Roasting, In Practice · Part 2 Rate of rise is a useful reading. The crash and the flick are not defects, and there is no mechanism by which they could be. Chasing a smooth curve is chasing the shape of a graph. 01 What the number actually is Rate of rise is a rate of change, not a measurement. The software takes bean temperature, subtracts the reading from a moment ago, and divides by the interval. It is a derivative of another curve, which means every error in the original is amplified in it. It depends heavily on your probe. Probe thickness, how deep it sits in the bean mass, and how much the software smooths the signal all change the shape of the line. Two machines roasting the same coffee identically will draw different rate of rise curves. What it is good for is comparison. On one machine, with one probe and one setting, it tells you whether today's batch is moving like yesterday's. That is a real and valuable use. 02 The crash is not a defect A cra...

Roasting, In Practice · Part 3 Development time ratio is not a weak metric. It is the wrong kind of number. Reaction rates depend on absolute temperature and duration, and a ratio of two durations carries neither. 01 What the number claims to describe The definition is simple. Take the time from first crack to the end of the roast, divide it by the total roast time, and express it as a percentage. Somewhere between fifteen and twenty-five percent gets quoted as the target range. The claim behind it is that this ratio tracks flavour development. Too low and the coffee is underdeveloped, too high and it is baked. The number is treated as a quality gauge that transfers between coffees, machines and batch sizes. It does not transfer, and the reason is not a lack of evidence. The problem is structural. A ratio of two durations cannot describe a process governed by temperature and time, because dividing one by the other throws both away. 02 The chemistry says some...

Roasting, In Practice The roaster who named the crash also advises using a thinner bean probe before forming strong opinions about how crashy your curves look. Both statements cannot be load-bearing at once. 01 State the claim fairly first The position is well known and clearly argued. Scott Rao observed across thousands of roasts that the best cups came from roasts whose rate of rise declined smoothly, and named the sharp drop at first crack the crash and the rebound after it the flick. The industry adopted both terms within a few years. The supporting evidence offered is sensory. He has written that most of his students can taste a roast blind and tell whether its curve crashed, and he presents that as validation of the idea. He also notes that crashed roasts tend to flick and tend to be underdeveloped, so the three often appear together. That last observation is the weak point, and it is his own. If crashing, flicking and underdevelopment usually arrive together, ...

Roasting, In Practice · Part 1 You do not control roast level. You control how much heat goes in, how much air moves through, and how long it lasts. Everything else is a consequence. 01 There are only three inputs Heat is the first. Gas pressure, burner output, element power, whatever the machine calls it. This is the energy going into the batch, and it is the only thing that raises bean temperature. Airflow is the second. How fast air moves through the drum. It carries heat to the beans, carries chaff and moisture away, and decides how much of the energy you applied actually arrives. Time is the third. How long the beans stay in there, and how that time is distributed between the stages. Time is not set directly; it is the result of the other two, and that distinction matters more than it sounds. Batch size belongs here too. Loading a 30 kg machine with 12 kg is a different machine from loading it with 28 kg. The relationship between drum mass, bean mass and...

Roasting, In Practice Not one controlled study has ever shown that an RoR crash harms a cup of coffee. Not a weak study. Not a disputed study. None at all, in more than a decade of the whole trade repeating it. 01 What is actually on the record There are observations. A consultant tasted thousands of roasts and noticed that the ones he liked tended to have smoothly declining curves. That is a real observation from a serious practitioner, and it is where the idea began. There are classroom demonstrations. Students taste roasts and say whether the curve crashed. The samples are chosen by the person who teaches the idea, the tasters know what they are being asked to find, and the results are not recorded in any form anyone else can check. There are blog posts and forum threads by the thousand. Nearly all of them cite each other. Trace any chain back far enough and it ends at the same original observation, not at a measurement. What does not exist is a controlled compar...