Notes on Iron Age copper-smeltingtechnology in the Transvaal

SYNOPSIS

by H. M. FRIEDE*, Ph.D. and R. H. STEEL* (Visitors)

Earlier reports on Iron Age copper metallurgy and copper-smelting techniques in the Transvaal are surveyed,and a description and tentative classification of the copper furnaces are given. The analyses of furnace slags andsherds of glazed crucibles from archaeological sites of the Transvaal and the neighbouring Orange Free State pro.vide evidence of copper smelting there in Iron Age times. .

An investigation conducted by the Department of Archaeology of the University ofthe Witwatersrand providedinformation on the details of furnace construction and smelting processes, on the composition of the materialsused and produced, and on the making and handling of tools and accessories, such as bellows and tongs, needed byIron Age smelters.

Three models (the KaondejVenda furnace, the Rooiberg crucible furnace, and the Uitkomst furnace) were con-structed and operated as a demonstration of Iron Age copper smelting as it was practised in the Transvaal.

SAMEVATTING

Daar word 'n oorsig gegeeoor vroeere verslae oor die kopermetallurgie en kopersmelttegnieke van die Ystertyd.perk in Transvaal, asook 'n beskrywing en tentatiewe klassifikasie van die koperoonde. Die ontleding van oondslakkeen skerwe van geglasuurde kroese afkomstig van argeologiese terreine in Transvaal en die aangrensende Oranje.Vrystaat lewer bewys van kopersmeltery daar gedurende die Ystertydperk.

'n Ondersoek wat die Departement Argeologie van die Universiteit van die Witwatersrand ingestel het, hetinligting verskaf oor besonderhede van die oondkonstruksie en smeltprosesse, die samestelling van die materialewat gebruik en gelewer is, en oor die maak en hantering van gereedskap en bybehore soos blaasbalke en tangewat die smelters van die Ystertydperk nodig gehad het.

Daar is drie modelle (die KaondejVenda-oond, die Rooiberg-kroesoond en die Uitkomst.oond) gebou en bedryfas 'n demonstrasie van hoe koper gedurende die Ystertydperk in Transvaal gesmelt IS.

INTRODUCTION

It is not known wit,h any certaintywhen and how the knowledge ofmining and metal extraction foundits way into Southern Africa. Manyarchaeologists think that migratingNegro people, probably Bantu-speaking, crossed the Zambesi Riverin the early centuries of the firstmillenium A.D. These invadersbrought with them a rich culturalheritage: the practice of subsistencefarming and domestication ofanimals, the art of making pottery,and the techniques of iron andcopper metallurgy. The possessionof iron weapons and implementswas probably a decisive factor forthe fast and apparently easy penetra-tion of the 'new' people into SouthernAfrica, where only Stone Age huntersand gatherers or pastoralists hadpreviously lived. It is fitting, there-fore, that the culture brought anddeveloped by the invaders from thenorth, lasting for nearly 2000 yearsup to the present, is called 'TheMrican Iron Age'l.

IRON AGE METALLURGY INSOUTHERN AFRICA

Onc of the earliest sites south ofthe Zambesi where relics of Iron

*University of the Witwatersrand, Johan-nesburg.

Age people have been found isMabveni (approximately 50 km westof Zimbabwe). A number of copperand iron beads have been excavatedthere. Charcoal associated with thesefinds has been dated2 to the end ofthe second century A.D.

Recently, several Early Iron Agesites were found in the Transvaal,among them Silver Leaves Farm3(near Tzaneen) and Broederstroom3(near Hartbeespoort Dam). TheBroederstroom site, dated to themid-fifth century A.D. could havebeen a centre of metal productionbecause large accumulations of ironslag, tuyere fragments, and furnacedebris were found there by Mason,together with a beautiful littlecopper chain, which may be theearliest proof of copper productionin South Africa3.

As opposed to the large number ofIron Age iron-smelting sites in theTransvaal, remains of copper smelt.ing and copper working are lessfrequently found. One reason forthis may be that the clay-builtcopper furnaces had to be brokendown once the smelting process hadbeen completed, because the copperextracted could not be removed inany other way. Most of the IronAge copper artefacts preserved comefrom the two larger copper centresof South Africa: the Messina-Lim-

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

popo Valley area in the northernTransvaal, and the Phalaborwa-Gravelotte area in the EasternTransvaaI4-6.

Less well known is the pre.European copper industry of theCentral and Western Transvaal.James Chapman, who travelled inthe Transvaal in 1852, marked amountain range on one of hismaps 'Copper Mountain', and re-marked that the copper found inthis area is 'converted with greatingenuity into a very fine wirewhich is worked up into armlets' 7.This Copper Mountain of Chapman issituated near the Dwarsberg, which,according to A. Boshier8, was alsoa great copper centre in the IronAge. Rooiberg, better known forits tin mines, may also have beena place where copper was pro-duced9-11.

IRON AGE COPPER FURNACESIN THE TRANSVAAL

The published archaeologicalliterature gives detailed informationon no more than a dozen Iron Agecopper-smelting furnaces, most ofthem from the Eastern TransvaaI4-6,one or two from the Rooibergarea 9-11, and two from UitkomstCave12 (about 20 km north ofKrugersdorp) .

It appears that two, or possibly

NOVEMBER H175 221

Site: Uitkomst Uitkomst Uitkomst Uitkomst Uitkomst Harmony Messina Kolwezi,5/67 I 5/67 II 5/67 5/67 E. Tvl 4/73 N. Tvl Congo

Position: Cave Bed 3, Cave Bed 3, Cave Bed Cave Bed Cave Bed C.VI 1955slag adhering surface slag 3 3, 2 furnace PAC

to tuyeres scraped from base 1 collectedtuyere by

R. Mason

Constituents: Expressedas %

Silicon SiO2 70,24 70,06 63,1 5,10 64,66 66,35 41,95 32,10Aluminium Al2O3 9,86 10,70 0,15 15,24 12,98 10,10 7,58Ferric Iron Fe2O3 2,56 2,56 0,81 0,13 1,21 0,87Ferrous iron FeO 8,48 7,ll 7,0 7,18 9,62 6,30 18,61 4,73Magnesium MgO 1,63 1,18 5,97 1,68 1,39 1,02 5,16Calcium CaO 1,92 2,59 27,65 1,20 6,14 21,08 1,15Sodium Na20 0,18 0,53 0,00 0,52 2,01 0,9 0,10Potassium K.O 2,17 2,79 0,71 3,84 1,79 1,2 1,78Titanium TiO2 0,49 0,43 0,10 0,80 0,90 0,51Phosphorus P2O5 0,39 0,45 0,38 0,22 0,27 0,12 1,41Chromium Cr203 0,01 0,01 0,10 0,05 0,01 0,00

MnO CombinedManganese Mn30. 0,62 0,70 0,72 0,73 0,96 0,12 H2O 0,09Ignition loss 0,12 0,05 33,20 0,10 0,56 0,34Total iron Fe 8,38 7,32 5,44 6,15 8,38 5,74 14,47 4,28Total copper Cu 4,1 2,8 0,69 13,2 ],38 0,64 2,29 35,61

Nickel 111 p.p.m. 149 p.p.m.

Analyst: NIM NIM Schneider12 NIM NIM NIM Stan]ey19 NIMAA972/4 AA972/5 AB89f5 AB89/3 AA972/3 AA972/10

three, types of copper furnacescan be recognized in the Transvaal.

One type, resembling a beehiveor a low dome, is found mainly inthe Eastern Transvaal and is re-presented by four smelting furnacesin the Phalaborwa region, whichhave been described by Van del'Merwe and Scully13. These furnacesare 600 to 900 mm in height andin diameter at the base, taperingto a chimney hole at the top ofabout 300 to 450 mm in diameter.A single arched opening in the frontallows the entrance of a tuyere.

The furnaces found by Evers atHarmony (Letaba) also belong tothis type, their measurements (590by 720 mm and 480 by 420 mm)giving them a ground plan thatcould be described as 'near-circular'4.

The second type has an ovalor elliptical ground plan. The dimen-sions vary. For example, the baseof a Rooiberg furnace measuredll450 mm (major axis) to 300 mm(minor axis), and the Uitkomst Bed3 furnace measured14 330 to 250 mm.Such furnaces have either one ortwo blow holes12, 13.

It appears that the copper-smelting furnaces were often smallerversions of iron-smelting furnaces,or were used for iron smelting aswell as for copper smelting. The'circular beehive' type corresponds

to the 'Loole type' of Kiisel'sclassification, and the 'oval' typepossibly' to his 'Buispoort type'6.

In specifying various types ofsmelting furnaces (iron furnaces aswell as copper furnaces), one has,of course, always to bear in mindthat the observed differences instructural characteristics may cemore apparent than real. Theycould well just be variations andmodifications of a single basic design.No fundamental differences in func-tional characteristics, as displayed,for example, in the 'natural draftfurnaces' of Central Africa15 or inthe 'tapping furnaces' of the SinaiPeninsula16 have so far been ob.served among the smelting furnacesof South Africa.

However, a technical problem isposed by the final smelting stagenecessary for copper refining, whichrequires fire-resistant crucibles, asustained high temperature abovethe melting point of copper (aboutBOO QC) combined with efficient

reducing conditions in the furnace,and some special devices for fastand safe discharge of the liquefiedcopper into moulds. The commonmethod was apparently a quickbreaking down of part of the furnacewalls at the end of the meltingprocess to give access to the veryhot crucible at the furnace bottom,

TABLE IANALYSIS OF IRON AGE COPPER SLAGS

but this was a very cumbersome andwasteful method. It is possible thatthe Rooiberg 'stone-circle cruciblefurnace'lo was a more efficientmelting furnace. This interestingproblem of crucibles and cruciblefurnaces will be discussed later.

It is unfortunate that not a singlewell-preserved copper furnace hasyet been discovered in the Messina-Limpopo Valley area, where sOmuch evidence of copper smelting(slag, crucibles, artefacts) has beenfound, especially at the Mapun-gubwe excavation sites. Informationon the features of the furnacesused there would permit valuableconclusions to be drawn on therelationships between the ancientcopper technology of the Transvaaland that of countries to the north.

ANALYSIS OF SLAGS ANDCRUCIBLE GLAZES

A number of copper ores17, slagpieces, glazed sherds, and copperornaments18 found in the Transvaalwere analysed to give an analyticalbackground to the investigation ofIron Age copper-smelting tech-nology.

The main evidence of coppersmelting on a prehistoric sit3 is,of course, the presence of copperin various objects found inside ornext to furnace remains (slags,

222 NOVEM6ER 1975 JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

Site:

I

Vechtkop, Dist.

I

Dist. Makgwareng,Heilbron, O.F.S. Harrismith, O.F.S. Dist. Lindley

Dimensions of Int. diameter, mm 40

I

35 110Crucible Int. depth, mm 51 30 90

Thickness/wall, mm 20---25 25-33 20

Elements Indications IndicationsCopper Very strong (2,6 %) Very strong (1,8%) 0,38%Iron Very strong Very strong 3,8 % (total as Fe.O.,)Zinc Medium Very strongManganose Weak Medium MnO 0,1%Titanium Weak StrongCalcium Very strong Very strong CaD 10,1%Tin Not detected Very strong TiO.. 0,6%So'dium - - Na.O 1,7%Potassium - - K.O 6,4%Magnesium - - MgO 4,1%Aluminium - - AI.O. 7,0%

Analyst: M. Milner

I

M. Milner Dr Verbeek21The Corner House Lab. The Corner House Lab.Johannesburg. Johannesburg.

tuyeres, crucibles). Unfortunately,not much evidence of this type hasso far become available from Trans-vaal sites. Stanley19 published theanalysis of a slag sample from theIron Age workings at Messina (TableI). Copper slag has been found atPhalaborwa13 and at RooiberglO, but,as far as is known, no analysis ofsuch slags has been published inthe literature. An analysis of aslag specimen excavated at Uit-komst Cave has been given byMason12. Since enough slag materialfrom Uitkomst is available in thecollections of the Archaeology De-partment of the University of theWitwatersrand, some additionalspecimens of slag were submittedfor analysis.

Table I gives the values for threesamples of furnace slags and twoslag pieces from tuyercs. All thesesamples, which come from Bed 2and Bed 3 of Uitkomst Cave, showa relatively high copper content,indicating that the slag was pro-duced by a process intended toextract copper from orc.

Four of the Uitkomst samplesshow a reasonable agreement incomposition (iron content 5,4 to8,4 per cent, copper content 0,7 to4,1 per cent, lime content 1,2 to2,6 per cent), whereas a fifth sampleshows different values (for copper13,2 per cent, and for lime 27,6 percent). It may be possible that thisodd sample was enriched by inclu-sions of ore concentrate.

For comparison, analytical valuesfor slag samples from three other

there. Van Riet Lowe20 excavateda number of sandstone crucibles atthe Vegkop site (district Heilbron),and thought it possible that thesecrucibles were used for iron smelt-ing. However, X-ray-fluorescencescan analysis of glazed surfaceparticles from two crucibles foundin the Orange Free State (now keptin the Department of Archaeologyat the University of the Witwaters-rand) showed the presence of largeramounts of copper, tin, and zincin the glaze, pointing to the smeltingof copper or copper alloys (Table II).These findings are confirmed by theanalysis of chippings from cruciblefragments found by Maggs21 atMakgwareng site (Lindley District,O.F.S.).

IRON AGE COPPER-SMELTINGTECHNIQUES IN SOUTHERN

AFRICA

copper-smelting sites (Harmony inthe Eastern Transvaal, Messina inthe Northern Transvaal, and Kol-wezi in the Congo) are included inTable I. The two Transvaal samplesshow copper values falling into themajority range of the Uitkomstslags, whereas the Congo sampleshows a very much higher coppercontent (35,6 per cent). This isprobably the result of the use of aricher type of ore, or of differentextraction methods.

The relative uniformity of thevalues for titanium, manganese, andphosphorus in the slag samplesfrom Uitkomst may point to afairly constant ore supply. The lowvalues for nickel (149 p.p.m. and111 p.p.m.) in two samples appearto exclude the use of ore of thenickel-copper type, which wouldyield a slag of a much highernickel content.

The presence of substantialamounts of calcium, sodium, andpotassium in the slags analysedcould be regarded as evidence forthe use of flux in the smelting pro-cess. However, these elements couldfind their way into the slag fromores containing alkaline-based com-pounds such as the calcite found inthe malachite ores from Phalaborwa.Such ores can be regarded as self-fluxing, where the addition of fluxmaterial would be superfluous.

Evidence of copper smelting or,more accurately, of copper meltingin the Northern Orange Free Stateis given by the analysis of the'glazing' adhering to crucibles found

TABLE IIX-RAY-FLUORESCENCE ANALYSIS OF CRUCIBLE GLAZE ON STONE CRUCIBLES FROM ORANGE FREE STATE

There are not many reports onthe construction of copper-smeltingfurnaces and on the smelting pro-cesses used in Iron Age SouthernAfrica.

The only report to give detailedinformation on the subject is thatpublished by Chaplin15 in 1916.Chaplin, an Inspector of RhodesianMonuments, arranged a demonstra-tion of copper smelting by Kaondetribesmen at Solwezi, a place westof the main Copper Belt in Zambia.The Kaonde tribe had a traditionof copper smelting that had diedout there about 1914, whereas, inthe Northern Transvaal, the Ba-venda had stopped their metal

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY NOVEMBER 1975 22:t

working long before. Chaplin founda group of old men who had takenpart in smelting with their fathers,and still remembered and couldpractise the traditional methods.The following is an abstract fromChaplin's report15.

The kiln was prepared by digging ahole approximately 20 cm wide. Roundthe excavated hole a wall about 40 cmhigh was built from sections of broken-down ant-hills and puddled clay. Asmall hole at the base of the furnaceprovided entrance for a tuyere. Thebottom of the furnace was filled withashes, above which a charcoal layerwas placed. The charcoal was firedand brought to (red-hot) glow withthe help of an airblast from a pair ofskin-bellows connected to the tuyere.Then ore was piled on the hot charcoaland covered with more charcoal. Afterthree hours of firing when the copperwas smelted down, the furnace wallswere broken up with a long pole.The liquid (crude) copper at the furnacebottom solidified rapidly and couldbe taken out after cooling.

The copper was then refined in asecond similar furnace. A clay potfilled with ashes was placed into thefurnace which was heated as described,and when the temperature in thefurnace was high enough, pieces of thecrude copper were placed on the glowingashes in the pot. After two hours offiring the furnace was quickly brokendown, the pot taken out (probably withthe help of tongs) and the still liquin.copper poured into a mould.The Kaonde furnace, as de-

scribed by Chaplin resembles thestone-walled bowl hearth used forcopper smelting since time im-memorial, e.g., by the Chalcolithicsmelters of the Sinai16 (third mil-lenium B.C.).

There is some evidence that afurnace similar to the Kaondefurnace described by Chaplin wasused in the Messina-Limpopo Valleycopper centre. Unfortunately, VanWarmelo, whose book22 is otherwiseone of the best reports on earlymining and metal working in theTransvaal, gives little informationon the smelting process, but Stayt23had the good luck to meet an oldtribesman in Vendaland who hadbeen a copper worker in his youth.His description of building a kiln,smelting the ore in it, resmeltingthe extracted crude copper in apotsherd-crucible, and pouring itinto a mould closely resembles manydetails in Chaplin's report.

INVESTIGATION OF IRONAGE COPPER SMELTING

Since the information on IronAge copper smelting in Southern

~4 NOVEM~~R 1\175

Africa is rather scanty, it wasdecided to investigate the subjectby some experimental work thatcould give information on the fol-lowing points.1. Details of the furnace structure

(structural material, dimensions,furnace building, etc.).

2. Details of the smelting process(temperatures, bellows action,air volume, proportion of char-coal to ore, methods of chargingand discharging, time of opera-tion, etc.).

3. Chemical composition of thematerial used and produced inthe smelting process (ore, char-coal, slag, intermediate pro-ducts, refined copper)- Deter-mination of minor and traceelements in these materials.

4. Economic aspect (efficiency ofthe process).

5. Influence of 'flux' addition onthe quality of the copper.

6. Making and handling of thetools and accessories used bythe ancient smelters (bellows,tuyeres, tongs, crucibles,moulds, wire-drawing equip-ment).

7. Special problems of cruciblefurnaces.

Of necessity, the project had tobe on a small scale. It was thereforelimited to an investigation of coppersmelting in only three types offurnaces:

an experimental furnace based onthe information given in Chaplin'sreport on the Kaonde furnace15(Model A),an experimental furnace based oninformation given in Wagner'sreport on Rooiberg crucible furn-acesl0 (Model B), anda furnace representing a full-scalereplica of the Uitkomst furnaceas described by Mason12 (ModelC). This experimental furnace wasconstructed from a termite mound,following a Rhodesian tradition24.(The original Uitkomst furnaceswere probably made from clay).Funds were provided by theDepartment of Archaeology.It was necessary, for practical as

well as economic reasons, to usemodern material and equipment inthe experiments, e.g., prefabricatedrefactory units, asbestos insulatingmaterial, an air compressor. The

choice of ore used in the experi-ments was arbitrary, since only twosuitable types of ore were availablein large quantities. Of course, suchmodifications diminish the valueof the experimental work in somerespects, but one can still obtainmuch comparative information anda better general understanding ofIron Age copper technology.Material Used

In all the twenty smelting experi-ments undertaken, use was made ofan enriched malachite ore providedby the Palabora Mining Company.There were some difficulties in theprocessing of this ore, which, afterdressing, had a copper content of17,5 per cent. The conditions of thesmelting procedure were varied ina number of experiments, but ineach case only a small quantity ofreduced copper, in the form of tinyglobules and of thin irregular-shapedcopper layers coating the surface ofthe slag, were obtained. The bulk ofthe copper was present in the slagas copper oxide (black cupric oxide,as well as red cuprous oxide). Toimprove the yield, the oxide slagcake was re-smelted at about 1l00°C,with an addition of silica to neutral-ize the high alkalinity of the slag.This procedure improved the copperyield, but the yield was still low(16,4 per cent).

Fortunately, a small quantity ofhigh-grade (40 per cent) azuritemined at Nchanga Mines (Zambia)was obtained from the South MricanCopper Development Association. Amixture of equal parts of thisazurite and malachite ore fromPalabora Mining Company gavemore satisfactory results. A fairamount of copper pellets and copperlumps could be recovered from theslag (yield 24,2 per cent).

Ore provided by the Messina(Transvaal) Development Companywas also tried, but its copper contentwas too low for the purpose.

The charcoal used in the experi-ments was commercial Natal 'braai-vleis' charcoal, reportedly made frommimosa wood. The larger pieces ofcharcoal were broken up to walnutsize, and the coaldust was sifted off.

Furnace Model A (KaondeSmelting Furnace)

This furnace, like the Model B

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND MET ALLURG¥

0. 10. 20. 30.ems

Fig. I-Floor plan of smelting furnace, model A

Plate I-Experimental furnace A, with potentlometer, thermocouple, and aircompressor.

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY NOVEMBER 1975 225

furnace, was erected on a smallpiece of open ground on the campusof the University of the Witwaters-rand.

Furnace ConstructionA shallow dished hole (18 cm in

inner diameter and 3 cm deep) wasdug into the ground and coated withclay. A shaft composed of threeceramic rings (each 13 cm in height,18 cm in inner diameter, and 22 cmin outer diameter) was erected abovethe dished hole. The rings werecemented together and to the groundwith a sealing compound. A secondcylinder composed of two largerrings (each 21 cm in height, 26 cmin inner diameter, and 32 cm inouter diameter) was built round theinner furnace shaft. The annularspace betwe6n the two ceramic ringcylinders was filled with vermiculiteinsulating material. The top of thefurnace could be covered with anasbestos sheet cut into two sectionsto allow adjustment of the chimneyopening. A rectangular opening (7cm by 10 cm) was cut into the bot-tom rings for the insertion of atuyere, which was sealed in withclay. See Fig. 1 and Plate 1.

After each firing, the furnacewas rebuilt and any cracks formedwere sealed with clay.

TuyeresVarious types of tuyeres were

used, reproducing patterns of tuyeresfound at Iron Age sites. The tuyeres(approximately 30 cm in length,2,5 to 4 cm in inner diameter, and4,5 to 7 cm in outer diameter) weremade either from alumina clay orfire clay.

Air Supply by Blow BellowsThe pair of bellows used in the

experiments was patterned on theZulu bellows kept in the AfricanaMuseum (Johannesburg), the airfrom the bellows being dischargedthrough cut-off ox horns. The effect-ive air delivery from these bellowsas measured by flowmeter was 60to 90 l/min when the bellows wereworked at medium speed (at about70 double strokes). This appears to below, since the volume of each ofthe bellows, found by the mass ofwater filling it, was 5 litres. Onereason for the low efficiency maybe the poor craftsmanship of present-day African skin workers, andanother reason, the relative in-

226 NOVEMBER 1975

efficiency of the inexperienced menworking the bellows. Furthermore,the bellows used were made in themodern way, by the sewing togetherof skins that had been taken fromcarcasses by cutting them openventrally lengthwise, whereas thebellows bag of the ancient metalworkers was pulled off in one pieceby drawing the skin over the head,body, and hindquarters of the animal(goat or buck). In this way, verylittle sewing was required to obtainpractically airtight bellows.

It is possible that a pair of well-made bellows could supply 100litres of air per minute, or evenmore when the skins of largebucks (e.g., sable antelopes25) areused.

Air Supply by Compressor

The petrol-driven compressor useddelivered approximately 120 litresof air per minute. Two controlvalves were built into the deliveryline leading from the air compressorto the tuyere and, by regulation ofthese valves, fairly constant airvolumes could be obtained from aminimum of 60 l/min to a maximumof 120 l/min - roughly comparablewith the blowing capacity of oneand two pairs of blow bellowsrespectively. From the values ob-tained, it was calculated that anair volume of 0,42 l/cm2/min isrequired to operate a smeltingfurnace of 18 cm diameter at 1100 QC.

The corresponding air volume re-quired for larger furnaces may berelatively lower.

Measurement of FurnaceTemperatures

The temperatures in the furnacewere measured at regular intervalsby a potentiometer registering thevoltage potential of a thermocouple(Ni/Cl' v Ni/AI) dipped into thefurnace centre at the level of thetuyere mouth. The correspondingtemperatures were read from atable. However, after comparisonof anum bel' of instrument readingswith the observed glow of the furnacecharge, the appearance of the flamesabove the firebed, and the 'feel' ofthe heat developed, it was .notdifficult to control the smeltingprocess sufficiently well by sensoryobservation, as done by the IronAge metal workers.

Details of Smelting ProgessAfter the furnace bottom had

been covered with wood ash, woodsticks, and dry leaves, the fire waslighted. A pile of charcoal brokenup to a size of 10 to 20 mm washeaped up opposite the tuyere andthe airblast was started. When thepile showed a dark-red glow, layersof ore pieces and charcoal woreplaced on the firebed and the air-blast increased until the fire glowedcherry red (800 to 900 QC). The

charcoal was replaced as it burnt off.At the end of the firing (approxi-

mately three hours after starting),the airblast was increased againuntil the charge showed a dullorange colour (950 to 1O00QC). Thenthe blast was stopped, the furnaceallowed to cool down, the top sectionof the furnace broken off, and themixture of charcoal, ash, and copperslag at the furnace bottom takenout.

The copper-slag cake eJntainedthe smelted-out copper in the shapeof pellets of various sizes, rangingfrom tiny globules to nuggets ofapproximately 5 mm in diameter.Sometimes fairly large chunks ofcopper could be picked out, but ina number of experiments thesmelted-out copper showed up onlyin 'specks' or as a thin coat stickingto the slag and the charcoal cinder.

If the reduction conditions (carbonmonoxide atmosphere in the furnacechamber) were unsatisfactory, muchcopper oxide was formed and thesmelting had to be repeated. If thesmelt was successful, the crudecopper pieces could be picked outeasily and hammered free from slag,cinder, and other impurities.

Furnace Model B (CrucibleFurnace)

Use of CruciblesThe use of crucibles for the smelt-

ing of metals is nearly d as theart of metal production' . Egypt-ian tomb reliefs and toplb paintingsof the twenty-fifth and fifteenthcenturies B.C. show clearly howmolten copper or Qronze ~as pouredfrom crucibles into moulds26.

Evidence of Iron Ag~ cruciblesmelting in Southern Africa is givenby a number of crucibles andcrucible fragments found at archaeo-logical sites such as Zimbabwe27,

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

Inyanga28, Mapungubwe29, Rooi-berg1O, and several places in theOrange Free State2O, 21.

Bryant30 notes that, for the smelt-ing down of metal, 'a sandstonebasin, five or six inches wide,shaped like the half of an egg'was used in Zululand. Somewell-preserved sandstone cruciblesfound in the Orange Free State arekept in the collection of the Archae-ology Department of the Universityof the Witwatersrand. The effectivepouring volume of four of thesecrucibles ranges from 10 to 40 m!.This appears to be a low value, butonly a small volume was sufficientto melt copper ingots weighing upto 300 g, enough to make a numberof bangles, earrings, and other smallornaments valued by the Bantu.

Use of Crucible Furnaces

There are only a few reports onhow crucibles were heated to thehigh temperatures (1l1O to 1200°0)at which copper will liquefy sothat it can be poured into moulds.

Ohaplin15 describes how the smelt-ing furnaces of the Kaonde wereadapted to melt copper. A rathervague description of the methodsused by the Bavenda for makingcopper rods is given by Stayt23:'the crude copper [produced in thesmelting kiln] was hammered intosmall cobbles and re-smelted in apotsherd about 7 inches in diameter,which was put over the impressionin the ground so that the moltencopper could be manipulated easilyand poured out into the mouldsprepared for it'.

In their report, Wagner andGordon10 give some details oncrucible furnaces used at Blaauw-bank-Rooiberg:

[The furnaces I are built of stonesset in a rough circle which had ap-parent,Jy been excavated out of theground. Big stones were used, eachseparated from the next by smallerstones. . . a gap of four inches wasleft in the circle, possibly for insertingthe tuyere. The most perfect furnacemeasured approximately 22 inchesacross. . . the stones forming the sideswere about 8 inches deep, of whichabout half was above ground and halfbelow. On one side of this furnace therewas found in place, what is evidentlya worked stone of a peculiarly taperedshape. This stone had been shaped. . .to act as a support for a smelting potor crucible. Similarly tapered stonesmay have originally formed a completecircle.

In the Waterberg Mountains (Cen-tral Transvaal), one of the presentauthors found the remains of afurnace that had features similar tothose of the Rooiberg furnace de-scribed above.Construction of Model B Furnace

The construction of the cruciblefurnace was based on the observa-tions mentioned above. A ring oflarge stones was built round adished hole of 20 cm diameter, thespaces between the stones beingfilled with smaller stones and sealedwith casting clay. The dished baseand the furnace walls were coatedwith fire clay. Two openings atopposite sides of the furnace wallsallowed the insertion of tuyeres.Four tapering stones were placedround the centre, leaving the spacefacing the tuyeres open. The top ofthe furnace was left open, but couldbe covered with stone slabs orasbestos sheets. See Fig. 2 andPlate H.

Operation of Model B FurnaceThe crucible was filled with small

pieces of crude copper until it wasabout three-quarters full. A thin

layer of dry leaves was placedabove the copper. The fire wasstarted as described for the operationof smelting furnace A. When thecharcoal in the furnace began toglow, the prepared crucible wasplaced on the support stones setround the centre of the furnace. Thecrucible was covered with a potsherdand more charcoal was heapedround the crucible up to the cruciblerim. Air was blown in from thebellows or the compressor until theglow in the furnace showed ayellow-orange colour (about 1l00to 1200°0). Burnt-off charcoal wasreplaced continuously to keep upthe reducing atmosphere in thefurnace.

The crucible was kept at toptemperature for about half an houruntil all the copper was molten.Then the potsherd lid was removedand the liquid copper in the cruciblestirred for a short time with astick of green wood. (This pro-cedure, still employed in moderncopper refineries for the removal ofthe oxygen absorbed by the copper,was used by the Iron Age smelters of

0 11)

Fig. 2-Floor plan of crucible furnace, model B

NOVEMBER 1975 227JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

20 30

Plate II-Experimental furnace B (crucible furnace). The two gemsbok horns onthe left lead from the blow bellows to the tuyere. The crucible in the centre is set

on four supporting stones.

Africa 'to bring the coppertogether'31.)

Up to this stage, the melting ofthe copper is a relatively simpleprocess, but the removal of the veryhot crucible is a fairly difficult pro-cedure, since the taking out of thecrucible and the pouring out of theliquid copper must be done veryquickly.

Paintings and reliefs in Egyptiantombs show Dynastic foundrymengrasping a red-hot crucible withflat stones pressed against it, orgripping the smelting pot withboughs (which were probably soakedin water)26.

It is likely that the metalworkersof Africa used similar methods toprotect their hands-such as wrap-ping them in 'thick cloths' or wetpieces of skin. The use of 'tongsmade from bark' is also mentioned32.However, all such methods are rathercumbersome and dangerous. It islikely that iron tongs were usedearly in the metallurgy of theMrican smiths, who knew well howto make iron tools. Angas illustrated

228 NOVEMBER 1975

a pair of 'native forceps' in a pictureshowing a Zulu blacksmith at work33.Holden mentioned that the smiths(of the Kaffirs) 'use tongs of veryrude construction (made of iron)'34.A photo in Van Warmelo's book TheCopper Miners of M usina22 showsan 'old pair of tongs' approximately50 cm long. Based on this design,a pair or such tongs was madefrom wrought iron for use in thework described here. No difficultywas experienced in removing thehot crucible from the furnace withthese tongs, and then pouring themolten copper into moulds that hadbeen prepared from sand or intoholes dug with a stick into theground, using the often-describedtechniques of the metal workersof the Northern Transvaal22, 23.

One of the copper rods cast fromthe copper smelted and melted inthe experimental furnaces was sub-mitted to Messrs McKechnie BrothersB.A. for analysis. The rod con-tained much iron (4,4 per cent) andsome nickel (0,02 per cent), but onlysmall traces of zinc (0,008 per

cent) and less than 0,001 per cent ofeach of the following elements:tin, lead, antimony, bismuth, andarsenic. These values are similar tothose of some copper objects foundat archaeological sites in the Trans-vaalH.

Furnace Model C (UitkomstType)

In some reports from Rhodesia24and Zambia15, the use of termitemounds or of parts of such moundsfor the construction of smeltingfurnaces is mentioned. It seemedinteresting to investigate such amethod. There were suitable termitemounds at Melville Koppies NatureReserve (Johannesburg), and per-mission was obtained from the CityCouncil of Johannesburg to use oneof them.

The making of a workable smelt-ing furnace from a termite moundtakes only a few hours. A slit wascut 8 cm wide, reaching from thetop of the mound, along its majoraxis, to a depth of 36 cm into theinterior of the mound. Then, a

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

FURNACE CHAMBER

....

PLAN AT TUYEREshowing slit at liplevel between buttresses

LEVE L 2

- TUYE~f LEVEL 2

section through furnace centre

9 IP 2P 3.0 4,0

flask-shaped furnace chamber wasexcavated by scraping away soilfrom both exposed sides of the slit.The walls of this chamber werelined with clay, and the openingsat both ends of the slit were sealedwith termite soil, leaving holes forinsertion of the tuyeres. Then twoearth buttresses were built, onejoining each side of the furnace, forstrengthening the structure, andfinally two platforms were con-structed opposite the tuyere holesto accommodate the bellows workers.Details of the furnace structure areevident from Fig. 3.

The design of the furnace isbased on that of the Bed 3 furnaceexcavated at Uitkomst Cave byMasonl2, who kindly provided themeasurements of this furnace. How-ever, for structural and materialreasons, some dimensions of thefurnace model were modified slightly.

A few trial firings were undertakento prove that copper smelting waspracticable in this furnace. The

Fig. 3-Furnace model C

furnace stood up well to firing,reaching yellow-orange heat (about1200°C).

An interesting feature of theoriginal Uitkomst furnace was theposition of the tuyeres, which wereplaced at a slight angle to the leftof the longer axis of the furnace,whereas the tuyeres are usuallyaimed radially direct at the firein the centre of a smelting furnace.If this off-centre position of thetuyeres in the Uitkomst furnacewas intentional, it was a sophisti-cated design indeed, creating aturbulent, circular flow of hot airround the central fire, This tuyerearrangement was reproduced in themodel furnace C. It was observed,when the fire in the furnace waslighted and air blown in throughthe tuyeres, that the heat in thefire bed was more uniformly distri-buted and that the furnace tempera-ture rose faster than in the experi-mental runs in models A and B. Itappears also that the cellular struct-

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

'\'\,",

:-';

0 10 20 30 40.' ..

ure of the termite mound layerbetween the furnace lining and theearth buttresses insulates the furnacechamber very well, a feature thatcontributes to the higher efficiency ofthis furnace.

It was not difficult in this furnaceto reduce copper ore (malachite-azurite) to crude copper. From theslag cake produced, many well-formed copper pellets could beeasily separated. The furnace wasalso used for the refining of crudecopper, which was placed in a claycrucible into the centre of the furnacechamber. The copper in the cruciblebecame liquid after firing for onehour. Mter the clay wall at oneof the slit ends had been brokendown, the hot crucible was takenout with tongs, and the copperpoured out into a mould.

CONCLUSIONS

(I) The number of analyses ofcopper slags found at varioussmelting sites was insufficient

NOVEMBER 1975 229

Plate Ill-Experimental furnace C (working replica of copper-smelting furnacefound in Uitkomst Cave by Mason), showing the platforms for the bellows workers.The smoke rising from the fire in the furnace chamber hides the chimney opening.

to permit definite conclusionsto be drawn on the smeltingprocedures and on the proven-ance of the ores used. A furtherinvestigation, by neutron-activation analysis, on minorelements and trace elements incopper slags is being conductedat present in a combined pro-ject by the Department ofArchaeology and the NuclearPhysics Research Unit of theUniversity of the Witwaters-rand. This project may givesome of the required informa-tion.

(2) The investigation on the smelt-ing of copper in model furnaceshas facilitated interpretation ofmany aspects of prehistoriccopper production in SouthAfrica, and the quantitiesinvolved in prehistoric Trans-vaal copper production cannow be assessed moreaccurately. The information ob-tained from these experimentswill also assist in the evaluationof the material effects of pre-historic copper production.

230 NOVEMBER 1975

ACKNOWLEDGEMENTS

The authors wish to thank thefollowing, who assisted them invarious ways: Professor R. J. Mason,Mr T. M. Evers, and Miss Z. Martinof the University of the Witwaters-rand; Mr J. Stanko, Dr R. E.Robinson, and Mr H. Stoch of theNational Institute for Metallurgy;Mr E. L. Psaros and Mr I. Ogilvieof Messrs McKechnie Brothers S.A.(pty) Ltd; Mr J. J. Schoeman ofPalabora Mining Co.; the Messina(Transvaal) Development CompanyLimited; Mr P. B. Lee of the S.A.Copper Development Association;and Mr A. P. Grobelaar of CullinanRefractories Limited.

REFERENCES

1. SHINNIE, P. L., In: FAGAN, B. TheAfrican Iron Age. 1971. p. 226.

2. ROBINSON, K. R. Arnoldia, vo!. 3,no. 1. 1967.

3. MAsoN, R. J. Background to theTransvaal Iron Age. J. S. Afr. Min.Metall., vo!. 74. 1974. pp. 211-216.

4. EVERS, T. M., and VAN DEN BERG,R. P. Ancient mining in SouthernAfrica with reference to a coppermine in the Harmony Block, N.E.Transvaa!. Ibid., p. 222.

5. MoRE, C. E. Some observations onancient mining at Phalaborwa. Ibid.,pp. 227-232.

6. KUSEL, U. S. Extractive metallurgyin Iron Age South Africa. Ibid.,pp. 246-249.

7. CHAPMAN,J. Travels in the interior ofSouth Africa. London, 1868. vo!. 1.p.22.

8. BOSHIER, A. K. Mining genesis.Mining survey, Chamber of Mines ofSouth Africa.

9. BAUMANN, M. Ancient tin mines ofthe Transvaal. J. Ghem. Metall. Min.Soc. S. Afr., vo!. 19. 1919. pp. 120-131.

10. WAGNER, P. A., and GORDON, H. S.Further notes on ancient bronzesmelters in the Waterberg District,Transvaa!. S. Afr. J. Sci., vo!. 26.1929. pp. 563-574.

11. WHITE, H., and OXLEY OXLAND, G.St. J. Ancient metallurgical practicesin the Rooiberg area. J. S. Afr. Inst.Min. Metall., vo!. 74. 1974. pp.269-270.

12. MAsoN, R. J. Prehistory of the Trans-vaal. Johannesburg. 1962. pp. 387-391,462.

13. VAN DER MERWE, N. J., and SCULLY,R. T. K. The Phalaborwa story.Wld Archaeol., vo!. 3, no. 2. 1971.pp. 178-195.

14. MAsoN, R. J. Personal communica-tion.

15. CHAPLIN, J. H. Notm on traditionalsmelting in Northern Rhodesia. S.Afr. Archaeol. Bull., vo!. 16, no. 62.1961. pp. 56-57.

16. ROTHENBERG, B. Timna, valley of theBiblical copper mines. London, 1972.pp. 235-239.

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

17. FRIEDE, H. Unpublished report,Dept of Archaeology, University ofthe Witwatersrand, 1974.

18. FRIEDE, H. Notes on the compositionof pre-European copper and copperalloy artefacts from the Transvaal.J. S. Afr. Inst. Min. Metall., vol. 75.1975. pp. 185-191.

19. STANLEY, G. H. Notes on ancientcopper workings and smeltings inthe Northern Transvaal. DurhamUniv. Philos. Soc., vol. 3. 1910. pp.307 -312.

20. LowE. C. van Riet. The stone hutsof Vechtkop. J. R. Anthrop. Inst.,vol. 57.1927. pp. 217-233.

21. MAGGS, T. M. O'C. Early farmingcommunities on the Southern High-

NIM reportThe following report is available

free of charge from the NationalInstitute for Metallurgy, PrivateBag 7, Auckland Park, 2006.

Report no. 1742The determination of tin in ores,

Vacuum metallurgyThe Fifth International Sym-

posium on Vacuum Metallurgy andElectroslag Remelting Processes isto be held in Munich from 11thto 15th October, 1976.

The purpose of this internationalconference is to continue in thetradition of earlier events held since1964 in giving experts from allparts of the world the opportunityof acquainting themselves with thestate ofthe art in the field of vacuummetallurgy. Both theoretical andpractical themes are to be discussed.The field of electroslag remelting is tobe integrated into the conferencefor the first time because of itsclose technical relatiomhip withsimilar processes. Particular atten-tion is to be given to the problemsin this field because of the growing

veld. Unpublished Ph.D. thesis, Uni-versity of Cape Town, 1974. p. 127.

22. VAN WARMELO, N. J. The copperminers of Musina. Pretoria, Dept ofNative Affairs, Ethnological Publica-tion no. 8. Plate V, pp. 81-82.

23. STAYT, H. A. The Bavenda. London,1931. pp. 62-68.

24. ROBINSON, K. R. Two iron smeltingfurnaces from the Chibi NativeReserve, Southern Rhodesia. S. Afr.Archaeol. Bull., vol. 16 no 61. 1961.p.20.

25. BARBER, H. M. The perforatedstones of South Africa. J. AntMop.Inst., Feb. 1893. p. 304.

26. HODGES, H. Technology in the ancientworld. Pelican Books, 1971. pp. 65,121.

residues, and concentrates by X-ray-fluorescence spectrometry.

The method described is suitablefor the determination of tin in aconcentration range from 0,002 to60 per cent. The samples for analysisare briquetted at a pressure of

interest in the refining of importantnon-ferrous metals and improve-ment of their properties.

The lecture themes are as follows:L The physical chemistry of

vacuum metallurgy.2. The vacuum treatment of

molten steel.3. The vacuum treatment of

molten non-ferrous metals, in-cluding vacuum distillation.

4. Melting and remelting pro-cesses under vacuum (vacuuminduction furnace, VAR, EB,etc.).

5. Metallurgy and use of theplasma beam.

6. The physico-chemcial prob-lems of the ESR process.

7. The properties and specialapplications for vacuumtreated

Competition for student membersEach year the South African

Institute of Mining and Metallurgyoffers a prize (or prizes should theentries warrant it) of up to RIOO forthe best paper or dissertation on atopic appropriate to the interests of.the Institute. The competition is

open to all Student Members of

the Institute.

A Student Member who is in full-

time study at a university may

submit the dissertation or thesis he

has to write in part fulfilment of his

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

27. CATON-THOMPSON, C. G. The Zim-babwe Culture-ruins and reactions.Oxford, 1931. p. 117.

28. SUMMERS, R. Inyanga. Cambridge.1958. p. 101.

29. GARDNER, G. A. Mapungubwe. Pre-toria, 1963. vol. 2.

30. BRYANT, A. T. The Zulu people.Pietermaritzburg, 1949. p. 388.

31. THOMPSON, L. C. Ingots of nativemanufacture. Nada, vol. 26. 1949. p. 8.

32. MOFFAT, R. Missionary labours andscenes in Southern Africa. London,1842. pp. 453, 466, 467.

33. ANGAS, G. F. The Kaffirs illustrated.London, 1849. plate 23.

34. HOLDEN, W. C. The past, present andfuture of the Kaffir races. London,1866. vol. 1, p. 242.

15 t, cellulose powder being usedas a binder. Calibration curves areestablished by the use of chemicallyanalysed samples. The coefficientof variation at the concentrationlevel of 20 per cent is 1 per cent.

and remelted materials (incI.ESR).

8. The characteristics of ESRplants and operating experi-ence.

9. Jointing processes undervacuum (brazing, diffusionwelding, EB welding).

10. Heat treatment under vacuum.11. Characteristics of vacuum

plants for metallurgical pro-cesses.

12. The economy of vacuummetallurgical processes incomparison with conventionaltechniques.

Further information is availablefrom Dr.-Ing. Gerhard Frey, c/oLeybold-Heraeus, D-8000 Munich50, Lerchenstr. 5, Fed. Rep. ofGermany.

university degree, provided that it is

presented in a manner and on a

topic suitable for publication in the

journal.

Entries for 1975 should reach the

Institute by 31st December, 1975.

NOVEMBER 1975 231