Case Study 2 — The Boulangerie That Survived: A 200-Year-Old Starter and What's Actually In It

A claim, a measurement, a surprise

There is a bakery in San Francisco — one of several in the city that make this claim, and one of the few with documentation that supports it — whose owners say their sourdough starter has been continuously fed since 1849. The Gold Rush. The same bubbling jar of flour-and-water culture, fed and re-fed and re-fed for 175 years and counting, used every day to leaven the bread sold at the front counter. Tourists come specifically to taste a loaf made from a starter older than the state itself.

The claim is not strictly mythical. The bakery's records — incomplete, but not absent — trace the starter through several owners, several relocations within the city (including, twice, after the 1906 earthquake and again after a 1980s building fire), and several near-disasters when the starter was at risk of being lost. Each owner, in a kind of relay-race oral tradition, has fed and propagated the same line of microbes. The starter currently in the back-of-house refrigerator has, in the most literal sense, been alive for 175 years.

But what does that mean? The yeast cells you would see if you put a drop of the starter under a microscope tomorrow — none of those individual cells are 175 years old. Saccharomyces cerevisiae divides roughly every 90 minutes under good conditions; the average yeast cell in an active starter has a lifetime measured in hours. The cells alive today are the great-great-great-great-(many greats)-grandchildren of the cells that came across the country in a covered wagon in 1849.

What persists is not the cells but the lineage — the genetic descendant of the original community, plus all the genetic drift, mutation, and possibly horizontal gene transfer that has happened over thousands of generations of doubling.

This raises a real scientific question. Is the starter today the same starter as in 1849? The answer is interesting, and it points to something genuinely profound about microbial communities and what we mean by "the same" living thing.

The 1971 measurement

In 1971, a microbiologist named Leo Kline, working at the Western Regional Research Center of the U.S. Department of Agriculture in Albany, California, took samples from several San Francisco sourdough bakeries that all claimed deep historical lineages. He cultured the organisms, identified them, and published the result in Applied Microbiology — a paper now famous in food-science circles. He found that the dominant lactic acid bacterium across all of these starters was a species he was the first to characterize formally and which he named, after the city where he found it, Lactobacillus sanfranciscensis.

L. sanfranciscensis is a fascinating organism. It is a heterofermentative bacterium — meaning it produces both lactic acid (the dominant souring acid) and acetic acid (vinegar), along with CO₂. It specializes in metabolizing maltose — which is exactly the sugar that wheat amylases produce from starch — while leaving glucose for the yeast to use. This metabolic complementarity is part of why the partnership is stable: the bacterium and the yeast are not competing for the same food. They are sharing.

Kline also identified the dominant yeast as Candida humilis (later renamed Kazachstania humilis), not Saccharomyces cerevisiae as the popular literature had assumed. K. humilis is a wild yeast that, like L. sanfranciscensis, specializes for the cool, acidic, maltose-rich conditions of a long-fermented sourdough. It is acid-tolerant, which most ordinary baker's yeast strains are not. The two organisms have, in effect, co-evolved into a stable consortium.

Subsequent research, with modern DNA-sequencing techniques, has refined this picture. The "San Francisco sourdough community" is not unique to San Francisco — L. sanfranciscensis and K. humilis (or close relatives) appear in long-fermented sourdoughs around the world. What seems to be happening is convergent selection: any starter that is fed regularly with wheat or rye flour, kept at moderate temperatures, and allowed to acidify will tend to end up with this particular community of organisms, regardless of where in the world it started. The starter community is shaped by the conditions, not by the geographic origin of the inoculum.

Which means that the 175-year-old San Francisco starter and a starter you build from scratch in your home kitchen over two weeks will, given enough time and consistent conditions, both end up dominated by the same organisms.

This is one of the most surprising and least-discussed facts in fermentation science. The "lineage" of a starter — the genetic continuity from 1849 to today — is real, but the community composition is not held in place by lineage. It is held in place by the selective conditions the baker provides. A starter is a community-shaped-by-environment, not a genetic family heirloom.

What this means for Maya's starter

Maya Okonkwo's 14-week-old starter, started from a packet of whole-wheat flour and the wild microbes of her Atlanta kitchen, will — assuming she keeps feeding it consistently with wheat flour at moderate temperatures — converge over time to a community that is microbiologically very similar to the San Francisco one. The convergence is gradual; in the first weeks, all sorts of organisms compete and die off; by month three or four, the dominant species are typically a Saccharomyces or Kazachstania yeast plus L. sanfranciscensis or a close relative. The pH stabilizes around 3.8–4.0. The microbial population stabilizes at around 10^8 yeast cells per gram and 10^9 lactic acid bacteria per gram.

If Maya's starter were sequenced today and compared to the San Francisco bakery's 175-year-old starter, the two communities would be identifiable as the same general consortium. The specific strains might differ slightly — there's room for some genetic drift across populations — but the species composition would be largely overlapping, and the metabolic outputs (the relative amounts of lactic acid, acetic acid, ethanol, CO₂, and various flavor esters) would be in the same range.

This is why bakers around the world can produce sourdough bread that tastes recognizably like sourdough, even though no two starters share an ancestor. The bread tastes the way it tastes because of the chemistry of the consortium, and the consortium converges to the same chemistry under similar conditions.

What the heritage adds

Does this mean a 175-year-old starter has nothing special about it? Not exactly. There are a few things lineage probably does contribute, even if the species composition is similar.

Strain-level diversity within a species. Even within L. sanfranciscensis, there is genetic diversity at the strain level — different strains produce slightly different ester profiles, slightly different acid balances, slightly different growth rates. A 175-year-old starter has had time for many generations of selection, and the specific strains living in it have been optimized for the specific feeding regimen, temperature, and flour of the bakery. Maya's 14-week-old starter contains strains that haven't yet been selected as deeply for her specific kitchen conditions. Over years, her starter would acquire its own character.

Less-dominant species and rare metabolisms. A long-established starter sometimes harbors small populations of unusual organisms — a particular yeast species in low abundance, an acetic-acid bacterium that contributes a specific flavor — that haven't yet established in a young starter. These minority members can contribute outsized flavor effects. Whether they survive at population levels low enough to escape rapid culturing techniques but high enough to influence flavor is an active research question.

The history itself. A 175-year-old starter has cultural significance independent of its microbiology. The bread the bakery sells is not just food; it is a connection to the Gold Rush, to the city's history, to the lineage of bakers who have kept the starter alive. This is real and it matters. But it is a different kind of value from "the bread tastes objectively better than what you can make at home."

When the starter is sequenced and the bread is taste-tested, the differences between a 175-year-old commercial sourdough and a careful home cook's three-year-old starter are smaller than most people would guess. The differences between a sourdough loaf and a commercial-yeast loaf are large. The differences within the family of well-made sourdoughs are small.

A complication: starter survival across catastrophe

The history of long-lived starters is also a history of near-disasters. The San Francisco bakery's starter has survived, by various accounts:

• The 1906 earthquake and fire, when the original bakery was destroyed. The starter was reportedly removed in a covered jar by a baker fleeing the building; it was kept on a back porch for six weeks while new premises were established.
• The Great Depression, when the bakery nearly closed and was kept open by a single owner who slept above the shop.
• A 1980s small fire that damaged the back kitchen but did not reach the refrigerator where the starter was stored.
• A 1990s ownership transition in which the starter was nearly thrown out by a temporary kitchen worker who did not know its significance, before a longtime employee intervened.

Each of these incidents, by the strict accounting of "did the same population of organisms persist?", might or might not count as continuity. If the starter went six weeks without feeding after the 1906 fire — which it did, by the bakery's own oral history — the population would have largely died off, and what restarted it was the small population that survived in dormancy at the bottom of the jar. This is similar to the way species survive ecological catastrophes: a small remnant population repopulates the system. The dominant community after the catastrophe is a recovered population, not a continuous one.

In the strictest microbial sense, then, the "175 years of continuous starter" may have had several near-extinction events with founder-effect recoveries. The dominant species are still the dominant species (because the conditions favor them), but the specific strains may not be direct descendants of the 1849 strains. This is genetic continuity at the species level but possibly not at the strain level.

This is fine. It is not a knock against the bakery's claim. It is simply what microbial lineages look like under real-world conditions — punctuated by founder effects, near-extinctions, and recoveries, with the species composition stable because the selective environment is stable.

The story of the starter is, in this sense, a microcosm of the story of any long-lived domesticated organism — a wheat variety, a cheese culture, a sourdough yeast, a cattle breed. The lineage matters. The conditions that maintain the lineage matter more.

What Danny saw at the restaurant

Daniel Reyes-Park, the food-science student, has been thinking about this. The restaurant where he works in Chicago keeps its own house sourdough, started by the chef five years ago when the restaurant opened. Danny tracked the chef down recently and asked him: where did the inoculum come from? The chef said he had used flour, water, and a small piece of an active starter from a friend who runs another fermentation-focused restaurant in Brooklyn. The friend's starter, in turn, had come from a wild fermentation begun in 2015, when the friend was experimenting with rye-flour-only inoculation. The 2015 starter, before that, traces back to nothing in particular — just a lucky catch of wild yeast and bacteria from a Brooklyn kitchen.

Five years of feeding. Brooklyn → Chicago → Danny's hands. The Chicago starter and the Brooklyn starter, sequenced today, would be "the same starter" in the sense that they descend from a common 2015 inoculum. They would also be "different starters" in the sense that they have lived in different kitchens, with different flours, at different feeding schedules, for several years. The species composition would likely be very similar. The strain-level differences would be measurable but small.

When Danny told the chef about the San Francisco bakery's claim of 175 years, the chef said: "That's a great story. And it doesn't change much in the bread. The conditions you give it determine what lives in it. The story is for the customer; the chemistry is for the cook."

This is, in essence, what this case study has been trying to articulate. The microbiology of sourdough is not about ancient lineages, however romantic they are. It is about the conditions a baker provides. Get the conditions right, give it weeks rather than days, and your starter will converge to something microbiologically indistinguishable from the storied ancestral starters. The romance is real and is allowed to be real. The chemistry is durable and does not depend on the romance.

Maya's 14-week-old starter, kept properly, will become as good as anyone's. So would yours.

Analyze this

1. The case study argues that a starter is "shaped by environment, not by lineage" — that two starters with no common ancestor can converge on the same microbial community given similar conditions. What does this imply for the popular belief that "old starters are better than new ones"? Identify two ways this belief is partly true (despite the environmental-shaping argument) and two ways it is largely false.

2. The 1971 Kline study identified L. sanfranciscensis and K. humilis as the dominant organisms in San Francisco sourdoughs. Subsequent studies have found these or closely-related organisms in long-fermented sourdoughs worldwide. What selective conditions in a sourdough starter favor these particular organisms? Why do they outcompete other yeasts and bacteria over time?

3. After the 1906 fire, the San Francisco starter reportedly went six weeks without feeding before being revived. What would have happened to the microbial population during those six weeks? Specifically: which organisms would have died first? Which would have survived? What does this say about the founder effect in microbial recovery?

4. A friend of yours is shopping for a "sourdough starter to keep at home" and is considering paying $75 to receive a sample of a "200-year-old San Francisco starter" by mail. Based on this case study, write a 200-word response advising her: should she buy it, or build her own? What will she gain or lose with each choice?

5. The case study mentions that Maya's starter would, given enough time, become "microbiologically indistinguishable" from the San Francisco starter under similar feeding conditions. What does "indistinguishable" mean here? At what level (species, strain, individual organism) is the indistinguishability real? At what level might subtle differences remain? Could a sufficiently-precise analytical method (say, full DNA sequencing of all organisms present) tell the two starters apart? What would the differences mean for the bread they produce?