Ethno-archaeometry in eastern Senegal: The connections between raw materials and finished ceramic products

(1)

Article

Reference

Ethno-archaeometry in eastern Senegal: The connections between raw materials and finished ceramic products

CANTIN, N., MAYOR, Anne

Abstract

For an ethnoarchaeological study of the ceramic traditions in the Falémé Valley (eastern Senegal),we have developed an ethnoarchaeometric approach to describe the variability in the clayeymaterials used by the potters and to stimulate discussion of the connections between raw materials and finished products. The aim is to test archaeometricmodels based on the identification and interpretation of compositional groups and ultimately, resolve archaeological questions regarding the pottery from excavations in the region. Using a data set of 47 specimens of rawmaterials, temper and finished products resulting fromdocumented production sequences, our study demonstrates the range of clayeymaterials used by potters in the FaléméValley and their possible identification in finished products. In addition, by systematically comparing the technical behavior with the compositional groups for the stages of the production sequence that vary (raw material procurement, clay preparation and firing), this study also shows the limits for interpretation of compositional groups in term of social-cultural aspects.

CANTIN, N., MAYOR, Anne. Ethno-archaeometry in eastern Senegal: The connections between raw materials and finished ceramic products. Journal of Archaeological Science: Reports , 2018, vol. 21, p. 1181-1190

DOI : 10.1016/j.jasrep.2017.01.015

Available at:

http://archive-ouverte.unige.ch/unige:104159

Disclaimer: layout of this document may differ from the published version.

(2)

Ethno-archaeometry in eastern Senegal: The connections between raw materials and ﬁ nished ceramic products

N. Cantin

^a,

⁎ ^{, A. Mayor}

^b

aIRAMAT-CRP2A, UMR 5060 CNRS, Université Bordeaux Montaigne, France

bLaboratoire Archéologie et Peuplement de l'Afrique (APA), Unité d'anthropologie du Département de génétique et évolution de l'Université de Genève, Switzerland

a b s t r a c t a r t i c l e i n f o

Article history:

Received 5 July 2016

Received in revised form 22 December 2016 Accepted 9 January 2017

Available online xxxx

For an ethnoarchaeological study of the ceramic traditions in the Falémé Valley (eastern Senegal), we have developed an ethnoarchaeometric approach to describe the variability in the clayey materials used by the potters and to stimulate discussion of the connections between raw materials andﬁnished products. The aim is to test archaeometric models based on the identiﬁcation and interpretation of compositional groups and ultimately, resolve archaeological questions regarding the pottery from excavations in the region.

Using a data set of 47 specimens of raw materials, temper andfinished products resulting from documented production sequences, our study demonstrates the range of clayey materials used by potters in the Falémé Valley and their possible identification infinished products. In addition, by systematically comparing the technical behavior with the compositional groups for the stages of the production sequence that vary (raw material procurement, clay preparation andfiring), this study also shows the limits for interpretation of compositional groups in term of social-cultural aspects.

Keywords:

Contemporary potters Chaînes opératoires Clayey sediments Ethnoarchaeometry West Africa

1. Introduction

Ceramic analyses generally aim to reconstruct the production sequences, from the selection of raw materials tofiring the vessel. This is a necessaryfirst step to understand the socio-cultural, economic or ritual meanings of the compositional groups observed. However, it is often difficult to differentiate the influence of the different stages of production on the composition offinished products, not including tapho- nomic factors that make the problem even more difficult. As already demonstrated, pottery analyses are not necessarily informative about the source of the raw material, since clayey sediments can undergo different treatments, and various added tempers may change the composition of the paste (Arnold et al., 1991). Similarly, it is not always possible to determine thefiring procedures and conditions from thefinished products (Gosselain, 1992; Livingstone, 2001).

Therefore, it is useful to compare data obtained through the observa- tion of modern pottery-practices and data gained through laboratory analyses to be able to better understand past behavior beyond what can be inferred by common sense and beyond the abusive interpretation of analytical results. Unfortunately, references linking descriptions of production sequences and analytic data ofﬁnished products are still sorely lacking. In this context, the objective of our research is to combine

two approaches–ethnoarchaeology and archaeometry–to challenge the interpretation of compositional groups identiﬁed in the laboratory.

This strategy can be designated by the term ethno-archaeometry, a termﬁrst proposed in the early 1990s (Gosselain, 1992) but increasing- ly used since the start of the 21st century (for exampleBuxeda et al., 2003; Cau Ontiveros et al., 2015).

In this paper, we will compare compositional groups obtained in laboratory to ethnographic variability of ceramic practices observed among current women potters of the Falémé Valley in eastern Senegal, and discuss correlations and discrepancies.

2. Methodology and data set 2.1. Context

This study was conducted as part of the international research program“Human settlement and paleoenvironment in Africa”led since 2012 in eastern Senegal (P.I. Prof. E. Huysecom, University of Geneva).¹ One of its aims is to construct interpretive reference databases of use to archaeologists through the ethnoarchaeological study of craft practices like pottery, iron metallurgy and architecture.

During threeﬁeldwork seasons undertaken from 2012 to 2014 in the Falémé Valley (Fig. 1), one of us (A. Mayor) and Ndèye Sokhna Guèye² Journal of Archaeological Science: Reports xxx (2017) xxx–xxx

⁎ Corresponding author.

E-mail addresses:[email protected](N. Cantin), [email protected](A. Mayor).

1www.ounjougou.org.

2Our dear colleague from UCAD in Dakar sadly died accidently in June 2014.

JASREP-00784; No of Pages 10

Contents lists available atScienceDirect

Journal of Archaeological Science: Reports

j o u r n a l h o m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / j a s r e p

(3)

documented ceramic traditions and their connections with apprenticeship networks, settlement history and the establishment of pre-colonial kingdoms (Huysecom et al., 2013, 2014, 2015). A fourth mission in 2015 (Huysecom et al., 2016) including both of the authors allowed us to complete the documentation and build upon the downstream sector of the valley studied twenty years ago byA. Gelbert (2003).

2.2. Fieldwork

In theﬁeld, the method has consisted in working with the potters, collecting their raw materials from deposits, geo-referencing and sampled them for laboratory analyses. We have also documented the treatment of the clayey materials and sampled the tempers used to prepare the paste. This was followed by coded description and photography of the different stages of the production sequences, from shaping toﬁring,

as well as the identiﬁcation of tools and hand movements, according to a methodology practiced previously in Mali (Gallay et al., 1998; Mayor, 2010, 2011). Finally, one or two examples of theﬁnished products were acquired.

Monitoring of the complete process takes time (several days) and requires that the potter is active at the time of our visit, something that is becoming rare in the Falémé Valley. Another difﬁculty resulted from the fact that our research strategy aimed to cover the entire Falémé Valley along its Senegalese course (Fig. 1). This extensive strategy had the advantage of enabling comparison of craft practices linked to different apprenticeship and matrimonial networks, in different ethnolinguistic groups, but required a high degree of mobility, resulting in relatively short periods spent in each of the villages. These constraints explain the difﬁculties in collecting data in the ideal situation described above, and why it was necessary to complement our data set with Fig. 1.Location of clay sources used by the potters studied in the Falémé Valley (eastern Senegal), sampled for analysis.

(4)

incomplete production sequences that included only the raw material and aﬁnished product (new or used) previously made by the potter.

The ethnographic data set results fromfieldwork with 56 potters in 26 villages. Samples for laboratory analysis come only from 16 of these 26 villages. They represent six complete production sequences (clayey material, grog andfinished products), seven incomplete production sequences (clayey material andfinished products) and three additional isolated raw materials. In total, we now have available an initial data set of 47 samples coming from 15 raw material sources (Table 1). We could locate 14 deposits where we have been able to go with potters to collect clayey sediments. In one case (Toumboura), the potter didn't have time to go to the clay source, so we sampled clay from her own reserve without having coordinates.

2.3. Laboratory analyses

Several commonly used methods for the study of ancient pottery have been applied to describe the raw materials andﬁnished products.

For each specimen, in the lab, a representative homogenized sample

(20 g) of raw clayey material was taken. This was also done for the temper and pottery. The powders obtained after grinding were divided into two fractions: one was set aside for mineralogical analysis and the other ignited for 1 h at 950 °C for chemical analyses and determination of loss- on-ignition. For the pottery, a section was set aside to obtain thin-sections for petrographic analyses.

The mineral composition of the raw materials (clayey sediment and grog) was determined by X-ray diffraction analysis of the powders. The equipment used was a BRUKER, D8 Advance, conﬁgured in a Bragg- Brentano geometry with Cu-K_αradiation and equipped with the PSD Lynxeye detector. Measurements were acquired at 40 kV, 40 mA from 3 to 60° with a step of 0.01° per second. The clay minerals were separated by sedimentation (on unmilled sediment) after sieving at 60μm. The fractionsb10μm were placed on glass slides. These oriented specimens underwent different treatments (ethylene glycol [EG], heated at 350 °C and 550 °C) to identify the different clay minerals (Moore and Reynolds, 1989). Analysis of the diffractograms was done with the EVA program using the ICDD PDF-2004 database. The mineral composition of the pottery was determined by XRD on the powders under the same conditions

Table 1

List of specimens analyzed: clay raw materials, temper andﬁnished products.

Deposit Village Clayey materials Grog Vegetal temper⁎ Ceramic Lab reference Geographical code (see map)

Balakonko Balakonko x BDX17188 M06

Dalaoulé Dalaoulé x BDX17374 M13

Faraba/Fodé Bineya Faraba x BDX17183 M02

Fodé-Bineya x BDX17189

x BDX17189-B

x BDX17192

x x BDX17193

x BDX17194

Farabana Farabana x BDX17375 M15

x BDX17376

Goundafa Goundafa x BDX17186 M04

x BDX17369

Kenieba Kenieba x BDX17366 M12

x x BDX17367

x BDX17368

Khossanto Khossanto x BDX17191 M07

x BDX17195

Koussan Koussan x BDX17360 M11

x BDX17361

Lalli/Sitabanta Sitabanta x BDX17352 M09

x BDX17353

x BDX17354

x BDX17355

Madina Foulbé Madina Foulbé x BDX17357 M10

x BDX17358

x BDX17359

Nayé Bellé x BDX17378

x x BDX17379

x BDX17380

Nayé x BDX17381 M14

x BDX17382

x BDX17383

x BDX17384

Sadatou/Falala Falala x BDX17185 M05

x BDX17187

x BDX17196

Sansandé Sansandé x BDX17184 M03

x BDX17370

x BDX17371

x BDX17372

x BDX17373

Sitabanta x BDX17356

Toumboura Toumboura x BDX17362

x BDX17363

x BDX17364

x BDX17365

Tourokhoto Tourokhoto x BDX17190 M08

⁎ Vegetal tempers have not been processed yet.

N. Cantin, A. MayorJournal of Archaeological Science: Reports xxx (2017) xxx–xxx 3

(5)

as for the raw materials, and by petrographic analysis of thin-sections using a LEICA DM2500P petrographic microscope coupled with an image acquisition system.

Grain size analysis of the raw materials was made by laser granulometry, the results of which are presented in grain size classes following the size categories proposed byKonert and Vandenberghe (1997) (clays [b7μm]; silts [7–63μm]; sands [63μm-2 mm]). The grain size distribution was obtained with a Horiba LA950 laser diffraction analyzer using 2 g of sample previously treated with hydrogen peroxide and sodi- um hexametaphosphate for 12 h, and dispersed into water with 60 s of ultrasoniﬁcation in the micro-granulometer. The Mie solution to the Max- well equation (Mie, 1908) was used to calculate the particle size distribution from the measured scattering of light using refractive indices of 1.333 for water and 1.55–0.01 for sediment.

Finally, the geochemical composition of the raw materials and pottery was determined by SEM (Jeol 6460 LV), equipped with an EDS detector (Oxford Instruments X-Max) to quantify major and minor elements. These measurements were made on powder pellets 8 mm in diameter and 3 mm thick, in low vacuum mode at 20 Pa. The results presented are the mean of four measurements on 300 × 400μm²area per pellet. We also used an ED-XRF equipment (SEIKO, SEA6000 VX) to determine trace elements. The measurements were made on the same pellets using calibration curves established with international geostandards.

3. Results

3.1. Ethnographic results

In the Falémé Valley, pottery is made only by casted women, daughters and wives of blacksmiths. Faced with the economic attraction of gold mining and the influx of cheap imported plastic and metal con- tainers, local knowledge is now disappearing. Pottery-making is today practiced mainly by older women who often work by commission and no longer teach their daughters. Only a few functions, such as water jars, couscous steamers and incense-burners, are still required, while the use of clay cooking pots has largely been abandoned several decades ago. Nowadays, potters practice their craft primarily for the inhabitants of their village of residence and a few neighboring villages. There are no weekly markets that would favor the circulation of pottery over longer distances. Only water jars for marriages reflect more distant and diver- sified provenances, paralleling the pattern of marriage networks.

The clayey raw materials are collected by the potters in different geographic and geological contexts (Fig. 1), at distances ranging from 0.1 to 18 km from their workshops, but most commonly between 1 and 3 km. With few exceptions, communities of potters have been exploiting the same sources for generations. These sources can be alluvia deposited during the Pleistocene and Holocene by the Falémé River or by its main tributary the Kobakoye, or surface sediments associated with seasonal streams or termite mounds. It should be noted that termite activity can bring deeply buried sediments to the surface (Kristensen et al., 2015). One source, clearly different from the others, is an extraction site exploited by mine shafts and subterranean galleries several meters long.

We can therefore isolate four groups of raw material and patterns of exploitation.

1. River bank deposits and gullies are exploited by potters living close to the Falémé (Goundafa, Sansandé, Toumboura, Sitabanta, Nayé, Tourokoto) or Kobakoye (Madina-Foulbé, Dalaoulé), and generally every community has its own source, close to the village, exploited throughout the dry season. Sometimes the source is located at the site of an ancient ruined village previously settled near the river (Sitabanta, Nayé). In one case (Madina Foulbé), potters have faced since 2014 the systematic breakage of their products duringﬁring, and consider now that their usual raw material has become

unsuitable for making pottery. Likely following the exhaustion of the source used for several generations, they mustﬁnd a new source to be able to continue their craft.

2. Surface sediments taken from seasonal streams are exploited by peo- ple living far from the Falémé river (Balakonko, Farabana, Kéniéba, Khossanto), along a general pattern of one source for one village.

3. Both termite mounds sites (Koussan, Sadatou-Falala) are shared by potters from two or even more villages, in areas at a distance from the Falémé. It is considered good quality raw material and distances between sources and workshops are the highest.

4. The unique subterranean site discovered is exploited jointly by the artisans of two villages (Fodé Bineya and Faraba), who respect many rites and taboos associated with this site. Clayey sediment is extracted once a year during the dry season after a sacriﬁce, and stored by each potter in dedicated places close to the shafts.

In all the communities of potters encountered in the Falémé Valley, the paste is made of clayey sediments to which sieved grog and crushed vegetal temper (fonio straw or dried grass) are added. The clay is soaked in water for two to three days prior to use, without speciﬁc preparation (although the largest impurities are removed by hand). The micaceous material from Faraba-Fodé Bineya is the only exception. This clay is stored unprocessed, dry, then crushed in a mortar and sieved just before use to obtain a powder (Fig. 2). Water is added and both tempers are then incorporated into the plastic mixture.

Then, two groups of paste preparation can be isolated, one including all the producing villages soaking the clay, and the other including only the villages of Faraba and Fodé Bineya, who crush and sieve their raw material. All the potters are using the same combination of sieved grog and crushed vegetal temper.

Concerning the shaping techniques, all of the potters questioned, regardless of their cultural and linguistic identity, use the same technique to shape pots, that is molding on a convex form over an upturned con- tainer with a round base, the surface of which is sprinkled with grog powder. After drying, the base is removed from the mold and turned over, the rim dampened and the walls built by superimposition of large highly plastic coils. Once dry, the pottery is scraped and coated prior toﬁring with a sticky vegetal maceration made of different species depending on the environment. According to the potters, this coating makes the pottery more“solid”and more“beautiful”. Only the tools used for shaping and decoration vary slightly between communities.

Unlike shaping techniques, which are highly uniform,firing techniques show variability (Fig. 3). In the villages located upstream, pots are arranged in a bonfire on the ground surface. Wood and bark are used for fuel. Firing is interrupted after 30 to 60 min and is followed by a post-firing process: the vessels are taken out still red from the

Fig. 2.Paste preparation in Fodé Bineya, showing the processes of pounding and sieving the clayey material before adding the vegetal temper (photo A. Mayor).

(6)

ﬁre and sprayed with a bark (Ximenia americanaorParkia biglobosa) or crushed fruit maceration, giving them a dark brown to black color, due to the partial reduction of the surface.

Another procedure is used in the villages located downstream. The pots are arranged in a depression in the ground and covered with typi- cally slow-burning fuel. This can be cow dung only, or a mixture of dung and variable proportions of other kinds of fuel, such as kitchen waste, bark, wood, wood shavings, corn or millet rachis, or corn or millet stalks.

In some cases, the fuel mixture does not include dung at all. While most potters use dung and corn rachis when available, the other kinds of fuel easier tofind are used opportunistically. Finally, the bonfire is covered with straw to help start thefire, which is set in the evening once the wind has dropped. Thefiring is characterized by a long soaking time of several hours, followed by slow cooling. The pots are recovered the next morning, or even the afternoon if still hot. The length of the process depends on several parameters, including the number of pots, the kind of fuel, the presence or absence of wind, etc. In this procedure, as cooling takes place in an oxidizing atmosphere, the pottery is mainly orange, with gray-black reduction spots of different size. Sometimes, in a single bonfire, pots may be fully orange, while others look fully black.

In summary, our observations allow us to distinguish two different technical procedures, and the use of a wide variability of fuels for one of them.³

Looking at ethnographic pottery practice along the Falémé River, it is important to note that there aren't any homogeneous traditions with distinctive traits characterizing all the steps of the“chaîne opératoire”. There are rather different patterns for raw material sources, paste preparation, shaping techniques andﬁring procedures, which can be correlated to environment, technical choices and socio-cultural history.⁴

3.2. Archaeometric results

The results of the grain size analysis of the sampled clayey sediments, presented in a ternary diagram (Fig. 4), show a distribution in two groups. Theﬁrst includes the samples from the site of Fodé-Bineya, very rich in mica crystals visible to the naked eye and with the highest percentage of sand. About 85% of the particles are larger than 63μm.

The second group includes all of the other sediments, which have b20% sand and a clay fraction of 20–40%; the silt fraction is dominant in these samples. Among the raw materials with the higher percentage of silt, these from Madina-Foulbé should be noted, withN40%.

The mineralogical analysis (Table 2) shows that all the sediments contain a signiﬁcant proportion of kaolinite, a mineral common in these climatic contexts (Ekosse, 2010; Galan, 2006). Ca-smectite minerals are also found at all of the sources, except for Nayé, located at an abandoned village on the banks of the Falémé.

For analytic indicators relating to ceramicﬁring, the results can be divided into three different groups. In theﬁrst group, we see the

persistence of kaolinitic-type clay minerals, which normally disappear at 550 °C (Maggetti and Rossmanith, 1981). This was observed at Fodé Bineya, Khossanto, Keniéba and Toumboura (Table 2). For these sites, loss-on-ignition measurements indicate elevated results, similar to those obtained for the clayey raw materials, confirming the presence of clay minerals in the pottery and suggesting that the maximalfiring temperature was low. Moreover the persistence of vegetal temper in some of these pots confirms a low temperature.

At the opposite extreme, in the second group,ﬁring was sufﬁciently high to transform the kaolinite into mullite, a mineral known to appear at temperatures around 1100 °C (Brindley and Lemaitre, 1987). This was observed for the pottery at Bellé, Nayé, Falala and Sansandé (Table 2). These results are consistent with those obtained for loss-on- ignition, near zero for these sites, indicating complete dehydroxylation of the clays.

Finally, the results for the intermediary group indicate that kaolinite clay minerals had disappeared but were not transformed into mullite, and therefore indicate a maximal ﬁring temperature higher than 550 °C but lower than 1100 °C.

Geochemical analysis (Appendix 1) indicates that most of the clayey materials are rich in iron (N7%, and evenN10% at Nayé and Khossanto), except Faraba-Bodé Bineya (only 2–4%). In addition, signiﬁcant variability exists between the clay material compositions in the region (Fig. 5).

The compositions of materials from the bank sources of the Falémé and its tributary Kobakoye form a relatively uniform group (with a ratio of silica/alumina around 3.0, and a high level of Cr). The sample from Keniéba, collected in Holocene deposits of a seasonal stream has a slightly higher amount of MgO (1.8–2%). The compositions of sources located far from the river are more varied. The subterranean site shared by the potters from Fodé Bineya and Faraba shows a higher amount of potassium (N4%), mainly in the mica component. The samples collected at the termite mounds shared by the potters of Sadatou and Falala are very rich in silicon (72%) and poor in aluminum (15%). This low percentage of aluminum is not related to the low percentage of clay particles (37–38%) but rather to a strong component of quartzitic silt elements (demonstrated by XRD). The composition of these sediments also diverges very clearly from the data set studied. Finally, the termite mound at Koussan shows the clay materials highest in CaO (1.5–2%).

The source of this calcium is not related to the presence of plagioclase feldspar, which is minor at this site, nor to the presence of calcite, which is absent in all the samples. Its presence is possibly related to the smectite group. Potassium is also higher at this site (correlated to Rb), reﬂecting the important presence of microcline feldspar.

As far as petrographic analysis is concerned, four types deﬁned by different mineral and particle size compositions can be observed among theﬁnished products (Table 3).

Type 1: The Fodé Bineya pottery is characterized by an abundance of coarse inclusions (400μm à 1 mm) that include potassium-rich alkali feldspars, muscovite mica and, to a lesser degree, quartz.

Type 2: This type includes pottery from villages both near and far from the river (Goundafa, Sansandé, Sitabanta, Nayé, Bellé, Madina Foulbé, Farabana, Kéniéba, Koussan and Khossanto). Although some in- ternal variability is observed, this type as a whole is deﬁned by an

3Aware of the high variability in ceramicﬁring temperatures (Gosselain, 1992;

Livingstone, 2001), we did not measure temperatures with a thermocouple in the ﬁeld.

4Socio-cultural and historical aspects will be analyzed elsewhere.

Fodé Bineya Khossanto Kéniéba Toumboura Farabana Madina Foulbé Koussan Sitabanta Goundafa Sansandé Falala Nayé Bellé

PROCEDURE 1 PROCEDURE 2

interrupted firing slow cooling firing

post-firing no post-firing

bonfire depression

wood, cork fuel blending of fuel without dung (or few) blending of fuel with dung dung fuel (alone or dominant with husks)

GROUP 1 GROUP 2 GROUP 3

kaolinite no kaolinite, no mullite mullite

important loss of ignition intermediate loss of ignition nearly no loss of ignition

< 550 oC intermediate temperature > 1100 oC

Fig. 3.Correspondence between ethnographic and mineralogical data forfiring procedures. Sites noted in bold correspond to observations of thefiring process on thefield whereas the others relieve from oral enquiries.

(7)

abundance of non-plastic elements in which angular quartz around 50μm dominates. Some coarser inclusions can also be seen, particularly in the products from Kéniéba and Khossanto. Few lithic elements are present.

Type 3: The Falala pottery is close to type 2, but differs in the round- edness of the quartz grains, probably reﬂecting an eolian origin given the location of the site far from the river.

Type 4: The Toumboura pottery has a low percentage of non-plastic inclusions, which is directly correlated with the results obtained by the chemical analysis (very high proportion of Al203).

In summary, while types 2, 3 and 4 are relatively similar, with some variability in the form or abundance of non-plastic elements, type 1 is signiﬁcantly different from them. For all the types, grog temper is easy to identify, even if its composition is nearly the same as that of the paste. The grog grains are generally highly variable in size within a single pot (from a few hundred microns toN2 mm in diameter), although the Keniéba and Toumboura pottery are exceptions, with two distinct size classes. Vegetal temper is identiﬁable by very visible traces in the paste and a characteristic trait of porosity. Sometimes it is still present, as seen in the pottery from Fodé Bineya.

4. Discussion 4.1. Raw material

The potters of the Falémé Valley can obtain raw materials from very different kinds of sources, and sometimes from sources requiring signif- icant investment, as in the cases of deposits very distant from the village, or even physical risks, as in the case of the mine galleries of Faraba-Fodé Bineya. Artisans are able to successfully make vessels with very different clayey materials, but not all of them are suitable.

This can be seen for example in Madina-Foulbé, where potters started unsuccessfully to collect a clay material with a very high silt content.

While the characteristics of this material are of huge consequence for the potters (all the pots broke during theﬁring process), its mineralogical and geochemical characteristics don't change its attribution to the compositional group“Falémé”, although located at the edge of the grain size distribution (Fig. 4).

Potters collect their raw material from the alluvial deposits of the river and its tributaries, and, in areas far from the river, from seasonal streams, termite mounds and a subterranean mine. Comparison of the compositional groups and the types of sources for clayey materials demonstrates a few correlations. The materials from deposits along the

Falémé and Kobakoye, used by different communities of potters at different parts of the river and on both banks, form a group of fairly consistent geochemical composition (Figs. 5 and 6). This group differs from the compositions observed in most of the sources located further from the river. There is thus a partial correspondence between source types and their analytic signature. Materials coming from seasonal streams don't represent a geochemical group but they have compositions some- what distinct from the Falémé river ones. The termite mounds material show different characteristics for each source. However, in all the analyses, the micaceous material from the subterranean mine is clearly isolated from the others, playing the role of a provenance marker (Figs. 5 and 6).

These preliminary results, still based on a small number of samples, demonstrate that if the compositional groups of unprocessed clayey materials are fairly well correlated with the different types of sources, it is nonetheless only possible in a small number of cases to use them as true provenance markers.

4.2. Paste preparation

Looking at the production sequence of pottery, the only variability observed is in the treatment of clayey sediments, which are soaked in water or crushed into powder and sieved, this latter process being used only for the material from the subterranean mine. Comparison between the compositions of unprocessed clay andﬁnished products demonstrate a good overall correlation, with some slight differences.

First, analyses of particle size shows that petrographic type 2fin- ished products are consistent with the raw materials used to make them, at least for the alluvial deposits of the Falémé and Kobakoye riv- ers, which have a high silt fraction. The particle size compositions for Falala and Toumboura pottery (types 3 and 4) are slightly different from the others, although the unprocessed clay sediments are not dif- ferentiated from the other samples. This anomaly remains difficult to explain, as it does not appear to correspond to any specific clay preparation. The general goodfit in particle size between clayey sediments and finished products can be explained ethnographically by the lack of specific processing of this raw material, apart from a short period of soaking in water. Surprisingly, the composition of the pottery from Fodé Bineya (type 1) is also very similar to that of its raw material, despite the specific processing treatment used to lessen the particularly coarse particle size of the clay by crushing and sieving it. This means that although the raw material is systematically converted to powder before being used (Fig. 2), elements of the original sand fraction and large non-plastic Fig. 4.Ternary diagram of grain size composition of the clayey raw materials and thin section in XPL of the corresponding ceramic productions (scale bar = 1 mm).

(8)

Table 2

Mineral composition obtained by XRD on powders and oriented slides for clay minerals.

quartz K feldspar Pl feldspar hematite goethite mullite spinel smectite chlorite vermiculite Interlayered illite talc kaolinite BDX17188

BDX17378 grog x x x x ? x/– x

BDX17380 ceramic x x x x

BDX17374 Dalaoulé clayey raw materials x x xx x x xx

BDX17185 clayey raw materials xxx x x x x x xxx

BDX17187 clayey raw materials xxx xx x x x/– xx xxx

BDX17196 ceramic xxx xx x x x/–

BDX17183 clayey raw materials x xx x x xx xx

BDX17189 clayey raw materials x xx x x xx xx

BDX17189–B crushed claye raw

materials x xx x x xx xx

BDX17192 grog x xx x xx

BDX17194 ceramic x xx x xx xx

BDX17375 clayey raw materials x xx x xx x xx

BDX17376 ceramic x xx x x

BDX17186 clayey raw materials x x x x xx

BDX17369 ceramic xx x x

BDX17366 clayey raw materials xx x x x x x xx

BDX17367 grog xx x x x x

BDX17368 ceramic xx x x x/– xx x

BDX17191 clayey raw materials xx x x x x x x x

BDX17195 ceramic xx x x x x

BDX17360 clayey raw materials xx x x/– x x x xx x

BDX17361 ceramic xx x x/– xx

BDX17357 clayey raw materials x x x x x xx

BDX17358 ceramic xx xx x x

BDX17359 ceramic xx xx x x

BDX17381 clayey raw materials x x xx x x x xx x xx

BDX17382 grog x xx x x/– x/–

BDX17384 ceramic x xx x x x

BDX17184 clayey raw materials xxx x x x x xx xxx

BDX17370 clayey raw materials xxx x x x x xx xxx

BDX17371 grog xxx x x x

BDX17373 ceramic xxx x x x

BDX17352 clayey raw materials xxx x x x x x

BDX17353 grog xxx x x x

BDX17354 grog fine xxx x x x

BDX17356 ceramic xxx x x x

BDX17362 clayey raw materials xxx x x/– x x xxx

BDX17363 grog xxx x x/– x x

BDX17365 ceramic xxx x x/– x x xxx

BDX17190 Tourokhoto clayey raw materials xxx x x x x x xx xxx

Minerals

Lab ref Clayey minerals

Site Type of material

Bellé

Goundafa

Kenieba Falala

Faraba–Fodé Bineya

Farabana

Sitabanta

Toumboura Khossanto

Koussan

Madina Foulbé

Nayé

Sansandé

Balakonko clayey raw materials x x x x x x x xx

7N.Cantin,A.MayorJournalofArchaeologicalScience:Reportsxxx(2017)xxx–xxx

Pleasecitethisarticleas:Cantin,N.,Mayor,A.,Ethno-archaeometryineasternSenegal:Theconnectionsbetweenrawmaterialsandﬁnishedceramicproducts,JournalofArchaeologicalScience:Reports(2017),http://dx.doi.org/10.1016/j.jasrep.2017.01.015

(9)

inclusions (400μm to 1 mm) are still found in the texture of the ceramic paste. These results show that petrographic analysis would not have de- tected the difference in paste preparation, although resulting from very distinct practices.

Secondly, the comparison of chemical compositions shows that the same characteristics are found in bothﬁnished products and the raw materials, despite the general use of grog. A“Falémé”group gathers all the deposits of the Falémé and the Kobakoye. The Fodé Bineya pottery is completely distinguished from this group by a very high percentage of K2O. The vessels from Keniéba and Koussan are also separated from the“Falémé”group by a slightly higher percentage of K2O, and that of Falala by its high percentage of SiO2.

The presence of grog is often viewed as a barrier to the application of chemical analysis for pottery tempered with such a material. In the Falémé valley industry, grog does not influence the chemical composition of the pottery. Regardless of the origin of raw materials, the composition of grog is not significantly different from those of clayey sediments and ceramics (Fig. 7). This confirms that grog was composed of broken sherds of local pottery made out of the same clay, as men- tioned by the potters. This phenomenon can be explained

ethnographically by the clear overlap of the production and consumption spheres and by the low mobility of the vessels given the lack of markets and the small scales production. In a more general perspective, it can be assumed that differences in the composition between grog and the plastic elements of the paste in ceramic assemblages can be considered as a useful proxy to discuss the coexistence of different traditions in the same region, the blending of products in consumption units, and the interactions between different communities producing pottery.

In brief, the analysis of theﬁnished products allows identiﬁcation of the broad types of raw materials, but not so easily the differences in clay preparations.

4.3. Firing

Several publications have shown that temperature can vary significantly from onefiring session to another, from one place to the other in the same bonfire, and even from one part of the vessel to another, comprehensively preventing inference about technology, like the kinds offiring structures or fuel used (Gosselain, 1992; Livingstone, 2001). Further, the oxidation-reduction conditions should also be

Table 3

Petrographic characteristics ofﬁred clay observed on thin-sections.

Petro type

Site Lab ref Abundance Granulometry Form Mineral and lithic non-plastic

inclusions

Organic inclusions (porosity)

Grog

1 Fodé Bineya BDX17193 +++ 400μm–1 mm +++; 100μm + Angular K-Feld (perthites), mus,qtz, Fe-ox + +/−

2 Belle BDX17380 +++ 50 +++;−150μm + Angular qtz ++ ++

Nayé BDX17384 +++ 50μm +++ Angular qtz, mus ++ ++

Farabana BDX17376 +++ 50μm +++ Angular qtz, detritic rocks + ++

Goundafa BDX17369 +++ 50–60μm +++ Angular qtz ++ ++

Sansandé BDX17373 +++ 50μm +++ Angular qtz + +

Sitabanta BDX17356 +++ 20–50μm +++ Angular qtz, K-Feld, Pl-Feld + ++

Madina Foulbé

BDX17358 +++ 50μm +++; 100–150μm + Angular qtz, pisolithes, mus ++ ++

BDX17359 ++++ 50μm +++ Angular qtz, pisolithes ++ ++

Kenieba BDX17368 +++ 30–50μm +++; 100–600μm + Angular qtz, Pl-Feld, argilite, mus ++ ++

Khossanto BDX17195 +++ 30–50μm +++; 500–800μm + Angular qtz, qtzite, silex ++ +/−

Koussan BDX17361 +++ 50μm +++ Angular qtz, Fe-ox ++ +

3 Falala BDX17196 +++ 50–200μm +++ Rounded to

sub-angular

qtz ++ ++

4 Toumboura BDX17365 ++ 50μm +++ Angular qtz, micaschistes +/− ++

0 1 2 3 4 5 6

0 0,05 0,1 0,15 0,2 0,25

SiO2/Al2O3

K2O/Al2O3

Bellé Toumboura Sitabanta Sansandé Goundafa Nayé Tourokhoto Madina Foulbé Dalaoulé Khossanto Kenieba Balakonko Farabana Falala Koussan Fode Bineya

Falémé riverTermite mounds

seasonal streams

Fig. 5.Variability in geochemical composition of ceramic products and clayey raw materials based on SEM-EDS results.

(10)

taken into account when reconstructingﬁring techniques (Maggetti et al., 2011).

In our ethnoarchaeological study, two contrasting technical procedures forfiring pottery have been identified: short interruptedfiring in a bonfire using wood and bark fuel, followed by a post-firing process, and longfiring in a pit using different slow-burning fuels including vary- ing amounts of dung, sometimes complemented with other combusti- bles, followed by slow cooling. The laboratory analyses, as far as they are concerned, evidence three groups: thefirst characterized by the persistence of kaolinite minerals, indicating a very low maximalfiring

temperatureb550 °C; the second by the presence of mullite, indicating high maximal temperaturesN1100 °C; and the third without kaolinite or mullite, indicating intermediate temperatures. The correspondence between ethnographicfiring procedures and analytic groups is again partial, with thefirst procedure associated with thefirst group, but the second procedure divided among three groups (Fig. 3). This shows the great variability in temperatures for the same procedure, which is in agreement with some earlier studies. This variability appears therefore here to be highly correlated with the kind of fuel and more specif- ically the quantity of dung used duringfiring.

Fig. 6.(a) Principal components analysis of theﬁrst and second components on the log central data of six elements (Cr, Cu, Zn, Rb, Sr, Zr) obtained by XRF analysis, for clayey raw materials and ceramic products. (b) Projection of the variables.

Fig. 7.Boxplots of Al2O3component (in %) in clayey raw materials, grog and ceramics for the different geographical sets.

(11)

Thus, inferences aboutfiring technologies are not impossible, but need cautious interpretations. In our case study, pitfiring for several hours with dung fuel allows temperatures higher than 1100 °C to be reached, a temperature unattainable by short interruptedfiring with wood and bark. But regardless of the procedure used it is important to be aware that all thefinished products arefired, functional and do not appear to have significantly different use-lives. As with earlier studies, this example serves as a warning: interpreting the three analytical groups as ceramic traditions would have been erroneous.

5. Conclusion

In the context of an ethnoarchaeological study conducted since 2012 on ceramic traditions in the Falémé Valley, in eastern Senegal, we have developed an ethno-archaeometric approach to describe the variability in the clayey materials used by the potters and to stimulate discussion about the connections between raw materials andﬁnished products.

The aim was to challenge archaeometric models based on the interpretation of compositional groups and ultimately, resolve archaeological questions regarding the archaeological pottery being discovered during excavations in the region. This article has focused on issues about ceramic production, while the sociocultural meanings of the compositional groups will be addressed in a separate publication.

Despite a relatively small initial data set, the present study, by systematically comparing the technical behavior of the potters with the compositional groups obtained in the laboratory, has clearly shown a number of limits in the interpretation of compositional groups.

For example, the results of grain size analysis and geochemistry do not match clearly with all the different types of raw material sources, even if the raw material from the subterranean mine is a clear provenance marker and if ceramics produced with the alluvial deposit from the Falémé river have a common geochemical signature. Similarly, temperature mineral groups match only partially withfiring procedures observed in thefield, due to the high temperature variability for a same procedure, depending namely of the proportion of cattle dung used as a combustible. On the other hand, the study has shown that compositions of raw material andfinished products are very coherent despite the overall use of grog, both having been made out of the same clay, a result that can be understood in the situation of small scale productions, the absence of markets and the rather low mobility of pots.

Further analysis based on a larger data set coming from the Bedik Country, collected in 2016, should enable these promising initial results to be strengthened and completed with a discussion about connections with socio-cultural aspects.

Geochemical composition of raw materials andﬁred clay determined by SEM-EDS on pressed pellets for majors and minors elements;

traces determined by ED-XRF are presented in a supplementary data associated with this article can be found in the online version, athttp://dx.

doi.org/10.1016/j.jasrep.2017.01.015.

Acknowledgements

Authors would like to thank several institutions for their help and funding: Swiss National Fund for scientiﬁc research (SNF nr 101211_163022); Swiss-Liechtenstein Foundation for archaeological research abroad (SLSA); Laboratory Archaeology and Population in Afri- ca (APA) at University of Geneva; IFAN and University Cheikh Anta Diop (UCAD) of Dakar; CNRS and University of Bordeaux Montaigne; the LaScArBx Labex, a research program supported by the French National Research Agency (ANR-10-LABX-52); Région Aquitaine. Thanks also to the PACEA laboratory (UMR5199) for the laser diffraction analyses

and to A. Queffelec in particular. Thanks also to B. Spiteri for realization of thin sections at IRAMAT-CRP2A (UMR5060). Finally, many thanks to all the potters in the Falémé Valley for their kindness, to Rebecca Miller for her help with translation and to both of the reviewers for their useful comments.

References

Arnold, D.E., Neff, H., Bishop, R., 1991.Compositional analysis and“Sources”of pottery: an ethnoarchaeological approach. Am. Anthropol. 93, 70–90.

Brindley, G., Lemaitre, J., 1987.Thermal oxidation and reduction reactions of clay minerals. In: Newman, A.C.D. (Ed.), Chemistry of Clays and Clay Minerals, pp. 319–370.

Buxeda, I., Garrigos, J., Cau Ontiveros, M.A., Kilikoglou, V., 2003.Chemical variability in clays and pottery from a traditional cooking pot production village: testing assump- tions in Peruela. Archaeometry 45 (1), 1–17.

Cau Ontiveros, M.A., Montana, G., Tsantini, E., Randazzo, L., 2015. Ceramic ethnoarchaeometry in western Sardinia: production of cooking ware at Pabillonis.

Archaeometry 57 (3), 453–475.

Ekosse, G.-I.E., 2010.Kaolin deposits and occurrences in Africa: geology, mineralogy and utilization. Appl. Clay Sci. 50, 212–236.

Galan, E., 2006.Genesis of clay minerals. In: Bergaya, F., Theng, B.K.G., Lagaly (Eds.), Hand- book of Clay Science, pp. 1129–1162.

Gallay, A., Huysecom, E., Mayor, A., 1998.Peuples et céramiques du Delta intérieur du Niger (Mali). Terra archaeologica III. von Zabern, Mainz.

Gelbert, A., 2003.Traditions céramiques et emprunts techniques dans la vallée duﬂeuve Sénégal. Maison des sciences de l'Homme–Episthèmes, Paris.

Gosselain, O.P., 1992.Bonﬁre of the enquiries. Potteryﬁring temperatures in archaeology:

what for? J. Archaeol. Sci. 19, 243–259.

Huysecom, E., Jeanbourquin, C., Mayor, A., Chevrier, B., Loukou, S., Canetti, M., Diallo, M., Bocoum, H., Guèye, N.-S., Hajdas, I., Lespez, L., Rasse, M., 2013.Reconnaissance dans la vallée de la Falémé (Sénégal oriental): la 15e année de recherche du programme international“Peu-plement humain et paléoenvironnement en Afrique de l'Ouest”.

Jahresbericht SLSA 2012. Zurich et Vaduz: Fondation Suisse-Liechtenstein pour les recherches archéologiques à l'étranger, pp. 25–112.

Huysecom, E., Loukou, S., Mayor, A., Jeanbourquin, C., Chaix, L., Chevrier, B., Ballouche, A., Bocoum, H., Guèye, N.-S., Timpoko-Kiénon-Kaboré, H., Rasse, M., Tribolo, C., 2014.

Vallée de la Falémé (Sénégal Oriental) et Parc National des îles Eotilé (Côte d'Ivoire):

la 16ème année de recherche du programme “Peuplement humain et paléoenvironnement en Afrique”. Jahresbericht SLSA 2013. Zürich et Vaduz:

Fondation Suisse-Liechtenstein pour les recherches archéologiques à l'étranger, pp. 113–176.

Huysecom, E., Chevrier, B., Mayor, A., Canetti, M., Chaix, L., Garnier, A., Guèye, N.S., Lespez, L., Loukou, S., Pollarolo, L., Rasse, M., Ballouche, A., Bocoum, H., Camara, A., Sankaré, F., Guardiola, M., Guindo, N., Hajdas, I., Jeanbourquin, C., Sanogo, K., Tribolo, C., 2015.La construction du cadre chronologique et paléoenvironnemental de la moyenne vallée de la Falémé (Sénégal oriental): les résultats de la 17ème année d'activité du programme international“Peuplement humain et paléoenvironnement en Afrique”.

Jahresbericht SLSA 2014. Zürich et Vaduz: Fondation Suisse-Liechtenstein pour les recherches archéologiques à l'étranger, pp. 61–134.

Huysecom, E., Chevrier, B., Mayor, A., Pelmoine, T., Cantin, N., Chaix, L., Leplongeon, A., Lespez, L., Loukou, S., Rasse, M., Truffa Giachet, M., Bocoum, H., Camara, A., Hajdas, I., Lebrun, B., Tribolo, C., 2016.Transitions et continuités dans la vallée de la Falémé (Sénégal): résultats de la 18ème année de recherche du programme international

“Peuplement humain et paléoenvironnement en Afrique”. Jahresbericht SLSA 2015.

Zürich et Vaduz: Fondation Suisse-Liechtenstein pour les recherches archéologiques à l'étranger, pp. 103–160.

Konert, M., Vandenberghe, J., 1997.Comparison of laser grain size analysis with pipette and sieve analysis: a solution for the underestimation of the clay fraction. Sedimen- tology 44, 523–535.

Kristensen, J.A., Thomsen, K.J., Murray, A.S., Buylaert, J.-P., Jayn, M., Breuning-Madsen, H., 2015.Quantiﬁcation of termite bioturbation in a savannah ecosystem: application of OSL dating. Quat. Geochronol. 30 (B), 334–341.

Livingstone, S.A., 2001.Bonﬁre II: the return of potteryﬁring temperatures. J. Archaeol.

Sci. 28, 991–1003.

Maggetti, M., Rossmanith, M., 1981.Archaeothermometry of kaolinitic clays. Revue d'Archéométrie 5, 185–194.

Maggetti, M., Neururer, C., Ramseyer, D., 2011.Temperature evolution inside a pot during experimental surface (bonﬁre)ﬁring. Appl. Clay Sci. 53, 500–508.

Mayor, A., 2010.Ceramic traditions and ethnicity in the Niger Bend, West Africa.

Ethnoarchaeology 2 (1), 5–48.

Mayor, A., 2011.Traditions céramiques dans la boucle du Niger. Ethnoarchéologie et histoire du peuplement au temps des empires précoloniaux, Frankfurt am Main, Af- rica Magna Verlag. J. Afr. Archeol. Monogr. Ser. 7.

Mie, G., 1908.Beiträge zur Optik trüber Medien, speziell kolloidaler Metallösungen. Ann.

Phys. 25, 377–445 (Leipzig).

Moore, D.M., Reynolds, J.R.C., 1989.X-ray Diffraction and the Identiﬁcation and Analysis of Clay Minerals. Oxford University Press, New York.

Ethno-archaeometry in eastern Senegal: The connections between raw materials and finished ceramic products

Article

Reference