Gold and Silver Refining at Sardis

Crawford H. Greenewalt, Jr., with an addition by Nicholas Cahill


Gold and silver occur together in a natural alloy, which often is recovered in the form of tiny grains and nuggets from alluvial deposits in riverbeds and streambeds. The alloy is commonly called electrum; and electrum recovered from alluvial deposits is called secondary or placer gold. Secondary or placer gold occurs in the region of Sardis, most famously in the Pactolus stream that flowed through the ancient city, but also in the beds of other streams on the north side of Mt. Tmolus (modern Boz Dağ).1

Coinage and Its Impact on Metal Refining

The invention of coinage in the seventh century BC and the use of electrum for the earliest coins (see Kroll, “The Coins of Sardis”) may have stimulated efforts to manipulate the natural alloy by increasing its silver content and by separating its gold and silver. Evidence for such manipulation includes electrum coins of the late-seventh and early-sixth centuries BC that contain a higher percentage of silver than is normal in secondary gold, thereby implying that the silver content had been artificially increased, and the appearance shortly before the middle of the sixth century BC (during the rule of Croesus) of coins issued separately in pure gold and pure silver. Ramage and Craddock have argued that the cementation-parting process was directly related to the emergence of coinage as an accepted medium of exchange, that the need for coins of consistent purity—difficult to achieve with alluvial gold because of its wide-ranging silver content—prompted development of a process that would separate silver and gold throughout alluvial gold, i.e. in the core, as well as at the surface. (A related process to enhance the surface of alluvial gold and make its surface a deeper yellow color by removal of surface silver had long been known).

The Refining Process at Sardis

Recovering gold and silver from electrum was accomplished at Sardis during the first half of the sixth century BC in simple installations located outside the city walls and close to the Pactolus Stream (Fig. 1).2 Two metallurgical techniques are directly attested by the archaeological remains: cementation, which was used to separate gold from silver and other metals present in secondary or placer gold; and cupellation, which was used to recover silver in metallic form after the cementation process, and perhaps also before cementation, to remove base metals (like copper) that also occur, together with gold and silver, in secondary or placer gold.

Cementation was conducted at sector PN in small brick ovens. In preparation for the parting process, larger pieces of electrum were hammered into thin sheets to create broad surfaces for attack by a parting agent, which was probably salt (possibly also sulfides, including alum). The sheets and small grains of alluvial gold were then sandwiched between layers of salt, perhaps combined with “carriers” of clay or brick dust, inside coarse pottery containers. The containers resembled—and may have included—ordinary Lydian cooking pots (Nos. 61, 63, Fig. 2). Filled containers were placed in ovens and heated at relatively low temperatures of about 800° C for many hours, perhaps even days. During the heating, salt vapors attacked the electrum and converted its silver to silver chloride, which was absorbed by nearby clay materials, including “carrier” clay or brick dust, the pottery containers, and the furnace bricks. After treatment, the gold was basically pure.

After cementation, the gold was tested, or “assayed,” to determine its purity, then collected and delivered to some other part of the city for consolidation in large amounts (the absence of crucibles at sector PN indicates that consolidation was done elsewhere). To test for purity, the gold pieces would have been melted on fragments of coarse vessels, like those used for cementation (many such fragments, bearing traces of melted gold, were recovered in excavation at sector PN), and rubbed on a touchstone, like No. 16, found in a Lydian house; the streak made by rubbing would then be compared to the streak made by a gold needle of known purity.3

The silver removed from the electrum during cementation was valuable, and to recover it, the clays of the parting vessels, salt, brick and clay dust (“carriers”), and furnace bricks that had absorbed silver salts during cementation were first smelted with lead. Two fragments that may belong to a smelting furnace were recovered in excavation at sector PN. Smelting transformed the silver salts to metallic silver and combined it with the lead. The silver and lead were then separated through cupellation. The “cupels” (better called “bowl hearths,” since their form differs from those described by literary sources), were porous-walled, saucer-shaped hollows in the ground surface (Fig. 3). The silver-lead alloy was placed in the bottom of the hollows, covered with fuel, and heated in an oxidizing atmosphere up to about 1100° C, by means of bellows. The combination of heat and oxygen oxidizes base metals; it transformed the lead into lead oxide, or litharge, and isolated the silver, often in concentrations of button-like form (Figs. 4, 5), which could easily be removed. As many as 200 of the “bowl hearths” were exposed in excavation, as well as the tip of an iron blow-pipe and bellows nozzles designed to withstand intense heat (tuyeres, e.g. No. 15).4

The furnaces, fragments of parting vessels (some containing gold drops), fragments of gold sheet, and other associated materials are the earliest known evidence for the cementation process anywhere.

  • Fig. 1

    Reconstruction of the gold refinery at sector PN at Sardis, showing production of foils (lower left), cementation (top), and cupellation (bottom) (©Archaeological Exploration of Sardis/President and Fellows of Harvard College)

  • Fig. 2

    Lydian cooking pot set up in an ancient cementation oven at sector PN (©Archaeological Exploration of Sardis/President and Fellows of Harvard College)

  • Fig. 3

    Cupels for recovering metallic silver from lead at sector PN (©Archaeological Exploration of Sardis/President and Fellows of Harvard College)

  • Fig. 4

    Litharge “cake,” the byproduct of cupellation, showing the hollow in its top surface where a “button” of silver has collected and been removed (©Archaeological Exploration of Sardis/President and Fellows of Harvard College)

  • Fig. 5

    Litharge and Foils from Sector PN. (Photograph by Crawford H. Greenewalt, jr.)

Lydian Gold and Electrum: Final Questions

The Pactolus stream was famous in antiquity for its gold; Herodotus (1.93) describes it as one of the two things worth seeing in Lydia (Fig. 6). This must have provided the Lydians with tremendous wealth, allowing them to monumentalize their city with colossal fortifications and terraces, and to build their military power to conquer western Anatolia.

It has long been assumed and stated as fact that the source of electrum for Lydian electrum coins and for the refinery at Pactolus North was alluvial gold with a high silver content, found naturally in the Pactolus and other streams in the vicinity of Sardis.5 Ancient authors often do not distinguish between gold and electrum, or gold and “white gold” as the Greeks sometimes called this material; when Herodotus describes the gold flakes in the Pactolus, he could mean either pure gold or electrum. Modern scholars have believed that this abundance of electrum from the Pactolus explains why the Lydians first minted coins in electrum, rather than pure silver and gold.

Recent analysis of natural gold from these streams and other sources around Sardis, however, reveals that the ore contains virtually no silver, but is essentially pure gold.6 A number of questions spring to mind. First, if the Lydians had pure gold, why did they mint electrum coins, and where did that electrum come from? and second, if the gold from the stream was pure, what were the Lydians refining at this industrial complex?

From the earliest years of the Mermnad dynasty the Lydians had a strong interest in the region around Troy. Gyges “rich in gold” controlled “the whole of the Troad” where famous gold mines such as those at Astyra and Cremaste were located.7 Gold mines between Pergamon and Atarneus were being worked in the reign of Croesus, who as a youth served as a governor nearby at Adramytteion.8 Although Herodotus focuses his account on the campaigns of the Lydian kings against the Ionians, Lydian interest in Mysia and the Troad began as early as or earlier than their campaigns to the west, and may have been of greater strategic importance. And the gold from this region does contain significant quantities of silver.9 We might hypothesize, therefore, that the electrum used for Lydian electrum coins was not derived from the Pactolus but from Lydian mines in the Troad; the pure gold from the Pactolus may have been used for other kinds of artifacts or as bullion, since it was pure.

If the installation at Pactolus North was not refining natural electrum from the Pactolus, what was it refining? We may recognize this as a fairly small-scale atelier surrounded by apparently domestic complexes, rather than a large-scale or state-controlled operation; and it was located not in the center of the Lydian city (as was believed when the sector was excavated), but in a suburban quarter (see Cahill, “City of Sardis”). It is possible that rather than refining natural ore, this was a facility for separating now-obsolete electrum coins and other objects into pure gold and silver.10 (NDC)

  • Fig. 6

    Alluvial gold from near Sardis (©Archaeological Exploration of Sardis/President and Fellows of Harvard College)


  • 1An early source for gold dust (ψῆγμα [χρυσοῦ]) in the Pactolus stream is Herodotus (1.93, 5.101); other ancient Greek and Roman sources are collected in Pedley 1972, 70–72 nos. 230, 242–255. The ancient word “electrum” is used by some modern scholars only for man-made artifacts (not for the natural product; Cowell and Hyne in Ramage and Craddock 2000, 168, 174 n. 1).
  • 2The refining installations are located at excavation sector PN (Pactolus North). They were excavated in 1968 and 1969 under the supervision of A. Ramage (Cornell University); results are published in Ramage and Craddock 2000.
  • 3Touchstones were commonly black and fine grain to set off the gold streaks, and in antiquity were often called “Lydian stones” because of their association with Lydian gold (Theophrastus, De Lapidibus 1.4; 7.46–47). Touchstones are discussed in Moore and Oddy 1985, and Ramage and Craddock 2000: 247–248.
  • 4Platinum Group Element (PGE) inclusions, which, in the form of iridium-osmium-ruthenium alloys, often occur in alluvial gold and regularly appear in Lydian coins of both electrum and refined gold, are absent in the gold foils recovered at sector PN. The inclusions could have been removed in the initial cupellation of alluvial gold and electrum, and the refined gold may have been destined for items other than coins. The subject is discussed in Ramage and Craddock 2000, 238–244, and Craddock et al. 2005.
  • 5 Head 1887, xxxiv; Ramage and Craddock 2000, 17-23, and many other discussions.
  • 6Cahill et al. Forthcoming.
  • 7Gyges: Archilochus fr. 15; Abydos and Astyra: Strabo 13.1.22, 14.5.28. Cremaste: Xen. Hell. 4.8.37.
  • 8Lydian Mines near Pergamon and Atarneus: Strabo 14.5.28, [Aristot] De Mirab. Ausc. 52. Adramytteion: Strabo 13.1.65; Croesus at Adramytteion: Nicolaus of Damascus, FGrHist 90 F65; Pedley 1972, no. 64; cf. Strabo 12.8.3, 13.1.8.
  • 9Cahill et al. Forthcoming.
  • 10Cahill et al. Forthcoming.