Chasing a buttery pale ale back to a cold fermentation

22 August 2026 · 3 min read

The pale ale I bottled in July had a flaw I could not place at first. It was clear, well carbonated, and the hops were there, but under all of it was a slickness on the tongue and a smell that I eventually recognized from the movie theater: butter, or butterscotch. That is diacetyl, and the frustrating part is that yeast produces it in every fermentation and then cleans it up again if you give it the chance. I had not given it the chance.

Here is what I did wrong. The garage was warm, so I set the fermentation chamber to 62 °F, cooler than usual for that strain, thinking a cleaner beer would come out of it. Fermentation finished slowly, and on day six I saw a stable gravity of 1.011 over two readings and crash-cooled to 36 °F for a week before bottling. Cooling the yeast off that early stopped the cleanup before it had started. A cold fermentation plus an early crash is almost a recipe for the stuff.

I confirmed it with the forced test that I had read about and never bothered with. I took two samples: one went into a covered jar in a pan of water at 140 °F for twenty minutes, the other stayed at room temperature. After both cooled I smelled them side by side. The heated one reeked of butter; the cool one only hinted at it. Heat converts the precursor into diacetyl quickly, so if the heated sample smells worse, there is more on the way and the beer is not ready.

For the rebrew in August I used the same grain bill and hops, 4.5 pounds of pale malt, 6 ounces of light crystal, and a modest late addition, and changed only the fermentation schedule:

Pitch and hold at 64 °F for the first three days, through the most active phase.
Raise the controller 2 °F a day to 68 °F and hold for three days once the gravity is within a few points of the expected finish.
Run the forced test on a sample; only crash-cool when the heated jar smells clean.

That batch went from 1.048 to 1.010, the forced test on day nine came back clean, and the bottles tasted like the beer I had meant to make the first time. The slow ramp at the end costs me three or four days per batch. Given that the alternative is a case of butterscotch, that is a bargain.