Does roasting reduce caffeine in hojicha?

Roasted deep, a little less

As the roaster's drum turns, a thin, toasty smoke drifts out. Within the published trials, roasting does not cut caffeine by much.

In a prefectural tea station's report, the light roast barely fell at all. The dark roast came out a few per cent to just over a tenth lower than before roasting.

What caused that fall, though, cannot be read out of the report. Next comes what the trial actually recorded.

Air and rotation, recorded

Hot air blows out of the roaster's mouth. What was recorded is the temperature of that outgoing air and the speed of the drum, nothing more.

The roasting ran in two stages, and the air at the second stage was the same for light and dark. What separated them was the drum's speed. The slower it turns, the longer the leaf stays inside.

How hot the leaf itself became is written nowhere. After the air comes a record from higher temperatures.

High heat inside an oven

Open an oven door and dry heat meets the face. Even an old record of longer heating at higher temperatures shows caffeine barely changing.

In a report that heated unrefined green tea in a constant-temperature oven, amino acids, vitamin C and sugars fell sharply. Caffeine, by contrast, hardly differed, the text says.

But the air leaving a roaster and the inside of an oven are measured in different places. The two records do not sit on one ruler. Next comes the band of real roasting temperatures.

A band still left open

Where bōcha is roasted, the inside of the drum runs far hotter. No public data has measured caffeine in that band.

An experiment roasting bōcha in a drum set the inner surface to around two hundred degrees, on manufacturers' advice. No public data measuring caffeine at those temperatures turned up in this search.

The sublimation temperature so often quoted is not in the official chemistry database either. The sources go as far as this: the band is still open. Next come stems before roasting, and coffee.

Stems, and coffee's split answer

Shake a packet of kukicha and the thin stems rustle. Unroasted stem tea looks lower than leaf, but the figure was not a like-for-like comparison of leaf and stem.

In the same station's table, unroasted stem tea came to about three quarters of first-flush tea. That table, though, adds one line for stems to a run of tea seasons; it was not built to compare leaf with stem.

For coffee, the same question splits. One study found the darker roast lower with brewing held level, and higher with extraction yield held level. Raw material and brewing differ, so it cannot be carried across to hojicha. Only the shape is shared: change what is held steady, and the answer changes.

Last comes where the missing part went.

Where the missing part went

Spread the roasted leaf on a sieve and fine dust gathers underneath. Where the part lost at the dark roast went is still not known.

Broken down by heat, or crumbled and lost as dust: the report does not say. Through the roasting, it was caffeine that stayed largely still. The other compounds that do move are taken one at a time in the pieces on roast depth.

On the mechanism, the sources go this far and no further. The figures and trial conditions are gathered just below. The leaf on the sieve is still faintly warm.