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. More than 300 of them have been identified in oolong. Most were not in the fresh leaf. During manufacture some are built by the plant as a live defence response and some are assembled by heat in the roast. A taster meets the result as one smell.

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 stored pool, the stress route and the roast

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 stress route

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 enzymatic side of that bruising 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.

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).

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. 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 volatiles, indole, jasmine lactone and trans-nerolidol, are produced by the manufacturing process itself (J. Agric. Food Chem. 2015, 63, 6905 to 6914⁶).

The finished leaf demonstrably carries a large locked pool. These papers do not settle what becomes of that pool in the cup.

The roast

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).

Adding hours at 105°C did not significantly move the aroma substances; one hour at 130°C did. 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. 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. 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 paper also fitted reaction orders at 100°C (212°F), 110°C and 120°C (248°F), heated for up to 12 hours, in a theanine-and-glucose model reaction. 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 favoured by heat (Foods, 2025⁸). The fits show which reactions the heat can drive and how their rates answer to temperature, not what any particular tea will do.

What one cup smells of

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 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.

Season and picking are treated in The Old Rule About Spring and Autumn Tea, Tested Against 266 Cups.