Ho-onji temple at Hiekoba in Arita, where Baek-pa-sun's great-grandson built a pagoda for her in 1705

Gosu: the Japanese Underglaze Behind Arita Blue and White

In 1618 porcelain stone was found at Izumiyama in Arita. A Korean potter named Baek-pa-sun, brought to Japan two decades earlier in Hideyoshi's invasions, led her clan and several hundred potters over to the new ground and ran the community there until she died at ninety-six. Arita calls her the mother of its porcelain.

Four hundred years on, the only company in Arita still making the blue that porcelain is painted with is run by her thirteenth-generation descendant, Sosuke Fukaumi. It is called Fukaumi Shouten, and the pigment it makes is gosu.

呉須, gosu

Gosu is a cobalt-bearing pigment painted onto unfired or bisque-fired clay and then covered with a clear glaze. The colour develops under the glaze rather than on top of it, so the line stays crisp and the surface stays smooth.

Blue and white porcelain dish painted in gosu with a persimmon and lattice pattern, fired on Amakusa clay to 1280C in reduction

Gosu on porcelain, fired on Amakusa clay to 1280C in reduction. Photograph: Fukaumi Shouten.

In its oldest form it is a stone. The ore is called asbolite, and cobalt is only part of it. Silicic acid, manganese, aluminium, iron and copper are in there too, in proportions that change from one deposit to the next, and that variation is why antique blue and white is not one blue. The grey-blues, the near-blacks and the violet-leaning blues all come from whatever else happened to be in the ground alongside the cobalt.

A 100g jar of Japanese gosu underglaze paste, blue and white label, on a pale ground

What we stock is the blended kind. Cobalt oxide, iron oxide, nickel oxide, manganese dioxide and kaolin, ground together in a known proportion rather than dug up at whatever proportion the seam happened to offer. Everything in it still came out of the ground. It was worked out in Arita in the Meiji period by the family that still makes it, and it is what Arita blue has meant for well over a century. What you gain is a blue that does the same thing twice.

Underglaze, and where gosu sits in it

The words in this corner of ceramics get used loosely.

Glaze is the glass. Mostly silica with fluxes to melt it, and its job is to form a hard, sealed, usually shiny surface.

Underglaze is colour that goes on before the glaze. It carries colouring oxides and stains with only a little of the glass-forming material, so it does not melt and flow on its own, and it stays exactly where your brush left it. A clear glaze goes over the top and the fired colour sits sealed beneath it.

Slip is liquid clay. Mostly clay body, so it shrinks with the pot and belongs on greenware. Put it on bisque and it tends to flake off, because the pot has already done its shrinking and the slip has not.

Engobe sits between slip and underglaze: less clay than a slip, more flux, which lets it bond to drier work including bisque. Commercial underglaze is, in effect, a coloured engobe with brushing additives.

Overglaze, enamel, china paint and lustre all go on top of a fired glaze and are fired again lower, which makes them a different family altogether.

Gosu is an underglaze in the strict sense: colour below the glaze, fired once with it.

How gosu differs from the underglaze you already own

Thick brush strokes of blue gosu underglaze paste

The paste, undiluted. Photograph: Fukaumi Shouten.

It is a concentrate. A bottle of commercial underglaze is formulated to be brushed straight from the bottle at a fixed strength. Gosu arrives as a paste you thin with water to whatever the job needs. More work, more control. The same jar gives you a hairline so pale it is almost grey and a wash dark enough to read as black.

It carries no brushing medium. Commercial underglazes contain gums and binders so they flow off a brush and stick to bisque. That medium can also sit on the surface and interfere with how the clear glaze is absorbed, which is why potters sometimes re-bisque decorated work before glazing. Gosu has no such layer to burn off. The trade is that nothing is holding it to the pot either, which is why the maker suggests a little gum arabic if the piece will be handled between painting and firing.

It is already ground. Fukaumi grind fine, by their own account on a stone mill, and they are explicit that their gosu requires no grinding.

Cobalt is among the most stable colourants in ceramics. The common complaint about commercial underglazes is that pinks, warm tones and some yellows fade or disappear through a cone 6 glaze firing. Cobalt blue does not do that, which is why blue and white survived four hundred years of kilns far less controlled than yours. If you have been let down by a colour burning out, blue is the colour least likely to do it.

What is harder about it. You are doing the dilution, so your first tile is a real test. There is no shade card that matches your kiln. And the jars look almost identical to each other, which matters when three of them are blue.

Where the blue came from

Before the Meiji period, Japan imported all of its gosu.

The oldest source was the Islamic world. That cobalt was known in China as Huiqing, or Huihuiqing, because Islam was also called Huiism there. When those imports stopped during the Ming dynasty, China went looking for its own and found it: mines in Yunnan, Guangdong, Jiangxi and Zhejiang, including Shaoxing and Jinhua. In Japan, Chinese natural gosu became known as Kara gosu.

Through the Edo period it reached Japan from China by way of Nagasaki, and Edo blue and white was painted with it. Japan had a little of its own, mined from around 1665 out of the Tertiary sediments in the belt running from Seto to Tajimi and Toki in what is now Aichi Prefecture, under the control of the Owari clan. It was full of impurities and not much good for painting, and when high-purity cobalt oxide became cheap to import in the Meiji period the mine closed.

Even the word is uncertain. Chinese names for their own domestic gosu were different, which suggests the word was coined in Japan; one theory has it as a corruption of Wuzhou, near the mouth of the Yangtze. It was not in common use until the late Edo period. Before that, potters called it chawanyaku.

That account is Fukaumi Shouten's, drawn in part from the director of the Arita History and Folklore Museum, and published as History of natural gosu.

What is in the jar

The kaolin in that blend is there for a reason that will make sense to anyone who has had underglaze lift off a pot: it improves adhesion between the pigment and the clay beneath it. Alumina is sometimes added as well.

In the kiln, those components combine into a spinel structure, and the spinel dissolves into the glassy part of the glaze above it. The colour is not sitting on the pot under a window. It has gone into solution in the glass.

Every raw material in gosu melts above 1300C, which is hotter than most of us fire. It works anyway because the gosu and the glaze are fired together, and the glaze brings the whole thing into melt at a lower temperature than the pigment could manage alone.

The chemistry above is from Fukaumi Shouten's Ingredients of Gosu, which cites Masataro Onishi's Traditional Techniques of Pottery (Rikogakusha, 1978) and the History of Seto City (1967).

How it is made

Six stages, and about a month from raw material to a sealed bag.

Raw material for gosu being scooped from a stainless bowl at Fukaumi Shouten, with a small dish of dark blue pigment alongside

Raw material at the works in Arita. Photograph: Seitaro Iki. Provided by Fukaumi Shouten.

The raw materials are prepared with cobalt oxide as the main ingredient and mixed to an even batch. That batch goes into a baking tray and is fired. What comes out is dried, crushed and ground, then wet-ground and blended, dried again, and finally milled on stone to bring the particle size down.

A stone mill at Fukaumi Shouten with wet blue gosu running from it

The mill. Photograph: Seitaro Iki. Provided by Fukaumi Shouten.

That last stage is the one the company singles out. They grind on marble millstones, and say theirs is the only gosu in Japan still made that way, which is why it suits hand-painting. Their best grade comes off the stones at a particle size of one micron.

The milling room at the Fukaumi Shouten gosu works in Arita, with wet floors and blue-stained machinery

The milling room. Photograph: Seitaro Iki. Provided by Fukaumi Shouten.

What one micron means. Ceramic raw materials normally arrive at 200 to 325 mesh, which is 74 to 44 microns. One micron is a different order of fineness, and it is the reason the company gives for their gosu suiting hand-painting. It matches what their customers say about it too: Tsuji Seijisha call it extremely fine grained and easy to paint with.

That figure belongs to their gosu, which their catalogue keeps as a separate range from their coloured underglazes. Of the ten we stock, three are gosu in that sense: Kaiheki, Kaiheki in Toku 2 and Ancient 3 No.3. The other seven come from the coloured underglaze range, which Fukaumi describe as made for fine detail work.

That account of the process, and the particle size, are from the company profile Fukaumi Shouten supply to their stockists.

The family that made modern gosu

Baek-pa-sun and her husband Fukaumi Soden were brought to Japan in the 1590s, during Toyotomi Hideyoshi's invasions of Korea, and first made pottery at the Kotouge kiln in Uchida village, Takeo. The family name came from a misunderstanding: the lord of Takeo, Goto, heard their home town of Gimhae as Shinkai, and wrote it with the characters that read Fukaumi. Soden died in 1618, the year the porcelain stone was found. In 1705 Baek-pa-sun's great-grandson built a pagoda for her at Ho-onji temple, the temple at the top of this page.

The chemistry comes later. In 1870 a German chemist named Gottfried Wagener was living at Dejima in Nagasaki, and Arita invited him up to advise as an expert in chemistry. Working with him, the head of the family, Heizaemon Fukaumi, and his sons Suminosuke and Takeji worked out how to make gosu rather than mine it. Inexpensive manufactured gosu with a stable colour began to be used in Arita in place of the very expensive natural gosu imported from China. They also perfected Hongamanishiki, a polychrome palette that survived firing at 1300C. It gave Arita a blue that behaved the same way every time.

What followed is the part museums know. In 1875 the Fukaumi sons Suminosuke and Takeji founded Koransha with the Fukagawa, Tsuji and Tezuka families. In 1876 Koransha won a gold medal at the Philadelphia Centennial Exhibition. Suminosuke went on to found Seiji Kaisha, and the company says his work, made under the name Toshikian Kizo, is held by the Tokyo National Museum and the V&A. Takeji was unofficially selected as an Imperial Household Artist in 1897 and died of tuberculosis before the honour was confirmed.

After the war, Tatsuji Fukaumi, born 1911, decided that what made Arita ware Arita ware was not the pots but the colour. He closed the family kiln and turned the business into a pigment works. It was incorporated as Fukaumi Shouten Co., Ltd on 10 June 1961.

Full version: History of Arita Ware: The 400-Year Legacy of the Fukaumi Family and Baekpasun.

The cobalt problem

Cobalt is element 27, and its name comes from the German Kobold, a goblin, because the ore misled the miners who found it. It is a rare metal, not because it is scarce in the crust but because it almost never forms deposits of its own. It comes up as a by-product of copper and nickel mining, in minerals with names like cobaltite, heterogenite, linnaeite and smaltite. Its supply is governed by the price of copper, not by anybody's demand for blue.

The distribution is extreme. On the 2021 figures Fukaumi cite from JOGMEC, the Democratic Republic of the Congo holds about 48% of world reserves, with Australia at 20% and Cuba at 7%, and produces about 70% of world output. One country, two thirds of the supply.

And pottery is no longer the customer that matters. Cobalt goes into lithium-ion battery cathodes, superalloys for jet engines, cutting tools and permanent magnets. The shift to electric vehicles has made battery demand the dominant force in that market, and a pigment works in Arita is bidding against it. Fukaumi put it bluntly: cobalt is becoming blue gold, and they manage it with procurement routes running back to the Meiji era and long-term stockpiling.

The same thing is happening to tungsten carbide trimming tools, for the same reason. A studio material, priced by industries that have never heard of ceramics.

Their piece is Cobalt Sourcing and Supply Risks, citing JOGMEC's Mineral Resources Material Flow 2021.

Why it matters that they are the last one

In March 2026, the company reports, the number of gosu manufacturers in the Hizen ceramic region fell from two to one. Fukaumi Shouten, now led by the thirteenth generation, Sosuke Fukaumi, is the only company still making gosu in Arita. They also say they are the only one in Japan still grinding it on marble millstones. Both are their own statements and we have not verified them independently.

By the company’s own figure, sales of Arita porcelain have fallen by 82% over the past three decades.

A red rack of small fired porcelain sample dishes at Fukaumi Shouten, each painted in a different gosu

Fired samples at the works, one dish per colour. Photograph: Seitaro Iki. Provided by Fukaumi Shouten.

That is most of why we brought it in. A pigment made in one town for four hundred years, by a family descended from the potter who started that town's porcelain industry, now made by one firm.

GOSU pv. Film: Fukaumi Shouten.

What we stock

A jar of gosu underglaze paste held in a hand, showing the size of the 100g pot

Ten pastes, 100g each, imported directly from Arita.

The three gosu, from the cobalt blue range ground to one micron. £18 each:

Seven from the coloured underglaze range, which Fukaumi make for fine detail work. At £18:

And at £22:

All ten are in the Japanese Underglazes collection.

How to use it

Hexagonal test tiles showing Ancient 3 No.3 gosu on Amakusa porcelain, fired in oxidation to 1230C

The maker's own test tiles, at several dilutions. Photograph: Fukaumi Shouten.

The maker's own guidance first, then ours.

For line work, roughly one part paste to three parts water.

For a wash, a teaspoon of paste in 80ml of water gives a strong colour, the same teaspoon in 250ml gives a pale one. Those two numbers bracket almost everything you will want to do.

A little gum arabic helps the colour stay put if the piece will be handled between painting and firing.

Firing: 1230 to 1300C, oxidation or reduction, according to the maker. That covers a normal stoneware or porcelain glaze firing, and these work in an electric kiln with nothing special done.

Storage: airtight. If a jar dries out it is not ruined. Grind it down and mix it back to a paste with water, which is a quiet advantage of buying pigment as paste rather than as ready-mixed bottled underglaze.

On how long it keeps, the clearest statement comes from a customer rather than the maker. Tsuji Seijisha, the Arita kiln that has produced porcelain for the Imperial Household since the Edo period, say of it: “Fukaumi Shouten’s gosu is extremely fine grained and easy to paint with. It doesn’t peel when overlayered and is usable even if stored for up to five years.”

Our own notes

Thin coats, built up. Cobalt is powerful, and a blue that looks right in the jar can fire almost black. Start paler than you think and add a second coat for depth.

Use a soft, pointed brush and keep it loaded. Japanese brush painting on porcelain works by letting the brush taper the line rather than by pressing.

Test tiles before a real piece. The dilution is in your hands rather than the manufacturer's.

We have not fired our own tiles through the whole range yet, so treat the colours on your screen as a guide to the pigment, not a promise about the fired surface.

Questions people ask

What is the difference between underglaze and glaze?
Glaze is the glass that seals and finishes the pot. Underglaze is colour that goes on before it and does not melt and move. Used together, the colour sits sealed under a clear coat.

Do you have to put a clear glaze over underglaze?
For anything functional, yes. Fired on its own, underglaze stays matt, slightly porous and vulnerable. Leaving it bare is a legitimate choice on sculpture and a bad one on a mug.

Can you put underglaze on greenware, or does it have to be bisque?
Either works. Bisque is easier to handle because the pot is stronger and your brush water will not soften it. Greenware suits work where you want the decoration to shrink with the body.

How many coats of underglaze?
Commercial underglazes are usually two to three, three for pale colours. Gosu does not work that way, because you set the strength when you dilute it. One well-loaded pass of a strong dilution does what three coats of a pale one would.

Does underglaze change colour after firing?
It can, and it is a common disappointment. Warm colours, pinks and some yellows are the ones that fade or vanish through a cone 6 glaze firing. Cobalt blue is the opposite case: it is among the most stable colourants in ceramics, which is why blue and white porcelain exists at all.

What is the difference between underglaze and slip?
Slip is liquid clay, mostly clay body, and it belongs on greenware because it shrinks with the pot. Underglaze carries far less clay and much more colour, and will sit happily on bisque.

Can you mix underglazes together?
With each other, yes. These are pigments in water, so mix and test. Mixing underglaze into a glaze is a different matter and generally a poor idea: the colour gets lost and the glaze behaviour changes.

Will a coloured glaze over the top still show the underglaze?
Usually not. Most coloured glazes cover underglaze completely. Clear or genuinely transparent is the glaze you want.

What temperature does it fire to?
1230 to 1300C, oxidation or reduction, per the maker. A normal cone 6 to cone 10 glaze firing.

Does it need grinding before use?
No. Fukaumi grind it, and say so explicitly. You thin it, you do not mill it.

How far does 100g go?
A long way, because you dilute it. A teaspoon makes 80ml of strong wash or 250ml of pale.

It says red. Is gosu not blue?
Gosu in the strict sense is the cobalt blue, and Fukaumi’s catalogue keeps it as its own range. Three of ours come from it: Kaiheki, Kaiheki in Toku 2 and Ancient 3 No.3. The other seven come from their coloured underglaze range, made for fine detail work. Red Chinese Gosu No.3 carries gosu in its name and still sits in that second range.

Is it food safe?
We do not make food safety claims about any material, and nobody selling you a pigment honestly can: what matters is the fired glaze over it, your clay body and your firing. Food Safety in Ceramic Glazes sets out where the responsibility actually sits.

Where to find them

All ten are in the shop at Railway Arch 848, Brayards Road, Peckham SE15 2AG, and online. If you want to see the blues next to each other before committing to one, come in.

Where to go next

Sources and credits

The photograph at the top of this page, and all photographs of fired ware, brushed paste, test tiles and the works at Arita, and the film, are provided by Fukaumi Shouten and used with their permission. Photographs by Seitaro Iki and Shinichiro Nakashima. The jar photographs are our own.

Fukaumi Shouten's own writing, by Sosuke Fukaumi, thirteenth generation of the family:

Works cited in turn by those pages: Masataro Onishi, Traditional Techniques of Pottery, Rikogakusha, 1978; Seto City, History of Seto City, Ceramics History Volume 3, 1967; JOGMEC, Mineral Resources Material Flow 2021.

Application and firing guidance supplied by the maker with the products.