Every roast ends with one irreversible action: the roaster opens the drop door and the batch falls into the cooling tray. The bean-probe reading at that instant is the drop temperature — also called end temperature, finish temperature or, on some control systems, simply drop. It is the most-recorded number in roasting, the one written on every batch sheet and the one most likely to be quoted between roasters. It is also the most widely misunderstood.
The misunderstanding is simple to state. Drop temperature is a timestamp, not a measurement of the coffee. It tells you where your probe was when you pulled the trigger. It does not, on its own, tell you how dark the coffee is, how developed it is, or how it will taste. Two roasters can drop at the same displayed number and produce coffees that look and taste nothing alike — and the same roaster can produce two identical-tasting coffees at drop readings a full ten degrees apart.
What drop temperature actually is
A bean-temperature probe does not measure beans. It measures itself. A thermocouple or RTD reports the temperature at and near its own tip, which is a blend of the bean mass tumbling against it, the air moving past it, and radiant heat from the drum wall. When you record a drop temperature, you are recording the equilibrium state of a metal probe sitting in a specific spot inside a specific machine.
That is not a criticism. A probe reading is an excellent relative instrument. Within one machine, with one probe, one bean mass and one airflow regime, the drop reading is highly repeatable and tracks real changes in the coffee closely. The problem only appears when the number is treated as absolute — as a property of the coffee rather than a property of the measurement system.
The three things drop temperature genuinely controls
- The end of energy input. Once the beans leave the drum, the exothermic pyrolysis reactions slow sharply and then stop as the mass cools. Drop temperature sets the ceiling on how far those reactions get to run.
- The total thermal load, in combination with time. Heat applied is an integral, not a point. Drop temperature is the final value of that integral, which is why it correlates with roast degree even though it does not define it.
- The starting condition for cooling. A batch dropped hotter has more residual energy to shed, which is why how quickly you cool the batch matters more at higher drop readings. Slow cooling from a high drop can add meaningful development after the coffee has technically left the roaster.
What it does not determine
Drop temperature does not determine color. It does not determine development. It does not determine roast level. All three are outcomes of the entire heat history — how fast you got there, how long you spent in each phase, how much moisture left when — with the drop reading as the final punctuation mark rather than the sentence.
A useful reframe: drop temperature is when you stopped, not what you made. Two roasts that arrive at the same endpoint by different routes are different coffees.
Why drop temperature is a poor proxy for roast level
Roast level is fundamentally a description of how far the bean's chemistry has progressed — measured most objectively by ground and whole-bean color, which is what Agtron and equivalent color scales exist to quantify. Drop temperature is a proxy for that, and a leaky one. Consider two batches of the same coffee on the same machine:
- Batch A races to the endpoint with a steep, unchecked rate of climb and drops at 208 °C (roughly 406 °F) about forty seconds after first crack.
- Batch B approaches slowly on a gentle decline and drops at the same 208 °C, but two minutes and twenty seconds after first crack.
Same drop number. Batch B will be visibly darker, structurally more fragile, lower in acidity and — if the approach was too flat — likely to taste baked and papery. The identical endpoint reading conceals two entirely different roasts. This is why roast level as a category should be verified by color and cupping, with drop temperature used as the control lever that gets you there rather than as the definition of where you arrived.
How drop temperature interacts with development and rate of rise
The endpoint only becomes meaningful when read alongside two companions: how long the coffee spent after first crack, and how fast it was climbing on the way in.
Development time and development time ratio describe the post-crack window. Drop temperature and development time are coupled — you cannot change one without changing the other unless you also change the heat applied. Raising your drop reading while holding heat constant lengthens development; lowering it shortens development. Roasters who chase a target endpoint without watching the clock frequently drift into longer, flatter roasts as their machine heats up across a production day.
The approach matters just as much. Rate of rise — how many degrees per minute the probe is gaining — determines the character of the final stretch. A batch arriving at the endpoint at a brisk climb is still actively developing when it drops; one crawling in at near-zero climb has effectively been coasting, and pushing it a few more degrees adds color without adding much sweetness. The same endpoint reached at a high climb rate and at a stalled climb rate produce very different cups.
All of this sits on a foundation set right at the start. The turning point — where the probe stops falling after charge and begins to climb — establishes the thermal baseline for everything downstream. A batch that turns high and early arrives at any given drop reading sooner and less developed than one that turns low and late.
Why one roaster's numbers do not transfer
This is the single most practical thing to understand about drop temperature: it is machine-specific, and often probe-specific within the same machine.
Probe diameter and thermal lag
A thin probe has low thermal mass and responds quickly. A thick, heavily sheathed probe has high thermal inertia and lags badly. Published comparisons of the sheath diameters commonly fitted to drum roasters — roughly 1.6 mm, 3.2 mm and 6.4 mm — consistently report that the thinnest probe turns first and then reads higher for the rest of the roast, with one widely cited comparison putting the heaviest probe around 13 °F (roughly 7 °C) below the thinnest at the endpoint of the same batch. That lag propagates all the way to the drop: a slow probe under-reads a rising curve, so it shows a lower number at the moment the coffee is genuinely ready.
Placement and immersion
A probe fully immersed in the bean mass reads closer to true bean temperature than one sitting partly in the airstream. Placement low in the drum, on the side the beans are thrown toward, gives the most bean-weighted signal; a probe mounted in the drop door or high in the chamber reads a heavier blend of air. A rule of thumb often quoted is to let the probe extend into the chamber by at least ten times its own diameter, so the sensing tip sits in beans rather than close to the cooler mounting boss. Batch size changes immersion too — a half-full drum may leave a probe intermittently exposed, shifting readings by several degrees for reasons that have nothing to do with the coffee.
Calibration and what you can actually correct
RTDs are broadly linear and can often be field-calibrated to remove offset. Thermocouples generally cannot be trimmed the same way. Either way, calibrating a probe against a reference bath corrects sensor error; it does not correct placement error or lag error, which are properties of the installation rather than the sensor. Two correctly calibrated probes in different positions will still disagree, legitimately, throughout the roast.
| Variable | Effect on the displayed drop number | Effect on the actual coffee |
|---|---|---|
| Thicker probe sheath | Reads lower on a rising curve; smooths the trace | None — measurement artifact only |
| Probe higher in the chamber | Reads higher (more air-weighted) | None — measurement artifact only |
| Smaller batch size | Often reads higher and noisier | Real: faster heat transfer per bean |
| Longer post-crack time | Reads higher for the same heat setting | Real: darker, more developed, less acidity |
| Slower cooling after drop | No effect on the recorded number | Real: added development after the drop |
Typical ranges by intended roast level
The figures below are widely cited orientation ranges for drum roasters with a bean-immersed probe. Treat them as a starting bracket to calibrate against your own machine, never as targets to copy.
| Intended level | Commonly cited drop range | What is usually happening |
|---|---|---|
| Light / filter | roughly 198–206 °C (about 388–403 °F) | Well past first crack, no oil, high acidity retained |
| Medium | roughly 206–215 °C (about 403–419 °F) | Sweetness building, acidity softening, surface still dry |
| Medium-dark | roughly 215–222 °C (about 419–432 °F) | Approaching or entering the second crack window |
| Dark | roughly 222–232 °C (about 432–450 °F) and above | Oil migrating to the surface, origin character receding |
First crack is generally cited around 196 °C and second crack around 224 °C on a typical bean probe, but both are ranges rather than fixed points and shift with density, moisture and machine. If you are roasting into or past the second crack window, the margin for error narrows sharply — the exothermic surge there can carry a batch several degrees further than intended in the seconds it takes to open the door.
Reading the drop without the number
Experienced roasters cross-check the display against their senses, and when the two disagree, the senses usually win. Pull the tryer and look for:
- Color and evenness. Compare against a physical reference set of roasted samples in consistent light, not memory.
- Surface texture. The centre crease closing and the wrinkled surface smoothing out are reliable progression markers; a first sheen of oil marks a distinct threshold.
- Sound. First crack is a loud, wet snap; second crack is a finer, drier, faster crackle. The density and tailing-off of crack activity says more about phase position than any single reading.
- Smell. The shift from bread and toast to unmistakable coffee happens around first crack; a sharp, acrid or smoky note late in the roast is a signal to move.
- Smoke and chaff. Rising smoke volume and darkening smoke color track the endpoint closely, particularly in the last minute.
Using drop temperature well
The productive way to use the number is as a within-machine consistency tool. Fix your probe, fix your batch size, fix your airflow regime, then log drop temperature alongside development time, the climb rate at drop, and a color measurement. Adjust one variable at a time and cup the results. Over a few dozen batches the endpoint reading becomes a genuinely powerful lever — one you can trust to a degree or better on your own equipment.
What it will never be is portable. When another roaster tells you their drop number, the useful follow-up questions are about probe type, probe position, batch size relative to drum capacity, development time and measured color. Without those, the number is a coordinate on a map you do not have.