How to Make a Sourdough Starter from Scratch, Day by Day

Day 1: mix

Equal weights of flour and water. Whole-grain rye or whole wheat wakes faster than white. Bread flour still works. Stir to a thick paste. Scrape the sides. Cover loosely. A sealed jar under pressure is how you make a mess, not a culture.

Leave it at room temperature. A cold kitchen is a slower kitchen. You are not on a clock yet.

Mark the level. You will want that mark tomorrow, and you will want it again on the day nothing happens.

Day 2: usually quiet

A few bubbles. A smell like wet flour. Or nothing. Nothing on day 2 is normal. Do not dump it. Do not add yeast. Stir, discard about half if it has separated, and feed equal flour and water again.

You are diluting what is there and giving it fresh food. That is the whole method.

Day 3: the false rise

This is the day nobody warns about. The jar can double, or more, and then collapse. It smells off. Nail polish, old fruit, something sharp. That rise is usually bacteria, not the yeast that will leaven a loaf. It is a false start.

Bakers throw the jar out here. Don't. The rise falls. The smell fades if you keep feeding. Dumping a day-3 explosion and starting over is how a two-week project becomes a month.

Discard down to a small amount. Feed again. The false rise does not mean it is ready to bake.

Day 4 and day 5: the stall

After the false rise, the jar often goes quiet. Fewer bubbles. Little or no growth. This is the stall. It looks like failure. It is the handoff. The early bacteria have been diluted. The yeast and the lactic acid bacteria have not taken over yet.

Keep the same move. Discard. Feed. Cover. Room temperature. A stalled jar on day 5 is still in the arc. A stalled jar you “help” with commercial yeast is no longer a sourdough starter.

If the top dries out, the cover is too loose or the room is very dry. Scrape the dry cap off. Feed the paste underneath. Mold is the exception: fuzzy spots in color, and the jar goes out. Bubbles and a gray liquid are not mold. Gray liquid is hooch. Stir it in or pour it off, then feed.

Day 6 to day 14: it starts to repeat

Growth comes back. Slower than the day-3 spike, and cleaner. The smell moves from sharp to yogurt, flour, a little alcohol. The rise starts to happen on a rhythm you can see against the mark.

Feed on that rhythm. Once it is rising reliably, move to the house ratio. A 1:5:5 feeding is 1 part starter, 5 parts flour, 5 parts water. That is the build the rest of the method assumes. A 1:10:10 at 18°C is a 32-hour culture, not a levain you mix the same evening. Do not use a huge ratio as the test for a young jar. The young jar cannot eat that much food on time.

Peak looks the same as it will for the life of the jar. Doubled or a little more. Domed on top. Webbed inside. Just before it falls.

When it can actually leaven bread

Ready is not the first day it bubbles. Ready is two or three feeds in a row where a 1:5:5 doubles in a predictable window and smells clean. Then it can raise a loaf.

Before that, any bread you mix is a guess. The inoculation is weak. Bulk runs long. The crumb comes out dense and you blame the formula.

Use it at peak, not after it has fallen. Mix before the peak and bulk stretches. Mix after the collapse and bulk compresses. Acid already spent part of the window in the jar.

True hydration on the loaf counts the flour and water inside that starter. A new baker's first dense loaf is often a hydration error, not a dead jar.

Keep a small jar. Feed what you will bake with. Daily feeding is for a culture you are using. A starter in the fridge, fed when you bake, is enough after it is established. The from-scratch phase is the only stretch that wants a feed every day.

Key numbers

• Start: equal flour and water, by weight.
• Day 3: false rise. Not yeast. Do not bake it. Do not dump it.
• Day 5: stall is normal.
• House feeding once it is alive: 1:5:5.
• A 1:10:10 at 18°C is a 32-hour culture, not a same-day levain.
• Ready: doubles on a repeated 1:5:5, clean smell, used at peak.