Testing and Writing a Recipe.

A recipe that works for its author may fail for a reader for reasons the author never noticed: a hotter oven, a smaller pan, a different flour, and above all different measuring. Volume measures are imprecise for solids: a cup of flour scooped and levelled can weigh 20 % more than one spooned in gently, and both differ from cook to cook (I.5). Weights remove most of this. Times are approximate too, because pans, burners and ingredients differ, so a good recipe gives cues alongside times — "until golden and fragrant", "until a skewer slides in", "until the oil separates", "until 74 °C at the centre" (XXIX.4, XVIII.2).

Testing is how a recipe becomes reliable. The author cooks it several times from the written text, exactly as written, noting every place they had to guess; then someone else cooks it, ideally in a different kitchen, while the author watches and does not help. Each round corrects the text — a missing step, an unclear size, a time that was too short. The final recipe lists ingredients in the order used, by weight (with volume as a courtesy), names sizes and specifications (XXXI.3), gives temperatures and cues, and states yield and timing honestly. The workshop measures flour ten ways.

Below: the flour ten cooks would put in for "500 g" of flour measured by the method you choose (deterministic spread at the method's typical variation, from published measuring studies), and the hydration of a 325 g-water bread dough each would get, against the ±2-point band where the dough behaves as intended.