This article was downloaded by: [University of New Hampshire]On: 19 October 2014, At: 08:51Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Azania: Archaeological Research inAfricaPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/raza20

Betwixt Tichitt and the IND: thepottery of the Faïta Facies, TichittTraditionK.C. MacDonald aa Institute of Archaeology , University College , LondonPublished online: 11 Apr 2011.

To cite this article: K.C. MacDonald (2011) Betwixt Tichitt and the IND: the pottery of the FaïtaFacies, Tichitt Tradition, Azania: Archaeological Research in Africa, 46:1, 49-69

To link to this article: http://dx.doi.org/10.1080/0067270X.2011.553485

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to orarising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Betwixt Tichitt and the IND: the pottery of the Faıta Facies, TichittTradition

K.C. MacDonald*

Institute of Archaeology, University College, London

This paper examines both decorative and formal change in the ceramics of theTichitt tradition of Mauritania (c. 1900-400 BC), and this tradition’s expression inthe Middle Niger, the Faıta Facies (c. 1300 � 200 BC). Using attribute-basedcomparisons, a wide range of assemblages from Mauritania and Mali are utilisedto demonstrate how temporal divisions may be discerned in this sequence.Particular attention is paid to the definition of Early and Late Faıta ceramicphases and the origins of finewares in the Middle Niger. It is notable that TichittTradition ceramics feature frequent and early examples of cord roulette use in theWest African Sahel.

Ce document examine les aspects decoratif et formel de la ceramique de latradition Tichitt de la Mauritanie (c. 1900-400 av.J-C), et un expression de cettememe tradition au Niger Moyen, le facies Faıta (c. 1300-200 av.J-C). Utilisant desanalyses basees sur les attributs, plusieurs assemblages de Mauritanie et le Malisont utilises pour demontrer comment des divisions temporelles peuvent etrediscernees dans cet sequence. Une attention particuliere est pretee a la definitiondes phases des ceramiques Faıta et les origines des « finewares » au Niger Moyen.Il est notable que les ceramiques de tradition de Tichitt sont parmis les premiersutilisant les cordelettes dans le Sahel d’Afrique occidentale.

Keywords: Ceramics; Late Stone Age; Tichitt; Faıta; Middle Niger; Inland NigerDelta (IND)

‘I, with my partie, did lie on our poste, as betwixt the devill and the deep sea.’

Robert Monro, 1637 His expedition with the worthy Scots regiment called Mac-keyes

Pottery classification in the West African Sahel

Pottery has been used since the inception of modern archaeology to measure time

and divide (cultural) space. The seriation of ceramic assemblages has been amply

demonstrated to provide real results and it remains a cornerstone of the latest

Darwinian approaches � even if there are disagreements as to processes driving

‘battleship curve’ distributions (O’Brien and Lyman 1999, 2000; Shennan and

Wilkinson 2001). Likewise, across space, even with the notional dissolution of

Culture History, archaeological ‘cultures’ (however re-branded) and type-variety

ceramic systems remain in broad use outside Sub-Saharan Africa. Once again, the

material existence of stylistic boundaries as archaeological phenomena is not

challenged, but the means of explaining these ‘entities’ is now largely outside the

*Email: kevin.macdonald@ucl.ac.uk

Azania: Archaeological Research in Africa

Vol. 46, No. 1, April 2011, 49�69

ISSN 0067-270X print/ISSN 1945-5534 online

# 2011 Taylor & Francis

DOI: 10.1080/0067270X.2011.553485

http://www.informaworld.com

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

bounds of the original culture historical approach (Lyman et al. 1997). In the words

of O’Brien and Lyman (2000, 393), ‘There is nothing inherently wrong with any of

the units devised by archaeologists of the twentieth century. What is inherently

wrong is how they are often misused. . .’. The stigma attached to the material

classificatory units of culture history is therefore not that they are inaccurate, but

that their interpretation can be either anthropologically naıve or, worse, may

succumb to nationalistic or racist agendas (Veit 1989; Ucko 1995).In the West African Sahel the systematic classification of pottery assemblages got

off to a slow start. In part this was due to the paucity of archaeology in the region

before the radiocarbon era, which meant that archaeologists were never purely

reliant on ceramics as markers of time. However, without ceramic sequences one is

left without a means of usefully estimating the date of surface assemblages and over

space one is left without a means of assessing social boundedness or relatedness.

Until the work of Patrick Munson and Susan McIntosh the region had few (if any)

ceramic reports featuring adequate statistical treatments that would allow either the

temporal phasing of assemblages or their rigorous comparison with other

assemblages (Munson 1971; McIntosh and McIntosh 1980; McIntosh 1995).

Interestingly, neither used what could be called a traditional typological approach.

Both authors, children of the New Archaeology, used attribute-based approaches,

comparing the individual decorative elements of sherds, rim forms and pastes.

Munson (1971) went a step further and, forming codes out of chains of attributes

held by individual sherds, used computer matching to form different levels ofattribute clusters (e.g. nine to ten out of ten attributes match, seven to eight out of

ten attributes match, etc.). Such attribute clusters are essentially explicit, statistically

derived types (sensu Spaulding 1953). Attribute-based analytical approaches have

spread in our region over recent years, and some attempts have even been made at

utilising Munson’s (1971) methods to arrive at attribute clusters (e.g. Schmidt et al.

2005). However, most work merely tends to assess the percentage presence of

attributes, such as cord roulette types, rim forms, and so forth. The difficulty is

making such attribute-based work useful at a regional level and in arriving at

definitions of artefact aggregates over time and space.

For the classification of artefact aggregates I employ here two terms that merit

definition: tradition and facies. The first is employed in the manner intended by its

creator, the Americanist Gordon Willey (1945, 53), with a tradition subsuming ‘a

line, or a number of lines, of pottery development through time within the confines of

a certain technique or decorative constant.’ In the case of the present study, for

example, such constants include certain rim forms, cord roulettes and ceramic pastes.

I employ the term ‘facies’ in a way comparable to that in which Ford and Willey(1941) used the term ‘horizon’ � essentially as a term denoting a ceramic phase, being

archaeological assemblages of the same broad temporal horizon or period, holding

similar sets of attributes and occurring within a restricted area. The term ‘facies’ was

chosen in deference to Francophone colleagues who were already using this term in

the region for a similar classification of related assemblages.

In creating new ceramic attribute clusters, facies and traditions for West Africa, I

am well aware that we are caught, as it were ‘betwixt the devil and the deep sea’. On

the one hand, we risk creating classifications that will be misunderstood, misapplied

or stretched beyond their rightful boundaries. On the other, there is a continual

danger of casually equating pots with peoples. However, if we do not formulate some

50 K.C. MacDonald

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

forms of aggregate classifications we will lack the baseline components used

elsewhere in world archaeology to create narratives and to drive more sophisticated

research questions about social boundaries and identities within such aggregates.

In the following article I attempt to use attribute-based analysis to refine ourdefinition of a pivotal ceramic tradition at one of the key transitional points of West

African history. In so doing, it is my hope to move the ceramic agenda for the region

between Tichitt and the Inland Niger Delta (IND) forward in a constructive manner,

as a basis for more sophisticated and critical inquiries.

The archaeological problem

Between Dhar Tichitt-Oualata (c. 1900 � 400 cal. BC), arguably West Africa’s firstlarge scale polity, and the Empire of Ghana (c. cal. AD 300 � 1100) with its Inland

Niger Delta (IND) bread basket, there rests the Faıta Facies. Identifiable from its

pottery, this archaeological entity may be found at Mali’s border with Mauritania

(the frontier post of Faıta), beneath the massive tell sites of the Mema, and in the

southern Macina, in the initial occupational horizon of Dia-Shoma. A few Faıta

Facies sherds are even documented from Jenne-jeno (see McIntosh 1995, Plate 13).

The Faıta Facies is, I believe, the nexus between the Tichitt Tradition and the first

millennium AD civilisation of the Inland Niger Delta (IND). Corresponding as itdoes to the first millennium BC transitional period, a time of both socio-political and

technological change, it is important that our definition of the Faıta phenomenon is

explicit.

Since initially proposing the existence of the Faıta Facies in my doctoral

dissertation (MacDonald 1994), new data on Tichitt and its relations to the south

have accumulated. There has been fresh fieldwork at southern Mauritanian Tichitt

sites in the Tagant (Ould Khattar 1995a, 1995b) and Dhar Nema (MacDonald et al.

2003, 2009; Person et al. 2004, 2006). Amblard-Pison (2006) has also published asynthesis of her long-standing fieldwork at Dhars Tichitt and Oualata (undertaken

from 1980 to 1996).

Regarding the Faıta Facies itself, further work has been undertaken on its

occurrence in the Mema by Togola (pers. comm.) at Akumbu, and by a Japanese-

Malian team at Kolima Sud-est (Takezawa and Cisse 2004). Finally, there have been

a number of excavation campaigns at Dia, where a substantial Faıta Facies site has

been tested at the base of Dia Shoma (Bedaux et al. 2001, 2005; MacDonald and

Schmidt 2004) (Figure 1). All of the foregoing has served to muddy the waters indefining what is, or is not, Faıta � and even publications to which my name is

attached may sometimes appear contradictory and confusing when viewed against

my dissertation.

Part of the difficulty lies in the misinterpretation of my initial division of Mema

pottery and stone tool assemblages into four Ceramic Late Stone Age (or

Neolithique) facies. These were not four distinctive entities, but rather two facies

(or phases) of two distinct traditions: Ndondi Tossokel and Faıta (Tichitt Tradition,

discussed here) and Kobadi and Beretouma (Kobadi Tradition, MacDonald 1994,1999). This hypothetical temporal relationship was clearly stated at the time

(MacDonald 1994, 107�118), but it bears repetition here.

Also, a decade ago, while a member of the international Dia project, I was

encouraged to collapse the Ndondi Tossokel facies and Faıta facies into one entity,

Betwixt Tichitt and the IND 51

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

as ceramic elements which could be assignable to both were found in the 800 to 0

cal. BC basal layer at Dia-Shoma. So, some pottery styles initially defined as

Ndondi-Tossokel, and now termed Early Faıta, were simply termed ‘Faıta’ in

MacDonald and Schmidt (2004). Additionally, disagreements about what was, and

what was not, Fineware (also known as Deltaware), saw a good deal of what I

would consider Late Faıta pottery classed as Deltaware in the final report (Schmidt

et al. 2005). Therein the ‘Faıta’ descriptor was somehow only reserved for large,

coarse Faıta storage vessels, a mere fraction of the assemblage. While I was one of

the co-authors of this report, I did not agree with this aspect of its contents.

Fortunately, I have been able to undertake an independent re-analysis of a portion

of the Dia Shoma assemblage and will take a fresh look at the Dia early pottery

sequence in the present article.

In essence, what I will be arguing below, using only assemblages that I have

been able to examine physically myself, is that the Tichitt Tradition first appeared

in the Mema region around 1300 cal. BC. Such pottery is termed the Early Faıta

Facies (ex � Ndondi Tossokel). Within a few hundred years notable changes occur

in this particular ceramic trajectory, becoming the Late Faıta Facies sometime

between 800 and 400 cal. BC. The Late Faıta Facies brings us some important new

elements: accordion pleat roulettes, proto-finewares that segue into the Middle

Niger Deltawares, rammed earth architecture, and iron metallurgy � forming a

fundamental point of transition from the Tichitt Tradition to the Phase I/II

ceramics of the Inland Niger Delta around 200 cal. BC. Let us begin by examining

the ceramic sequence of the Tichitt itself.

Figure 1. Map of regions and sites mentioned in the text (NdT� Ndondi Tossokel).

52 K.C. MacDonald

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

The Tichitt ceramic sequence

Although a good deal of time has passed since Patrick Munson’s (1971) original

eight-phase classification of the Tichitt sequence, there is still much to recommend it,

particularly if one is willing to revise and compress its copious number of phases, as

Vernet (1993, 264) has done. Munson’s ceramic seriation is elegant and results in a

clearly unfolding sequence. Unfortunately, most scholars who have worked in the

region since 1971 have made no effort either to embrace it or to refute it. Whether

this reluctance to employ Munson’s pottery sequence stems from difficulty of access

(it exists only in Munson’s unpublished PhD; Munson 1971), linguistic barriers, or

other reasons, the study of Tichitt since 1971 has been the poorer for lacking an

appreciation of the diagnostic temporal elements of its ceramics. In advocating the

utility of Munson’s original sequence, it can be argued that his dates came from

contextual excavations at a range of sites, and that his pottery chronology was

anchored upon a methodologically sound seriation of carefully chosen ‘single

component’ ceramic assemblages.

This difficulty of operating without a ceramic sequence is shown to good effect by

Amblard-Pison’s (2006) recent Tichitt-Oualata synthesis. Her Tichitt-Oualata

pottery assemblages, representing over 2000 years and 28 sites, are analysed as a

totality and without reference to Munson’s earlier ceramic seriation. Moreover,

decorative tools used to produce motifs are only separated and presented

quantitatively for a single site, while for other sites only their ‘effects’ are catalogued

(e.g. ‘incisions’, ‘chevrons’, ‘impressions’, ‘dents’, ‘flames’, ‘lignes ondes’, etc. . .). No

wonder then that Amblard-Pison’s portrayal of Tichitt is one without chronology, a

socio-economic and cultural entity that arrives fully formed along the falaise,

multiplies, dwindles and disappears. Without a pottery chronology as a benchmark,

especially if one’s radiocarbon chronology is based on surface-collected potsherds,

then one is apt to see only stasis. In dismissing Munson’s excavation-based

developmental sequence for Tichitt � e.g. the gradual advent of stone architecture,

or the shift from unfortified to fortified sites � Amblard-Pison’s (2006) new dates

derived largely from surface sherd organics are used uncritically. In an area of almost

total aeolian deflation one cannot assume that a sequence of surface sherd dates

from a single site represents continuity of occupation, scale of occupation, or

association with architectural features.

Having had the opportunity in 1993 to go over and re-record the entirety of

Munson’s original Tichitt assemblages with him, and to excavate stratified

assemblages in Dhar Nema with Robert Vernet in 2000, it is possible both to

compress and summarise this sequence in an updated form. The date ranges utilised

stem primarily from Munson’s original excavated samples, coupled with dates from

more recently excavated deposits in adjoining regions.

Early Tichitt: Khimiya and Goungou phases, c. 1900-1600 cal. BC

Munson (1971, 1976) originally viewed the Khimiya and Goungou phases as a

mobile, pre-cereal agriculture period for Tichitt pastoralists. Direct, stratified dates

of Tichitt’s distinctive dry-stone architecture are rare, and none come from this

period. Munson therefore did not believe that the stone compound walls of Tichitt

existed in the earliest phases. Amblard-Pison (2006, Figure 11) musters only

Betwixt Tichitt and the IND 53

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

five ‘early village’ or ‘architectural’ dates, including a Dakar laboratory-dated

charcoal sample from the Tichitt ‘type site’ of v.72 (also known as Akreijit or Seyyid

Ounquil) of 37769120 BP (Dak-52; 2440-2030 cal. BC; see Endnote 1), which

appears aberrant when viewed against the spread of dates from this locality. Theother dates are surface sherds from sites v.72, v.149 and v.304. None of these

determinations inspire confidence in attributing fully developed Tichitt architecture

to this period.

However, I concur with Amblard-Pison (2006) that recent research has clearly

demonstrated that important agricultural developments did, in fact, take place

during the Early Tichitt period, with morphologically domestic millet being dated to

1950-1690 cal. BC from village 72, Dhar Tichitt (Pa-1157; 35009100 BP; Amblard

1996, Amblard-Pison 2006) and 1740-1610 cal. BC from Djiganyai, Dhar Nema(GX-25359-AMS; 3370940 BP; MacDonald et al. 2009). But, whether these dates

indicate that Dhar Tichitt arrived as a fully formed agro-pastoral economic package,

as Amblard-Pison (2006) argues, or represents a rapid local domestication of millet

during the Pre-Tichitt or Early Tichitt period, remains to be proven (MacDonald

et al. 2009).

Early Tichitt pottery is most remarkable for its widespread employment of

twisted cord roulettes, which dominate its decor (52-64% of total motif occurrences;

Table 1 and Endnote 2). This is amongst the earliest appearances of West Africancord roulettes, falling geographically within the core zone for this innovation

proposed by Livingstone-Smith (2007; see also Manning 2011). More rarely, cord

wrapped-elements are still applied in an impressed form, usually in neat rows, and

not rouletted. It is possible that the cord-wrapped tools involved were not cylindrical

and therefore could not have been rouletted. As is the case throughout the Tichitt

Tradition, there are rare examples of matt impressions, more often on body sherds

than rims, such as those illustrated by Holl (1986, 83) (for a further discussion of

early mat-formed pottery see Manning 2011; Mayor 2011).Undecorated vessels are most numerous during the Early Tichitt period, most

such vessels being lightly burnished. It should be noted that Tichitt pottery of all

periods, to some degree or another, utilised red slip, particularly on the lips of vessels.

However, aeolian sand erosion makes quantification of this attribute very unreliable.

The forms of the Early Tichitt pottery remain much as they were during the Pre-

Tichitt period, being a variety of globular simple rimmed vessels, mostly slightly

closed in rim angle. Such rare everted vessels as do occur are only closed vessels with

slight inflections at their lip. Throughout the Tichitt tradition chaff remains thedominant temper, whether using Cenchrus (cram-cram) or Pennisetum (millet) chaff.

The distinctive pottery of the Early Tichitt period is now documented from Dhar

Tichitt and Dhar Oualata, and in small quantities from Dhar Nema, although not

yet from the Tagant (cf. Ould Khattar 1995, 1995b; MacDonald et al. 2009).

Classic Tichitt: Nkahl, Naghez and Chebka phases, c. 1600-1000 cal. BC

Classic Tichitt represents this tradition’s floruit, with most of Tichitt’s mainpopulation centres coming into being during this period. Tichitt’s distinctive and

vast settlements, with their conjoined clusters of stonewalled compounds, spread at

this time across Tichitt-Oualata and as far afield as the Tagant (Ould Khattar

1995a). Likewise, test excavations from Dhar Nema place occupation at the tell site

54 K.C. MacDonald

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

Table 1. Tichitt pottery percentages. The numbers presented in this table are rounded percentages of the samples analysed. None of the pottery samples

comprised fewer than 25 or more than 80 rim sherds. The samples from Tichitt preferentially feature excavated assemblages.

Fabric Rims Decor

Site Phase

Grog/

Bone Sand Chaff Sponge Simple Everted Thickened Collared PFR PFI

CR

Twist

CR

Pleat

CR

Braid

CR

Knot Punct Stylus Plastic Mat/ Net Comb Undec

Goungou B Early 0 7 93 0 100 0 0 0 0 7 64 0 0 0 0 0 0 11 7 11

Goungou C Early 0 0 100 0 98 2 0 0 0 2 52 0 0 2 0 2 0 0 0 42

Naghez Classic 0 0 100 0 50 43 7 0 0 4 67 0 0 4 0 18 0 0 0 6

La Baidha I Classic 3 0 97 0 36 70 15 0 8 6 43 0 0 6 1 27 3 0 0 6

Seyid Ouinquil

(aka Akreijit, v.72)

Classic 3 0 97 0 29 57 14 0 5 11 43 0 0 11 0 18 11 0 0 1

Dhar Arriane Late 0 0 100 0 39 57 2 2 11 2 44 0 0 2 0 22 7 4 0 9

Bledd Initi Late 4 0 96 0 4 73 8 15 13 13 28 0 3 3 3 19 0 0 13 5

Key to decor type abbreviations (see Endnote 2):PFR �peigne filete roulette, also known as rolled cord-wrapped roulettePFI �peigne filete impression, also known as multiple impressed cord-wrapped rouletteCR Twist � twisted cord rouletteCR Fold � folded strip rouletteCR Braid �braided cord rouletteCR Knot �knotted cord roulettePunc � stabbed stylusStylus �dragged or incised stylus, including wavy or geometric designs made with stylusPlastic �applied plastic nubbins and filletsMat/Net �pottery which appears to have been impressed on a mat or netComb � rocker, stamped or dragged combUndec �undecorated

Betw

ixt

Tich

itta

nd

the

IND

55

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

of Djiganyai and the small settlement of Bou Khzama during the Classic Tichitt

period. As we shall see below, it is also during this period that the Tichitt Tradition

enters the Mema region.

Following Munson (1971), I would argue that the growth of Tichitt’s settlementswas a gradual one, but existing radiocarbon dates cannot sustain his initial three-

phase division. Classic Tichitt pottery, with a major direct dating programme on

diagnostic sherds, could probably eventually be divided into at least two distinctive

phases. However, on the present state of evidence it is probably best to err on the side

of caution and leave our definition inclusive. For pottery, the main trend from Early

Tichitt to Classic Tichitt is one of extreme diversification (Table 1). Forms of pots

become markedly differentiated, with a proliferation of new everted rim types,

including at least 12 distinct rim forms. These largely replace slightly closed simplerimmed vessels, with simple rims now being more often employed for open bowls and

tightly closed pots. Thickened rim forms also appear, largely via added clay on the

upper body and lip, creating a variety of bulbous profiles. These were mostly

employed for more robust closed vessel forms. It is tempting to think that the

diversification of pottery forms during this period may be linked with increasing

display of socio-economic difference within Classic Tichitt’s expanding stonewalled

settlements.

Yet, it is in decorative terms that Classic Tichitt pottery diversifies the most.Cord-wrapped elements return as a decorative component, this time definitely in

cylindrical form, since they begin to be used as roulettes. Interestingly, cord-wrapped

roulettes are almost inevitably employed on everted rimmed vessels in a vertical

manner, descending from the collar onto the upper body of the vessel. This motif is

one of the most distinctive Classic Tichitt motifs. This same tool is also often used in

a linear impressed manner in rows on the neck or upper body of vessels. Dragged

stylus or grass also proliferates, used in arched geometric patterns and wavy lines.

Finally, there is a distinctive flowering of applied plastic motifs. They consist ofeither coin-sized nubbins applied in dispersed rows around the upper bodies of

vessels, or clay banks (fillets) in the same area, usually impressed by cord-wrapped

elements or sliced by a stylus.

Finally, during the Classic period there is a slight increase in paste variety, with

grog or bone more frequently being added to chaff tempered vessels and a small but

steady stream of sherds not tempered with chaff at all, but rather with combinations

of grog and bone (3�4% of sherds). Likewise, at Djiganyai in excavated Classic

Tichitt layers, grog-tempered sherds, without any chaff temper, comprise between24% and 36% of the assemblages (MacDonald et al. 2009). It is tempting to think

that the dominant chaff-tempering of Tichitt’s pottery results from activities

scheduling, with most pots being made directly after the harvest, and only a few

being made without chaff in other periods of the year. Alternatively, such pots may

merely have been made at localities without conveniently accessible crop processing

waste.

Late Tichitt: Arriane and Akjinjeir phases, c. 1000-400 cal. BC

The end of Tichitt remains as ambiguous as its beginning. Munson (1971, 51) sees

this phase as an extreme refuge period, with settlements restricted to those that were

‘very small, very poorly constructed, very ‘impoverished,’ heavily fortified, and so

56 K.C. MacDonald

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

well hidden among the high rocks that it would have been almost impossible to have

found any of them without the aid of aerial reconnaissance.’ Growth in Tichitt

Tradition settlements outside the core region during this period, whether in the

Tagant or along the northern banks of the Middle Niger, attests to the ‘emptying’out of the Dhars and the Hodh basin at this time. Munson (1980) ascribes this to

environmental collapse and Berber incursions.

Regarding pottery, the assemblage from the Dhar Tichitt fortified settlement of

Bledd Initi is similar in many ways to Classic Tichitt assemblages, except for the

appearance of dragged comb motifs, thumbnail punctuate, and high collared vessel

forms, and perhaps one final Tichitt innovation, braided cord roulettes that also

appear at this time in Tichitt-derived assemblages in the Mema and the Macina

(Tables 1�3). On the other hand, applied plastic motifs decrease or disappear duringthis period, as do most simple rim forms.

It is interesting that some of the Late Tichitt ceramics display certain hallmarks

that were to continue in the region up to the time of Koumbi Saleh (from c. AD 1000;

Berthier 1997). The Dhar Nema ‘Groupe III’ pottery as described by Person et al.

(2006), and directly dated to the first millennium AD, has also been directly dated by

MacDonald et al. (2009) to the Final Tichitt period (two direct dates falling in the

770�400 cal. BC time range). This would seem to indicate a long continuity of this

Late Tichitt pottery style in the region. Such large, sparsely decorated, high-collaredpottery is also known from first millennium AD Berber sites including Tegdaoust

and Essouk. As in the Late Tichitt period, their decoration is sparse, usually

featuring twisted cord roulettes. I therefore believe Munson’s Final Tichitt

assemblages may ultimately be found to represent a syncretism of incoming Berber

(Maure) pottery styles and with that of the final Tichitt inhabitants.

The Faıta Facies in the Mema

The existence of the Tichitt Tradition in the Mema was first signalled, albeit

unknowingly, by Raymond Mauny at the Sixth Pan-African Congress of Prehistory

at Dakar in 1967. Distinctive Tichitt tradition vessels appear like strange guests in

Mauny’s (1972) illustrations of Kobadi vessels in his ‘Planche II,’ nos. 8�10. These

vessels share forms and decorative motifs with vessels he illustrates in the same

article from a number of Tichitt-Oualata sites (Mauny 1972, Planche V, no. 1 and 17,

Planche VII, no. 1): despite these similarities, a connection was not made in the text

of Mauny’s paper.I found this connection more apparent while working with these same collections

in 1989. It became a good deal clearer while working in the Mema later that same

year, and was confirmed in a comparative visit to Patrick Munson’s Tichitt

collections in 1993. It should be noted, however, that not everyone agrees with this

assessment. Writing about the notion of the Tichitt Tradition in the Mema, and

making reference to my initial writings on it (MacDonald 1996, 1998), Amblard-

Pison (2006, 292) remarks: ‘. . . il faudrait disposer des elements qui font avant tout

l’originalite des populations des Dhars Tichitt et Oualata. Cette affirmation ne peutdonc, actuellement, etre consideree que comme une hypothese basee en fait sur la

seule proximite de ces deux regions.’

It is not stated exactly which ‘elements’ Amblard-Pison refers to here: my initial

articles made quite clear the correspondence of pottery forms, decorative elements

Betwixt Tichitt and the IND 57

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

Table 2. Mema pottery percentages. The numbers presented in this table are rounded percentages of the samples analysed. None of the pottery samples

comprised fewer than 25 or more than 80 rim sherds. See Table 1 for key to decor type abbreviations.

Fabric Rims Decor

Site Fine

Grog/

Bone Chaff Sponge Simple Everted Thickened Handles Pot Lids PFR PFI

CR

Twist

CR

Fold

CR

Braid Stylus Plastic Comb Undec

Early Faita Assemblages

Ndondi Tossokel 1119�1 0 81 0 19 37 63 0 0 0 48 6 0 0 0 18 12 3 3

Saberi Faita 0 73 8 19 54 43 3 0 0 55 7 14 0 0 7 3 7 7

Ndondi Tossokel 1119�6 0 55 39 5 61 33 6 0 0 39 14 19 0 0 10 0 4 14

Transitional Assemblage

Kolima Sud�Est 6 35 53 6 32 39 29 0 0 23 47 5 16 0 5 9 0 16

Late Faita Assemblages

Ndondi Tossokel 1119�3 9 82 9 0 22 39 35 0 4 18 46 0 10 0 0 4 4 18

1251�2 4 82 8 4 35 29 28 0 8 9 58 5 13 2 5 5 2 3

Faita Ouest 2 44 54 0 30 39 31 0 0 9 54 0 14 0 0 0 0 23

Diaguina 4 10 86 0 38 36 18 2 6 0 33 0 28 0 5 7 0 28

Akumbu LSA 4 38 58 0 13 30 53 4 0 3 10 10 43 0 7 0 0 27

58

K.C

.M

acD

on

ald

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

and their manner of employment between Classic Tichitt ceramics and those of the

Ndondi Tossokel (Early Faıta) facies. I also made reference to the predominant

origin of stone raw materials used in the Mema from the Mauritanian Dhars

(especially the distinctive siltstone phthanite), as well as correspondences in

associated projectile point, stone bracelet, and hachette forms (cf. MacDonald

1994, 1996). The only element for which there is not a link are dry stone walls and

schist grain bin pillars, the distinctive Tichitt architectural tradition upon whichAmblard Pison (2006, 53�118) focuses a good deal of attention. However, if this is to

be her criterion, given that the Mema is a floodplain and without native stone

resources comparable to those of the Tichitt escarpment, it would be rather difficult

to ever prove an archaeological link on that evidential basis.

Let us then examine the Faıta Facies more carefully in relation to Tichitt. Early

Faıta pottery occurs at relatively ephemeral sites beside the ancient Mema

floodplain, particularly in an area known locally as ‘Ndondi Tossokel’. These

single component sites, although eroding, feature small intact middens containing

cattle bone, broken pottery and worked, imported stone (largely phthanite and

sandstone from the Mauritanian Dhars, used both for chipped and polished

implements) (MacDonald 1994). The forms of polished implements at the Early

Faıta sites conform to Amblard’s (1984) typology of polished stone axes and rings

from Tichitt (cf. MacDonald 1994, Table 5.3). Pottery and stone assemblages

identical to those from Ndondi Tossokel also occur atop Saberi Faıta, the last

inselberg of the Tichitt chain, situated on the Mauritanian border with Mali. Spreadover 6 ha, this site also features grinding equipment, which is absent from Early Faıta

sites along the Mema floodplain (MacDonald 1994).

Early Faıta assemblages are only directly dated at the site of Kolima-Sud, a

deeply stratified floodplain site, where Faıta pottery and stone artefacts co-occur

with ceramics from the Kobadi Tradition. As has been extensively argued elsewhere

(MacDonald 1994, 1999), this site was first occupied by the Kobadi Tradition as a

fishing site, and subsequently had an additional seasonal presence of Tichitt-derived

pastoralists, as witnessed by the appearance of Early Faıta pottery, cattle remains,

phthanite, and cattle figurines in Layer III and II of the stratigraphy (MacDonald

1994). From Layer III a cattle molar has been directly dated by bioapatite to

3084973 BP (GX-19814-A-AMS; 1440-1260 cal. BC; MacDonald 1994, 130). This

date is consistent with the stylistic resemblance of the ceramics to those of the

contemporary Classic Tichitt period.

The pottery of the Kobadi Tradition and the Early Faıta Facies can be clearly

separated by their fabric (Kobadi ceramics are sponge-tempered), their forms(Kobadi ceramics were predominantly simple rimmed) and by their decorative motifs

(Kobadi potters relied heavily on potter’s combs and spatulae as decorative tools).

The pottery of the Early Faıta facies is relatively straightforward to define, and,

excepting its fabric which is more often tempered with grog (55�81%) than with

chaff (5�19%), it conforms stylistically to the Classic period of the Tichitt Tradition

(Tables 1 and 2). Its rim forms are mostly everted (33 to 63%) and conform to

Munson’s (1971) types 3, 4, 5 and 7. Thickened rims of the Tichitt type are present

but comparatively rare (3�6%). In terms of decoration, rouletted cord wrapped

elements predominate (39�55% of motifs). They are employed in the Tichitt fashion,

descending vertically from the neck of everted vessels, or horizontally on simple

rimmed vessels (Figure 2, A-C, E). Likewise, simple incisions used to create the same

Betwixt Tichitt and the IND 59

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

wavy or arched geometric patterns documented at Tichitt (Figure 2, D), and applied

plastic nubbins and fillets, sometimes cut by simple incisions, are shaped and

emplaced on the upper body of the pot, just as in Classic Tichitt assemblages.

Impressed cord-wrapped elements and rouletted twisted cord roulettes are also

present. Quantitatively, these five dominant motifs are also the principle decors of

the Classic Tichitt Tradition (Tables 1 and 2). The sparse presence of rocked or

impressed comb in the Ndondi Tossokel phase is probably due to the incorporationof some Kobadi Tradition ceramics into these assemblages, as indicated by the

distinctive sponge temper of these same sherds.

The only qualitative difference between Tichitt pottery from the Mauritanian

Dhars and that of the Early Faıta Facies is that of temper, where grog and bone

temper (55�81%) dominates chaff (0�39%), instead of vice versa. The explanation for

this may be, as previously stated, one of access to crop processing waste. As it has

been hypothesised (MacDonald 1994, 1999) that Early Faıta represents the seasonal

presence of a transhumant pastoral segment from the Mauritanian Dhars, they

would have been unlikely to have much access to chaff.

Around 900 BC, a broad settlement transformation took place in the region.

Instead of a seasonal pastoral presence with ephemeral camps and/or short-term co-

occupations with fisherfolk (as at Kobadi and Kolima-Sud), this extension of the

Tichitt Tradition began to make its own more permanent settlements. These ranged

from the 10 ha site of Kolima Sud-Est, with rammed earth architecture (MacDonald1994; Takezawa and Cisse 2004), to the founding layers of the tell complexes at

Akumbu (MacDonald 1994, 92, T. Togola pers. comm.) and Dia Shoma (the latter of

which also features earthen architecture, Bedaux et al. 2005). Kolima-Sud Est, with

both Early and Late Faıta Facies ceramics, has been dated via excavation and

charcoal radiocarbon dates to between c. 900 and 400 cal. BC, with raw dates of

2722, 2648 and 2521 BP (NB error ranges and laboratory numbers were not given in

the initial publication, Takezawa and Cisse 2004). Given these dates, it is not

surprising that Faıta Facies ceramics are also associated with the beginnings of iron

metallurgy in the region, with a smelting site associated with the settlement of Faıta

Ouest, and slag visible on the surface of both this site and Kolima Sud Est

(MacDonald 1994). At Dia Shoma, in the Macina, this association is better

documented with the stratigraphic association of Faıta Facies ceramics with iron

slag, and the find of a tuyere in Horizon I (Schmidt 2005).

Late Faıta Facies pottery begins in the Mema region during the occupation of

Kolima Sud-Est, between 900 and 400 cal. BC, exhibiting both parallels with, and

strong innovations from, the preceding period (Table 2). It is tempting to attributethese divergences to syncretism with the Kobadi Tradition, which no longer existed as

a materially visible entity in the first millennium BC. Similarities between Early and

Late Faıta ceramics include the continued use of cord-wrapped elements as a primary

decorative tool. However, there is a great diversification in thickened rim forms in Late

assemblages, either with a bulb at their extremity or with divergent inner and outer

walls crowned by a flattened lip (Table 2 and Figure 3, A, B, D, E). Concurrently, pot

lids with knob-shaped handles appear and become relatively common (Figure 3, F). In

terms of decor there are also important innovations, most notably the beginning of

folded strip roulettes (previously known as Twine 4 or accordion pleat roulette:

McIntosh 1995; Haour et al. 2010). These roulettes largely replace twisted cord

roulettes in Late Faıta assemblages and, of course, go on to become the principal cord

60 K.C. MacDonald

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

roulette form in the early ceramics of the Inland Niger Delta (McIntosh 1995; Schmidt

et al. 2005). Folded strip roulettes are especially dominant in the early layers of

Akumbu (‘Akumbu LSA’), where they account for 43% of all decor (Table 2). A few,

rare, braided cord roulettes are known from the site of 1251-2, and are similar to those

from late period Dhar Tichitt sites. One striking innovation in the Late Faıta Facies is

a shift to using cord-wrapped elements more frequently as tools for multiple

impression rather than as roulettes. These tools appear to have had a flexible core,

being often bent into curved shapes. This core is often clearly evident between the

wraps of more loosely constructed tools (Figure 3). At all sites with the exception of

‘Akumbu LSA’ impressed cord-wrapped elements comprise 33�54% of total motifs.

It is important to note that fine paste, burnished, and often otherwise

undecorated vessels are frequently associated with the more coarse wares of the

Late Faıta Facies, being a source of some confusion in defining its assemblages. This

paste represents the earliest ‘fineware’ known from the Middle Niger region, with

most examples being highly burnished and undecorated (Figure 3, C).

In essence, there are three separate fabrics present in Late Faıta assemblages:

coarse chaff-tempered pottery (usually used on large, ]40 cm diameter, thickened

rim vessels), medium coarse grog/bone-tempered vessels similar to the dominant

fabric in the Early Faıta period, and a very finely sorted and well-fired grog/sand-

tempered fabric, usually reserved for smaller well-burnished, undecorated vessels. As

all of these fabrics occur together in the well-stratified Faıta assemblage from Dia,

they do not appear to be temporal variants. They may have been made by different

Figure 2. Examples of Early Faıta Facies rim sherds: A �everted rim, red slip and cord-wrapped roulette, Ndondi Tossokel 1119-1, B� everted rim (jar form), red slip and cord-wrapped roulette, 1251-1, C� everted rim, cord-wrapped roulette, Kolima Sud, D� evertedrim, incised undulating line and cord-wrapped roulette, 1251-1, E� simple rim, cord-wrappedroulette, Ndondi Tossokel 1119-1.

Betwixt Tichitt and the IND 61

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

groups of potters, or by a single pottery tradition with different functional properties

in mind for vessels made using each fabric.

Finally, it is interesting that while Late Faıta phase vessels were, like those of the

central Tichitt Tradition, once again frequently tempered with chaff (8�86% of

assemblages, Table 3), it is not millet chaff, but rather that of fonio (Digitaria sp.).

Figure 3. Examples of Late Faıta Facies rim sherds: A� thickened rim, impressed cord-wrapped roulette on interior and exterior, Faıta Ouest, B� thickened rim, rocked cord-wrapped roulette on interior and exterior, Kolima Sud-Est, C� everted rim, burnishedfineware, 1251-2, D� thickened rim, impressed cord-wrapped roulette, 1251-2, E� thickenedrim with folded strip roulette, Akumbu AK3, first occupation layer, F� pot lid, impressedcord-wrapped roulette, G� everted rim, impressed cord-wrapped roulette, Kolima Sud-Est.

62 K.C. MacDonald

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

This grain has been identified macroscopically and abundantly at Kolima Sud Est,

where it is the only domestic grain identified (Takezawa and Cisse 2004). K.

Neumann (pers. comm) has also identified it as the cereal type used to temper the

Faıta Facies ceramics found by Togola in the basal layers of unit AK3 at Akumbu

(see Endnote 3). Millet has, thus far, not been recovered from first millennium BC

contexts in the Mema. It may well be that incoming populations from the

Mauritania Dhars found that millet was not suitable to the well-watered and clay

rich soils of the Mema during this time period, or perhaps that fonio was already in

cultivation in the area when they arrived.

The Faıta Facies in the Macina: the Dia Shoma assemblage

The Horizon I, first millennium cal. BC, deposits of Dia Shoma represent the largest

occupation site of the Faıta Facies yet known, shown via extensive test excavations to

be over 18 ha in expanse (Bedaux et al. 2005). At present, Faıta Facies ceramics are

only known at one further site in the region, the tell site of Tondodi, currently being

eroded by the Niger near Diafarabe. However, given the size of Dia’s early

occupation layer it is likely that many more Faıta phase settlements exist in the

region.

In my review of the data from Dia Shoma I differ in two respects from the official

joint accounts (Bedaux et al. 2001, 2005): site chronology and the classification of its

pottery. As with all large projects, final reports are inevitably a contentious affair and

ultimately a compromise of viewpoints. I should like to take this opportunity to

suggest some alternatives. First, regarding chronology, it is worth examining the

early, Horizon I and II, dates for the site. The final report places these periods

respectively between 800 to 0 cal. BC, and cal. AD 0 to 500. However a review of the

available dates shows that 15 of the 16 relevant determinations concentrate in two

distinct clusters between 800 and 400 cal. BC and between 200 cal. BC and cal. AD

200 (Figure 4). I should like to propose that there were, in fact, two hiatuses in

occupation at Dia Shoma, with the ]18 ha Faıta occupation occurring between 800

and 400 cal. BC, then a hiatus, a smaller occupation between 200 cal. BC and cal.

AD 200, followed by another hiatus. This is particularly clear in the dates for Unit B,

the ceramics of which I shall be considering below.

Regarding the ceramics themselves, there was a lack of consensus between

analysts about what did and did not constitute ‘Deltaware’: a particular form of

fineware initially documented at Jenne-jeno between 200 cal. BC and cal. AD 400

(McIntosh 1995). Additionally, in the final pottery report, the term Faıta was

erroneously only applied to large, thickened-rim, chaff-tempered vessels of the type

documented from the Late Faıta Facies in the Mema region. Other Faıta phase rims

were incorporated into the ‘Deltaware’ fabric category. In 2001 I visited Leiden

(where the collections from the first two seasons were then held) to re-record the rims

from Unit B, one of the most comprehensive early sequences from the site, for my

own future reference and to ensure comparability with my existing datasets. I was

therefore able to disentangle fabrics according to my own system previously used on

assemblages from Dhar Tichitt, Dhar Nema and the Mema. I present a portion of

this re-analysis as Table 3, focusing only on Deltawares/ Finewares and types

previously recognised in the Early and Late Faıta assemblages of the Mema.

Betwixt Tichitt and the IND 63

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

The Horizon I (800�400 cal. BC) assemblage combines Early and Late Faıta

elements, indicating that this period of time comprises the transition between the

two. The everted and simple rimmed, medium coarse grog-tempered vessels with

cord-wrapped roulettes typical of early assemblages are present (]10%), as are later

elements such as the burnished undecorated finewares (]5%), and the large, chaff-

tempered, thickened rim vessels decorated with folded strip roulettes or impressed

cord-wrapped elements (]4%). There are also various other medium coarse everted

rimmed vessels with applied plastic nubbins combined with a variety of braided cord,

folded strip and twisted cord roulettes (9%). Yet, surprisingly, the bulk of the

ceramics in Horizon I are, in fact, very fine, highly fired and burnished ‘Deltaware’

fabrics (c. 50%). This is the style of fineware first documented by Susan McIntosh

(1995, 153) at Jenne-jeno and ‘nicknamed ‘chinaware’ (because of the high-pitched

clinking reminiscent of fine china, made when two sherds were knocked together)’.

Deltaware is more refined in burnished finish and of greater hardness than the Faıta

finewares. A particularly prevalent early Deltaware type (]27%) comprises small,

flat-everted rimmed vessels with a band of red slip on their lip, followed by neatly

applied cord rouletting over the remainder of the vessel, (of folded strip, twisted or

Atmospheric data from Reimer et al (2004);OxCal v3.10 Bronk Ramsey (2005); cub r:5 sd:12 prob usp[chron]

1500CalBC 1000CalBC 500CalBC CalBC/CalAD 500CalAD 1000CalAD

Calibrated date

B104 (II) 1996±42BP

B114 (II) 1930±120BP

B117 (II) 1933±47BP

B95 (IA) 2470±40BP

B122 (IA) 2550±50BP

A94 (II) 2050±110BP

A115 (IB) 1990±50BP

C77 (IB) 2070±100BP

C92 (IB) 2450±30BP

C105 (IA) 2220±100BP

D79 (I) 2380±80BP

F81 (IA) 2470±70BP

K115 (IA) 2390±80BP

K118 (IA) 2522±47BP

K126 (IA) 2485±52BP

Figure 4. OxCal3 Graph of Dia-Shoma Horizon I and II radiocarbon dates.

64 K.C. MacDonald

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

Table 3. Ceramic attribute clusters from Dia Shoma, Unit B, Horizons I-III. Numbers to the left in columns are percentages of the total rims

from that horizon. Numbers to the right are raw counts. Rim forms are illustrated in Schmidt et al. (2005, Figure 7.1.2).

Wares Horizon II

(Fabric [Association], Rim, Angle, Decor)Horizon I

800�400 BC200 BC�AD

200Horizon III

AD 500�1000

Deltaware, Simple, Open, Slip+Burnish band, then PFR or CR Folded or Twisted 1.5 4 7.6 30 9.6 152Deltaware, Simple, Open, Burnish, Red Painted with Crosshatch Pattern, CR Folded or Twisted 0 0 3.5 14 2.1 32Deltaware, Simple, Closed, Slip+Burnish Band, then PFR or CR Folded or Twisted 0.8 2 3.8 15 2.9 47Deltaware, Simple, all other attributes undefined 8.6 23 17.5 69 16.7 266Deltaware, Everted (cf. E3/E4), Closed, Burnished only, Undecorated 0.4 1 1.8 7 5.6 89Deltaware, Everted (cf. E3/E4), Closed, Burnished only, CR Folded or Twisted 0.4 1 0.3 1 2.3 36Deltaware, Everted (cf. E3/E4), Closed, Slip+Burnish, and CR Folded or Twisted 9.7 26 17.2 68 5.9 93Deltaware, Flat Everted (cf. X2/X4), Closed, Slip on Lip + Burnish, CR Folded, Twisted or

Braided27.7 74 23.5 93 5.6 89

Deltaware, Smaller Everted (cf. E2/E9), Closed, Slip+Burnish only 1.5 4 1.1 4 5.1 80Deltaware, Smaller Everted (cf. E2/E9), Closed, Slip+Burnish, PFI 0 0 1.1 4 0.2 2Deltaware, Bottle or 90 degree Jar neck, Burnished 0.8 2 3.1 12 1.5 23

Medium Coarse Grog (Faita), Simple, Open, With one of: PFR, or CR Folded, Twisted, Knottedor Braided

2.2 6 0.5 2 0 0

Medium Coarse Grog (Faita), Everted (cf.E3/E10), Closed, Usually slipped on lip, PFR descendingfrom neck

8.2 22 0.8 3 0.1 1

Medium Coarse Grog (Faita), Everted (cf.E3/E10), Closed, Usually slipped on lip, CR Folded orTwisted

4.5 12 1.1 4 0.2 2

Medium Coarse Grog (Faita), Flat Everted (cf. X1/X3), Closed, CR Braided + Plastic Nubbins 4.5 12 0.5 2 0.2 2Fineware Grog/Sand (Faita), Various Everted (cf. X2, X7, E6), Closed, Burnished, Undecorated 5.2 14 2.3 9 1.3 21Coarse Chaff (Faita), Thickened (cf. S1, S5, E8a), Closed, Decorated int + ext with CR Folded or

Twisted, or PFI4.5 12 0 0 0.3 4

Total Rims in Sample 267 395 1589% of Total Rims not classed in the list of types given above 19.5% 14.3% 41.4%

Betw

ixt

Tich

itta

nd

the

IND

65

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

braided cord variety). Interestingly, different Deltaware attribute clusters grow

dominant during Horizons II and III (Table 3).

As to decorative tools, cord roulettes dominate in Horizon I, particularly folded

strip and twisted cord varieties. To a lesser degree, but still quite common, are cord-

wrapped roulettes and impressions. Braided cord roulettes occur but are very rare

(50.5%). According to the final report (Schmidt et al. 2005) geometric painting of

Deltaware was quite common in Horizon I. I record no instances of it from Unit B,

although it does become increasingly important in the unit’s later Horizons (II and

III).

In Horizon II (200 cal. BC to cal. AD 200) Faıta elements diminish considerably,

and may indeed be intrusive from earlier deposits. Deltawares comprise a remarkable

]77% of the ceramics in Horizon II in Unit B. The number of simple rimmed

Deltaware vessels increases, as do new very mildly everted forms (e.g. types E2 and

E9; Schmidt et al. 2005, Figure 7.1.2). By Horizon III (cal. AD 500 -1000)

Deltawares still occur frequently (]55%), but are diminishing.

In drawing conclusions from these data, I should note that I believe that the

Deltawares of the Inland Niger Delta developed out of the finewares first

documented in the Faıta phase of the Mema. Those from the Mema are not as

well finished as their later counterparts, but the consistency and composition of the

paste appears identical. The co-occurrence of these two fabrics in Horizon I adds

some weight to this hypothesis. It must, of course, also be stressed that there was

extensive pit digging at Dia Shoma, which, despite careful stratigraphic excavation,

no doubt brought some earlier material up to more recent layers. Despite this caveat,

demonstrated by the presence of small amounts of Faıta materials as high as

Horizon III, I do not believe that this fundamentally affects the sequence shown in

Table 3. This is shown by the visible seriation of types across these three distinct

occupational horizons. The unexpectedly early occurrence of Deltawares between

800 and 400 cal. BC at Dia Shoma � and their absence from the Mema (Togola 2008)

� implies that their development was in the Macina region, rather than further north

or south. We must continue to research, from both technical and social perspectives,

why ceramics of such quality (comparable to Roman terra sigillata � though pre-

dating it) came into existence along the ancient Niger.

Conclusions

In the course of this paper I have characterised the ceramics of the eastern trajectory

of the Tichitt Tradition, the Faıta Facies, employing datasets viewed through the eyes

of a single researcher. Attribute links between assemblages ranging from the

southeastern escarpments of Mauritania and the Inland Delta have been demon-

strated, and sequences of ceramic change proposed. I do not pretend that this is an

absolute key to this long and complex material culture trajectory. Nor do I view these

archaeological entities as having tidy affiliations with singular ancient cultural

identities. However, this study will hopefully serve as a new material culture

foundation, utilising a single terminology, that can be tested, refined and developed

in more sophisticated ways through future work.

66 K.C. MacDonald

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

The Tichitt Tradition, as defined here, traverses many key frontiers in the

archaeology of West Africa: the beginnings of cereal agriculture, early complex

societies, the advent of complex architectural forms, and the origins of iron

metallurgy. It is apparent that many of these transitions and their explanations are

contentious. Therefore, I can think of few archaeological entities in West Africa so

deserving of further study.

Acknowledgements

The research contained in this paper covers almost two decades of study and was funded by anumber of bodies including primarily (in chronological order): the Thomas J. WatsonFoundation, the United States National Science Foundation, the United Kingdom’s Arts andHumanities Research Council and the Institute of Archaeology, University College London. Ishould particularly like to acknowledge the assistance of the following colleagues inundertaking this research: Patrick Munson, Robert Vernet, Rogier Bedaux, and my latefriend Tereba Togola.

Notes

1. All dates appearing in this paper were calibrated, or recalibrated, using OxCal3.2. Terms used in referring to fibre roulettes are drawn from the nomenclature agreed in Haour

et al. (2010).3. Subsequent to his doctoral field research at Akumbu in 1989/1990 (Togola 2008), Tereba

Togola returned to Akumbu for a further excavation season, to complete unit AK3 onMound B. I examined this new Akumbu material with Tereba in 1996. The basal layers atAK3 contained predominantly Late Faıta ceramics with folded strip roulettes andthickened rims. A sample of these rims was confided to me by Tereba for identificationof their grain impressions (one example appears in Figure 3). Unfortunately, the secondAkumbu excavation season was unpublished at the time of his death.

Note on contributor

K.C. MacDonald (PhD Cambridge 1994) is Reader in African Archaeology at the Institute ofArchaeology, University College London. He has directed fieldwork in Mali, Mauritania andthe United States on a variety of topics, ranging from the Late Stone Age to the historicarchaeology of West Africa and the African Diaspora.

References

Amblard, S. 1984. Tichitt-Walata, Republique Islamique de Mauritanie. Civilisation et industrielithique. Paris: A.D.P.F.

Amblard, S. 1996. Agricultural evidence and its interpretation on the Dhars Tichitt andOualata, south-eastern Mauritania. In Aspects of African archaeology: Papers from the 10th

Congress of the Pan-African Association for Prehistory and Related Studies, ed. G. Pwiti andR. Soper, 421�427. Harare: University of Zimbabwe Publications.

Amblard-Pison, S. 2006. Communautes villageoises neolithiques des Dhars Tichitt et Oulata(Mauritanie). Oxford: Archaeopress.

Bedaux, R, K.C. MacDonald, A. Person, J. Polet, K. Sanogo, A. Schmidt and S. Sidibe. 2001.The Dia Archaeological Project: Rescuing cultural heritage in the Inland Niger Delta(Mali). Antiquity 75: 837�48.

Bedaux, R. J, Polet, K. Sanogo and A. Schmidt 2005. Recherches archeologiques a Dia dans leDelta Interieur du Niger (Mali): Bilan des saisons de fouilles 1998-2003. Leiden: CNWSPublications

Betwixt Tichitt and the IND 67

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

Berthier, S. 1997. Recherches archeologiques sur la capitale de l’empire de Ghana. Oxford:Archaeopress.

Ford, J.A. and G.R. Willey. 1941. An interpretation of the prehistory of the eastern UnitedStates. American Anthropologist 43: 325�363.

Haour, A., K. Manning, N. Arazi, O. Gosselain, N.S. Gueye, D. Keita, A. Livingstone Smith,K.C. MacDonald, A. Mayor, S.K. McIntosh and R. Vernet. 2010. African pottery roulettespast and present: Techniques, identification and distribution. Oxford: Oxbow Books.

Holl, A. 1986. Economie et societe neolithique du Dhar Tichitt (Mauritanie). Paris: EditionsRecherche sur les Civilisations.

Livingstone-Smith, A. 2007. Histoire du decor a roulette en Afrique subsaharienne. Journal ofAfrican Archaeology 5: 189�216.

Lyman, R. L., M.J. O’Brien and R. C. Dunnell. 1997. The rise and fall of culture history. NewYork: Plenum.

MacDonald, K.C. 1994. Socio-economic diversity and the origins of cultural complexity alongthe Middle Niger (2000 BC to AD 300). PhD diss., University of Cambridge.

MacDonald, K.C. 1996. Tichitt-Walata and the Middle Niger: evidence for cultural contact inthe second millennium BC. In Aspects of African archaeology: Papers from the 10th Congressof the Pan-African Association for Prehistory and Related Studies ed. G. Pwiti and R. Soper,429�440. Harare: University of Zimbabwe Publications.

MacDonald, K.C. 1998. Before the Empire of Ghana: Pastoralism and the Origins of CulturalComplexity in the Sahel. In Transformations in Africa: essays on Africa’s later past, ed. G.Connah, 71�103. London: Cassell/Leicester University Press.

MacDonald, K.C. 1999. Invisible pastoralists: an inquiry into the origins of nomadicpastoralism in the West African Sahel. In The prehistory of food: Appetites for change ed. C.Gosden and J. Hather, 333�349. London: Routledge.

MacDonald, K.C. and A. Schmidt. 2004. The Faıta Facies and the ‘obscure millennium’ of theMiddle Niger. In Acts of the XIth Congress of the Pan-African Association Prehistory andRelated Fields, Bamako, February 07-12, 2001, ed. K. Sanogo and T. Togola, 222�229.Bamako: Soro Print Color.

MacDonald. K.C., R. Vernet, D.Q. Fuller and J. Woodhouse. 2003. New Light on the TichittTradition: A preliminary report on survey and excavation at Dhar Nema. In ResearchingAfrica’s past: New contributions from British archaeologists ed. P. Mitchell, A. Haour and J.Hobart, 73�80. Oxford: Oxford University School of Archaeology.

MacDonald, K.C., R. Vernet, M. Martinon-Torres and D.Q. Fuller.. 2009. Dhar Nema: Fromearly agriculture to metallurgy in southeastern Mauritania. Azania: Archaeological Researchin Africa 44: 3�48.

Manning, K.M. 2011. Potter communities and technological tradition in the Lower TilemsiValley, Mali. Azania: Archaeological Research in Africa, 46, 1: 70�87.

Mauny, R. 1972. Contributions a l’inventaire de la ceramique neolithique d’AfriqueOccidentale. In Actes du 6e Congres Panafricain de Prehistoire, Dakar 1967, ed. H-J Hugot,72-79. Chambery: Les Imprimeries Reunies.

Mayor, A. 2011. Impressions de vanneries et technique du martelage sur forme concave:Anthropologie et histoire d’une technique dans la Boucle du Niger. Azania: ArchaeologicalResearch in Africa, 46, 1: 88�109.

McIntosh, S.K. and R.J. McIntosh. 1980. Prehistoric investigations in the region of Jenne, Mali:Part I. Oxford: British Archaeological Reports

McIntosh, S.K. (ed.) 1995. Excavations at Jenne-Jeno, Hambarketolo, and Kaniana (InlandNiger Delta, Mali), the 1981 season. Berkeley: University of California Press.

Munson, P.J. 1971. The Tichitt Tradition: a late prehistoric occupation of the southwesternSahara. PhD diss., University of Illinois at Urbana-Champaign.

Munson, P.J. 1976. Archaeological data on the origins of cultivation in the SouthwesternSahara and their implications for West Africa. In Origins of African plant domestication, ed.J.R. Harlan, J.M.J. De Wet and A.B. Stemler, 187�209. The Hague: Mouton.

Munson, P.J. 1980. Archaeology and the prehistoric origins of the Ghana Empire. Journal ofAfrican History 21: 457�466.

O’Brien, M.J. and R.L. Lyman. 1999. Seriation, stratigraphy and index fossils: The backbone ofarchaeological dating. New York: Springer.

68 K.C. MacDonald

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014

O’Brien, M.J. and R.L. Lyman. 2000. Applying evolutionary archaeology: A systematicapproach. New York: Kluwer Academic/Plenum.

Ould-Khattar, M. 1995a. La fin des temps prehistoriques dans le sud-est Mauritanien. PhDdiss., Universite de Paris-I, Pantheon-Sorbonne.

Ould Khattar, M. 1995b. Les sites Gangara, la fin de la culture de Tichitt et l’origine deGhana. Journal des Africanistes 65: 31�41.

Person, A., T. Ibrahim, H. Jousse, A. Finck, C. Alabret, L. Garenne-Morot, V. Zeitoun, J-F.Saliege and S. Ould M’Heimam. 2004. Environnement et marquers culturels en Mauritaniesud-orientale: le site Bou Khzama (DN4), premiers resultants et approache biogeochimique.In Du nord au sud au Sahara: Cinquante ans d’archeologie francaise en Afrique de l’Ouest etau Maghreb, ed. A. Bazzana and H. Bocoum, 195�213. Paris: Sepia.

Person, A., H. Jousse, A-F. Maurer and T. Vallette 2006. Les sites du Neolithique final duDhar Nema (Mauritanie): Relations peuplement � environnement. In Senegalia: Etudes surle patrimoine Ouest-Africain, hommage a Guy Thilmans, ed. C. Descamps and A. Camara,297�307. Saint-Maur: Sepia.

Schmidt, A. 2005. Metaux. Recherches archeologiques a Dia dans le Delta interieur du Niger(Mali): Bilan des saisons de fouilles 1998-2003, ed. R. Bedaux, J. Polet, K. Sanogo, and A.Schmidt, 257-262. Leiden: CNWS Publications.

Schmidt, A., N. Arazi, K. MacDonald, F. Cosme and R. Bedaux. 2005. La poterie. Recherchesarcheologiques a Dia dans le Delta interieur du Niger (Mali): Bilan des saisons de fouilles1998-2003, ed. R. Bedaux, J. Polet, K. Sanogo, and A. Schmidt, 191�256. Leiden: CNWSPublications.

Shennan, S.J. and J.K. Wilkinson. 2001. Ceramic style change and neutral evolution: A casestudy from Neolithic Europe. American Antiquity 66: 577�593.

Spaulding, A.C. 1953. Statistical techniques for the discovery of artifact types. AmericanAntiquity 18: 305�313.

Takezawa, S. and M. Cisse. 2004. Domestication des cereales au Mema, Mali, In Acts of theXIth Congress of the Pan-African Association Prehistory and Related Fields, Bamako,February 07-12, 2001, ed. K. Sanogo and T. Togola, 105�121. Bamako: Soro Print Color.

Togola, T. 2008. Archaeological investigations of Iron Age sites in the Mema Region, Mali(West Africa). Oxford: British Archaeological Reports.

Ucko, P. 1995. Introduction: Archaeological interpretation in a world context, In Theory inarchaeology: A world perspective, ed. P. Ucko, 1�28. London: Routledge.

Veit, U. 1989. Ethnic concepts in German prehistory: A case study on the relationship betweencultural identity and archaeological objectivity, In Archaeological approaches to culturalidentity, ed. S. J. Shennan, 35�56. London: Unwin Hyman.

Vernet, R. 1993. Prehistoire de la Mauritanie. Nouakchott: Centre Culturel Francais A. deSaint Exupery-Sepia.

Willey, G.R. 1945. Horizon styles and pottery traditions in Peruvian archaeology. AmericanAntiquity 10: 49�56.

Betwixt Tichitt and the IND 69

Dow

nloa

ded

by [

Uni

vers

ity o

f N

ew H

amps

hire

] at

08:

51 1

9 O

ctob

er 2

014