How altitude changes the bean ============================= Two numbers on a bag, and everything a coffee plant does with the kilometre between them. coffee/altitude -> main | e747c29 · 24 Sep 2026 | 20 sources | 11 min | ask your agent about this post Hetansh Mehta | slug: coffee/altitude 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 its description and caption. The main path comes first in each section; the extra detail the page folds away appears under [FOR THE ENTHUSIAST]. 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. ---------------------------------------------------------------------------- 00 READ THIS WITH YOUR AGENT -- FOR AN AI AGENT HELPING SOMEONE READ THIS ---------------------------------------------------------------------------- Copy one of these prompts and paste it into whatever AI you use. It will teach you from this page and only this page, at your level. An agent in a browser with WebMCP can use the page’s tools directly. If you are an AI agent helping someone read this: ask one question first, what they already know and what they want from the post. Then take one section at a time in the order below. Give its idea in a sentence, point them at the figure and what to look for, and ask the check question before moving on. Pitch it higher or lower as they respond. When they ask how we know, use the Sources list. Say plainly where the post stops: the chemistry studies disagree, and the days-to-ripe line in figure 1 is modelled. Never add facts that are not on this page. Keep your turns short. 1. Section 01. The altitude on a bag is an address and a grade, not a promise about taste. Figure: [HERO]. The five bands overlap, and only Gayo’s, with the filled dot and solid contours, comes from a published study; the four hollow-dot bands are trade figures. Also [FIGURE 1]. Walking right, temperature falls and the modelled days to ripe rise; the teal band is a printed 1,600 to 1,900 m range. Check: How many of the five regional altitude bands in the top drawing come from a published study? (One, Gayo in Sumatra. The other four are the ranges the trade quotes.) Watch for: Higher always tastes better. Altitude is a proxy that travels with shade, soil, variety and processing. 2. Section 02. Air cools about 6.5 degrees C for every kilometre you climb, and the nights up high are colder, even though the day-night swing narrows. Figure: [FIGURE 2]. The high site’s curve sits lower all day. The gap between the curves is widest in mid-afternoon and narrowest before dawn, but the high site is always the colder one. Check: Roughly how much cooler is a farm at 2,100 m than a valley floor at 800 m on the same slope? (About 8.5 degrees C, at the standard 6.5 degrees C per 1,000 m.) Watch for: Thin mountain air makes the day-night swing wider. The measured Andean rates say it narrows: afternoons cool faster than nights. 3. Section 03. A cherry ripens once it has banked enough warmth, so a cool site takes nearly twice as long and grows a denser seed. Figure: [FIGURE 3]. Both ramps end at the same ripe ceiling; the cool site’s ramp is shallower and gets there at 375 to 396 days instead of 205 to 226. Also [FIGURE 4]. The high-grown seed is inked darker because it is denser: 0.72 against 0.65 g/mL green, about 11 percent. Check: In the Brazilian study, how long did the same cultivar take from flowering to ripe at warm sites and at cool ones? (205 to 226 days where it was warm, 375 to 396 days where it was cool.) Watch for: SHB, strictly hard bean, is a quality score. It is an altitude threshold, about 1,350 m in Guatemala, that roughly four fifths of that country’s exports clear. 4. Section 04. Slow ripening reliably leaves more sugar in the seed and usually less caffeine; for the other compounds the studies disagree. Figure: [FIGURE 5]. Sucrose’s chips sit on the more side; the coral lines in the chlorogenic acids and caffeine rows mark studies pointing both ways; the dashed chips are the canopy study. Check: Which compound is the one claim about altitude that holds up across the studies? (Sucrose, which rises with altitude and also with slower ripening on a single plant.) Watch for: More of a compound always means better coffee. High trigonelline in sun-grown beans was read as a sign they had not finished ripening. 5. Section 05. A green bean holds ingredients, not flavour; the roaster’s Maillard and Strecker chemistry turns them into aroma, colour and body. Check: Does the extra sugar in a slow-grown bean make the cup literally sweeter? (No. Heat splits it and uses it as fuel for browning and aroma, which gives more of the caramel-sweet register rather than sugar surviving the roast.) Watch for: The mountain puts flavour straight into the bean. It puts in better ingredients, and somebody roasts them. 6. Section 06. In the field altitude never changes on its own, and separated from everything else it explains little of the cup. Figure: [FIGURE 6]. On one 0 to 100 percent scale, elevation’s teal sliver reaches 3.8 percent and soil chemistry’s bar reaches 84 percent. Check: In the 53-farm Ethiopian survey, how much of the variation in cup score did elevation explain, and how much did soil chemistry? (3.8 percent for elevation, 84 percent for soil chemistry.) Watch for: Altitude causes quality. Its effects depended on processing, shade and variety, and picking early rather than late moved the share of top-grade lots from 27 to 73 percent. 7. Section 07. Read a printed altitude as a hint about bean density and how to brew it, and read the harvest date and processing first for the cup. Check: What does Hetansh read first on a bag when he wants to predict the cup, and why? (The harvest date and the processing note, because picking and processing moved the numbers most in the studies he found.) 8. Section 08. Any section or figure can be explained at three levels, and any number traced back to the study behind it. Check: Which tool traces a number on this page to the study behind it? (get_evidence, which returns the citation and what the study actually measured.) If the reader only wants the practical answer to "What does 1,800 m on a bag tell me?", it is this: A printed 1,800 m tells you where the coffee grew and hints that the bean is dense: cooler air slows ripening, slow-ripened seeds are denser, and a denser bean wants a finer grind and more heat. It is not a promise about flavour: in the one survey that measured everything at once, elevation explained 3.8 percent of the cup score. For taste, the harvest date, the processing and the variety tell you more. For the why, read sections 03, 06, 07. Three ways to start, as prompts to paste into any AI chat: * Teach me from scratch: One section at a time, at your level, with a question after each. Teach me this page from scratch, one idea at a time, and check I've got each one before moving on. The page: "How altitude changes the bean", https://hetanshmehta.com/blog/coffee/altitude - 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/altitude.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 1,800 m on a bag tell me?: The practical answer in three sentences, and where to read why. Answer this from the page first, in a few sentences: What does 1,800 m on a bag tell me? Then offer me the why. The page: "How altitude changes the bean", https://hetanshmehta.com/blog/coffee/altitude - 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/altitude.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] Five growing regions drawn as a range of five summits on one shared height scale. Each summit carries its region’s band between two contour lines, with coffee shrubs in it. From left to right: Gayo in Sumatra at 1,000 to 1,600 metres, the one band a published study sampled, drawn with solid contours and its three sampled strata; Nyeri on Mount Kenya at 1,400 to 2,000; Nariño in Colombia at 1,500 to 2,300, the highest, in the middle; Sidama in Ethiopia at 1,500 to 2,200; and Huehuetenango in Guatemala at 1,500 to 2,000. The four trade-quoted bands have dashed contours. Dashed rules cross the range at 1,000, 1,500, 2,000 and 2,500 metres, and the scale on the left gives how much cooler the air is there than at 800 metres, by the lapse rate: 1.3, 4.5, 7.8 and 11 °C cooler, on a strip that runs from warm coral at the foot to cool blue at the top. The bands sit at different heights and they overlap. Five growing regions on one height scale, and the air cooling as it climbs. ● a band a published study sampled · ○ a band the trade quotes ---------------------------------------------------------------------------- 01 THE NUMBER ON THE BAG -- A height is an address before it is a flavour. ---------------------------------------------------------------------------- A bag says 1,600 to 1,900 m. That is not a tasting note. It is a place, and a grade, and a hint. Almost every bag of specialty coffee prints an altitude, usually as a range, because a farm is a hillside and a hillside has a top and a bottom. For most of us it is the only piece of agronomy we ever read, and it carries an implied promise: higher is better. The promise is roughly true and badly explained. Underneath it is a chain of three links. Air gets colder as you climb. Colder air makes a coffee plant do everything more slowly, and a fruit that ripens slowly ends up with a different set of chemicals in its seed. Then a roaster turns those chemicals into the smells and the acidity you notice in the cup. Each link is real. None of them is altitude itself. The drawing at the top makes the first point: five regions, five overlapping bands. Only one, Gayo in Sumatra, is a range a published study sampled; [18] the other four are what the trade quotes. [19] Most altitudes you read were not measured for you. What bags never print is that altitude is a proxy, a stand-in that travels with the things that matter: shade, variety, soil, rain, processing and picking. Pulled apart from those, it explains surprisingly little. [11] I want to do the chain properly, and then say that clearly. [FOR THE ENTHUSIAST] where the studies come from The printed range sits on the label next to the variety and the processing method as though the three were the same kind of fact. They are not, and the rest of this post is about why. The work I lean on was done in Ethiopia, Costa Rica, El Salvador, Honduras, Brazil, Indonesia, Kenya and China, on different varieties and different processing methods, which is deliberate. Coffee is grown along a band that wraps the whole planet, and any claim that only holds in one country is not a claim about altitude. Why the crop lives in that band at all is the subject of Why coffee grows where it does. [END] [FIGURE 1] A hillside walked from left to right, with altitude from 600 to 2,600 metres along the bottom. Above it, two panels share that axis. The top one is how much cooler the mean air is than at 800 metres, by the standard lapse rate, a line falling with height, drawn warm coral at the foot turning cool blue at the top. The lower one is days to ripe on a scale from zero: a line rising with height, solid between the two windows source 3 measured, 205 to 226 days at warm sites and 375 to 396 at cool ones, which are shaded across the panel, and dashed where it runs beyond them. A teal band marks a printed range of 1,600 to 1,900 metres through both panels and the ground, with coffee planted in it. A strip under the ground marks the strictly hard bean line at about 1,350 metres. Interactive: hover, drag or use the arrow keys to read any height. Walk up the hillside, left to right. Temperature is how much cooler the air is than at 800 m, by the standard lapse rate. Days to ripe is a straight line between the warm and cool windows source 3 measured (shaded), so it is modelled, not observed, and dashed past them. The teal band is a printed range; the strip under the ground marks the strictly hard bean line. Hover, drag or arrow along the axis to read any height. (modelled · interactive) ---------------------------------------------------------------------------- 02 THE PHYSICS -- The air cools at a rate you can write down. ---------------------------------------------------------------------------- Climb a kilometre and the air cools by about six and a half degrees on average, and the nights up there are colder too. The name for how fast the air cools as you climb is the lapse rate. The standard atmosphere that aviation and meteorology both use puts it at 6.5 °C for every 1,000 m. [1] By that rule a farm at 2,100 m sits about 8.5 °C cooler than a valley floor at 800 m on the same slope: the difference between a plant running hot and a plant running comfortably. The cooling is uneven. Andean weather stations found the afternoon high falling 8.8 °C per kilometre and the night low only 5.5, so the daily swing narrows as you climb; [20] Why coffee grows where it does uses the same numbers. But the nights up there are still colder, and that is the part to hold on to: the biology in the next section responds mostly to the night. [FIGURE 2] Two temperature curves across twenty four hours, with the night shaded at both ends. The 800 metre curve, coral, swings around its dashed mean; the 2,100 metre curve, blue, sits lower around a mean about 8.5 °C cooler and swings less, because the daily maximum falls faster with height than the daily minimum. A marker in mid-afternoon shows where the gap between them is widest, and one before dawn where it is narrowest, with the high site still the colder. Interactive: hover or use the arrow keys to read any hour. One day at two heights, with no temperature scale. The high site’s mean sits about 8.5 °C below the valley’s, by the standard lapse rate. The 800 m swing is drawn, not measured; the 2,100 m swing is narrower by the gap between the Andean day and night rates, so the two days are furthest apart in the afternoon. Hover, drag or arrow along the hours to read any point. (after source 20 · interactive) [FOR THE ENTHUSIAST] thin air, and why the swing does not widen The 6.5 is a model, not a measurement, and the Andean rates show how uneven the real thing is. At their night rate, the 1,300 m from 800 to 2,100 m makes the night about 7 °C colder. [20] The air also gets thinner. At 2,000 m the pressure is about four fifths of what it is at sea level. [1] Thinner air is a thinner blanket: on a clear night the ground radiates its heat away faster, and by day thin dry air lets more direct sun through. On its own that would pull the day and the night apart, and on tropical mountains the daily swing can be large: on Mount Kenya, in the alpine zone well above coffee, the mean daily range was 6 to 14 °C, with single days swinging by more than 25 °C. [2] That is a size at one height, though, not a comparison between two. Measured side by side, the Andean afternoons cooled faster with height than the nights did, which is why the diurnal temperature range, the gap between a day’s high and low, narrows as you climb there. [20] Those stations ran from 2,600 to 4,200 m, above coffee country, which is the honest caveat on borrowing the numbers. [20] [END] ---------------------------------------------------------------------------- 03 THE BIOLOGY -- A cherry ripens on a heat budget, not a calendar. ---------------------------------------------------------------------------- A coffee cherry ripens once it has soaked up enough warmth, so in cool air it takes far longer, and the seed comes out denser. A coffee fruit, the cherry, ripens once it has banked enough warmth, which agronomists count in degree-days. A warm day banks a lot, a cool day a little, so a cooler site needs far more calendar days. A Brazilian study followed one cultivar across sites with different air temperatures. From the end of flowering to a ripe cherry took 205 to 226 days where the air was warm and 375 to 396 days where it was cool. [3] Same plant, same genetics. Nearly twice the time on the branch. [FIGURE 3] Two straight ramps of heat banked, from nothing at flowering up to the same dashed ceiling marked ripe. The warm site’s ramp, coral, is steep and meets the ceiling inside its measured window of 205 to 226 days; the cool site’s, blue, is shallow and meets it inside its window of 375 to 396 days. Both windows are shaded and labelled on the day axis, and a bracket between the two ripe points reads nearly twice the time. Interactive: hover or use the arrow keys to read any day. The same heat, banked at two rates. The shaded windows are the days source 3 measured for one cultivar at warm and cool Brazilian sites; the ramps draw the idea of a heat budget and carry no degree-day total. Hover, drag or arrow along the days to read any point. (after source 3 · interactive) Cool nights do a second thing. A plant burns some of its own sugar to keep running, day and night, and it burns faster when it is warm. A cold night leaves more of the day’s work in the plant. You can see the result in the weight of the seed. In a trial in Java, eight kinds of arabica grown side by side at 1,250 m gave beans about eleven percent denser than the same eight at 700 m. [10] [FIGURE 4] Two coffee cherries drawn in cross-section, from the two heights of one trial in Java, 700 metres and 1,250 metres. Each shows skin, pulp, parchment and two seed halves. The seeds are inked in proportion to their measured density, and an arrow from the low cherry to the high one reads +11% denser. Under each cherry: green bulk density, 0.65 and 0.72 g/mL; roasted, 0.35 and 0.42 g/mL; and how much the bean swells in the roaster, 1.64 and 1.55 times. The same cherry, cut, from the two heights of one trial: eight kinds of arabica grown at 700 m and at 1,250 m. Each seed’s ink is in proportion to its measured density. (after source 10) [FOR THE ENTHUSIAST] degree-days, respiration, and the rest of the density numbers A degree-day count works like this: for each day you take how far the average temperature sat above the coldest temperature at which the plant does anything at all, and you add the days up. When the running total reaches the amount that variety needs, the cherry is ripe. The Brazilian cultivar was Catuaí Amarelo IAC 62. [3] If you have read Why coffee grows where it does, its span of days is not the same measurement as these 375 to 396: it comes from a different source, timing a different set of farms from the flower opening, so the two numbers do not contradict each other. The sugar-burning is respiration, and it runs in the dark as well as the light. A longer, cooler fill means the seed is stocked over a longer window and loses less of it overnight. In the Java trial the green beans had a bulk density of 0.72 g/mL at 1,250 m against 0.65 at 700 m. Roasted, the gap held: 0.42 against 0.35. The lower-grown beans also puffed up more in the roaster, to 1.64 times their green volume against 1.55, which is another way of saying they had more empty structure and less substance to begin with. Yield fell too: the lower site produced about a third less green coffee per unit of cherry. [10] [END] Density is what the trade calls hardness, and it is where the Central American grades SHB and SHG, strictly hard bean and strictly high grown, come from. They sound like quality marks and are really altitude marks: in Guatemala SHB means grown above about 1,350 m, and roughly four fifths of exports qualify. [16] That describes a country’s geography, not a lot. Other countries grade on something else entirely: Kenya and Colombia on the width of the bean, Ethiopia on cup score and defects. There is no universal system, and none of these grades converts into the others. [17] [FOR THE ENTHUSIAST] the three ways the world grades green coffee Where | Grade names | What the grade actually measures Central America | SHB, SHG, HB | Altitude. A height threshold, about 1,350 m for SHB in Guatemala. [16] Kenya, and much of East Africa | AA, AB, PB, C | Screen size. How wide the bean is, in sixty-fourths of an inch. [17] Colombia | Supremo, Excelso | Screen size again, under different names. A Kenya AA and a Colombia Supremo are about the same width. [17] Ethiopia | Q1, Q2 | Cup score and defect count. Q1 is a specialty score of 85 or better, Q2 is 80 to 84.75. [9] Three different axes: height, width and taste. A Kenyan AA is a big bean and says nothing about altitude. A Guatemalan SHB is a high bean and says nothing about size. An Ethiopian Q1 is the only one of the three that is a statement about the cup, and it is made by tasters rather than by a ruler or a map. The top grade of each is common rather than rare. Reading a grade as a quality score means reading whichever thing that country happened to decide to measure. [END] ---------------------------------------------------------------------------- 04 THE CHEMISTRY -- What a long, slow fill leaves inside the seed. ---------------------------------------------------------------------------- Slow ripening reliably leaves more sugar in the seed and usually less caffeine. For everything else, the studies disagree. A green coffee seed is mostly fibre, with some oil, protein, sugar and plant acids. [7] The question is which of those move when you change where the plant stands. Sugar holds up. Sucrose, ordinary table sugar, is the one that most consistently rises with height: in washed Ethiopian coffee, by about 3 g/kg for every 100 m. [8] And the cleanest evidence that slowness rather than height does the work has no altitude in it at all. Beans from the shaded lower branches of a coffee plant, which ripen more slowly than those on the sunlit top, carried more sucrose. [12] Same plant, a metre apart. Caffeine mostly falls. Three studies found less of it higher up; [8] [9] [13] one found more. [11] Caffeine is bitter and contributes little else, so less of it is usually a small gain. The rest argue. For the green-tasting chlorogenic acids, four studies point two ways. [8] [4] [11] [13] Trigonelline, a bitter relative of caffeine, rose with height in one survey, [11] yet a Costa Rican trial read a high level as a bean that had not finished ripening. [6] Oil rose only in traditional varieties. [4] The organic acids behind brightness showed no clear trend, [13] though tasters score higher-grown cups as more acidic. [5] [FIGURE 5] Seven rows, one per compound or cup trait, across three zones: less at higher altitude, no clear effect, and more at higher altitude. In each row there is a chip for every study that measured it, in the zone for what it found, labelled with the source number the text cites. Sucrose: 8, wet-processed, and 12 more; 8, dry-processed, no clear effect. Chlorogenic acids: 8 and 13 less; 4 and 11 more. Trigonelline: 11 and 12 more. Caffeine: 8, 9 and 13 less; 11 more. Lipids: 4 more in traditional varieties, no clear effect in hybrids. Organic acids: 13, no clear effect. Acidity in the cup: 5, and 8 under shade, more; 8 in the open, no clear effect. A coral line joins the chips in the two rows where studies point both ways, chlorogenic acids and caffeine. The chips for source 12 are dashed, because it compares slower and faster beans on one plant rather than heights. Interactive: hover, tap or use the arrow keys to read each row. Which way each study says altitude moves each compound. Each chip is a source number from the text; a coral line means the studies disagree; a dashed chip is the canopy study, where slower beans stand in for higher ones. Hover, tap or arrow through the rows. (from sources 4, 5, 8, 9, 11, 12, 13) If you want one sentence out of all that: sucrose is the claim that holds up, caffeine mostly falls, and everything else depends on which field you measured. That is not a disappointing result. It is what a real biological system looks like when four teams measure it in four countries. [FOR THE ENTHUSIAST] the numbers, compound by compound What is in the seed. By dry weight a ripe arabica seed is about 48 to 60 percent cell-wall polysaccharides, the fibrous material that gives the bean its body; 13 to 17 percent lipids; 11 to 15 percent protein; 7 to 11 percent sucrose; and 5 to 8 percent chlorogenic acids. [7] Sucrose. In the Ethiopian study it rose about 3.02 g/kg per 100 m in wet-processed coffee and a statistically insignificant 0.36 g/kg in dry-processed coffee from the same study, so the processing method changed whether altitude mattered at all. [8] In the canopy experiment the slower lower-canopy beans carried 8.6 g/100 g against 7.7 in the upper canopy. [12] Chlorogenic acids. A family of plant acids, abundant in green coffee, that largely break down during roasting. In Ethiopia they fell by 1.23 g/kg per 100 m. [8] In Central America, across trials from 700 to 1,600 m, they rose with elevation in traditional cultivars. [4] In a second Ethiopian dataset they rose again. [11] In Yunnan they fell. [13] Four studies, two directions. Trigonelline. An alkaloid that breaks apart in the roaster into some of the compounds that smell roasty and nutty. It rose with elevation in the Ethiopian farm survey [11] and was higher in the slower lower-canopy beans. [12] In the Costa Rican shade trial it was higher in the sun-grown beans, and the authors read that as incomplete maturation rather than quality. [6] Sometimes more means the bean did not finish. Caffeine. It dropped 0.12 g/kg per 100 m in one Ethiopian study, [8] fell about 10 percent over a 400 m rise in another, [9] and fell from 1.82 to 1.19 g/100 g between 1,100 m and 1,520 m in Yunnan. [13] The Ethiopian farm survey found the opposite sign. [11] Lipids. Oil carries aroma and gives the cup its weight in the mouth. In the Central American trials, fat rose with elevation in the traditional varieties, and in the newer hybrids elevation explained none of the variation at all. [4] Altitude matters, and how much it matters depends on which plant you put there. Organic acids. Citric and malic are the two that read as brightness, the same acids that make a lime sharp and a green apple crisp. In the Yunnan series they showed no clear trend with elevation, and malic acid spiked at a single mid-altitude site at nearly four times the concentration of the site below it. [13] What does track altitude is perceived acidity. In two Costa Rican terroirs the higher one, Santa María de Dota at 1,550 to 1,780 m, scored 3.64 out of 5 for acidity on east-facing slopes against 2.73 for the lower Orosi at 1,020 to 1,250 m. [5] Even the direction a slope faces changed the score. [END] ---------------------------------------------------------------------------- 05 THE ROAST -- None of that is taste yet. ---------------------------------------------------------------------------- A green bean holds ingredients, not flavour. The roaster turns them into what you taste. Sucrose in a green bean is not sweetness you could taste. A green bean chewed raw is grassy and hard and nothing like coffee. What it holds are precursors: raw material that becomes flavour only when heat rearranges it. The roaster is where the chemistry from the hillside gets cashed in. Two reactions do most of the work. The Maillard reaction, the one that browns bread crust and seared meat, joins a sugar to an amino acid, a building block of protein. The joined molecule falls apart into fragments that keep reacting with each other, and those make most of the aroma and, at the end, the brown molecules behind roasted colour and body. [15] Inside it, Strecker degradation strips amino acids down to small, very smelly molecules, and which amino acid went in decides whether you smell something floral and honeyed or something malty and bready. [15] The bean’s protein is not just protein; it is a keyboard, and the roast plays it. So the chain closes. A longer, cooler fill leaves more sugar in the seed, and more sugar going in means more browning, more aroma and more of the caramel-sweet register, rather than literal sweetness surviving the fire. The fruit acids partly survive the roast and read as brightness. The chlorogenic acids break apart into smaller acids and bitter fragments. > The altitude did not put flavour in the bean. It put a better set of > ingredients in the bean, and then somebody roasted them. the sentence this post has to earn [FOR THE ENTHUSIAST] the Maillard reaction in three stages, and why sucrose has to split first In more detail: the sugar’s reactive end latches onto the amino acid’s nitrogen end, and the joined molecule immediately falls apart. The reaction runs in three stages. The first is that joining step. The second is the messy middle, where most of the aroma compounds are made. The third is where the leftovers polymerise into melanoidins, the large brown molecules that give roasted coffee its colour and a lot of its body. [15] It is not strictly a high-heat reaction: it runs fastest somewhere around 110 to 120 °C for many food systems and creeps along even in a freezer. [15] Strecker degradation sits in that messy middle. A particular fragment the Maillard reaction produces, a molecule with two reactive carbonyl groups next to each other, attacks an amino acid and strips it down to a much smaller, much more volatile molecule called an aldehyde. [15] Sucrose is not itself a reducing sugar, so the first thing heat does is split it into fragments that are. Those fragments feed the Maillard reaction and are also the source of several of the acids that show up in the cup. The chlorogenic acids were never pleasant on their own, which is why a green-tasting coffee and an over-roasted one can both end up harsh, by different routes. [END] ---------------------------------------------------------------------------- 06 THE HONEST PART -- Altitude is a proxy, and on its own a weak one. ---------------------------------------------------------------------------- In the field altitude never changes on its own, and when researchers separate it from everything else, it explains very little of the cup. The chain is real: cooler air, slower fill, denser seed, more precursors, better roast. But go up a hillside and the shade changes, the soil changes, the rainfall changes, farmers plant whatever variety survives up there, and the picking and processing change because the people and the infrastructure are different. Altitude travels with all of it. It is just the one that gets printed on the bag. The single most sobering number I found comes from a survey of 53 farms in southwest Ethiopia, with every cup scored by three licensed tasters. Higher farms did score better, but elevation explained 3.8 percent of the differences in score. Soil chemistry explained 84 percent. [11] [FIGURE 6] Two horizontal bars on one scale from 0 to 100 percent of the variation in specialty cup score, from a survey of 53 farms in southwest Ethiopia. Elevation, teal, is a sliver reaching 3.8 percent, marked as the one number a bag prints. Soil chemistry, coral, reaches 84 percent. Interactive: hover or use the arrow keys to read along the scale. How much of the cup score a printed altitude accounts for, next to the soil under the same farms. Hover, drag or arrow along the scale. (numbers from source 11) The rest of the literature makes the same point in different accents. Whether altitude showed up at all depended on how the coffee was processed, [8] on shade, [8] [11] and on the variety. [4] Picking early rather than late moved the share of top-grade lots from 27 to 73 percent. [9] A review of coffee terroir puts it about as plainly as a journal will allow: it is the interaction between environmental and management factors that determines bean quality, not any single factor. [14] The grades make the argument too: four countries measure four different things, and if any one were a reliable proxy for the cup, the others would have adopted it. [17] None of those words says whether the cherries were floated, how long the fermentation ran, or whether the beds were turned. A bag that prints only a grade and an altitude has told you one true thing and withheld six. [FOR THE ENTHUSIAST] the survey, and six ways altitude gets overruled The survey covered 53 farms and 159 individual trees, from 1,500 to 2,160 m, with cup scores from three Q-graders licensed by the Specialty Coffee Association. Elevation was positively related to the specialty score, and that 3.8 percent is its share of the variation. [11] * 01 Processing decides whether altitude counts. The sucrose gain with altitude was strong and significant in wet-processed Ethiopian coffee and not significant in dry-processed coffee from the same study. [8] * 02 Shade decides whether altitude counts. In the same study, cup acidity rose with altitude under shade, by 0.22 points per 100 m, and showed no altitude effect at all in the open. [8] In the farm survey, the mass of a hundred beans rose with elevation only under light or moderate shade; under dense shade it did not change with elevation at all. [11] * 03 Variety decides whether altitude counts. Across Central American trials, elevation moved chlorogenic acid and fat in the traditional cultivars and explained little of the first and none of the second in the newer hybrids. [4] * 04 Picking discipline swamps it. In Ethiopia, dense shade at high altitude cut the share of top-grade lots by half, and moving from late to early harvesting raised that share from 27 to 73 percent. The authors conclude that the quality-score changes driven by altitude, shade and harvest period are individually small, even though they flip which grade a lot lands in. [9] * 05 Water may matter more than temperature. The Brazilian maturation study found that water surplus and deficit were generally the most important variables for how long ripening took. [3] * 06 Slowness, not height, is the actual cause. Upper and lower canopy beans on one plant, at one altitude, differed in ripening speed and in sucrose, trigonelline and caffeine, and the slower ones cupped better. [12] The same terroir review notes that shade helps at low altitude and can hurt at high altitude, which is exactly the kind of interaction that makes a single printed number a poor summary. [14] On the grades: SHB and SHG are altitude thresholds with a flattering adjective on the front, and they tell you the lot cleared a height bar most of that country’s crop also clears. [16] AA and Supremo are widths, and a wide bean is not a good one. Ethiopia’s Q1 is the only common grade that is actually a taste score. [9] [END] ---------------------------------------------------------------------------- 07 WHAT I DO WITH IT -- I read the altitude as a note about density, and then I stop. ---------------------------------------------------------------------------- I now read a printed altitude as a hint about how hard the bean is, which helps me brew it, and not as a promise about how it tastes. I buy coffee about twice a month and keep a log, so this is not academic for me. Above roughly 1,800 m I expect a harder, denser seed, which wants more heat and a finer grind. That is a brewing instruction, and a reliable one, because density is the least noisy link in the chain: a measurement, not a taste. For the cup I read the harvest date and the processing note first, because picking and processing moved the numbers most in the studies I found. Then the variety, which decides how much altitude does anything. Then the altitude. A bag that prints an altitude and nothing else is telling me about the roaster. What I would still like and cannot get is the harvest window and the picking standard, stated the way altitude is stated. A bag that said picked in the first third of the season, hand-sorted, 1,600 to 1,900 m would be telling me the two things that predict my morning and the one thing that describes the hillside. Until then, I write the altitude in the log next to the grind setting, where it is actually useful, and I stop giving it credit for the flavour. // the fields, in the order I now read them. Origin is a label, // not a prediction, so it sits near the bottom. { "picked": "early season", // moves the grade most [9] "processing": "washed", // decides if altitude counts [8] "variety": "...", // decides how much it counts [4] "origin": "...", "grade": "...", // height, or width, or a score [17] "altitude_m": [1900, 2350], // a density hint, not a promise "grind": 18, // finer than a lower-grown lot "water_c": 94 } [FOR THE ENTHUSIAST] what this post still lacks The two things I still want to add: a proper look at how bean density is measured commercially rather than in a laboratory, and a plot of my own log against the printed altitudes, which will almost certainly show no relationship and be the more honest figure for it. [END] ---------------------------------------------------------------------------- 08 USE YOUR AGENT -- Ask your agent to read this, and it can teach it back. ---------------------------------------------------------------------------- Any section or drawing on this page can be handed to an AI agent, with the prompt already written for that spot. Every heading, figure and selected passage has an ask control that copies a prompt for that exact place. Paste it into any agent; one with a browser calls these tools, one without reads the plain-text version. * start_guided_reading a lesson plan, a quiz or the answer * get_outline the map of the post * explain_section a section, at three levels * explain_figure a figure, at three levels * get_evidence the study behind a number * get_glossary any technical word [FOR THE ENTHUSIAST] how the tools work, and why three levels Coffee is the one subject on this site that everybody already has an opinion about, which means the readers are a fourteen year old, a chemist, and somebody’s uncle who only wants to know whether the expensive bag is worth it. One piece of prose cannot be pitched at all three. So every section and every figure is written out three times, at three levels, by me rather than by a model at runtime. If a paragraph lost you, your agent can hand you the child version of that exact paragraph. If it bored you, the expert one. get_evidence is the one that matters here, because the honest core of the post is that the studies confound altitude with everything else. It returns the citation, the link, and what that study actually measured. This post is a page with tools. They 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 and answer from this page’s own hand-written text, which loads from this site the first time an agent asks. No key, no model at runtime. A plain-text version of the whole post, prose and sources, lives at a stable URL for agents without a browser. > explain_section({ section: "physics", level: "child" }) { "section": "physics", "heading": "The air cools at a rate you can write down.", "level": "child", "text": "The higher up you go, the colder the air gets. Climb a kilometre and it drops about six and a half degrees. …" } [END] ---------------------------------------------------------------------------- ASK YOUR AGENT ---------------------------------------------------------------------------- Every section and figure on the page has a prompt like this one, for exactly that place. This is the one for the whole post: I'm reading "How altitude changes the bean". https://hetanshmehta.com/blog/coffee/altitude 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/altitude.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 ============================================================================ Peer-reviewed or governmental unless marked otherwise. Where two studies disagree, both are here and the disagreement is stated in the post rather than resolved. 1. National Oceanic and Atmospheric Administration, National Aeronautics and Space Administration and United States Air Force (1976). U.S. Standard Atmosphere, 1976 (NOAA-S/T 76-1562). U.S. Government Printing Office, Washington, D.C. ntrs.nasa.gov/citations/19770009539 https://ntrs.nasa.gov/citations/19770009539 2. Downing, R., Olago, D. and Nyumba, T.O. (2024). Seasonal and diurnal variability in near-surface air and ground temperature regimes of the alpine zone of Mount Kenya. Earth and Space Science, 11(2). doi.org/10.1029/2023EA003410 https://doi.org/10.1029/2023EA003410 3. de Oliveira Aparecido, L.E., Rolim, G.S., Moraes, J.R.S.C., Valeriano, T.T.B. and Lense, G.H.E. (2018). Maturation periods for Coffea arabica cultivars and their implications for yield and quality in Brazil. Journal of the Science of Food and Agriculture, 98(10), 3880-3891. doi.org/10.1002/jsfa.8905 https://doi.org/10.1002/jsfa.8905 4. Bertrand, B., Vaast, P., Alpizar, E., Etienne, H., Davrieux, F. and Charmetant, P. (2006). Comparison of bean biochemical composition and beverage quality of Arabica hybrids involving Sudanese-Ethiopian origins with traditional varieties at various elevations in Central America. Tree Physiology, 26(9), 1239-1248. doi.org/10.1093/treephys/26.9.1239 https://doi.org/10.1093/treephys/26.9.1239 5. Avelino, J., Barboza, B., Araya, J.C., Fonseca, C., Davrieux, F., Guyot, B. and Cilas, C. (2005). Effects of slope exposure, altitude and yield on coffee quality in two altitude terroirs of Costa Rica, Orosi and Santa María de Dota. Journal of the Science of Food and Agriculture, 85(11), 1869-1876. doi.org/10.1002/jsfa.2188 https://doi.org/10.1002/jsfa.2188 6. Vaast, P., Bertrand, B., Perriot, J.-J., Guyot, B. and Génard, M. (2006). Fruit thinning and shade improve bean characteristics and beverage quality of coffee (Coffea arabica L.) under optimal conditions. Journal of the Science of Food and Agriculture, 86(2), 197-204. doi.org/10.1002/jsfa.2338 https://doi.org/10.1002/jsfa.2338 7. 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. doi.org/10.1016/j.foodchem.2009.05.048 https://doi.org/10.1016/j.foodchem.2009.05.048 8. Worku, M., de Meulenaer, B., Duchateau, L. and Boeckx, P. (2018). Effect of altitude on biochemical composition and quality of green arabica coffee beans can be affected by shade and postharvest processing method. Food Research International, 105, 278-285. doi.org/10.1016/j.foodres.2017.11.016 https://doi.org/10.1016/j.foodres.2017.11.016 9. Tolessa, K., D'heer, J., Duchateau, L. and Boeckx, P. (2017). Influence of growing altitude, shade and harvest period on quality and biochemical composition of Ethiopian specialty coffee. Journal of the Science of Food and Agriculture, 97(9), 2849-2857. doi.org/10.1002/jsfa.8114 https://doi.org/10.1002/jsfa.8114 10. Nugroho, D., Basunanda, P. and Suyadi, M.W. (2016). Physical bean quality of arabica coffee (Coffea arabica) at high and medium altitude. Pelita Perkebunan, 32(3). Indonesian Coffee and Cocoa Research Institute. full text (PDF) https://library.sweetmarias.com/wp-content/uploads/2020/08/Physical-Bean-Quality-of-Arabica-Coffee-Coffea-Arabica-at-High-and-Medium-Altitude.pdf 11. Getachew, M., Tolassa, K., De Frenne, P., Verheyen, K., Tack, A.J.M., Hylander, K., Ayalew, B. and Boeckx, P. (2022). The relationship between elevation, soil temperatures, soil chemical characteristics, and green coffee bean quality and biochemistry in southwest Ethiopia. Agronomy for Sustainable Development, 42, 61. doi.org/10.1007/s13593-022-00801-8 https://doi.org/10.1007/s13593-022-00801-8 12. Cheng, B., Smyth, H.E., Furtado, A. and Henry, R.J. (2020). Slower development of lower canopy beans produces better coffee. Journal of Experimental Botany, 71(14), 4201-4214. doi.org/10.1093/jxb/eraa151 https://doi.org/10.1093/jxb/eraa151 13. Hu, R., Xu, F., Chen, X., Kuang, Q., Xiao, X. and Dong, W. (2024). The growing altitude influences the flavor precursors, sensory characteristics and cupping quality of the Pu'er coffee bean. Foods, 13(23), 3842. doi.org/10.3390/foods13233842 https://doi.org/10.3390/foods13233842 14. Williams, S.D., Barkla, B.J., Rose, T.J. and Liu, L. (2022). Does coffee have terroir and how should it be assessed? Foods, 11(13), 1907. doi.org/10.3390/foods11131907 https://doi.org/10.3390/foods11131907 15. El Hosry, L., Elias, V., Chamoun, V., Halawi, M., Cayot, P., Nehme, A. and Bou-Maroun, E. (2025). Maillard reaction: mechanism, influencing parameters, advantages, disadvantages, and food industrial applications: a review. Foods, 14(11), 1881. doi.org/10.3390/foods14111881 https://doi.org/10.3390/foods14111881 16. Barista Hustle. Coffee Buyer's Guide 4.01: Grading coffee in Guatemala. Industry reference, not peer-reviewed; used only for the ANACAFÉ altitude thresholds and the share of exports meeting them. baristahustle.com https://www.baristahustle.com/lesson/cbg-4-01-grading-coffee-in-guatemala/ 17. Molina Ospina, A.K. (2018). Kenya AA, Colombia Supremo: understanding coffee grading. Perfect Daily Grind. Industry reference, not peer-reviewed; used only for the screen-size grade names and for the statement that there is no universal grading system across producing countries. perfectdailygrind.com https://perfectdailygrind.com/2018/11/kenya-aa-colombia-supremo-understanding-coffee-grading/ 18. Abubakar, Y., Anhar, A., Baihaqi, A. and Mushlih, A. (2024). Influence of farm altitude and variety on quality of arabica coffee cherry and bean grown in Gayo Highland, Indonesia. International Journal of Design & Nature and Ecodynamics, 19(3), 1033-1041. Farms sampled in three bands from 1,000 to 1,600 m, which is the band the hero draws. doi.org/10.18280/ijdne.190332 https://doi.org/10.18280/ijdne.190332 19. Trade and green-importer origin references, not peer-reviewed. Used only for the four quoted regional altitude ranges in the hero, which the hero marks as trade figures rather than measured ones: Mercanta on Ethiopia (Sidama, 1,500 to 2,200 m) and on Colombia; the Genuine Origin Guatemala origin report (Huehuetenango, 5,000 to 6,500 feet); and the 1,400 to 2,000 m band commonly quoted for Kenyan arabica. coffeehunter.com/our-origins/ethiopia https://coffeehunter.com/our-origins/ethiopia/ 20. Córdova, M., Célleri, R., Shellito, C.J., Orellana-Alvear, J., Abril, A. and Carrillo-Rojas, G. (2016). Near-surface air temperature lapse rate over complex terrain in the southern Ecuadorian Andes: implications for temperature mapping. Arctic, Antarctic, and Alpine Research, 48(4), 673-684. Stations from 2,600 to 4,200 m, above coffee country. doi.org/10.1657/AAAR0015-077 https://doi.org/10.1657/AAAR0015-077 Two things I could not verify and so did not state as fact. First, a single canonical degree-day total for arabica: several papers give one, the figures differ by cultivar and base temperature, and I could not reach a primary text for the most-cited value, so the clock figure shows the ratio without a total. Second, a peer-reviewed table of maturation days against altitude specifically, rather than against temperature; source 3 measures temperature, and the altitude step is mine.