Why a Green Oolong and a Heavy Roast Never Want the Same Water
The brewing guide states the temperature bands by oxidation level. Here is the extraction chemistry, measured in real studies, that makes them true.
The brewing guide states the numbers plainly: 85 to 90C for a green-leaning oolong, 96 to 100C for a heavy roast. Ask why, and the honest answer sits in food chemistry, not tradition. Three things change as a leaf moves from lightly oxidized to heavily oxidized and roasted, and each one alone would be reason enough to change the water. Together they explain why treating every oolong the same way either scalds the delicate end of the category or shortchanges the dark end of it.
The bands the guide gives, and the question it does not answer
A greener oolong, tie guan yin and gaoshan among them, wants 85 to 90C. A dark, roasted yancha or dong ding wants 96 to 100C, up to a full boil. That is stated as a working rule on the brewing page, drawn from experienced tea sellers. It is correct. It does not say why a fifteen-degree gap matters, or what actually happens inside the leaf at each end of it. That gap has been measured, in three separate places: how fast caffeine and catechins leave the leaf, how the leaf's aroma compounds respond to heat, and how the leaf's physical shape after processing slows the whole exchange down.
Caffeine does not respond to heat evenly across the category
A 2025 study in the journal Foods2 put six tea types through matched brewing conditions, two temperatures (80C and 100C) and three durations, and measured what actually came out in the cup. Oolong was the one outlier. At a two-minute steep, its caffeine content jumped from about 4mg to about 14mg per 200ml between the two temperatures, a statistically significant swing. Green, white, black, and the other types tested did not move nearly as much over the same twenty-degree gap (Ayakdas and Agagunduz, Foods, 2025). The study did not sort its oolong samples by oxidation level, so it cannot say whether a lightly oxidized sample or a roasted one drives the effect more. What it does establish is that oolong, as a category, answers to water temperature more than a reader might assume from watching a green or black tea brew.
An older paper gives a plausible reason why oxidation level itself should matter this much. Spiro, Jaganyi, and Broom measured caffeine extraction rates across a range of temperatures for an unoxidized green tea and a fully oxidized black tea, and calculated the activation energy for each, the energy cost of speeding the reaction up. Green Chun Mee came in at 62 kilojoules per mole; black Assam Bukial at 40 (Food Chemistry, 1992). A higher activation energy means the extraction rate is more sensitive to a change in temperature, not less. Oolong sits between these two endpoints on the oxidation spectrum, and the paper did not test it directly, so the honest claim is a bracket, not a measurement: a jade-style, lightly oxidized oolong should behave nearer the 62 end, more sensitive to a few degrees, and a dark-roast oolong nearer the 40 end, more forgiving of them. The 1992 measurement and the 2025 oolong-specific result point the same direction without confirming each other outright, which is exactly the caution a single study deserves.
Catechins have a ceiling, and pushing past it does not help
Heat does not just pull more out of the leaf. Past a point, it changes what comes out. A Turkish study brewing green tea at 75, 85, and 95C across steep times from one minute to forty-five found that EGCG, the catechin most associated with a green tea's bitterness and its health chemistry, peaked at 85C and three minutes, at 50.69mg per 100ml, then fell at longer times and higher heat as the molecule converts to non-epistructured catechins through epimerization, a structural rearrangement that keeps the mass but changes the compound (Saklar, Ertas, Ozdemir, and Karadeniz, Journal of Food Science and Technology, 2015). Caffeine, measured in the same infusions, plateaued within a few minutes at every temperature tested and held there instead of reversing.
This matters for oolong because catechin load tracks oxidation level directly. A lightly oxidized leaf, chemically closer to green tea, still carries most of its original catechin pool intact. Not so past the midpoint of the spectrum: kill-green catches a heavier leaf only after oxidation has already spent a large share of that pool turning it into theaflavins and, later, thearubigins, the pigmented compounds a fully fermented cup runs on, well before any kettle is involved. So the greener end of the oolong spectrum is the end still carrying enough raw catechin to run into the same ceiling the Turkish study found: brew it too hot or too long and you are not extracting more flavor, you are converting what you already had into something duller. The roasted end has spent that pool already, one real reason it tolerates a hotter, less exacting hand.
Aroma is a bet placed at the kettle, and the two ends bet differently
Temperature does not only govern what dissolves. It governs what survives. A 2023 comparison of roasted and unroasted oolong, examined by gas chromatography, found the two styles built from almost entirely different aroma chemistry. Unroasted high-mountain oolong ran heavy on floral terpenes, linalool, geraniol, and nerolidol among them, compounds that account for close to half its detected volatiles and that are genuinely heat-labile: they evaporate and degrade under sustained high heat rather than simply dissolving faster. Roasted oolong ran instead on heterocyclic compounds, pyrazines, pyrroles, and furans, the class responsible for a roasted, caramel-like character, which made up as much as 55 percent of its volatile profile against roughly a third in the unroasted leaf (Wang et al., Current Research in Food Science, 2023).
Those heterocyclic compounds are not merely heat-tolerant, they are Maillard products, meaning heat is what created them in the first place during roasting. Hot water cannot damage what fire already built. But hot water can absolutely strip a lightly oxidized oolong of the linalool and geraniol that are its entire reason for being brewed cool. Pour boiling water on a jade Tieguanyin and you are not being generous with it. You are cooking off the one thing that made it worth choosing over a roasted style.
The leaf's shape adds a second clock, independent of chemistry
Oxidation is not the only variable running. Leaf shape sets a separate timer. Rolled ball-style oolongs, the tieguanyin and gaoshan pellets most lightly oxidized leaf is finished into, present almost none of their surface to the water at first contact. A twisted or open leaf, the shape favored for the more oxidized yancha and dancong styles, unfolds most of its surface immediately. The same 1992 measurement puts a number on the hindrance: at 80C, the diffusion coefficient of caffeine inside the leaf ran 96 to 132 times smaller than its diffusion coefficient in plain water, depending on the tea. A tightly rolled pellet compounds that hindrance with simple geometry: water has to work its way past several folded layers before it reaches the core.
This is why the rinse step matters more for a rolled leaf than an open one, and why the first infusion of a ball-rolled tieguanyin often reads thin no matter how hot the water is. The leaf has not physically opened yet. Shape and oxidation happen to correlate in oolong, tightly rolled shapes cluster at the lighter end of the spectrum, open twisted shapes at the heavier end, so the two effects usually point the same direction: patience for the green end, speed for the dark end. But they are two different clocks, one chemical and one physical, running at once.
The fifteen-degree gap is measured behavior, not tradition
Put the three findings together and the fifteen-degree gap in the brewing guide stops being a rule of thumb and becomes a description of measured behavior. A jade-style oolong carries more raw catechin, more heat-labile aroma compound, and often a rolled shape that slows water's first contact with the leaf, three separate reasons to keep the kettle under 90C and let the pellet take its time. A heavily oxidized and roasted leaf has already spent that catechin pool, built its aroma from heat-stable Maillard compounds instead of fragile terpenes, and usually opens faster besides, three separate reasons it can take a full boil without complaint. None of this changes the numbers the guide already gives. It explains why guessing past them costs you something real, not just tradition ignored.