Scaling a Cheese Recipe Without Wrecking It
Cheese recipes are written for a batch size the author happened to use — four litres, two gallons, seven-point-six litres because that is what a US two-gallon jug holds. Yours will rarely match, and the scaling is where beginners quietly go wrong.
Some things scale linearly. Some do not scale at all. Knowing which is which is most of it.
What scales with the milk
Culture, rennet, calcium chloride and salt all scale in direct proportion. Double the milk, double the amount. That is a simple ratio and the culture ratio calculator does it exactly:
| Recipe says | You have | Scale | Use |
|---|---|---|---|
| ¼ tsp per 4 L | 10 L | 2.5× | 0.63 tsp |
| 1 packet per 7.6 L | 19 L | 2.5× | 2.5 packets |
| ½ tsp per 10 L | 4 L | 0.4× | 0.20 tsp |
| ¼ tsp per 4 L | 38 L | 9.5× | 2.38 tsp |
The arithmetic is trivial and the errors are not. Halving an odd fraction in your head, or converting gallons to litres mid-recipe, is where a batch gets a third of the culture it needed — and an under-cultured make does not just taste flat, it fails to acidify properly, which is a safety consideration as well as a flavour one.
What does not scale
This is the part that catches people scaling up.
Time does not scale. A 30-minute ripening is 30 minutes whether you have 4 litres or 40. Culture activity is a rate, and giving a larger batch proportionally longer over-acidifies it.
Temperature does not scale. 32°C is 32°C. What does change is how long the milk takes to get there and how long it holds — a large volume heats slowly and cools slowly, so the same recipe run at scale spends longer in transit between temperatures.
Cut size does not scale. A pea-sized curd is pea-sized regardless of vat volume, because it is about the distance whey has to travel out of each piece.
Press weight does not scale linearly, because pressure is force over area. A wheel twice the diameter has roughly four times the surface, so matching the recipe's pressure needs about four times the weight, not twice.
The thermal-mass problem
The single biggest practical difference when scaling up is that large volumes change temperature slowly, and that changes the make in ways the recipe does not mention.
Forty litres takes far longer to warm from fridge temperature to renneting temperature than four does, and during that time the culture is already working. It also resists cooling, so a cook step that the recipe expects to end promptly keeps going. The usual symptoms are a make that runs more acidic than intended and a curd that firms faster than the timings suggest.
The practical response is to watch the cheese rather than the clock — flocculation, clean break, curd feel — and to treat the recipe's times as expectations to check rather than instructions to follow.
Scaling down has its own problems
Going smaller is not simply easier. Two issues dominate.
Measurement resolution. A 4-litre batch might need 0.4 mL of rennet or a fifth of a teaspoon of culture. Kitchen spoons do not resolve that, and the proportional error from a slightly heaped measure is much larger on a small batch than a big one. A small syringe and a set of fine measuring spoons solve most of it.
Surface-to-volume ratio. A small vat loses heat faster and a small wheel dries faster, because both have proportionally more surface. A recipe's aging time assumes a wheel of roughly the size it was written for; a half-size wheel will reach the same moisture sooner, which the aging calculator can show you in weight terms.
Culture packets and the "DCU" problem
Freeze-dried cultures are often sold in units sized for a specific milk volume — a packet "for 10 litres", or measured in DCU. Where a recipe calls for a packet and you have a different volume, scale by the ratio as above; where it calls for a fraction of a packet, opening and re-sealing repeatedly is a real practical issue, because moisture ingress kills viability.
The common workaround is to make up a bulk mother culture, or to buy packets sized closer to your usual batch. Both are better than an opened packet living in a cupboard for six months and delivering an unknown fraction of its stated activity — which presents exactly like a failed make.
A sanity check after scaling
Two numbers worth predicting before you start, so you can tell whether the scaled make behaved.
Run the milk through the yield calculator and note the expected finished weight — whole cow milk at 3.6% fat and 3.2% protein to a 38% moisture cheese predicts about 10.1 kg per 100 kg of milk. And note the acidity you expect at draining, so you can check with the acidity calculator rather than guessing.
If the scaled batch lands well below the predicted yield, or acidifies faster than the small version did, the cause is usually one of the non-scaling variables above rather than the ingredient amounts — and knowing which of the two went wrong is the difference between fixing it and repeating it.