Koju: The Unregistered Tea Cultivar That Smells of Grapes

by Christian Mauerer, founder of One with Tea, who filmed with tea growers across Shizuoka and Kyushu in May 2025.

A tea farmer in Shizuoka bought a Taiwanese pan-firing machine and then could not find a Japanese tea worth putting in it. He withered and fired dozens of varieties, more than a hundred aroma trials, and none of them were right. Then he found a bush growing by itself in the corner of a field that smelled of grapes.

That bush is Koju. It is not in the national register, it has no official parentage, and for most of its life exactly one farm has made tea from it. We do not sell it and I have never tasted it. I am writing it up anyway, because it is the clearest record I have found of how a cultivar actually gets born, which is almost never the way the textbooks describe.

What is Koju tea?

Koju (香寿, sometimes romanised Kōju or Kôju) is a Japanese tea cultivar from Honyama, Shizuoka. The grower Takahashi Tatsuji of Marutaka Farm selected it from a single seedling tea plant growing wild in the corner of one of his fields. It is not registered with Japan's Ministry of Agriculture, Forestry and Fisheries, and the specialist cultivar shop Shinkoju states this plainly: 「＊農林水産省の品種登録はされておりません。」 It is a 民間(個人)育種品種, a privately bred cultivar. The leaf is made into a withered, lightly oxidised, pan-fired tea that the farm files under 本山の香り茶, the fragrant teas of Honyama.

Key Takeaways

• Koju (香寿) is a Japanese tea cultivar from Honyama, Shizuoka, made into a lightly oxidised, pan-fired semi-fermented tea with an aroma of white grape.
• Koju is not a registered variety. Koju appears nowhere in the Chanorin register of national tea cultivars, and Shinkoju records it as privately bred. Its 来歴, its provenance, is simply 実生選抜, seedling selection.
• It was found, not bred. Takahashi Tatsuji had been making tea in Shizuoka for nearly fifty years when he noticed one bush in a field corner that smelled different, then spent years propagating it from cuttings.
• The grape aroma is the whole point. Shinkoju calls it 「お茶とは思えない程の芳醇な果実香」, a fruit aroma you would not believe came from tea, and 「まるでマスカットのような」, just like muscat.
• The name means fragrance and long life, after the white down on its buds, not after any effect of drinking it. Heavy white down suggested 白眉, white eyebrows, and so 長寿, longevity. 香 plus 寿 gives 香寿.
• It is made like Taiwanese tea, not like sencha. Six to ten hours of withering, a pan heated over 200°C to stop oxidation, rolling, then a finish in the pan-firing machine.

Koju or Koshun? Two different plants

Koju (香寿) and Koshun (香駿) are different cultivars, and in English they get mixed up constantly. Koshun is a registered Shizuoka cultivar with a documented cross behind it. Koju is an unregistered seedling selection with no documented parents at all. The two share a first character, 香, meaning fragrance, and nothing else.

What makes the confusion worse is that the same farm is known for both. At the 2016 Nihoncha Awards in Shibuya, Marutaka took three teas to the Tokyo Tea Party tasting: their Honyama fragrant tea Koju, their Koshun sencha, and a semi-fermented hojicha. The Koshun sencha won the Japanese Tea Grand Prix Special Award in the aromatic tea category.

So if you have read that a farm in Honyama won an aroma prize, that was Koshun. Koju was also among the teas the farm took there as award recipients, one of twenty put to a public tasting vote, but it is a different plant from the one that won that category. Our profile of the Koshun cultivar covers the registered one.

How one tree in the corner of a field became a cultivar

Marutaka Farm sits at Koseto, in Aoi Ward, Shizuoka City, inside the Honyama district. About twenty years ago, by his son's account, Takahashi Tatsuji started working outside Yabukita. The trigger was a piece of equipment: he bought a 台湾製釜炒り機, a Taiwanese-made pan-firing machine, and then needed something to fire in it.

The farm's own page describes the search in a single sentence. He withered zairai plants and every named cultivar he could get hold of, and checked the aroma 「何十種、百回以上」, across dozens of varieties, more than a hundred times. His son Takahashi Kazuaki, who now runs the shop, names Oiwase and Okumidori among the ones that did not work.

Then he found it. Not in a breeding plot and not in a nursery catalogue. A tea plant 自生していた, growing of its own accord, 畑の隅, in the corner of a field. He took cuttings and spent several years multiplying it into a garden.

Kazuaki's description of his father is the part I keep returning to: 「とにかく父は研究熱心なんです。遊びも仕事のうちというか、お茶への情熱がすごくて」, roughly, my father is relentlessly curious, there is no line between play and work, his passion for tea is enormous. That is the actual mechanism behind this cultivar. Not a breeding programme. One man who could not walk past an unusual bush without picking a leaf and smelling it.

Worth remembering what he was working against. Twenty years ago, Kazuaki says, 「萎凋が進んだ煎茶はあまりよくないものとして扱われていた」: sencha that had gone withered was treated as a fault, not a style. Almost no farm in Japan was making semi-fermented tea on purpose. He knows of one other, in Ikumi in Shimada City, who had started ten years earlier.

What does the name Koju mean?

Fragrance and long life. 香 is aroma, 寿 is longevity, and the farm wrote down exactly how it got there: 「香りに特徴があり、白毛が多いことから、（白眉→長寿）香寿と名付けました。」

Two things went into the name. The aroma, which is 香, ka or ko, fragrance. And the buds, which carry unusual amounts of 白毛, white down. Heavy white down on a bud suggested 白眉, hakubi, white eyebrows, which in Japanese and Chinese imagery belongs to someone very old. White eyebrows lead to 長寿, choju, long life, and long life gives you 寿, ju. Fragrance plus longevity: 香寿.

The farm carries the same idea into the product subtitle with a classical phrase it uses as a house motto, 〜茶は寿（いのちなが）らふ〜. That is old poetry about long life, not a statement about what the tea does to anyone, and I am repeating it only because it explains the character choice.

Most cultivar names in Japan are descriptive or geographic. Yabukita means north of the thicket. Our article on Yamatomidori covers a plant named for the old province it came from. Koju is named for a physical feature of its own buds, which tells you the person naming it was looking at the plant very closely.

Why does Koju smell like grapes?

Nobody has confirmed why. The producer and three independent tasters all describe white grape or muscat, and the one compound anybody has proposed does not match that tasting note. The farm says 「果実を思わせる香り」, an aroma that suggests fruit. Shinkoju says 「まるでマスカットのような甘く爽やかな香り」, a sweet, fresh aroma just like muscat. Florent Weugue of Thes du Japon, who reviewed it in 2016, describes white grape with a discreet touch of mint in the brewed aroma. On the palate he finds white grape again, with an impression of sweet red fruits.

The retailer LUPICIA offers a chemical explanation. They attribute the aroma to アントラニル酸メチル, methyl anthranilate, the compound shared with grapes and melons. That is a retailer's statement rather than a published assay of this specific cultivar, and it should be read that way.

The compound itself is well documented independently. PubChem records methyl anthranilate as "a clear colorless to tan liquid with an odor of grapes" and "grape-like or orange aroma". It is the methyl ester of anthranilate, it is a natural constituent of Camellia sinensis, and it is one of the signature volatiles of jasmine tea.

Here is the part nobody seems to have noticed. The Japanese tasting word is muscat, and muscat aroma does not come from methyl anthranilate. A 2010 study in BMC Plant Biology puts it plainly: the sweet, floral flavour typical of Muscat grape varieties is "due to high levels of monoterpenoids (geraniol, linalool and nerol)" (Emanuelli et al., BMC Plant Biology, 2010).

Methyl anthranilate belongs to a different grape entirely, the Concord register, the one you meet in American grape candy and grape soda. So the tasting note and the proposed compound are pointing at two different grapes.

That is not a gotcha. It is a fair description of where the evidence stops: nobody has published an assay of Koju's volatiles. I am leaving it open rather than picking whichever version sounds better, because the honest answer is that the aroma is real, attested by the producer and three independent tasters, and chemically unconfirmed.

Where did Koju come from? The Meiji black tea theory

Nobody knows, and no genetic test has been run. The leading theory is that Koju is a natural cross between a Japanese tea plant and an Assam-type variety left growing wild in the Honyama mountains after the Meiji-era push to make black tea. Every source that states it marks it as a theory, and so does this one.

Weugue's 2016 note calls Koju a selection from a zairai seed plant, probably 印雑 inzatsu type, a Japanese and Assam crossing. LUPICIA words it the same way and is careful with the verb: 「日本の品種と外来品種であるアッサム系の品種の自然交配で生まれたと考えられている」, considered to have been born from a natural cross between a Japanese variety and a foreign Assam-type variety.

The mechanism is the interesting part, and it is a piece of local history. LUPICIA's note on the region: 「本山は明治期に紅茶作りが奨励されたこともあり、多くの外来品種が山中などに残って自生しています。」 Honyama was pushed to make black tea during the Meiji period, and many of those imported varieties are still growing wild in the mountains there. Which means there is loose Assam-lineage pollen in those hills, more than a century after the black tea push ended. Pollen from one of those survivors landing on a local seed-grown bush would produce exactly what Koju appears to be.

Two pieces of physical evidence fit. The plant grows unusually large leaves, which is an Assam-type trait, and both the farm and Shinkoju describe it in the same words: 「葉が大きく育つのが特徴で直立型のお茶」, large-growing leaves, upright habit. And the aromatic Assam-lineage cultivars behave this way, which is why our page on Benifuuki, a deliberate Japanese and Assam cross, is the useful comparison.

Fit is not proof. The plants in those mountains have never been catalogued. No marker study has been run on Koju. A seedling that simply appeared in a field corner is the hardest possible case to reconstruct. Compare Z1, where even a plant inside the national breeding system carries an origin the register leaves half-explained.

How Koju is made: withering, kill-green, rolling, pan-firing

Marutaka publishes its own process page for 本山の香り茶, which is unusual and makes this one of the better documented small-farm methods in Japan. Four steps.

1. Withering, 萎凋 icho. Leaf is laid out in sun or indoors. The sun raises leaf temperature and wilts it, enzymes inside the leaf go to work, aroma appears, and the green deepens. The farm adjusts for the strength of the sun and for whether the pick is ミル芽, tender bud, or 硬葉, hard leaf. They turn the leaf over to even out the sun and agitate it indoors to stimulate it. Kazuaki gives the duration: 「萎凋は6〜10時間くらい」, roughly six to ten hours.
2. Kill-green, 殺青 sassei. The withered leaf goes into a pan heated to over 200°C, which destroys enzyme activity and stops oxidation where the maker wants it. Water inside the leaf boils off with a パチッパチ crackle. The farm explains why it does not burn: there is still moisture inside the leaf, so in that short window it cannot scorch. Steam comes out of the front of the pan.
3. Rolling, 揉捻 junen. A rolling machine draws the internal moisture to the surface and twists the leaf into shape. The one thing to avoid is over-rolling, which tangles the leaf into 玉, balls.
4. Drying, 乾燥. A hot-air dryer runs first, because leaf straight off the roller is wet on the surface and would stick to the pan. Then it is finished in the pan-firing machine. The farm notes this is the stage where the aroma really arrives, described as a ふわっ〜 of good smell coming through the hot air.

Worth noting how unusual that last step is at all. MATRA, which sells the tea, calls pan-firing 「今では日本国内でも1%未満になってしまった幻の製法」, a phantom method now used for under one per cent of tea made in Japan, and says the farm brought in a well-known Taiwanese tea master to share technique. That is a retailer describing its own product, so treat the figure as an order of magnitude rather than a measurement, but the direction is not in doubt: nearly all Japanese tea is steamed, not pan-fired.

The result is leaf that has been curled rather than needled, described as 丸くカールした勾玉状, curled into the shape of a magatama bead, and a liquor the farm calls 透明感のある黄金色, a clear golden colour.

One consequence matters more than it sounds. 「葉の組織が壊れにくい製法ですので、普通の煎茶に比べ、成分の抽出がゆっくりです」: the method leaves the leaf tissue largely intact, so extraction is slower than it is with ordinary sencha. Steeping runs longer, it is 煎がきく to around four infusions, and the retailer MATRA calls it a スロースターター, a slow starter. Spent leaf keeps its shape well enough that the farm suggests brewing in glass and watching it reopen.

Is Koju green tea or oolong?

Neither label fits cleanly, which is the honest answer and also the interesting one.

By oxidation level it sits between them. The farm lists 発酵度 as 軽発酵, light oxidation, and shelves it under 半発酵茶, semi-fermented tea. Weugue reached the same place from the other direction: a semi-oxidised tea with light oxidation that could almost be classified as kama-iri cha, Japanese pan-fired green tea, and he called it a must-try Japanese wulong.

By method it is a Japanese tea made with Taiwanese logic. The withering step and the pan-firing finish belong to the oolong tradition. The plant, the place and the maker are Japanese. LUPICIA sells it as 本山釜炒り新茶, Honyama pan-fired new tea, and lists the ingredient as 緑茶, green tea, which tells you the category system has not really made room for it.

If you need one sentence: Koju is a Japanese tea, lightly oxidised on purpose, finished in a pan, from a plant that is not in the register. Every one of those clauses is unusual on its own.

How do you brew Koju?

Start at 80°C for about two minutes, 3g of leaf to 150ml of water. That is the producer's own recommendation and the coolest, slowest of the three published methods. The other two go hotter and shorter, and slow-extracting leaf is why a twenty degree spread still gives you a drinkable cup.

Start with the producer's numbers. Marutaka gives 3g of leaf to about 150ml of water at 80°C, drawn off completely, for about two minutes, across infusions one to three. For cold brew they give 10g to 800ml, left in the fridge for six to ten hours.

LUPICIA pushes harder: 4 to 5g per 150ml cup, 90 to 100°C, 45 seconds to a minute, one to two infusions.

Weugue went tighter and cooler: 3g to 80ml at about 80°C for 90 seconds, which he found gave a very crisp, clear result, and he noted that stronger parameters were harder to handle.

Every source that mentions cold brewing recommends it. Weugue says the characteristic flavours stand out just as strongly cold, perhaps more, playing on the aftertaste. The HONYAMA88 interviewer, tasting a cold brew that had steeped overnight, was 圧倒される, overwhelmed, by the muscat-like flavour and aroma.

What does Koju taste like?

Fruity and gently spreading, with white grape on the palate, sweet red fruits behind it and a short finish. Four independent tastings, and they agree more than most cultivar notes do.

The farm's own description is restrained: 「フルーティーで口の中でやさしく広がります」, fruity, spreading gently in the mouth, with コク and うま味, body and umami, in the cup. Shinkoju is less restrained: a fruit aroma you would not believe came from tea, like muscat, and 「滋味は濃厚ではありませんが奥深さがあり」, a liquor that is not thick but is deep.

Weugue's 2016 note is the most detailed in any language. Dry leaf: slightly tangy, lemony, a sense of juicy fruit. Brewed aroma: white grape with a discreet touch of mint, which he then admits also reads as a pasta smell, and says it is nicer than it sounds. On the palate: white grape more clearly, with an impression of sweet red fruits, and a feeling of freshness despite rich flavours. The liquor stays clean across infusions, which he credits to the quality of the withering and oxidation.

He also records the one criticism I have seen anywhere, which is worth repeating because almost nothing else about this tea is critical: a possible lack of length in the finish. He argues the shortness is part of why it reads as refreshing. I would rather quote a reviewer's reservation than pretend a tea has none.

Where is Koju grown?

Koju is grown in Honyama, and sold by the farm that found it. Honyama sits in the valleys of the Abe and Warashina rivers above Shizuoka City, and it is a premium tea district with a long history. Tradition holds that the priest Shoichi Kokushi brought tea seed back from Song China and sowed it here around 1241, which is why Honyama is often called 静岡茶発祥の地, the birthplace of Shizuoka tea. Take that as tradition rather than settled history, which is how the sources themselves frame it.

The Koju garden is grown 自然仕立て, natural form, meaning the bushes are not clipped into the usual rounded hedge. That looks romantic and is mostly just hard. A natural-form garden cannot be machine-picked, so five or six pickers are brought in and the leaf comes off by hand at the right moment. Marutaka runs about ten gardens in different places, and Kazuaki's current worry is not weather or price, it is that the pickers are ageing and there are fewer of them every year.

Can you buy Koju?

Yes, from three places, and in small amounts. Marutaka sells 30g of hand-picked Koju direct for 1,200 yen including tax. LUPICIA has carried a 2026 Honyama pan-fired version, 40g for 1,800 yen. MATRA sells 15g and states 「日本でただ一軒しか作っていない」, that only one farm in all of Japan makes it. The same listing notes that the Takahashis have recently begun 株分け, dividing plants out to several other growers, so that may not stay true.

We do not sell Koju and we have no relationship with Marutaka Farm. If you want it, buy it from the people who found it. What we do sell is organic Japanese sencha, sourced the same way, by going and asking.

Frequently asked questions

Is Koju a registered tea cultivar?

No. Koju does not appear in the Chanorin register of Japan's national tea cultivars, and Shinkoju states that it has no plant variety registration with the Ministry of Agriculture, Forestry and Fisheries. It is described as a 民間(個人)育種品種, a privately bred cultivar, with a provenance of 実生選抜, seedling selection.

What is the difference between Koju and Koshun?

They are different cultivars that share the character 香, fragrance. Koshun (香駿) is a registered Shizuoka cultivar with a documented cross behind it. Koju (香寿) is an unregistered seedling selection from a single plant found growing wild in a field corner in Honyama. Marutaka Farm is known for both, which is part of why they get confused.

Is the grape aroma in Koju chemically confirmed?

No. The proposed explanation is methyl anthranilate, the compound behind grape aroma, which LUPICIA says the cultivar contains. PubChem describes methyl anthranilate as having an odour of grapes, and it occurs naturally in Camellia sinensis. No published assay of Koju's volatiles exists, so treat the chemistry as a plausible explanation rather than a measured one. The aroma itself is attested by the producer and by several independent tasters.

How is Koju classified for sale?

Inconsistently, because it sits between categories. The producer lists it as 軽発酵, lightly oxidised, and sells it as a semi-fermented tea. The reviewer Florent Weugue called it a lightly oxidised tea that could almost be classified as kama-iri cha, Japanese pan-fired green tea, and described it as a Japanese wulong. LUPICIA lists the ingredient simply as green tea.

Who created the Koju cultivar?

Takahashi Tatsuji of Marutaka Farm in Honyama, Shizuoka, found the original plant growing on its own in the corner of a field and propagated it from cuttings over several years. He had been making tea in Shizuoka for close to fifty years at that point. The farm is now represented by his son Takahashi Kazuaki, a certified Japanese Tea Instructor.

Can you cold brew Koju?

Yes, and every source that mentions it recommends it. Marutaka gives 10g of leaf to about 800ml of water, refrigerated for six to ten hours. Reviewers report the grape character comes through as strongly cold as hot, and in at least one account more strongly.

A note from me

I have never tasted Koju. We do not sell it. Across five weeks filming with growers in Shizuoka and Kyushu, nobody mentioned it to me once. That is what you would expect of a tea from one farm with a small planted area.

What pulled me in was the shape of the story. We talk about cultivars as though they arrive from research stations with numbers attached, and a lot of them do. Our guide to Japanese tea cultivars is mostly full of those. Koju is the other route. It is older, rarer and much harder.

A man buys a machine he cannot use properly. He spends years smelling leaves off every strange bush he passes. He finds one plant nobody had noticed, in the corner of his own field. Then he spends several more years turning that one plant into a garden, and names it after the white hairs on its buds.

The two aromatic cultivars I wrote up recently, Musashikaori and Sayamaakari, both came out of prefectural programmes with paperwork behind them.

This one has a father, a son, and about a hundred aroma trials. That is worth recording while the people who did it are still around to be asked.

Buying tea for a shop, cafe or studio? We supply organic Japanese tea at wholesale, with the same attention to where each lot comes from. See our wholesale program, or start with the organic sencha we do sell and tell us what you are pouring.

Christian
Founder, One with Tea

Sources

1. 丸高農園 (Marutaka Farm), the producer's own Koju page: product page, in Japanese. The primary source for almost everything here: the natural seedling origin, the Taiwanese pan-firing machine and the hundred-plus aroma trials, the field-corner discovery and the cuttings, the naming from 白毛 through 白眉 to 長寿, large leaves and upright habit, 軽発酵, the golden liquor, the 30g price, and the producer's own brewing numbers hot and cold.
2. 丸高農園, 本山の香り茶 製造工程: production process page, in Japanese. The four stages in the farm's own words, including the over-200°C pan, the crackle, the reason the leaf does not scorch, why the hot-air dryer runs before the pan, the curled leaf, and the slow extraction and four infusions.
3. 丸高農園, farm diary, 2 and 4 December 2016: Nihoncha Award 2016 entry, in Japanese. Koju as one of the awarded teas at the Tokyo Tea Party, twenty teas in the public tasting vote, and the Koshun sencha taking the Japanese Tea Grand Prix Special Award in the aromatic tea category.
4. HONYAMA88, interview with 高橋一彰, 28 February 2023: column, in Japanese. The twenty-year timeline, Oiwase and Okumidori among the failures, the six to ten hour withering, natural-form gardens and five or six hand pickers, the ageing picker problem, the cold-brew tasting, and the description of his father.
5. 心向樹 (Shinkoju), 品種茶一覧: cultivar reference list, in Japanese. The decisive line on registration, 「＊農林水産省の品種登録はされておりません」, plus 民間(個人)育種品種, 実生選抜, the muscat note, and the suitability of the pan-firing method.
6. LUPICIA, 本山釜炒り新茶 香寿: product page, in Japanese. The methyl anthranilate attribution, the Japanese and Assam natural-cross theory stated as a theory, the Meiji black tea history that leaves foreign varieties growing wild in the Honyama mountains, Shoichi Kokushi and the 1241 tradition, nearly fifty years of tea making, and their own hotter and shorter brewing numbers.
7. Japanese Tea Sommelier (Florent Weugue, Thes du Japon), 5 July 2016: Koju, a unique semi-fermented tea from Hon.yama, in English. Verified live. The unregistered-cultivar description, the zairai and inzatsu theory, the full tasting note, the Japanese wulong framing, the brewing parameters, the cold-brew recommendation, and the one reservation about length.
8. PubChem, methyl anthranilate, CID 8635: compound record. "A clear colorless to tan liquid with an odor of grapes" (NTP, 1992) and "grape-like or orange aroma", plus its status as a natural constituent of Camellia sinensis.
9. Emanuelli F, Battilana J, Costantini L et al., "A candidate gene association study on muscat flavor in grapevine (Vitis vinifera L.)", BMC Plant Biology, 9 November 2010, DOI 10.1186/1471-2229-10-241: open-access full text. Cited for one line only: muscat flavour in grapes is carried by monoterpenoids, geraniol, linalool and nerol, which is what makes the methyl anthranilate attribution worth questioning.
10. MATRA, Koju listing: product page, in Japanese. The one-farm-in-Japan claim, the recent sharing of plants with other growers, and the slow starter note. Caveat: this page spells the breeder 高橋達治, where the farm's own site, LUPICIA and Shinkoju all give 高橋達次.
11. Chanorin registered cultivar list (附表１ 茶農林登録品種一覧, 2005 edition), hosted by minorien: list, PDF in Japanese. Cited for an absence: 香寿 does not appear in it. Checked against a Yabukita control that returns fourteen hits, so the zero is a real zero. Note the site soft-404s, so this was validated on contents rather than status code.