Yes, it does, and that is the part the method's own name hides. Honey processing is built around skipping the submerged fermentation tank that defines a washed coffee. But the days a depulped, mucilage-coated seed spends on a raised bed before it is dry turn out to be a managed handoff from one population of microbes to another, not a quiet dry-out, and a 2024 study that tracked exactly that handoff, day by day, found a bacterial population rising to dominate the bed within three days and then losing ground to a yeast population that kept climbing for the rest of the drying window it measured.
Nobody doing the honey-process write-up for a bag of coffee says that, including us. Our own page on the method calls the result of a long, cool dry "deepening fermentation character" in one clause and says "there is no fermentation tank stage" a few sentences earlier, which is true and also exactly the kind of sentence that makes a careful reader stop and ask which one we mean.
Why does "no fermentation tank" sound like no fermentation at all?
Because a tank is the only fermentation most coffee writing ever shows a picture of, and some of the trade's own reference writing says so outright. Sweet Maria's glossary puts it more starkly than we do: "fermentation is bypassed altogether and the fruit-covered coffee beans are sent straight to dry." Cafe Imports' own processing writeup describes the same mechanical sequence in gentler words — the skin stripped off, the mucilage left intact, the beans sent straight to the bed. Our own honey-process page describes it correctly too: the cherry skin comes off, some or all of the sticky mucilage stays on the seed, and it dries that way on a raised bed or patio instead of going into a water-filled tank first. All three are right about the mechanics, and none of the three says what is actually alive on that bed while the moisture leaves, because none of them set out to measure it. We have not either, until now.
The honey-grade spectrum itself, white through black by how much mucilage stays and how slowly the bed runs, is already on our page and I am not repeating it here. What neither that page nor the trade's general description explains is the mechanism underneath the color names: what is actually consuming that mucilage while the bed dries, and whether it behaves anything like fermentation at all.
Which organisms actually take over while a honey lot dries?
A team of researchers at Yunnan Agricultural University in Kunming, China, led by Xiaojing Shen, set out to answer exactly that question. Their paper, published in the journal Fermentation in August 2024, studied what they call semi-dry fermentation processing: ripe cherries are floated to sort them, de-pulped, and sun-dried, a method the paper itself describes as "a hybrid of wet and dry fermentation processing." De-pulping is the mechanical step that strips the skin and most of the pulp but does not wash the mucilage off the seed underneath it, which is the same move Sweet Maria's and Cafe Imports describe above for a honey lot. Shen's paper never uses the word "honey," but the mechanics match. Her team sun-dried Coffea arabica at temperatures of 10 to 18 degrees Celsius and sampled the microbial community four times, three replicates each: day 0, day 3, day 6 and day 9.
The succession has a clear shape. Bacteria move in first: the phylum Proteobacteria ran from 42.01 percent of the whole community at day 0 to 98.77 percent by day 3, and within it a single genus, Tatumella, went from 7.48 percent to 92.17 percent over that same stretch. From there the balance shifts toward fungi. The phylum Ascomycota reached as high as 84.23 percent of the community, with that peak also landing around day 3, and inside it a yeast genus, Hanseniaspora, gradually increased across the whole nine days, from 2.50 percent on day 0 to 21.50 percent by day 9 — still climbing when the study's own sampling stopped. Bacteria arrive first and peak early; yeast builds more slowly and has not finished building when the measurement ends.
Chemically, the bed is not static either, and it does not move in lockstep with the microbes. Shen's team screened 1,551 non-volatile compounds in the drying beans. Comparing day 3 against day 0, 29 changed significantly; day 6 against day 3, 57 changed, the single largest swing of any stretch measured; day 9 against day 6, only 9 changed, nearly all of them down. Measured start to finish, day 9 against day 0, 117 compounds had moved significantly: 32 up, 85 down. Three chemical families carried most of that movement — lipids, organic acids and their derivatives, and phenylpropanoids and polyketides, a large plant-compound family that includes many of the flavonoids and phenolic acids a coffee seed carries. The biggest single increases included zofenopril, 7-ethoxycoumarin and xanthylic acid; the biggest decreases included amygdalin and kaempferol-3-O-glucoside. None of this chemistry was cupped. Shen's paper measured compounds, not flavor, and does not claim otherwise. But lipids carry body and mouthfeel in a brewed cup, and organic acids carry brightness and acidity, and a nine-day dry that moves a measurable share of both families in one direction is at minimum a plausible chemical reason a honey-process cup reads differently from a washed one. Plausible, not proven — nobody tasted these specific beans.
That is a fermentation by any definition that does not require a tank: a population succession consuming substrate and leaving a measurably different chemical profile behind. It happens in the open air, on a bed, with no water standing anywhere near it.
Is this the same succession that happens to a natural, or a washed coffee?
No, and the difference is worth being precise about, because we have published the other two and I do not want this piece to blur into either of them. A washed coffee submerges whole, depulped seeds in a water-filled tank, and our own piece on mucilage covers what actually happens there: Sylvie Avallone and colleagues found the mucilage's structural polysaccharides showing, in their words, "no apparent degradation" through a fermentation run, in a study published in the Journal of Agricultural and Food Chemistry in 2001. A year later the same group concluded the breakdown is "correlated to acidification and not to enzymatic pectolysis." A tank does not digest mucilage with enzymes so much as it sours its way through it.
A natural is a different process again, and a bigger one: the whole cherry, skin and pulp and all, dries intact around the seed, usually for two to three weeks rather than nine days. We have written at length about what happens on that kind of bed, using Ethiopia Buku Abel as the working example: acetic acid bacteria and yeasts dominate, not the lactic acid bacteria a wet tank favors, and the succession there is Enterobacteria climbing in the first 24 hours before yeasts take over for most of the two- to three-week dry.
Honey, or semi-dry, processing sits structurally between those two and resembles neither one exactly. The skin comes off like a washed coffee's does, so there is no fruit pulp buffering the seed the way a natural has. But the mucilage stays on and the whole thing dries in open air rather than underwater, so the organisms doing the work are not the lactic acid bacteria of a tank. Shen's team found a different bacterial genus altogether, Tatumella, leading the way before yeast takes over, on a shorter clock than a natural's three weeks. Three related processes, three different microbial casts, and the method's name tells you almost nothing about which one you are getting until somebody actually looks.
Does this explain why a black honey tastes different from a white honey?
It explains the clock, not the whole color chart, and I want to be exact about which part is Shen's finding and which part is mine. Shen's team studied one farm's semi-dry lot over nine days. They did not compare honey grades, did not vary how much mucilage was left on the seed, and say nothing in their paper about white, yellow, red or black honey specifically. That comparison is not in the study. What is in the study is a timed population curve: a bacterial population that peaks early, around day 3, and a yeast population that is still climbing when the measurement stops at day 9.
Put that curve next to what our own page already says about the grades, and the two facts line up in an obvious way even though no single source states them together. Our page puts white and yellow honeys at having "most of the mucilage mechanically removed, on the order of half or more," and says they are "dried quickly in open sun." Red honey "retains most of the mucilage, roughly three quarters by common definition," and "often dries more slowly or under partial shade." Black honey "keeps essentially all of the mucilage and dries slowest of all, frequently under cover." If Shen's curve is anything close to typical for a bed carrying retained mucilage, a fast white-honey dry is cut off while the bacteria are still in charge and before the yeast has done much of anything, while a slow black-honey dry runs well past the point where yeast has taken over and kept working. That would track with the cup: white and yellow honeys read closer to washed, clean and comparatively restrained, while reds and especially blacks read winey and more deeply fermented, which is exactly what a longer yeast-dominated stretch would produce. I think that is a reasonable connection to draw. I also have not seen anyone run the actual experiment, grade by grade, with the microbiology tracked alongside the cupping, and until somebody does this is an inference sitting next to a measurement, not a single proven fact.
Shen's own paper ends by asking for almost exactly that experiment, in its own words. It suggests that a starter culture could be used to steer a semi-dry ferment deliberately instead of leaving the outcome to whatever microbes happen to be on the cherry and however long the weather lets the bed run, and it closes by saying plainly that "further research is needed on the controlled fermentation of coffee." A grade-by-grade honey trial, cupped against a tracked microbial curve the way this one tracked a single lot, is the specific shape of that further research, and it has not been run yet, by us or by anyone else.
There is a cost to managing a population rather than running a quiet dry, and our own page says so without connecting it to any of this: the beans have to be turned by hand through the dry specifically to head off "clumping, mold, and uncontrolled fermentation," and that is exactly what a bed risks if nobody turns it while Tatumella and then Hanseniaspora are working through it. A natural lot and a washed lot each carry their own version of that labor. A honey lot's version is watching and turning a bed that is, by the measurements above, doing something biologically active for most of nine days, not sitting still.
Do we roast a honey-processed coffee at Inheritance?
Not right now, and I have given a worse answer to that question than I should have. For a while my standard line to a customer asking about honey process was that it sits "in between" washed and natural, which is true and also a way of not actually answering. It let me skip past what I had not looked into, which was what is biologically doing the "in between" work. I do not have that excuse anymore.
The honest state of our shelf is this: the honey-processed lots we have carried are archived, not buyable, and I am not going to point you at a product page that cannot ship you the coffee. We have carried Elías & Shady Bayter, a Colombian pink bourbon honey that read hazelnut, blueberry and orange, and Rwanda Kibirizi, a honey lot that read honeycomb, lemon and chocolate cake. We even carried one lot, Java Puncak Sirna, whose own label used the word "fermentation" outright: a five-day sealed anaerobic ferment before the honey-dry, reading rum cake, raisin and clove. All three are gone. What we do have, if you want to taste what a longer, more open fermentation does to a cup without the mucilage still attached, is Ethiopia Hambela Buku, a natural that cups 92.3 and reads ripe strawberry, mango and melon, with a honey-like sweetness and a bright citrus finish. It went through the longer whole-cherry version of this same kind of open-air succession, not the shorter semi-dry one Shen's team measured, and the fruit character on that bag is the natural's own version of the same underlying idea: microbes working a bed for days at a stretch, with the length of the dry setting how far they get to go.
Inheritance Coffee, Pensacola. Read more on how we think about processing on our honey-process page, or taste the longer open-air fermentation for yourself in Ethiopia Hambela Buku, cupping 92.3.
