On a shop shelf, a packet marked light roast sits beside one marked dark. Picked up, they rustle dryly in the hand. Turned over, neither says how deep the roast went. Light and dark are written on the front alone.

What follows traces what changes with depth, and what depth alone does not settle, through colour, aroma and taste in turn.

Light and dark, shelf by shelf

Set side by side, two packets seem to differ even in the shade of their paper. The line between light and dark is drawn by each shop.

It is a little like a mountain that each village at its foot calls by a different name. The labelling rules go as far as names such as hojicha and hold no words for depth. One long-established tea house writes that the finer the tea, the lighter the roast, while other shops make dark roast their signature.

Studies, too, divide the levels in their own ways, so the sources support comparison only within a single shelf. How research levels line up with shop words continues in Can hojicha roast levels be compared across brands?. After the words on the packet comes the colour in the cup.

Gold, and a reddish brown

Poured into a white cup, the colour is the first thing to catch the eye. The deeper the roast, the redder it turns.

Like autumn leaves colouring day by day, the red builds with roasting time. Tea merchants describe light-roast liquor as gold and dark-roast liquor as reddish brown. When a prefectural research station measured the colour of the leaf, the dark roast came out clearly redder too.

Colour figures shift with the way they are taken, though. The sources reach as far as the difference visible on white paper, and draw no line where dark roast begins. After colour comes the aroma in the steam.

Toasty steam, on the lighter side

Close to the cup, a toasty smell rises with the steam. That toastiness does not necessarily grow stronger the deeper the roast.

When one prefectural station held the temperature and changed only the depth, the toasty-aroma compounds came out higher in the light roast. In another prefecture, an experiment held the depth and raised the temperature, and compounds of the same family rose. The two ask different questions, so quoting just one turns the conclusion upside down. Depth and aroma are taken up at leisure in Does dark roast hojicha taste more toasty?.

Deeply roasted, the tea kept almost the same mix of aromas whatever the temperature. It is the light roast that shows how it was made. The precursors of the floral notes sit in the leaf bound to sugars. Roasting heat frees them into the air, and their share falls as the roast goes on.

Keeping the floral notes means stopping early, and that is as far as the sources go. Why stem tea tends to get a lighter roast can be checked in Why stem hojicha gets a light roast. After the aroma comes the taste left on the tongue.

Less astringency, caffeine much the same

With one sip, a dry astringency lingers under the tongue. The deeper the roast, the less of what causes it, while caffeine hardly changes.

As a ripening persimmon loses its pucker, roasting draws the astringency out. In the prefectural station’s figures, the dark roast held about a third as much of its source as the light. Caffeine dipped only a little in the dark roast, and the light roast stayed close to unroasted leaf.

The report does not show why caffeine fell; the figures are gathered in Does roasting reduce caffeine in hojicha?. Coffee, said to turn more bitter as the roast deepens, goes the other way here. Some merchants recommend dark roast with milk, but the sources reach only as far as its lower astringency. How it sits in a latte is covered in Dark roast hojicha for lattes: what changes in milk.

Whether to change the water or the time by depth continues in Hojicha water temperature: brew light and dark roasts apart?. The figures and trial conditions are set out together further on.

The kettle water, set by harvest

The kettle begins to sing, and there is a short pause before the water meets the pot. No rule ties the heat of the water or the steeping time to the depth of the roast.

Brewing guides published by merchants and associations, read together, offer no advice that splits the water by roast. One tea farm does give separate figures, but it divides them by harvest. Hojicha from the second or third picking gets boiling water and a quick steep. First-flush hojicha gets slightly cooled water and a slower one.

Two products often mentioned side by side do not compare depth with depth either. A lightly roasted stem tea and a tea roasted deep over direct fire differ in raw material and in the amount of leaf. Their water and times can be set side by side in Hojicha water temperature: brew light and dark roasts apart?.

“Roasted over strong heat” describes the method, while “dark roast” describes the finish. One tea house roasts the stems sorted out from gyokuro and sencha over strong heat. Beside them on its shelf sits a tea roasted lightly over a gentler flame. The difference is taken up in High heat vs dark roast hojicha: two Kyoto examples.

A merchant selling powders by roast level also pairs stems with the light end and twice-roasted leaf with the dark. The roast word on a packet travels together with leaf or stem and with the harvest. The sources support only this much: it is the raw material that decides the water. To line up the colours by eye, the next place is the kitchen stove.

Three depths, roasted in the kitchen

Over the flame, the leaf in the pan begins to crackle softly. Roasted from the same leaf in three batches, the differences in depth show plainly even in a home kitchen.

The pan, the amount of leaf and the flame stay the same, and only the time before the heat goes off changes. Without a hōroku, the earthenware pan made for roasting tea, a frying pan will do. The leaf changes colour in moments, so the stove is not left while it roasts.

Back-to-back batches run faster in a pan that is already warm. Between rounds, the pan cools on a damp cloth. Roasting carries on in the leftover heat after it leaves the flame, so each round ends only when the pan has cooled.

Once cool, the three go on white paper from light to dark and are photographed in the same place and light. Brewed with the same pot, the same water and the same wait, each aroma gets a single word in a notebook. The middle cup may well turn out the most toasted.

Which one tastes best is something the sources cannot decide. The full steps are written out in Roasting hojicha at home in three levels, light to dark. With three cups in a row, the rest is reading on in the order of what is left to know.

The cup set down, the next page

A faint brown ring is left at the bottom of the cup. What to read next is best chosen by what is still unknown.

Who decides the words on the packet is the starting point of Who decides light and dark roast hojicha? Labels, 18 shops. How research levels meet shop words is laid out in Can hojicha roast levels be compared across brands?.

For aroma there are Does dark roast hojicha taste more toasty? and Why stem hojicha gets a light roast, and for milk, Dark roast hojicha for lattes: what changes in milk. Caffeine has its own page in Does roasting reduce caffeine in hojicha?. Each one carries its trial conditions in full.

The brewing guides are compared in Hojicha water temperature: brew light and dark roasts apart?. For roasting at the stove, Roasting hojicha at home in three levels, light to dark and How to roast hojicha at home: horoku or frying pan are worth keeping close. By the time the next cup is brewed, the choice will likely be made.

The numbers in detail

Labelling rules
The Japan Tea Central Association’s Green Tea Labelling Standard defines 17 names, hojicha among them. Roast-level words appear neither among its 11 mandatory label items nor among the 7 optional ones.
18 sellers’ labels
Of 18 hojicha sellers’ official websites, 4 used words for roast depth. None published a roasting temperature or time. The sites were checked as of 24 September 2026.
Shops’ own words
Nakamura Tokichi Honten writes that the higher the price, the lighter the roast and the paler the brown. Ito En puts “dark roast” in a product name. Ryuhoen is the merchant that describes the liquor as golden and red-brown.
Colour values
The colour trial is by the Tea Research Station of the Kochi Prefectural Agricultural Technology Center. In its 2014 sample, the redness value a* of first-flush leaf, ground and measured as powder, was 1.34 for the light roast and 5.11 for the dark. A 2020 Ishikawa report measured whole leaf and recommends watching lightness L* and yellowness b* rather than redness.
Aroma trial (Kochi)
The Kochi trial roasted first-flush leaf (2013) with the exhaust temperature held at 100°C. The drum turned at 17, 12 and 8 revolutions a minute to make the three levels. The two pyrazines used as markers of toastiness are absent from unroasted leaf; in first flush and karibancha they were higher in the light roast, and second flush showed no difference. Of the floral compounds, linalool peaked at the light roast and geraniol at the medium.
Aroma experiment (Ishikawa): In a 2020 experiment by the Industrial Research Institute of Ishikawa and others, stems were roasted at drum-surface temperatures of 180–220°C. All eight aroma-active pyrazines rose the hotter the roast. Deeply roasted samples kept almost the same mix of compounds whatever the temperature.
Astringency values
Astringency was compared by brewing 1.9 g of first-flush leaf (2013) in 200 mL of boiling water for 5 minutes, per 100 mL of liquor. Total catechins were 34.89 mg unroasted, 21.86 mg light, 14.59 mg medium and 6.98 mg dark. Some individual catechins rise at the light roast before falling.
Caffeine
Caffeine comes from a separate 2014 sample, with unroasted leaf set at 100. First flush scored 101 light and 94 dark, karibancha 99 light and 87 dark, so the dark roasts were 6–13% lower.
29 brewing guides
The account of coffee roasting and bitterness is from the All Japan Coffee Association. For brewing, 29 published pages were reviewed, from 18 merchants and makers and 11 public bodies and associations. They were checked on 22 September 2026. Only Fujita Chaen split water temperature or time, and it did so by harvest.
Split by harvest
Fujita Chaen gives 3–4 g of leaf in boiling water for 30 seconds for ordinary hojicha made from second or third flush, and 2–3 g at 80–90°C for 1 minute for first-flush hojicha.
Two often-quoted products: The two often-quoted products are Maruhachi Seichajo’s Kenjō Kaga Bōcha, lightly roasted stems, and Minokamo Chaho’s direct-fire dark-roast hojicha. The first is 6 g in 260 mL of boiling water for 25 seconds; the second about 5 g in 180 mL at about 80°C for 1 minute.
High heat and dark roast
Maiko no Chahonpo in Kyoto sells karigane, the stems sorted out when making gyokuro or sencha, roasted over strong heat, beside a light-roast hojicha roasted gently over lower heat. The powders by roast level come from a Maruyasu Chagyo announcement: stem hojicha for light, stems and leaf for medium, twice-roasted leaf for dark.
Residual heat
The note that roasting carries on in the leftover heat after the pan leaves the flame follows Chakoizumi’s frying-pan method.