Marble supports

Why do ancient marble statues lean on tree stumps and drapery that the bronze originals never needed?

Walk through a gallery of classical sculpture and the same object keeps appearing: a tree stump at the athlete's ankle, a bundle of drapery falling improbably to the ground, a little bar of stone bridging a wrist to a thigh. The bronzes have none of it. The stock explanation is that marble is weak and would snap under its own weight, so the copyist propped it. That sounds right, and it is worth doing the arithmetic before believing it, because the arithmetic is not hard and it does not say what you expect.

Take the classic worry, an outstretched arm. Model it as a stone rod about sixty centimetres long and eight thick, cantilevered from the shoulder, and ask what tension the self-weight puts at the root. That is an ordinary beam sum, and I get something under a single megapascal — while marble's tensile strength runs to roughly ten. So a life-size marble arm carries perhaps a twentieth of what the stone can take. The folk explanation fails on its own terms. Static self-weight is not the problem.

Then what is? Start with how marble fails. It is brittle: it does not bend, warn you, and hold; it reaches its limit and separates. And the limit is not really a material property in the way steel's is — it is set by the largest flaw in this block, and marble is veined, bedded and full of them. When a material's strength is a lottery and its failure is total and instant, you do not design to the average. You design with a margin that looks absurd, because the tail of the distribution is where you live.

Now add the loads that are not self-weight. A statue is roughed out, turned, transported, hoisted, set on a base, cleaned, knocked by a passing cart, shaken by a tremor. The moment the figure tips off vertical, its whole mass acts through a lever, and the critical section is not the wrist at all — it is the ankles. A life-size figure in marble runs to something like two hundred kilograms, and all of it stands on two slender columns of stone the width of your fist. Notice where the supports actually go in the gallery: at the leg, at the base, at the hip. That is not where an arm would break. That is where the figure would.

And there is time. Marble is made of calcite grains that expand unequally along different crystal directions when heated, so every day and night of thermal cycling works the grain boundaries looser. The modern demonstration is visible on buildings: thin marble cladding panels bow outward over a few decades and lose much of their strength, and some famous towers had to be re-clad. A statue must survive not one loading but a few hundred thousand.

Set the bronze beside all this and the contrast is not a matter of degree. Cast bronze has a tensile strength of a couple of hundred megapascals — more than twenty times marble's — and it is ductile, so it deforms visibly before it fails. Better still, a large ancient bronze is not solid. It is cast hollow, in pieces, with walls of the order of a centimetre, then joined; so the metal sits where the stress is and the figure weighs a fraction of its stone twin. An outstretched bronze arm is a light lever on a strong ductile joint. An outstretched marble arm is a heavy lever on the worst crack in the block. The same pose is two entirely different engineering problems.

Which is the trap the copyist inherits. A pose composed for bronze — weight on one leg, an arm out, a spear held clear — was designed against a material that tolerated it. Transfer that geometry into marble, as point transfer encourages, and you have imported a shape nobody carving from scratch would have chosen. The stump is the patch.

Two honesties before we call it settled. Not every support is structural: a tree stump beside a hunter, a dolphin beneath a sea-born goddess, a quiver at the hip are iconographic as well as load-bearing, and some struts are workshop scaffolding never cleaned off, or nineteenth-century restorations. And the inference many people run backwards — strut, therefore Roman marble copy of a lost Greek bronze — is far weaker than it is usually stated. Greek marble originals carry supports too, and the older picture of a mechanical copying industry churning out replicas has been substantially reworked by specialists who now see far more adaptation and far fewer straight copies.

The correction is that "marble is too weak" is the wrong shape of claim. Marble is easily strong enough for the average day; it is not strong enough for the worst day, and with brittle failure the worst day is the only one that matters. Supports buy margin against flaws, handling and time, not against arithmetic.

It also helps to keep three separate advantages of bronze apart, because they are usually merged into one. Bronze is stronger in tension; bronze is ductile, so it fails gradually and visibly; and bronze is hollow, so the same figure imposes far less load in the first place. Marble did not merely lose a contest of strength — it was handed a shape from a competition it was never in.

• Why some struts were left visible and polished while others were clearly meant to be removed.

Every term the collection defines is gathered in the glossary.