Choux is cooked twice and the second cooking is steam

The paste is cooked before it is baked

Choux begins on the hob, which is unusual for a pastry. Water and butter are brought to a boil, flour is added all at once, and the mixture is stirred over heat until it forms a smooth mass that pulls away from the pan. What is happening during that stirring is starch gelatinisation: the starch granules in the flour absorb the hot water and swell into a paste that can hold a great deal of liquid. That gelatinised starch is what allows choux to carry so much egg without collapsing.

Continuing to cook the paste after it comes together drives off some of the water. This step is often described vaguely, and the reason for it is precise. The paste has to end up dry enough that the eggs added afterwards can bring the mixture back to the correct consistency; a wet paste takes fewer eggs, and fewer eggs means less protein to set the structure later.

Eggs are added to a texture, not to a number

The quantity of egg is the single most consequential variable in choux and it cannot be specified reliably by weight, because it depends on how much water was driven off in the previous step, which depends on the pan, the heat and the cook. This is why experienced recipes describe a target consistency instead: the paste should fall from a spoon in a thick ribbon and a trough drawn through it should close slowly rather than either holding open or vanishing.

The eggs are doing two jobs. Their water becomes steam in the oven and provides the lift, and their proteins coagulate to build the walls that hold the resulting cavity open. Too little egg and there is not enough water to expand and the buns stay small and heavy. Too much and the paste is too slack to hold its piped shape, spreads, and rises into something flat and misshapen. The margin is narrower than it sounds, which is why the last egg is added in fractions.

In the oven, everything happens by vapour pressure

A piped bun contains no yeast, no baking powder and no beaten-in air worth mentioning. What it has is a lot of water, held in a starch and egg matrix. In a hot oven, the surface sets quickly while the interior heats past the boiling point of water, and the steam produced has no escape route through the sealed skin. It pushes outwards, the still-elastic dough stretches, and the bun inflates.

While that is happening, the egg proteins coagulate and the gelatinised starch firms, converting the stretched shell into a rigid structure. The interior remains largely hollow because the water that was there left as steam and pushed the material to the edges. This is a race, and the whole method is arranged to win it: a hot oven at the start to generate steam fast, and enough time afterwards at a moderate temperature to dry the shell out so it does not soften and sag.

Which is why the collapses happen when they do

A bun that rose beautifully and then sank as it cooled was almost certainly underbaked. The shell had inflated but had not dried enough to be self-supporting, so when the internal steam condensed as it cooled, the pressure dropped and the walls fell inwards. The remedy is more time at a lower temperature rather than a hotter oven, and piercing each bun after baking to let the remaining steam escape helps considerably.

Opening the oven door early causes the same collapse by a different route, dropping the temperature at the moment the structure is still soft and the internal pressure is doing the work. Soggy buns that never crisped are usually a humidity problem — too many on a tray, in an oven that could not clear the moisture they gave off. There is no leavening safety net here. Everything about the result comes from managing water.

The variations are all the same physics

Éclairs, profiteroles, gougères, churros and the fried Spanish and Latin American versions of the same paste all rely on this mechanism, with the differences arising from shape and cooking medium rather than from anything fundamental. Fried choux behaves comparably: the hot oil sets a skin, the interior steams, and the pastry inflates. The oil temperature matters for the same reason the oven temperature does.

The one addition that changes the character is a craquelin or a coating, applied before baking, which restrains the surface slightly and gives a more even, rounded rise as well as a crisp top. It is a genuine structural intervention and not just decoration, though it also makes the pastry less forgiving of an under-set base. As with most of choux, the fix for almost every problem is the same: drier paste, hotter start, longer finish.