A crackled bowl looks damaged the first time you see it. The surface is covered in a web of hairline lines, as though the glaze had failed and someone decided to keep the piece anyway. Run a fingernail across it and the surface is smooth. The lines are not scratches. They are fractures inside a layer of glass, and the bowl does not leak, because the cracks stop when they reach the clay body underneath.
Crackle is not an accident that was tolerated. It is a difference in shrinkage that was deliberately set, then allowed to happen on a schedule. That distinction is the whole craft.
Two materials, two contraction rates
A glazed pot is a sandwich of two materials that are bonded together at high temperature and then cooled together. The clay body and the glassy glaze each have their own coefficient of thermal expansion: they contract by slightly different amounts for each degree of cooling. At the moment the glaze solidifies, the pair is stress-free. Below that point, every further degree of cooling asks them to become different sizes while remaining stuck to one another.
If the glaze wants to contract more than the body, it is stretched. Glass is strong in compression and weak in tension, so beyond a modest strain it relieves the tension by cracking into a network of polygons. This is crazing, and it is the useful case. If the glaze wants to contract less than the body, it is squeezed instead; the stored compression is released by flakes of glaze popping off the rim or the shoulder, which is a failure nobody wants. Potters therefore aim deliberately at the stretching side of the line, and they aim at a small, controlled amount of it.
This is why a celadon glaze is described by its depth rather than its colour. The same iron-bearing glass that gives the pale green also carries a particular expansion behaviour, and the crackle is one of the things that makes light scatter inside the layer instead of bouncing off a single flat surface.
The crack is scheduled, not instant
Most of the visible network appears as the piece cools through the range where the glaze has hardened but the body is still changing. Quartz in the clay goes through a structural change at about 573 degrees Celsius, with a small but abrupt volume change of its own, and that step can add a second burst of stress. The sound of a freshly fired crackled pot is real: fine ticks continuing for hours or even days after it leaves the kiln, as the network keeps extending to release what is left of the strain.
How far the cracks run depends on how much stress is stored and how thick the glaze is. A large difference produces many small cells; a small difference produces a few long, widely spaced lines. Thick glaze holds more strain and tends to break into a denser web. Because the fissures are narrow and open to the surface, tea, ink or plain dust can settle into them over years of use, which is why an old crackled cup often shows its pattern more clearly than it did on the day it was made. The lines are drawn by time, not by a brush.
How the difference is tuned
Glaze is essentially silica plus a flux that lowers its melting point, plus alumina that keeps it from running off the pot. Adjusting the flux changes the expansion. Alkali oxides such as sodium and potassium raise the coefficient of thermal expansion; more silica or boron lowers it. A potter who wants a tighter crackle adds alkali, or fires a little thicker, or cools a little faster through the critical range. A potter who wants no crackle at all works in the other direction until the two contraction rates nearly match.
Getting it right is a calibration exercise, not a recipe. The same glaze on two different clay bodies can craze on one and fit perfectly on the other, because the body is half of the equation. Test tiles exist for exactly this reason: a glaze that looks beautiful on a small flat sample may behave differently on a curved wall where the coating is unevenly thick. Wall tiles make the same problem visible at a larger scale, where a whole panel must keep one surface behaviour across hundreds of separate pieces.
Reading the network
Two things are worth separating. The crackle is inside the glaze; it does not pass through the clay, so the vessel remains watertight. And it is a network, not a single flaw: once one crack has released the local tension, the next one forms roughly at right angles to it, which is why the pattern tends toward irregular polygons with many T junctions rather than parallel lines.
Chinese potters gave names to different densities of crackle and cultivated several within one family of glazes, sometimes staining the lines so the contrast could be read at a distance. What they were managing was never a decorative motif. It was a measurable mismatch between two materials, held within a narrow window where the failure stays beautiful and stops exactly where it should.