The Forgotten Science of the Sourdough Starter

Somewhere on a kitchen counter is a glass jar of flour and water, bubbling slowly. The person feeding it thinks of it as a culinary curiosity. It is, more accurately, a continuously running biological reactor that has been operating, in the broad sense, for at least six thousand years. The yeasts and bacteria inside it are doing the same chemistry their ancestors were doing in the kitchens of Mesopotamia, and the rules that determine which strains end up living together are not the rules the bread books describe.

What is actually in the jar

A mature sourdough starter is dominated by two kinds of microbes in roughly stable proportion: lactic acid bacteria, mostly from the genus Lactobacillus and its recently split descendants, and wild yeasts, almost always Candida, Kazachstania, or Saccharomyces species. The ratio is not subtle: bacteria outnumber yeasts by a factor of one hundred to one. The bacteria do most of the chemistry; the yeasts do most of the gas.

What surprised microbiologists when DNA sequencing reached this question in the 2010s is how community-specific the composition is. Two starters fed the same flour, in the same kitchen, by the same person can settle into different stable equilibria. The Stanford-led Global Sourdough Project sequenced five hundred starters from around the world in 2021 and found that the dominant lactic acid bacteria correlated more strongly with the baker's hands than with the flour, the water, the climate, or the geography. The hands, it turned out, were the most reliable inoculation source.

The rules of cohabitation

Why do bacteria and yeasts cohabit at all? In most natural environments they compete; in a starter, they cooperate, and the reason is metabolic. Lactobacillus species ferment maltose preferentially, leaving glucose untouched. The yeasts ferment glucose preferentially, leaving maltose untouched. Each one's leftovers are the other's preferred substrate, and neither species can dominate because the resources are partitioned.

The acidity is the second mechanism of stability. The bacteria produce lactic and acetic acid, which drop the pH to around 3.5 within hours. Most environmental microbes, including pathogens, cannot tolerate this. The yeasts and bacteria that survived the early generations are by definition acid-tolerant, and their continued production of acid is what keeps the system free of contamination. A sourdough starter is, in the technical sense, a self-sterilizing reactor: it generates the conditions that prevent its own takeover.

The age claim, examined

Bakers will sometimes tell you their starter is two hundred years old, traced back to a great-grandmother. The cultural lineage is real; the microbial lineage is not. Population genetics work on bacterial cultures shows that the dominant strains turn over within months under typical home feeding. What persists is the equilibrium, not the individuals: a "two-hundred-year-old starter" is two hundred years of careful selection pressure for the same niche conditions, which produces the same kind of community even if every cell is twentieth-generation.

This is not a flaw of the metaphor. Most living systems we think of as continuous (a forest, a coral reef, a city) are also turnover-and-replacement, not preservation. The continuity is in the form, not the matter.

The geography that does not exist

San Francisco sourdough is the most famous regional sourdough, and the lactic acid bacterium responsible for its characteristic tang, Lactobacillus sanfranciscensis, was named after the city in 1971. For a long time, the city was assumed to be its native habitat. The 2021 global survey found L. sanfranciscensis in starters from Italy, Belgium, Brazil, and Australia. It is one of the most common members of the mature-starter community worldwide. The "regional" identity of San Francisco bread is real, but the microbe that produces it is not regional at all; it is a generalist that thrives in any starter that selects for it.

Why bread is older than agriculture

For most of the twentieth century, the consensus was that bread followed agriculture: people domesticated grains, then learned to make flour and bread. The 2018 discovery at Shubayqa 1 in Jordan complicates this. Charred breadcrumb fragments dated to around 14,400 years ago, predating the local domestication of wheat by roughly four millennia. The makers were Natufian hunter-gatherers, grinding wild grains and other tubers and baking flat breads in stone-lined hearths.

If the bread is that old, the fermentation is plausibly that old too. A flour-and-water mixture, left in a warm place, will ferment spontaneously within a day or two. Once people noticed that the leavened dough rose and tasted better, the practice would propagate. Sourdough may not be a technology that humans invented so much as one that we noticed and kept feeding.

What feeding does

The instructions in a recipe book ("discard half, add equal parts flour and water, let stand twelve hours") are reverse-engineered from microbial population dynamics. Discarding half resets the carrying capacity. Adding flour resets the substrate. Twelve hours is roughly the time for the population to grow back to saturation. Skip a feeding and the bacteria continue producing acid until the pH is too low even for them; the population crashes; the starter "dies." It can usually be revived from a small portion that retained viable spores or a few hardy cells, but the community equilibrium is disturbed and may take weeks to settle.

Backslopping, the practice of keeping a small amount of an old batch to inoculate the next, is the technique that propagates these communities through history. Industrial bakers do it. Yogurt makers do it. The kombucha SCOBY is the same idea. Once you see the pattern, fermentation cultures look less like recipes and more like the oldest continuously running biotechnology infrastructure in human civilization, predating writing, agriculture, cities, and almost every other thing we usually identify as the start of history.

The flour is not the inoculant

One of the most counter-intuitive findings of the modern sequencing work is that the microbes in a mature starter are mostly not from the flour. Flour does carry microbes, but the species that show up in week-one starters disappear by week four, replaced by the acid-tolerant specialists. Where do these specialists come from? The leading hypothesis is the baker's skin and the kitchen environment. Hands, jars, cloths, and the air all carry low-level populations of Lactobacillus, and once the starter selects for them, they bloom.

This is why two bakers using the same flour in the same kitchen still grow distinguishable starters: they are inoculating with different skin and household communities. Your starter is, in a small but real sense, a portrait of you.

The right respect

None of this changes how you feed the jar. The recipe book is approximately right because the equilibrium is robust. But understanding what is actually in the jar, and what older lineage of practice it belongs to, changes how you regard it. The flour-and-water on the counter is connected, by an unbroken chain of technique passed hand to hand, to the bread bakers of Shubayqa fourteen thousand years ago. We have not invented many things that have run that long. It is worth feeding it carefully.