Sourdough stability
A Socratic walk-through of sourdough stability — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why does a jar of flour and water left on the counter for years stay wholesome when the same mixture left alone would rot?
A jar of flour and water on a warm counter is about as hospitable a place as a microbe could ask for: wet, starchy, unsalted, room temperature, open to the air. Leave it alone and in a week it is a foul, mouldy mess. Feed it flour and water every day, change nothing else, and the same jar is still good ten years later — and bakers hand these things down for generations.
The tempting explanation is that the starter is somehow clean. It is not. It is teeming, far more crowded than the jar that rotted. So the difference cannot be the absence of microbes. What is it?
Reasoning it through
REASONING #Begin with what the daily feeding actually does, mechanically. You throw most of the jar away and top it up with fresh flour and water. That is a serial dilution: every microbe in the jar is cut back by the same factor, and only those that can regrow to fill the gap before the next feed get carried forward. Anything that grows too slowly in this particular medium is diluted out of existence within days, no matter how harmless the environment is to it.
So the feeding schedule is itself a filter, and it selects on one criterion — speed of growth on wet flour at room temperature. Who wins that? Lactic acid bacteria, mostly, which are already on the grain, and a few acid-tolerant yeasts. Both ferment the maltose that flour's own amylases liberate from starch, and they do it fast.
Now the second filter, which is the one that does the protecting. The lactic bacteria excrete lactic and acetic acid, and within hours the pH of a mature starter falls to somewhere near 3.5 to 4.0 — a recalled range, and it varies with flour, hydration and temperature. Almost nothing that spoils food, and no foodborne pathogen I know of, grows at that acidity. The canonical threshold in food safety is pH 4.6, below which Clostridium botulinum will not grow, which is why acidified foods are defined by that line.
But the acidity is not simply a hostile pH number, and this distinction matters. Lactic and acetic acid are weak acids: at low pH a large fraction of each molecule is undissociated and electrically neutral, so it diffuses straight through a cell membrane. Inside the cell, where the pH is near neutral, it dissociates and dumps a proton. The invader must then spend energy pumping protons back out, continuously, and it starves doing so. This is why acetic acid is far more punishing than its pH alone suggests, and why a stiff, cool starter — which makes proportionally more acetic acid — keeps more aggressively than a wet, warm one.
There is a third filter that gets less credit than it deserves: sheer numbers. A mature starter carries something on the order of a billion lactic bacteria per gram, with yeasts perhaps a hundredfold fewer — recalled magnitudes, not measurements. A stray spore landing on the surface is not facing an empty field, it is facing an established population that has already eaten the available sugar and acidified the medium. Priority and abundance do a great deal of the work that we credit to chemistry.
Now the control case, the jar left alone. It is not that nothing grows — plenty does at first. But nothing renews the sugar. The lactic bacteria exhaust the maltose, sit in their own acid, and decline. As cells lyse and proteolysis releases ammonia and amines, the pH can drift back up. And at the surface, where there is oxygen, the organisms that are acid-tolerant — moulds and film yeasts — have their opening. That is the honest failure mode of a neglected starter, and it is worth naming precisely: it fails as fuzz on the top, not as putrefaction throughout.
The analogy
THE ANALOGY #Think of a busy market stall that never closes. It is not defended by a lock; it is defended by being permanently occupied and permanently stripped — there is no counter space and no stock left over for anyone else to set up on, and the regulars restock and re-clear it every morning. Stop restocking for a fortnight and the same stall becomes an empty doorway, and someone else moves in.
A market stall's occupants merely take up room, whereas the starter's residents actively poison the site with organic acids, so the exclusion is chemical as well as spatial — and unlike a stall, the site becomes uninhabitable to the occupants themselves once the feeding stops.
Clarifying the model
THE MODEL #The mental model to correct is "the starter stays clean". It is the opposite of clean. It is kept crowded, acidified and nutritionally stripped, and those three conditions are hostile to everything except the organisms that create them. Stability here is not the absence of change but a steady state maintained by continuous input — take the input away and the state collapses, exactly as the neglected jar shows.
It is also worth separating two things the word "selection" is doing here. The acid selects on tolerance; the feeding schedule selects on growth rate in this medium. They usually favour the same organisms, which is why changing schedule, hydration or temperature genuinely shifts a starter's dominant species over some weeks.
One caveat about how much I have simplified: whether particular starters also owe some protection to bacteriocins — targeted antimicrobial peptides — rather than to ambient acid alone is a real and unsettled question. And species composition varies enough between starters that generalising from one to all of them is not safe.
The falsification test follows directly from the account. If exclusion is chemical and numerical rather than an active antagonism, then neutralising the acid should abolish the protection. Take a mature starter, split it, buffer one half back to about pH 6 without removing its residents, and challenge both with the same dose of a marked spoilage organism. The account predicts the acidified half resists and the neutralised half does not. The refuting observation would be an invader that grows happily on flour at pH 3.6 in a sterile control, and is still suppressed in the acid starter — that would mean something more specific than ambient acid is doing the work.
A picture of it
THE PICTURE #How to readStart at the left. The upper path is a loop, and the loop is the point — the jar is only ever in one of three conditions and the daily feed keeps returning it to the first. Follow the branch out of Acidified instead, the one taken when nobody feeds it, and the sequence runs one way to spoilage. Nothing about the microbes differs between the two paths, only whether the cycle is closed.
What became clearer
WHAT CLEARED #A sourdough starter is not preserved, it is maintained. Daily feeding is a selection regime that keeps one guild of fast-growing acid producers dominant, and their acid — working by crossing membranes and draining the invader's energy, not merely by reading low on a pH meter — keeps everything else out. Crowding does the rest. The wholesome jar and the rotten jar hold the same organisms and the same flour; the difference is entirely that one of them is a cycle and the other is an ending.
Where to go next
ONWARD #- Why a stiff, cool starter sours differently from a wet, warm one, and what that does to the bread.
- How the yeasts and the lactic bacteria in a starter avoid competing, given they share the same flour.
Key terms
TERMS #| Term | What it means |
|---|---|
| Backslopping | carrying a portion of a finished ferment into the next batch as the inoculum. |
| Lactic acid bacteria | the acid-producing bacteria that dominate a sourdough starter, mostly Lactobacillus and its relatives. |
| Undissociated acid | the neutral form of a weak acid, which crosses cell membranes and acidifies the cytoplasm from within. |
| Steady state | a condition held constant by continuous input rather than by the absence of change. |
Every term the collection defines is gathered in the glossary.