Clay Shrinkage: Divide, Do Not Add a Percentage

Measure your own clay, do not trust the bag

A supplier quotes shrinkage as a range across a firing range, and the number that matters is your clay at your cone in your kiln. The test costs one tile: roll a slab, mark two lines exactly 100 mm apart with a needle, dry it, fire it to the cone you actually use, and measure the marks again. Eighty-eight millimetres means twelve per cent; eighty-six means fourteen. The tool takes those two measurements directly and reports the percentage, which is the input every other calculation needs.

Two details make the test worth doing properly. The marks must be needle-fine, because a two-millimetre pencil line on a hundred-millimetre span is two per cent of uncertainty on its own — a sixth of the answer. And the tile must be fired flat on the same shelf and in the same position as your work, since a piece near a kiln wall can shrink measurably differently from one in the middle.

Shrinkage happens twice, and only one part is optional

Clay loses size in two stages. Drying removes the water between the particles and takes most of the shrinkage, typically five to eight per cent; firing removes the chemically bound water and begins vitrification, adding another three to eight depending on the body and the temperature. The figure the calculator wants is the total from wet to fired, because that is what a test tile measures and what a thrown size has to allow for.

The firing half is the one you can move. The same body fired to a lower cone shrinks less and stays more porous; fired higher it shrinks more and vitrifies. A stoneware that gives twelve per cent at cone six can give fourteen at cone ten, which on a 30 cm bowl is six millimetres of difference — enough to matter for a lid, a stacking set or anything that has to fit a made object. If you change cone, the shrinkage number changes with it and the test tile has to be redone.

Volume shrinks by the cube, and that is what surprises people

Twelve per cent of linear shrinkage is not twelve per cent of capacity. Every dimension drops to 0.88 of its wet value, so the volume drops to 0.88 cubed, which is 0.681 — a mug thrown to hold 400 mL wet holds 273 mL fired. Nearly a third of the capacity disappears, and the loss is invisible in any measurement taken across the rim.

So a piece made to a capacity has to be sized on volume and not on width. Divide the target capacity by 0.681 to get the wet capacity, then throw to that — a 400 mL fired mug is a 587 mL wet one. The calculator handles linear dimensions, which are what you measure on the wheel; the cube is the step to do in your head afterwards, and it is the step that catches out anyone making a set of measures, a teapot or anything sold by the litre.

Frequently asked questions

What shrinkage should I assume if I have not tested?

Twelve per cent is the usual working figure for a mid-fire stoneware and it will be roughly right, but roughly right is the wrong standard for a lid or a matched set. Earthenware fired low tends to run eight to ten, stoneware ten to thirteen, porcelain twelve to fifteen and some porcelains higher. Use the middle of the range for the first piece and the tile for everything after.

Does a piece shrink the same in every direction?

Not exactly. Throwing aligns the clay platelets around the wheel's axis, and aligned platelets shrink slightly more across their alignment than along it, so a thrown pot usually loses a touch more in height than in diameter. The effect is small — a fraction of a per cent for most bodies — and irrelevant to a bowl, but it is visible on a tall narrow form and on anything slab-built where the rolling direction is consistent.

Does glaze change the final size?

It adds a thickness rather than a shrinkage — a few tenths of a millimetre a side, which matters only where two glazed surfaces have to meet. A lid that fits a bisque pot to the millimetre will bind on the glaze firing, so glaze fit is designed with clearance and the seating rim of a lid is usually left unglazed. That is a tolerance decision and not a shrinkage one, which is why the calculator does not model it.