Oolong Aroma

Aroma in oolong names the volatile fraction of the leaf and of the cup: the compounds that reach the nose rather than the tongue, present in quantities measured in micrograms. More than 300 of them have been identified in oolong. Most were not in the fresh leaf. They arrive by three different mechanisms during manufacture, some built by the plant as a live defence response, some held in sugar-bound storage forms, some assembled by heat in the roast. A taster meets the result as one smell.

This guide documents that volatile fraction: where each class of compound comes from, which processing stage drives it, what the published record has actually measured, and where the record stops. It covers the stress-induced route, the glycoside pool and the evidence against the received account of it, the roast as a trade rather than an addition, and what a sensomics study finds in a finished cup.

Four neighbouring subjects are presumed here rather than re-argued. The bruise that opens two enzymatic locks at once is set out in Two Reactions Start When the Leaf Is Bruised, Not One. Catechins, theaflavins, thearubigins and astringency, which are taste and not smell, are the subject of Oolong Polyphenols. The oxidation spectrum and the roast axis as processing decisions are documented in Oolong Oxidation and Roast. Kettle temperature and vessel belong to Oolong Brewing and to Why a Green Oolong and a Heavy Roast Never Want the Same Water.

The three routes by which an oolong comes to smell of anything

The three run on three different mechanisms. The first is a storage form and the enzyme that would open it. The second is gene expression in a leaf still alive and still responding to injury. The third is thermal chemistry in a leaf that has stopped responding to anything, because kill-green has by then denatured the enzymes and ended the live response, which is the stage that closes route two and opens the way to route three.

The chemistry itself overlaps, and the overlap is the reason the routes are worth separating. Geraniol and linalool sit in the bound pool and also in the free set; the plant makes the primeverosides by glycosylating those same volatiles. So naming a compound settles almost nothing on its own. It is the route, and the stage that drove it, that decide whether the cultivar or the maker put it in the cup.

The stress route: the maker provokes, the leaf answers

Work from Yang Ziyin's group at the South China Botanical Garden, Chinese Academy of Sciences, identifies two stress factors as the inducers of aroma formation in oolong processing: mechanical damage (损伤) and low-temperature stress (低温胁迫). Continuous wounding through the zuoqing stage raises the expression of three biosynthesis genes, each yielding a different compound by a different pathway. CsTSB2 yields indole (吲哚). CsLOX1 yields jasmine lactone (茉莉内酯). CsNES yields (E)-nerolidol ((E)-橙花叔醇). Low temperature and wounding together produce a significant synergistic effect, because jointly they raise the content of jasmonic acid (茉莉酸), the upstream signalling molecule, along with the expression of its synthesis genes and of the key transcription factor CsMYC2. The group's conclusion is narrower than a slogan and worth keeping narrow: the enzymatic formation of aroma during oolong processing (香气的酶促形成) is principally a stress response (Zeng, Zhou, Su and Yang, Trends in Food Science & Technology 106, 2020¹; reported in Chinese by 科学网²). The qualifier matters here. It leaves the roast route, which is thermal and not enzymatic at all, entirely outside the claim.

Read the craft accordingly. The maker turning and shaking the leaf through the night is injuring a living plant, repeatedly, at a controlled rate and in a cool room. The fragrance rising off the pile is the plant's alarm. The enzymatic side of that bruising, and why zuoqing is best understood as the management of an oxygen-limited reaction by hand, is treated in Two Reactions Start When the Leaf Is Bruised, Not One. Wounding delivered by an insect rather than by a hand is treated in the Taiwanese tea whose bite mark is the flavour.

The locked pool, and what the record does not settle

The tea leaf stores aroma compounds in a non-volatile, sugar-bound form: glycosidically bound volatiles, chiefly β-primeverosides and β-glucosides of the floral alcohols listed in the table above. The unlocking enzyme, β-primeverosidase, is a disaccharide-specific β-glycosidase. Its specificity is very narrow for the sugar, the β-primeverosyl moiety it is named for, and it will not touch the same alcohol carrying a single glucose instead (substrate specificity of β-primeverosidase, PubMed³). The alcohols it releases are the floral set: the aglycones of the β-primeverosides found in tea leaves are linalool, geraniol, methyl salicylate, linalool oxide, benzyl alcohol, 2-phenylethanol and 3-hexenol. The enzyme is localized in the cell wall, and the primeverosides are held separately in the vacuole (Mizutani et al., Plant Physiology 130(4), 2002⁴). In an intact leaf they never meet.

Note carefully which barrier that is, because it is the distinction the whole section turns on. Bruising the leaf ruptures cell membranes, which is what lets polyphenol oxidase reach the catechins and lipoxygenase reach the free fatty acids, the two reactions Two Reactions Start When the Leaf Is Bruised, Not One sets out. The cell wall is a different structure and a tougher one. An enzyme held in the wall and a substrate held in the vacuole are not brought together by a bruise that opens the membranes between them.

The textbook account runs from there: break the cell, let enzyme meet substrate, and oolong's floral smell is the pool being unlocked. Two direct measurements cut against that account, and they agree with each other.

Wang, Kubota, Kobayashi and Juan quantified 16 glucosides and primeverosides at four stages of oolong manufacture, at plucking, after solar withering, after indoor withering and in the finished tea, in the Taiwan cultivars Chin-shin oolong and Chinhsuan (Jin Xuan). Most of these glycosides increased from the solar-withering stage onward and reached their highest level in the finished oolong, and the authors noted that not one of the sixteen fell at any stage, which they marked as the difference from black tea manufacture (J. Agric. Food Chem. 2001, 49, 5391 to 5396⁵). A pool being drained does not end fuller than it began.

The same study went further and measured the free alcoholic aroma compounds those glycosides would yield on hydrolysis. They came through manufacture almost unchanged or slightly lower, and they accounted for only about 12 and 17 percent of each cultivar's total aroma compounds. What rose markedly in the finished tea was jasmine lactone and indole, which are not glycoside products at all. They are the stress-route compounds of the section above. Jin Xuan as a bred cultivar rather than a flavoured tea is covered in Jin Xuan and the milk that isn't milk.

Gui, Fu, Zhou, Katsuno, Mei, Deng, Xu, Zhang, Dong, Watanabe and Yang put the question directly: does enzymatic hydrolysis of the bound volatiles contribute to oolong's volatiles at all? Their answer was no. Through the enzyme-active stages of manufacture the leaf cell walls are not disrupted, so the bound volatiles and the β-glycosidases cannot interact, and the characteristic potent odorants, indole, jasmine lactone and trans-nerolidol, are produced by the manufacturing process itself, by de novo biosynthesis (J. Agric. Food Chem. 2015, 63, 6905 to 6914⁶).

What follows is a boundary, and the Order keeps it. The finished leaf demonstrably carries a large locked pool. These papers do not settle what becomes of that pool in the cup. It is tempting to say the surviving glycosides are what keep the sixth and seventh infusions aromatic. That has not been measured, and a guide that asserts it is inventing a finding. The Order holds nothing sacred that it cannot measure, and it states nothing it cannot verify twice.

The roast: a trade, not an addition

A roasting study on Tieguanyin tested 105°C (221°F) for 1, 3 and 5 hours, and a high-temperature treatment of 130°C (266°F) for 1 hour applied after 5 hours at 105°C (Sensory and chemical characteristics of Tieguanyin oolong tea after roasting⁷).

Long hours are not the variable. Temperature is. Adding hours at 105°C did not significantly move the aroma substances; one hour at 130°C did. Read that comparison for exactly what the design supports and no more: every sample in the study was roasted, with no unroasted control, so it measures what more time at 105°C costs, not what the first hour costs. The sensory panel is the check on reading it too kindly, since floral aroma was already declining across the 105°C series while the instruments registered no significant change in the substances themselves.

A second study put kinetics on the heat-built compounds, and it is worth being precise about which part of it was tea. Gao, Wang, Wang and Song roasted Qidan, a Wuyi yancha (rock tea) cultivar, at a single temperature, 110 ± 5°C (230°F), sampling every two hours out to twelve. The reaction orders come from a companion model system rather than from leaf: theanine and glucose in water at pH 5.5, run for five hours at 100°C (212°F), 110°C and 120°C (248°F). In that beaker the pyrazines followed zero-order kinetics at all three temperatures, 2-methylfuran was zero-order at 100°C and first-order at 120°C, and benzaldehyde shifted from zero-order at 100°C to first-order at 110°C and above, its formation sharply accelerated by heat (Foods, 2025⁸). A model reaction is not a roasting basket. It shows which reactions the heat can drive and how their rates answer to temperature, not what any particular tea will do.

The useful reading for a drinker: charcoal roast is not a dial that adds roastiness on top of the floral set. Above a threshold it trades one compound class for another, and the traded-away class does not come back. α-Farnesene was not reduced at 130°C. It was gone.

What one cup smells of, and why concentration is not smell

A sensomics study of a floral honey-like Fenghuang Dancong from Fenghuang Mountain, Chaozhou, Guangdong Province identified 17 key odorants by gas chromatography-olfactometry. Those scoring intensity 3 or above included α-ionone, geraniol, 4-methylpent-3-en-2-one, (E)-2-nonenal, heptanal, limonene, linalool and (Z)-2-penten-1-ol (sensomics characterization of Fenghuang Dancong cup aroma⁹).

The three highest odour activity values in the cup aroma were α-ionone at OAV 2875, geraniol at OAV 1738, and decanal at OAV 850. Set those against the measured concentrations: α-ionone 28.75 μg/m³, geraniol 20.86 μg/m³, decanal 212.61 μg/m³.

Decanal was present at roughly seven times α-ionone's concentration and scored less than a third of its OAV, because an odour activity value divides concentration by the compound's odour threshold. A compound present in quantity may contribute almost nothing, and a compound present in trace may dominate the cup. Any aroma figure that reports how much of a compound is present, without reporting the threshold, has not told you what the tea smells of.

The strong evidence in that study is the recombination test. A model aroma built from the identified odorants was compared against the real sample by paired t-tests, and no sensory descriptor differed significantly (p > 0.05). The list was complete enough to rebuild the smell.

That profile describes one tea, from one mountain, in one cultivar family. Fenghuang Dancong is a population of cultivars, several of them named for the smell they carry, as set out in Honey Orchid to Duck Shit, and Whether the Names Are Real.

How to read an aroma claim

A cultivar name on a bag tells you which plant was picked. It does not tell you which of the three routes dominated the finished aroma, because the same cultivar processed differently ends up in a different place. Chin-shin oolong and Chinhsuan were measured through the same manufacture, and both accumulated glycosides rather than spending them, despite differing from each other in the glycosides their fresh leaf carried in the first place.

Season and picking are a separate axis again, treated in The Old Rule About Spring and Autumn Tea, Tested Against 266 Cups, as is the distribution of floral compounds within a single plucked shoot, treated in The Stem Concentrates Floral Compounds the Leaf Does Not.

A roast claim is the most informative label on a bag, because the roast is the one stage of the three with a measured threshold behind it. "Lightly roasted" is consistent with hours at 105°C, which the record says costs the aroma substances no significant further ground while cutting astringency, though the floral note edges down with every added hour. A cup that reads as roasted, nutty and caramelized is a cup whose Maillard heterocycles were built at a temperature its floral set did not survive. Where no roast level is stated, the taste of the cup, not the copy on the bag, is the evidence, and the two are worth checking against each other before the bag is bought again.