The Custard That Won't Set
You are developing a new creme brulee for a 200-cover brasserie. Your standard formula uses 1 L of 35% cream, 8 egg yolks, and 120 g sugar, baked in a water bath at 150°C to an internal temperature of 82°C, then chilled. The new supplier delivers cream labeled 35% but your first test batch, made to the exact same formula and method, comes out loose and weeping, with a visible watery layer after overnight chilling. You ran a second batch with a fresh dozen eggs from a different lot and got the same result. The walk-in is reading 4°C. Your pastry commis suggests adding two more yolks and baking to 85°C. The brasserie launches in five days and you have 12 L of the new cream on hand.
1. List the plausible causes of the loose, weeping custard and explain how you would isolate the true cause before changing the formula.
Read the worked answer
Weeping (syneresis) and a loose set in a baked custard point to one of a few things:
- Fat or solids variation in the cream. A cream labeled 35% can differ in actual fat, and homogenization/aging affects how the fat network sets. If the new cream is genuinely lower in fat or differently processed, the custard has less structure and more free water.
- Egg variation. Ruled out as the sole cause because a fresh lot gave the same result - but keep it as a secondary variable.
- Over- or under-coagulation. Baking past the coagulation window squeezes water out (over-coagulation = weeping). Baking under sets loosely. Internal temp of 82°C is near the top of the safe window for a yolk-set custard; small oven variance could push it over.
- Water bath / oven behavior. A water bath that runs hot or a convection oven with uneven zones can overcook edges while the center stays loose.
Isolation plan (before touching the formula):
- Measure the new cream's actual fat (or request a spec sheet and a retained sample). Compare against the old supplier.
- Run a controlled side-by-side: same formula, same eggs, old cream vs. new cream, same oven and bath, same probe placement. This isolates the cream.
- Probe internal temperature at the center of several ramekins to confirm you are actually hitting 82°C and not overshooting at the edges.
- Hold a batch overnight at 4°C and observe syneresis.
Only after the side-by-side tells you which variable moved do you change the formula.
2. Evaluate the commis's suggestion to add two yolks and bake to 85°C.
Read the worked answer
It is a plausible but risky fix that treats a symptom.
- Adding yolks increases egg protein and emulsified fat, which will firm the set and can bind more water. That may mask a low-fat cream. But it changes the flavor (more eggy), the cost, and the texture profile, and it does not address why water is separating.
- Baking to 85°C is the dangerous part. Yolk-set custards coagulate in roughly the 80–85°C range; pushing to 85°C risks over-coagulation, which causes weeping by contracting the protein network and expelling water. It could make the problem worse while appearing to set firmer.
Verdict: Do not adopt it as the fix. If the side-by-side confirms the cream is the problem, the correct responses are to source compliant cream, adjust fat/solids deliberately with a tested formula change, or standardize on a measured fat target - and to bake to a verified internal temperature inside the coagulation window, not above it. If a formula change is needed, test it as a deliberate variable, not as a panic patch five days out.
3. You have 12 L of the suspect cream and a launch in five days. What is your sourcing and contingency plan?
Read the worked answer
- Do not build launch stock on unverified cream. Quarantine the 12 L pending the fat/spec result; use it only for products tolerant of variation (e.g., a cooked cream base where you can adjust), not for the flagship brulee.
- Get the spec and a retained sample from the new supplier in writing; compare to the old supplier's fat and processing. If it is out of spec, that is a supplier claim, not a formula problem.
- Secure a verified supply for launch: revert to the proven supplier for the brulee, or approve the new cream only after a passing side-by-side.
- Lock the method: standardize probe-verified internal temperature, bath temperature, and oven zone checks; write it on the station card.
- Contingency menu: if verified cream is short at launch, feature a dessert that does not depend on a delicate baked custard set (e.g., a set cremeux or a chilled set dessert with a wider tolerance) for the first week, and launch the brulee once supply is confirmed.
- Communicate the supply risk to the chef and front of house now, not on launch day.