The Bean

Here’s the reframe that changed how I think about coffee: coffee is not a bean. It’s the seed of a fruit. We call it a bean because roasted it looks like one, and that one lazy word has done more to confuse people about coffee than anything else. See it as a fruit instead — a red cherry on a shrub, picked ripe, flesh stripped off to get at the pit — and the whole thing snaps into focus. Why coffee tastes fruity. Why how you remove the fruit matters as much as who roasts it. Why altitude and ripeness get talked about the way a winemaker talks about a vintage. What follows is the plant, the seed, and everything that happens before a roaster ever sees it.

The plant

Coffee is the genus Coffea, in the same family as gardenia and quinine. Left alone it’s a shrub or small tree that would happily reach 15 or 30 feet; on a farm it’s pruned to head height so it can be picked by hand. The flowers are small, white, and smell like jasmine — a farm in bloom is one of the better-smelling places on earth.

Then the fruit sets and takes its time: roughly seven to nine months from flower to ripe cherry for Arabica. Green, then yellow, then deep red (some varieties ripen yellow and stay that way). A tree carries cherries at several stages of ripeness at once, which is exactly why picking selectively — going through the same tree repeatedly, taking only the ripe ones — beats stripping the branch or running a machine down the row. Ripeness is sugar, and sugar is everything downstream. A mature tree yields a few pounds of cherries a year, which cooks down to about a pound of roasted coffee: a year of one tree’s work is a few weeks of your mornings.

Arabica vs. Robusta — an honest comparison

There are well over a hundred Coffea species. Two of them are essentially the entire commercial world.

Coffea arabica is what specialty coffee is built on — a little over half of world production: the USDA’s July 2026 forecast for 2026/27 is 105.9 million 60-kg bags of arabica against 83.8 million of robusta, about 56% to 44%, and arabica has run between about 53% and 58% of the crop since 2021/22 (Coffee: World Markets and Trade). The older figure of 60–70%, still widely quoted, is out of date. Arabica is the older, fussier, better-tasting species. It’s tetraploid (four sets of chromosomes, where most Coffea are diploid) and largely self-pollinating, which is a bigger deal than it sounds: selfing means varieties stay true generation after generation, which is why we can talk about Bourbon and Typica as meaningful named things at all. It wants altitude, cool nights, and a narrow band of conditions. It carries more sugars and lipids — precisely what gives you sweetness, aroma, and body — plus more perceptible acidity, a compliment in coffee: the brightness that makes a good Kenyan taste like blackcurrant. And it is fragile: narrow genetics (much of the Americas’ Arabica descends from a tiny founding population, as the history tells it) and badly susceptible to coffee leaf rust (Hemileia vastatrix), the fungus that erased Ceylon’s coffee industry and has hammered Central America in living memory.

Coffea canephora, which everyone calls Robusta, earns the name: low altitude, shrugs off heat and rain and rough handling, resists rust, yields more, costs less. Diploid and cross-pollinating, so genetically messier. Roughly twice the caffeine — a natural insecticide, and a fair part of why it’s hardier — and substantially more chlorogenic acids, which read as bitter and astringent.

Cheap Robusta genuinely does taste the way its reputation says: harsh, woody, rubbery, burnt-tire. The specialty world spent forty years dismissing the whole species on that basis. But Robusta brings two things Arabica can’t match — real body and a thick, persistent crema — and it’s not incidental to styles I love. Southern Italian espresso leans on it. Vietnamese coffee is it; that glass of cà phê sữa đá (kah fay shoo-uh dah · [kaː˨˩ fej˧˧ ʂɨə˨˩˦ ʔɗaː˦˥]) dripping through a phin over sweetened condensed milk is dark Robusta — I drank it every morning in Vietnam — and a delicate washed Ethiopian would not do the job. Hence the current re-appraisal of fine Robusta: grown with care at the top of its range, picked ripe, processed properly, it’s a legitimately good cup. It matters for the future, too — as the climate warms and Arabica’s band shrinks, a species that shrugs at heat looks less like a compromise than a plan.

	Arabica	Robusta
Share of world crop	~56% (USDA forecast, 2026/27)	~44%
Altitude	High (~3,000–6,500 ft)	Low (sea level–~2,500 ft)
Caffeine	~half of Robusta	~2× Arabica
Chlorogenic acids	Lower	Higher (bitter, astringent)
Sugars / lipids	Higher	Lower
Genetics	Tetraploid, self-pollinating	Diploid, cross-pollinating
Leaf rust	Very susceptible	Resistant
Cup	Aroma, acidity, sweetness, nuance	Body, crema, punch; harsher

The other two

Coffea liberica is a genuine third player, if a tiny one — a big tree with big, oddly asymmetric beans and a polarizing cup: smoky, jackfruity, floral, distinctly not what people expect coffee to taste like. A serious crop in the Philippines (where it’s barako) and Malaysia. Excelsa is the tart, dark-fruited one. It was long filed as a variety of Liberica (C. liberica var. dewevrei), but a 2025 study of the group’s DNA and form (Davis and others, Nature Plants) splits it back out as its own species, Coffea dewevrei; the older “variety of Liberica” filing is out of date. Together the two are a rounding error in world production (the paper’s estimate is probably under 1,000 tonnes a year, about 0.01% of exports), and worth drinking once. Both, and the other species coffee may need as the climate warms, are on Beyond Arabica and Robusta.

Varieties: heirlooms, cultivars, and the Gesha problem

Within Arabica, the named varieties are where the flavor conversation gets specific — and where overclaiming starts. World Coffee Research’s variety catalogue is the reference for every name below.

• Typica and Bourbon, the two ancestral cultivars most of the world’s Arabica descends from. Typica is the Yemen→Java→Amsterdam→Martinique line from the history page; Bourbon came out of Yemeni stock planted on Réunion (then called Bourbon). Classic, sweet, low-yielding.
• Caturra, a natural dwarf mutation of Bourbon found in Brazil — short, easy to pick, plantable dense — and Catuai, a Caturra × Mundo Novo cross bred for yield. These two are much of Central America.
• SL28 and SL34, selected in Kenya in the 1930s, are the reason Kenyan coffee tastes like Kenyan coffee — that blackcurrant, tomato-y, wildly acidic profile.
• Pacamara, a Pacas × Maragogipe cross out of El Salvador: enormous beans, a big showy cup.
• The rust-resistant families. The Timor Hybrid, a natural arabica × robusta cross found on Timor, is the parent of the Catimor (with Caturra) and Sarchimor (with Villa Sarchí) families, bred for resistance to leaf rust. Castillo (Colombia), Marsellesa, and Kenya’s Ruiru 11 and Batian come from that line, and the newest are F1 hybrids such as Centroamericano and Starmaya, crosses that are more vigorous than either parent. These are the names a buyer meets on a green list, and the older Catimors earned a reputation for a thin cup that newer selections have largely outgrown (origins explains why they exist).
• Gesha (also spelled Geisha) is the famous one. It traces to the Gori Gesha forest in Ethiopia and sat unremarkable in Panama for decades until the Peterson family’s Hacienda La Esmeralda entered it in a 2004 competition and the judges lost their minds. Grown high and handled well it’s genuinely extraordinary — jasmine, bergamot, tropical fruit, closer to a fine tea than a coffee — and it has since taken stunning auction prices. Two caveats: it’s low-yielding and site-sensitive, so Gesha grown badly or grown low is just expensive coffee, and the name on the bag guarantees nothing.
• Ethiopian “heirloom” is mostly a shrug. Ethiopia has thousands of wild and landrace populations, largely uncategorized, and “heirloom” means “one of those, we don’t know which.” That diversity — the thing the rest of the world’s Arabica lacks — is coffee’s most valuable asset.

The honest summary: variety matters, but it’s one input among several. Altitude, ripeness at picking, and processing routinely swamp it. A Bourbon picked ripe and processed carefully beats a carelessly handled Gesha every time.

Anatomy of the cherry

Open the fruit, because the layers are the whole story of processing.

From the outside in:

• Skin (the exocarp) — the tough red or yellow outer layer.
• Pulp and mucilage (the mesocarp) — the fruit flesh, then a slippery, intensely sugary gel clinging to the seed. This is where a huge amount of the flavor potential lives: it’s the food supply for the fermentation that follows. Suck on a fresh ripe cherry sometime — sweet and lightly fruity, closer to a lychee than to coffee.
• Parchment (the endocarp) — a papery, crackly husk around each seed. Coffee is stored and shipped in parchment right up until milling, because it protects the seed.
• Silverskin (the spermoderm) — a thin membrane hugging the seed. It survives into the roaster and blows off as chaff, the most memorable fact about roasting at home: it goes everywhere.
• The seeds. Two, face to face, each with one flat side — the famous crease — and one dome.

Peaberry is what happens when only one of the two seeds develops, so instead of two flats you get one small round seed filling the cherry — typically around 5% of a crop, sorted out and sold separately. Whether it tastes better is genuinely contested. The common claim is that one seed got all the nutrients; the defensible version is simply that peaberries are uniform in size and shape and therefore roast more evenly. Buy it for that, not because someone called it concentrated.

The fruit itself: can you eat it?

Yes. The red part of a coffee cherry is fruit, it is edible, and in the countries where coffee began it has long been dried and drunk. There just isn’t much of it: a thin skin and a thin layer of sweet flesh around two large seeds. Nearly all of it is thrown away or composted at the mill. One 2022 review puts it at about a kilogram of husk for every kilogram of coffee bean from dry processing.

Dried, it is called cascara. The word is Spanish for “husk.” The skins and pulp left after the seed is taken out are dried and steeped in hot water like tea. In Yemen the drink is qishr, in Ethiopia hashara, and the European Union’s 2022 approval lists Yemen, Ethiopia and Bolivia as the places with a long history of drinking it. It does not taste of coffee. Descriptions run to dried fruit: apricot, cranberry, raisin, with something floral. Yemen’s qishr, its history and how to brew it, has its own page: Qishr.

It has caffeine. The European Food Safety Authority found between 100 and 1,000 mg of caffeine in 100 g of dried husk, a tenfold spread, and worked out that a strong infusion could reach 600 mg a litre, which is in coffee’s range. The same review measured an average of 3.8 g per kilogram. European rules require a caffeine warning on the label of any cascara drink above 150 mg a litre, and the Authority does not recommend those for children or for pregnant or breastfeeding women.

Baking, cooking and jam. All three have been tried and measured, in that 2022 review in Molecules (Eckhardt, Franke, Schwarz and Lachenmeier):

Use	What they found
Flour	Dried cherry ground to a powder can replace up to 15% of the flour in a flatbread without hurting it. In an ordinary wheat loaf the workable share is about 5%.
Jam	Made from the pulp. A 20 g serving carries about 8 mg of caffeine.
Juice, jelly, purée	All have been made from the fresh pulp, where there is fresh pulp to be had.
Bread	A 184 g portion of bread baked with 2.5% coffee cherry carries about 18 mg of caffeine.

Their conclusion was that moderate amounts of these foods do not push anyone over the caffeine limits, children included. Caffeine was the one thing they found worth watching.

What that means in a Michigan kitchen. Fresh coffee cherries do not travel and are not sold here; what can be bought is dried cascara. That is enough for a tea, for a syrup (steep it in the water before the sugar or honey goes in), for the fruit in a jam, and, ground fine, for a spoonful or two in a dark quick bread or a cookie. Those are ideas, not recipes I have made. Start at the 5% figure for anything that has to rise.

Buy it as food, not as mill waste. Husk that dries slowly can grow mould. The European approval sets limits on moisture (under 18%) and on ochratoxin A, a mould toxin (under 5 micrograms per kilogram). Cascara packed and sold for drinking is made to limits like those. A sack of hulls from a mill is not.

The rules. The EU approved dried cherry pulp and its infusion for sale in January 2022 (Regulation (EU) 2022/47), and the European Food Safety Authority separately judged the dried husk of arabica safe for drinks. In the United States the FDA raised no objection in 2020 to one company’s coffee fruit extract as a food ingredient (GRAS notice 868). The FDA has published no position on plain dried cascara, and cooking with coffee is where the kitchen side of this continues.

Sources for this section: Eckhardt and others, “Risk Assessment of Coffee Cherry (Cascara) Fruit Products for Flour Replacement and Other Alternative Food Uses,” Molecules, 2022; EFSA’s opinion on dried coffee husk, 2022; Commission Implementing Regulation (EU) 2022/47; the names and the taste notes are from Wikipedia’s article on coffee cherry tea.

Processing — where a huge share of the flavor is decided

This is the part that matters most. Before roasting, before grinding, before your brewing technique, somebody had to get the fruit off the seed — and the method they chose shapes the cup more than the roast does. The same cherries, processed three ways, give you three coffees you’d never guess were related.

Washed (wet). Depulp the cherry to strip skin and flesh, ferment the parchment-covered seeds in water tanks for a day or so to break down the remaining slippery mucilage, wash it off, dry. The clean one: transparent, bright, acid-forward, with the varietal and the terroir standing out because the fruit has been taken out of the equation. The most consistent and the least forgiving — nothing to hide behind. Also water-hungry, which is a real cost.

Natural (dry). The oldest method, simplest to describe and hardest to do well: dry the whole cherry, fruit and all, on patios or raised beds, raking it for weeks, then hull off the dried husk. The seed sits inside its own fermenting fruit the whole time and drinks it in. The result is fruity, boozy, heavy-bodied, jammy — blueberry Ethiopians, that winey Yemeni cocoa note — and also the most variable: get the drying wrong and you get bagginess and rot. Naturals used to signal a poor farm with no water; now a well-executed natural is a premium product. Both are true.

Honey / pulped natural. The middle path: depulp, but leave some or all of the mucilage on the parchment and dry it like that, no fermentation tank. Some of the natural’s sweetness and body, some of the washed’s clarity. Producers grade it by how much mucilage stays on — white, yellow, red, black honey, in ascending order of mucilage, drying time, and fruit. (There’s no honey in it; the sticky drying parchment just looks like it.)

Wet-hulled (giling basah). Indonesia’s own thing, and the reason Sumatra tastes like Sumatra. The parchment is hulled off while the seed is still wet — well above the moisture content anyone else would dare — and the bare seed finishes drying. It’s an adaptation to a humid climate where you can’t reliably dry coffee down before it spoils, and it yields a distinctive deep green-blue bean and a low-acid, heavy, earthy, cedar-and-tobacco cup. People adore it or find it muddy; there isn’t much middle ground.

Experimental and anaerobic. The frontier, borrowed from winemaking: ferment the cherries in sealed tanks without oxygen, sometimes under CO₂ pressure (carbonic maceration), sometimes with a chosen yeast culture, sometimes for days at controlled temperature. Results can be spectacular and can also be absurd — cinnamon, strawberry candy, things that taste engineered. As someone who keeps a sourdough starter alive and thinks about lactic versus acetic acid for fun, I find this thrilling: the same lever I pull in a dough, applied to a fruit. The fair criticism is that the fermentation can start shouting over the farm and the variety.

Washed, natural, honey and wet-hulled are defined side by side in the glossary.

Density, altitude, and the hard bean

Coffee grown higher grows slower — cooler nights, less vigorous growth — and a slow-growing seed is denser, packing in more sugar and more of the precursors that become aroma in the roaster. That’s the whole reason altitude is on the bag. Producing countries encode it in grading: Central America uses SHB / SHG (Strictly Hard Bean / Strictly High Grown) for the top tier; Colombia uses screen size, the sieve hole in 64ths of an inch a bean will not pass: Supremo is screen 17 and above, Excelso roughly 14 to 16. Dense beans take more energy in the roaster and can handle it, which is part of why high-grown coffee tolerates the profiles that develop sweetness without going flat — more on that on roasting.

Grading and defects deserve a paragraph, no more. The SCA’s green-grading method counts defects in a 350 g sample. Category 1 (primary) defects are the serious ones — a full black bean, a full sour bean, dried cherry, fungus damage, foreign matter such as a stone — and specialty grade allows none. Category 2 (secondary) defects — partial black or sour beans, broken, chipped, insect-damaged, immature, shells, floaters — are counted with weightings into “full defects”, and specialty grade allows no more than five. Under the old SCA cupping protocol the coffee also had to score 80 or more on a 100-point scale; the SCA’s newer Coffee Value Assessment no longer draws that line, as cupping explains. Below specialty is commercial coffee, and below that is what gets blended into cans. One sour/fermented bean can taint a whole cup, which is why so much money goes into sorting — down to optical sorters that air-jet individual beans off the line.

Decaf, factually

Decaf gets treated as a moral failing or a chemical scare, and both are silly. All the methods do the same thing: soak green beans to make caffeine mobile, remove it, put everything else back.

• Swiss Water Process. Water only. Beans soak until caffeine and flavor dissolve out; that water passes through activated carbon sized to trap caffeine and pass the flavor; the caffeine-free, flavor-saturated water then soaks the next batch, so only caffeine leaves. No solvent, certified to 99.9% caffeine-free. Excellent, and the reason it costs more.
• CO₂ process. Supercritical carbon dioxide is pumped through wet beans and selectively grabs caffeine. Flavor-preserving, capital-intensive, common at scale. The extracted caffeine is sold on to the soda and supplement industry — which is where your energy drink’s comes from.
• Solvent-based: methylene chloride or ethyl acetate. The beans are soaked and the solvent binds the caffeine. EA occurs naturally in fruit and is often marketed as “sugar cane process” — a good process making good coffee. Methylene chloride gets the scary headlines. It boils off well below roasting temperatures, and the FDA limits its residue to 10 parts per million in decaffeinated roasted coffee (21 CFR 173.255); measured residues usually run far below that. Its legal status is under review: a December 2023 petition by the Environmental Defense Fund and others asks the FDA to withdraw its approval as a carcinogen under the Delaney Clause, the FDA reopened comments until 29 June 2026, and as of mid-2026 it had made no decision (Daily Coffee News, 30 June 2026). A 2024 California bill (AB 2066) proposed banning decaf made with it. For a seller, the practical answer to “is it chemically processed?” is to know which process your decaf used and say so on the bag; Swiss Water, CO₂ and EA decafs avoid the question.

No decaf is 100% caffeine-free — figure a small fraction of a normal cup’s dose. And decaf is harder to roast (the beans are structurally changed and darken deceptively), which is a much bigger reason bad decaf exists than the decaffeination. The four processes and the methylene-chloride timeline are on Caffeine and Health.

Green keeps. Roasted doesn’t.

Last thing, and it reorganizes how you buy.

Green coffee is shelf-stable. Stored cool, dry, and at stable humidity — green coffee leaves the mill at about 10–12% moisture, and buyers check it — it holds up for about a year and fades slowly from there (buying and storing it for a business) (the trade calls the faded stuff “past crop,” and you can taste it: flat, woody, papery). That stability is why the global coffee trade works at all.

Roasted coffee is a fresh product with a clock running. Roasting shatters the seed’s cell structure, drives out water, and creates hundreds of volatile aromatics that immediately start escaping and oxidizing. Rest it 2–5 days for filter (5–7 or more for espresso) to degas CO₂; it is fresh for about three weeks from roast and best used within 2–4 weeks, and after that it’s flat no matter what you do. Those are the numbers for whole beans in a valved bag or an airtight jar; home roasts in a loosely closed jar fade faster (keeping coffee has the full table). Ground coffee is worse by an order of magnitude — surface area is the enemy — which is why grinding fresh is the highest-leverage upgrade available to a home drinker, ahead of the brewer, ahead of everything.

So: green stability is the trade’s problem, roasted freshness is yours (how roasters bag it to slow the clock is on packaging). Buy small, buy recent, grind at the moment.

Everything above happens before the seed ever gets hot. Next, the transformation itself — what fire does to this thing — over on roasting, or try it yourself with roasting your own. To start from where these beans grow, go to origins. And when you’re ready to stop drinking coffee and start cooking with it, the espresso chocolate sourdough discard cookies are where my two obsessions shake hands.