Copper: From the Mountain to the Ingot

Copper was smelted on Cyprus for some three thousand years, and a model that mixed those centuries would show a process that never existed. Two excavated places let this one keep to a single date. Charcoal from the furnace walls and slag at Phorades gives radiocarbon dates in the 16th and 15th centuries BC, and the pottery agrees (Knapp et al. 2001; Knapp 2012). At Enkomi, near the east coast, a building that Porphyrios Dikaios excavated in Area III had copper worked in its western rooms in Level IB, about 1550 to 1450 BC (Crewe 2007, App. 3; Kassianidou 2012).

The two places did different jobs. Phorades smelted ore to matte and nothing more; matte was made into metal and cast in the towns (Kassianidou 2008; Knapp 2012), and Enkomi’s hearths for heating crucibles are that town end. The ore came from Kokkinorotsos, “red rock”, a ridge of rust-coloured gossan about half a kilometre north-west of the smelting place and visible from it (Kassianidou 2013a, p. 39).

What happened to the copper next, the ships, the letters to Pharaoh and the gods standing on ingots, is told in Under a Kilogram; the Iron Age kingdoms that mined the same hills are in A Kingdom for Fifty Talents.

Cypriot copper lies in the pillow lavas of the Troodos, in lenses of massive sulphide capped by gossan, the crust of iron oxides left where rainwater leached the copper out and carried it down. Gossans are red, yellow and ochre, several metres to 40 m thick (Constantinou 2012; Kassianidou 2013a, p. 36), and they show from a distance, which is the best guess at how Bronze Age prospectors found ore; how they actually searched is not recorded.

Below the gossan a leached zone, up to 30 m at Apliki, holds 0.06 to over 1% Cu. Under it the copper carried down from above has been redeposited in the enrichment zone, where bornite and chalcocite give ore of up to 20% Cu; deeper still lies the primary ore, pyrite with chalcopyrite, poorer in copper (Constantinou 2007, p. 342; Antivachis 2015). The one lump of Phorades matte, a copper sulphide (Knapp 2012), shows that the smelters worked sulphide ore, most likely from the enrichment zone.

No section through Kokkinorotsos is published, so the zones are borrowed from the Apliki deposit, where drilling has measured a column of about 130 m in a body 100 to 150 m wide (Antivachis 2015). The model is a block of hillside 30 by 20 m with the zones squeezed in, the enrichment zone about 12 times thinner than at Apliki, so that the sequence fits under one adit. The plate shows the true scale.

A popular story has the island named after its copper, but the metal takes its name from the island. Latin aes Cyprium, “Cypriot bronze”, was shortened to cyprium and then cuprum; the Greek for the metal was chalkos, and where the name Kypros comes from is unknown (Knapp 2008, pp. 325–326; Kassianidou 2013a, p. 35).

One Bronze Age mine on Cyprus is known from its contents. At Ambelikou, near the north-west coast, adits driven in the 1940s broke into ancient galleries holding Middle Cypriot pottery of about the 19th century BC, stone axes and perforated stone hammers. No ancient timber was found, so the galleries were probably unpropped (Webb & Frankel 2013, p. 113). No Late Bronze Age mine is known. At Apliki, beside the only known miners’ village (Kassianidou 2018), “no evidence of Late Bronze Age exploitation was recorded in the mine itself” (Kassianidou 2013a, p. 41), and every dated timber from a Cypriot mine is of the 9th century BC or later. The adits on Kokkinorotsos are undated.

So the adit in the model is a reconstruction by analogy, and all its sizes are our estimate: 1 m wide and 1 m high, running 10.5 m into the hill at 15° to a chamber 2 by 1.6 m and 1.2 m high in the enrichment zone. Ambelikou gives the form, without dimensions; the Iron Age one-person adit at Agia Varvara-Almyras, 2 m high and under 2 m deep, gives the scale (Kassianidou 1998). Digging it removes about 17 m³ of rock, which makes about 25 m³ of spoil.

The tools are stone ring-hammers, axes, pounders and anvils, as at Ambelikou and at Apliki, where some pounders are stained with iron oxide (Kassianidou 2007). There is no Cypriot evidence for fire-setting, and freshly broken sulphide takes up oxygen fast enough to foul the air of a gallery (Bruce 1948, cited by Kassianidou 1998). No Bronze Age lamp has come from a mine, so the model shows none. The Almyras adit flooded after winter rain, and mining was probably work for summer and autumn (Kassianidou 1998).

At the mouth the ore was broken and sorted by hand, and our model keeps one part in 2 of what is dug. At Mavrovouni a gossan outcrop of almost an acre turned out to be an undated dump of ore “cobbed off and rejected before smelting” (Bruce 1948, p. 10, cited by Kassianidou 2013a, p. 36). The timbered gallery of the photographs, with pine props, lamp niches, ladders and iron picks, belongs to the Iron Age and Roman mines (Bruce 1937) and appears here only as a ghost.

Sulphide ore is usually roasted before it is smelted: heated in air and kept below about 700 to 800 °C, so that some of the sulphur burns off and some of the iron turns to oxide, ready to form slag (Rose et al. 2020, p. 80). Phorades had no crushing tools and no roasting layers, and the excavators put both jobs on the eroded part of the site or “somewhere else perhaps near the mine” (Kassianidou 1999).

Every roasting installation known on Cyprus is later: a clay-lined rock-cut pit of the Late Hellenistic period at Almyras, roasting waste of the 10th to 8th centuries BC on the Agrokipia spoil heaps, an Archaic roasting site on Kokkinorotsos itself. Andreas Hauptmann notes that roasting debris can hardly be told from natural gossan, so a Bronze Age roasting place could lie unrecognised on any red hillside (Hauptmann 2011, p. 227). Whether the smelters roasted in heaps, in pits, or partly inside the furnace is open.

The model offers one possibility as an option: a heap near the mine on a bed 2 by 1 m, the size used when an Alpine team roasted 165 kg of chalcopyrite over 43 hours with 400 kg of wood (Rose et al. 2020, pp. 82–83). At that rate one ingot’s ore would take about 1,150 kg of wood, which the tree count below leaves out. The furnace model assumes a roasted charge.

The excavation, by the University of Glasgow and the University of Cyprus under Bernard Knapp, with Michael Donnelly in the field and Vasiliki Kassianidou on the metallurgy, found no wall at all. The smelters had built a bank of river cobbles into the channel of the ancient creek, levelled it with gravel into flat working surfaces, and piled their slag against the creek’s natural bank. North of the heap, fragments of tuyères, the clay nozzles that carried air into a furnace, lay thick over about 30 m² around a stone-lined pit, perhaps the pit the furnaces were tapped into (Knapp et al. 2001, p. 204; Kassianidou 2008).

Ploughing had cut down the working surfaces, so no furnace was found in place. Tiny snail shells in the cobble bank suggest short seasons of work, probably in the slack months of the farming year, and a spring nearby gave water “even in the hottest months” (Knapp et al. 2001, pp. 204, 206). The charcoal is almost all Pinus brutia, the pine of the Troodos foothills (Kassianidou 2008). One find has no known use: a conical diabase boulder, foreign to the spot, set on hard clay with a little bone at its foot (Kassianidou 1999).

The final publication has not appeared, only interim accounts (Knapp & Kassianidou 2008; Kassianidou 2022), so the layout is ours: the bank on the east, the creek bed on the west, and two furnace emplacements on the platform, one in use and one being relined. The excavators write of furnaces in the plural (Kassianidou 2011, p. 43). The plan keeps the one published relation: the pit lies north of the heap.

More than six thousand pieces of furnace came out of Phorades, of coarse clay full of chaff with bits of gossan worked in. They come from free-standing cylinders built on the ground: a thick clay disk first, then the wall raised on it in slabs, with fingerprints where the clay was pressed down round the foot. Walls are 2.5 to 4.5 cm thick, bases 4 to 7.5 cm, the outside diameter 42 to 44 cm, and the height “at least 30 cm”, since no fragment keeps a full profile (Hein et al. 2007; Knapp et al. 2001, p. 206; Kassianidou 1999).

No wall fragment has a hole for a tuyère, so the tuyères went in through the open top (Knapp et al. 2001, p. 207). The clay set the limit of the fire: Hein, Kilikoglou and Kassianidou found that “a temperature of 1200 °C must have been already critical” for it (Hein et al. 2007). Some fragments carry layer on layer of new clay where a burnt wall was relined.

Our furnace is 43 cm across outside with walls 3.5 cm thick, leaving a shaft 36 cm wide, on a base 6 cm thick. Its height of 55 cm is our estimate, a shaft about one and a half times its width, as in experimental furnaces of this kind; the plate draws the published 30 cm in solid line and the rest dashed. The slag cakes prove a tap, so the furnace has a clay-stopped arch at its foot, 10 by 8 cm, opening onto a pit 50 cm across and 14 cm deep: inferred from the cakes, sized by us. No published drawing of a Phorades furnace is known to us; ours follows the excavators’ words.

Two other furnaces often stand in for it. Furnace 8 at Agia Varvara-Almyras, the only Cypriot furnace rebuilt to full height, is conical, about 85 cm high, with one tuyère through the side and a clay door broken out at each run (Fasnacht, in Peege 2018, pp. 271–280). It dates from about a thousand years after Phorades. The side-blown furnace that R. F. Tylecote drew for Kition and Enkomi in the early 1980s is rejected, because the Phorades walls have no holes (Kassianidou 2012; 2016).

Phorades yielded some fifty nearly whole tuyères and about six hundred fragments (Knapp 2012): straight clay pipes 5 to 7 cm across with a bore of 2.5 to 3.5 cm, the best about 30 cm long and still incomplete, and one bent at an elbow (Hein et al. 2007; Kassianidou 2008). A few were used, then wrapped in fresh clay and used again, a habit known only from Phorades. Ours are 30 cm long and 6 cm across with a 2.5 cm bore. Two go into each furnace, resting on the rim and pointing down at about 76°; the count and the angle are our estimate.

Inside some bores, well back from the tip, are slag droplets sucked in on the back-stroke of a bellows (Kassianidou 2011). Which kind of bellows is less clear. The excavators first assumed leather bags. Kassianidou later identified four sherds from the site as probably from clay pot bellows with a bridged nozzle, a form found only on Cyprus, while warning “If I am correct” (Kassianidou 2011, pp. 43–44). Her complete comparison, Π.Γ.096 in the Pancyprian Gymnasium Museum in Nicosia, has no known findspot. It is 45 cm across and 29.5 cm high with a nozzle 5.1 cm wide outside, and she dates it early on grounds she calls “admittedly circumstantial” (Kassianidou 2011, pp. 42–45). Christopher Davey notes that pot bellows are generally absent from smelting sites and suit long runs of melting in towns (Davey 2021).

In the tomb of Rekhmire at Thebes, painted about 1450 BC, men stand with one foot on each of a pair of pot bellows and lift the skins with cords. The model works its pot bellows that way, the probable reading, with bag bellows one click away. The air is our model: each pot pushes about 12 litres a stroke, 50 per cent of it reaching the tuyère, so 2 pairs at 30 strokes a minute deliver about 720 L/min of air and burn charcoal at 10.8 kg/h.

Our furnace is lit and left on its own draught for 1.5 h, climbing towards 900 °C, the heat at which the Alpine team started its bellows (Rose et al. 2020, p. 83). Then the bellows start, and every 12 minutes a charge goes in on top, roasted ore with 2 kg of charcoal to each kilogram: 24 charges, 25 kg of ore. The hot zone settles at about 1,200 °C. Matte melts from about 1,050 °C and the iron-rich slag runs from about 1,180 °C. Matte droplets sink through the lighter slag, and the slag eats the wall, 1.5 cm of it in this run, so that about 5.4 kg of the slag is the furnace’s own clay. After 5.1 h of blowing the plug is broken.

Every Phorades cake is plano-concave, flat on top and hollowed beneath, with river pebbles stuck in its edge. Kassianidou reads the shape as the whole content of the furnace run into a pit, where the matte settled to the bottom and the slag cooled slowly on top of it, taking its hollow from the matte (Kassianidou 2008; 2012). Our pit gives a cake of 20 kg over a button of 2.74 kg of matte at 65% copper, which holds 1.8 kg of copper; 0.44 kg more stays in the slag. The real slag, iron silicate and iron oxide with fine prills of matte caught in it, holds 2.7 per cent copper oxide and was never crushed to recover it (Knapp et al. 1999; Kassianidou 2008; 2012).

The heat depends on the air supply. With one pair of bellows at 30 strokes a minute the hot zone stalls near 950 °C, the slag never runs, and the charge sinters round the tuyère tips. With three pairs it climbs to about 1,320 °C, and the wall burns through within hours. The relined walls and re-wrapped tuyères of Phorades show furnaces worked close to the limit of their clay.

The commonest picture of a Bronze Age smelt has bright copper pouring from a furnace into a mould. At Phorades what ran from the tap was slag and matte, and the matte had at least two more fires to go through before it was copper. The lump that survives is silver-blue, 73.5 per cent copper, 2.6 per cent iron and 23.9 per cent sulphur (Knapp 2012); our model assumes a leaner 65%. Experiments that follow the whole route get grams: the Alpine team took 12.3 kg of roasted matte through the next smelt and got 46 g of copper (Rose et al. 2020, pp. 86–87). How the Bronze Age smelters did so much better is not known. Our figures are tuned to land inside the published ranges.

The heap’s 3.5 tonnes (Knapp et al. 2001, p. 204; Kassianidou 2011, p. 43) also appear in print as “three tons”, with the smelt “repeated at least 150 times” (Kassianidou 2012). At 20 kg a cake, 3.5 tonnes is about 175 cakes. Run 175 times, our furnace leaves 3.6 t of slag and 310 kg of copper in matte, about 9 ingots. The excavators reckoned 650 lb, “the equivalent of ten oxhide ingots” (Knapp et al. 2001, pp. 207–208), or about 350 kg (Kassianidou 2008). All are minimums, since the creek has carried part of the heap away. No Bronze Age slag heap much larger is known on the island.

With two furnaces taking turns, one smelt a day, which is our assumption, 175 smelts are 3–6 seasons of one to two months. The Uluburun ship, wrecked about 1300 BC, carried 354 oxhide ingots, about 10 tonnes of copper (Pulak 2010): about 32 Phorades on one cargo.

One 29 kg ingot takes 19 smelts, 475 kg of ore from 950 kg of rock, and 1,150 kg of charcoal. At the 7 kg of wood to 1 kg of charcoal adopted by the Troodos survey, that is 8,100 kg of pine, 21 trees of the survey’s typical size, cut from about 475 m² of good forest (Urwin 2013). The survey’s range of ratios gives 3 to 60 trees an ingot, 15 typically, for the smelt alone. A display in the Cyprus Museum gave 300 kg of charcoal for each kilogram of copper, and the tourist copper itinerary repeats it: 8,700 kg of charcoal and about 160 trees an ingot, some ten times the survey’s own ratios (Urwin 2013, pp. 333–335). The claim that smelting deforested the island sixteen times rests on about four million tonnes of slag of all periods, most of it Roman (see climate and landscape).

Counting every task from mining to casting, one ingot costs about 150 person-days in our model, anywhere from 80 to 250, a figure good only to an order of magnitude.

The building in Enkomi Area III went up early in Late Cypriot I with massive walls, and Dikaios called it the Fortress (Dikaios 1969–1971). In Level IB metalworking is “evident in almost all the rooms of the western part” (Kassianidou 2012). The model takes Room 103, the only room in the building with no outside wall. Inside it was 3.8 m east to west and at first 3.2 m north to south; while Floor X was in use the south wall moved about a metre south, giving 4.1 m. It had four doorways (Crewe 2007, App. 3).

On Floor IX a “semi-circular furnace with an external face of stones, plaster and clay” was built against the north wall, reaching nearly to the centre and full of tuyère fragments; Kassianidou gives it a diameter of 1.4 m (Crewe 2007, App. 3; Kassianidou 2012). Dikaios called every such feature a smelting furnace. Kassianidou disagrees: a smelting furnace needs a diameter of about 40 cm, these were full of ash that a smelt would have taken into its slag, and smelting sulphide ore in a closed room is unlikely. She reads them as “hearths in which crucibles, used either for refining or for melting metal were heated” (Kassianidou 2012), and so does the model. The little slag in these rooms is nodular and crusted with copper, nothing like the Phorades cakes.

The two descriptions of the hearth do not fit. A half-circle 1.4 m across projects 0.7 m from the wall, while “nearly to the centre” of a room 4.1 m deep means about 2 m. The model takes the one number: 1.4 m along the wall and 0.7 m deep. The room’s walls, 0.7 m thick and 2.4 m high, are our estimate, and it has no roof, because whether it had one is unknown. In the fill above Floor IX, in the south-west corner, lay the earliest Cypro-Minoan tablet yet found, Inv. 1885, written in the island’s undeciphered script and dropped in a metalworkers’ room (Crewe 2007, App. 3).

Between Phorades and an ingot lie at least two more fires. The matte was roasted again and smelted to black copper, an impure metal (raw copper from Iron Age Almyras is 80 to 95 per cent copper; Fasnacht, in Peege 2018, p. 270). The black copper was then melted in a crucible under a blast of air until the remaining iron and sulphur burnt off, leaving copper of about 99 per cent (Hauptmann et al. 2002). Where the matte became black copper in Late Cypriot I is not known: nothing at Phorades shows it, and at Enkomi it has not been demonstrated. In our mass balance 52 kg of matte gives about 33 kg of black copper and 29 kg of refined metal, with losses of our own estimate, since none are published.

The Level IB crucibles are “open bowls with a convex base”, often slagged inside, and none has been measured (Kassianidou 2012); ours, 18 cm across and 7 cm deep, holds about 0.57 litres. The air came down through elbow tuyères, used at Enkomi from the building’s first level: bent clay pipes with a flat base so they can be set down, the bore narrowing from 2 to 2.5 cm to about 1 cm at the nozzle, only moderately slagged by the lower heat of melting (Kassianidou 2012). No bellows of this date survive there, so the model reuses the Phorades pot, marked speculative; the tongs follow a later Enkomi pair (bronze tongs).

A 29 kg ingot is 3.3 L of copper, 6 fillings of our crucible. Ingots were not cast in one pour of freshly smelted copper, as is often believed: they are porous, with slag inclusions from remelting, and their insides show cooling rims and successive batches of metal from the same ore (Hauptmann et al. 2002; Gale & Stos-Gale 2012, p. 70).

No oxhide ingot of Late Cypriot I date has been found on Cyprus. The first fragments on the island are of the 14th century, at Enkomi and Maroni-Vournes, and most are of the 13th (Kassianidou 2013b). Only one complete ingot has come from a controlled excavation on the island, the British Museum’s dig at Enkomi in the 1890s; otherwise ingots turn up as fragments in workshops and hoards (Kassianidou 2016).

The shape of the earliest ingots is known from Crete. In the Royal Villa at Hagia Triada nineteen were found walled up in basement compartments, averaging 29.13 kg (Evans 1935, pp. 651–652); one in the Heraklion Museum, Χ721, is 45 by 39 cm. These are the Type 1 “pillow” ingots, both faces convex and the long sides curving in. Their copper is not Cypriot by its lead isotopes (Stos-Gale 2011), but four of six ingots of the same date from Zakros, unmeasured, are consistent with Cyprus (Kassianidou 2013b). So our ingot takes its shape and size from Hagia Triada and its copper from Zakros: 45 by 39 cm and 29 kg, 2.7 cm thick at the centre and 2.1 cm at the edge, the thickness worked out from the weight. A copper oxhide ingot of almost the same size and weight, 44.5 by 36.5 cm and 28.6 kg, said to come from Asia Minor, is in the Metropolitan Museum (Met 11.140.7). The later four-handled ingots from the Gelidonya wreck average 60 by 45 cm and 4 cm thick (Bass 1967, p. 53), but the types were in use together and do not date an ingot (Bass 1967; Jones 2007). The name “oxhide” is modern; the projecting corners are handles. The model offers a four-handled ingot as an option, drawn after the complete one from Enkomi in the British Museum: 70.5 by 41.5 cm and 36.92 kg, longer and narrower than the Gelidonya average.

No ingot mould is known from Cyprus. The only oxhide mould found anywhere is a block of shelly limestone from a palace courtyard at Ras Ibn Hani in Syria, of the late 13th century, burnt, cracked and spattered with copper that is Cypriot by its lead isotopes (Craddock et al. 1997; Jones 2007, pp. 240–241). Michael Jones argues that ingots were normally cast in open moulds of sand, used once (Jones 2007, pp. 240–242), and the model sets a sand bed in the floor of Room 103, marked speculative. A carrier takes the slab on one shoulder, as in Egyptian tomb paintings; a fragment of a Cypriot bronze stand shows a man carrying an ingot. What ingots weighed against is in weights and ingots; the Metropolitan Museum’s ingot weighs 28.6 kg.

The excavated core is short. At Phorades: the creek-bank platforms, the heap, the stone-lined pit, the furnace’s form, walls and base, the tuyères from the top, the cakes, the matte, the pine charcoal and the diabase cone. At Enkomi: Room 103, its floors, its hearth, the open crucibles, the elbow tuyères and the tablet. The ore zones are geology. Everything else is filled in, and each piece says how.

Some parts are inferred from what survives: the tap and pit from the cakes, the pot bellows from four sherds, the sorting from the waste. Some are our estimate, a number no source gives: the furnace’s height, the tuyères’ count and angle, the adit, the crucible, the walls of Room 103, the layout of Phorades. The furnace’s figures are our model. And some parts are borrowed, each labelled with its place and date: the hill’s zones from Apliki; the adit’s form from Middle Cypriot Ambelikou and its scale from Iron Age Almyras; the ring-hammers from Ambelikou and 13th-century Apliki; the roasting heap from an Alpine experiment of 2020; the pot bellows from an unprovenanced Cypriot piece; the tongs from the 12th-century foundry hoard at Enkomi; the ingot from Late Minoan IB Crete; the several pours from 14th-century ingots; the Almyras furnace from the Iron Age; the timbered gallery from mines six hundred years and more after Phorades. Two pictures are drawn only to be rejected: Tylecote’s side-blown furnace and the furnace that pours copper.

At Phorades there is almost nothing to see: a creek bank with a scatter of slag and tuyère fragments. Kokkinorotsos is a red ridge under pines, with adits of unknown date and Archaic slag nearby; the model’s as-found state shows both. Apliki has been swallowed by a modern open-cast pit. Enkomi lies in the north of the island, under Turkish occupation since 1974.

The big slag heaps that visitors see are later. The great black heap at Skouriotissa grew in the Late Roman centuries, its charcoal dated between the 3rd and 7th centuries AD, and the smaller red heap there is of the 1st century BC (Urwin 2013; Kassianidou 2013c). The tourist itinerary’s claim that some of its slag “was dated to 2000 BC” has no published support; the oldest dated slag heap on the island, at Kokkinorotsos, is no older than the 8th century BC. The industry that left these heaps is the subject of A Kingdom for Fifty Talents.

In the Republic of Cyprus a visitor can see the gossans and flooded pits at Mathiatis, the pit and smelting site at Almyras, the Skouriotissa pit and heaps from the road, and the Museum of Mining Heritage at Katydata, with miners’ tools, models of galleries and a replica ingot. The Cyprus Museum in Nicosia holds a complete oxhide ingot of 39.18 kg, bought on the Famagusta market in the 1930s and said to come from Enkomi (Kassianidou 2016), and the bronze god standing on an ingot from the Enkomi sanctuary; see the compendium entry on copper metallurgy.

Nobody knows where the matte became black copper, or whether the ore was roasted, or where. No Bronze Age gallery has been recorded on Cyprus, and no ingot of this date has been found on the island. Whether Phorades used pot bellows or bags rests on four sherds. The two descriptions of the Room 103 hearth do not fit each other, and nobody can say whether the room had a roof. The heap is printed as three tonnes and as three and a half, and the full report on Phorades has not appeared.

The numbers have the widest gap. Our furnace makes 1.8 kg of copper in matte a smelt because it was tuned to, and experiments that follow the same route end with grams. The furnace itself is the least certain object in the model. No Phorades furnace survives above 30 cm, and everything drawn above that line is ours.