iron-smelting furnaces and the metallurgical traditions of the south

Iron-smelting furnaces and metallurgicaltraditions of the South African Iron Age

by H. M. FRIEDE*. Ph.D.(Visitor)

SYNOPSISA short survey is given of the Iron Age smelting sites and smelting furnaces that have been found in South Africa.An "ttempt is made to classify South African smelting furnaces by the inclusion of functional characteristics such

as the numbers and positions of tuyeres. An affinity is suggested between the Venda (eastern Transvaal) type offurnace and an East African type.

The principles of the smelting process in South African Iron Age furnaces and the characteristics of the smeltingproducts are discussed. Results of metallographic and chemical analyses of 19 iron specimens found at South Africansites are given and compared with those from sites in Zimbabwe-Rhodesia, Mocambique, and Nigeria.

The carburization and steeling of iron during primitive smelting and forging are discussed.

SAMEVATTINGDaar word 'n kort oorsig gegee oor die smeltplekke en smeltoonde uit die Ystertyd wat in Suid-Afrika gevind is.Daar word 'n poging aangewend om die Suid-Afrikaanse smeltoonde te klassifiseer deur die insluiting van funk-

sionele eienskappe SODSdie getal blaaspype en hul posisie, en 'n ooreenkoms tussen die oond van die Vendas (Oos-Transvaal) en die Oos-Afrikaanse tipe aan die hand gedoen.

Die beginsels van die smeltproses in Suid-Afrikaanse smeltoonde uit die Ystertyd en die eienskappe van diesmeltprodukte word bespreek. Die resultate van metallografiese en chemiese ontledings van 19 ystermonsters watby Suid-Afrikaanse smeltplekke gevind is, word aangegee en met die vir monsters afkomstig van smeltplekke inZimbabwe-Rhodesie, Mosambiek en Nigerie vergelyk.

Die karburisering en verstaling van yster tydens die primitiewe smelting en smeding word bespreek.

Introduction

The term South African Iron Age was first deliberatelyused by R. Mason1 in 1952 to indicate 'the period subse-quent to the introduction of iron-working but prior tothe appearance of European metal artefacts within thearea defined'. The study of the cultures and the tech-nology of this period has led to many fruitful archaeo-logical and archaeo-metallurgical investigations, especi-ally over the past ten years, when the existence of anEarly Iron Age industrial complex in South Africa wasrecognized and studied.

Early Iron Age Smelting Sites

The production and the use of iron formed an import-ant characteristic of the African Iron Age2. Unfortun-ately, not many iron artefacts and only a few iron-smelting sites from the Early Iron Age, lasting from the4th century A.D. to the 11th century A.D., have beenfound in South Africa.

The Early Iron Age smelting sites discovered aregenerally characterized by the presence of smelting slagand sometimes ::tlso of tuyere remains and pieces of ironore. The only place in South Africa where large slagfloors and furnace debris from the Early Iron Age havebeen excavated is the Broederstroom archaeologicalsite, near Hartebeespoort Dam (district Brits), whichhas been dated to the middle of the 5th century A.D.3.Besides many small pieces of iron ore and corroded iron,a large block of crude iron and some fairly well-preservediron artefacts were found at this site. A detailed de-scription of these finds will be given in a forthcomingpublication by R. Mason4. The results of chemical andmetallographic analyses of some of the metallurgical

*Archaeological Research Unit, University of the Witwatersrand,Johannesburg.

372 AUGUST 1979 JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

material from Broederstroom were reported previously5.Additional analyses of iron objects are given here inTables I and H.

There are indications that more Early Iron Age sitesare to be found in the Broederstroom-Magaliesbergarea, and it is hoped that more-substantial remainswill be found there that will allow some conclusions onthe structure of furnaces in the Southern African EarlyIron Age - a subject on which the information up to nowhas been meagre.

Furnaces from the Middle and Later Iron Age

By the 11th century A.D. iron production appears tohave spread over wide areas of South Africa - from theNatal coast to the western Transvaal, and from theLimpopo River to the Vaal River. Many fairly well-preserved iron-smelting furnaces of this period have beenfound there. They are all low-shaft furnaces, beingbellows-blown and non-slag tapping, but they varyconsiderably in structure, size, and functional detail.Two main types can be recognized: one type widelyspread in the Transvaal resembles in many respects thebeehive- or oval-shaped furnaces found in Rhodesia; asecond type, called Venda furnace, is found at manyplaces in the eastern and north-eastern Transvaal, butdoes not show clearly recognizable affinities with furn-aces in the neighbouring regions. The characteristics ofthe Venda furnace are three equally spaced, radiallyarranged tuyere slits and straight or slightly outwardssloping walls6. At one time it was thought that theVenda furnaces showed some affinity with a Madagascar!Eastern furnace type, but it seems more likely that thespecial features of the Venda type point to a relationshipwith the Central!East African intengwe (ichingtengwa)furnace type, 'small blast furnaces shaped like a churchbell put upside down'. These intengwe had three slits for

gJ

~..

"«z0eoz«u$z

'"..iI:

~~::>i

~0~'"~..~E«.......'"«Z0eo

~'"..

~...«Z«u:;:..«'"8......«.....::>i

8 ~C ....e ;n~ ~~ r:i.

"~

~....;na-I"<

~

*00t:~r:i.

~

~~15<.>0!:~

~""<.>

'".;,

§'s.C,g

§"I....

~;:)~..(

*'"00-....

;;;0"I~~

~'ii'"15<.>0!:

'"::E

"2~I!J

~'"~~

,$1~u,S

.g'i(0l!JoS

'..";;

Hu;oS

..6 ::~~ 5>" '".o,!S ..1i~ ,§

~::E ~g:E '6~5 '80'" '";:,s ]0 ~ '6'ist:i '0~g ]z~ ~~oo

'"

,0000

u":~g.9'-

,'">c

"'""...""".cc~00

«'ii

Q«

"8'".;,

8"'s.",g

I"<~~

"I....~I"<;:)

~..(

~t:8"I~~

~'ii'"80!:'"::E

~

*",00-....;;;0"Ir:i.

"~

~~

15<.>0!:'"::E

,;

";

""0,S

"t:it:~"2~;g'"6'"~~;~'EJ:~"~~'"2~'s'"~

'"....0

'"t:::

~....;;;0"I

~~'ii

'"15<.>0!:

'"::E

~;"'".."'015>

'6.:::11'"~;"'"8'""'"'":c

....0

'"t:::

*'"00-....;;;:;:r:i.

"~

~'ii'"15<.>0!:

'"::E

,~

~~

,.1!~

11.

~"~~'"i's~

~....;;;0"Icl.~

~~

15<.>0!:'"::E

,~~"'"..C

'2~'6'"C'"'"~'6'0]'"'"

*",00-t:8"I

~~

~'ii'"15<.>0!:

'"::E

~;i'§11.oSt:i~~'"~C

'~~

"I0'.;,I"<0,~

",'0r.0

'"'":I!

~8

'".;,

~ ~ ~ ~ ~ ~"', "',00 00 00 00

~" ~" ~" ~= ~" ~":::8 :::8 ~8 ~8 ~8 ~8

2t;;

6 6O;!i 02)0,- 0,-~~ ~~.8,!1 .8,!18 J:: 8 J::

&'i~ ~~

..6 6 6 t:ig,;! g,;! g,;! ~~~~ ~&'i ~~ g,~.8,!1 .8,!1 .g,!1 ~~8~ 8~ 8~ ~ti~O &'io ~o 50

'""I~I"<....;;;-I"<;:)

~..(

.. ..t:i t:i~.o ~.o

"'C ..."g~ g~~~ ~~C

"~ti ~ti50 50

"'"'":c..2'8~

a-M~I"<....

~;:)~..(

"N

'"«

'"I"<~....0-I"<;:)

~..(

!!' !!'" "~.o ~.o..." ..."g~ g~

"'" "'"~~ ~~~ti ~ti5'0 50

~'"«

I"<~;::0-I"<;:)

~..(

C~

~~;::0-I"<;:)

~..(

*00....;;;0"I

~~

~~15<.>0

!:

'"::E

,~

~"'"..C

'015>'6'""'"'""N

'6'0]'"'"

..'"t:i 15~.o'"..."'"0" '"o~ ...

'"~ ...2

~~ I>~~.; oS.;6~ £~

"~~ ~; ~'"~<I:: «

'"00~....M

'";:)~..(

~....;na-I"<cl.

"~

~'ii'"15<.>0!:

'"::E

~8~I>>0B..

,S

'ii.:!~

~....81

;:;..'s~:;

~"]

'8C,g

~;"'"'""'"'EI>11.

'"I"<s:0

;;~E;;;;:)~..(

~t:~I"<

~~

~'ii'"15<.>0!:

'"::E

~;"'".."'015>'6'"C

'"'EJ:

....c;,

0

~ ~!i] ~t

...!i !!,'ii...~ ,,~,g.; '§.;",~ o'~j;:;Q ~Q

~15C,g

'"~"I....~"I;:)~..(

'""II"<

- ...I"<

.c Ng "S.c .c<.> <.>

'" '"

C2 ~§; ,.N

~ ~~.9 ';~'"

&'i,5~ "'00 "'a-

"0 ~d~ ]:::~ ~..;

C"'~:; §~.; g

2 N-5o'~,§ d~,§J:: ,-~,,:;,~ ~"";; ,~

'" 0 8~ '"~~ -e~e §;,$1 B§;; ~~~ ~~'~ ~6,5 ~ ~2 "S2 ~ "'.9'~ i!'5::: 81>~ .!;

'"0 ~;:;" oS

8 gj, ,~§::'",S e.c" ;\O~ £'~a 8.1h I:~g '8~~ :;;,~~

"" -~" 2"'.; ~...:; 15~~ ~~~ ~

_00

00u "8

"0oo,~ ~N0- "",,,88 2~0~~ 5n a-

'"~;:; ~.;,I> q

6 S5 .ss< 6

;:'.~

.:;~"'~-c08

<!] :E" u

~ ...... ..] ]'0 '00 0~ ~

'"I"<

t

N

"S.c<.>

'"

",~

.c~

.r:.;-c08

"'..81>'-.0::El>!!,'ii]~'0 .;o'~~Q

]:au

........

!i.0'00~

'"I"<00

I"< I"<

....0

~'~....0

'"'is~":sj~

~a-~""""-<.>

8

:!; '"I"<",'I"<

~

'"<!~

~

'" '"

00, 00,OM,

'i'0

"I:*5: 2 e ~00

I!J ~ ~<P:i ~ ~ $

88 'g 11."~~ ~ g g

"" I"< '" '"""" - '" '"

~

20<.>

">.E-<

.,.;

'"I"<

P:i

~

'";::t:

~;"'".9,5.g155

'"~'6'0]~

11.

0

"'"'"

.\i",.g~

~~i

~

>,.2pa

~ 1><':~ .:;!!,

'"!J,.!!

2 ';'8,] :!!@~Po. ~"':1oS '3~~t:i a£';::g 'o5~~ 11~~C '6.~E2

"~'"~ 'iiI!J,!::

~ ~~~~ ~~g:;; 8'~;>- J':",g

'" e.c:§..co e'~ ~,<;::00

"'" '"~- .g,:::.g<!&S~lE2", "",'8~~ ~~1i2-5 ';;H;~ ~..2~<.>

'" <':.c0~ ~~,~.'8 ~- 2

~.:! ~lE"'~ ",11...2

u~

"8"I.;,

I

~I- 11"< 1"1 I'" I'" I'" \ loo la- I~ I~ I~ I::: I:!;I~ I~ I~ I~I~ \g IN I~ I~ I~ I~

~:;; ~>, ...- 0~ g« !:

'"::E

~~15

~

"::E

".. ~ i1I; ~'"E-«I:: «

"I~~

"I....

~;:)~..(

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

};~~ ~

11 215'".z. '~5 ~

,5 g'>, '0

'"~; ~

~ !J,-g g;

'"g>" 2~ ~,g, ';1~ """'"0' 6

'is .; top: 86 i'~ 8 ~'0 g'~~ ~e ~

"" '"B" 1i 6';; ;<.>" 'ii.;~ 61:.0 :s ,:::6

'"ri 11 "O~8 ~t'~ g8.~~ 06 <.> ~ ~

8 £~,;: ~J':, ~ I>'" ~E-<'" '0':;'0 ..oS,!!> I§ 0, ~ :s::11. I!J~':; B;8 ;:";;..$ .!!&'" 0

"'" "8 ~ ~-6.,~ ~"'";; 't;.g

g'"'8'0 '~'"

§ E

"~ g;';; 'a,,~..

"'2=.c2

~'E ,S'E 5 '{E11.:;; ~..2Po. Po...2

....

"'"';::if :g~2 '-"I§ 0"', ;'q<.> ",0 ,'"§ ]] ~~.c.o.o

-'"8 H ]i0..5 oa

~

C ~°a ~

!J,'o ~« 8. <.>

'" '".;,

~""<.>

'".;,

C

",~~ :&,,~ C

'"0

'" .=

'"<.>0:cC,g

I"<,'"0 -

"~

" "I

"...... -

.9 ~"

,'ii ;:Ju ~

..(

"I'"~"I....

~;:)~..(

.8]

<.>

'"

C0 C~ '8,- 0> ...

S u

if:§.

]'

"'"C"~<!~..;,!S.:!''''is......00e';;

.; B8:: g'o5 ~'e~H';:: ~3J: 8 a

u~,~~ 600-: 0""">

~

~!!'I>;g

8

'"-..$ >, ~<I::

"", 8"'..$ ,- C~ ~§I> ~2d ~Jj] ~ ~

~.. ~t:i 0.g ,iJ :!?2 '" "~~ ~ ;~~ <! ::;:;;

"~o ::E ::E

8"'s.'3~ 8 8

:J;: ::t:

I"<;;'";:)~..(

>,.2

".:1'"

N"S.c<.>

'"

"'>,".0j",g.gCC,,"

.cO':::~~

'""c~el'gb~';::'0!~§~

,!:: C>,'~

11

80N

B"0;:;..

;.::1

gI"<

&S,

"""&'i

~:::

'"

-5..;~ ;~C ::t:

'"~ S~::t: =~11 ~::t:

""','s,.,.;'0

g;

'o~ ~~e &s :E2

"1:1 H81;! "".c", 0'",g>; ~~0:J;:"'6~~ h

81 gB~ 0 <.>,S

~-e <P:i~ B ~11~] 'e~~c ~~

>,C~,g0...::E~~§"'~~'"ga-,"'~g"'6"g,., 6""c"60!J,o.oe~

;,,~'i~ ..~~~.9

'"0

~<.>

'";::t:

~'"~"~11

u~;j Q

8 « a8~Q ~" "Ic;« ~ 11 .;,

~~.0,sN

""g ,,]'os]' ~~; El ~ g "',!!~ ~ ~::E g>I>,- 0

'";,!!!,

z::E ::E~ E-<Z

§'s.C,g

]:au

I"<

iz

AUGUST 1979

~

'Q'-gM"':.:COO;:;-~~g:;"""..."'_"I

oil~.0~"".c0.......;C;B"~~

u~a8"I.;,

.9....0~'6~'s;:J>;..~'ii~....0da~'""Q~....0:J15~~~

~

373

SECTION N-S

Fig. I-Reconstruction of iron-smelting furnace 7/63 from Melville Koppies Nature Reserve, Johannesburg

WL Excavated working level BC Bellows compressed BE Bellows extended P PlatformSL Slag level CW Clay walls

Fig. 2-lron-smelting furnace 7/63 from Melville Koppies Nature Reserve, Johannesburg

374 AUGUST 1979 JOURNAl- OF THE SOl,JTH AFRICAN INSTITUTE OF MINING AND METALLURGY

tuyeres, which were adjusted to slant inwards after thefurnace floor had been raised with ash. Such furnaces7were used by the Vipas, a tribe living between LakesTanganyika and Rukwa (Tanzania).

A possible third type is the M elville Koppies typ3(Figs. I and 2). Only one example of this type is know11:it was excavated on a hill in the Johannesburg Cityarea8. This furnace, the lower part of which is sunkinto the ground, could be a developed bowl type, whichis a primitive type of smelting furnace. However, beingdated to the 11th century A.D., the Melville Koppiesfurnace could belong to an intermediate type thatcombines the features of the other types found in theTransvaal, or it could merely be a local variety of one ofthese types.

A typology based only on structural features may notbe the most suitable for smelting furnaces, and theclassification could perhaps be functional so that itwould also show technological development. The mainclasses of induced-draught furnaces and forced-draughtfurnaces are, of course, based on functional features.Similarly, the Southern African forced-draught(bloomery) furnaces could be subdivided according tothe position and the numbers of their tuyeres. A charac-teristic of many Rhodesian and some South Africanfurnaces is the position of the tuyeres in the furnacewall opposite a large front hole that serves for firing anddischarge. In these furnaces, the hottest zone of thefurnace would be near the back wall, the heat falling offtowards the front portion.

If the ends of the tuyeres in a two- tuyere furnacewere placed to face one another with a gap betweenthem, the efficient heat and combustion zone would beround the centre of the furnace and spreading upwardsin a cylindrical or conical zone. Should the tuyeres at the

same time be directed slightly downwards and set offat a slight angle towards the wall instead of towards thefurnace centre, then a turbulent circular flow of hot aircould be created round the furnace interior with obviousadvantages in the air and heat regime9 (Fibs. I and 2).

Another efficient arrangement is the three-tuyeresystem of the Venda furnace (Fig. 3), in which eachtuyere blows against the interstice between the othertwo tuyeres. Here there would be an enlarged centralcombustion zone and, through internal preheating of thetuyeres, the heat economy would be better.

It would be interesting to make both a theoretical andan experimental study of the influence of the varioustuyere arrangements, and also of the various furnaceshapes and sizes and so obtain a better understandingof the efficiency of the smelting procedures practisedduring the South African Iron Age.

Smeltin~ Sla~s

Slag floors and slag heaps are the most commonremains found on old smelting sites, since the iron sili-cates of slag are fairly resistant to weathering and tochemical attack by soil acids. The analysis of slags cangive an indication of the ore sources used in iron smelt-ing, and also some information on the smelting processitself. However, such analyses must be interpreted withcaution since they are influenced by sampling procedures,variations in ore and fuel material, use of scrap iron andflux, and similar factors. In a previous papers, the authorreported the analyses of 31 slag samples from smeltingsites in Southern Africa. The majority of the samplesshowed a high iron content (40 to 50 per cent Fe), butin 11 samples the iron content was low (4 to 8 per cent).The slags of high iron content, which are typical ofwell-conducted bloomery-furnace smelts, consist prob-

Fig. 3-Venda furnace near Iron Mountain, Low Country, northern Transvaal. Photo taken by H. Cros in1888. Reproduction by courtesy of Africana Museum, Public Library, Johannesburg

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY AUGUST 1979 375

Site Catalogue Silicon Total Metallic Manga- Alumi- Phos- Sulphur Analyst Notesno. iron iron nese nium phorus

% % % % % % %Broederstroom 24/73 KA 0,61 83,7 ::1::20 - 0,19 0,016 - Iscor Fragment,

Research semi-quantitativeLaboratory analysis

Broederstroom 24/73 Azc 3,8 58,2 ::1::10 - 0,44 0,067 - Iscor Fragment,Research semi-quantitativeLaboratory analysis

Broederstroom 24/73 Aj - 92,6 88,1 - - - - National Iron blockLl63 Institute

forMetallurgy

Rooiberg B 0,17 - - Nil - Traces 0,01 Schoch Soo Table I,no. 14

Rooiberg I - - - - - 0,01 0,014 Schulz Soo Table I,no. 15

Rooiberg II - - - 0,01 - 0,03 0,01 Schulz See Table I,no. 16

Niamara I - - - 0,01 - 0,32 0,01 Schulz Ref. 24

Niamara II - - - TracesI

- 0,2 - Schulz See Table I,no. 21

ably mainly of the compound fayalite (containing 57per cent Fe). One of the likely reasons for the productionof low-iron slag is the use of a siliceous low-grade ironore, as it is sometimes found in banded iron ores (forexample, in the Magaliesberg area).

Slag deposits can also give an indication of the ironoutput of smelting furnaces. Slag heaps containing 30 to50 tons of slag have been found near smelting furnacesin the Phalaborwa area1O, 11. Impressively large slagheaps have also been reported from a Botswana smeltingsite near Tautswel2. It appears that iron smelting was asizable industry in the Later Iron Age.

Carburized (Steeled) Iron

Much information on the traditions and the develop-ment of iron smelting and working can be obtained fromchemical analyses and metallographic examinations ofiron artefacts found at archaeological sites. Such methodswere already used some fifty years ago by M. Baumann13(1919) and G. H. Stanley14 (1931). The results of thisearly, as well as of more recent, research work are re-ported in Tables I and II.

The 'steely' character of Iron Age iron objects isevident from the analyses of these 25 samples. (The termsteel should be reserved for modern homogeneous high-purity carbon steels and should not be used for thecarburized iron of the African Iron Age.) Two hypothesesof the procedures to obtain steely (high-carbon) iron inAfrican Iron Age smelting furnaces can be presented.

Hypothesis.. Intentional CarburizationThe first hypothesis involves a high-temperature

process producing high-carbon iron by intentionalcarburization in a smelting furnace. This process re-quires sustained high temperatures (above 1300°0).This could be achieved either in high furnaces of naturaldraught as were used in West and Oentral Africa, or infurnaces with many tuyeres like the forced-draught

'blast' furnaces of East Mrica2, 7,11,15,16, Thereare indications that natural-draught furnaces were usedin Rhodesia17, but no remains of such furnaces have yetbeen found in South Africa. High-carbon iron (of thecast-iron type) could probably have been produced insmall quantities in bloomery furnaces under favourableconditions (high temperature, suitable ratios of ore tofuel, prolonged firing times, efficient bellows, carefularrangement of tuyeres, good insulation of the furnace,etc.). Under such conditions, temperatures of more than1300°0 could have been obtained - notes on 'whiteheat' in bloomery furnaces are found in the reports ofearly travellers. Maximum temperatures of 1300°0 havealso been recorded in smelting experiments recentlyconducted at the Archaeological Research Unit of theUniversity of the Witwatersrand, but it appears to bepractically impossible to control the reduction processin small bloomery furnaces at such high temperatures.Also, the tuyeres and furnace walls will not stand up tohigh temperatures. In any case, high-carbon iron (of thecast-iron type) would be too brittle for the making ofimplements and weapons. It would also be very difficultto reduce the high carbon content to a satisfactory levelby oxidation-de carburization and long hammering.

Hypothesis: Unintentional Carburization

The sponge-iron (bloom) produced in bellows-blownlow-shaft (bloomery) furnaces is generally a mixture ofsmall solid-iron particles, slag, and unburnt charcoal,with occasional pieces of unreduced ore. Depending onthe conditions of the smelting process, either the ironproduced could have had a low carbon content(bloomery or wrought iron), or it could have containedsome high-carbon iron resulting from accidentalcarburization. Occasional local overheating in the pre-sence of unburnt charcoal and consequent absorption ofthe carbon by reduced iron probably happened fairly

CHEMICAL ANALYSES OF IRON SPECIMENS

TABLE II


frequently in bloomery furnaces. As a result of thisreaction, areas of high-carbon content (above 0,8 percent) can be found in many iron specimens from IronAge smelting sites. The heterogeneous structure of thesteely iron appears to be - as G. H. Stanley14 hadpointed out in 1931 - characteristic of the products ofSouthern African Iron Age bloomery furnaces. This isalso evident from the results of the metallographicanalyses reported in Table 1. (See also A3, A4, B of theAddendum.)

The influence of the forging (smithying) procedures O~lthe carburization of crude iron is discussed later in thispaper.

Examination of Iron Artefacts

Of the materials listed in Table I, specimens 1 to 13are from three CentraljWestern Transvaal sites: Broeder-stroom, Olifantspoort, and Platberg, and specimens 14to 19 are from other parts of the Transvaal and theOrange Free State as recorded by previous authors inthis field. Included for purposes of comparison is relevant

material from the literature on the Zimbabwe cultures:Zimbabwe, Niamara, and Manyikene (specimens 20to 22) and t~e Nok culture (specimens 23 to 25).

The Nok culture of Nigeria (Taruga) belongs to ageographic and ethnographic area having no directconnections with the Southern African Iron Age com-plex. However, apart from the Broederstroom site, itis the only African Early Iron Age area from whichdetailed information on the furnace structure and themetallurgical character of the iron products is avail-able18, 19. In both respects, there are remarkable simi-larities between the Western African and the SouthernAfrican technologies. Whether such similarities point toany affinities beyond that of a common source for theAfrican Iron Age metal technology can be decided onlywhen much more cultural, archaeological, and metal-lurgical information becomes available.

A number of iron artefacts found by R. Mason at theIron Age sites of Broederstroom, Olifantspoort, andPlatberg (see Table I, nos. 1 to 13) were submitted toProfessor C. E. Mavrocordatos (Department of Metal-

Fig. 4-lron Age artefactsTop: Chisel from Rooiberg MinesBottom (L. to R.): Blade (Marikana, Rustenburg dist.), adze (Marikana, Rustenburg dist.), scraper (Mapungu-bwe, Soutpansberg dist.), barbed arrowhead (Mapungubwe, Soutpansberg dist.), pin(Olifantspoort, Rustenburgdist.)


Fig. 5-Photomicrograph of section of iron bit Ll62 fromBroederstroom. Junction of an inner (Iow carbon) and anouter band (high carbon). Scale X170. By courtesy of C. E.

Mavrocordatos

Fig. 6-Cross-section cut from bottom of iron block Ll63from Broederstroom site. Black areas are cavities producedby the irregular shapes of insufficiently consolidated nuggets.

Scale X2. By courtesy of C. E. Mavrocordatos


lurgy, University of the Witwatersrand) for metallo-graphic examination. A summary of his results is givenin the Addendum. (The designations of the specimenscorrespond to those given in Table I.)

Distribution of Carbon in Iron SpecimensThe irregular distribution of low-carbon areas, high-

carbon areas, and entrapped slag is a typical feature ofAfrican bloomery iron. The carbon-rich areas are theresult of natural carburization by occluded charcoalparticles in the furnace fire and by carbon trapped in thepores of the bloom when it is hammered out in the forge.Additional carburization takes place when carbon isabsorbed and diffused into the reduced iron from the hotcharcoal bed of the forge. The resulting heterogeneousstructure of the iron is observed in nearly all the speci-mens analysed (Table I).

Zonal Structure in Iron Age ArtefactsZonal structures are often found in Iron Age artefacts.

A laminated structure is present in a very early bladefrom Taruga (Nigeria), which is approximately dated tothe 4th century B.C. In that specimen (no. 25J), threeseparate pieces of metal are joined together. Similarly,in an iron piece from Broederstroom (no. 1), an innerlow-carbon band is sandwiched between two high-carbon outer bands. In a specimen from Rooiberg (no.16), carbon-rich zones alternate with low-carbon zones.A report28 on a hoe found at Magwareng (no. 19) describesit as a 'complex structure of six welds and a lap jointwhere thin sheets of strips of metal were joined together'.

In the specimens mentioned, the carbon content ishigher in the outer layers, but in other specimens thereverse is the case, the carbon content being higher inthe centre zone (nos. 15, 20, 21). (See also AI, A2, Binthe Addendum.)

One could assume that the laminated structures arethe result of deliberate carburization (cementation) ofthin iron strips in the hot charcoal bed of the forge.However, doubts have been expressed as to whetherbloomery iron could be intentionally carburized to high-carbon steely iron by the African blacksmith in a directforging process. R. F. Tylecote16 comments: 'The variablecarbon content (of Iron Age artefacts) has led many totalk about intentional carburization. But the fact is thatreheating can remove or add carbon according to thecarbon level of the area of the bloom heated and theprecise position of the metal in the hearth in relation tothe tuyere'.

If R. F. Tylecote's objection is accepted, one has to lookfor a more empirical explanation of the carburizationprocess. It could be that the blacksmiths observed differ-ences in the appearance, hardness, and strength in ironpieces forged at different places in the hearth and recog-nized the influence of varying temperatures and heatingtimes. The pronounced effect of these factors has beenshown in a recent study by Maddin2°, who found that,after nine hours' firing at 1150°C, the concentration ofcarbon at 1,5 mm below the surface of an iron piececan reach 2 per cent. The corresponding figures at 920 °Care 0,02 per cent and 0,48 per cent carbon after onehour's firing and after nine hours' firing respectively. Soboth low-carbon and high-carbon iron could be obtained.

Much more analytical, metallographic, and experi-

mental work is required before the problems of carburiza-tion in the metal produced by the early African black-smiths are completely understood.

Welding Methods~here are indications that, at the beginning of the

AfrICan Iron Age, a rather sophisticated method ofwelding was already known. A pile of crude-iron pieceswas wrapped up in a fire clay envelope and heated to ahigh temperature in a forge fire. The clay envelopecontaining the iron was then taken out of the fire andwas broken up by hammering, the hot iron being weldedtogether. In this way, excessive oxidation of the crudeiron during a long heating period in the forge could beavoided1s, 21. This appears to have been a method bywhich a very clean iron could be obtained. A cylindricaliron block, 960 g in mass, found at the Early Iron Agesite of Broederstroom (5th century A.D.), as well asmetal pieces from the even earlier site of Taruga (5thcentury to 3rd century E.C.), may have been producedin this way5. IS. A sample taken from the Broederstroomblock had a total iron content of 92,6 per cent and ametallic iron content of 88 per cent (Table 11).

Quenching MethodsAnother field of Iron Age metallurgy of which we know

little is the treatment of hot forged iron by quenchingand tempering. Some metallurgists think that suchsophisticated procedures were unknown to the Africanblacksmith. However, at least one report22 on iron-working by the Nyiha smiths of south-western Tanzaniamentions that 'the hot beaten iron was plunged intowater. . . to harden it'. The presence of martensite in aspecimen from Rooiberg (Table I, no. 16) and of Wid-manstiitten structures in a number of specimens (Table I,nos. 3, 4, 6, 24, 25) could also point to accelerated coolingafter the metals had been heated to elevated tempera-tures, but relatively low outside air temperatures couldbe responsible for the effect.

Chemical AnalysesThe few iron specimens that were analysed chemically

give only limited information (Table 11).Some iron pieces from the Early Iron Age site of

Broederstroom contain fairly high amounts of impurities,but this is not necessarily a characteristic of iron pro-duced during the African Iron Age, since metal fromTaruga (Nigeria) dated to between the 5th and the 3rdcentury B.C. shows an extraordinary degree of purity16.Products from the Later Iron Age, e.g. from Rooiberg,also contain very little impurities. In the samples fromNiamara the phosphorus content is high. Such differ-ences can be explained as being the result of the varyingcompositions of the iron ores used for smelting.

Some implements made in the later phases of theSouth African Iron Age appear to have been high-quality products. M. Baumann13 remarks that iron chiselsfound in the ancient Rooiberg tin mines were of 'excep-tional purity - equal in quality to the best Swedishiron'. Similarly, Sir Theophilus Shepstone23, an adminis-trator of Natal, wrote in a report of 1853 that ironproduced by African blacksmiths was 'very good andpossibly superior to Swedish iron'.

Conclusions(1) Furnaces of the South African Iron Age can be

classified into two or three structural (possibly alsofunctional) types belonging to one functional mainclass, the forced-draught, low-shaft bloomeryfurnace.

(2) There is no firm evidence that iron ore was smeltedin South African bloomery furnaces at sustainedhigh temperatures (over 1250°C) for the deliberateproduction of high-carbon (steeled) iron. Uninten-tentional accidental carburization of reduced ironin smelting furnace and forge, and irregular distribu-tion of high- and low-carbon iron in bloom, crude,and forged iron, are inherent characteristics of theSouth African bloomery process.

(3) Steeled iron could also be made by the heating,beating, and pile-welding of selected iron strips ofdissimilar carbon contents.

(4) From archaeological investigations and frommetallographic and chemical analyses of iron piecesand artefacts found at the 5th century A.D. Broeder-stroom site, it can be concluded that all the basicelements of Iron Age metal-working technology(blo~ng and firing procedures, forging and pilingtechnIques) were known to the earliest iron workersof South Africa. Our present knowledge is insuffi-cient to indicate whether the techniques of ironproduction in South Africa changed relatively littleduring the Iron Age, or whether improvements infurnace construction and production methods(influenced by contact with East African andEuropean cultures) led to a real technological pro-gress in the later stages of the Iron Age. This is afield for further investigation.

AcknowledgementsI am very grateful to Professor R. J. Mason (Director

of.the Archaeological Research Unit, University of theW1twatersrand) for the use of the facilities and collec-tions of his unit, and to Professor C. E. Mavrocordatos(Department of Metallurgy, University of the Witwaters-rand), who generously allowed me to use the results ofhis research work on Iron Age artefacts.

Sincere thanks are also due to the following peoplewho helped in various ways: Mr H. Stoch and Mr T. W.Steele (National Institute for Metallurgy), Dr F. E.Malherbe (Iscor), Dr H. Klein (Frobenius Institute,Frankfurt am Main), Mrs B. Gillooly, and Mrs S. Hughes.

References1. MASON, R. South African Iron Age pottery from the

southern Transvaal. S. Afr. Archaeol. Bull., vol.7. 1952. p.70.2. PHILLIPSON, D. W. The later prehistory of Eastern and

Southern Africa. London, Heinemann, 1977. pp. 120-123,147,148,176.

3. MASON, R. Background to the Transvaal Iron Age-new discoveries at Olifantspocrt and Broederstroom. J. S.Afr. Inst. Min. Metall., vol. 74. 1974. pp. 211-216.

4. MASON,R. Personal communication, 1978.5. FRIEDE, H. Iron Age m~tal working in the Magaliesberg

ar~a. J. S. Af~. I'f!'8t:¥ir:' .Metall., vol. 77.1977. pp. 224-232.6. KUS~L, U. PrImItIve Iron smelting in the Transvaal.

Stud~es by the National Cultural History Museum, Pretoria,no. 3. 1974.

7. WEMBAH-RASHID,J. A. R. Iron working in Ufipa. NationalMuseum of Tanzania, Dar-es-Salaam, 1973. pp. 20-34.

8. MASON, R. Prehistoric man in Johannesburg. Dept. of


Archaeology, University of the Witwatersrand, Johannes-burg. Occasional paper no. 6. 1971. p. 35.

9. FRIEDE, H. and STEEL, R. Notts on Iron Age coppersmelting technology in the Transvaal. J. S. Afr. Inst. Min.Metall., vol. 76. 1975. p. 229.

10. MASON, R. The origin of South African Society. S. Afr.J. Sci., vol. 61. 1965. p. 262.

11. VAN DER MERWE, N. J. The carbon-14 dating of iron.Chicago, University of Chicago Press, 1969. pp. 103, 113.

12. LEPIONKA, L. A preliminary account of archaeologicalinvestigations at Tautswe. Botswana Notes and Records,vol. 3. 1971. p. 24.

13. BAUMANN, M. Ancient tin mines of the Tra:J.svaal. J.Ghem. Metall. Min. Soc. S. Afr., vol. 19. 1919. p. 122.

14. STANLEY, G. H. Some products of native iron smelting.S. Afr. J. Sci., vol. 28. 1931. pp. 131-134.

15. THOMAS, H. B., and SOOTT, R. Uganda, London, OxfordUniversity Press. 1935. illustr. facing p. 220.

16. TYLEOOTE, R. F. History of metallurgy. London, MetalsSociety, 1976. pp. 43, 47, 53, 55.

17. PRENDERGAST, M. D. A new furnace type from theDarwendale Dam Basin. J. Prehist. Soc. Rhodesia, vol. 7,no. 14. 1975. pp. 16-19.

18. TYLEOOTE, R. F. Iron Sm<:Jlting at Ta.rI15a, Nigal'ia. J.Hist. Metall. Soc., Univeraity of Newcagtle-upo:1-Tyne,1975. pp. 49-56.

19. TYLEOOTE, R. F. The origin of iron smelting in Africa.W. Afr. J. Archaeol., vo!. 5. 1975. pp. 1-9.

20. MADDIN, R. et al. How the Iron Age began. Sc. American,vol. 237. 1977. pp. 124, 125.

21. TYLEOOTE,R. F. Iron smelting in pre-industrial communi-ties. J. Iron Steel Inst., vol. 203. 1965. p. 346.

22. BROOK, B. P. Ironworking amongst the Nyiha of southwestern Tanganyika. S. Afr. Archaeol. Bull., vol. 20. 1965.pp. 98, 99.

23. SHEPSTONE, Sir T. Letter. Natal Mercury, 29th Dec.,1853.

24. SOHULZ, E. H. Zusammensetzung und Aufbau einigerMetallfunde der Afrika-Expedition von Leo Frobenius1928/1930. Paideuma, Frobenius Institute, Frankfurt amMain, vol. 5. 1950 pp. 59-65.

25. FROBENIUS, L. Erythraea. Berlin, 1931. pp. 285-293.26. MASON, R. Prehistory of the Transvaal. Johannesburg,

Witwatersrand University Press, 1962. p. 462.27. DAVIDSON, A. C. On the material of an ancient native

iron hoe. S. Afr. J. Sci., vol. 31. 1934. pp. 509-510.28. MAGGS,T. M. O'C. Iron Age communities of the southern

Highveld. Occ. Publications of the Natal Museum, no. 2.1976. p. 112.

29. WIELD, D. V. Report of Archaeology Section of G.E.A.,I.I.G.M. o~ large loop specimen. Manyikeni Zimbabwe,S. MocambIque, 1977.

Addendum: Metal1o~raphic Examinationby C. E. MAVROCORDATOS*

Several items found at excavations in Olifantspoort,Broederstroom, and other sites were submitted tometallographic examination. The results of this investiga-tion are summarized below.

A. General FeaturesI. All the specimens were laminated, generally in a

longitudinal direction. Occasionally signs of 'up-setting' of previously laminated blocks were0bserved.

2. The laminations, as a rule, were separated by oxideor slag layers. The carbon content varied, usuallyabruptly, from lamination to lamination. Onseveral occasions where no oxide or slag springerseparated the laminations, diffusion of carbon fromthe high-carbon to the low-carbon lamination hadtaken place.

3. Several implements showed signs of 'carburization'or decarburization on the surface, and that not

*Department of Metallurgy, University of the Witwatersrand,Johannesburg.

:380 .AUGUST 1979 JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

4.

necessarily on both sides of the examined crosS-section. This rather suggests that such apparentcarburization was accidental rather than deliberate.All the specimens examined, both on this occasionand on another (see report 295/76), were in someform of annealed state. No bainitic and, certainly,no martensitic structures were observed.This strongly suggests that neither a deliberate noran accidental attempt at hardening was made, whichin turn suggests that the ability of carbon steels toharden was not known or appreciated by the nativeartisans. This again would tend to discredit theclaim that steel was produced deliberately. Thatthe items can be described as 'steels' does notnecessarily suggest the deliberate production ofsteel. This statement implies that, in the opinion ofthe writer, no conscious and/or deliberate control ofthe carbon content of the finished articles wasexercised.On the other hand, no article (except for the 'ingot'L163, which was discussed elsewhere) showed any'cast-iron' structure, and no implement was foundconsisting of almost 'pure iron' in its entirety. Onlyoccasionally, some laminates - or a single nuggetin the ingot LI63 - consisted of ferrite alone.

All the specimens consisted of pearlitic structures.These varied from very large grains of coarse tofine pearlite, to small grains of fine pearlite.Spheroidized and divorcing pearlite was observedin at least 5 implements (LB3, L139, L151, L86,and L2). This structure is produced by the reheatingto about 700°C of a previously fully annealedarticle, so as to consist of pearlite with or withoutproeutectoid ferrite or proeutectoid cementite(depending on the carbon content of the laminates);the temperature of reheating would normally nothave been below 650°C (very long times of heatingare necessary at that temperature) but wouldcertainly not have been above 723 °c. Steel articleswith such a structure are, of course, in their softestcondition.In the opinion of the writer such reheatings wereentirely fortuitous since very accurate temperaturecontrol could not have been employed. If the fulleffect of temperature on the properties of steel hadbeen appreciated, attempts at quenching, ratherthan at fully tempering, would have been made.On the other hand, there is no evidence of articleshaving been reheated to temperatures between723 °c and 900°C. But this, of course, is due to thefact that no article (except one) was examined allalong its length. For instance, on the fully finishedknife blade (item LI39), from which two specimenswere cut, the shaft consisted of 'divorcing' pearlite,a structure characteristic of a steel heated to atemperature below 723 °c, while the tip of the bladeshowed a structure of fine pearlite, indicating aheating temperature of about 900°C. If furtherspecimens were cut at intermediate positions, it iscertain that structures of undissolved ferrite andpearlite would have been found, proving that thearticle had been heated differentially.

5.

6.

7.

B. High Carbon ContentIt is perhaps very significant that, with the exception

of one implement in which only a patch of a hyper-eutectoid structure was observed (LBI), the only speci-men with a structure close to that of cast iron was thesection of the 'ingot' (LI63). In both these specimens, thegrain boundary cementite (LI63) and the Widman-stiitten cementite (LI63 and LBI) were associated withferrite, suggesting 'abnormality' (see below). The co-existence of grain-boundary and Widmanstiitten cement-ite reflects a comparatively fast rate of cooling coupledwith the absence of alloying elements. Both these ob-jects were rather massive, resulting in an appropriaterate of cooling to produce such structures by merelycooling in air.

The fact that the only really high-carbon (1,5 to 2 percent) materials were the 'nuggets' from which the'ingots' were made, and the fact that no more very high-carbon structures were observed on the laminates of theimplements (with the exception of the unfinishedarticle labelled LBI), indicate that the reduction of theiron ore in low-shaft blast furnaces produced sufficientcarburization for the nugget to perhaps reach cast-ironconsistency when it came into contact with solid carbonin the 'hearth'. When the nugget was subsequentlyreheated for the production of the laminated article,the carbon became oxidized, leaving the various articleswith greatly varying carbon contents, ranging fromalmost pure iron to a eutectoid composition (0,8 percent carbon). The fact that an almost pure iron nuggetwas found in the section of the ingot examined, and thatit was entirely surrounded by nuggets of almost cast-iron composition, suggests that the degree of carburiza-tion in the low-shaft blast furnace employed for thereduction was not the same on every occasion.

C. AbnormalityBoth specimens showing primary cementite (LI63 and

LBI) are typical of an 'abnormal' steel. This fact ismentioned here as a 'curiosity'. It would be interesting

to hear from others who have investigated such speci-mens whether they have observed the same phenomenon.

D. ConclusionsThe examination of specimens of ancient iron imple-

ments found in South Africa (some 13 articles) leads tothe following conclusions.

1. It would appear that the blacksmiths produced allthe implements by consolidating small 'nuggets' ofiron reduced from ore.

2. These 'consolidated' nuggets were then reheatedand hot-worked until eventually the final articlewas produced.

3. During this subsequent reheating, any excess carbonwas oxidized, leaving lamellae with various amountsof carbon.

4. In the majority of cases, but not always, the outersurfaces were poorer in carbon. In some cases, thereverse was true and then not necessarily on bothsides.

5. No evidence of deliberate 'carburization' wasfound. In one implement in which both outersurfaces were of a eutectoid composition, the transi-tion from a high-carbon surface to a low-carboncore was more or less sharp and in places it wasinterrupted by elongated oxide and slug stringers,indicating that the surfaces of high-carbon contentwere merely caused by the accidental joining of twohigh-carbon 'nuggets' to one or more lower-carbonnuggets, which were placed in the middle.

6. All the articles were in the 'soft' condition, consist-ing of pearlite or of ferrite and pearlite. Some werein a sub critical annealed condition.

7. No attempt at 'quenching for hardening' wasdetected.

8. There are indications that the iron produced by the'blast furnace' is high in carbon, but that it isdecarburized to below eutectoid composition duringthe hot-forming operations.

Welding workshops and exhibitionThe Welding Institute is to exhibit at the International

Welding, Cutting and Metal Fabrication Exhibition,which is being held in Birmingham from 24th to 28thSeptember, 1979.

During the period of the Exhibition, the Institute willarrange the following programme of six afternoon work-shop sessions to study topical aspects of welding tech-nology.

Tuesday, 25th September:-Thermal sprayed coatings for improved product

performance and greater cost-effectiveness in mech-anical components

-The application of robots in welding

Wednesday, 26th September:-Exploiting advanced welding processes-High-energy density processes-Safety in welding and the law

Thursday, 27th September:-Exploiting advanced welding processes-Friction welding and diffusion bonding-Automatic operation and control of welding in

heavy engineering.This arrangement will allow the busy visitor an oppor-tunity to tour the exhibition in the morning and attendone of the workshops in the afternoon.

Further details are obtainable from The WeldingInstitute, Abington Hall, Cambridga CBI 6AL, England


iron-smelting furnaces and the metallurgical traditions of the south

Documents