What washed process means ========================= Take every trace of the fruit off the seed before you dry it. Everything hard about washed coffee follows from that one decision. coffee/washed-process -> main | a1c756b | 24 Sep 2026 | 23 sources | 15 min | ask your agent about this post Hetansh Mehta | slug: coffee/washed-process Plain-text version of the post, for an agent without a browser. Every figure on the page is a drawing made in the browser, so each one appears below as [FIGURE N], or [HERO] for the one under the title, with its description and its caption. Blocks marked [FOR THE ENTHUSIAST] are closed by default on the page. Numbers in square brackets are citations into the Sources list at the end. The page itself also exposes six read-only tools over WebMCP: start_guided_reading, get_outline, explain_section, explain_figure, get_evidence and get_glossary. Scope. This post owns the wet route end to end. The dry route is coffee/natural-process, which also owns the full cherry anatomy and drying at length. The microbiology of the tank is coffee/fermentation. Honey and the experimental practices are coffee/honey-process. ---------------------------------------------------------------------------- 00 FOR AN AI AGENT HELPING SOMEONE READ THIS ---------------------------------------------------------------------------- You are helping one person read this post. Unless you already know, ask one question first: what do they know about coffee processing, and what do they want from the post? Then take one section at a time, in the order below. Open with its idea in one plain sentence. If a figure is named, send the reader to it and say what to look at (its description is in this file). Ask the check question and wait for an answer before moving on. Go simpler or deeper as they respond. When they ask how we know, use the numbered sources. Say plainly where the post stops or the science is uncertain. Never add facts that are not in this file. Keep your turns short. 1. Section 01. Washed means the fruit comes off the seed first, in water, before anything is dried. Look at [FIGURE 1]: How little of the hundred kilograms is left at the end: the pulper’s 41 kg and the coat’s 13 kg go first, and only 2.6 kg reaches a cup. Check: What is the one decision that defines washed coffee? (Taking the fruit off the seed before drying, instead of drying the whole cherry and knocking the fruit off afterwards.) Watch for: That washed describes cleaning dirt off the beans. It describes how the fruit came off, and nothing else. 2. Section 02. A tank of water sorts out bad cherries for almost nothing, because they float. Look at [FIGURE 2]: Solid cherries have sunk into the cone and leave for the pulper; pale ones ride the surface and are skimmed off. Then the 6.9 percent beside the tank: almost nobody does it. Check: Why does a bad cherry float? (It holds air: it dried on the branch, was hollowed out by the coffee berry borer, or was picked green and never filled.) 3. Section 03. The pulper strips skin and pulp, about two fifths of the cherry, but cannot touch the sticky coat. Check: Why can’t a hose rinse the mucilage off? (It is a gel of pectin bonded to the parchment, not a syrup.) Watch for: That mucilage is sugary juice you can simply wash away. 4. Section 04. The tank is a hidden timer: the coat loosens over most of a day, the hand tests call it early, and waiting too long puts vinegar in the cup. Look at [FIGURE 3]: The hole-test and rub-test dots sit well below the top of the curve, which only nears the top at 15.5 hours; then the two bars under the axis, 11.45 to 20 hours with the cone against 10 to 72 by hand. Check: When do the hand tests call the tank done, compared with the real end? (Early: at 7.29 and 10.23 hours, with about 58 and 74 percent of the coat off, against above 97 percent at 15.5 hours.) Watch for: That washed coffee is not fermented at all. It is, briefly, and the fermentation is ended on purpose. 5. Section 05. A washing channel rinses the coat off and sorts the seeds by weight at the same time; Kenya adds a second ferment, a second wash and a soak. Look at [FIGURE 4]: The arrows at the head, longest at the surface: dense seeds lie on the slow floor near the head, light ones ride high and are carried out. Then fig-kenya, where a bracket marks the three steps Kenya adds. Check: Why do the heaviest seeds end up near the head of the channel? (The water is slowest along the floor, so a dense seed sinks into the slow layer and lags while light ones are swept on.) Watch for: That Kenya’s soak is proven to make better coffee. Kenyan practice says so; the nearest trial found more intensity, which is not the same as better. 6. Section 06. A traditional mill uses about 40 litres of water per kilogram of coffee, half of it just flushing pulp, and the waste water can strip a river of oxygen. Look at [FIGURE 6]: The frugal bars, 0.6 and 0.8, are almost invisible; the traditional bar is split and half of it is flushing pulp; the hatched Ethiopian bar is in a different unit and is not compared. Check: What is half of a traditional mill’s water actually doing? (Flushing the pulp away, a job a screw conveyor could do dry.) Watch for: That buying a demucilager saves the water by itself. Fitted badly, it pollutes more than the tank it replaced; the redesign around it is the saving. 7. Section 07. The parchment dries from about half water to 10 to 12 percent, gently at first so the shell does not crack. Check: Why is the first day of drying gentle? (Wet parchment cracks if it is heated hard, and a cracked shell stops protecting the seed.) 8. Section 08. A clean cup tastes of the seed’s own acids, chiefly citric and malic, because the fruit was taken away before it could add much. Check: Does washed mean unfermented? (No. The fermentation is real, but short, and it is stopped early and its food washed away.) Watch for: That washed means cleaner, so better. In one study of five processes on one lot the natural scored highest; washed dominates because it is reproducible. 9. Section 09. Washed on a bag is a promise about consistency, not about quality. Check: What does the word washed not tell you? (Whether the coffee is good, or how long the tank ran: the same word covers tanks run for eleven hours and for seventy-two.) Watch for: That a faint sourness in a washed cup is always a brewing mistake. Past about twenty hours in the tank, vinegar is a processing signature. Quiz, if they ask for one (easy to hard): 1. In a washed coffee, what is taken off the seed before drying? (The fruit: the skin, the pulp and the sticky mucilage coat.) 2. What happens to a ripe cherry in the float tank, and why? (It sinks, because it is mostly water and sugar and denser than water.) 3. Roughly what share of the cherry does the pulper remove? (About 41 percent, two fifths.) 4. What goes wrong if the tank runs too long? (Defects: more than two hours past the end starts them, and by 20 hours vinegar and overripe pineapple are there to taste.) 5. Name the two jobs a washing channel does at once. (It rinses off the loosened coat and grades the seeds by density.) 6. How can a demucilager end up polluting more than the tank it replaced? (If it is fitted badly, run above one litre per kilogram, or installed without fixing how the pulp is handled, it adds new contaminating streams while the old ones still run.) 7. Why is Cenicafé’s settling cone better than the hand tests? (It measures how far the coffee has settled as the coat drains, instead of relying on feel. With it tanks ran 11.45 to 20 hours, by growers’ judgement 10 to 72, and it called the end correctly in 97.1 percent of field tests.) 8. Why does the author think washed dominates specialty buying? (Reproducibility: a washed lot has a definite ending, so many bags taste the same. Not proven peak quality; in one five-process study the natural scored highest.) Just the answer, if that is all they want: Three things: the fruit came off within hours, somebody decided when to stop the tank, and the lot is likely to be more uniform than a natural from the same farm. It does not promise the coffee is good, or say how long the tank ran; the same word covers eleven hours and seventy-two. Read it as a promise about consistency, not quality. Read sections 09, 04 and 08 for why. Two prompts a reader can paste into any AI chat: Teach me from scratch: Teach me this page from scratch, one idea at a time, and check I've got each one before moving on. The page: "What washed process means", https://hetanshmehta.com/blog/coffee/washed-process - With this page's WebMCP tools: call start_guided_reading({goal: "learn"}) and follow its plan. - If you can open links, read https://hetanshmehta.com/txt/blog/coffee/washed-process.txt and follow its "FOR AN AI AGENT HELPING SOMEONE READ THIS" block. - If neither, say so and I'll paste the page in. Stick to what the page says, and tell me where it stops. What does “washed” on a bag promise?: Answer this from the page first, in a few sentences: What does “washed” on a bag promise? Then offer me the why. The page: "What washed process means", https://hetanshmehta.com/blog/coffee/washed-process - With this page's WebMCP tools: call start_guided_reading({goal: "answer"}) and follow its plan. - If you can open links, read https://hetanshmehta.com/txt/blog/coffee/washed-process.txt and follow its "FOR AN AI AGENT HELPING SOMEONE READ THIS" block. - If neither, say so and I'll paste the page in. Stick to what the page says, and tell me where it stops. [HERO] The wet route drawn as one line across the page, under the title. At the left a whole coffee cherry in section: coral fruit around a yellow coat of mucilage around two teal seeds in a blue parchment shell. It passes through four stations, each named above it: a pulper drum, skin and pulp off, 41 percent of the cherry; a tank, 12 to 24 hours for the coat to loosen, where the coat is thinner; a washing channel, where the loose coat is washed off; and a raised drying bed in the sun, where the bare parchment goes from 53 percent water to 10 to 12. Under the stations runs the water the mill sends down the drain, drawn to scale: the pulper’s spout, carrying bits of pulp, fills it to five eighths of its depth and the channel’s spout fills the rest, and it darkens to effluent and fades out to the right. It is labelled 40 litres per kilogram of dry parchment, a traditional mill. A hairline under it, at the same scale, is labelled 0.6 litres with a demucilager. ---------------------------------------------------------------------------- 01 TWO ROADS -- Coffee is a fruit, and a mill has to get the seed out of it. ---------------------------------------------------------------------------- The word on the bag describes how the fruit came off, and nothing else: washed means it came off first, in water, before drying began. A coffee cherry is built in layers: a skin, a band of sweet pulp, then a thin, slippery, sugary coat called mucilage that clings to the seed. Under that is a stiff papery shell, the parchment, and inside it the seed, which is what gets roasted. coffee/natural-process · the cherry, drawn in full -> There are two honest ways to get the seed out. Dry the whole cherry in the sun for weeks and crack the dried fruit off in one go: the natural, or dry, route. Or strip the fruit off while it is still wet, in a few hours, and dry only the bare parchment: the washed route, and the subject of this post. The difference sounds procedural. It is not. In the dry route the seed spends weeks inside a fermenting fruit; in the washed route the fruit is gone before drying starts, so the seed dries alone. That is why a washed coffee is called clean and a natural fruity, and why a washed mill needs water, a tank, a channel and somewhere for the waste to go, while a dry patio needs sunshine. [FIGURE 1] Where a hundred kilograms of coffee cherry goes, as five columns on a scale of 0 to 100 kilograms. The first column is the whole cherry, 100 kg, in layers: coral pulp at the top, the yellow mucilage coat below it, and the rest. The pulper takes out 41 kg of skin and pulp; the tank and channel take out 13 kg of mucilage; a hatched block with no number stands for drying, hulling, sorting, shipping, roasting and brewing; and the last column, 2.6 kg, is what ends up in a cup. Interactive: step along the columns for a sentence about each. Where a hundred kilograms of cherry goes. Step along it with the pointer or the arrow keys. The fruit’s 41 and 13 kg are from source 2; the 2.6 kg that reaches a cup is from source 1, a different study; the hatched block between is not split in either, so it carries no number. [sources 1, 2 · interactive] Two numbers frame everything below. Of a hundred kilograms of cherry, about 2.6 end up as something a person actually consumes. [1] And the fruit thrown away is not inert: pulp and mucilage are sugar, acid and pectin, and washed into a river they eat the oxygen out of it. ---------------------------------------------------------------------------- 02 FLOATING -- The first thing a good mill does is throw part of the crop away. ---------------------------------------------------------------------------- A tank of water is the cheapest sorting machine there is, and almost nobody uses it. Cherries are tipped into water before anything mechanical touches them. This is flotation: a ripe cherry is mostly water and sugar, denser than water, so it sinks. One that dried on the branch, was hollowed out by an insect called the coffee berry borer, or was picked green and never filled, holds air. It floats. The floaters are skimmed off. That matters because an underdeveloped seed lacks the sugars to brown in a roaster, so it stays pale and tastes of paper, cereal or raw peanut. One of those in a cup is enough to notice. [FIGURE 2] A receiving tank in cross-section. Ripe cherries, drawn solid, have settled row on row into the cone at the bottom, above an outlet marked to the pulper. Hollow, dried, bored and underripe cherries, drawn pale, ride the water surface, and an arrow skims them over the left wall onto a heap marked skimmed out. Beside the tank, or under it on a phone, two numbers: 6.9 percent of 70 surveyed Colombian farms sorted cherries in water, and the tank ran 1.3 hours shorter when the sort was skipped. Why a tank of water is a sorting machine. The step is called hydraulic classification, flotation or siphoning depending on who is speaking. [from sources 2, 3] Yet in a Colombian field study of seventy farms, only 6.9 percent sorted their cherries in water. [3] [FOR THE ENTHUSIAST] what the sort is worth, and what skipping it does to the tank How much is it worth? Less than you would hope, and more than the practice suggests. In the main harvest, when pickers are going through a field that is mostly ripe at once, immature fruit is generally under three percent of what comes in. [2] That is a small share, but it is entirely defect and entirely removable, and it goes before any equipment, labour or water has been spent on it. The study was Cenicafé’s, across seventy representative farms in four departments. [3] It found a second-order effect too: skipping the sort shortened the fermentation stage by about 1.3 hours on average, because an unsorted mass is not the same material. [3] A mill that skips the tank is not only keeping its defects, it is also making the next stage harder to time. ---------------------------------------------------------------------------- 03 DEPULPING -- The pulper takes the fruit off and leaves the glue. ---------------------------------------------------------------------------- A machine strips two fifths of the cherry in one pass, and cannot touch the sticky coat underneath. A pulper squeezes cherries between a rotating textured drum and a fixed plate, set just far enough apart that the seed slips through and the skin does not. Skin and pulp fall out of one side; the seeds, still in their parchment and still coated in mucilage, fall out of the other. It is the least mysterious operation in the whole chain, and the pulp it removes is about 41 percent of the cherry. [2] What does not leave is the mucilage. It is not a syrup you can rinse off but a gel of pectin, the same family of molecule that sets jam, and it is bonded to the parchment. The pulper cannot take it off and neither can a hose. > WHERE THIS POST STOPS. Whether the pulper runs with water or dry changes > the whole water bill; that is section 06. Leaving the coat on the seed > on purpose is the honey route, a different post. coffee/honey-process · leaving the coat on -> [FOR THE ENTHUSIAST] the mass balance, the caffeine, and what the coat is made of Mucilage is 22 percent of the weight of depulped coffee and 13 percent of the weight of the whole cherry, [2] which puts the depulped coffee at 59 percent of the cherry and the pulp at the remaining 41. In one pass the mill has separated out two fifths of everything it carried in, and it now has to put that somewhere. Some of the coffee goes with it. A study that sampled every stage of a wet mill found that about 40 percent of the cherry’s caffeine leaves with the pulp. [16] Caffeine is a defence compound and the fruit holds a lot of it, which is one reason pulp is a difficult thing to throw into a stream. The coat is between 0.4 and 2 millimetres thick. [3] Its solid framework is a cell wall: roughly 30 percent pectic substances, about 9 percent cellulose and about 15 percent other non-cellulosic sugars. [8] ---------------------------------------------------------------------------- 04 THE TANK -- The tank is a timer nobody can see. ---------------------------------------------------------------------------- Most of a day in which nothing visibly happens, and most of the quality of a washed coffee is either kept or lost. Depulped coffee sits in a tank, usually concrete or tile, and waits. Microbes that were already on the fruit eat the sugars in the mucilage and make acids and alcohols, the tank turns sour, and the coat loosens until a rinse will take it off. Which microbes, in what order, and whether any of it reaches the seed is its own post. coffee/fermentation · what is actually living in the tank -> For the wet route the tank is process control, and the only questions are how long, how you know, and what happens if you are wrong. HOW A PRODUCER KNOWS IT IS DONE. There are two traditional tests. Push a stick into the coffee and pull it out: if the hole holds its shape, the coat has lost its stickiness. Or rub a handful of parchment between your palms until it stops slipping and squeaks like clean glass. Cenicafé, Colombia’s national coffee research centre, timed both against the real endpoint. Both stop the tank with a quarter to nearly half of the coat still on; it is really off, above 97 percent, at about 15.5 hours. [3] The instrument they built instead is almost comically simple: a perforated cone of coffee left in the tank, which reads how far the bed has settled as the coat drains away. [3] With the cone, tanks ran 11.45 to 20 hours. With the growers’ own judgement, on the same coffee, they ran 10 to 72. [3] That is the spread the tank has when nobody is measuring it. [FIGURE 3] Mucilage removed against hours in a fermentation tank, 0 to 30 hours. Four measured points: none at 0 hours; 58 percent at 7.29 hours, where the hole test calls it; 74 percent at 10.23 hours, where the rub test calls it; above 97 percent at 15.5 hours. A smooth line joins them and runs flat after 15.5 hours. Dashed rules mark 15.5 hours, coat off, and 20 hours, vinegar; the stretch between is shaded yellow and everything after 20 hours coral. Under the hours axis two bars show how long tanks actually ran in the field: 11.45 to 20.0 hours when a settling cone ended them, and 10 to 72 hours, running off the right edge, when growers judged it. Interactive: scrub along the hours for a sentence about each stretch. When the coat is really off, and how early the hands call it. Only the four dots are measured; the line between them is drawn smooth, and the readout names the dots either side rather than quote the line. Scrub along the hours. [source 3, and 2 for the cup · interactive] [FOR THE ENTHUSIAST] the timings, the cone, and how accurate it is The hole test called it at 7.29 hours on average, with about 58 percent of the mucilage actually removed. The rub test called it at 10.23 hours, with about 74 percent removed. True removal, above 97 percent, arrived at 15.5 hours plus or minus 0.9. [3] As mucilage drains away the bed of coffee gets denser, from about 827 kilograms per cubic metre when freshly depulped to about 702 when washed, and it settles by 11.9 to 13.1 percent of its starting volume. [3] So the cone is a perforated truncated cone holding half a litre of depulped coffee, left sitting in the tank. When the empty space above the coffee inside it passes 85 millimetres, more than 96 percent of the mucilage is gone. Across 45 controlled lots it landed at 96.7 percent with a standard error of 0.22, and in the field it called the point correctly in 97.1 percent of tests. [3] Temperature sets the clock. Holding a tank at 15 degrees Celsius rather than letting it run at ambient stretched the process by more than 24 hours. [5] A cold mill is a slow mill. This is one reason Kenya and Ethiopia ferment for a day or more while a warm Colombian tank is finished overnight, and why no mill can copy another mill’s schedule. [FOR THE ENTHUSIAST] pH, the tank’s walls, and what really happens to the pectin pH is a measure of acidity: the lower it is, the more acid. A field study across seven fermentation batches on four Nicaraguan farms found the tank starting at pH 5.5 to 5.7 and dropping sharply to about 4.6 as the process came to an end, with glucose falling throughout and either ethanol or lactic acid rising sharply at the finish. [4] A controlled study on Colombian coffee ended lower again, below 3.5 in one treatment and close to 4.0 in another; the same study’s 15 degree tank ended at a different final pH from the ambient one. [5] A third dataset, from a large trial covering both depulped and mechanically demucilaged coffee, starts higher, at pH 6.0 to 6.5, and falls to about 4.0 with the main drop after 36 hours. [7] They agree on the direction and not on the numbers. Even the tank’s building material shows up. In the Colombian field survey, stainless steel tanks ran slower than plastic ones, because steel conducts heat out through the walls and leaves the mass cooler. [3] And the pectin is not doing what you would guess. The obvious story is that microbes secrete pectin-digesting enzymes and dissolve the coat. When somebody checked, that turned out to be mostly wrong. Comparing the cell wall polysaccharides of unfermented and 20-hour fermented beans, the bulk pectic material showed no apparent degradation. What changed was subtle: a slight drop in the pectins’ viscosity and average molecular weight, and partial de-esterification of the water-insoluble fraction. The authors concluded that removing mucilage by natural fermentation is the result of a restricted pectolysis whose mechanism remains to be worked out. [8] The coat comes off, but not because it has been eaten. THE COST OF BEING LATE. Past the endpoint the tank buys nothing and costs a lot. More than two hours late is enough to start putting defects in the cup, the ones the Colombian trade calls vinagre and fermento. [3] By 20 hours, vinegar and overripe pineapple are there to taste. [2] [FOR THE ENTHUSIAST] how bad it gets, and the yield the tank costs The overfermentation trials put numbers on the slide: at 40 hours, 37.5 percent of samples were graded fermented or stinker; past 64 hours, more than three quarters of the cups were called nauseous. [2] There is a quieter cost as well. The seed is alive and breathing throughout, and a normal fermentation loses around 1.5 percent of the bean’s dry matter to respiration. [2] That is yield, gone, in exchange for a coat coming loose. THE MACHINE THAT SKIPS THE WAIT. A mechanical demucilager removes the coat by force instead of by time: depulped coffee is pushed through a narrow gap against a fast rotor, and the beans scour each other clean in minutes. Cenicafé’s version, part of its Becolsub system, does the job on about 0.8 litres of water per kilogram. [13] What it trades away is the open question of this post. Cenicafé’s own tasting panels found no quality advantage for the tank. [2] The largest study to follow both routes end to end found they are not the same process: different microbes, different chemistry, different cups. [7] Both things can be true: the machine will not make a worse coffee, and it will not make the same coffee. My own reading is that the disagreement is mostly about what is being asked. Can a careful mechanical mill hit specialty grade? Yes, and that is settled. Does the tank add a character some buyers are paying for? It seems to, and nobody has quantified it well enough to price it. [FOR THE ENTHUSIAST] the three studies behind that disagreement Cenicafé’s comparison, using the sensory panels of the Colombian coffee federation, came down clearly: the samples tracked 1.81 to one in favour of the mechanical route, and the authors state categorically that fermentation does not influence the quality of the coffee relative to mechanical demucilaging. [2] A separate controlled study found no significant quality difference between temperature-controlled and spontaneous fermentation either, with all treatments averaging above 82 points on the Specialty Coffee Association scale. [5] Against that, the largest study to track both routes end to end found that mechanically demucilaged batches and depulped batches developed different microbial communities, different metabolite profiles and measurably different sensory results, with extended fermentation increasing fruity and acidic notes. [7] ---------------------------------------------------------------------------- 05 THE CHANNEL -- Moving water rinses and grades at the same time. ---------------------------------------------------------------------------- The one stage where water is truly needed also sorts the coffee by weight, for free. Once the coat has loosened it has to be washed off, classically in a washing channel: a long, shallow concrete run with water flowing down it. This is where the mucilage finally leaves the coffee, and the only stage in the wet route where water is genuinely indispensable. [2] The second job comes free. Water in an open run moves fastest at the surface and slowest along the floor, so a dense seed sinks into the slow layer and lags while a light one rides high and is swept along. The batch sorts itself, heavy near the head and light at the tail. That is density grading, the float tank’s principle in a moving stream, and the heavy fraction is the first grade. [FIGURE 4] A washing channel in side section, water running left to right down a slight slope. At the head, five arrows show the speed of the water: longest at the surface, labelled fast at the surface, and shortest at the floor, labelled slow on the floor. Dense seeds, teal, lie on the floor near the head, labelled dense, sits low and lags: the first grade. Light seeds, coral, ride high, spread along the run and past a dashed take-off line near the right end, where an arrow carries them out, labelled light, rides high, carried out. Why a moving channel sorts. The velocity profile does the work: slow along the floor, fast at the surface, so weight decides which layer a seed lives in. [mechanism, not measurement] How the channel is built swings the water bill ninefold: about 39 litres per kilogram in an open channel with no reuse, 4.2 when the coffee is rinsed four times inside the tank instead. [2] Section 06 has the whole bill. [FOR THE ENTHUSIAST] the middle design, and why four rinses beat a running flow A semi-submerged channel, which recirculates its water, does the same job at about 6.1 litres per kilogram of dry parchment coffee. [2] Same coffee, same grading. The four-rinse method has a second property that matters more than the volume. Fill the tank, stir hard, drain, and repeat three times: the first rinse carries 66 percent of the dissolved organic matter from the mucilage and the first two carry 90 percent. [2] The pollution is concentrated into a small volume you can actually catch and treat, instead of diluted across forty litres you cannot. KENYA, AND THE REASON FOR DOING IT TWICE. Kenya runs the most elaborate version of the wet route in commercial use: pulp, ferment, wash, ferment again, wash again, then soak the clean parchment in clean water before drying. The first fermentation runs one to four days depending on the weather, and the soak about twelve hours. [12] [FIGURE 5] The usual washed route and Kenya’s, as two rows of steps. The usual route: pulp, ferment for 12 to 24 hours, wash, then a dashed arrow straight on to dry. Kenya’s: pulp, ferment for 1 to 4 days, wash, ferment again, wash again, soak for about 12 hours, dry. A bracket over Kenya’s second ferment, second wash and soak reads what Kenya adds. Interactive: step along Kenya’s sequence for a sentence about each step. What Kenya adds to the usual route: a second ferment, a second wash and a soak. The post has durations for three of the steps and no others, so they are drawn as steps rather than hours. Step along it with the pointer or the arrow keys. [sources 3, 12 · interactive] Two things explain it. Kenya’s coffee is grown high and the mills are cold, so a tank that would finish overnight in Colombia takes a day or more, and one long fermentation is harder to judge than two shorter ones. And the second wash is a correction: it gets the last of the coat off, and leftover mucilage is exactly what discolours parchment and sours a cup during drying. [2] The soak is the step people actually mean when they say Kenyan. Kenyan practice holds that it improves colour and taste; [12] the nearest controlled trial found a longer soak made the cup more intense, which is not the same thing as better. [7] Kenya’s reputation for a piercingly clean, blackcurrant-edged cup is real, and the soak is the step most often pointed at, but pointing is not the same as proving. [FOR THE ENTHUSIAST] what the evidence on the soak actually says, and what it plausibly does The FAO’s description of Kenyan practice states plainly that soaking parchment for about twelve hours after fermentation improves the coffee both in colour and in taste. [12] That is an industry judgement rather than a controlled result, and it is worth marking as one. The nearest controlled evidence is a large processing trial that applied a 24-hour soak to half of every batch and found it increased overall flavour intensity in the cup. [7] Increased intensity is a real, measured effect. It is not the same claim as increased quality, and I have not found a study that makes the stronger claim well. What the soak plausibly does is mechanical rather than microbial. Parchment coming out of a tank is carrying acids, sugars and microbes in its surface water. Sitting it in clean water, and changing that water, dilutes all of it away before drying locks whatever is left onto the bean. ---------------------------------------------------------------------------- 06 THE WATER -- This is the bill, and it is mostly avoidable. ---------------------------------------------------------------------------- A traditional wet mill spends about forty litres of water per kilogram of coffee, and half of it is not washing anything. The kilogram here is of dry parchment coffee: washed and dried, but still in its shell. A conventional Colombian mill spends about 40 litres on each, and half of that goes on flushing pulp away, a job a screw conveyor could do dry. [2] Pulping without water saves a further eighth. The part that is genuinely necessary, washing, is the smallest share. [FIGURE 6] Horizontal bars of water use in litres per kilogram of dry parchment coffee, on one axis from 0 to 66, in three groups. Whole mill: a traditional mill at 40 litres, its bar split into pulping 12.5 percent, flushing the pulp away 50 percent, and washing and moving beans 37.5 percent; Becolsub at 0.8; demucilaged, washed and cleaned at 0.6. Washing only: an open channel with no reuse at 39; a semi-submerged channel at 6.1; four rinses in the tank at 4.2. Per kilogram of green coffee, a different unit: the Ethiopian survey of the Gidabo basin at 63, hatched. The frugal bars are almost invisible next to the others. Interactive: scrub along the litres to see which of the six parchment figures fit that budget. The same job, six ways, and one that cannot be compared. Whole-mill figures sit apart from washing-only ones, and the traditional bar is split into its three uses: half of it is flushing pulp. The last bar is per kilogram of green coffee, a different and larger unit, and is not converted. Scrub along the litres. [sources 2, 13, 14 · interactive] At the frugal end, the machines. A demucilager that strips the coat, washes and cleans in one pass has been shown to do the whole job on 0.6 litres. [2] There is a catch, and Cenicafé’s own manual states it: a demucilager fitted badly, or run above one litre per kilogram, or installed without fixing how the pulp is handled, will pollute more than the tank it replaced. [2] The machine is not the saving. The redesign around the machine is the saving. [FOR THE ENTHUSIAST] the three-way split, Becolsub, and the zero-discharge mill The unit is written L/kg dpc. A conventional mill’s 40 splits three ways: 12.5 percent in pulping, 37.5 percent in washing the beans and moving them around, and 50 percent in flushing the pulp away. [2] Cenicafé’s packaged Becolsub system, which pulps without water, demucilages mechanically and carries the pulp and mucilage away together on a screw conveyor, runs at about 0.8. [2] [13] An ecological mill designed to discharge nothing at all comes in under 0.5. [13] That is between one fiftieth and one eightieth of the traditional figure. WHAT COMES OUT THE OTHER END. The used water, the effluent, is not dirty in the ordinary sense. It is clear-ish, smells sweet at first and vinegary later, and is extremely rich in dissolved sugar, pectin and acid. That richness is the problem: bacteria in a river feed on it and use up the oxygen. Cenicafé’s way of putting it is the most vivid: the pulp and mucilage in one kilogram of cherry can remove all of the oxygen from 7.4 cubic metres of clear water. [2] That is a small swimming pool per sack. And it happens at scale, every processing season, in the headwaters. Below one discharge on the Gidabo River in southern Ethiopia, which supplies more than 1.58 million people, dissolved oxygen reached zero. [14] Zero dissolved oxygen means nothing with gills is alive in that reach. [FOR THE ENTHUSIAST] BOD and COD, the Gidabo survey, and why treatment is hard The measure is BOD, biochemical oxygen demand: how much dissolved oxygen the bacteria in a water body will consume while eating what you put in it, in milligrams per litre. A companion measure, COD or chemical oxygen demand, does the same thing with a chemical oxidant and catches material the bacteria cannot reach. Raw coffee wastewater has been reported at BOD up to 20,000 mg/L and COD up to 50,000 mg/L, with a pH below 4. [15] One Colombian review puts the polluting power of coffee effluent at 60 to 240 times that of domestic wastewater. [13] The Gidabo survey sampled during the processing season and found mills using an average of 63 litres of water per kilogram of green coffee, with 90 percent of the 285 mills surveyed having no working recycling. Below one discharge, pH fell from 7.48 to 5.05, BOD rose from 30 to 300 mg/L and COD from 64.5 to 2,669 mg/L. [14] Treatment is possible and it is not easy. Anaerobic lagoons are the common answer and they are frequently undersized: the Gidabo survey measured an average retention time of 1.99 days, which is not long enough to degrade the load. [14] Coffee wastewater also contains tannins, phenolics and alkaloids, including all that caffeine that left with the pulp, and those actively inhibit the biological degradation you are relying on. [18] There is a substantial literature on turning the waste into something useful instead, from compost and biogas to pectin recovery, [17] and Cenicafé’s own route is to mix pulp and mucilage and feed them to worms. [2] The redesign works where it is adopted: handling pulp dry avoids 72 percent of the potential contamination at a stroke. [2] But I want to be careful not to turn this into an accusation pointed at farmers. The frugal designs cost money, the traditional ones do not, and the person choosing is often working under five hectares. [3] Washed coffee has a real environmental cost, the cost is technically solvable, and solving it is a capital expense borne by the poorest party in the chain. [FOR THE ENTHUSIAST] how much the redesign controls, and the national scale Pulp handled dry avoids 72 percent because the pulp is 72 percent of the potential contamination and the mucilage only 28. [2] Four rinses plus dry pulp handling controls about 85 percent, and a properly built mechanical mill controls about 92, by physical means alone with no treatment plant at all. [2] Scaled up, the numbers get large: Colombia’s traditional mills were estimated to need 46 million cubic metres of water a year, roughly the annual domestic consumption of a city of 840,000 people. [2] The farm sizes are from the Colombian field survey: 74.6 percent of its farms were under five hectares. [3] ---------------------------------------------------------------------------- 07 DRYING -- Half water, down to about a ninth. ---------------------------------------------------------------------------- The washed parchment has to dry for days, gently at first, before it can be stored. Parchment leaving the washing channel is about 53 percent water by weight, [2] and it has to reach roughly 10 to 12 percent: [22] too wet and moulds grow in the sack, too dry and the bean goes brittle and loses aromatics. The first day is gentle, because wet parchment cracks if it is heated hard, and a cracked shell stops protecting the seed. Drying is where the two routes converge, and the dry route’s post covers it in full. coffee/natural-process · drying, at length -> [FOR THE ENTHUSIAST] the export standards, Kenya’s named stages, storage and dryers National export standards cluster tightly around the same window: Rwanda specifies 10 to 12.5 percent for fully washed grades, Colombia caps green coffee at 12 percent, Uganda at 12.0 percent for washed arabica. [22] Kenyan practice targets 11 percent and describes the run in named stages, from skin drying at 55 down to 45 percent, through white, soft black and hard black, to a conditioning stage that lands at 11. [12] Cenicafé’s storage work found that parchment kept in cooled, aerated bins held 10 to 12 percent for a year and kept its cup quality for up to ten months, while sacks left in ambient conditions drifted up to 12.9 percent and lost it. [2] Mechanical dryers manage the cracking problem with high air temperatures against very wet grain, where the evaporation itself keeps the bean cool: air at up to 80 degrees can leave the grain below 40. [2] Controlled thin-layer experiments on parchment arabica at 50, 60 and 70 degrees and 10 to 30 percent relative humidity map the trade-off between speed and how far the moisture actually falls. [23] On a farm without a dryer the same job is done by shade first, then sun, and by not spreading the layer too thick. ---------------------------------------------------------------------------- 08 THE CUP -- Clean is not an absence. It is citric and malic acid. ---------------------------------------------------------------------------- A washed coffee tastes of the seed’s own chemistry, because the fruit was taken away before it could add much. The trade’s words for washed coffee are clarity, structure, brightness and a clean finish. Separately they are vague. Together they describe one thing: the flavours that show up are the ones the plant put in the seed while the cherry was ripening, and the ones the fruit could have added were removed before they had the chance. That does not make it unfermented. A day in a tank is long enough for lactic acid bacteria to take over [10] and for yeasts to leave a real mark on the cup. [11] What washing does is stop that early and take the food away, so far less of the fruit’s sugar is ever converted into anything. The acids left are mostly the plant’s own, chiefly citric and malic, rather than ones the microbes made out of the pulp. That is what cuppers call a clean cup, and it is a deliberate choice rather than a default. coffee/fermentation · what the microbes make -> [FOR THE ENTHUSIAST] what the seed itself is doing, and where the chemistry studies disagree Two measured differences support the same picture. In a full wet mill sampled stage by stage, the green beans held the highest concentration of organic acids and sucrose of anything in the line, at roughly 4.96 and 5.07 grams per 100 grams of dry weight depending on the mill type. [16] And the seed is behaving differently in the two routes: germination activity inside the bean peaks about two days after wet processing begins and about a week after dry processing begins, which is a different metabolic history written into the same seed. [19] Where the studies disagree is on composition, and they disagree sharply. One controlled study of wet processing found no influence of the processing method at all on chlorogenic acid content. [20] Others report wet-processed coffee higher in chlorogenic acids and trigonelline and lower in sucrose than semi-dry. I have not found a reconciliation, and the reason is probably that processing is entangled with origin, variety, altitude and roast in almost every dataset that exists. AND THE THING THAT COMPLICATES THE SALES PITCH. Washed coffee dominates specialty buying, and the usual explanation is that it is better. The evidence for that is weaker than the market suggests: when a Colombian study compared five processes on one lot, the natural scored highest, and statistically differently from the others. [21] One study on one lot is not a verdict, but it is not nothing either, and it points the opposite way to the price list. My own reading of why washed dominates has less to do with peak quality and more to do with variance. A washed lot has a definite ending: somebody decides the coat is off and washes it, and from then on the coffee is a bare seed drying in the open. A natural has no such moment; fermentation carries on inside the drying fruit until the water runs out. That makes washed the route you choose when you need a thousand bags that taste the same, which is what a roaster buying to a profile needs. Reproducibility is a real virtue, and it is not the same virtue as being the best cup in the room. [FOR THE ENTHUSIAST] that five-process study, and four places this story stops being true The five were natural, honey, conventional aerobic wet, aerobic with a prior fermentation in cherry, and anaerobic with a prior fermentation in cherry, profiled by gas chromatography and by a sensory panel. Fifty-one volatile compounds were identified, and the families that separated the processes were ketones and pyrroles. [21] 1. Washed does not mean unfermented. It means the fermentation was ended on purpose, and early. 2. Clean does not mean characterless. A washed coffee shows origin more clearly precisely because there is less fruit chemistry sitting on top of it. 3. The water figures are Colombian and the effluent figures are Ethiopian. Both are well measured and neither is a global average, because no global average of this exists. 4. Almost every comparison between processing methods is confounded by origin, variety, altitude, harvest date and roast. Treat single-lot comparisons as observations, not results. ---------------------------------------------------------------------------- 09 WHAT I DO WITH IT -- I read washed as a promise about consistency, not about quality. ---------------------------------------------------------------------------- The word tells me the fruit came off fast and somebody decided when to stop. It does not tell me the coffee is good. I buy coffee about twice a month and keep a log of it, so this is a practical question for me. Washed now tells me three things and no more. That the fruit came off within hours. That somebody made a decision about when to stop, which means a person was paying attention on a specific morning. And that the lot is likely to be more uniform than a natural from the same farm, because a definite ending produces less spread than an indefinite one. It does not even tell me whether the fermentation was controlled: the same word covers tanks run for eleven hours and tanks run for seventy-two. [3] When a bag says washed and nothing else, I read that as a category, not a claim. Two things changed in how I brew. Washed lots are the ones where I chase acidity, because citric and malic are what is actually there, and a slightly higher water temperature or a slightly finer grind brings them forward. And I stopped treating a faint sourness in a washed coffee as my brewing error: vinegar and overripe fruit in a washed cup is a processing signature with a known cause, and past twenty hours in a tank it is what you get. [2] What I would most like on a bag, and cannot get, is the fermentation time and how it was ended. Washed, tank ended by cone at 14 hours, mechanically demucilaged, water recirculated would tell me more about my morning than the altitude, the variety and the tasting notes put together. [FOR THE ENTHUSIAST] the processing fields I keep in my log, and what this post still lacks Until a bag says it, I write the processing note in the log and leave a column empty next to it. // what "washed" would have to say to be useful. Most bags fill // the first line and nothing else. { "process": "washed", // a category, not a claim "mucilage_off": "tank", // tank or machine, they differ [7] "tank_hours": null, // field data span 10 to 72 h [3] "ended_by": null, // clock, hand, or a measurement [3] "soaked": null, // the Kenyan step, if any [12] "water_reuse": null, // 0.6 or 40 L/kg, nobody says [2] "dried_to_pct": 11, // the one number that is standard "grind": 16, "water_c": 94 } Two things I still want to add: a proper look at whether soaking has ever been tested against a no-soak control on the same lot in Kenya, which I could not find, and a water figure that reports litres per kilogram of green coffee as well as per kilogram of parchment, so the Ethiopian and Colombian numbers can sit on one axis honestly. ---------------------------------------------------------------------------- 10 YOUR AGENT -- Use your agent with this page. ---------------------------------------------------------------------------- Every section and every figure has an ask agent button: pick how deep to go, and it copies a prompt about just that part. Selecting a passage offers one too. Paste it into any assistant. start_guided_reading: a teaching plan, a quiz, or just the answer get_outline: every section, figure and claim explain_section: a section at child, student or expert level explain_figure: a figure, and how to read it get_evidence: the study behind a number get_glossary: every technical word, plainly [FOR THE ENTHUSIAST] how the tools work, and an example answer Processing is the part of coffee where the audience splits hardest. Somebody wants to know why their bag says washed, somebody is a food science student who wants the pectin chemistry, and somebody runs a mill and wants the water figures. One piece of prose cannot be pitched at all three at once, so every section and every figure on this page is also written out three times, at three levels, by me rather than by a model at runtime. The six tools are registered with WebMCP when the browser has it, which today means behind a flag, and the page does nothing at all when it does not. All six are read-only, answer from this page’s own data, and make no network call beyond loading that data from this site. An agent without a browser can read the plain-text version of the whole post, prose and sources, at. washed-process.txt ASK YOUR AGENT ABOUT THE WHOLE POST. Paste this into any AI chat: I'm reading "What washed process means". https://hetanshmehta.com/blog/coffee/washed-process Explain the whole post simply, for someone with no science. - With this page's WebMCP tools: call get_outline(), then explain_section({section, level: "child"}) for each section. - If you can open links, read https://hetanshmehta.com/txt/blog/coffee/washed-process.txt. - If neither, say so and I'll paste the post in. Stick to what the page says, and tell me where it stops. ---------------------------------------------------------------------------- SOURCES -- 23 references ---------------------------------------------------------------------------- Peer-reviewed, governmental or from a national coffee research institute unless marked otherwise. Where two studies disagree, both are here and the disagreement is stated in the post rather than resolved. 1. Rotta, N.M., Curry, S., Han, J., Reconco, R., Spang, E., Ristenpart, W. and Donis-González, I.R. (2021). A comprehensive analysis of operations and mass flows in postharvest processing of washed coffee. Resources, Conservation and Recycling, 170, 105554. https://doi.org/10.1016/j.resconrec.2021.105554 2. Roa Mejía, G., Oliveros Tascón, C.E., Álvarez Giraldo, J., Ramírez Gómez, C.A., Sanz Uribe, J.R., Dávila Arias, M.T., Álvarez Hernández, J.R., Zambrano Franco, D.A., Puerta Quintero, G.I. and Rodríguez Valencia, N. (1999). Beneficio ecológico del café. Cenicafé, Chinchiná, Colombia, 273 pp. English edition: Coffee processing at the farm level: an ecological and profitable case. National coffee research institute, not peer-reviewed; the source of most water, mass and contamination figures here. https://www3.ctahr.hawaii.edu/hawaii/downloads/Low_Water_use_processing.pdf 3. Peñuela Martínez, A.E., Pabón Usaquén, J.P. and Sanz Uribe, J.R. (2013). Método Fermaestro: para determinar la finalización de la fermentación del mucílago de café. Avances Técnicos Cenicafé, 431. National coffee research institute, not peer-reviewed. https://biblioteca.cenicafe.org/handle/10778/479 4. Jackels, S.C. and Jackels, C.F. (2005). Characterization of the coffee mucilage fermentation process using chemical indicators: a field study in Nicaragua. Journal of Food Science, 70(5), C321-C325. https://doi.org/10.1111/j.1365-2621.2005.tb09960.x 5. Peñuela-Martínez, A.E., García-Duque, J.F. and Sanz-Uribe, J.R. (2023). Characterization of fermentations with controlled temperature with three varieties of coffee (Coffea arabica L.). Fermentation, 9(11), 976. https://doi.org/10.3390/fermentation9110976 6. Peñuela-Martínez, A.E., Moreno-Riascos, S. and Medina-Rivera, R. (2023). Influence of temperature-controlled fermentation on the quality of mild coffee (Coffea arabica L.) cultivated at different elevations. Agriculture, 13(6), 1132. https://doi.org/10.3390/agriculture13061132 7. Zhang, S.J., De Bruyn, F., Pothakos, V., Contreras, G.F., Cai, Z., Moccand, C., Weckx, S. and De Vuyst, L. (2019). Influence of various processing parameters on the microbial community dynamics, metabolomic profiles, and cup quality during wet coffee processing. Frontiers in Microbiology, 10, 2621. https://doi.org/10.3389/fmicb.2019.02621 8. Avallone, S., Guiraud, J.-P., Guyot, B., Olguin, E. and Brillouet, J.-M. (2001). Fate of mucilage cell wall polysaccharides during coffee fermentation. Journal of Agricultural and Food Chemistry, 49(11), 5556-5559. https://doi.org/10.1021/jf010510s 9. Avallone, S., Guiraud, J.-P., Guyot, B., Olguin, E. and Brillouet, J.-M. (2000). Polysaccharide constituents of coffee-bean mucilage. Journal of Food Science, 65(8), 1308-1311. https://doi.org/10.1111/j.1365-2621.2000.tb10602.x 10. de Oliveira Junqueira, A.C., de Melo Pereira, G.V., Coral Medina, J.D., Alvear, M.C.R., Rosero, R., de Carvalho Neto, D.P., Enríquez, H.G. and Soccol, C.R. (2019). First description of bacterial and fungal communities in Colombian coffee beans fermentation analysed using Illumina-based amplicon sequencing. Scientific Reports, 9, 8794. https://doi.org/10.1038/s41598-019-45002-8 11. Elhalis, H., Cox, J., Frank, D. and Zhao, J. (2020). The crucial role of yeasts in the wet fermentation of coffee beans and quality. International Journal of Food Microbiology, 333, 108796. https://doi.org/10.1016/j.ijfoodmicro.2020.108796 12. Food and Agriculture Organization of the United Nations. Annex 5: Coffee post-harvest handling and processing in Kenya. Intergovernmental, not peer-reviewed; used for Kenyan fermentation and soaking times, the named drying stages and the 11 percent target. https://www.fao.org/4/x6939e/X6939e11.htm 13. González-Freire, A. and Martínez-Hernández, C.M. (2022). Improvement and use of effluents for the benefit of coffee. Revista Ciencias Técnicas Agropecuarias, 31(2). https://www.redalyc.org/journal/932/93271464008/html/ 14. Ulsido, M.D., Geleto, M.Z. and Berego, Y.S. (2024). Waste water management in wet coffee processing mills and their impact on the water quality status of Gidabo River and its tributaries, southern Ethiopia. Environmental Health Insights, 18, 11786302241260953. https://doi.org/10.1177/11786302241260953 15. von Enden, J.C. and Calvert, K.C. (2002). Review of coffee waste water characteristics and approaches to treatment. PPP Project, German Technical Cooperation Agency (GTZ), Vietnam. Development agency report, not peer-reviewed; used only for the upper-bound BOD, COD and pH figures for raw coffee effluent. https://www.researchgate.net/publication/238084098_Review_of_coffee_wastewater_characteristics_and_approaches_to_treatment 16. Figueroa Campos, G.A., Sagu, S.T., Saravia Celis, P. and Rawel, H.M. (2020). Comparison of batch and continuous wet-processing of coffee: changes in the main compounds in beans, by-products and wastewater. Foods, 9(8), 1135. https://doi.org/10.3390/foods9081135 17. Rattan, S., Parande, A.K., Nagaraju, V.D. and Ghiwari, G.K. (2015). A comprehensive review on utilization of wastewater from coffee processing. Environmental Science and Pollution Research, 22(9), 6461-6472. https://doi.org/10.1007/s11356-015-4079-5 18. Ijanu, E.M., Kamaruddin, M.A. and Norashiddin, F.A. (2020). Coffee processing wastewater treatment: a critical review on current treatment technologies with a proposed alternative. Applied Water Science, 10, 11. https://doi.org/10.1007/s13201-019-1091-9 19. Bytof, G., Knopp, S.-E., Kramer, D., Breitenstein, B., Bergervoet, J.H.W., Groot, S.P.C. and Selmar, D. (2007). Transient occurrence of seed germination processes during coffee post-harvest treatment. Annals of Botany, 100(1), 61-66. https://doi.org/10.1093/aob/mcm068 20. Joët, T., Laffargue, A., Descroix, F., Doulbeau, S., Bertrand, B., de Kochko, A. and Dussert, S. (2010). Influence of environmental factors, wet processing and their interactions on the biochemical composition of green Arabica coffee beans. Food Chemistry, 118(3), 693-701. https://doi.org/10.1016/j.foodchem.2009.05.048 21. Flórez Arenas, A., Villegas Hincapié, A.M. and Taborda Ocampo, G. (2022). Effects of post-harvest process on volatile and sensory profile for coffee in Colombia. Coffee Science, 17, 1-14. https://doi.org/10.25186/.v17i.2016 22. International Coffee Organization (2018). National quality standards. Document ICC-122-12, 122nd session of the International Coffee Council. Intergovernmental, not peer-reviewed; used for the national moisture limits. https://www.ico.org/documents/cy2017-18/icc-122-12e-national-quality-standards.pdf 23. Phitakwinai, S., Thepa, S. and Nilnont, W. (2019). Thin-layer drying of parchment Arabica coffee by controlling temperature and relative humidity. Food Science & Nutrition, 7(9), 2921-2931. https://doi.org/10.1002/fsn3.1144 Three things I could not verify and so did not state as fact. First, the exact cupping penalty for suppressing yeasts in a wet fermentation: the direction is clearly reported in source 11 but I could not reach a primary text giving the point difference, so the post states the direction only. Second, a controlled trial of the Kenyan soak against a no-soak control on the same lot: source 12 asserts the benefit as practice and source 7 measures a related 24-hour soak, and the gap between those is not closed. Third, a like-for-like water figure: source 2 reports litres per kilogram of dry parchment and source 14 per kilogram of green coffee, so the last bar in the water figure is hatched and labelled rather than converted. ---------------------------------------------------------------------------- Merged to main from coffee/washed-process · a1c756b · 24 Sep 2026. Siblings on this branch cover the dry route, the microbiology of the tank and the honey route.