Journal

Roast notes · 2026 season

Four entries, written across the season and closed at the end of it. Whole thing on one page — there is nothing behind a link here.

Closed archive04 entries · March – August 2026
8 August 2026Entry 01 / 04

Closing the 2026 season

Six coffees, 282 published data points, four notes. That is the season.

The Huehuetenango was the hardest to get right and took eleven batches before the molasses stopped turning into something flatter. The Nyeri needed the least from us and would probably have been fine roasted by anyone. Both are on the shelf at the same price, which tells you something about how little the difficulty has to do with the cup.

We are closing this archive rather than letting it run on. Next season starts with new lots and a new page, and these four notes stay where they are.

19 June 2026Entry 02 / 04

We stopped rounding the numbers

The first version of this site published drop temperatures to the nearest degree. It looked tidier. It was also the one piece of the whole thing that was not true.

A drop at 226.1 °C and a drop at 226 °C are not the same roast, and if we are going to claim the data is the product then rounding it for the sake of a neat column is exactly the kind of small dishonesty the site exists to avoid.

Every figure on a coffee page is now the logged value. Where a number looks awkward, that is because it was awkward.

2 May 2026Entry 03 / 04

Why the Toraja is the only one we cover

Sari dries under cover and pulls the beds in every afternoon before the rain. It costs him about a third of his drying capacity and roughly two extra weeks per harvest.

We asked him once whether it was worth it. He said the valley has been drying on open concrete for forty years and everybody there knows what the rain does to a lot, and that the difference is not subtle if you have tasted both.

He is right, and it is the clearest example we have of a decision you can taste but cannot see on a label. The clove note holds because the parchment never got a soaking halfway through. Nothing about the roast produces that. It arrived in the bag.

14 March 2026Entry 04 / 04

The Kivu naturals nearly went wrong

First three batches of the Nyamasheke stalled between 168 and 174 °C and sat there for the better part of ninety seconds. On the fourth we charged three degrees lower and opened the damper a notch earlier, and the rise came back.

What we had wrong was the bean density, not the drum. A natural that dried twenty-one days on a bed carries more moisture at the centre than the paperwork suggests, and it takes the heat differently. The curve on the coffee page is batch five. Batches one to three went to the cupping table and stayed there.

That is the whole reason we publish these. If you only ever see the good one you learn nothing about the coffee.