Four different jobs that all involve heat
It is worth separating them, because tea writing runs them together. Kill-green — steaming or pan-firing early in green tea manufacture — applies heat specifically to destroy the oxidative enzymes and stop oxidation before it starts. Drying reduces the leaf’s moisture to a level at which it is stable in storage, and is the final step of nearly every manufacture. Roasting or baking applies heat beyond what drying requires, deliberately, to generate new aroma and to change the character of the tea. And re-firing applies gentle heat to a finished tea some time later, to drive off reabsorbed atmospheric moisture. Same physical input, four different intentions, and the temperature, duration and equipment differ accordingly. A tea can undergo all four at different points in its life. Temperature and duration distinguish them more than equipment does. Kill-green is brief and hot enough to destroy enzymes quickly; drying is longer and cooler; roasting is longer still and is judged by what it produces rather than by a target moisture level; re-firing is the gentlest of all. The same pan, oven or drum may be used for several of them, which is part of why the vocabulary blurs, and the intention behind each is what a description should be reporting.
What the Maillard reactions do
When reducing sugars and amino acids are heated together they react, through a long cascade rather than a single step, to produce a large family of new compounds — including the nitrogen-containing heterocycles associated with roasted, nutty and popcorn-like aromas, furans associated with caramel and sweetness, and the brown pigments that darken the material. This is the same chemistry that browns bread crusts and roasted coffee, and it is why roasted tea smells of things that are not in the fresh leaf at all. Caramelisation of sugars alone contributes as well, at higher temperatures. The critical point for a taster is that these compounds are created rather than concentrated: a roasted oolong has a genuinely different aromatic inventory from the same leaf unroasted, not a stronger version of the same one. The cascade is also why roasting is so difficult to reverse or adjust. It is a sequence of reactions rather than a setting, so a tea taken too far cannot be brought back, and a tea roasted lightly can always be taken further later. Roasters therefore work upwards in stages with rests between, tasting as they go, which is the same logic a cook applies to browning and is one of the reasons a good roast is expensive.
Why the amino acid pool matters here too
Because the reaction needs both partners, the amino acid level in the leaf affects what roasting can produce. This links two parts of the catalogue that look unconnected: the same free amino acids associated with umami in an unroasted green tea are consumed as substrate when the leaf is roasted, becoming aroma compounds instead. It is a real exchange, and it explains why heavily roasted teas do not show the savoury sweetness of a shaded green and why their sweetness reads as caramel and malt rather than as broth. It also explains why hojicha, made by roasting bancha or kukicha at high heat, tastes so unlike the tea it was made from — and why roasting is sometimes used on material whose delicate qualities were never going to be the selling point. It also explains why roasting is applied to the materials it is applied to. Later-harvest leaf, stem and coarser grades have less of the delicate character that a light treatment would preserve, and they respond well to a process that generates new aroma rather than protecting old. Hojicha, genmaicha and the roasted stem teas are all built on this reasoning, and they are better understood as products designed around a material than as lesser versions of something finer.
The distinctions within roasting
Roast level, roast method and roast recency are three separate variables and are frequently confused. Level runs from a light finishing bake through medium to heavy and to full charcoal roasting, and this catalogue records it as a descriptor on the tea. Method matters: traditional charcoal roasting in Wuyi and Anxi is a slow, low, repeated process over hours or days that behaves quite differently from electric baking, and practitioners consider the two distinguishable — charcoal roasting is generally described as producing a rounder integration and electric baking a cleaner, more controllable result. Recency matters most of all to a drinker: a freshly roasted tea is often at its least attractive, showing the roast on the surface, and improves over months as it settles and integrates. A shop selling a tea it roasted last week is selling something that is not finished. Recency has a further consequence for buyers, which is that the same tea is a moving target through the year after it is made. A shop that roasts its own stock may sell a tea in three noticeably different states over eighteen months, all accurately described by the same name. Asking when a roasted oolong was fired is therefore as reasonable as asking when a green tea was picked, and it is a question good sellers of the category expect.
What roasting can and cannot fix
It can transform a plain tea into an interesting one, add body and a perception of sweetness, and give a tea the structure to survive years of storage — traditional roasting is a preservation technology as much as a flavour one, which is why the roasted oolong traditions and the ageing traditions overlap. It can also cover faults, and this is worth naming plainly: a heavy roast will mask a thin, coarse or slightly stale base, and some heavily roasted tea on the market is heavily roasted for that reason. What it cannot do is restore aromatics that have already gone, and a re-fired stale tea is a roasted stale tea. The buyer’s tell is whether the roast sits over the tea or is integrated with it, which is most easily judged in the later infusions where the roast character has thinned and whatever is underneath is exposed. Brewing a roasted tea also demands a different approach from an unroasted one. Roast character comes out early and strongly, so a first infusion can read as almost entirely roast with the tea underneath invisible; the second and third are where the base becomes legible. Judging a heavily roasted oolong on its opening steep is therefore judging the firing rather than the tea, which is one reason these teas are so often dismissed by drinkers who tried them once.
Steaming against pan-firing, and why it matters for aroma
The kill-green method is the earliest fork in green tea manufacture and its aromatic consequences persist all the way to the cup. Steaming, dominant in Japan, uses moist heat that halts the enzymes quickly and preserves the fresh green, marine and vegetal character of the leaf. Pan-firing, dominant in China, applies dry heat to the leaf surface and initiates a small amount of the same browning chemistry described above, which is why Chinese green teas so often show chestnut, bean and toasted notes that Japanese ones do not. Neither method is more authentic. They are two solutions to the same problem, adopted in different places, and the difference in the resulting cup is one of the largest and most reliable distinctions in the whole category. Steaming depth is a further variable within the Japanese method and is worth knowing because it is frequently stated on labels. A longer steaming breaks the leaf down more, producing a cloudier, sweeter, fuller cup that brews quickly, while a shorter one keeps the leaf intact and the liquor clearer and more delicate. The same cultivar from the same field can therefore be sold as two quite different products, and the depth is the maker’s decision rather than a regional given.