South African Archaeological Society

South African Archaeological Society
The Early Iron Age and the Spread of the Bantu
Author(s): Thomas N. Huffman
Source: The South African Archaeological Bulletin, Vol. 25, No. 97 (Jun., 1970), pp. 3-21
Published by: South African Archaeological Society
Stable URL: http://www.jstor.org/stable/3888762 .
Accessed: 09/03/2011 15:11
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=saas. .
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
South African Archaeological Society is collaborating with JSTOR to digitize, preserve and extend access to
The South African Archaeological Bulletin.
http://www.jstor.org
THE EARLY IRON AGE AND THE SPREAD OF THE BANTU*
THOMAS N. HUFFMAN
Universityof Illinois, Urbana,Illinois, U.S.A.
Of all the great cultural movementsof the world,
the Bantu migrationis one of the most strikingand
impressive. The close similarities among widely
separated Bantu languages have interested scholars
for many years. To be so close linguisticallyand yet
so geographicallydispersed,it has been argued, the
Bantu must have spread rapidly and quite recently
from a common centre. At Europeancontact, Bantu
speakerswere moving to the north and south, indicating the expansionwas still in progress.
Africanistsusually have tried to explain the rapid
dispersalof the Bantu in one of two ways: conquest
(Johnston, 1913; Wrigley, 1960)or populationexplosion (Murdock,1959;Oliver,1966).The majormigration theories,withoutexception,arebasedon linguistic
data, resulting in imaginativepatterns of expansion
to explainthe distributionof similarlanguages.Other
sources of information-especially archaeological
data-are merely adjusted until they fit the already
set linguisticinterpretations.
Linguistics,however,contributesto culturehistory
in a numberof ways: (1) by determiningthe common
origin and subsequent diversificationof languages,
thus implying a diversificationof people; (2) by
suggestingdegreesof relationshipbetweenlanguages
throughlexicostatistics;(3) by determiningborrowed
features in a language, indicating culture contact;
and (4) by suggestingelementsof culturecontent and
physicalenvironmentfroma reconstructedvocabulary
(Swadesh,1960).Linguisticsand archaeologytogether
supply informationon population size, density, and
movements,as well as economic patterns, However
the linguistic and archaeological data should be
evaluated separatelybefore collating their results.
Recently, several hypotheses about Bantu origins
have been suggested which are based on untested
archaeological premises, and these in turn have
fostered other hypotheses. Most literatureon Bantu
origins has overlooked alternative archaeological
interpretations and has often elevated untested
hypotheses to the level of fact. This is probably
because too little attention has been devoted to the
evidence behind the assumptionsscholarscommonly
make about the Bantu and early Iron Age archaeological remains. Recently Posnansky (1961b; 1968),
Phillipson (1968a), and Hiernaux (in Fagan, 1969)
have challengedthese commonlyacceptedhypotheses.
I propose to pursue their line of reasoning by synthesizing the pertinent early Iron Age evidence and
relatingit to the newly developingviewpoint.
The Problem
The Iron Age connotes different ideas to many
people. It is generallyused in Africa, however, as a
* Received May 1969.
3
S. Afr. Archaeci. Bull. 25: 3-21.
culturaltermdescribinga particularway of life instead
of a temporal term including all remains between
certainboundarydates.Althoughthe termemphasizes
a single trait, several others are equally important;
for example, semi-permanentvillages of pole and
daga huts and grain bins, cereal agriculture,animal
husbandry,characteristicpottery, and often copper
ornamentsas well as simple iron tools and weapons.
Even though directproof is often lackingfor some of
these traits, there is sufficient indirect evidence to
support their use as characteristictraits of the Iron
Age in East and southernAfrica. The use of the term
'Iron Age culture'in this paper implies these characteristics.
Certain Iron Age cultures in sub-SaharanAfrica
are generally accepted as archaeologicalindications
of the Bantu expansion. Several inferences from
linguistics, archaeology, and physical anthropology
supportthis correlation:(1) All Bantu languagesare
related to a proto-Bantulanguage with a homeland
in the southernCongo (Guthrie, 1962) and then to a
pre-Bantulanguagesomewherein WestAfrica(Greenburg, 1955; Guthrie, 1962; Murdock, 1959). (2) Languagesare usuallyspreadby people and not diffusion.
(3) The Iron Age is recent in southern Africa and
representsa migrationratherthan a diffusion.(4) Late
Stone Age Industriesin southernAfrica are thought
to be the productsof peoplelike the modernBushmen.
(5) The only substantiatedclaimsfor Iron Age peoples
in southernAfricaat Europeancontact are for BantuspeakingNegroes. And (6) there is a generalcorrelation between the distributionof Bantu languagesin
East and southernAfricaand IronAge archaeological
remains.On the basisof thesepremisesthe correlation
between the Bantu language and the early Iron Age
is extremelyplausible.
In recentyears an archaeologicaltheory has arisen
to explainthe latterpart of the Bantu expansion.The
formal statement of this theory is actually very
sketchy, and in reality there are probably several
theories. But there is a core of thought, and I have
takenthe libertyof consideringit the majorhypothesis
with severalminor variants.
Accordingto Clark(1962; 1967)the early Iron Age
pottery in East Africa (Dimple-basepottery) and in
southern Africa (Channel-decoratedpottery) are
virtuallythe same. Dimple-basepotteryhas even been
characterized as simply Channeled pottery with
'dimpledbases' (Oliver,1966).Cole (1963) and Fagan
(1965) have popularizedthis theory and formalized
its majordimensions:(a) Dimple-basepotterywas the
first Iron Age ceramiccomplex south and east of the
tropical forest, and (b) it was carriedsouth where it
degeneratedat Kalambo Falls and was then trans-
GOKOMERE
GOKOMERE
A
R
x1
(KALUNDU)
(ALUNDU
L/
A
(KAPWIRIMBWE)
(KAPWIRIMBWE)
V/
E
TA2
M
/
KALAMBO
KALAMBO
i
E
WALE)
i
,KWALE)
/s
I
/
A///
1/
DIMPLE-BASE
East Africa
RELATIVE SPACE
Southern Africa
Fig. 1. The Dimple-base/Channelware continuum
hypothesis. The names in parentheseshad not been
definedwhen the hypothesiswas originated.
formed into the Channel-decoratedpottery along the
Zambezi (Situmpa pottery). Throughoutthe rest of
the paper this hypothesis will be referredto as the
Dimple-base/ChannelWare continuum (abbreviated
DB/CW).
Figure 1 is a diagram of the DB/CW hypothesis.
If the right arrowaccuratelyportraysthe transformation of styles over space, then the left arrow is a
logical consequence;it representsthe impliedcorrelation of change in Dimple-basepottery in East Africa
over time. Because ceramic change has a closer
analogy to linguisticdiversificationthan to biological
diversity,when one tradition splits up, the resulting
traditionsretain many related traits and only gradually drift furtherapart. Consequently,the makers of
Dimple-base pottery that remained in East Africa
should have undergone a similar sequence of style
change through time as did those travelling from
East Africa to the Zambezi.The Iron Age industries
in parentheseswere not yet definedwhen the DB/CW
hypothesiswas originated.
At the time of Cole's and Fagan's publications,
sites with Channel-decoratedpottery (i.e. Gokomere,
Ziwa, or Situmpa pottery) were radiocarbondated
to the first centuriesA.D. whereasonly one site with
Dimple-base pottery was dated and that indirectly
to the second millenniumA.D., some 700 to 800 years
later. After noting the discrepancyin these dates,
Oliver (1966) wonderedwhy the movement was not
interpretedsomewhatdifferently,i.e. south and north
from the Zambezi. In a review of Fagan's Southern
Africa Duringthe Iron Age, Posnansky(1967b)asked
the same question-why north to south?
The fundamentalproblemcentresin the core of the
theory itself, i.e. Dimple-baseand Channel-decorated
4
potteryare basicallythe sameand spreadacrossAfrica
in a single movement.
If Dimple-base and Channel-decoratedpottery
sites are plotted on a map, they coveran areaapproximately 1,825 miles north to south, 1,675miles east to
west, and span at least 800 yearsof prehistory.Considerable caution is necessary before characterizing
all thepotteryas a wholein an areaso large.According
to the continuum hypothesis, however, the relationship betweenall the earlyIron Age potteryis basedon
a common decorative technique-channeling-and
it is subdividedinto waresby the presenceor absence
of dimpledbases. Instead of using multiplecombinations of attributes that probably would not have
occurredtogether by chance, the postulated connection dependson two very generaltraits.
Several regional variants in Zambia have recently
been definedby Phillipson(1968a) that do not easily
fit into the DB/CW continuumand cast considerable
doubt on it. Phillipson'sdata plus severalnew radiocarbon dates (Table I) have suggesteda re-evaluation
of the DB/CW hypothesis is appropriate.There no
longer appears to have been a 700- to 800-year
discrepancybetween the early Iron Age in East and
southern Africa. These radiocarbon dates allow at
least two different interpretationsof the ceramics.
(a) If one assumes the dates are relatively accurate
and the sitescontemporary,then a unilinearmovement
of a single style would have been extremely fast;
therefore,the earliest Iron Age pottery should look
about the same everywhere.Or, (b) one could assume
the dates are misleadingbecause of the variabilityin
radiocarbondatingand becausemore time is logically
necessaryfor a movement on the scale of the Bantu
expansion. One would then expect considerable
ceramic variability due to transformationthrough
time. Both choices must be tested, however, before
eithercan be accepted.
QualitativeComparisonof Six EarlyIronAge
CeramicStyles
Close examination of the available evidence
indicatesearlyIron Age ceramicswerealreadyregionally diversified and not the same when they first
appearedin East and southernAfrica (cf. Phillipson,
1968a).To illustratethis, six widelyseparatedceramic
groups have been selected for comparison: two in
East Africa (Dimple-baseand Kwale), one in central
Africa (Kalambo),two in southernZambia(Kalundu
and Kapwirimbwe),and one in Rhodesia(Gokomere)
(fig. 2). These six were chosen because(1) each consists of a numberof sites with similarpottery,(2) they
are relatively well dated, (3) they are recognizedas
the earliest Iron Age in their regions, and (4) they
are all well describedin the literatureor avilable for
first-handstudy.
At least one possiblesourceof errorexists,however:
some of the sites within any given group may not be
exactly contemporary. Consequently, what is considered to be normal variation within a group may
actuallybe change throughtime. The time dimension
TABLE 1
RADIOCARBON DATES
Dimple-base
Ndora
Cyamakuza
Urewe
Yala Alego
Kwale
Bombo Kaburi
Kwale
Rwanda
Rwanda
Kenya
B-755
B-758
GX-1186
N-436
A.D.
A.D.
Same JA.D.
Sample XA.D.
250? 100
300?80
390?95
320?110
N-435
A.D. 270? 110
Kenya
N-437
A.D.
400?235
Tanzania
Tanzania
N-347
N-291
A.D.
220?115
N-292
A.D. 270? 110
A.D. 260?i 110
Zambia
GrN-4646
GrN4647
A.D. 345?40
A.D. 430?40
Kapwirimbwe
Zambia
A.D.
A.D.
A.D.
Leopard'sHill Cave
Zambia
GX-1012
GX-1013a
GX-1013b
SR-126
A.D.
505?95
425? 110
410? 84
535?125
SR-65
SR-123
GX-i 114
GX-11 15
A.D.
300?90
Kalambo
Site C
Kapwirimbwe
Kalundu
Kalundu
Gundu
Gokomere
Mabveni
ZimbabwePeriod I
CighwaHill
Gokomere
Kinsale Farm
Zambia
Zambia
Rhodesia
Rhodesia
Rhodesia
Rhodesia
Rhodesia
SR-43
SR-79
M-913
SR-1 19
SR-26
SR-1 17
cannot be handled by this method (Spaulding,1960),
but it will be consideredin the next section. I have
tried to eliminate variation due to space within a
region.
Beforebeginningthe comparison,a few briefwords
are in order about terminology.The phrase 'broadline incision' is used here in place of 'channeling'or
'grooving'. Channeling and grooving connote a
woodworking technique that gouges out material
from the surfaceof the vessel.After examiningseveral
thousand so-called channel decorated sherds, no
clear evidenceof an excisingtechniquewas found. In
fact, most sherds were clearly incised, not excised.
Therefore,broad-lineincision is a more appropriate
term.
Ceramic terminology in African archaeology is
obviously not sufficientlyexplicit. The applicationof
termslike type and ware has been very loose and has
fostered enormous misconceptions. Type and ware
5
A.D. 455?95
A.D. 440?85
A.D.
540?85
A.D.
180?120
A.D. 570? 110
A.D.
A.D.
A.D.
A.D.
320?150
410?95
530?120
540?95
are shorthandmethods for referringto a large body
of data that would otherwisetake up a considerable
amount of space in description.However, the inclusion of Kalambo, Kalundu, and Gokomere under
ChannelWare (or dividing all of the early Iron Age
into Dimple-baseor ChannelWaregroups)is exceedingly naive and a gross misinterpretationof the data.
These units are so large they obscure the range of
variation within the early Iron Age and prevent us
from making meaningful comparisons. Phillipson
(1968a)has recentlyavoidedthis pitfall by regrouping
the ceramicsin Zambia on a regional basis, and has
shown the necessityof restricting'channelpottery'to
Gokomere, Ziwa, and Dambwa-like traditions. His
outline will be basicallyfollowed here.
Many societies often conceive of their ceramic
inventory in terms of functions. This is especially
true in the area we are concerned with here. When
Tonga villagerswere asked to identifyvariouspottery
L
Dimple-bose
2. Kwale2
3. Kalambo
4. Kapwirimbwe
5. Kalundu
6. Gokomere
4A
\
0
690
miles
(666
1200
Fig. 2. Distributionof six early Iron Age ceramictraditions.
6
fragmentsduringa recentexcavationin Zambia,they
invariably answered with functional terms: e.g.
drinking-cup,storage-jar,etc. Their ability to discriminate function was based on the identification
of the vessel shape. The thought process is the significantpoint here, of course,not whetherthe workmen
were actually correct in their identification.Fagan
(pers. comm.) has reportedthe same kind of answer
from Ila near Basanga, Zambia, as has Summers
(pers. comm.) for Shona at Inyanga and Zimbabwe,
and Willet (1967) for Yoruba in West Africa. In
these cases shapes formed a meaningful basis for
categorization.
Therefore,each ceramicgroup will be presentedin
termsof its rangeof shapesand the decorationwill be
discussed within this framework.However, because
of the nature of the data, it has not been possible to
describethe total ceramic categories for any group,
and undoubtedlysome designand shapecombinations
have been over-representedand others omitted.
Enough of the materialhas been published,however,
to make it possible for others to duplicatethis study.
Table I gives a complete list of radiocarbondates
for the six groups. Figures 2-7 attempt to present a
reasonable amount of variation without losing the
'feel' of each style. The column of reconstructed
shapes simply provides one example of a particular
class; dotted outlines indicate educated guesses. The
shapesshould not be taken as a type standard,but as
an aid to understandingdecorationplacementand the
generalmorphologyof the vessel. Size is shown more
or less relatively. Table II briefly summarizes the
results of the comparison.
Key to figures3-8
B
Bevel
BLI
Broad-lineincision
DS
Dentate stamp
F
Flute
FLI
Fine-lineincision
FRCP False-reliefchevronpunctates
GB
Graphiteburnish
IN
Incision
P
Punctate
SB
Stab
Kwale
Soper (1967a; 1967b; 1968) has recently described
a new early Iron Age tradition in south-east Kenya
and north-east Tanzania; he (1968) has also drawn
attentionto similaritiesbetweenKwale, Dimple-base,
and pottery from Sandawelanddescribedby Smolla
(1957).
Figure 4 shows a few of the major shapes and
decorationpatterns from Kwale. The small number
of shapes only reflectsthe limitedpublicationsof this
recent discovery. The decoration patterns on the
first jar form are reconstructions.
Figure 4 emphasizes such combinations as (1)
globularjars with multipleflutes, a line of DS below
the last flute, flutes boundedby DS on the neck, and
IN with pendant motifs in the shoulder area, and
(2) bowls with multipleflutesor bevelledlips bordered
on the bottom by DS, and a small band with pendant
motifs a short space below. Concave bases are common. Most rims were redrawnfrom Soper (1967a;
1967b).
Dimple-basePottery
M. LeakeyoriginallydescribedDimple-basepottery
from central Kavirondo, Kenya, collected by Owens
(Leakey, Owen, & Leakey, 1948). The pottery was
compared to fragments recovered earlier from
Nsongezi rockshelterby O'Brien(1939). However, a
later excavation by Pearce and Posnansky (1963)
suggestedthe Nsongezi assemblagewas considerably
later than the pottery describedby Leakey (Posnansky, 1967a).
Dimple-base pottery has also been reported by
Hiernaux and Maquet (1957; 1960) from Ruanda/
Urundi and Kivu, by Nenquin (1959) for Kasai,
Congo, by Posnansky(1961a)from Uganda,by Fagan
& Lofgren (1966) on Lolui Island, and by Chapman
(1967) from Kantysore Island in the Kagera River.
A comparison of Dimple-base pottery with Kwale
has recently been made by Soper (1968). Figure 3
illustratestypical attributesof Dimple-baseceramics
and emphasizes such combinations as (1) globular
jars with multiple bevelled and fluted lips, crosshatchedrims, and large bands of BLI in the shoulder
area, and (2) bowls with multiplefluted and bevelled
lips, IN belowthe rim,andcurvilineararealdecoration
in the shoulderarea. Concave bases are common on
all forms. Most of the rimswereredrawnfromLeakey,
Owen, & Leakey(1948).
KalamboFalls
The full excavation report for Kalambo Falls has
not yet been published although Fagan (1966a) has
brieflydescribedthe earlyIron Age potteryfrom two
of the sites. Figure 5 applies only to the sites around
the southernend of Lake Tanganika(fig. 2) that are
included in Phillipson's (1968a) Kalambo Falls
group. It illustrates(1) jars with IN or bevelledlips,
undecoratedrims, and necks filled with bands of IN
and DS, and (2) bowls with bevelledand fluted rims
and occasionallybands of DS and IN below the rim.
The rims were taken from unpublished drawings
kindly suppliedby Prof. J. Desmond Clark.
7
Kapwirimbwe
The earliest Iron Age on the Lusaka Plateau is
representedby collections from Leopard's Hill and
Kapwirimbwe(Phillipson, 1968a). Figure 6 shows
(1) a jar form with IN rims, (2) a jar form with an
internallythickenedlip, undecoratedrim, and necked
filled with multiple bands often in a herringbone
pattern,and (3) bowls with thickenedlips, GB lip and
rims, and a band of IN boundedby P below the rim.
Several rims were redrawnfrom Phillipson (1968b).
KanduluWare
The earliest Iron Age on the Batoka plateau has
been called the Kalundugroup by Phillipson(1968a).
Reconstructed
Shapes
Variation
>
< |>
of Lip and Rim
1
}
Decoration
)
)4
W
Combinations
From Actuol Vessels
N
IN
N
1
I_
IN
_
_
'
_L
L
IS
t
v
9
IN
B~~~LI
g
t
-
BLI
INL
\
\
81\
'-00:
N
INI
(K
'
|
ti
.__-~~~~~
IN
BLI
_
DN
~
.
Fig. 3. Attributesof Dimple Based pottery.
8
._
_B
LI
__
_ _
Reconstructed
Shapes
Variation of Lip and Rim
Decoration
Combinations
F
~~~~~~F
~
F
---- --- - -Os
---
S
-
F
-----OS
ODS
CH
,
- -'---------
F~~~~~~~
F
F_
I
_____,
O
DS
F
F
11WDS
^'
DS
.
A,
F
>
'
D
F
F
~~~~~~
Fig.
4. Attributes--of Kwae
-ptte
S
D
-
R
F
F
F
S
-
os~~~
y.
-D
F
__F...D
which she called Class A. Gokomerepotteryhas also
been reported by Robinson from Zimbabwe (Summers, Robinson & Whitty, 1961), Mabveni (1961a)
and Gokomere (1963). Gokomere sherds have also
been found mixed with Bambata Ware in several
rock-shelters in south-west Rhodesia (Laidler,
1938; Schofield,1940; Robinson, 1964; 1966b).
Figure 8 illustrates(1) jars with oblique DS on the
rim, a line of P followed by BLI in the neck, and a
line of P on the shoulder,and (2) bowls with bands
of DS below GB lip andrims.The drawingsweremade
fromcollectionsin the National Museumsof Southern
Rhodesia,Bulawayo.
The precedingdescriptionsand summaryof the six
early Iron Age ceramic traditionssuggest early Iron
Age pottery was not the same throughoutEast and
Gokomere
southernAfrica when it first appeared.If the radioSchofield (et al., 1940) first described Gokomere carbon dates are accurate, then several regionally
pottery from a rock-shelterat Gokomere Mission, distinct ceramic traditions existed by the fourth
Rhodesia. However, Caton-Thompson (1931) had century A.D. (cf. Posnansky, 1961; Hiernaux, 1968;
earlier uncovered the same tradition at Zimbabwe, Phillipson, 1968a).
It was originally described by Inskeep (1962) from
Kalunduand laterby Fagan (1967a).Anotherdeposit
has recently been uncovered at Gundu (Fagan &
Huffman, 1967). Kalundu Ware was originally
thought to be an early phase of the Kalomo culture,
but it is undoubtedlya distinctcomponentin Kalomo
culture mounds (Phillipson, 1968a).
Figure 7 is largelytaken from unpublishedmaterial
from Gundu, but it also applies to Kalundu. It
emphasizes(1) a jar form with incised rims and IN
and P on the shoulder,(2) a pot form with an internally thickenedlip, GB rim, and a broad band of IN
bounded by P in the neck, and (3) a bowl form with
GB on the inside and outside, and a band of IN and
P a short space below internallythickenedlips.
9
TABLE
II
SUMMARY OF THE COMPARISON
DB KW KF KAP KL GK
Lip
decoration
Rim
decoration
Rim form
Lowerrim/
Upper neck
decoration
JARS
Neck
decoration
Shoulder
decoration
MultipleF or B
BLI
Deep IN
DS
x
x
P
F
DS-single horizontalline
oblique
IN- cross-hatch
oblique lines
horizontallines
interlockingparallellines
GB
x
x
x
x
x
x
x
x
x
x
x
x
P
x
IN- single line
cross-hatch
BLI
FRCP
Small band of IN & P
x
x
x
BLI or F- parallel
curvilinear
with pendantIN
IN- cross-hatch
herringbone
oblique lines with pendantP
interlockingparallellines
DS- horizontalline
oblique
P
Multiplebands
FRCP
x
x
x
x
x
x
x
x
x I
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
10
x
x
P
Concave
x
x
x
x
x
x
x
x
x
?
x
single DS line
Base form
x
x
x
Recurved
Largeband of BLI-parallel
curvilinear
Smallband- IN & P
IN with pendantP
single IN line
FRCP
x
x
?
T1ABLE
11
DB KW KF KAP KL GK
Lip
decoration
BLI
IN- cross-hatch
x
x
GB
Rim
decoration
Lip/Rim form
Lower rim/
Upper neck
decoration
POTS
Neck
decoration
Shoulder
decoration
Base form
BLI
IN- cross-hatch
horizontallines
GB
x
x
x
x
x
x
x
x
x
x
x
x
x
Thickened
P
SB
IN- cross-hatch
small band with P
interlockingparallellines
herringbone
FRCP
DS
BLI
BLI & P
Smallband- IN & SB
IN & P
IN & DS
interlockingIN
cross-hatchIN
DS & DS
Multiplebands
Hatchedmeanders
x
x
Areal motifs-P & SB
curvilinearBLI with P
Smallband- BLI
cross-hatchIN
IN & P
IN & FRCP
Single line of DS
GB
x
x
x
x
Concave
x
11
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
?
x
x
x
x
TABLE Il]
DB KW KF KAP KL GK
Inside
GB
MultipleF or B
x
Lip
P
x
decoration
Rim
decoration
Decoration in
shoulderarea
All bowls base form
x
x
x
x
x
x
X
IN- cross-hatch
DS- oblique
BLI
GB
x
Thickenedor pulled over, outside
x
x
Smallband-cross-hatchIN
BLI
IN & SB
P & IN
DS & DS
DS with pendant IN & P
IN & FRCP
herringboneIN
x
x
x
x
x
x
_
__
I
x
x
x
x
x
x I x Ix
x
x
x
x
x
x
x
x
x
x
x
P
Shoulderedbowls
x
x
x
Thickened,inside
Necked bowls
Neck decoration
x
x
BLI
IN- oblique
GB
Lip/Rim form
Lowerrim/
Upper shoulder
decoration
?
x
Band of oblique DS & DS or DS & IN
multiple bands
x
>
x
Band of cross-hatchIN
BLI- curvilinear
Band of DS or IN with pendantIN & P
DS- oblique
interlockingparallellines
P
FRCP
Largeband of interlockingIN & P
Hatchedmeanders
Hatchedmeanders
Main decoration-aboveshoulder
on shoulder
below shoulder
x
x
x
x
x
Concave
x
x
12
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Reconstructed Shapes
Decoration Combinations FromActual Vessels
Variationof Lip and Rim
LI
iOLI
U U
U
U
,/t-uq,
IN
DS
~~___
9L
bLI
.
()SXXXIIIN
f
L
t
w
I
IN
/OsLa8t
)1i
J/I
I
J
~~~BLI~~
~~~~~
wBI
IN
-
B
-
<<
--BLI
B~~LI
IN
BLI
,A
-ZBLIKN
IN
0
////BLI
-
BLI
9LI
BLI
c /
'?rDS
IN
~ ~ ~~
~
|.,
\X~~~~~~~~~_-ZzX~
LI
S-
D(
LI
--
0BLI
r
a~BLI
IN
.BLI.BLI.BLI
-n-u--u--~~~~~~~~~~~~~~~~~~~~~IN
IN
/
u"',"
"D
IN
INXXIN
~
VO~(I
\II
__XXX
^
till'.
IIN
t
/2
|
=
\~~~~~~~~~~~~~~~~~L
V8L~~~~~~~~~~*
\
wvv
=\BLI
=
;\
IN
p"
t7v~tIpN
OL.
t
IN
LI
I I
I !!
X~~~~~~~~~~~~>XXX><IN
S
%
^
.
I
BL
aIN
R53
.DS
IN
AA
A
S
B~LI
IN
II\N
I~~~~~~~~~~~~~~~~~~~~~~~S
IN(LI*XV
Fig.
5.
Attributes of Kalambo
/////IN
.LI
BI
pottery.~~~~~~BL
Fig. 5./AttibutesofKaamboptte
(i) in the manner of firing in the Gokomere and
Kalundu traditions-bowls are fired at a lower temperature and/or in more of a reducing atmosphere
than cooking-jars; (ii) in the selection of paste in
Kwale pottery-bowls have finer temperingmaterial
than jars (Soper, 1967); and (iii) in Gokomere,
Kalundu, and Kapwirimbwebowls were commonly
Still, early Iron Age pottery is quite similar and was
obviously made by closely related peoples, and it
forms a meaningful unit distinct from other industries
in their area. For instance: (a) All six groups contain
numerous shape categories with a clear division
between most jars and bowls. (b) A technical difference, undoubtedly due to function, can be seen:
13
Reconstructed
Variation of Lip and Rim
Shapes
Decoration
IN
IN
) 1
\
i
~
~
\
l
~
~~~
w
\
KF
~ ~ ~ ~ ~~~
g
n
o
c
~
NI
From Actual Vessels
Combinations
p
N
IN1
~
iN
I
~
C
P
>
~~~~~~~~~~~~~~~~~
IN
_~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
_ __
F
ININ
IN
/~
/~~~~~~~~~~~~~~~ \
,.
|
~ ~~~
X
~~~~Fig
\
>
awrmbeptey
6.Atiueso
Fig.~~~~~~~~~~~
nXn
6. Attribts
~ ~~~~~~~~
s
of' Kapwirmw
coated with graphite but cooking-jars were not
because repeated use in a fire would burn off the
graphite. (c) Almost every group has a composite
vessel class, and its own version of a globular pot
with a concave neck and textured lip or rim (fluted,
bevelled, incised, or stamped). And (d) in all six groups
there is an overwhelming use of bordered, decorative
bands filled by texturing. Ceramics in the East
African Neolithic, however, are basically limited to
deep beaker-bowls (Leakey, 1943; 1945; Clark,
1967b) while bag-shaped vessels with roulette decoration on the neck are connected with pastoralists
during the second millennium A.D. (Posnansky,
/
\1
/
i
pottery
1967a). Prehistoric Hottentot pottery is also mostly
in the form of bagged-shaped vessels with or without
handles (Sydow, 1967; Rudner, 1968).
The Time Factor
According to the DB/CW hypothesis, Dimple-base
pottery degenerated at Kalambo Falls and then
transformed into the Gokomere pottery of Rhodesia.
If the radiocarbon dates are inaccurate, their interpretation is ostensibly compatible with a unilinear
movement of people. However, to verify such a
hypothesis, Dimple-base Ware must look more and
more like Gokomere pottery as time passes (or the
14
Reconstructed
Shapes
of Lip and Rim
Variation
Decoration
JI11~~~~~~~~~~N
s
l
\
\
/
jl,,,,,,.,,j
,P
\'E4
71I 1N
I
,
J\\\
\\\\P
IN
1t/
))
,z~lilililil
N
P
)II)
1
INN
f 5f II N
lirlz=
\
From Actual Vessels
~
I
J~~~~~~~~~~~~IIIIIL
Iff/p(~
Combinations
_
p
aG
i
'NN
N
\\Ip
_.iN
IN
-
\ ip
{I
I Ii
\
Il-P
Iip
I
IDL
|
_
"
{
N
^~~~~~~~l~
t
t
4
t
\
St a
\
X
N
/
8L
GBBLI
p
ap
~~~~~I
GS
GG
GB
GB
lG
........~~~~~~~~~~~~~
.
Fig. 7. Attributes of Kalundu pottery.
15
BLIIN
N
*GSt
GB
Ei:3=~
S
IN
GB
Gp
\
lort
,.--s
GB
Reconstructed
/
Variation
Shapes
\
}
of Lip and Rim
)
}
Combinotions
Decoration
DS
)
From Actual Vessels
)~~~~~~~~~~~~~~~~DS
}
'
,,
D5
f
DS
GBS
- - -
DS
-DS
1)~)
\\~~
_
~
\~~
-
~~~
~ ~ 1~
-----5
\70
Fig.
F
-.rE-
I N
Os'
t
I
*
-
_lB
--
.'
n
o-, m-
~N
,s
P a o b
--
of Gokmeeoter
a
b
P
~~~~~~~~~~~~~~~~~~~~~~~~~~~
D ___
s
16~~~~~~~~~~~~~~~~~~~~~~~~~
DR
D.
_S
__n-
-BL
~.
DS
:-DS
Ann
_L
I
S
' D5
- -
--DS
reverse)(see fig. 1). The availableevidence, however,
indicates several sets of divergingceramic traditions,
when we considerthe assemblagesthroughtime.
At Nsongezi rock-shelter,Dimple-basepottery was
found to overliea Wilton hearthdated to the eleventh
century A.D. (Pearce & Posnansky, 1963; Posnansky,
1967a). Comparedto classic Dimple-basethe pottery
had undergonesome modificationsbut it still looked
like Dimple-base. Dimple bases still persisted and
basic decorative motifs had their antecedentsin the
earlier phase (Posnansky, 1967a). Even if this case
was unreliable(only one radiocarbondate),Posnansky
(1967a) has described the same situation on Lolui
Island.The evidenceis not too clearbut it suggeststhe
changes were centred around themes alreadypresent
in earlierDimple-baseassemblages,and that Dimplebase pottery at A.D. 1000 did not look any more like
Gokomerepotterythan it did at A.D. 300.
The same phenomenon recurs at Kalambo where
Fagan (1966a) found pottery very similar to the
earliest Iron Age in the area but dating to the sixteenth century A.D. And the same kind of change is
seen in the Lusaka region (Phillipson, 1968a). The
pottery from TwickenhamRoad is strikinglysimilar
to the Kapwirimbwegroup, 500 to 600 years earlier.
Not only are the shape categories almost identical,
decorativepatternsmimic each other on corresponding shapes. There are two dates for the lower levels
of Twickenham (Gx-1329 940?110 A.D., Gx-0662
1055?110 A.D.) and four for the early phase of
Kapwirimbwe(Table 1). Twickenhampottery does
not look any more like Dimple-base or Gokomere
potterythan did the earlierKapwirimbwephase.
The change in Gokomerepottery over time further
strengthensthe argument against the DB/CW continuum. Leopard's Kopje Phase I is undoubtedly
derived from the 600-year earlier Gokomere phase
(Robinson, 1966a); and parallels the change from
Ziwa, to Ziwa2 (Summers, 1958) in Mashonaland
and from Kumadzulo to Dambwa in the Victoria
Falls region (Vogel, pers. comm.). The VictoriaFalls
IronAge is one of the best datedsequencesin southern
Africa. Thereare six radiocarbondates from Kumadzulo (Vogel, pers. comm.) ranging from the fifth to
seventh centuries A.D., and seven from Dambwa
which range from the seventh to the ninth century
A.D. (Fagan, 1966b; 1967; 1969) which resolves any
doubt about their chronological ordering. Neither
Leopard'sKopje 1, Ziwa2,or Dambwalook any more
like Dimple-basethan did their ancestors.
The evidence for stylistic change over time within
early Iron Age ceramics is particularly strong in
Rhodesia and southernZambia, and very suggestive
for the other groups. Clearly,the transitionalphases
that are expectedwith the rapidmovementof a single
style are lacking. No Gokomere-likepottery has ever
been reported overlying a Dimple-base deposit in
East Africa, nor has Dimple-basepottery been found
stratigraphicallyover a Gokomere component in
southern Africa. The later phases of early Iron Age
pottery look less like each other than the earlier
17
phases, instead of merging into each other. Consequently, Dimple-base pottery could not have
degenerated into Kalambo pottery or transformed
into Gokomere. Neither could Gokomere pottery
have evolved into the Kalundu, Kapwirimbwe,
Kalambo,or Kwale or Dimple-basetraditions.
The comparison of the six ceramic groups has
demonstratedconsiderablediversitywithin the early
Iron Age. If the radiocarbondates are accurate,there
is no evidence for a rapid movement of a group of
people in a single direction. Should the dates be
misleading, however, the continued distinctiveness
of the traditions over at least 600 years still argues
against a unilinear movement. Therefore, we are
forced to conclude that the DB/CW continuum
hypothesisis untenableno matter how the dates are
interpreted.
The Alternative
Despite representingseparate entities, the early
Iron Age ceramicsdo look similarwhereverthey are
found, although they diverge more as time passes.
Therefore,one would expect the pottery to look even
more alike further back in time since it would be
nearerto a common source.
Many scholarshave come to the same conclusion.
Posnansky (1961; 1968) first suggested a common
centre of origin in 1961. Oliver(1966) offereda north
and south movementfrom the Zambezi as an alternative to the DB/CW hypothesis.And the grouping
of regional variantsin Zambialed Phillipson(1968a)
to postulate a north-west origin for the Zambian
early Iron Age. The evidence clearly indicates the
early Iron Age radiated from a common source,
metaphorically,somewhat like pellets bursting from
a shotgun (fig. 9).
This common origin and subsequentdiversification
is similar to the Co-Traditionmodel developed in
South Americato handle a seriesof traditionsarising
from one ancestralphase (Rouse, 1957).In American
archaeologya tradition is a series of phases while a
phase equals the smallest unit that includes all the
social and material aspects possessed by the people
in its components(Rouse, 1957).
The Co-Traditionmodel is amenable to the linguisticas well as the archaeologicalevidence.Guthrie's
(1962) research is now fairly well known and his
historical hypothesis about a proto-Bantu centre
south of the tropicalforestis familiarto most scholars.
Guthrie'sown data led him to postulate that Bantu
Africa was populatedfrom an eliptical area south of
the tropical forest where proto-Bantu was spoken.
Not all linguists agree on the place, time, or significance of Guthrie's nuclear area (called PB-X), but
most Bantuists agree on a secondary dispersion of
Bantu languages called Eastern Bantu or PB-B by
Guthrie that is a later split than Western Bantu
or PB-A (Ehret & Vansina, pers. comm.). If the
sudden appearance of the early Iron Age in East
and southern Africa can be connected to any linguistic movement,it would most likely be the spread
split, then we would expect PB-B pottery to contain
more specificattributesof the north and south tradition than from a later tradition like Kalundu or
KWALE Kapwirimbwe.The potterycould easily have changed
somewhat before the Kalambo, Kapwirimbwe,and
Kalundutraditionsleft the source area.
The southern tradition raises some interesting
problems.At the presentit is the only style extending
into the Republic of South Africa from Zambia and
Rhodesia; if no other early tradition is found, then
KALAMBO the southerntraditionmust be connectedwith Southeast Bantu speakers. While the earliest Iron Age
CO-TRADITION
people are often assumed to have brought herds of
SOURCE
domestic animals with them, the makers of the
AREA
southerntraditiondo not appearto have emphasized
cattle in theireconomy. So far no conclusiveevidence
KAPWIRIMBWE has been producedin the form of actual remainsor
figurines from any Gokomere, or Ziwa phase site.
Other domestic animals are known to have been
present,however,for a goat horn core was uncovered
KALUNDU
by Robinson at Gokomere and caprine teeth were
found at Mabveni(1961a). Even the word for cow in
South-eastBantudoes not havethe normalBanturoot.
The South-east Bantu root *-komo comes directly
from Khoisan and only indirectly from Central
Sudanic,while the normalBantu root *-qombecomes
directly from Central Sudanic (Ehret, pers. comm.).
Thereis no evidenceof an emphasison cattle herding
in south-westRhodesia until the appearanceof LeoGOKOMERE
pard's Kopje II. This culture is probablynot related
Fig. 9. Radiation from a common centre: a co-tra- to the same ceramictraditionas Kumadzulo,Gokomere, or Ziwa (Huffman, 1968), but appearsto be a
dition.
later intrusionconnectedto the same influxof people
of Eastern Bantu, or PB-B (Ehret & Vansina, pers. whose remains were uncovered at Bambandyanalo
comm.).
(Fouche, 1937; Gardner, 1963) and at Zimbabwe
If the early Iron Age ceramicswere made by Bantu duringPeriod II (Caton-Thompson,1931; Summers,
speakers, then one should be able to reconstruct Robinson, & Whitty, 1961). This movement might
PB-B, proto-Bantupottery,and eventuallypre-Bantu, representthe firstappearanceof the Shona (Huffman,
in the same manner as an early language is recon- 1968).
structed.The generaltraits that were sharedthroughWe have only been reviewingthe evidencefor the
out the earlyIron Age most likely representattributes latter part of the Bantu expansion. It is not at all
of PB-B pottery, and one localized region should certainwhetherpottery,iron, a semi-sedentaryvillage
contain the majorityof them.
pattern,and food productionfilteredinto the Co-TraA close examinationof the early styles reveals a dition spheretogetheror from differentdirectionsand
north/south split in the ceramics,most likely repre- at differenttimes. Therefore,more than one Co-Tradisenting two separate, major traditions within the tion conceivablycould have been present,although I
Co-Tradition(fig. 10). The northernbranch is made would not favour this interpretation.The data more
up of Dimple-base,Kwale, and Sandawelandpottery. strongly suggests all early Iron Age pottery in East
The southernbranch consists of Kumadzulo, Goko- and southernAfricahas its originin the same Co-Tramere, Ziwa, the site of Happy Rest (Matokoma) in dition.
the north Transvaal(De Vaal, 1943), and probably
The study of the Iron Age is still in its infancy,
Castle Peak in Swaziland(see Beaumont in Fagan, especially its contribution to Bantu origins. The
1967). When comparingtwo words in linguistics,the correlationof Bantu speakerswith the early Iron Age
wider spreadof the two is the oldest, all things being is a plausible assumption,even though the evidence
equal (Sapir, 1916). Following this analogy the wide no longersupportsthe DB/CW hypothesis.Hopefully
range of these two traditions when compared to further constructivecriticism will be stimulated by
Kalunduor Kapwirimbwesuggests a very early split the suggestion of radial dispersionfrom a common
within the Co-Traditionsource area and a possibly centre. The Co-Traditionmodel is merely a working
earlymovementout of the centre.Of course all things hypothesisto help organizenew data into a meaningful pattern,If futureevidenceis unexplainableby this
are not equal and thereare many possibilities.
But if these two traditionsdo representa very early model, then the model can be modifiedor discarded.
DIMPLE-BASE
18
0
600
1200
miles
Fig. 10. Branchesof the co-tradition.
19
Summaryof Conclusions
Evidence from physical anthropology, linguistics,
and archaeology plausibly suggest Bantu-speakers
were responsiblefor the early Iron Age in East and
southernAfrica.Six earlyIron Age ceramictraditions
wereexaminedto test the Dimple-base/ChannelWare
continuum hypothesis. This examination indicates
the Gokomere, Kalundu, Kapwirimbwe, and
Kalambo traditions could not have originatedfrom
Dimple-base pottery as implied by the hypothesis.
Althoughmany attributesare sharedamong the early
Iron Age traditions,particularcombinationsare very
different.The Dimple-base/ChannelWare hypothesis
also states that early Iron Age peoples migratedfrom
the Great Lakes region of East Africa south into
Zambia and Rhodesia. The evidencepresentedhere,
however,agrees with the recent work of a few other
scholars and suggests that early Iron Age people
radiated outward in several directionsfrom a common source area. This radial dispersioncan be conceptualizedas a Co-Traditionand might possibly be
connectedto Guthrie'sPB-B.
Acknowledgements
This paperhas benefitedgreatlyfrom commentsby
Dr. Charles Keller, Mr. Patrick Munson and Mrs.
Cheryl Munson, and Dr. Arthur Rohn from the
University of Illinois and Dr. Brian Fagan at the
Universityof California,Santa Barbara.Miss Cherie
A. Bohat did the excellent drafting. I owe special
thanks to Mr. Roger Summers of the National
Museums of Southern Rhodesia for the facilities at
the Bulawayo Museum and for his continual assistance.I am also deeplygratefulto Mr. Keith Robinson
for making available his collections and for many
valuable discussionswith him. Mr. David Phillipson
kindly allowed me to view pottery from Kalambo
Falls, Kapwirimbwe, and Twickenham, as did
Mr. Joseph Vogel for material in the National
Museum of Zambia. Discussions with Mr. Robert
Soperhave tremendouslyhelpedmy understandingof
Kwale and Dimple-basepottery. I am most grateful
to Dr. J. Desmond Clark, University of California,
Berkeley,for the drawingsof Kalambo pottery. This
paper was considerablyrevised after review copies
were sent to Dr. Charles Alexander, Chris Ehret,
Roland Oliver, David Phillipson, Larry Robins,
Keith Robinson, Robert Soper, Roger Summers,
Jan Vansina,JosephVogel, and Frank Willett. I owe
special thanks for their commentsand in many cases
new information.Any errorsin this paper, of course,
are the sole responsibilityof the author.
CLARK,J. D. 1967a.A Recordof earlyagricultureand
metallurgyin Africa from archaeologicalsources.
In Gabel & Bennett, eds., Reconstructing African
culturehistory.Boston: Boston UniversityPress.
CLARK,J. D. 1967b.The Problemof Neolithic culture
in sub-SaharanAfrica. In Bishop & Clark, eds.,
Background to
evolution in Africa.
Chicago:
Universityof ChicagoPress.
CLARK,J. D. & FAGAN,B. M. 1965. Charcoals,sands,
and Channel Decorated pottery from northern
Rhodesia. American Anthropologist. 67 (2): 354-371.
COLE,S. 1963. The Prehistoryof East Africa. New
York: MentorBooks.
DANIELS,S. G. H. 1965. The statisticaldetermination
of potterytypes from the recentlyexcavatedsite of
Dambwa near Livingstone,Northern Rhodesia. In
Snowball, G. J., ed., Science and medicine in
central Africa. London: Pergamon Press.
DE VAAL,J. B. 1943. 'n SoutpansbergseZimbabwe.
S. Afr. J. Sci. 40: 303-322.
FAGAN,B. M. 1964. The Greefswaldsequence:Bam-
bandyanaloand Mapungubwe.J. Afr. Hist. 5 (3):
337-60.
FAGAN, B. M. 1965. Southern Africa during the Iron
Age. New York: FrederickA. Praeger.
FAGAN, B. M. 1966a. Early Iron Age pottery in
easternand southernAfrica. Azania.1: 101-109.
FAGAN, B. M. 1966b. Radiocarbon dates for sub-
SaharanAfrica-IV. J. Afr. Hist. 7 (3): 495-605.
FAGAN,B. M. 1967a. Iron Age cultures in Zambia-
vol. I. London: Chatto & Windus.
FAGAN, B. M. 1967b. Radiocarbon dates for sub-
SaharanAfrica-V. J. Afr. Hist. 8 (3): 513-527.
FAGAN, B. M. 1969. Radiocarbon dates for sub-
SaharanAfrica-VI. J. Afr. Hist. 10 (1): 149-169.
FAGAN,B. M. & HUFFMAN,
T. N. 1967.Excavationsat
Gundu and Ndondi-1967.
Archaeologia Zambiania. 2: 3-5.
FAGAN,B. M. & LOFGREN,L. 1966. Archaeological
Reconnaissance on the Ssese Islands, Uganda.
Uganda J. 30 (1): 81-86.
FOUCHE, L. 1937. Mapungubwe, ancient Bantut
civilization on the Limpopo. Cambridge: Cambridge
UniversityPress.
vol. II. Pretoria.
GARDNER,G. A. 1963.Mapungubwe,
J. H. 1949. Studies in African linguistic
GREENBURG,
classification. Southwestern J. of Anthropology.
5: 309-317.
H. 1955. Studies in African linguistic
classification. New Haven.
GUrHRIE,M. 1962. Some developmentsin the prehistory of the Bantu languages. J. Afr. Hist. 3 (2):
GREENBURG, J.
273-282.
HIERNAUX,J. & MAQUET,E. 1957. Cultures prehistorique de l'age des metaux au Ruanda-Urundi
et au kivu (Congo Belge). Bulletin des Seances de
l'Academie Royale des Sciences Coloniales. 2 (6):
1126-1 149.
HIERNAUX,J. & MAQUET,E. 1960. Cultures prehistorique de l'age des metaux au Ruanda-Urundi
et au kivu (Congo Belge). Part II. Memoires de
Bibliography
G. 1931. Zimbabwe Culture.
CATON-THOMPSON,
Oxford.
CHAPMAN,S. 1967. Kantsyore Island. Azania. 2:
175-191.
CLARK,J. D. 1962. The spread of food production in
sub-Saharan Africa. J. Afr. Hist. 3 (2): 211-228.
20
l'Academie Royale des Sciences d'Outre Mer. 10
(2): 5-102.
HIERNAUX, J. 1968. Bantu expansion: the evidence
from physical anthropology confronted with
linguisticand archaeologicalevidence.J. Afr. Hist.
9 (4): 505-515.
HUFFMAN, T. N. 1968. The Place of the Leopard's
Kopjeculturein the Iron Age sequenceof Rhodesia.
Read at African Studies Association Annual
Meeting, 1968, Los Angeles.
INSKEEP,R. R. 1962. Some Iron Age sites in Northern
Rhodesia. S. Afr. Archaeol.Bull. 17 (67): 136-80.
JOHNSTON, Sir H. H. 1913. A Survey of the ethnographyof Africa: And the formerracial and tribal
migrations in that continent. J. roy. Anthr. Inst.
43: 375-414.
LAIDLER, P. W. 1938. South African native ceramics;
their characteristicsand classification. Trans. roy.
Soc. S. Afr. 26: 93-171.
LEAKEY, M. D., OWEN, W. E. & LEAKEY, L. S. B.
1948.Dimple-basepotteryfrom centralKavirondo.
CoryndonMem. Museum.Occ. paper 2.
MARCUS, H. G. 1967. A historian's view of Bantu
origins.Paperpresentedto AfricanStudiesAssociation Annual Conference.New York.
MURDOCK, G. P. 1959. Africa, its peoples and their
culture.New York: McGraw Hill.
NENQUIN, J. 1959. Dimple-based pots from KasaiBelgian Congo. Man. 242: 153-154.
NENQUIN J. 1963. Notes on some early pottery cultures in north Katanga.J. Afr. Hist. 4 (1): 19-32.
O'BRIEN, T. P. 1939. The prehistoryof the Uganda
protectorate.Cambridge:CambridgePress.
OLIVER, R. 1966. The problem of the Bantu expansion. J. Afr. Hist. 7: 361-376.
PEARCE,S. W. & POSNANSKY,M. 1963.The re-excavation of Nsongezi shelter,Ankole. UgandaJ. 27 (1):
85-94.
PHILLIPSON, D. W. 1968a. The early Iron Age in
Zambia: regional variants and some tentative
conclusions.J. Afr. Hist. 9 (2): 191-211.
PHILLIPSON, D. W. 1968b. Z5: Finds from Kapwirimbwe, Lusaka. InventoriaArchaeologicaAfricana.
POSNANSKY, M. 1961a. Dimple-base pottery from
Uganda. Man. 61 (art. 168).
POSNANSKY, M. 1961b.Iron Age in East and central
Africa-points of comparison. S. Afr. Archaeol.
Bull. 16: 134-136.
POSNANSKY, M. 1967a. The Iron Age in East Africa.
In Bishop & Clark, eds., Backgroundto Human
Evolution.Chicago: Chicago Press.
POSNANSKY, M. 1967b. Review of Fagan's Southern
Africa during the Iron Age. In Archaeology.20
(1): 69-71.
POSNANSKY, M. 1968. Bantu Genesis-archaeological
reflections. J. Afr. Hist. 9 (1): 1-1 1.
ROBINSON, K. R. 1961a. An early Iron Age site from
the Chibi District, Southern Rhodesia. S. Afr.
Archaeol.Bull. 16 (63): 75-102.
ROBINSON, K. R. 1961b. A note on hollow-based
pottery from Southern Rhodesia. Man. 61 (art.
105): 88.
21
K. R. 1961c. Rhodesian School Boys
ExplorationSociety:BuffaloBendReport.Salisbury:
ROBINSON,
Govt. Printer.
K. R. 1963. Further excavations in the
Iron Age deposits at the Tunnel Site, Gokomere
Hill, Southern Rhodesia. S. Afr. Archaeol. Bull.
18 (72): 155-71.
ROBINSON, K. R. 1964. Domosanga Rock Shelter,
MtetengweRiver Beit Bridge, SouthernRhodesia,
excavationresults.Arnoldia.1: 7.
ROBINSON, K. R. 1966a.The Leopard'sKopjeculture;
Its position in the Iron Age of SouthernRhodesia.
S. Afr. Archaeol.Bull. 21 (81): 5-51.
ROBINSON,
K. R. 1966b. Bambata Ware; Its position
ROBINSON,
in the Rhodesian Iron Age in the light of recent
evidence. S. Afr. Archaeol.Bull. 21 (82): 81-85.
ROUSE,I. 1957. Culture Area and Co-Tradition.
SouthwesternJ. of Anthropology.12: 133.
RUDNER, J. 1968. Strandloperpotteryfrom South and
Southwest Africa. Ann. S. Afr. Mus. 49 (2).
SAPIR,
E. 1916. Time perspective in aboriginal
American culture, A study in method. Canada
Geol.Survey,Memoir.90.
F. 1941. A report on the pottery from
BambataCave. S. Afr.J. Sci. 37: 361-372.
SCHOFIELD,J. F., GARDNER, G. A., & WELLS. 1940.
Recent archaeology of Gokomere, Southern
Rhodesia. Trans.roy. Soc. S. Afr. 28: 219-253.
SMOLLA, G. 1957. PrehistorischeKeramik aus Ostafrika. Tribus.6: 35-64.
SOPER, R. C. 1967a. Iron Age sites in northeastern
Tanzania.Azania.2: 19-36.
SOPER, R. C. 1967b.Kwale: An early Iron Age site in
south-easternKenya. Azania.2: 1-18.
SOPER, R. C. 1968. Early Iron Age pottery typesfrom
East Africa: comparativeanalysis. Presentedto VI
Pan African Congressof Prehistory.Dakar.
SPAULDING, A. C. 1960. The dimensionsof archaeology. In Dole & Carneiro, eds., Essays in the
Scienceof Culturein Honorof LeslieA. White.New
York: Thomas Y. CrowellCo.
STUIVER, M. & VAN DER MERWE, N. J. 1968. Radiocarbon chronologyof the Iron Age in sub-Saharan
Africa. CurrentAnthropology.9 (1): 54-58.
R. 1958. Inyanga:Prehistoricsettlementsin
SUMMERS,
SouthernRhodesia.Cambridge:CambridgePress.
SUMMERS,R., ROBINSON, K. R., & WHITTY, A. 1961.
Zimbabwe Excavations: 1958. Occ. Papers NatI.
Mus. S. Rhod.3 (23A).
SWADESH, Morris. 1960. Linguisticsas an instrument
of prehistory. SouthwesternJ. of Anthropology.
15 (20).
W. 1967.The Pre-Europeanpotteryof SouthSYDOW,
west Africa. CimbebasiaMemoir.1.
VOGEL, J. 0. In press.Kamanagoza:An introduction
to the Iron Age cultures of the Victoria Falls
Region.
WILLET, F. 1967. Pottery classification in African
archaeology. West Afr. Archaeol. Newsletter. 7:
44-55.
C. C. 1960. Speculationson the economic
WRIGLEY,
prehistoryof Africa.J. Afr. Hist. 1 (2): 189-203.
SCHOFIELD,J.