Drop a spoonful of starter in water. If it floats, bake. If it sinks, wait. It's the single most-repeated piece of sourdough advice, and it's also the most misunderstood. The float test answers one narrow physical question — it does not answer "is my starter ready."
The float test measures buoyancy, nothing more. A dollop of starter floats when it contains enough trapped CO₂ to make its average density lower than water. That's it. The test doesn't measure acidity, so it can't tell you whether the culture has turned too sour for a mild loaf. It doesn't measure gluten development, so it can't tell you anything about your dough's structure. And it doesn't directly measure yeast population — it measures the byproduct of yeast activity that happens to still be physically trapped in the matrix at the moment you scoop the sample.
Gas retention and yeast activity are correlated, which is why the test works at all. But correlation is not the same as precision. A starter can be actively fermenting and still fail the test, and a starter that's already declining can still pass it. The test is a proxy, one step removed from the thing you actually care about — is this the best moment to mix my dough.
That gap matters more than it seems. Most home bakers treat a float as confirmation that everything about the starter is in order — flavor, strength, timing — when it only confirms one narrow physical fact. Two starters can both float and be in completely different states: one climbing toward peak with a mild, yeasty aroma, the other already sliding past it with acid building and rise capacity dropping. The spoon in the glass can't tell them apart.
A starter doesn't need to be at peak to trap enough gas to float. Somewhere around 60–70% of its total rise, the bubble structure inside is already dense enough to lift a spoonful above the waterline. That means a starter can pass the float test a full hour or more before it reaches true peak — before yeast population maxes out, before gas production rate is at its highest, before the flavor and rise balance is actually optimal.
If you use the float test as your only readiness check, you'll consistently bake earlier than you should. The bread will still rise — it just won't rise as much as it would have at true peak, because the yeast hadn't finished building toward maximum activity yet.
The opposite failure is just as common and less talked about. Trapped gas doesn't vanish the instant yeast activity starts declining. After peak, the culture's gas-holding structure degrades gradually — bubble walls weaken, gas escapes, acidity climbs — but that process takes time. In practice, a starter that has already passed its true peak can keep floating for another one to two hours.
So the float test doesn't give you a point in time. It gives you a window — and a wide one. "Yes, it floats" can mean anywhere from 70% risen to two hours past peak. That's not a precision test; it's a rough floor. Anyone using it to pinpoint the exact moment to mix dough is asking it to do a job it was never built for.
Beyond timing, the float test fails outright for entire categories of starters.
Stiff, low-hydration starters (around 50% water): these rarely float even at full activity. A stiff dough-like matrix doesn't let gas bubbles distribute and merge the way a loose, high-hydration starter does — the gas gets trapped in smaller, more isolated pockets instead of forming the larger buoyant structure the test relies on. A baker maintaining a stiff starter (a common choice for rye or for slower, more stable cultures) who relies on the float test will conclude their starter is dead when it's actually fine.
Rye-heavy and whole-grain starters: these have a fundamentally different gas-retention structure than a white-flour starter. Bran particles disrupt the gluten-adjacent matrix that holds bubbles in place, and the fermentation itself tends to run faster and produce a looser, more liquid consistency. The result is unpredictable float behavior — sometimes floating well before peak, sometimes barely floating at peak at all.
Old, weak, or fridge-cold starters: a starter just pulled from the refrigerator, or one that's been neglected for a stretch of infrequent feeds, may have reduced vigor without being unhealthy. It can under-perform the float test on a feed or two even as it's recovering perfectly normally. Judging it "not ready" and discarding it, rather than just feeding it again and giving it time, is a common overreaction to a test result that was never that informative to begin with.
In each of these cases, the failure isn't the starter — it's the test being applied outside the conditions it was ever reliable under. The float test was popularized around a fairly typical white-flour starter at moderate hydration, and it travels poorly outside that profile. If your starter doesn't match that profile, a sink tells you very little, and you need to lean more heavily on the visual and aroma signs than on the water glass.
None of this means the float test is useless. Treat it as one weak signal among several, and pair it with information that's actually specific to the moment you're in.
Track volume growth from a marked starting point — a rubber band on the jar right after feeding gives you a hard reference line, and tripling from that line is a far more precise signal than a spoonful in a glass. Watch the surface: a starter that's still doming is still climbing toward peak, while one that has flattened or begun to sink slightly in the center has already crossed it. And use your nose — a peaked starter smells yeasty and mildly tangy, not sharply acetic or like nail-polish remover.
A full breakdown of all six visual and tactile peak signs — volume, surface, bubble structure, aroma, texture, and sound — lives in our companion article, Sourdough Starter Peak: 6 Signs Your Levain Is Ready. If you only take one test with you from this article, take that one: it's built around signs that scale with time, not a binary that stays "yes" for hours.
Because the float test's window is so wide, the most reliable fix isn't a better spoon test — it's not needing the test to carry that much weight in the first place. S.D Timer calculates your feed time backward from your target bake time, using your ambient temperature, inoculation ratio, and flour type, so peak lands inside a much narrower window than "somewhere in the next two hours." The float test becomes a final sanity check instead of your main tool.
📅 Plan Your Next Bake with S.D TIMERThe float test tells you one thing: there is currently enough trapped gas in this sample to float. It doesn't tell you if that gas was trapped ten minutes ago or two hours ago, and it can mislead you entirely if your starter is stiff, rye-heavy, or freshly out of the fridge. Use it as a floor, not a finish line — confirm activity with it, then find true peak with volume, surface, and aroma, or with a schedule that gets you close before you ever need to test anything.