How to Read a Cheese Recipe

A cheese recipe looks like a cooking recipe and isn’t one. It has ingredients and numbered steps and times, so it invites you to treat it like a cake — follow the instructions, get the result. Then you follow the instructions exactly and get something that isn’t what the book promised, and there’s no obvious reason why.

The reason is that most of the numbers in a cheese recipe are descriptions of what happened in somebody else’s kitchen, and only a few of them are actual instructions. Learning which are which is the whole game. This page is the translation guide.

The one thing to internalise: the clock is a proxy

Take a line from any hard-cheese recipe:

Add the culture, stir, and ripen at 90°F for 45 minutes.

What that step is actually trying to achieve is a certain amount of acid development — the milk needs to reach roughly a particular pH before you add rennet, because the rennet’s speed and the curd’s final texture depend on it. The forty-five minutes is the author’s estimate of how long that took with their milk, their culture, their dose, and their kitchen.

Change any of those and the time changes. Fresher milk, colder room, a slightly light culture dose, a different season — the acid arrives sooner or later, and the clock lies to you either way. Follow it faithfully and you’ll add rennet to milk that isn’t ready, or milk that’s gone too far, and the cheese will be subtly wrong in a way no step of the recipe mentions.

So: the pH is the instruction. The minutes are an estimate. Where a recipe gives you both, believe the pH. Where it gives you only minutes, understand that you’ve been handed an estimate and watch the milk instead.

This is the difference between the entry-level books and the technical ones, and it’s the single biggest available step up in skill. It’s also the argument my nine books are having with each other.

What to watch instead of the clock

pH, if you have a meter. Cheap meters need calibrating far more often than their instructions claim; an uncalibrated meter is worse than no meter, because it’s confidently wrong.
The flocculation point — for the rennet set, the moment the milk first shows a faint graininess when you turn a bowl on its surface. Multiply the time from adding rennet to that moment by a factor the recipe gives (typically 3 to 5 depending on the cheese) and cut then. This is a measurement of your own vat, which is why it works when a flat “cut after 45 minutes” doesn’t.
The clean break — the curd splitting cleanly along a knife rather than tearing raggedly.
How the curd feels in the hand after cooking: squeaky and springy, or already grainy.

Three of those need no equipment at all.

The numbers that ARE instructions

Some numbers in a cheese recipe are not estimates and must be respected.

Rennet strength. Rennet is sold at different strengths and dosing “a quarter teaspoon” is meaningless without knowing which. The unit is IMCU (International Milk-Clotting Units). A recipe written for single-strength rennet and executed with double-strength gives you a fast, brittle, bitter curd. Check the bottle against the book, and if the book doesn’t say, work from the bottle’s own dose per gallon.

Temperatures during the cook, especially the top temperature. This isn’t a proxy for anything — the temperature directly determines how much the curd contracts and how much whey it expels, and for the alpine cheeses it also selects which bacteria survive. A cooked-pressed cheese held ten degrees low is a different cheese, not a slightly worse one.

Salt, as a percentage of curd weight. Recipes that say “sprinkle with salt” are asking you to guess at one of the two variables that decide the whole outcome. If a recipe gives a percentage, weigh the curd and hit it.

Calcium chloride, if you’re using pasteurised milk. Pasteurising knocks out some of the soluble calcium the rennet network needs, so store milk gives a weak, mushy curd. A small CaCl₂ addition buys much of it back. This is a real correction, not an optional extra — but don’t overdo it, because too much makes the curd tough and can taste of it.

Aging temperature and humidity, which are the whole reason for having a cave rather than a refrigerator. A cheese aged at fridge temperature and fridge humidity isn’t aged slowly; it’s a different, drier, cracked thing.

Culture: what the packet is actually telling you

Recipes call for cultures in ways that assume you know the shorthand.

Mesophilic vs thermophilic is the fundamental split — middle-temperature organisms that die if you cook the curd hot, versus heat-loving ones that need it. A mesophilic culture in an alpine recipe gets cooked to death partway through, and the cheese simply stops developing. This one substitution ruins more wheels than anything else.

DVI (direct vat inoculation) means the freeze-dried packet you sprinkle straight in. A mother culture means you propagated it yourself in milk first, and a recipe written for one needs its dose converted for the other — mother culture is typically 1–2% of the milk by volume, where DVI is a tiny measured dose.

Adjunct cultures are the extras doing specific jobs rather than the main acidification: Propionibacterium for the eyes and sweetness in Swiss types, Brevibacterium linens for a washed rind’s smell and orange colour, Penicillium candidum for a bloomy rind, P. roqueforti for blue, Geotrichum candidum as a rind-conditioner. If a recipe lists one of these, it isn’t decoration — leave it out and you have made a different cheese with the same shape.

Lipase deserves a warning: it’s how you get the sharp, piquant note in provolone and some Italian cheeses, it works fast, and it is very easy to overdo. Excess lipase reads as soapy or rancid and cannot be undone.

Yield, and why your wheel is smaller than the book’s

Recipes state a yield — “makes about 1 lb from 2 gallons” — and yours will often come out under it. It usually isn’t a mistake. Yield tracks the fat and protein content of the milk, which varies by breed, by season, and by how the animals are fed. Jersey milk yields more cheese per gallon than Holstein. Late-lactation milk differs from spring milk. Store milk that’s been standardised to a fixed fat percentage is its own case.

A useful habit: record what you actually got, every time, alongside the milk you used. After a few makes you’ll know your own yield, and the book’s number stops mattering. Mary Karlin’s book ships record-keeping forms for exactly this reason, and I ignored that advice for far too long.

The steps a recipe won’t tell you

Things I’ve learned the recipe assumes you already know:

“Stir gently” has a meaning. Fresh curd tears, and torn curd loses fat into the whey — visible as a whey that’s cloudier and whiter than it should be. That fat is yield you’ll never see again.
“Let it drain” is not a fixed length of time. It’s until it stops, and that depends on how you cut and how warm the room is.
The wheel gets flipped more at the start than later, because that’s when moisture is still moving and gravity will otherwise pull it all to the bottom.
Pressing weight is pressure, not pounds. Twenty pounds on a small mould is a completely different thing from twenty pounds on a big one, because the curd surface differs. This is the number home recipes get wrong most often, and it gets its own treatment on the presses page.
Nearly every failure is a moisture failure or a pH failure, and usually both, since they’re coupled. When something’s wrong, start there rather than hunting for an exotic cause.

The short version

Believe the pH, the temperature, the rennet strength and the salt percentage. Treat every duration as an estimate. Write down what you actually did.

Do those three and you’ll get more out of any recipe in any of the books than following one of them perfectly.

Next: the family tree for the shape of the whole craft, or my bookshelf for which books say which of the above.