UNIVERSITY OF THE WITWATERSRAND

INSTITUTE FOR ADVANCED SOCIAL RESEARCH

ADDITIONAL SEMINAR PAPER

TITLE: Transvaal and Natal Iron Age Settlement Revealed byAerial Photography and Excavation.

BY: MASON, R.J.

NO: A13

Reprinted from

AFRICAN STUDIESQuarterly Journal

Published by

WITWATERSRAND UNIVERSITY PRESS

JOHANNESBURG

TRANSVAAL AND NATAL IRON AGE SETTLEMENT REVEALED BYAERIAL PHOTOGRAPHY AND EXCAVATION

R. J. MASON*

INTRODUCTION

In 1820 John Campbell visited a thrivingIron Age settlement built by the Humtsetribe at that time under the regent Liqueling,and known as Kurrichane or Kaditshwene(spelling uncertain), believed to be near thepresent town of Zeerust in the SAW Trans-vaal, 2, 9, 14 located by J. D. Seddon on airphotos, and partially excavated by R. J.Mason in July 1966 and 1967. Campbellestimated the population at approximately16,000. Barely four years later T. L.Hodgson, then living at Matluassi MissionStation near the present town of Wolmarans-stad in the Southern Transvaal, recorded thedestruction of a town, probably Kurrichane,in intertribal warfare.5 We quote fromHodgson's letter:

1824 January 23rd—Visited by a party ofBomananya, a tribe residing five days'journey northwest of us . . . they do notknow Kurrechain, but knew Liquelling,the regent of the Marootse nation, whomthey state to have been killed by thelate invading enemy who destroyed histown . . .

In 1828 Robert Moffat, travelling to theeast of Kurrichane, described remains ofinnumerable, recently destroyed settlementssimilar to Kurrichane.' - Moffat '5 record wasthe start of Iron Age research in the Trans-vaal. 9 Most, or all, of the Iron Age structuresdiscussed in the present paper probably pre-date the tribal wars of the 1820's. Mysubject in the present paper is confined toIron Age structures. Other aspects of theIron Age are to be discussed in later papers.

Development of the Transvaal Iron AgeProject

The present Iron Age project began todevelop in 1950 when I excavated a numberof small Iron Age cave sites near Johannes-burg. In 1963 at Melville Koppie I began anintensive study of Iron Age settlementconcentrating on the lateral clearance ofsites with full record of the position andspatial relationship of artefacts as a guide tothe activity that produced them togetherwith C14 dating and ethnographic study.Since then we have developed the IronAge project with larger scale excavations andtopographic surveys at Phalaborwa, Klip-riviersberg, Matluassi,Waterval, Kaditshweneand Olifantspoort. The present paper isintended to summarize work done todate.

Throughout the work we have soughtto find behavioural evidence in terms ofmaterial artefacts and their spatial disposi-tion on sites, associated foodwaste and topo-graphic location of living sites. Pilot surveysamong living people associated with the siteshave been made in the Soutpansberg andZeerust areas.

The excavation programme was co-ordinated with settlement analysis on aerialphotographs. In 1964, I made a count ofIron Age settlements on air photos covering1,211 square miles of the Transvaal.11 MrSeddon subsequently analysed the distribu-tion of Iron Age settlements in terms of my1964 criterion, a contiguous walling system{see J. D. Seddon, p. 189 of this issue). In1967, I discarded this criterion and, with the

assistance of a team of six students, analysed*Dr R. J. Mason is Senior Lecturer, Sub-Department of Archaeology, University of the Witwatersrand.

Reprinted from AFRICAN STunits Vol. 27, No. 4 19H8

Iron Age settlements on air photos covering47,733 square miles of Southern Transvaal-Northern Natal in terms of five classesdefined on a basis of plan-form (fig. 1 andfig. 2). Iron Age settlement distribution wasplotted relative to present settlement dis-tribution (see figs. 3 and 5).

I consider that settlement plan-form is adirect function of social organization, andthat one of the keys to the understanding ofIron Age social behaviour is the plan-form of settlements. The division of eachsettlement into areas for industrial or socialactivity is directly reflected in the settlement

plans depicted in fig. 2, rather than a hypo-thetical division of major and minor areas.We now have counts of settlements in eachclass in different areas over the region2G°-3PE. and 25-27 3S.

Air-photo Analysis of Iron Age SettlementThe basis of classification emerged from

study of range of variation in plan-forms onthe photographs. We found that a total of6,237 structures which we attributed to theIron Age in the area, could be reduced tovariantsclasses:

within the following five main

"ION ACE SETTLEMENTS IN

VAAl AND LIMPOPO BASINS

Fig. 1 Air-photo analysis.

Class 4b Structure Flight 369 No.7684. Lobaisi Dist., Botswana.

Class 1 Structure Flipht 3S0 No. 8162.Zeerust Dist.. W. Tvl.

Class 2 Structure Flight 406 No.6611. Koster Dist., W. Tvl.

Class 4a Structure Flight 369. No.3225. Zeerust. W. Tvl.

Class 3 Structure AOC 12364 2:9:49Kliprivicrsberg, S. Tvl.

1 ig. 2. Air photos of Structure Classes for Transvaal and Natal Iron Age settlement revealedby aerial photography and excavation by R. J. Mason.

TRANSVAAL AND NATAL IRON AGE SETTLEMENT

Definition of Transvaal Iron AgeStructure Classes, see Fig. 2

Class 1 Isolated circle or a few adjacent butisolated circles.

Class 2a Adjacent circles forming large cir-cular enclosing wall. Blank spacein middle.

Class 26 Two or more adjacent large circlescomposed of interlocking smallcircles.

Class 3 Clear enclosing wall and innercircles.

Class 4a No enclosing wall; scattered circlesforming a closely related whole.

Class 4b No enclosing wall; closely relatedcircular structures covering a largearea.

Class 4c A dispersed set of isolated circlesseeming to form a related whole.

Class 5 Scattered irregular walls—no defin-ite plan identified.

Owing to difficulty of definition, no specificcount of slope terracing could be made.

The new experimental classification there-fore proceeds from Class 1, as the smallestand simplest class or structure, to Class 4, asthe largest and most complex. Classes 2 and4 are subdivided in terms of the relative sizeof area occupied by the structures.

The classification of stone structures issubject to the same causes of error as otherarchaeological artefact classification. Weattempted to reduce error in the presentclassification by combining the differentperceptual capacities of seven differentworkers. Discussion between these sevenworkers, as the analysis proceeded, combinedto yield the following count:

Distribution of Iron Age settlements in Western, Southern and Eastern Transvaal, and Northern Natal.47,733 sq. miles 25°-27° S. and 26°-31°E.

COUNTS ARE ESTIMATES

Area Drainage

1. MaricoHexCrocodileElands

2. SandPienaarElands

3. Steelpoort . .SabiCrocodileKomati

4. Skoonspruit

5. KlipdrifMooi

6. Suikerbos

7. Eastern VaalWaterval

8. Olifants . .

9. Klip River . .Natal

Scattered Areas..

Class 1

338

24

392

10

163

212

381

184

477

58

2239

Class 2a

116

5

14

2

6

8

117

27

251

4

550

ONLY SUBJECT

Class 2b

2

—

—

—

2

—

—

—

4

Class 3

558

92

1335

36

324

525

66

5

46

21

3308

TO ERROR NOTED IN TEXT

Class 4a

110

4

44

27

74

49

16

4

49

5

382

Class 4b

10

—

10

—

—

1

—

—

—

21

Class 4c

2

—

7

—

—

—

—

—

—

1

10

Class 5

18

—

3

—

—

2

—

—

—

23 1

Total

1154

125

1792

88

567

794

585

220

823

89

otal 6237

Total structures show the following occur-rence :Most common Class 3—clear enclosing wall

and inner circles.Least common Class 26—two or more ad-

jacent large circles composedof interlocking small circles.

Classification of Occupation Areas within oradjacent to Structure Classes 1-5

Field studies of Iron Age structures andoccupation sites suggest the following clas-sification of occupation areas within struc-tures. The classification is tentative andremains to be confirmed or modified in futurework.

Cave settlements either domestic (e.g.Gatsrand Caves, Uitkomst Cave surface orBed 4), or industrial caves used as smeltingcentres, e.g. cave at confluence of Blaaubankand Jukskei R., Uitkomst Cave Beds 2 and 3.

Open settlements. Here stone walls demar-cate:

1. Industrial areas Furnace behind semi-circular wind-screen on edge of MelvilleKoppie Class 3 structure. Furnaces inlarge enclosure at Olifantspoort, Mase-kwapoort. Furnace behind semicircularwind-screen at Shankare.

2. Cattle pens large (see Klipriviersberg, fig.6). No direct evidence—inference fromearl}- historic accounts, e.g. Casalis3

p. 123.3. Cattle pens or small livestock pens (see

Klipriviersberg, fig. 6); interpreted onbasis of pens I observed at Himba settle-ments on the Kunene River. Note threestructures consisting of larger enclosure,smaller enclosure and 'tail'.

4. Men's eating places Klipriviersberg storehut and granary base, fig. 6, Kurrichanc,a larger version of stone wall and granarybase excavated in 1966.

AREAS OF IRON AGE SETTLEMENT CONCENTRATION

ft'&i^ffs^ \£££&L^

i™®??^A l ifl i i i l l

| v •••'-.•. [Not analysed [ | Alt Photo analyil* by | • • |l*oUt«d «itt»Archaeological RataarchUnit. Univarilly of lha Wltwit.nrtnd

Fig. 4 Areas of Iron Age settlement concentration.

TRANSVAAL AND NATAL IRON AGE SETTLEMENT

5. Stone enclosure between clay huts andouter enclosing walls Class 3 structureKlipriviersberg. Probably domestic en-closure for occupants or adjacent unit,demarcated from domestic area of neigh-bouring hut (fig. 6.)

6. Small stone huts probably roofless, foroccupation by herd boys guardingentrance to cattle pens? (See Klipriviers-berg, fig. 6). Small corbelled stone hutsas at Klipriviersberg.

7. Small low stone structuresCircular cooking areas paved (Kaditsh-wene 1966)

Paved fireplace and semicircular wind-screen anchor (Klipriviersberg, fig. 6).

Topographic Distribution of Iron Age Settle-ments

The present paper refers mainly to IronAge settlements in open country. Thereader should note that Iron Age settlersmade extensive use of caves as well.9 Thecount of 6,237 settlements made in the presentstudy is probably far below the actualnumber of settlements on the ground, andis also subject to measurable variation inperceptual differences between the workers

KUPRIVIERSBERG IRON AGE SITE

POSSIBLE FUNCTIONS OF ENCLOSURE

CI*Y will

Sim will

Fig. 6 Functional interpretation of Class 3 structure excavated at Klipriviersberg.

— ' • " • •

who made the count, as indicated by theKlipriviersberg test (below). The differencesin class proportions shown in the table(p. 3) are probably valid in a general sense,however, although the total counts willcertainly be modified in future work. Inearlier studies'1 I noted the tendency forIron Age settlers to build on high ground,the slopes or summits of hills or koppies.The main point emerging from the presentsurvey is the concentration of settlements inthe drainage areas of major rivers or streams(fig. 4). The table lists nine areas of Iron Agesettlement concentration, eight in the Trans-vaal and one in northern Natal. More detailedwork would probably define a greaternumber of minor areas of Iron Age concen-tration. It seems clear that Iron Age settlersavoided extensive grasslands such as thegrasslands of the western Karoo Highveld9

on the Lichtenburg Plain, possibly becauseof the rarity of surface streams and low reliefwhich denied water and natural topographicprotection to settlers in such territory. Thedistribution of Iron Age settlements in thenine areas provides a basis for planning theUniversity of the Witwatersrand Iron Ageproject. Future excavations will be plannedto detect similarities and differences in IronAge behaviour in the nine areas.

Cattle and Iron Age Settlement PatternsPart of the cause of historically recorded

endemic conflict between Transvaal Iron Agesettlements was probably cattle. The needto water cattle probably accounts for theconcentration of Iron Age settlements in thedrainage areas of major streams in the Vaaland Limpopo Basins. A. Leeds and A. Vaydaprovide a useful summary of relationshipsbetween cattle and social conflict in sub-sistence level economies.7

Problems of Iron Age Structure Classificationon aerial Photographs

The classification of Iron Age stone struc-tures like other artefact-classification is

subject to a margin of error due to thevariability of plan-form which necessarilyforms the basis of the vertical or slightlyoblique view of such structures as depictedon air photos. As noted, the classification ofthe 6,237 structures .in five main classesobserved in the Transvaal and northernNatal (25°-27°S., 26°-31°E.) represents to alarge extent a consensus of opinion by sevenworkers, working simultaneously on adjacentaerial photographs and classifying individualstructures after discussion. While it was notpossible to discuss in detail the classificationof every single structure, intermittent butpersistent discussion led to a consensus ofdiffering perceptual capacities. In order totest the effects of isolation on classification,the seven different workers studied the sameaerial photograph (Aircraft Operating Com-pany photo of Klipriviersberg area south ofJohannesburg, included site fig. 6 excavatedby University of Witwatersrand Archaeologyclass 1965-66). The photo was chosen becauseof the exceptional complexity of Iron Agestructures visible on it.

Test classification by workers, trained asa team, working in isolation on same air pho-to of Iron Age structures on Klipriviersberg

Worker1234567

Class 127292317121316

Class 362515052304444

Class 4a22

.—1213

Wide variation in individual counts isapparent, from 30 to 62 for Class 3, from 12to 29 for Class 1, and from 0 to 3 for Class 4a.On the other hand, all seven workers iden-tified Classes 1,3 and 4a with exception of oneworker who did not identify Class 4a. Withinthe wide range of variation in classification

TRANSVAAL AND NATAL IRON AGE SETTLEMENT

by individual workers, there is a certainmeasure of agreement.Class 1 Four of the seven workers counted

between 12 and 17, the remainingthree between 23 and 27.

Class 3 Two of the seven workers achievedthe same count of 44 structures, threeworkers counted between 50 and 52structures, counts by the remainingtwo workers ranging from 30 to 62.

Class 4a Here the greatest degree of agree-ment was reached: Two workerscounted 2 each, three workerscounted 2, and one worker counted 3.

The widest disagreement was in Class 3structure counts. A Class 3 structure isrepresented by a clear enclosing wall andinner circles. Reference to fig. 2, showing partof the actual photograph used in the test,will demonstrate the reason for the widedisagreement in count of Class 3, sinceadjacent enclosing walls and their innercircles are very frequently in actual contactwith one another. The worker who countedthe highest number of Class 3 structuresbased the count on single enclosing wallswith inner circles, while the worker whoachieved the lowest Class 3 count of 30 struc-tures merged adjacent enclosing walls in thecount.

Development of numerical taxonomic ormatrix analysis may reduce the margin errordemonstrated by the test classification of theKlipriviersberg structures. In the interim,the Klipriviersberg classification should beused as a measure of variability in thereader's assessment of the counts set out inthe total classification of 6,237 structuresnoted in the present paper.

Workers' Comments on Iron Age StructureClassification

No satisfactory method for the classifica-tion of terracing could be established. Clas-sification in terms of areas enclosed by ter-racing is suggested.

Class 3 (enclosing wall with inner circles)occasionally joined to one another, presentingdifficulty in counting.

Classes 4a and 4b were divided in terms ofa size classification. The lack of specificdifference led to confusion by differentworkers on different photographs depictingthis class.

Since Class 4 has no boundary wall, andwas frequently widely dispersed, it wasuncertain where the periphery lay, andseparate sites in this class may havebeen included in a count of one in theclass.

Class 5 included structures of uncertainclassification. Whether or not a particularstructure was uncertain in classificationdepended on the opinion of the individualworker.

Natural features such as soil marks,desiccated pans, and concentric vegetationgrowth often simulated stone wall struc-tures.

Air-photo Iron Age Structure Count comparedwith Ground Survey Structure Count

An aerial photo count of structures prob-ably represents a figure substantially lessthan the actual occurrence of structuresbecause of coverage by bush or other vegeta-tion. Our detailed knowledge of Iron Agestone structures visible on the ground atvarious sites in Johannesburg and theMagaliesberg proved that many Iron Agestructures are not visible on air photos. Forexample flight plan 406, photo No. 9376{Moedwil farm west of Rustenburg) shows adense plant cover approximately | mile inlength and £ mile in width. Our traverse ofthis area revealed a large Class 4a structureentirely invisible on the air photo. There canbe little doubt that the stone structures onthis spot encouraged plant growth which hasnow almost entirely obliterated them.

I traversed extensive Iron Age structureson the summits of the Matluassi Hills near

Matluassi Mission. These are entirelyobscured by vegetation on the air photos.

Many Iron Age sites are represented byovergrown entrenchments on the grasslandbetween Witbank and Carolina. These arenot visible on the air photos.

Comparative Study of Iron Age and Present-day African Settlement. Possible combinationof wood and stone in Iron Age structures.

The range of variation in height of stonewalls at Iron Age structures examined on theground in the Johannesburg area, at Melville,Waterval and Bedfordview, and at Olifants-poort (Rustenburg district) is between ap-proximately 18 inches and 9 feet. The Water-val, Melville and Bedfordview structures arebarely 18 inches in height and there is no evi-dence to show that the walls were originallyhigher. While no direct evidence could befound, it is possible that the stone walls atthese sites served merely to anchor stockadesor withy-fences. Though widely separated inspace and time from these Johannesburg arealocalities, I observed a combination of awooden stockade superstructure anchored instone wall foundations at three sites in theSoutpansberg (northern Transvaal), at Mago-ngoza's Kraal, Masekwapoort (both probably19th century sites) and at the present Vendachief's village at Mukumbani (Sibasa dis-trict) ('South African Iron Age and Present-day Venda Architecture and Pottery', p. 15of this issue). Comparative study of Iron Ageand present-day African settlement, begun in1964, is to be extended.

Functional Interpretation of MorphologicalClassification

Terminal Iron Age societies in the Trans-vaal described by early explorers such asJ. Campbell, were torn by inter-settlementconflict over cattle and other possessions.One major functional reason for Iron Agestructure design and site selection may havebeen defence against sudden-attack. Defence

appears to have depended upon sufficientwarning for counter-attack, not upon pro-longed siege. Reference to the structurecount totals (p. 169) shows that Class 3 com-prises 3308 of the 6237 Iron Age structurescounted. Class 3 structures comprise a clearenclosing wall with inner circular structures.Class 3 structures usually occur in smallgroups substantially smaller than the largestgroups, those in Class 4. The Class 3 en-closing walls probably functioned partly asdefensive walls, as well as walls for contain-ing livestock and perhaps small children.Class 4 structures lack enclosing walls, thedefensive or retaining function of enclosingwalls being replaced by the mere size of thesettlement complex as a whole.

As noted, since 1963 we have made detailedexcavations at Iron Age sites, representedby settlements in open country as distinctfrom Iron Age cave sites, at the followinglocalities: Klipriviersberg, Melville, Waterval,Kurrichane (in the southern or westernTransvaal). In addition, we have excavatedopen localities at Nareng and Shankare inthe Phalaborwa area. Excluding the latter,we find limited historical evidence concerningthe function of Class 3 and Class 4 structures.J. Campbell's 1820 account of 'Kurrichane'(Kaditshwene)2 indicated the social and poli-tical function of a Class 4 structure, and re-vealed that a certain amount of indus-trial activity in the form of metal productiontook place at such sites. Class 4 structurescould house up to 16,000 inhabitants, ifJ. Campbell's estimate of the Kurrichanepopulation is correct. My detailed excavationof a Class 3 structure at Klipriviersberg re-vealed the remains of clay-walled hutsapproximately 10 feet in diameter, spaced atregular intervals of 30 feet, the whole resembl-ing the illustration of a Rolong village by E.Casalis (3p. 123). The excavation of muchof the Klipriviersberg Class 3 structure sug-gested that its functions were confinedto social activity, protection of domestic

1 *•.}*.- .* v \ *

• ' ; : ' • • • ' ' •

;CJL.

TRANSVAAL AND NATAL IRON AGE SETTLEMENT

animals, and storage of domestic grain(fig. 6). I discovered grains of Sorghumsp. in the ash heap at this site. A well-preserved furnace and metal smelting en-closure was uncovered at the north-westcorner of a Class 3 stone structure on thenorth pediment of Melville Koppie. A morecomplete functional analysis of Iron Agestone structures is to be published at a laterdate when site excavation reports will bepresented. The evidence suggests a widerange of variation in the actual functionof the structures. I have given a generalaccount of the possible relationship betweenstructure and function in the TransvaalIron Age.9 J. D. Seddon presents a theoreticaldiscussion of Iron Age structure function inhis article 'An Aerial Survey of Settlementand Living Patterns in the Transvaal IronAge', p. 189 of this issue.

Distribution of Similar Structures in AfricaStructures similar to some of those visible

between 25°-27°S. and 26°-31°E. appear tooccur sporadically throughout the less for-ested parts of Africa south of approximately15°N. 3°W. in the north-west, and 2°S. 8°E.in the north-east. Henry Barth's plan of thechief's house at Aribinda, encountered be-tween Sokoto and Timbuktu, resembles aTransvaal Class 2a structure (*• p. 242), whileMary Leakey's Hyrax Hill plans resembleClass 1 structures.a

Dating of Stone Structures 25°-27°S., 26a-31°E.The reader should note causes of error in

C14 age estimations given by M. Stuiver andH. Suess.16 The earliest radiocarbon datefor Iron Age artefacts in the area 25°-27cS.and 26°-31 °E., under discussion in this paper,isA.D. 1060 ± 50 (Y-1338). This date refersto a metal smelting furnace of two-aperturetype, including stamped pottery, but notincluding stone walls. An identical piece ofpottery, about one-third of a bowl rim, wasfound within a complex Class 1 structure

about 300 yards north-west of the A.D. 1060furnace, and a smaller Class 1 structure liesabout 20 yards to the south of the A.D. 1060furnace. The earliest date directly relativeto a stone wall is A.D. 1650 ± 80 (Y-1323B)at Uitkomst Cave. The earliest dated com-plex Class 1 structure is at Waterval, A.D.1720 ± 40 (GrN-4797). The earliest datedClass 3 structure is at Klipriviersberg, A.D.1755 ± 45 (GrN-4926). These dates, limitedand open to a wide margin of error as theyare, suggest a generally second millenniumdate for the stone structures under discussionin the present paper. This statement is nomore than a guess, however, and no chrono-metric dating hypothesis may be regarded asfirmly established on less than 100 measure-ments. Early dates for stone structures inAfrica may be expected consequent on thedating of the East African Engaruka hillsidestructures from A.D. 330 ± 90 (GX-0348)onwards given by H. Sassoon.l 3 Somerecently completed radiocarbon age estima-tions for sites relevant to the present dis-cussion are given in the table, pages 10 and11.

The Culture Area Concept and Areas of IronAge Settlement Concentration

In 195281 applied the culture area conceptto Iron Age settlement in the SouthernTransvaal and Orange Free State, extendingit to Western Transvaal in 1962.9 J. D.Seddon's application of the territorial im-perative to Iron Age studies relates theculture area concept to studies of animalbehaviour. The nine areas of settlement con-centration delineated by the present aerialphoto analysis may be explained as expres-sions of distinct Iron Age behaviour linkedwith the environmental variation from areato area, or merely as distinct behaviouraldevelopments due to relative geographicalisolation. Large scale total analyses of arte-facts, settlement layout, foodwaste andsettlement ecology, already commenced, may

TRANSVAAL IRON AGE DATES—APRIL 1967—SAMPLES BY ARCHAEOLOGICAL RESEARCH UNIT,

UNIVERSITY OF THE WITWATERSRAND, JOHANNESBURG

Subject to error stated by M. Stuiver and H. Suess.u

Site

Klipriviersberg(just south ofJohannesburg)

Melville (Johan-nesburg suburb)

Uitkomst Cave

Waterval (Johan-nesburg suburbCorriemoorNorthcliff)

Shankare(Phalaborwa)

Lat.

26°11'S.

26°20'S.

25°55'S.

26°09'S.

23°S5'S.

Long.

28°02'E.

28 °E.

27°45'E.

27°56'E.

31°10'E.

Locality or structure class

From 60 cm depth in ashheap adjacent to Class 3Iron Age Structure Posi-tion Db/Ea/11.

Charcoal in collapsed claywall of hut 210/E(i) depth± 10 cm. Class 3 Iron Agestructure.

Ash layer on floor of stonehut 220/f depth 10-15 cm.Lower habitation layer.

Furnace. Iron Age 1 12*depth.

Furnace on edge of Class 3structure.

Furnace Bed 3. Behind stonewall at mouth of cave.

TU 15. grid ref. Charcoal onfloor about 10 cm belowsurface adjacent to stonewall of complex Class Istructure.

Charcoal from floor aroundbase of anvil next tofurnace behind stone windbreak.

C u Date

A.D. 1845 ± 50.

A.D. 1755 ± 45

A.D. 1770 ± 5 0

A.D. 1060 ± 50

A.D. 1860 ± 6 0

A.D. 1650 ± 8 0

A.D. 1720 ± 4 0

A.D. 1660 ± 80

Lab. No.

GrN-4796

GrN-4926

GrN-4927

Y-1338

Y-1768

Y-1323B

GrN-4797

Y-1766

Comment

Archaeologically similar to GrN-4926and GrN-4927. Date suggests con-tamination. With stamped potteryof Uitkomst style.

Probably same occupation as GrN-4927.With stamped pottery of Uitkomststyle.

Probably same occupation as GrN-4926.With stamped pottery of Uitkomststyle.

Not known if this date representscontamination by early carbon.This regarded as unlikely. Withstamped pottery of Uitkomst style.

Furnace and pottery similar to Y-1338.Recent date suggests contaminationalthough charcoal was collected onbase of furnace bowl. Completelysealed from present surface. WithUitkomst style stamped pottery.

Possibly acceptable date. Withstamped pottery of Uitkomst style.

Possibly acceptable date. Withstamped pottery of Uitkomst style.

Charcoal was mixed with iron chipsfrom artefact making on stone anvil.Possibility of contaminationunknown. Y-1766 is part of a largeIron Age settlement of unknownduration. With incised pottery ofMakapan style.

duration. WithMakapan style.

incised pottery of

Shankare(Phalaborwa)

Shankare(Pliukilmnva)

Nareng(Phalaborwa)

23°55'S.

23°55'S.

23°55'S.

31°10'E.

31°I0'E.

3l°10'E.

Charcoal from domesticrefuse on terrace onnorth slope of Shan-kare behind retainingwall of stones.

Charcoal from floor of 3-apcrturc furnace grid ref.N/15 O4-7/-.2'-.4' .Char-coal was sealed ± ' 2 cmbelow present surface.On north pediment ofShankare.

Charcoal on floor adjacentto furnace, dating tooperation of furnace, 30cm below present surface.Furnace of 3-aperturetype identical to GrN-4929.

A.D. 1860 ± 6 0

A.D. 1956=12 I

A.D. 1840 ± 2 5

A.D, 1790 ± 6 0

Y-1769

GrN-4929

GrN-4928

Y-1767

Date suggests contamination althoughimmense mass of domestic refuse onpediment of Shankare indicatesIron Age settlement of long duration.Date of A.D. I860 ± GO may there-fore represent terminal Iron Ageoccupation in this area. Withincised pottery of Makapan style.

Date suggests contamination by recentcarbon. GrN-4929 furnace is iden-tical to GrN-4928 furnace. Withincised potterv of Makapan style.

Note overlap with C.,( date by YaleLaboratory Y-1767. Both datesrefer to same charcoal sample fromisolated smelting site on north pedi-ment of Nareng —a small koppieeast of Phalaborwa. With incisedpottery of Makapan style.

12

reveal the nature and inter-relationships ofregional Iron Age developments. The prob-lem may be complex due to successive oroverlapping occupation within the sameregion. At least three distinct Iron Ageoccupations are registered in the stratificationof the Klipriviersberg structure. Iron Age set-tlers appear to have had no qualms about re-use of abandoned stone wallings. Some IronAge caves, such as Uitkomst, preserveevidence of separate Iron Age occupations.The early historic destructive traverse ofTransvaal Iron Age settlements by Mzilikazimay not have been the first conquest tolink isolated settlement areas under anephemeral overlay of uniformity.

Analyses of Iron Age FoodwasteIn 1937 L. Fouche published the first

analysis of Transvaal Iron Age foodwastefrom Mapungubwe. In 1951 I published thefirst analysis of Iron Age foodwaste fromSouthern Transvaal cave sites. C. K. Brainis now making more extensive studies of food-waste produced by excavations at Klip-riviersberg and 'Kurrichane' (Kaditshwene),and has so far identified domesticated goats,sheep, cattle and a range of undomesticatedanimals. Regrettably, most foodwaste isconfined to bone. Plant food remains arelimited to 3 seeds of Sorghum sp. I recoveredat Klipriviersberg.

Attribution of SettlementsThe following evidence indicates Bantu as

makers of stone structures:All artefacts found at excavation are

similar to those still made by Bantu, orhistorically attributable to them.

No exotic non-Bantu artefacts occur atTransvaal sites except beads or porcelain.

Though rare, hominid skeletal material atIron Age sites is negroid, with the exceptionof Mapungubwe, where skeleton material isBush Boskop, and artefacts, pottery andarrow heads are different from most IronAge assemblages.

Iron Age Settlement Patterns and Bantu Settle-ment Today

Bantu settlement in the Transvaal todayis being rapidly changed in terms of agricul-tural planning policy, but air-photo studyand field research shows that:

Venda tribesmen today still build stonestructures for their chiefs at Tshivhase andTshimbupfe similar to the Iron Age walledstructures found in the Soutpansberg andPhalaborwa.

The present-day Bantu homesteads in theNorthern and Eastern Transvaal are builtaround a unit of two or three cone andcylinder huts on the edge of an area enclosedby a clay or thorn bush wall. I have detectedsuch settlements overlapping underlying IronAge structures.11 Within the Klipriviersbergstone walls I discovered remains of clay-walled huts C14 dated to c. A.D. 1750, prob-ably similar to present-day Bantu huts.

In flights over Bantu settlement units inthe Eastern Transvaal I noted farmlands upto approximately fifty acres enclosed bythorn scrub fencing or hedges. There is notrace of the latter in the archaeological con-text, but the present-day concept of a fewhuts related structurally to an enclosed livingarea probably descends from the Iron Ageconcept, where the huts were actually en-closed within the living area. The change tothe present-day concept probably reflectsgreater security from external attack. B.Sansom suggested to me that greater com-petitiveness within the community accountsfor the dispersal of much of the SekukunilandPedi into small units of a few huts adjacentto an enclosed living area.

While stone walled settlements of Iron Agestyle were occupied till 1881 at Mapogstad,9

in other areas wooden stockades replacedstone walls early in the 19th century. A. P.Borcherd, who visited New Littakoo (Litha-kong), a wood and clay hut settlement in-cluding wooden stockades, near Kuruman in1801, noted that ruins of Old Littakoo were

TRANSVAAL AND NATAL IRON AGE SETTLEMENT 13

in stone. The Batlapin of New Littakoo didnot know who had built the stone walls ofOld Littakoo.

ConclusionThe 6,237 structures identified in our 1967

analysis probably represent only a part of thetotal Iron Age settlements in the area. Esti-mates of the dating range of settlement mustawait extensive radiocarbon measurement.Very few sites suggest prolonged occupationof the same site by different generations, themaximum recorded depth of refuse being lessthan 6 feet as determined at Kurrichane(Kaditshwene) in 1966. Some settlementswere re-occupied after a substantial lapse oftime. At Klipriviersberg I recorded at leastthree separate occupations. Borcherd's ob-servation at Littakoo (Lithakong) indicatesthat Iron Age communities building in woodoccupied areas formerly the territory of IronAge people building in stone. We concludethat the Transvaal Iron Age represents anexceptionally complex stage in the transitionfrom Stone Age to present-day society.Future work will be directed to large scaleexcavation of the localities indicated.

ACKNOWLEDGEMENTS

Thanks are due to the following:The African Studies Programme of the

University of the Witwatersrand, Johan-nesburg, for financial assistance.

The Government Printer, Pretoria, for per-mission to publish results from official airphotograph records, and Aircraft Operat-ing Company.

Dr J. Vogel of the Groningen Laboratory andDr M. Stuiver of the Yale Laboratory. MrJ. D. Seddon for discussion.

Mrs M. Briggs and Mr P. Wallace for detailedmapping of analysis results, and Miss J.Rouse for preparing the plates.

Mrs M. Briggs, Miss M. Coombes, Miss P.MacClarty, Miss R. Pelletier, Miss J. Karkand Mr D. J. Stewart for carrying out theair photograph analysis under my super-vision.

Students of the University of the Witwaters-rand, Johannesburg, and members of theSouth African Archaeological Society forhelp in excavations.

Mr A. Retief, Mr P. Meyer, Mr J. Vosloo, MrJ. P. Snyman and Mr F. Cilliers for per-mission to camp or excavate on their farmsKlipriviersberg, Kwarriefontein and Ver-genoegd.

The Johannesburg Council for NaturalHistory for financial assistance in theMelville excavations.

Dr J. Schoeman and the Phalaborwa MiningCompany for generous practical assistanceat Nareng and Shankare.

REFERENCES

1. BARTH, H. Travels and Discoveries inNorth and Central Africa, London, 1890,Vol. 2.

2. CAMPBELL, J. Travels in South Africa,London, 1822, Vol. 1.

3. CASALIS, E. The Bastttos or Twenty-threeYears in South Africa, Cape Town, 1965.(C. Struik reprint)

4. CLARKE, D. L. Matrix Analysis andArchaeology with particular reference tothe British Beaker Pottery, Atti del VICongresso Internazionale delle ScienzePreistoriche e Protostoriche, Vol. II, 37-42, 1965.

5. HODGSON, T. L. Letters to the WesleyanMissionary Society London No. 26 fromMatluassi 12.1.1824. Not yet published.

6. LEAKEY, M. D. and L. S. B. Report onthe excavations at Hyrax Hill, Nakuru,Kenya Colony, Trans Roy. Soc. S. Aft.30:271-409, 1945.

14

7. LEEDS, A. and VAYDA, A. Man, cultureand animals, Amer. Assn. Adv. Sci. 78:1965.

8. MASON, R. J. South African Iron Agepottery from the Southern Transvaal,S. Afr. Archaeol. Bull. 26(7) 70-79, 1952.

9. MASON, R. J. Prehistory of the Transvaal,Johannesburg, 1962.

10. MASON, R. J. The Prehistoric People ofMelville Koppies, Melville KoppiesNature Reserve Guide Book, Part 2,Johannesburg, 1963.

11. MASON, R. J. The Origin of SouthAfrican Society, 5. Ajr. J. Sci. 61 (7)255-267, 1965.

12. MOFFAT, R. Matabele Journals of RobertMoffat 1829-60, London, 1945.

13. SASSOON, H. Engaruka: Excavationsduring 1964, Azania I: 79-100, 1966.

14. SEDDON, J. D. Kurrichane: A Late IronAge Site in the Western Transvaal, Afr.Stud. 25: 227-231, 1966.

15. SMITH, A. The Diary of Dr AndrewSmith 1834-36, ed. P. Kirby, Cape Town,1940, Vol. II (Van Riebeeck Society21).

16. STUIVER, M. and SUESS, H. E. On therelationship between radiocarbon datesand true sample ages, Radiocarbon 8:534-540, 1966.

15

ions• • »

3.

ron4fr.

'rewwn,iety

theites

8:

SOUTH AFRICAN IRON AGE AND PRESENT-DAY VENDAARCHITECTURE AND POTTERY FROM THENORTHERN TRANSVAAL, SOUTH AFRICA

R. J. MASON

The products of some present-day Trans-vaal tribes may be related to the archaeolo-gical record in or near their present places ofsettlement. Numerous ethnographers havereported products of Venda activity whichresemble prehistoric remains in the same oradjacent territory. Before 1930, H. A. Staytnoticed erect stones in Venda walls and com-pared these with Rhodesian Iron Age features(Stayt 1931); G. Gardner published a planof a hut, carved door and surrounding stock-ade, excavated by him on Mapungubwe hill-top, dated to A.D. 1380 ± 60, which resem-bles present-day Venda structures in somerespects (Gardner 1963). In 1931 P. Kirbyvisited Chief Tshivhase's capital atMukumbani in central Vendaland and sawthe building of a massive stone structure tocelebrate the accession of a new chief, andalso reported wooden stockades at the site(Kirby 1956).

In 1963 I initiated a new Iron Age pro-gramme in the Transvaal, now part of theAfrican Studies Programme of the Universityof the Witwatersrand, Johannesburg. Ex-cavations of Iron Age localities have nowbeen made at three sites on the Witwaters-rand, one near Zeerust and two at Phala-borwa. Preliminary ethnographic studiesrelated to the Phalaborwa remains were madein the Soutpansberg, Northern Transvaal, inor on the fringe of traditional Venda settle-ments.

In 1962 and 1964 I visited two stone-walledsettlements in the Waterpoort and Masekwa-poort districts, central and western Soutpans-berg respectively, under the guidance of

Mr P. J. Uys, a keen amateur archaeologist.On the basis of local tradition, Mr Uysattributes both settlements to late nine-teenth century Venda builders. The settle-ments cover a few acres on steep mountainslopes similar to present-day Vendasettlements at Mukumbani and Tshimbupfe.Well-built stone walls demarcate living andworking areas at the sites. The former aresuggested by hut remains, the latter bysmelting furnaces. Remains of woodenstockades inserted into the enclosing stonewalls, seen at Masekwapoort, have not beenrecorded previously at prehistoric sites inSouth Africa to my knowledge (figs. 1 and

2).Late in 1964, and again in 1966, I visited

Mukumbani and Tshimbupfe in central andsouthern Vendaland. Here the presentchiefs' homes, administrative offices and thehomes of their families and tribal offices, areadjacent to deserted remains of their grand-fathers' settlements. At both sites the grand-fathers' settlements are located on the sum-mits or far up the slope of adjacent hills,indicating a movement down the slope fromthe paternal, nucleus settlement in relativelyrecent times. Professor J. Blacking kindlyinformed me that another reason for settle-ment movement was the death of a chief.This movement repeats a feature widespreadin the Transvaal, where early nineteenthcentury Iron Age settlements are to be foundmainly on hilltops, while the descendantshave today occupied valley floors or lowerslopes, presumably due to greater security atthe present time.

16

These two chiefs' personal houses atMukumbani and Tshimbupfe were built inEuropean style, but many of the surroundingwalls and stockades, including some I sawunder construction in September 1964,closely resemble stone walls and stockaderemains at the presumed late nineteenthcentury sites at Masekwapoort and Water-poort (figs. 3-7). The wall-building methodsin use today at Tshimbupfe and Mukumbanialso closely resemble techniques evident atIron Age stone-walled settlements through-out the Southern Transvaal (Mason 1962).Vertical stones have been erected in some ofthe walls. At Mukumbani some stones andwooden poles are surmounted by bovidhorns (figs. 4-5).

The Mukumbani and Tshimbupfe walls,built during or just before 1964, wereintended to separate areas for differentactivities. Some of the walls retain terraceson steep slopes. Others are speakers' ormusicians' podia. Those I saw at Tshimbupfewere built by Venda tribesmen at their chief'srequest. A team of eight men used Europeantools for the building of the walls, but fol-lowed the techniques evident at the pre-historic sites (fig. 8). Similar stone struc-tures may be seen elsewhere in the VendaReserves. Chief Ne-Tshimbupfe informed methat most Venda men were capable of build-ing in stone, though some were more expertthan others. "Like making porridge", he said.

Venda clay fireplaces, with shallow bowl-like depressions, seen near Tshimbupfe(fig. 9), exactly resemble clay structures weexcavated at Shankare Iron Age sites,Phalaborwa, where stone terraces and wallsare also to be seen.

Tshimbupfe is also a centre of pottery-production. Pottery purchased here in 1964

showed marked resemblances in incisedmotifs, rim finish, surface finish and form topottery excavated by us at Iron Age sites atShankare and Nareng, Phalaborwa, andalso in the uppermost bed in the Cave ofHearths at Makapansgat. Figs. 10, 11 illus-trate some prehistoric and present-day pot-tery from these sites.

We may conclude that both present Vendaarchitecture and Venda pottery show closeparallels to prehistoric materials in the sameor adjacent territories. We propose toinvestigate these relationships in more detail.

ACKNOWLEDGEMENTSI wish to thank the Department of Bantu

Administration for permission to visit Venda-land and publish the present note. Mr J.Witt introduced me to Phalaborwa, Mr P. J.Uys took me to the western and centralSoutpansberg sites and Professor J. Blackingkindly read the note. Dr N. J. van Warmelochecked the orthography and place names.Dr J. J. Schoeman kindly assisted with field-work in the Phalaborwa area. Mr P. Devittintroduced me to Mukumbani in 1964 andthe late Chief Tshivase kindly welcomed methere.

REFERENCESGARDNER, G. A. 1963. Mapttngubwe II, Van

Schaik, Pretoria.KIRBY, P. R. 1956. The building in stone of

a new kraal for the paramount chief of theVenda, 5. Afr. J. Sci. 52(7) 167.

MASON, R. J. 1962. Prehistory of the Trans-vaal, Witwatersrand University Press,Johannesburg.

STAYT, H. 1931. Notes on the Bavenda andtheir connection with Zimbabwe ZimbabweCulture. Oxford University Press, Oxford.

IRON AGE AND PRESEXT-DAY VEND A ARCHITECTURE AND POTTERY 17

Fig. 1. Masckwapoort. Remains of wooden stockade in late 19th century stone wall enclosing chief's burial hut.

Fig. 2. Masckwapoort. Remains of woodenstockade in late I9th century stone wall.

IS

Fig. 3. Mukumbani. Chiefs village enclosed by high stone wall.

Fig. 4. Mukumbani. Chief's village with stone wall, erect stones and horns.

IRON AGE AND PRESENT-DAY VENDA ARCHITECTURE AND POTTERY 19

Fig. 5. Mukumbani. Chief's village with stone wall and wooden posts.

Fig.6. Mukumbani. Wooden stockade in stone wall.

20

Fig. 7. Mukumbani, September 1964. Tshikona dance. Enclosing stockade in stone wall foundation andmusician's stone platform.

Fig. 8. Tshimbupfe, September 19(54. .Men building wall at Chief's village.

IRON AGE AND PRESENT-DAY VENDA ARCHITECTURE AND TOTTERY 21

Fig. 9. Clay fireplace at Tshimbupfe.

22

Fig. lO.Rim section of a pot from 6"-9" below surfacelevel at Nareng Iron Age smelting site. Phalaborwa.

Fig. 11. Rim section of a pot made by Xyamuka-madi, a Muvenda woman at Tshishongc, Tshim-bupfe, September 1964.