Cultures

Rennet sets the milk in half an hour and then its job is essentially done. The bacteria work for the next two years.

They do three things, and it’s worth separating them because recipes rarely do: they make acid, which drives the whole make; they make gas, which is either a feature or a disaster depending on the cheese; and they make flavour, mostly much later, during ripening.

The split that ruins the most wheels

Mesophilic means middle-temperature-loving. Thermophilic means heat-loving. The distinction sounds academic and is the single most consequential substitution in cheesemaking.

Mesophilic cultures are happy around 30 °C / 86 °F and are progressively knocked out as you go much above about 39 °C / 102 °F. They’re what you use for cheeses cooked gently or not at all — cheddar, gouda, most fresh cheeses, and nearly everything a beginner makes.

Thermophilic cultures want roughly 45 °C / 113 °F and tolerate the hot cook that alpine and Italian cheeses require — Gruyère, Emmental, Parmesan, mozzarella.

Now the failure. Put a mesophilic culture into a recipe that cooks the curd to 52 °C and you kill your own starter partway through the make. Acidification simply stops. The curd never reaches its target pH, the cheese doesn’t develop, and — because the wheel is under-acidified — you have also handed an advantage to whatever else is in there.

The recipe will not warn you. It names a culture and a cook temperature and assumes you know they belong together. This is a large part of why the pH is the real instruction: watch the acid, and if it stalls mid-make, suspect the temperature before anything else.

Homofermentative and heterofermentative

The other split, and the one that decides whether your cheese has holes.

Homofermentative organisms convert lactose almost entirely to lactic acid. Tidy, predictable, no gas. Most of your acidifying workhorses are these.

Heterofermentative organisms make lactic acid plus carbon dioxide and other products. That CO₂ has to go somewhere. In a cheese designed for it, it forms small openings and the texture is correct. In a cheddar, it is a defect with a name — early blowing — and it will lift and crack your wheel.

So “a culture that makes gas” is not good or bad in the abstract. It’s a specification. Using a gas-producing blend in a close-textured cheese is the same class of error as the meso/thermo mix-up, and it’s why you cannot casually substitute one culture packet for another because the shop was out.

The cast

You don’t need to memorise these, but recognising the names on a packet tells you what it will do.

Mesophilic acidifiers. Lactococcus lactis subspecies lactis and cremoris — the backbone of cheddar, gouda and most of what a home cheesemaker makes.

The buttery one. Lactococcus lactis biovar diacetylactis produces diacetyl, the compound that smells and tastes of butter. It’s why cultured butter and some fresh cheeses have that particular richness. It also produces gas.

The gas producers. Leuconostoc species — heterofermentative, used deliberately where small openings are wanted.

Thermophilic acidifiers. Streptococcus thermophilus, Lactobacillus delbrueckii subsp. bulgaricus, Lactobacillus helveticus — the alpine and Italian workhorses, and the yoghurt organisms too.

⭐ The eyes: what actually makes the holes

Emmental’s holes are the most famous thing in cheese and they are made by a third organism working after the cheese is already a cheese.

Propionibacterium freudenreichii doesn’t eat lactose — the starters already did that. It eats what the starters left behind: lactate. Through the Wood–Werkman cycle it converts lactate to pyruvate and then splits the output two ways:

• part becomes propionate, which is the characteristic flavour of Swiss-type cheese;
• part becomes acetate and carbon dioxide, and the CO₂ is what forms the eyes.

So the flavour and the holes are two outputs of one reaction. You cannot have the taste without the gas, which is a rather satisfying piece of design.

And it explains the strange temperature schedule those cheeses go through, which looks arbitrary until you know why:

Stage	Temperature	What’s happening
Initial ripening	10–14 °C, 2–3 weeks	Ordinary early ripening; propionibacteria mostly quiet
The warm room	20–24 °C	Propionibacteria go to work; the eyes form here
Cold room	4–6 °C	Stop it before the eyes get too large or the wheel splits

That warm room is not a convenience. It is a deliberate stage where you let a specific organism produce a specific gas, and then you stop it on purpose by chilling. Leave it too long and the gas pressure cracks the wheel — the split defect.

Two more details I enjoyed: P. freudenreichii preferentially uses L-lactate over D-lactate, and the subspecies shermanii is unusually resistant to both cold and heat stress — which is very likely why it’s the one that survives in Emmental technology, with its hot cook and its cold-warm-cold schedule. The organism suits the process because the process selected for it.

Also: a small wheel cannot form large eyes, because the gas pressure dissipates before the holes reach full size. Culture, temperature schedule and wheel size all have to agree.

🔴 What stops eyes forming at all

Worth separating from the size question, because they get conflated — and I conflated them myself until my own Swiss came out with none. Size caps how big eyes get. A total absence means the propionic fermentation never happened, and the documented causes are:

• No warm-room stage, or not warm enough — low temperature slows the metabolism, so less gas and fewer eyes.
• Too acidic — propionibacteria are inhibited by low pH.
• Too much salt, which suppresses them (and too little invites what you didn’t want).
• Cooked too hot.
• No P. freudenreichii in the blend — it’s an adjunct, not part of a standard thermophilic starter.
• And the unusual ones: residual antibiotics in the milk, elevated copper, bacteriophage crippling the lactic fermentation so insufficient lactate remains to eat, and some heterofermentative starters actively inhibiting propionibacteria.

Nearly all of those are fixable, which is the practical point: no eyes is a technique failure, not a verdict on your equipment.

Adjuncts, and lipase

Beyond the acidifiers, recipes call for organisms with specific jobs — the surface moulds and bacteria covered in rinds and moulds, and flavour adjuncts that contribute during ripening without doing much acidifying.

Lipase deserves its own warning. It’s an enzyme rather than an organism, it produces the sharp piquant note of provolone and some Italian cheeses by breaking down fat, it works fast, and it is very easy to overdo. Excess lipase reads as soapy or frankly rancid, and unlike most mistakes it cannot be corrected later. Measure it.

DVI versus mother culture

The practical form question, and the one my books argue about.

DVI — direct vat inoculation — is the freeze-dried packet you sprinkle straight into the milk. A measured dose of specified organisms, consistent every time.

A mother culture is one you propagate yourself in milk and keep going, typically dosed at 1–2% of the vat by volume against DVI’s tiny measured amount. The doses are not interchangeable, so a recipe written for one needs converting for the other.

My honest position, having started with packets: DVI buys reproducibility, which is exactly what you want while you’re still learning what “right” looks like — if the cheese fails, at least you know it wasn’t the culture. A living culture buys complexity and independence, and it makes you pay attention. Start with packets. Don’t stay there forever.

---

Next: the curd — cutting, healing, cooking and washing, which is where you actually decide how much water stays in.

Sources

• The Wood–Werkman cycle, lactate → propionate (flavour) and acetate + CO₂ (the eyes); the 10–14 °C → 20–24 °C warm room → 4–6 °C schedule; L-lactate preference; and shermanii’s cold and heat resistance: ScienceDirect — Propionibacterium freudenreichii and Journal of Dairy Science on split defect in Swiss cheese