Dry milling

Dry milling is everything that happens to coffee after it has been dried and before it is put in a bag for export. The centrepiece is hulling — cracking off the hard, dead outer layers so the green seed comes out — and everything around it is sorting: by size, by weight, by colour, and finally by eye.12

Wikipedia’s coffee production article puts the whole job in one sentence: “The final steps in coffee processing involve removing the last layers of dry skin and remaining fruit residue from the now-dry coffee, and cleaning and sorting it”.3

The name is a little misleading. Dry milling is not a processing method — it is not an alternative to washed or natural. It is the shared ending. A washed coffee, a natural, a honey and a pulped natural all arrive at the same shed and go through the same machines.4 What changes is only how much dead material has to come off.

The confusing sibling is wet milling, which is the opposite end of the timeline: the wet mill is the washing station where cherry is pulped and fermented within hours of picking. Wet milling happens at the start, dry milling at the finish, and a coffee usually spends weeks or months drying and resting in between.

What actually comes off

Hulling removes different things depending on what the farm did months earlier. This is the clearest statement of it, from Wikipedia:

“The first step in dry milling is the removal of what is left of the fruit from the bean, whether it is the crumbly parchment skin of wet-processed coffee, the parchment skin and dried mucilage of semi-dry-processed coffee, or the entire dry, leathery fruit covering of the dry-processed coffee.”

What arrived at the millWhat hulling takes off
Washed — dried parchment coffeeThe papery parchment shell only
Honey / pulped naturalParchment plus the dried, glassy mucilage stuck to it
Natural — whole dried cherriesThe entire husk in one pass: skin, dried pulp, mucilage and parchment together

Two more words worth knowing before the machines start:

  • Parchment (botanically the endocarp) is the thin, brittle, papery shell wrapped directly around the coffee seed. A washed coffee is stored and shipped inland as “parchment coffee” and only loses that shell at the dry mill — because the parchment protects the seed in storage.
  • Silverskin is the even thinner membrane underneath the parchment, fused to the bean itself and tucked into its centre crease. Hulling does not fully remove it. A separate optional step does.

The National Coffee Association describes both halves of the same split — that in natural process “the entire dried husk, including the outer layer, pulp, and parchment” comes off, while in washed coffee “the papery inner parchment is removed” after drying.5

Parchment, pergamino, and coffee that ships with its shell on

The layer this whole step exists to remove has its own trade name in two languages. Parchment — botanically the endocarp, in Spanish pergamino — is the brittle papery shell wrapped directly around the seed; Coffee cherry has the full anatomy. Coffee Review’s processing reference introduces it as “a last thin skin covering the bean, called the parchment skin or pergamino.”6

The shell is kept on deliberately, and for a long time. A washed coffee is dried, stored, moved and often bought and sold inside its parchment, because the shell is packaging: it stops the bean scuffing, slows the rate at which it trades water with the air, and shields it from oxygen. The normal sequence at origin is therefore dry, hold as parchment, and hull as late as the shipping schedule allows. Hulling is the act of unwrapping the coffee, and it is done at the last sensible moment rather than the first.

Hulling is not compulsory, though. Coffee Review is explicit that “coffee occasionally is sold and shipped in parchment or en pergamino, but most often a machine called a huller is used to crunch off the parchment skin before the beans are shipped.”6 A lot described as en pergamino is coffee traded with its shell still on — usually inland, between a farmer and a mill or exporter, and weighed and priced as parchment rather than as green coffee. Because parchment is bulkier and lighter than the green coffee inside it, the two are not interchangeable weights, and a price quoted for one says nothing directly about a price for the other.

Between the parchment arriving and the huller starting, some specialty mills insert a deliberate rest. A lot averaging 11% moisture is not a lot of beans that are all at 11% — it can be a mixture of 9% beans and 13% beans, and mixed beans hull unevenly, break more, and later roast unevenly. Sitting the parchment in a conditioned warehouse lets the wet and dry beans in the lot even out before anything mechanical touches them. The practice is well established; the specific durations and warehouse settings quoted for it are not.

Alongside moisture, the other number specialty buyers now ask a mill to report is water activity — how much of the water in the bean is chemically free rather than merely present, and therefore how much of it can grow mould or drive staling. That note records the same problem as this one: the figures in circulation are trade practice, and no published standard fixes any of them.

To verify

The dry-mill spec sheet everyone quotes traces to a single blog post. One bylined specialty blog is the only source found for the precise numbers attached to the front of the mill: arriving lots targeted at 10.5–11.5% moisture with anything above 12.5% rejected or re-dried; parchment rested 2–12 weeks in warehouses held at 18–22 °C and 55–60% relative humidity; an ideal water activity of 0.55–0.60; and specialty buyers increasingly requesting lots unpolished to protect flavour.7 Every one of these is plausible and consistent with what this wiki records elsewhere — the moisture band sits inside the 10–12.5% spread below, and 0.55–0.60 aw sits inside the range in Water activity. But no standards body, exporter document, extension manual or study was found stating any of them, so read them as one writer’s working figures rather than industry specifications.

The eight stations

A modern dry mill is a line of machines, each one throwing away a different kind of bad bean. Two independently fetched sources — a specialty-process reference and an Ethiopian exporter’s technical guide — describe the same sequence in the same order:

1. Pre-cleaning. Before anything mechanical touches the coffee, sticks, stones, string, nails and dust are pulled out. Ethiopian mills use “a pneumatic separator (airstream) that lifts loose dust, fiber, and broken parchment”.8 This step exists to protect the hulling machine as much as the coffee.

2. Hulling. The shell is cracked and rubbed off by “a disc huller or drum huller, which uses friction and impact to crack and abrade the outer shell”.9 Wikipedia notes the machinery spans everything “from simple millstones to sophisticated” hullers. What comes out the far side is green coffee — “the raw, unroasted seed ready for further sorting”.

3. Polishing (optional). Friction rollers scrub off the silverskin left clinging to the bean. Wikipedia gives the reason plainly: “This is done to improve the appearance of green coffee beans and eliminate a byproduct of roasting called chaff”. It is cosmetic, it is done for markets that want a uniform-looking bean, and it costs a little weight. A Peruvian mill’s own account of its line gives the order plainly: “the huller removes the outer brittle parchment shell from the coffee seed (i.e., bean). Once the parchment is removed, the coffee is polished to remove a layer of silver skin.”10 The specialty end of the market has partly turned against the step — buyer-facing trade writing reports lots now being asked for explicitly unpolished, on the argument that scrubbing a bean does nothing for the cup and only exposes more of its surface to air.7 The preference is real in how coffee is offered; the flavour reasoning behind it is unproven, and is flagged above.

4. Size grading. The coffee runs over stacked vibrating screens — “beans fall through successive screens until they reach a screen small enough to retain them”. Screen numbers are measured in sixty-fourths of an inch, so “screen 18” means a bean too big to fall through an 18/64-inch hole. A screen number on a bag is a statement about bean size, not bean quality — though the two correlate enough that most origins price them together.

5. Density sorting. Sized beans go onto a gravity table, described by Coffeester as “a vibrating, slightly inclined deck” and by Wikipedia as a machine that “shakes the sized beans on a tilted table, so that the heaviest, densest and best vibrate to one side of the pulsating table, and the lightest to the other”. This is where hollow beans, insect-bored beans and underdeveloped beans leave, because damage and immaturity both show up as missing mass.

6. Colour sorting. Optical sorters photograph every bean in flight and blow the wrong-coloured ones sideways with a puff of air, using “high-resolution cameras and, in advanced units, near-infrared (NIR) sensors”. This catches blacks, sours and ferment-damaged beans that weigh the right amount.

7. Hand sorting. People on illuminated tables or a conveyor “visually inspect every bean”. Wikipedia is blunt that at the top of the market this is not a fallback but the standard: “With most high-quality coffees color sorting is done in the simplest possible way: by hand”.

8. Bagging. The finished green coffee is weighed into export bags, “sewn, stacked, and held in a temperature-controlled warehouse” until it ships. This is also the moment traceability gets fixed in physical form: bags are weighed, closed and tagged with lot codes tied back to a farm or farmer group, a harvest date and the mill’s own quality readings, which is what allows a roaster on another continent to name the mill a coffee passed through.7 Specialty bagging now usually adds a sealed liner — GrainPro, Ecotact or similar — inside the jute sack.10

To verify

There is no single export bag weight. Three sources consulted for this note give three answers: 69 kg “typically for export” from the blog above,7 a 70 kg bag from the Peruvian mill describing its own line,10 and 60–70 kg jute bags with GrainPro liners as current specialty practice from the importer Nordic Approach, recorded in this wiki’s Coffee supply chain note. The 60 kg bag used all over this wiki’s origin notes is a statistical unit for reporting national crops, not a physical sack every exporter fills. The safe reading is 60–70 kg, set by national convention and the buyer’s contract; do not quote a universal figure.

Grading sits on top of all of this rather than being a ninth machine. Wikipedia’s definition: “Grading is the process of categorizing coffee beans by various criteria such as size of the bean, where and at what altitude it was grown, how it was prepared and picked, and how good it tastes”. Steps 4 to 7 generate the physical evidence; the grade name is the label put on it.

Moisture: the number that decides when milling can start

Coffee cannot be hulled until it is dry, and it cannot be stored if it is too wet — mould and staling both start above roughly 12–13%. Every source consulted agrees on the principle and none agree on the number.

SourceFigure given
ISIC — coffee industry scientific bodyMaximum 12.5% before storage, for washed and natural alike11
NCA — US National Coffee AssociationDried “to approximately 11% moisture to properly prepare them for storage”
Wikipedia — Coffee productionApproximately 10% water content before stability and distribution
CoffeesterOptimal window at hulling “generally cited as 11–12%
Ethio CoffeeWashed parchment arrives at 11.0–12.5%; export range 10.5–12.5%

The honest reading: somewhere between 10% and 12.5%, with 11% the most commonly repeated single figure and 12.5% the ceiling nobody wants to cross. Read alongside this wiki’s Washed process note, which records the same spread from the drying side — and the underlying reason from Cafe Imports, that below about 11% microbes can no longer work, which is precisely why that band is the target.

To verify

No standards body publishes a binding number. The Specialty Coffee Association’s green coffee standard — SCA-110 Green Coffee: Grading, Specifications, and Test Methods — is still listed under “Standards in Development” on the association’s own page and is not published.12 So the moisture figures above, and the screen numbers below, are trade convention and national rules rather than a single global specification. Anyone quoting “the SCA moisture limit” for green coffee is quoting something that does not yet exist.

Who owns the mill

This is the part that matters commercially, and the two fetched sources agree on the shape of it. Coffeester: “Dry milling is typically carried out at a dedicated dry mill (also called a hulling or threshing station), which may be located on a large estate, at a cooperative, or at an exporter’s central facility”. Ethio Coffee reports Ethiopian mills as “owned by a mix of private exporters and cooperative unions” and concentrated “on the outskirts of Addis Ababa”.

Three ownership models, three very different consequences for a smallholder:

  • The estate mill. A farm large enough to own its own huller controls the last step and can keep a lot separate right up to the bag. This is the normal arrangement behind competition lots — Daterra in Cerrado Mineiro and the Panamanian estates around Best of Panama all mill their own.
  • The cooperative mill. Members deliver parchment and the co-op mills and markets it. Traceability survives only as far as the co-op’s record-keeping allows.
  • The exporter’s central mill. One facility mills for many producers. This is the highest-capacity and lowest-cost option, and the one where a small farmer’s coffee most easily becomes indistinguishable from everyone else’s. It is also why vertically integrated traders such as Sucafina and origin exporters such as Daye Bensa treat milling capacity as a competitive asset, not a service they buy in.

The pattern is the same one this wiki’s Washed process note records for washing stations: in origins built on small farms, the shared facility rather than the farm is the real unit of quality. Dry milling is the second such chokepoint, and it comes after the coffee has already left the farmer’s control.

The mill also decides which market a lot is sold into, because a sorting line does not produce one output. Everything the gravity table, the optical sorter and the hand table reject still exists and still gets sold — as a lower grade, as material for soluble (instant) coffee, or into the domestic market at home. The export bag is the top fraction of what came through the door, not all of it.

To verify

Where the rejected fractions actually go is thinly documented. A 2020 post-harvest quality-assurance thesis hosted by UC Davis is the only source behind the specific claim that higher-value coffee is exported after dry milling while lower-value fractions go to instant coffee or domestic consumption.13 The document could not be read here — it exceeds the fetch size limit — and no second source, and no volume or percentage, was found. The direction of the claim is uncontroversial in the trade, since mills sell their rejects rather than binning them, but the specific routing is recorded here as a claim rather than a fact.

To verify

The Ethiopian grade thresholds. Ethio Coffee gives specific screen-to-grade mappings — “beans retained on size 14 and above are generally considered specialty grade,” size 18+ trading as G1, screens 14–15 as G2–G3. This is a single commercial exporter’s guide, not a government document. Ethiopian grades are set by state machinery — the Ethiopian Coffee and Tea Authority and the Ethiopia Commodity Exchange — and this wiki’s ECTA note already warns that the Authority publishes far less per-grade detail than trade sources imply. Treat these thresholds as one exporter’s working summary until an ECTA or ECX schedule is read directly.

To verify

“Peaberries are 10–15% of any harvest.” Coffeester states this while describing screen grading. A peaberry is a cherry that grew one round seed instead of two flat ones, and it is separated at the screening stage because its shape will not behave like a normal bean. No second source consulted gives a share, the figure sits well above the ~5% commonly repeated elsewhere in the trade, and no agronomic study was read. Treat 10–15% as one reference site’s number, not a measured rate. See Peaberry.

Why it matters

Dry milling is where quality stops being created and starts being defended. Nothing in the shed makes a coffee taste better than the cherry allowed.14 Every machine in the line does one thing: throw away beans that would drag the cup down.

  • It is what a defect count actually measures. This wiki’s Fine Robusta note records CQI’s rule that a lot must have zero primary defects in 350 g — no full blacks, no insect-bored beans, no stones. Zero. That number is not a farming outcome; it is a milling outcome. The gravity table, the optical sorter and the hand table are the only reason it is achievable at all, which is why the Fine Robusta note frames the standard as an argument about mill investment as much as about taste.
  • It is the last chance to keep a lot separate. A competition lot is worth its price because it is that farm, that picking, that fermentation. Running it through a shared mill after a commodity lot without a full clean-down mixes the two. Micro-lot traceability lives or dies at the huller, not on the farm.
  • It is the step Robusta mostly does not get. Almost all Robusta is dried whole as a natural and milled cheaply at volume — which is exactly why hitting specialty defect standards is so hard for the species, and why origin-level programmes rather than individual farms drive Fine Robusta.
  • It is where “screen 18 Geisha” comes from. The size, density and defect language on a specialty bag — screen numbers, grade names, defect counts — is generated here. Cup score is decided by cupping; everything else on the label is decided by the mill.
  • Indonesia does it differently, on purpose. Wet hulling (giling basah) strips the parchment off while the coffee is still far wetter than normal, producing the distinctive earthy, low-acid Sumatran profile. It is the one place where hulling is a flavour decision rather than a cleaning step, and it is the reason this note’s “milling is not a processing method” rule has an exception.

To verify

Nobody consulted here has published what milling does to a cup score. The research behind this note found no primary cupping data isolating the effect of specific milling choices — polished versus unpolished, one density recipe versus another — and no Best of Panama or Cup of Excellence lot sheet tying a named huller, mill or sorting recipe to a score, price or buyer. Competition records publish farm, variety, altitude and process; they do not publish mill parameters. Any claim that a particular hulling or polishing choice raised a score is, on the evidence available, unsupported. The trade reason is simple: those recipes are proprietary to the mill.

Footnotes

  1. Wiki Coffee — Coffee Milling — cited in the research report, not independently fetched

  2. World Bank — Coffee agronomy and processing (FAU14) — cited in the research report for the historical framing of the dry method, not independently fetched

  3. Wikipedia — Coffee production — source of the hulling, polishing, gravity separator, hand-sorting and grading definitions quoted above; confirmed via WebFetch

  4. Counter Culture Coffee — Coffee Basics: Processing — roaster; cited in the research report for the dry-mill finishing step, and already confirmed via WebFetch for this wiki’s Washed process and Natural process notes

  5. National Coffee Association USA — Coffee processing — US industry association; the natural-vs-washed split of what hulling removes, and the ~11% moisture target, confirmed via WebFetch

  6. Coffee Review — Processing: Fruit Removal and Drying — long-running coffee review publication and reference; confirmed via WebFetch; source of the parchment/pergamino naming, the “sold and shipped in parchment or en pergamino” practice, and the huller “crunching off” the parchment skin 2

  7. Bean Brew Digest — Dry Mill Processing Stages — bylined specialty coffee blog (James Okafor), not an industry body; confirmed via WebFetch as the source of the 10.5–11.5% arrival target and 12.5% rejection line, the 2–12 week rest at 18–22 °C / 55–60% RH, the 0.55–0.60 water activity figure, the unpolished-lot preference, and the 69 kg bag and lot-coding description — all flagged above rather than stated as fact 2 3 4

  8. Ethio Coffee — Ethiopian Coffee Dry Milling: Export Preparation Guide — commercial exporter; the eight-stage sequence, pneumatic pre-cleaning, moisture ranges and the flagged Ethiopian grade thresholds; confirmed via WebFetch

  9. Coffeester — Milling, Sorting & Grading Green Coffee — disc/drum hullers, 1/64-inch screens, gravity tables, NIR optical sorters, the 11–12% hulling window, mill ownership models and the disputed peaberry share; confirmed via WebFetch

  10. ElevaFinca — From Parchment To Green: The Dry Milling Process — green coffee sourcing company describing the Café Selva Norte dry mill in Peru; confirmed via WebFetch; source of the huller-then-polisher sequence, the screen/gravity/optical sorting line, and the 70 kg hermetically lined export bag 2 3

  11. ISIC (institute for scientific information on coffee) — Initial coffee bean processing — industry scientific body; 12.5% storage ceiling, silo storage, and “hulled to remove the parchment, and cleaned, screened, sorted and graded” confirmed via WebFetch

  12. Specialty Coffee Association — Coffee Standards — the trade’s standards body; SCA-110 green coffee grading confirmed as still “in development,” not published; confirmed via WebFetch

  13. Neil Rotta (2020) — Post-harvest Processing and Quality Assurance for Speciality Coffee (thesis PDF, hosted by UC Davis) — cited in the research report for the dry-mill phase and the export-versus-instant-coffee split of fractions; fetch attempted here but the file exceeds the size limit, so it was not read directly

  14. Sweet Maria’s library — Processing and Drying of Coffee: A Review (PDF) — cited in the research report for drying and flavour formation, not independently fetched