ISIJ International
Online ISSN : 1347-5460
Print ISSN : 0915-1559
ISSN-L : 0915-1559
Regular Article
Gold Refining by Cementation with Salt at Sado in Early Seventeenth Century Japan
Eiji IzawaTetsuya Nakanishi
ジャーナル オープンアクセス HTML

2014 年 54 巻 5 号 p. 1098-1105


In the seventeenth century Japanese gold production became the highest rank of the world and Sado gold contributed nearly half to the domestic production. The key technology of massive gold production was gold-silver parting by cementation process with salt. During the excavation of the Sado Bugyosho (magistrate office) site, in 1996, gold refining remains were found below the 1647 fire horizon. A total of 29 furnaces were found in an area of east-west 17 m by north-south 8 m. The largest furnace is an elongated shallow tray-like furnace, which resembles closely the cementation furnace illustrated in eighteenth- and nineteenth-century picture scrolls of the mine. Large number of fragments of earthenware such as clay dishes, clay plates and clay rods were unearthed. Many of the earthenware were discolored from the original reddish color.

The chemical compositions of discolored samples are characterized by increase in Na2O, K2O, S, Cl and Ag, and decrease in SiO2 and Fe2O3 indicating the gold-silver parting reaction by common salt.

1. Introduction

Sado Island lies in northeastern Japan (Fig. 1) and gold placer in the island was documented in a book of the twelfth century. Nishimikawa is the old placer mining area in the island. Hard rock mining for silver ores started at Tsurushi and Niibo in the mid-sixteenth century. The silver-sulfide ores were smelted using a lead addition technology and silver was extracted by cupellation. In the late sixteenth century the mining area was extended further north to Aikawa.

Fig. 1.

Map of Japan and Sado showing the locations of gold mines, kinza office and port mentioned in the text.

In 1601 Sado Island came under the direct control of the Tokugawa shogunate and Sado Bugyosho (magistrate office) was constructed near coastal area in Aikawa in 1603–1604.1) After that the Sado gold-silver mine certainly became the largest gold producing mine in the seventeenth century world. The key technology of massive gold production was gold-silver parting by cementation process with salt.

In 1996, during the excavation of the Sado Bugyōsho site, gold refining remains were found below the 1647 fire horizon.1) A total of 29 furnaces were found. The largest furnace is an elongated shallow tray-like furnace, which resembles closely the cementation furnace illustrated in eighteenth- and nineteenth-century picture scrolls of the mine. Large number of fragments of earthenware such as clay dishes, clay plates and clay rods were unearthed.

This paper discusses gold fineness with deposit types and the importance of gold-silver parting technology in early seventeenth century Japan. The gold-silver parting technology at Sado will be described on the basis of the observation and analysis of the earthenware remains unearthed from Sado Bugyosho site.

2. Compositions of Gold Grains in Various Gold Deposits

The major source of gold in ancient world was gold-quartz veins of mesothermal deposits (of Lindgren’s classification)2) and gold placer derived from the veins. Other source was silver sulfide-gold-quartz veins of epithermal deposits and their gold placer (Fig. 2).

Fig. 2.

Historically important gold deposit types.

The gold mineral in mesothermal veins is native gold, which composition ranges typically from 75 to 90% gold3) or 88–93% gold.4) On the other hand gold mineral is electrum in epithemal veins. The compositions of gold grains range typically from 50 to 80% gold3) or 60–65% gold.4) Figure 3 shows the compositions of gold grains from historically important areas in Japan.5)

Fig. 3.

Composition of “gold grains” in the ores.5)

Primary ores of epithermal gold-silver deposits in Sado Island consist mainly of quartz veins, which contain silver sulfide (acanthite; Ag2S) and electrum (gold-silver alloy) and the bulk composition of ore is silver dominant6) (Fig. 4).

Fig. 4.

Composition of electrum4) and gold value expressed as Au/(Au + Ag) of ores6) at Sado (Aikawa).

3. Smelting Process at Sado

3.1. The Eighteenth and Nineteenth Centuries

Processes of ore dressing and smelting are described in several documents7) and picture scrolls8) in the eighteenth and nineteenth centuries. Also Percy9) describes smelting at Sado in 1868.

Picture scrolls and documents describe the dressing and smelting processes performed in the Sado Bugyosho (magistrate office). Careful ore dressing by gravity separation resulted in two kinds of concentrates, that is, electrum rich one, mizusuji, and silver sulfide rich one, yurimono.

Smelting of electrum rich mizusuji resulted in bullion, which according to a document,7) contains about 40% gold. Smelting of silver-sulfide rich yurimono yielded auriferous silver, which gold content was probably several %. Silver was extracted from the auriferous silver by the sulfur process and the products were silver sulfide and bullion (gold-silver alloy of about 40% gold). Silver was recovered from the silver sulfide.

Bullion formed from both mizusuji and yurimono must be refined by salt cementation. Powdered bullion (0.1 to 1 mm size) is mixed with salt, which must contain bittern. From the early nineteenth century document7) bullion:salt ratio is estimated to be 1:5 in mass. A pile of salt-bullion mixture (about 8 kg) is put on a clay dish and is set in two lines in the cementation furnace. A clay plate is put on each salt-bullion pile (Fig. 5 top-left).

Fig. 5.

Picture scroll8) depicting the operation of cementation at Sado in the early nineteenth century.

Total 44 to 45 piles are set in the furnace with charcoal. Then charcoal is burnt for 8 hours. The reacted salt-bullion pile is dip into water to cool and wash (Fig. 5 bottom). The process is repeated four or five times, then gold value increases to about 70% Au, which value is suitable for koban in the early nineteenth century (66% Au).

3.2. The Early Seventeenth Century

In the early seventeenth century, miners performed ore dressing and smelting in their working sites, probably in similar processes as described above. A document of 1617 refers the existence of furnaces for silver separation from auriferous silver using sulfur in the mining settlement in Sado.10) Smelting products (bullion containing 40% gold and crudely refined silver) and placer gold (60% gold contained) were collected by the Sado Bugyōsho and sent to Edo to make gold coin (koban). In Edo, craftsmen, who were associated with kinza (monopolized office to making gold coins), upgraded bullion using salt cementation.

In 1621, sujikin-yakisho (cementation workshop) and koban-nobesho (factory for making gold coins, koban) were attached to the Sado Bugyōsho. Craftsmen came from Edo7) and the cementation process was firstly introduced to Sado. Ore dressing and smelting, however, were done by individual group of miners near the mining sites.

From 1621 to 1695 Sado produced 1450000 pieces (taels) of gold coin (Keicho koban: 86% Au),11) that was equivalent to 22.4 t of gold. The koban, bullion and placer gold were sent to Edo together with the large amount of silver. Sado gold contributed nearly half of the domestic production in the seventeenth century and Japanese gold production became the highest rank of the world (Fig. 6).

Fig. 6.

World gold production in the early modern period. Asian production was not included in Soetbeer’s data.12) Japanese production was estimated in this study.

This early refinery was burned down in 1647’s fire and no detailed record of the refinery was remained. The refinery site excavated in 1996, therefore, important to examine the cementation process operated in early seventeenth century Sado.

4. Excavated Materials and Analytical Methods

4.1. Furnaces

In the southern part of the Sado Bugyōsho site, a total of 29 furnaces were found in an area of east-west 17 m by north-south 8 m (Fig. 7(A)). Three furnaces, which shape are unclear, and four rectangular furnaces (about 1 m wide) are probably older ones. Fourteen furnaces (average 80 × 50 cm and 25 cm deep) with a dividing wall, three circular furnaces (average 65 cm and 24 cm deep) and an elongated shallow tray-like large furnace (4 m long, 66 cm wide and 18 cm deep) seem to be used almost in the same period. Most of the furnaces are purplish red in color and charcoal dust stuck on the fire-hardened floor and they were considered to be cementation furnaces.1)

Fig. 7.

Comparison of several cementation workshops. (A) Cementation workshop at Sado in the early seventeenth century. (B) Two rows of the small circular cementation furnaces at Edo kinza (the koban coinage factory)13) in the early nineteenth century. (C) A long cementation furnace at Sado shown in the picture scroll8) in the early nineteenth century.

The smaller furnaces form two rows and this furnace arrangement resembles that of cementation furnaces in the workshop associated to the Edo kinza in the early nineteenth century (Fig. 7(B)). The long furnace resembles closely the cementation furnace illustrated in eighteenth- and nineteenth-century picture scrolls of the Sado mine8) (Fig. 7(C)).

4.2. Earthenware

Interesting finds are a large number of fragments of earthenware such as clay dishes, clay plates and clay rods (Figs. 8 and 9). Most of the earthenware were crushed into pieces and reinforced in the walls of the partitioned furnaces, especially many larger pieces were found in the dividing walls.1) The number of clay dishes is over 1500 pieces. The original diameter of the clay dish seems to be more than 20 cm, the edge-height is about 5 cm and the depression is 0.6–1.6 cm (Fig. 8(A)). Clay plates are quadrangle or round-shaped, 1.1–1.9 cm thick and the original diameter seems to be more than 15 cm. All the clay plate occurs as fragments and are discolored (Fig. 8(B)). There are many clay rods of prism- and cylinder-shapes, 2–3 cm in diameter and originally more than 12–13 cm long (Fig. 8(C)). Some rods were less reacted and has reddish color (Fig. 10(A)) but many were highly reacted and discolored (Fig. 10(B)).

Fig. 8.

Excavated various earthenware. (A) Clay dish D11-80 (estimated size is 30 × 20 cm with 2–3 cm thick) reacted. (B) Clay plate D10-12 reacted and discolored. (C) Clay rods showing bleached surface. Some rods have less reacted surface with reddish color (scale is 14 cm-long ball-point pen).

Fig. 9.

Clay dish D11-105 (less reacted and red color). (Left) Bottom side of clay dish has red color. (Right) Cross section of the clay dish showing bleached part along the crack and fractures. Small white spots are quartz grains.

Fig. 10.

Clay rods. (A) Clay rod D11-31A (prism) showing red colored surfaces and inside. Only one side of the prismatic surface was bleached. (B) Clay rod D11-56 (prism) showing discolored surface. Inside is less reacted and reddish color.

Although the earthenware was originally rich in iron oxide and of an orange color, many were discolored and cracked probably by the furnace work.

4.3. Analytical Methods

The chemical compositions of earthenware samples were determined with a Rigaku RIX 3100 X-ray fluorescence spectrometer (XRF), using pressed powder pellets. Also XRF mapped images with semi-quantitative measurements were obtained using an X-ray analytical microscope (Horiba XGT-5000).

The constituent minerals were examined using a Rigaku Rint 2000 and Ultima IV X-ray diffractometer (XRD) with a Cu target.

5. Results and Discussion

5.1. Chemical Changes of Earthenware

Chemical compositions are shown in Table 1. The compositions of discolored samples are characterized by increase in soda (Na2O), sulfur (S), chlorine (Cl) and silver (Ag), and decrease in silica (SiO2) and iron (Fe2O3) (Fig. 11).

Table 1. Chemical composition of earthenware (XRF analysis).
Sample no.120517-3D11-105D11-94D11-31D11-671RD11-671WD10-12D11-56
SampleNosaka clayClay dishClay rod
Clay rod
Clay dishClay dishClay plateClay rod
Raw material
Weakly altered
Weakly altered
Weakly altered
Moderately altered
Highly altered
Highly altered
Highly altered
SiO2 (%)

Fe2O3* = Total iron as Fe2O3; Others = As,W,Cr,V,Zr,Rb,Sr,Ba,Y,Nb; Au<0.005, Sb<0.01, Sn<0.01.

Fig. 11.

Changes of chemical compositions of clay dish by cementation process. Dry basis data recalculated from Table 1. “Nosaka clay” = row material (120517-3), “Red color” = weakly reacted red clay dish (D11-105), and “Discolored” = highly reacted part (discolored fringe) of clay dish (D11-671W).

The main row material of the earthenware is local clay called “Nosaka clay”, which is hydrothermally altered and weathered volcanic rocks of siliceous composition (dacite). Iron-rich red clay seemed to be added to the Nosaka clay to make various earthenware such as clay dishes, tuyere etc.

XRF mapped images by an X-ray analytical microscope shows the variation of several elements within the earthenware. A cross section of the clay dish D11-671 was analyzed (Figs. 12 and 13).

Fig. 12.

XRF mapped images of silver for a cross section of clay dish D11-671.

Fig. 13.

Variation of silver and iron (semi-quantitative values) in the clay dish D11-671 after. Iron decreased at the fringe and increased along the reaction front.

It is clear that the increase of silver is conspicuous at the fringe of the dish (Figs. 12 and 13), where a pile of salt-bullion mixture did not contact. Iron also decreased very much in the fringe portion and increased toward inside (Fig. 13). An iron-enriched zone is seen as dark line on the cross section (Fig. 13). In the cementation reactions iron is volatilized as iron chloride and lost from the earthenware. Also iron seems to migrate inward of the earthenware.

5.2. Chemical Reaction of Cementation Processes

Complex reactions of cementation proceed under moderate temperatures (less than 800°C = melting point of common salt). Sodium chloride reacts with water (H2O) from charcoal burning, which results in formation of sodium ion and hydrogen chloride (gas). This reaction will be accelerated by silica.15) In the case of Sado, original earthenware contained quartz and cristobalite, and meta-kaolin (Al2Si2O7) as silica sources. Then albite-forming reaction occurred:   

2NaCl + H 2 O + 4SiO 2 + Al 2 Si 2 O 7 = 2NaAlSi 3 O 8 + 2HCl

XRD data are shown in Fig. 14, which indicates decrease in silica minerals (quartz and cristobalite) and increase in sodium feldspar (albite).

Fig. 14.

X-ray powder-diffraction patterns of clay rods. (Top) Less-reacted clay rod showing intense reflections of silica minerals (quartz and cristobalite) and weak feldspar (alkali feldspar and albite) reflections. (Bottom) Highly reacted clay rod showing decreased intensities of silica minerals and increased intensity of albite (sodium feldspar).

Under the oxidizing and at temperature conditions between 700°C and 800°C, iron(III) component in earthenware reacts with hydrogen chloride and form iron(III) chloride (gas).   

3HCl+1/2 Fe 2 O 3 = FeCl 3 + 3/2 H 2 O
Iron(III) chloride attacks silver and this forms silver chloride (melt) and iron(II) chloride (melt).   
FeCl 3 + Ag = FeCl 2 + AgCl

If chlorine (gas) is formed, it also attacks silver.   

2FeCl 3 = 2FeCl 2 + Cl 2
1/2 Cl 2 + Ag = AgCl

As described above cementation processes of gold-silver parting are principally a chain of gaseous reactions. The reactions will terminate when readily reactive silica or iron are exhausted.

5.3. The Function of Earthenware and the Difference of Cementation Procedure in the Early and Late Edo Periods

The changes in chemical composition of the earthenware fit well in reactions of cementation process.

The function of the clay dish and plate is to provide some reactants, such as silica and iron. In many place in the world, brick powder is mixed with salt and bullion. In case of Sado, clay dishes, clay plates and the surface of furnace act as reactants.

Gowland16) described gold refining process in Japan in 1868 that the powdered gold was mixed with common salt and clay. Remains of early seventeenth century Sado and also picture scrolls and documents of later period, however, show no indication of use of clay mixing.

There were a long furnace and many small furnaces in the Sado refinery of the early seventeenth century. Use of clay rod also was unique. The function of the weakly reacted rod may be the grate on which dishes sat.17) The clay rod, which has highly reacted surface but less reacted inside, was used as a core in the salt-bullion cones.18)

In the early seventeenth century bullion must be upgraded to 86% Au. Repeated operation required to use of many furnaces. Clay rods would be expected to accelerate reactions. Chemical changes of clay rod, however, seem to be small. Rod was enveloped in the reactants (salt-bullion mixture) and reacting gas was hardly penetrated in solid clay rod. The inefficiency of inner clay rods was probably noticed and the use of clay rod was abandoned in later years.

6. Transfer of Salt Cementation into Japan

In 1590’s a group of craftsmen engaged in gold metal work in Sakai (a major trading port of Medieval period in Japan). The craftsmen were originally masterless samurai warriors (rōnin). Later they move to Kyoto and worked for gold refining.14) Raw materials were bullion from mines and scrap gold. At the end of sixteenth century kinza or kobanza (monopolized office for making gold coins, koban) was founded in 3 places, Edo, Kyoto and Sunpu. The craftsmen were associated with koban office in Kyoto, Sunpu and Edo and engaged in gold refining at their own houses using cementation with salt. After 1698 private gold refining was prohibited and the craftsmen worked only at the workshops in kinza.

The end of sixteenth century was very active period of interaction between Japanese and Portuguese or Spanish. Seigern (liquation process) for extracting silver from copper was learnt from Portuguese.19) The gold-silver parting technology, cementation with salt, also possibly was transferred to the craftsmen in Sakai from Portuguese or Spanish people.

7. Conclusions

Unearthed remains from gold refinery in early seventeenth century Sado was examined and it has been concluded:

(1) The compositions of discolored earthenware are characterized by the increase in soda, chlorine and silver, and the decrease in silica and iron, that show the involvement of silica and iron with the cementation reaction of bullion and common salt.

(2) The gold-silver parting proceeds principally with gaseous reactions. Iron was volatilized as iron chloride and was lost from the earthenware, such as clay dishes.

(3) In the refinery of the early seventeenth century there were many small furnaces along with a long furnace. Repeated operations probably required to use of many furnaces, as bullion must be upgraded to higher gold value at the time.

(4) Clay rods were used as cores in salt-bullion cones, which were expected to accelerate reactions. However, clay rods did not work effectively and the use was not continued afterwards.

  • 1)  Aikawa Town Board of Education: The Sado Gold Mine Remains (Sado Bugyosho Site), Aikawa Town Board of Education, Aikawa, Sado, (2001), (in Japanese).
  • 2)   W.  Lindgren: Mineral Deposits, 4th ed., McGraw-Hill, New York, (1933), 212.
  • 3)   N. H.  Fisher: Econ. Geol., 40 (1945), 561.
  • 4)   N.  Shikazono: Min. Geol., Special Issue, (1981), No. 10, 259 (in Japanese).
  • 5)   E.  Izawa: Extended Abstracts of MMIJ 2012 Spring Meeting, Mining and Materials Processing Institute of Japan, Tokyo, (2012), 182 (in Japanese).
  • 6)  Japan Mining Industry Association (Nihon Kōgyōkyōkai): Ore Deposits in Japan, Part II, Japan Mining Industry Association, Tokyo, (1968), 447 (in Japanese).
  • 7)   Anonymous: Hitori Aruki (Walking alone), (early 19th century), located in Senshu-bunko, Sado High School.
  • 8)   Anonymous: Picture Scroll of Mining and Smelting at Sado Gold-silver Mine, (early 19th century), located in Kyushu University Museum.
  • 9)   E.  Percy: Metallurgy: Silver and Gold, Part I, John Murray, London, (1880), 552.
  • 10)   A.  Kobata: Nihon Kozanshi-no Kenkyu (Study of History of Japanese Mines), Iwanami-shoten, Tokyo, (1968), 68 (in Japanese).
  • 11)   T.  Takizawa and  Y.  Nishiwaki: Kahei (Coins), Tokyo-do, Tokyo, (1999), 241 (in Japanese).
  • 12)   A.  Soetbeer: Edelmetall-Producktion und Wertverhältniss zwischen Gold und Silber seit der Entdeckeckling Amerikas bis zur Gegenwart, (1879). Cited Engin. Min. J., ed. by W. R. Ingalls, (1909), 358.
  • 13)  Dainippon Kaheishi (History of Japanese Coins), Vol. 3, Ministry of Finance, ed. by K. Yoshida, Tokyo, (1876), Reproduced ed., (1925), 364 (in Japanese).
  • 14)  Kinza during the Tokugawa Period, ed. by Tokyo City, Tokyo City, Tokyo, (1931), 123 (in Japanese).
  • 15)   T.  Craddock: King Croesus’ Gold, ed. by A. Ramage and P. Craddock, British Museum Press, UK, (2000), 175.
  • 16)   W.  Gowland: J. Soc. Chem. Ind., (1896), 404.
  • 17)   T.  Craddock: King Croesus’ Gold, ed. by A. Ramage and P. Craddock, British Museum Press, UK, (2000), 27.
  • 18)   K.  Ueda: Proc. of 6th Int. Mining History Cong. 2003, Akabira City, Akabira, (2004), 282.
  • 19)   E.  Izawa: Metallurgy and Civilisation: Eurasia and Beyond, ed. by J. Mei and T. Rehren, Archetype Publications, London, (2009), 163.
© 2014 by The Iron and Steel Institute of Japan