Pastry
A water bath exists because an oven is too fierce for a custard
Surrounding a dish with water caps the temperature at its sides and slows the whole exchange, which is the only reliable way of getting the centre of a custard to set before the edges overshoot.
By Clara Lindqvist4 min read
A custard has a narrow target and an oven has none
Egg proteins coagulate over a range of a few degrees, and past the top of that range they contract, squeeze out the liquid they were holding and curdle. A baked custard therefore has to arrive at a specific temperature throughout and then stop. An oven set to a hundred and fifty degrees, meanwhile, is trying with some enthusiasm to bring everything in it to a hundred and fifty degrees, and it makes no distinction between the outside of a dish and the middle.
Left in direct oven heat, the sides and base of a ramekin race ahead. By the time the centre has reached setting temperature, the outer centimetre has passed well beyond it and has begun to break down, which is why an unprotected baked custard is rubbery and weeping at the edge and still loose in the middle. Nothing about the recipe fixes this. It is a heat transfer problem and it needs a heat transfer answer.
Water caps the temperature at the sides of the dish
Set the dish in a tray of water and the sides are no longer in contact with hot air but with a liquid that cannot exceed its boiling point. Whatever the oven is set to, the water surrounding the dish sits below a hundred degrees, so the maximum temperature the custard can ever see at its boundary is well within the range where egg sets rather than curdles. The oven can be hotter than the custard could survive and the custard is still safe.
The second effect is at least as important and gets discussed less. Water has a very high heat capacity, so warming the bath absorbs a great deal of energy, and the whole assembly changes temperature slowly. That slowness gives the centre time to catch up with the edges, which is the actual objective. A bath that caps the temperature but was filled with boiling water and set in a hot oven has given away half of what it was for.
Curdling and weeping are one failure at two stages
Overheating a custard produces two symptoms that look unrelated. Early on, the protein network tightens and begins to expel the water it was holding, and liquid pools at the surface or around the edge — syneresis, the same contraction that makes an over-set jelly weep. Push further and the network breaks into visible curds suspended in that liquid, which is the point at which the dessert is finished.
The reason a starch-thickened custard tolerates more abuse is that starch granules hold water independently of the protein and get in the way of protein chains linking too tightly. That is why a pastry cream can be boiled and a crème caramel can’t. It also means that adding a little starch to a baked custard buys real insurance, at the cost of a texture that reads as slightly less clean.
The practical details all follow from the mechanism
Hot water rather than cold, because the bath has to be doing its work from the beginning rather than spending twenty minutes warming up while the oven attacks the dish above the water line. Water coming at least halfway up the sides, because anything above that line is exposed to oven heat directly. A cloth or a sheet of paper under the dishes, because metal-to-metal contact conducts heat into the base far faster than water does and produces a set bottom under an unset middle.
Covering the tray loosely with foil is worth doing where a skin is unwanted, since it suppresses radiant heat from above and slows evaporation from the bath. And the custard comes out when the centre still has a distinct wobble, because a dish this size carries a great deal of stored heat and the setting continues for several minutes afterwards. Waiting until the middle looks done means removing something that is already past it.
Where the bath is unnecessary
Not every baked egg mixture needs one. A quiche-style filling with a high proportion of starch, flour or cheese is buffered enough to manage in direct heat, and a bread and butter pudding has bread throughout it doing the same job — the custard is in thin films between absorbent solids rather than in a single mass, so the heat has a much shorter distance to travel and the structure is far more forgiving.
The dishes that genuinely need the bath are the plain ones: a set custard, a crème caramel, a cheesecake baked for a smooth rather than a cracked top. Anything that is mostly egg and liquid, and mostly one thickness, and expected to be silky. It is worth noticing that these are also the desserts that look easiest on paper, which is a fairly typical arrangement in pastry.
Common questions
Can I skip the water bath if I lower the oven temperature?
Partly. A low oven reduces the gradient and helps, but it doesn’t cap the temperature at the sides of the dish and it doesn’t add the thermal buffering that slows everything down. The result is usually better than a hot oven and still less even than a bath.
Why did my cheesecake crack?
Because the outside set and contracted while the centre was still expanding, and the surface tore. A water bath moderates that by slowing the whole thing and keeping the edges from running ahead. Cooling it slowly in a turned-off oven helps as well, since a sudden drop in temperature contracts the surface sharply.
Does the water need to be boiling when it goes in?
Hot, not necessarily boiling. Cold water means the bath spends a long time warming while the top of the dish is exposed to full oven heat, which defeats the purpose. Hot tap water or water from a kettle left to stand for a minute is entirely adequate.
Deputy editor, The Last Bite
Clara writes about baking basics, chocolate & sugar, frozen, mostly the parts other people skip and is unreasonably interested in the detail nobody else checks.
