falls with it, and stress in a membrane is exactly what we perceive as tautness. Dry the room out and the chains re-bond, stiffness returns, the material contracts slightly, and the same fixed geometry now demands a much higher force.

Does that rule hold everywhere on the scale? Not quite, and this is where honesty earns its keep. Mechanical measurements on glue-sized linen — Marion Mecklenburg's work at the Smithsonian is the standard reference — find that above roughly seventy-five to eighty percent relative humidity the trend reverses and tension climbs steeply again. The size layer, not the linen, dominates the composite at that end, and it responds far more violently than the cloth beneath it. So the familiar "damp equals slack" is a real rule with a real ceiling, and paintings are damaged at both ends of the range rather than only at one.

Three refinements make the model behave properly.

First, equilibrium takes time, and the delay is the conservator's friend. A painting does not track the hourly weather; moisture has to diffuse through the fabric and its coatings, so brief swings are largely absorbed and only sustained changes move the tension much. This is precisely why a sealed backboard helps — it does not stop moisture, it slows it enough that the daily cycle never reaches the paint.

Second, the path matters. Cellulose shows sorption hysteresis: the moisture content reached at sixty percent humidity coming down from wet is higher than the value reached coming up from dry. So "equilibrium" here is not one curve but a pair of them, and a canvas carries a faint memory of its recent history.

Third — and this is the practical sting — the wood is playing the same game. The stretcher absorbs and loses moisture too, and swells and shrinks across its grain far more than along it. Some of what we read as the canvas moving is really the frame changing size underneath it.

That leads to the misconception worth correcting. A slack canvas invites you to knock out the keys and take up the slack. But if the slackness is a humidity state rather than a permanent stretch, you have just tightened the cloth at the bottom of its cycle; when the room dries the tension arrives on top of your correction, and the paint layer — the least extensible thing in the stack — pays for it. The old advice to key out only in a stable, moderate room is not fussiness. It is an instruction about which point on the equilibrium curve you are allowed to calibrate from.

• Why humidity cycling damages paintings more than a steady wrong value does.
• What lining a canvas actually changes — and why it is now a last resort.
• How wooden panel paintings solve the same problem with entirely different failure modes.

Every term the collection defines is gathered in the glossary.