How much weight do coffee beans lose when roasting

Why beans get lighter in the roaster

Most of the weight that disappears is water. Roasting consultant Rob Hoos notes that specialty green coffee starts at 8 to 12 percent moisture, while roasted coffee ends below 3 percent. A 2012 study of Arabica and Robusta by Bicho and colleagues in the journal Ciência e Tecnologia de Alimentos measured green beans at about 9.1 to 9.2 percent moisture and roasted beans at 1.1 to 3.1 percent, depending on how far they were roasted.

The rest is lost as the bean's own material breaks down. Hoos quotes the coffee science literature (Clarke and Macrae, and Illy and Viani) to the effect that some 5 to 8 percent of the weight at a medium roast is dry matter, lost mainly as carbon dioxide, and that the chemical reactions of roasting create more water, which also escapes as steam. Illy and Viani put the split of those breakdown products at about 70 percent water and 30 percent carbon dioxide. Bicho's team adds a third source you can see: the silver skin that comes off as chaff.

While the beans get lighter, they also get bigger. Bicho's study found that apparent density fell steadily as roasting went on, and at the darkest roast the roasted Arabica had less than half the density of the green beans. That is why a roasted batch fills a larger jar than the green coffee it came from, even though it weighs less.

Typical weight loss by roast level

Published ranges overlap but do not match exactly. Tanaka Coffee Lab says most roasts lose 12 to 22 percent. Clarke and Macrae, as quoted by Hoos, give 14 to 20 percent. Bicho and colleagues cite a usual range of 14 to 23 percent depending on the coffee's origin, its moisture, how it was stored and the roasting method, while their own samples lost 10.5 to 19.4 percent for Arabica and 10.1 to 16.7 percent for Robusta as the roast got darker.

Hoos gives working bands from his own roasting. Below 11 percent he suspects the coffee is under-roasted. From 11 to 13 percent is, almost without exception, light roast. From 14 to 16 percent is medium, 17 to 18 percent dark, and 19 to 21 percent extremely dark, and above 22 percent he advises avoiding it altogether. He is clear that other roasters draw the lines between light, medium and dark differently.

Tanaka Coffee Lab expresses the same idea as a roast index, the green weight divided by the roasted weight. It places Light and Cinnamon at 1.10 to 1.15, Medium and High at 1.15 to 1.20, City and Full City at 1.20 to 1.25, and French and Italian at 1.25 to 1.30 and above. An index of 1.20 is a weight loss of about 16.7 percent, and 1.25 is exactly 20 percent, which lines up with Hoos's medium to dark bands.

How to calculate it

The formula Hoos gives is: (green weight minus roasted weight) divided by green weight, times 100. If you charge 250 g of green coffee and drop 213 g, you lost 37 g, and 37 divided by 250 is 14.8 percent. The same batch has a yield of 85.2 percent (213 divided by 250), which is simply 100 minus the loss, and a roast index of about 1.17.

Weigh the roasted beans the same way every time. Tanaka Coffee Lab recommends weighing after the beans have cooled. Chaff is part of what the roast removes, so decide whether any stray chaff goes on the scale and stick to it.

Scale resolution matters more for small batches. On a 100 g popcorn popper batch, a 1 g reading error shifts the result by a full percentage point, which can be the difference between two of Hoos's bands. A scale that reads to 0.1 g makes small-batch numbers worth comparing.

The formula also works backwards. Hoos divides the roasted weight you need by one minus the loss: if a coffee usually loses 16 percent and you want 200 g roasted, you need 200 divided by 0.84, or about 238 g of green coffee.

Using weight loss to check consistency

The most useful thing about weight loss is comparing a batch against your own earlier batches of the same coffee. Hoos writes that the same coffee roasted the same way should lose the same weight, and that a result 0.5 percent or more away from the baseline means something is seriously wrong. In his quality control, batches that far out were also off in whole-bean or ground color.

Moisture shifts the baseline. Hoos gives the example of an Ethiopian coffee at 10 percent moisture that loses 12 percent; a bag of the same coffee at 9.5 percent moisture, roasted the same way, should lose about 11.5 percent. Comparisons are most meaningful within one bag or lot, and a new lot of the same origin needs a fresh baseline.

Weight loss also checks your roast profile against your palate. Hoos suggests that if a profile you think of as light ends with a weight loss that looks medium or dark, something is out of balance, and it is worth comparing the numbers with your first crack and drop times.

Common weight loss mistakes

Weighing hot or half-cleaned beans. Tanaka Coffee Lab recommends weighing after the beans have cooled, and a varying amount of chaff on the scale changes the result. Cool the beans and weigh them the same way each time.

Comparing across different coffees. A wetter green coffee and a drier one can lose different amounts at the same roast level, as Hoos's moisture example shows. Use weight loss to compare a coffee with itself, not one coffee with another.

Treating the bands as rules. Hoos presents his ranges as his own guide, and the published ranges differ by several points. The number describes what happened; how the coffee tastes decides whether the roast was right.

Not recording both weights every time. Hoos's advice is to weigh what goes in and what comes out on every single batch. CrackLog asks for the charge and drop weight when you drop the beans and computes the yield for you, next to the curve with both crack markers. The roast level stays your own call, picked from Light through Italian.

Frequently asked questions