(PDF) Materialising Ancestral Madang: Pottery Production and Subsistence Trading on the Northeast Coast of New Guinea

Permanent link to open access version: http://hdl.handle.net/10523/10586 Materialising Ancestral Madang documents the emergence of pottery production processes and exchange networks along the northeast coast of New Guinea during the last millennium before the present. This dynamic period in the Pacific’s human past involved important fluctuations to people’s mobility, social interaction, and technological organisation. It therefore remains crucial to understanding and historicising the expansive maritime subsistence trading networks that famously characterised the coast in the late 19th and early 20th centuries. This book investigates these transformations by exploring the archaeology of Madang District; the heart of the Madang exchange network that revolved around the production and distribution of distinctive red-slipped pots. Potsherds of this style have been previously found spanning a 200 km radius, reaching Karkar Island, the Bismarck Archipelago, and even the New Guinea Highlands. By combining archaeological survey, excavation, craft ethnography, and archaeometric analyses, the volume systematically delineates the production groups that were working within this broader community of practice. The study shows that pre-colonial potters made use of a range of local raw materials and were free to improvise with their forming and decorating techniques but learned and reproduced similar technological sequences over the past 500–600 years. It is likely that social restrictions permitted only potters from a small number of clans to produce ceramics and that the finished vessels were then distributed both informally within the local area and strategically during extensive trade voyages along the northeast coast of New Guinea. These results therefore cast light on an important but previously obscured aspect of Pacific culture history and provide a model for how craft production and exchange processes have manifested and co-modified across the generations

MATERIALISING ANCESTRAL MADANG: Pottery Production and Subsistence Trading on the Northeast Coast of New Guinea Dylan Gaffney 2020 ISSN 0110-3709 University of Otago Studies in Archaeology · No. 29 Series Editors: Glenn R. Summerhayes Richard Walter Les O’Neill Archaeology Programme University of Otago Dunedin New Zealand © 2020 ISSN 0110-3709 University of Otago Studies in Archaeology is a continuation of the monograph series previously known as ‘University of Otago Studies in Prehistoric Anthropology.’ Contents List of Figures List of Tables Acknowledgements CHAPTER 1. INTRODUCTION Production and exchange on the northeast coast of New Guinea Ethnographic configurations Archaeological models Bel culture history Language Oral tradition Geology Ceramic technology Theoretical foundations Aspects of production Aspects of exchange Objectives and approach Research objectives vii xiii xvii 1 3 3 3 4 4 5 7 7 9 9 10 11 11 CHAPTER 2. THE ARCHIPELAGO OF CONTENTED PEOPLE Physical geography of Madang Local setting Raw material zones Land systems Geology Geophysical processes Tectonic activity Volcanic activity Summary 13 13 13 13 17 17 17 19 19 CHAPTER 3. BEL PRODUCTION AND EXCHANGE People of the coast Settlement Aspects of Bel production Food Pottery Canoes Magic Aspects of Bel exchange Trade systems Modes of exchange Trade friends Sailing Magic Food Material culture Negotiating colonial disruptions Summary 20 20 22 22 23 26 27 28 28 28 30 30 31 32 32 37 41 CHAPTER 4. MODERN POTTING COMMUNITIES Production groups Yabob pottery Bilbil pottery Defunct production centres Procurement Clay Slip Temper Production Emic typologies Clay preparation Roughing out and forming Decoration Slipping Firing Finishing techniques Distribution Consumption Reuse and discard Summary Vai: a Madang style chaîne opératoire 42 42 42 43 44 44 48 49 50 50 52 53 53 55 55 56 56 57 58 58 58 CHAPTER 5. TRACES OF THE PAST Archaeological anthropology around Madang Ancestral Madang exchange networks North South East Chronologies Summary 63 66 66 68 69 72 74 CHAPTER 6: ARCHAEOLOGICAL INVESTIGATIONS Madang area survey Kranket Island Siar Island Yabob Island Malmal village Bilbil Island Surface collections Excavation procedures Nunguri excavations Nunguri, Bilbil Island Excavation Stratigraphy Excavated material Tilu excavations Tilu, Malmal village Excavation Stratigraphy Chronology Excavated material Summary 75 75 77 78 79 80 81 81 82 82 82 83 85 90 90 90 91 92 93 95 iv CHAPTER 7. PRE-COLONIAL POTTING I: PRODUCTION Methodological issues Style and production groups Technology as methodology Approaches to New Guinea pottery classification Technological classification Method The Madang classification Technical attributes Assemblage composition Results: Nunguri Test Pit 1 Exotics Results: Tilu Unit 1 Exotics Shovel Pit 1 Results: Surface survey Summary 97 97 98 100 100 101 101 102 105 108 108 123 124 124 134 134 136 138 CHAPTER 8. PRE-COLONIAL POTTING II: PROCUREMENT AND DISTRIBUTION Methodological issues Foundations Geochemical analyses and New Guinea ceramics Technology as methodology Procurement Distribution Method The Madang classification Macroscopic fabric analysis Sampling procedures for geochemical analysis Geochemical characterisation Statistical treatment of the clay chemical data Results: ethnographic samples Inclusions and tempers Clay composition Results: Nunguri Techno-fabric groups Techno-mineralogical groups Techno-compositional groups Results: Tilu Techno-fabric groups Techno-mineralogical groups Techno-compositional groups Results: surface survey Techno-fabric groups Techno-mineralogical groups Techno-compositional groups Results: exotics Mineralogy Clay composition Summary v 139 139 140 140 141 141 142 142 143 144 145 146 147 147 148 151 151 152 153 163 163 163 163 175 175 175 175 179 179 181 181 CHAPTER 9. PRE-COLONIAL POTTING III: DECORATING Methodological issues Decoration and New Guinea ceramics Technology as methodology Method The Madang Classification Technical attributes Results: Nunguri Results: Tilu Results: surface survey Results: exotics Summary 185 185 186 187 187 188 193 210 223 227 229 CHAPTER 10. MATERIALISING ANCESTRAL MADANG Vai: aspects of pre-colonial production Procurement Forming Decorating The magic of the chaîne opératoire Production groups and communities of practice Dadeng: aspects of pre-colonial exchange Local distribution Regional distribution Exchange, embodied knowledge, and communities of practice Summary 230 230 234 235 239 239 240 240 244 247 248 CHAPTER 11. BEL CULTURE HISTORY Debates in language, landscape, history, archaeology Austronesian-speaking migrants (east or west)? Yomba Island Adaptations to the coast Towards a Bel culture history The local sequence Conclusions Endnote 249 249 252 253 253 253 255 256 258 BIBLIOGRAPHY vi List of Figures Figure 1.1 Madang, the study area, in Near Oceania. 2 Figure 1.2 Model of production & exchange networks operating around northeast New Guinea 3 Allen’s wave-like model for the intensification and divergence of trade networks through time 4 Figure 1.4 Austronesian language clusters within the West Melanesian Oceanic subgroup 5 Figure 1.5 Major language stock boundaries in Madang Province including the Bel language family 6 Figure 1.6 Bel languages branching from North New Guinea Cluster 6 Figure 1.7 Oral testimonies of 23 Bel speakers regarding Yomba Island 7 Figure 1.8 Four timelines illustrating possible Bel movement from a Yomba homeland to Madang with their associated archaeological signatures 8 Two competing scenarios for the movement of potting communities and Austronesian languages onto the northeast coast 9 Figure 1.3 Figure 1.9 Figure 2.1 Madang on the northeast coast of New Guinea 14 Figure 2.2 Aerial view looking east over coastal plains to Yabob Island, Urembu, and Bilbil Island 14 Figure 2.3 Coastal Madang District, northeast New Guinea 15 Figure 2.4 Land systems of the Madang District 16 Figure 2.5 Geological zones around Madang District 18 Figure 3.1 Stilted houses on Bilbil Island, illustrated by Nikolai Miklouho-Maclay 1872 21 Figure 3.2 A stilted house used as a school on Kranket Island 21 Figure 3.3 A darem (spirit house) on Bilbil Island 22 Figure 3.4 Women at Bilbil village preparing newly harvested yams for storage 23 Figure 3.5 Fishing nets, called raj in Bilbil. Bongu village, Astrolabe Bay 24 Figure 3.7 Bilbil Island potters observed by Otto Finsch 25 Figure 3.8 A sample of Bel pottery 26 Figure 3.9 Bel canoes from the Madang coast 27 Figure 3.10 Coastal trade networks in northeast New Guinea 29 Figure 3.11 Wind chart showing major sailing winds used by Bel traders 31 Figure 3.12 A sample of inland Madang pottery 35 Figure 3.13 Stone axe from the Huon, collected by Richard Neuhauss 35 Figure 3.14 Jangar, an elder of Bilbil Island with shield from Karkar Island 36 Figure 3.15 A pestle and mortar for crushing sago, Karkar Island, German New Guinea 36 Figure 3.16 Two wooden bowls from Astrolabe Bay 37 Figure 3.17 Wooden ancestor figure, Bilbil Island, collected by Otto Finsch 38 Figure 3.18 Turtle shell armbands from Astrolabe Bay 38 Figure 3.19 Trade valuables from Arop/Long Island 39 Figure 3.20 Nikolai Miklouho-Maclay meets local people in Astrolabe Bay vii 39 Figure 3.21 The people of Riwo Island during German administration 40 Figure 4.1 Yabob-up-top potting village 43 Figure 4.2 Bilbil potting village. A view from offshore 43 Figure 4.3 Extracting the ‘Red’ clay near Yabob-up-top 44 Figure 4.4 Yeyeg selects appropriate clay from the ‘Red’ source near Yabob-up-top 45 Figure 4.5 Dorcas rolls the ‘Yellow’ clay into a portable ball, near Bilbil village 45 Figure 4.6 The ‘Black’ clay source near Bilbil village 46 Figure 4.7 Balls of clay stored after collection, Bilbil village 46 Figure 4.8 Canonical discriminant functions of clay samples from Bilbil and Yabob 48 Figure 4.9 Local ceramic resources used by modern Bilbil potters 49 Figure 4.10 Local ceramic resources used by modern Yabob potters 50 Figure 4.11 Range to ceramic resources used by modern Bel potters near Madang 50 Figure 4.12 Yeyeg mixes clay and temper at Yabob-up-top 52 Figure 4.13 Yeyeg moulds rim preform at Yabob-up-top 53 Figure 4.14 Yeyeg using a short, thick paddle to form shoulder carination, Yabob-up-top 54 Figure 4.15 Yeyeg demonstrates decorating the neck with impressions, Yabob-up-top 54 Figure 4.16 A common decoration applied by modern Bilbil and Yabob potters 55 Figure 4.17 Kasare Dadau holds a decorated magob pot awaiting slipping and firing 55 Figure 4.18 Dorkas Kana applying slip to unfired vessels 56 Figure 4.19 The sago paste ready to be applied to the unpolished vessel, Bilbil village 56 Figure 4.20 Sentie Noah polishes a fired bodi using sago paste 57 Figure 4.21 57 An assortment of standard bodi and tangeng, alongside tourist pots Figure 4.22 Dorkas Kana sells her pots in Port Moresby 58 Figure 4.23 58 Tourist pots for sale online Figure 4.24 Three broken pots are arranged to support an in use tangeng cooking vessel 59 Figure 4.25 59 Several bodi being reused as flower pots, Bilbil village Figure 4.26 A display of the vessel forming sequence, Yabob-up-top 60 Figure 4.27 A chaîne opératoire of a modern Madang style bodi pot 61, 62 Figure 5.1 Archaeological sites around coastal Madang, examined by Egloff 66 Figure 5.2 Previous excavations at Tilu (JCA) site 66 Figure 5.3 Loc. map showing sites where Ancestral Madang sherds have been reported 68 Figure 5.4 Madang pot traded into Megabo, Eastern Highlands 69 Figure 5.5 Woman and children on Arop/Long Island with Madang trade pot 70 Figure 5.6 Radiocarbon dist. with 2σ confidence, in association with Madang sherds 73 Figure 6.1 Archaeological places investigated during the 2014 field season 76 Figure 6.2 Aerial photograph of Kranket Island 77 Figure 6.3 Aerial photograph of Siar Island 77 Figure 6.4 Survey of the Madang District, 2014 78 Figure 6.5 Aerial photograph of Yabob Island 79 Figure 6.6 Aerial photograph of Malmal area showing Tilu site 79 viii Figure 6.7 Aerial photograph of Bilbil Island 80 Figure 6.8 Inscribed German brick, highest point of Bilbil Island 81 Figure 6.9 Looking east at Bilbil with beach at south and Nunguri site in the centre 82 Figure 6.10 Plan of excavations at Nunguri clan area, Bilbil Island 83 Figure 6.11 Excavations at Test Pit 1, Nunguri site, Bilbil Island 83 Figure 6.12 Stratigraphy of Test Pit 1, Nunguri site, Bilbil Island 84 Figure 6.13 Charcoal samples submitted to AINSE for dating, Nunguri, Bilbil Island 85 Figure 6.14 Plot of calibrated date distributions from Test Pit 1, Nunguri site, Bilbil Island 85 Figure 6.15 Stone artefacts from Test Pit 1, Nunguri site, Bilbil Island 87 Figure 6.16 Shell artefacts from Test Pit 1, Nunguri site, Bilbil Island 88 Figure 6.17 Density of artefact types to sieved deposit, Test Pit 1, Nunguri site, Bilbil Island 89 Figure 6.18 Density of obsidian and non-obsidian lithics to sieved deposit, Test Pit 1, Nunguri site, Bilbil Island 89 Water-rolled pottery sherds, Test Pit 1, Nunguri site, Bilbil Island 89 Figure 6.19 Figure 6.20 Excavations at Tilu site, 2014 90 Figure 6.21 91 Plan of excavations at Tilu clan area, Malmal village Figure 6.22 Stratigraphy of Unit 1, Tilu site, Malmal village 91 Figure 6.23 North wall of completed Unit 1, Tilu site, Malmal village 92 Figure 6.24 Charcoal samples submitted to AINSE for dating, Tilu, Malmal village 92 Figure 6.25 93 Plot of calibrated date distributions from Unit 1, Tilu, Malmal village Figure 6.26 Density of artefact types to sieved deposit, Unit 1, Tilu site, Malmal village 94 Figure 6.27 Shell artefacts from Unit 1, Tilu site, Malmal village 95 Figure 7.1 Analytical terminology of a ceramic chaîne opératoire 99 Figure 7.2 Procedures of a technological classification 101 Figure 7.3 Anatomical landmarks of a Madang style pot 102 Figure 7.4 Classificatory procedures to distinguish Madang style technical classes 103 Figure 7.5 Discrete technical attributes recorded in the ceramic database 104 Figure 7.6 Metric technological attributes recorded in the ceramic database 106 Figure 7.7 Portions of ceramic vessels represented at Nunguri, Test Pit 1, by number 109 Figure 7.8 Portions of ceramic vessels represented at Tilu, Unit 1, by number 109 Figure 7.9 Percentage of Madang style rim classes at Nunguri, Test Pit 1 by exc. spit 110 Figure 7.10 Class 1 Madang style rims at Nunguri, Test Pit 1 111 Figure 7.11 Class 2 Madang style rims at Nunguri, Test Pit 1 112 Figure 7.12 Class 3 Madang style rims at Nunguri, Test Pit 1 113 Figure 7.13 Class 4 Madang style rims at Nunguri, Test Pit 1 114 Figure 7.14 Class 5 Madang style rims at Nunguri, Test Pit 1 115 Figure 7.15 Rim course and profile by technical class at Nunguri, Test Pit 1 117 Figure 7.16 Lip profile and extra lip features by technical class at Nunguri, Test Pit 1 118 Figure 7.17 MCA of discrete formal attributes for Madang style rims, Nunguri, Test Pit 1 119 Figure 7.18 Rim thickness by technical class at Nunguri, Test Pit 1 119 ix Figure 7.19 Relationship between rim and neck thickness by technical class at Nunguri 120 Figure 7.20 Relationship between rim length and thickness by technical class at Nunguri 121 Figure 7.21 Minimum and maximum rim orientation by technical class at Nunguri 122 Figure 7.22 Relationship between rim orientation & inclination by class, Nunguri 122 Figure 7.23 Vessel orifice diameter by technical class, Nunguri, Test Pit 1 123 Figure 7.24 Exotic rim from Nunguri, Test Pit 1 124 Figure 7.25 Percentage of Madang style technical classes at Tilu, Unit 1 by excavation spit 125 Figure 7.26 Class 1 Madang style rims at Tilu, Unit 1 126 Figure 7.27 Class 2 Madang style rims at Tilu, Unit 1 127 Figure 7.28 Class 3–5 Madang style rims at Tilu, Unit 1 128 Figure 7.29 Rim course and profile by technical class at Tilu, Unit 1 129 Figure 7.30 Lip profile and extra lip features by technical class at Tilu, Unit 1 130 Figure 7.31 MCA of discrete formal attributes for Madang style rims, Tilu, Unit 1 132 Figure 7.32 Rim thickness by technical class at Tilu, Unit 1 132 Figure 7.33 Relationship between rim and neck thickness by technical class at Tilu 133 Figure 7.34 Relationship between rim length and thickness by technical class at Tilu 133 Figure 7.35 Relationship between rim orientation and inclination by technical class, Tilu 134 Figure 7.36 Vessel orifice diameter by technical class, Tilu, Unit 1 135 Figure 7.37 Exotic rims from Tilu, Unit 1. 135 Figure 7.38 Ring base from 60 cm below surface, Tilu, Shovel Pit 1 136 Figure 7.39 Rims collected from surface survey, Madang 2014 137 Figure 8.1 Several scenarios in which local exchange is difficult to chemically distinguish 142 Figure 8.2 Procedures of a technological classification 143 Figure 8.3 Ward’s HCA of eight ethnographic clay sources. 150 Figure 8.4 Group average HCA of eight ethnographic clay sources 150 Figure 8.5 PCA showing clay compositional data from eight ethnographic clay sources 151 Figure 8.6 Nunguri rim sherd techno-fabric groupings by excavation spit 152 Figure 8.7 Nunguri decorated body sherd techno-fabric groupings by excavation spit 152 Figure 8.8 Nunguri rim sherd techno-fabric groupings by technical class 153 Figure 8.9 Ward’s HCA of Nunguri sherds 159 Figure 8.10 Group average HCA of Nunguri sherds 160 Figure 8.11 PCA showing clay compositional data from Nunguri sherds by excavation spit 161 Figure 8.12 PCA showing clay compositional data from Nunguri sherds by technical class 161 Figure 8.13 PCA showing clay compositional data from Nunguri sherds by fabrics 162 Figure 8.14 PCA showing clay compositional data of Nunguri sherds and ethnographic clays 162 Figure 8.15 Tilu rim sherd techno-fabric groupings by excavation spit 163 Figure 8.16 Tilu decorated body sherd techno-fabric groupings by excavation spit 164 Figure 8.17 Tilu rim sherd techno-fabric groupings by technical class 164 Figure 8.18 Ward’s HCA of Tilu sherds 171 x Figure 8.19 Group average HCA of Tilu sherds 172 Figure 8.20 PCA showing clay compositional data from Tilu sherds by excavation spit 173 Figure 8.21 173 PCA showing clay compositional data from Tilu sherds by technical class Figure 8.22 PCA showing clay compositional data from Tilu sherds by fabrics 174 Figure 8.23 174 PCA showing clay compositional data of Tilu sherds and ethnographic clays Figure 8.24 Ward’s HCA of surface survey sherds 175 Figure 8.25 177 Group average HCA of surface survey sherds Figure 8.26 PCA showing clay compositional data from surface survey sherds by location Figure 8.27 177 PCA showing clay compositional data from surface survey sherds by tech. class 178 Figure 8.28 PCA showing clay compositional data from surface survey sherds by fabric 178 Figure 8.29 PCA showing clay compositional data of survey sherds and ethnographic clays 179 Figure 8.30 Ward’s HCA of exotic sherds 181 Figure 8.31 Group average HCA of exotic sherds 182 Figure 8.32 PCA showing clay compositional data from exotic sherds by location 182 Figure 8.33 PCA showing clay compositional data from exotic sherds & ethnographic clays 183 Figure 8.34 PCA showing clay compositional data from all sherds by location 183 Figure 8.35 184 PCA showing clay compositional data from all sherds by class Figure 8.36 PCA showing clay compositional data from all sherds by fabric 184 Figure 9.1 Procedures of a technological classification 187 Figure 9.2 Design configurations represented in the Madang style 189 Figure 9.3 Design configurations represented in the Madang style 190 Figure 9.4 Appliqué design motifs represented in the Madang style 191 Figure 9.5 Linear-gash incised design motifs represented in the Madang style 192 Figure 9.6 Groove incised, impressed and paddle impressed design motifs in the Madang style. 193 Figure 9.7 Locations of decoration on rim sherds 193 Figure 9.8 Examples of appliqué decoration on Madang style body sherds at Nunguri 195 Figure 9.9 Examples of appliqué decoration on Madang style body sherds at Nunguri 196 Figure 9.10 Examples of appliqué decoration on Madang style body sherds at Nunguri 197 Figure 9.11 Examples of incised decoration on Madang style body sherds at Nunguri 198 Figure 9.12 Examples of incised decoration on Madang style body sherds at Nunguri 199 Figure 9.13 Examples of incised decoration on Madang style body sherds at Nunguri 200 Figure 9.14 Examples of incised decoration on Madang style body sherds at Nunguri 201 Figure 9.15 Examples of paddle impression on Madang style body sherds at Nunguri 202 Figure 9.16 Percentage of application methods represented on the external body/ shoulder of sherds at Nunguri, Test Pit 1 by excavation spit 203 Figure 9.17 Percentage of Nunguri rims with decorated lips by excavation spit 206 Figure 9.18 Percentage of Nunguri rims with inner rim notching by excavation spit 207 Figure 9.19 Cases of application method on body/shoulder by technical class at Nunguri 209 Figure 9.20 Cases of application method on body/shoulder by techno-fabric at Nunguri xi 210 Figure 9.21 PCA showing clay chemical data of Nunguri sherds coded by body/ shoulder application method 211 Figure 9.22 PCA showing clay chemical data of Nunguri sherds coded by lip decoration Figure 9.23 PCA showing clay chemical data of Nunguri sherds coded by inner rim notching 211 212 Figure 9.24 Examples of appliqué decoration on Madang style body sherds at Tilu 214 Figure 9.25 215 Examples of appliqué decoration on Madang style body sherds at Tilu Figure 9.26 Examples of incised decoration on Madang style body sherds at Tilu 216 Figure 9.27 Examples of paddle impressed decoration on Madang style body sherds at Tilu 217 Figure 9.28 Percentage of application methods represented on the external body/ shoulder of sherds at Tilu, Unit 1 by excavation spit 218 Figure 9.29 Percentage of Tilu rims with decorated lips by excavation spit 221 Figure 9.30 Cases of application method on the body/shoulder by technical class at Tilu 223 Figure 9.31 Figure 9.32 Figure 9.33 Percentage cases of application method on the body/shoulder used by techno-fabric at Tilu, Unit 1 223 PCA showing clay chemical data of Tilu sherds coded by body/ shoulder application method 224 PCA showing clay chemical data of Tilu sherds coded by lip decoration 224 Figure 9.34 PCA showing clay chemical data of Tilu sherds coded by inner rim notching Figure 9.35 PCA showing clay chemical data of survey sherds coded by body/ shoulder application method. 225 227 Figure 9.36 PCA showing clay chemical data of survey sherds coded by inner rim notching 228 Figure 9.37 Exotic decorated body sherds from Tilu, Unit 1 228 Figure 10.1 Clay procurement zones around Madang including the major procurement hub around Bilbil and possible minor sources at Yabob and Mindiri 231 Figure 10.2 Different procurement choices represented by ‘exotic’ sherd tempers 233 Figure 10.3 Seriation of Madang style technical classes at Nunguri, Test Pit 1 235 Figure 10.4 Seriation of Madang style technical classes at Tilu, Unit 1 235 Figure 10.5 236 Comparison of Nunguri and Tilu seriations with associated absolute dates Figure 10.6 Chaînes opératoires of pre-colonial Madang style potteries Figure 10.7 Figure 11.1 Figure 11.2 Figure 11.3 241, 242 Diagrammatic model showing pre-colonial exchange of Madang style pots around the northeast coast 247 Duration of major ceramic traditions around the northeast coast of New Guimea 251 Diagrammatic representation of proposed Bel culture history sequence around Madang 254 Modern Bilbil pot designed for tourists 257 xii List of Tables Table 3.1 Important subsistence crops on the northeast coast 22 Table 3.2 Important subsistence animal species on the northeast coast 24 Table 3.3 Annual planting and trading cycle 32 Table 3.4 Trade items acquired by Bel according to oral testimonies and word lists Table 4.1 Clay sources in use by Bilbil and Yabob potters, Madang 47 Table 4.2 Colour of Bilbil and Yabob clays: unfired, 700°C and 1000°C. 47 Table 4.3 Ethnographic sand tempers collected from Bilbil and Yabob areas, Madang 49 Table 4.4 Emic pottery classification at Bilbil and Yabob 51 Table 5.1 Allen’s (1971) ceramic types based on a surface collection near Bilbil village 64 Table 5.2 Summary of sites recorded near Madang by Egloff 65 Table 5.3 Radiocarbon dates for Tilu Mound A and Mound B 66 Table 5.4 Variation in Madang rim forms identified by Egloff 67 Table 5.5 Madang ceramic decorative techniques, identified by Egloff 67 Table 5.6 Arop/Long Island style groups I & IV rim forms 69 Table 5.7 Lilley’s Madang style rim classes based on sherds exc. from the Vitiaz Strait 71 Table 5.8 Published radiocarbon dates in association with excavated Madang sherds 72 Table. 6.1 Summary of surface collections made during 2014 survey 81 Table 6.2 Radiocarbon determinations from Test Pit 1, Nunguri site, Bilbil Island 84 Table 6.3 Material by weight (g) recovered from Test Pit 1, Nunguri site, Bilbil Island 86 Table 6.4 Material by number (n) recovered from Test Pit 1, Nunguri site, Bilbil Island 86 Table. 6.5 Water-rolled pottery sherds, Test Pit 1, Nunguri site, Bilbil Island 90 Table 6.6 Radiocarbon determinations from Unit 1, Tilu site, Malmal village 93 Table 6.7 Material by weight (g) recovered from Unit 1, Tilu site, Malmal village 94 Table 6.8 Material by number (n) recovered from Unit 1, Tilu site, Malmal village 94 Table 6.9 Material by number (n) and weight (g) recovered from Shovel Pit 1, Tilu site, Malmal village. 95 33, 34 Table 7.1 Madang ceramic assemblages by number (#) and weight (g) 107 Table 7.2 Rims by number of sherds and minimum number of vessels 108 Table. 7.3 Tech. classes represented at Nunguri, Test Pit 1 by MNV and excavation spit 108 Table. 7.4 Rim and lip forms by technical class at Nunguri, Test Pit 1 116 Table. 7.5 Average rim and neck thickness (mm) by technical class at Nunguri, Test Pit 1 116 Table. 7.6 Kruskal-Wallis test of rim thickness (mm) by tech. class at Nunguri, Test Pit 1. 116 Table. 7.7 Post hoc test of rim thickness (mm) comparing individual technical classes at Nunguri. 117 Table. 7.8 Rim direction (MNV) by tech. class and excavation spit at Nunguri, Test Pit 1 121 Table 7.9 Average rim orientation and inclination (°) by tech. class, Nunguri, Test Pit 1 122 Table 7.10 Average orifice diameter (cm) by technical class, Nunguri, Test Pit 1 122 Table 7.11 Kruskal-Wallis test of orifice diameter by technical class at Nunguri, Test Pit 1 123 xiii Table 7.12 Mood’s test of orifice diameter by technical class at Nunguri, Test Pit 1 Table 7.13 Post hoc test of orifice diameter comparing individual tech. classes at Nunguri 124 Table 7.14 Technical classes represented at Tilu, Unit 1 by MNV and excavation spit 124 Table 7.15 Rim and lip forms by technical class at Tilu, Unit 1 131 Table 7.16 Average rim and neck thickness (mm) by technical class at Tilu, Unit 1 131 Table 7.17 Average rim orientation and inclination (°) by technical class, Tilu, Unit 1 131 Table 7.18 Average orifice diameter (cm) by technical class, Tilu, Unit 1 134 Table 7.19 Technical classes represented in Tilu, Shovel Pit 1 by MNV and depth 135 Table 7.20 Technical classes represented in surface collections by MNV and location 136 Table 8.1 Macro-tempers of Madang style ceramics 144 Table 8.2 Madang style sherds selected from Nunguri and Tilu for geochemical analysis 144 Table 8.3 Madang style rim sherds from surface collections for geochemical analysis Table 8.4 Exotic sherds from excavation and surface collection for geochemical analysis 145 Table 8.5 SEM results of mineral inclusions naturally present in Bilbil clay sources 148 Table 8.6 SEM results of mineral inclusions naturally present in Yabob clay sources 149 Table 8.7 SEM results of mineral inclusions in Nunguri, Spit 2 samples 154 Table 8.8 SEM results of mineral inclusions in Nunguri, Spit 5 samples 155 Table 8.9 SEM results of mineral inclusions in Nunguri, Spit 11 samples 156 Table 8.10 SEM results of mineral inclusions in Nunguri, Spit 13 samples 157 Table 8.11 SEM results of mineral inclusions in Nunguri, Spit 15 samples 158 Table 8.12 SEM results of mineral inclusions in Tilu, Spit 2 samples 165 Table 8.13 SEM results of mineral inclusions in Tilu, Spit 4 samples 166 Table 8.14 SEM results of mineral inclusions in Tilu, Spit 6 samples 167 Table 8.15 SEM results of mineral inclusions in Tilu, Spit 7 samples 168 Table 8.16 SEM results of mineral inclusions in Tilu, Spit 9 samples 169 Table 8.17 SEM results of mineral inclusions in Tilu, Spit 10 samples 170 Table 8.18 Techno-fabric groups of survey collected sherds 175 Table 8.19 SEM results of mineral inclusions in surface survey samples 176 Table 8.20 SEM results of mineral inclusions in exotic samples from Tilu, Nunguri, and Yabob. 180 Table 9.1 Application methods represented at Nunguri, Test Pit 1 by excavation spit 194 Table 9.2 Average body thickness (mm) by application method at Nunguri. 202 Table 9.3 Kruskal-Wallis test of body thickness (mm) by application method at Nunguri 202 Table 9.4 Mood’s test of body thickness (mm) by application method at Nunguri 204 Table 9.5 Appliqué configurations represented at Nunguri, Test Pit 1 by location 204 Table 9.6 Incised configurations represented at Nunguri, Test Pit 1 by location 205 Table 9.7 Impressed and paddle impressed configurations represented at Nunguri, Test Pit 1 by location 206 Table 9.8 Motifs represented at Nunguri, Test Pit 1 by excavation spit Table 9.9 Techno-style groups within the Madang style at Nunguri xiv 123 145 207, 208, 209 212 Table 9.10 Decorative application methods represented at Tilu, Unit 1 by excavation spit 213 Table 9.11 Average body thickness (mm) by application method at Tilu, Unit 1 217 Table 9.12 Kruskal-Wallis test of body thickness (mm) by application method at Tilu 217 Table 9.13 Mood’s test of body thickness (mm) by application method at Tilu, Unit 1 217 Table 9.14 Appliqué configurations represented at Tilu, Unit 1 by location 219 Table 9.15 Incised configurations represented at Tilu, Unit 1 by location. 220 Table 9.16 Impressed and paddle impressed configurations represented at Tilu, Unit 1 by location 221 Table 9.17 Motifs represented at Tilu, Unit 1 by excavation spit 222 Table 9.18 Techno-style groups within the Madang style at Tilu 225 Table 9.19 Decorative application methods represented from surface collections 226 Table 9.20 Decorative configurations represented in surface collections by location 226 Table 10.1 Cross-tabulation of chi-squared results of decorative method to Egloff ’s rim type 237 Cross-tabulation of chi-squared results of decorative method to Class 1–5 rims 238 Cross-tabulation of chi-squared results showing decorative method to rim class 245 Comparison of Gaffney’s fabric and Lilley’s paste group designation 245 Table 10.2 Table 10.3 Table 10.4 xv Acknowledgements I was first invited to Madang in June 2014 to take part in a research project led by Glenn Summerhayes. This project would form the basis of my University of Otago Master of Arts thesis (Gaffney 2016), upon which this volume is based. I am indebted to Glenn for his guidance, trust, and encouragement in the field and during subsequent writing periods at Otago and Cambridge. In Dunedin, my second supervisor, Anne Ford, provided much needed feedback on the analyses and my written work, continuing a trend of unbeatable teaching. The thesis owes a great deal to both of their dedication and stewardship. The Madang fieldwork was completed by Glenn and myself, along with Mary Mennis, Affrica Cook, Teppsy Benny, Judith Field, and the late Herman Mandui. Mary Mennis planted the seeds for the work, describing eroding midden from Bilbil Island, and throughout the project her stimulating comments on Madang culture and history have been invaluable. Herman was an inspiring mentor in the field and provided museum consent for the project. The original MA thesis was dedicated to Herman who sadly passed away a few months after the fieldwork. Affrica and Teppsy provided great friendship in the field. Judith kindly assisted with excavations on Bilbil and, along with Sindy Luu and Adelle Coster, undertook crucial residue analysis on the pottery remains. The research was supported by the National Research Institute (NRI), Port Moresby, and the Madang Provincial Government and was affiliated with the National Museum and Art Gallery of Papua New Guinea (NMAG-PNG), Port Moresby. Sir Peter Barter of the Madang Resort and son Andrew Barter were incredibly generous in providing accommodation during fieldwork. Over the past six years Sir Peter has been an outstanding support for archaeology in Madang Province. Busybee, Casper, Melissa, and Sebona from the Resort also accommodated us and provided transport during our fieldwork. Father Jan Czubra at Divine Word University and Jane Naso formerly at the Madang Museum were also extremely helpful and welcoming. Jane accompanied me during ethnographic work and interviews. his family were also incredibly welcoming and insightful about local history. On Yabob Island, Gabriel and Paul were welcoming guides. At Yabob-up-top, Yeyeg and her sisters along with Margaret were invaluable to describing their potting process. My sincere thanks go to Yeyeg for her time as her knowledge and skill in pottery making contributed substantially to the direction of the present volume. Various funding bodies supported the research. An Otago Research Grant to Glenn Summerhayes funded the fieldwork. The Australian Archaeological Association (AAA) Student Research Grant along with a University of Otago Master of Arts Grant funded my own fieldwork travel to Madang. The Skinner Fund, Royal Society of New Zealand, supported travel to Auckland and Sydney to visit various museum collections. An Australian Institute of Nuclear Science and Engineering (AINSE) Research Grant (ALNGRA15007) supported radiocarbon dating. Geraldine Jacobsen conducted the C14 dating and Sally Brandt assisted with correspondence on the dates. The Asian Migrations Research Theme Postgraduate Fund and AAA Student Scheme supported presenting preliminary results at AAA Conference 2014. A University of Otago Postgraduate Fund supported travel to WAC-8 in Kyoto to present the results of the analyses. The R and E Seelye Trust Masters Scholarship supported my study in Dunedin. The Kakano Fund of the Association of Social Anthropologists of Aotearoa/New Zealand funded the return of community reports to Bilbil and Malmal in 2016 (Gaffney and Summerhayes 2017). The Magdalene College Archaeology and Anthropology Fund has made the production of the physical monograph possible and I wholeheartedly thank Simon Stoddart for his support in this matter. Labwork was greatly expedited by the assistance of numerous student volunteers from 2014–2015. Greg Hil, Eleanor Moore, and Charles Radclyffe helped me with the cleaning of ceramics. Debbie Stoddart, Greg Hil, Evan Morcom, Georgia Kirby, Rebecca Adam, Jamie Hearfield, Merryn Chynoweth, Lucy Northwood, Laura Lawson, Teina Tutaki, Rhian Gaffney, and Emma Morris assisted me in cleaning At Malmal, Joeseph Barem gave permission for work on and provisionally sorting shell midden. Debbie Stoddart Tilu and helped to excavate, along with Elias and Fili. invested substantial time in identifying the Tilu shell, for Erasmus Bitolai, Mari Bow, Danu Kepi, and Maksil Mot which she must be especially thanked. Alana Kelly, Alix helped to relocate the Tilu site. Many thanks also to the Muir, Robert Henderson, Tessa Whitehead, and Nikole numerous others at Malmal who helped on the sieves and Wills helped to sort non-ceramic artefacts. Alana, Alix, bagging. At Bilbil, Kebei Balifun, Sungai Damun, Daguan and Rob were particularly charitable in their time photoPasagai, Sungasong Yalom, Garang Kisom, Napen Tegil, graphing decorated pottery sherds. Brian Kubei, Kasare Dadau and many others were essential to the excavations and provided consent for the work. In Geology, my advisor James Scott provided valuable suThanks also to Trevor Nomu for rowing all the way back pervision in scanning electron microscopy (SEM) and unto Bilbil just to collect a small bag of sand for geochemi- dertook preliminary petrographic analyses. Brent Pooley cal analysis. Interviews and demonstrations from Sentie assisted making SEM plugs and thin sections, while DaNoah and Dorcas Kana from Murpat clan were essential mian Walls assisted firing ethnographic clay samples. Liz for recording the potting process. At Siar Island, Peter was Girvan provided technical support in using the SEM. In our very helpful guide. On Kranket Island, Desmond and the Department of Statistics, Tim Jowlett consulted with xvii me on statistical analyses and R, and was incredibly sup- Walter and I particularly thank the anonymous reviewers portive and patient in approaching multivariate statistics. for their incisive comments on the draft manuscript. Les O’Neill, Glenn Summerhayes, and Richard Walter must In Biological Anthropology, Anna Gosling provided help- be thanked for their ongoing support of the Otago monoful advice on human remains and Hallie Buckley con- graph series, and their encouragement in seeing this volducted an examination of those remains. Monica Tromp ume to completion. Les O’Neill must be especially thanked conducted an analysis of human dental calculus, while for his expert editorial and design work that has brought Kathrin Naegele and Monica Tromp undertook screen- the volume into book form. ing for ancient pathogens and conducted ancient DNA sequencing at the Max Planck Institute in Jena. In the former Department of Anthropology and Archaeology at Otago, Catherine Waite, Marj Blair, and Heather Various museum staff facilitated accessing historical and Sadler provided invaluable administrative and technical ethnographic Madang collections: Moira White and Scott support. As always, Phil Latham was essential to organisReeves at the Otago Museum; Rebecca Conway at the ing laboratory facilities. I am grateful to my employers and Macleay Museum, University of Sydney; Grace Hutton at friends at Southern Pacific Archaeological Research, parTe Papa; Yvonne Carrillo at the Australian Museum; Fuli ticularly Karen Greig and Richard Walter, who kindly proPereira and Kolokesa Mahina-Tuai at the Auckland Mu- vided me the support to complete my original MA thesis seum. I am thankful for their assistance. while working full-time on New Zealand archaeological projects. I am also indebted to former graduate students Ben Shaw and Simon Bickler both gave valuable advice at Otago for years of friendship: Jessie Hurford, Tristan on the project and feedback on writing, for which I am Russell, Nick Sutton, Helen Heath, Evan Morcom, Naomi very grateful. Brian Egloff and Jim Specht both supplied Woods, Julia Lewis, Fran Allen, Alex Scahill, Megan Lawuseful unpublished reports on the region. Bruce Numode rence, Jeremy Moyle, Luke Tremlett, Peter Petchey, Gabe photocopied essential site reports from the NMAG-PNG. Vilgalys, Nick Hogg, and Sam Kurmann. Lastly, I sincerely Malcolm Ross and Simon Greenhill provided advice on thank old friends in Dunedin, new friends in Europe, and the linguistics. Thank you also to Simon Day for insight- my family for years of support, learning, and encourageful and important discussions on Yomba and the geology ment, and for reminding me that people not pots – if I am of the area. The original thesis was revised following in- permitted to borrow the archaeological adage – is what’s sightful examiner comments from Ian Lilley and Richard important. Dylan Gaffney Magdalene College Cambridge October 2020 xviii Chapter 1. Introduction 2007; Widgren and Håkansson 2014). Equally importantly, these groups also lacked much of the arable land from which social relations could be generated. For instance, along the Rai Coast land stores the creative potential for It is within production and exchange networks that ob- generating all kinds of interpersonal and political alliances, jects are made, products are moved, and communities with claims to the control of, or centrality to, these sets of are bound. By unravelling something about how materi- relations being paramount (Leach 2003: 216–217, 2012). Not als are transformed within these networks, and how they so for those specialist producers and middlemen groups vitalise and shed meaning along the way, we can begin to who often had limited cultivatable land at their disposal, understand broader social, technological, and economic and for whom networks of production and exchange berelationships within the landscape. Perhaps more than came key to their access to utilitarian goods, raw materianywhere else in the world, ethnographic examples from als, and food. For these ‘subsistence traders’ (Allen 1985), Melanesia have cast light on diverse modes of material attending to exchange relationships was not only socially culture manufacture and distribution, which have been motivated, but importantly required for subsistence maincrucial to theoretical and methodological developments in tenance, ensuring the population’s ongoing survival. the wider discipline (e.g. Gell 1992a; Gregory 1982; Mauss 1925; Sahlins 1972; Thomas 1991). The coastal fringe and In some places, subsistence traders came to thrive ecooffshore islands, in particular, played host to unique ar- nomically by holding sway over the movement of materirangements of these networks in the recent past, involving als, despite occupying ecologically marginal zones. Conannual cycles of cropping, specialist craft production, and trolling these networks allowed some subsistence trading long-distance trade voyaging, which redistributed objects groups to not only acquire surplus foodstuffs, but also hundreds of kilometres from their point of manufacture valuable and exotic items necessary for socio-political mato geographically disparate consumers. This is particularly noeuvring and the renewal of important life-cycle stages well known from Malinowski’s (1922) foundational de- (Martin 2019). Sahlins (1972: 282) notes that although scription of the Kula exchange network, which still oper- many of these specialist trading groups were marginally ates today, and sees named shell valuables moving in vari- situated with regard to the larger island of New Guinea, ofous directions around the Massim, off the southeastern tail ten possessing scarce resources for crop production, they of New Guinea. Equally important examples describe shell were usually the most materially affluent in the area, as exchange in the Bismarck Archipelago (Counts 1979; Par- a result of their extensive social networks and ability to kinson 1887; Salisbury 1966) and Solomon Islands (Cooper control the flow of goods and exchange rates, and so oc1971; Woodford 1908), Hiri pottery trade on the south coast cupied a precarious but privileged position off the coast. of Papua (Chalmers 1887; Dutton 1982; Groves 1960; Selig- Coastal exchange networks then formed crucial conduits mann 1910), Siassi middlemen in the Vitiaz Strait (Harding for people’s social connectivity and subsistence, and, at 1967), and the red-feather money of Santa Cruz (Beasley the regional level, allowed for the ratcheting of produc1936; Pycroft 1935). tion and consumption, while reducing pressure on locally scarce resources. Many of these groups occupied ecologically marginal zones and lacked ‘landesque capital’ – the potential to Despite these important contributions from Melanesian modify and invest labour in gardens with the intention ethnography, the long-term evolution of these networks that the land itself will then provide increased yields (see prior to the colonial period is patchily understood, and Bayliss-Smith and Hviding 2015; Clark and Tsai 2013; Kirch many of our interpretations remain ahistorical (Kirch 1991). By examining the temporal dynamics of how craft 1 Tim Ingold (2011: 1–20) in Redrawing Anthropology: Materials, production and subsistence trading transformed over the centuries it should be possible to model the interrelationMovements, Lines Follow the materials; learn the movements; draw the lines — Tim Ingold (2011)1 1 Chapter . Introduction ship between social, economic, and technological change – and how this is manifested archaeologically – amongst small-scale, lowly-stratified groups in the area. Moreover, to better appreciate the vibrant communities and material cultures that characterise the region today, we must explore the pre-colonial past, and the distinct, historically contingent processes that brought these interaction networks to life. As such, the present volume traces production and exchange back through time on New Guinea’s northeast coast. Specifically, it is an archaeological study that examines the emergence of the Madang exchange network over the last half-millennium before present, which was one of the most extensive, but perhaps least well-known, interaction systems around coastal Melanesia during the recent past. It is within this crucial period that many of the exchange networks observed ethnographically around New Guinea began to take form, and when specialised pottery making shifted increasingly towards mass production to fuel these networks (Skelly and David 2017: 522). ed material culture around the coast, between islands, and into the interior (Harding 1967). Potsherds ancestral to these red-slipped vessels have been found scattered across hundreds of kilometres, from Karkar Island in the north (Egloff 1975), along the northeast New Guinea coast (Lilley 1986), into the Bismarck Archipelago (Summerhayes 2001a), and even to the Central New Guinea Highlands (White 1972), suggesting that this network was in operation, and flourishing, prior to Europeans. However, very little work has been carried out around Madang itself, the nucleus of this network. For this reason, the deeper history of the Bel groups and their pottery-making, along with how their exchange network emerged, remain crucial but unanswered questions in Melanesian anthropology. To begin to address these questions, this volume will investigate Bel ceramic technology from its pre-colonial inception around Madang, up until the present day, taking an explicitly anthropological approach to technological production and exchange. The first chapter of this monograph will introduce the necessary background literature that In 1871–1872, when the Russian scientist Nikolai Mik- 1) models pre-colonial production and exchange around louho-Maclay first explored the northeast coast, the Bilbil the northeast coast of New Guinea, 2) describes important speaking Bel groups that today live around coastal Ma- contextual aspects of Bel culture history, and 3) informs dang (Fig. 1.1) produced brilliant red-slipped pottery and the underlying theoretical perspective. The chapter will commanded an expansive trading network (referred to conclude by outlining specific research objectives along here as the Madang exchange network), which redistribut- with the approaches taken in this volume to address them. Hawaii Ethnographic exchange network Mariana Islands Philippines New Guinea Marshall Islands PACIFIC OCEAN Solomon Islands Samoa Vanuatu French Polynesia Fiji Equator Australia Tonga New Caledonia New Zealand Sep NE W GU ik IN BISMARCK SEA Madang Vitiaz Strait New Britain EA Huon SOLOMON SEA Solomon Islands Pa a Torres Strait pu Hiri Port Moresby Kula Mailu 0 1000 km Figure 1.1. Madang, the study area, in Near Oceania. Significant ethnographic trading networks redrawn from Lilley 2017 and Skelly and David 2017. 2       · .  Production and exchange on the northeast coast of New Guinea Horizontal links Ethnographic configurations On New Guinea’s northeast coast, a number of interconnected trading networks operated with a locational focus on the Vitiaz Strait, Dampier Strait, Astrolabe Bay, and the Huon Peninsula. Harding (1967) has described these networks as involving a large number of low-density, relatively acephalous groups that contributed specialised products and foodstuffs to a larger Vitiaz Strait social sphere or ‘super-system’ (see Lilley 1986: 35). This super-system relied both on coastal production, as well as the import of goods from ‘ethnic blocks’ in the interior, via hinterland middlemen, to port communities, and vice versa (Fig. 1.2). Once objects reached the port communities, they would be distributed by specialised long-distance traders: across the Vitiaz and Dampier Straits by the Siassi Islanders (Harding 1967), to the Huon Gulf by the Tami Islanders (Hogbin 1947), and to Astrolabe Bay and Madang by the Bel2 groups (Lawrence 1964; Mennis 2006a). Because many social groups produced a variety of different specialised material culture, we can view the broader northeast coast as being a landscape in which the diversification of production had itself stimulated the ongoing movement of products. Vertical links Specialist overseas traders Port communities Hinterland middlemen Interior ethnic blocks The impetus for the movement of materials in this network was both economic and social (Harding 1967: 165). It was economic in the sense that specialised production and exchange was essential to the survival of many island groups, including the Siassi, Tami, and Bel subsistence traders, who needed to exchange objects such as pottery for crops and animals to support their populations, which often exceeded the agronomic capacities of their small islands. In fact, some communities on the mainland even under-produced crops to create intentional food shortages that attracted subsistence traders bringing exotic goods, which could become politically efficacious objects or be used in further exchange with inland groups (Harding 1967). Trading was also social in the sense that many groups imported variations of items that they themselves made, or could have easily produced, as a means to enter into the super-system and maintain social connectivity. These connections were taken seriously, and exchange partnerships were often cultivated over multiple generations, inherited from father to son. This sometimes even involved the adoption of a trade friend’s son from an early age, which encouraged people to become fluent in the language of their counterparts. Terrell and Welsh (1997: 555) call these ‘inherited friendships,’ as they are termed along the Sepik coast. Around the Vitiaz Strait, these fictive kinship bonds established how people would behave in trading relationships, and allowed individuals not necessarily related by blood or marriage to act with familial and affinal qualities (Harding 1967: 166). In particular, these bonds allowed trade friends to participate in gift-giving and delayed reciprocity; although Figure 1.2. Model of production & exchange networks operating around northeast New Guinea (adapted from Lilley 1986: 36, 2017; see also Harding 1967: 16). giftings were seen as distinct and unconnected events, an immediate return would be highly inappropriate as it did not create the potential for a return visit (Sahlins 1965). As well as being maritime traders, the Bel groups in particular were set apart from the Siassi and Tami in that they were also specialist pottery producers. In this way they approximate the Motu pottery makers, who maintained the hiri trade links along the south coast of Papua (Mennis 2014). Within the hiri system, women would produce earthenware pottery, while men would assemble trading canoes and sail them along the coast, exchanging the pots for large quantities of sago and other crops produced by groups living to the west in the Gulf of Papua (Dutton 1982). A similar schedule of carefully timed pottery production and trade voyaging was in play amongst the Bel, who sailed on waing voyages to undertake dadeng (trade) with horticulturalists along the Rai Coast and beyond. Archaeological models Archaeological research around coastal Melanesia has attempted to track these practices back through time. This corpus of work suggests that production and exchange in the Late Holocene was just as vibrant as these ethnographic observations would have us believe; a crucial 2 Elsewhere referred to as Bilbil/Bilibili. 3 Chapter . Introduction time of change, featuring socio-political transformation, population migration, and cultural diversification (Spriggs 1997: 154–161). From about 2000 years ago, there was a gradual breakdown of the long-distance kinship and exchange connections that characterised the Lapita period in the third and fourth millennia before present (CathGarling 2017; Summerhayes 2007). By the last millennium, these had been replaced by smaller, but highly complex, regional exchange networks. Jim Allen (1982, 1984b, 1985) developed a model, based on work along the south coast of Papua, which suggested periods of trading that gradually became more intense punctuated the last few millennia. This intensification was due to accelerating production to acquire surplus goods and increasing consumer demand, which led to interaction becoming unstable, eventually breaking down. As depicted in Figure 1.3, Allen’s model suggests production and exchange networks became increasingly complex, but reliably disintegrated into a series of geographically smaller networks through time. This has been likened to speciation by network decay, whereby regional socio-technological processes gradually fragmented and diversified to form the complex networks of people and materials seen at ethnographic contact (Thomas 2009). ceramic hiatus until 1700 years ago, which in turn was followed by the emergence of standardised pottery designed for increasingly complex exchange networks. Although the precursors to ethnographic pottery production and trading can be seen as far back as the early second millennium before present, these networks only took on their recent configuration within the last millennium, intensifying within the last 300 years. Lilley sees these changes around the northeast coast as consistent with Allen’s (1985) model for southern Papua, with the far-reaching, highly mobile Lapita period breaking down into a set of smaller but substantially more complex local and sub-regional networks, characterised by Type X, Sio, and Madang style potteries. Although numerous research projects along the south Papuan Coast (Allen 2017; Bulmer 1979; David et al. 2010; Egloff 1979; Irwin 1985; Skelly and David 2017; Summerhayes & Allen 2007; Urwin et al. 2018), in the Massim (Bickler 1998; Chynoweth et al. 2020; Shaw 2016; Shaw and Dickinson 2016; Shaw et al. 2016, 2020), around the Huon Peninsula (Lilley 1988a), and on the Sepik north coast (Beaumont et al. 2019; Gorecki et al. 1991; Swadling et al. 1989; Terrell & Schechter 2011a), have previously described the emergence of complex pottery production In revisiting the pre-colonial Vitiaz Strait networks, Lilley and exchange networks, until very recently almost nothing (2017) describes ‘proto-systems’ in which the configuration has been known archaeologically about the Madang area. of pottery production and exchange were markedly differ- Most of what we do know about Madang’s pre-colonial ent to that observed ethnographically by Harding (1967). past stems from linguistic, oral history, and geological Along the northeast coast at Sio and in the Siassi Islands of evidence. This existing evidence will now be briefly prethe Vitiaz Strait, a low intensity production and exchange sented to frame our current understanding of Bel history of Lapita pots around 3000 years ago was followed by a and how these groups have come to occupy their unique position along the coastal margin and on offshore islands around Madang. Bel culture history Language Complexity On the northeast coast of New Guinea, Non-Austronesian languages that typify the interior of the island, including those of the ‘Madang’ and Trans-New Guinea families, result from the divergence and interaction of communities within New Guinea throughout the past 50,000 years (Foley 2000; Pawley 2006; Wurm 1983). The comparatively recent occurrence of the Oceanic sub-branch of the Austronesian language family, associated with Lapita pottery, emerged in the Bismarck Archipelago having split from the Eastern Malayo-Polynesian languages spoken in Cenderawasih Bay between 4000–3000 years ago (Blust 2013; Kirch 1997). The presence of Oceanic languages on the northeast coast of New Guinea today may represent traces of these initial Austronesian-speaking migrant communities as they moved eastward along the north coast, but, if that is the case, they have been mostly overwritten by subsequent westward ‘backwashes’ from the Bismarck Archipelago (Hooley 1971; Pawley & Ross 1993; Z’Graggen 1975: 39). Sp ati a le xte e nt o Tim fs ys te m s Figure 1.3. Allen’s wave-like model for the intensification and divergence of trade networks through time (adapted from Allen 1984b). Ross (1988) subdivides the large Oceanic subgrouping of Austronesian into three genetically linked clusters, all 4       · .  with a central origin in West New Britain: the Meso-Melanesian Cluster, the Papuan Tip Cluster, and the North New Guinea Cluster (Fig. 1.4).3 The North New Guinea Cluster covers all of the Austronesian languages along the northeast coast including Madang, the Huon Peninsula, the Sepik, and most of New Britain west of the Willaumez Peninsula. These speech communities are primarily restricted to the coastal margins and offshore islands, representing their more recent arrival to the mainland (Fig. 1.5). Based on linguistic reconstructions, it is likely that North New Guinea Cluster Austronesian speakers initially moved from West New Britain to the northeast coast of New Guinea, perhaps settling Manam Island before moving westward along the Sepik coast (forming the Schouten Chain). Following this there was a southward movement from West New Britain to the Huon Gulf area (forming the Huon Gulf family). Later still, some groups moved from West New Britain across the Vitiaz Strait and around coastal Madang (Ross 1988: 160–183; see also a summary in Lilley 1999). This formed the Ngero/Vitiaz language network, which stretches from Karkar Island to West New Britain and the Tami Islands, and comprises five sub-families, including Bel (Fig. 1.6).4 The Bel sub-family is spoken around coastal Madang Province, from Karkar in the north to the Rai Coast in the south. This sub-family is further subdivided into individual languages as defined by the Bel people: Gedaged and Bilbil around Madang town, Takia on Bagabag Island and the south side of Karkar Island, Megiar and Matukar on the adjacent mainland, Ham inland between the Gogol and Nuru Rivers, and Mindiri, Biliau, and Wab on the Rai Coast. Within Gedaged there are four dialects, Siar, Riwo, Sek, and Kranket, and these are mutually intelligible (Z’Graggen 1976). Oral tradition Oral histories give some insight into the restricted distribution of these North New Guinea Cluster languages around Madang. The most common etiological story reported by the Bel refers to the disappearance of their homeland; an island generally known as Yomba (Mennis 1980b, 1981a, 1981b). This homeland is reported to have been located somewhere beyond Madang, in the Bismarck Sea, but sunk into the waves many generations ago, forcing the occupants to flee to the northeast coast. The most 3 Greenhill and Gray (2009) group only the North New Guinea, detailed accounts describe a full sequence of events: large Papuan Tip and Willaumez portion of the Meso-Melanesian earthquakes, Yomba’s eruption, subsidence of the island Cluster together using phylogenetic reconstruction, but more into the sea, and a resulting tsunami (Mennis 2006b; Fig. recent analyses at the Max Plank Institute, using lexical data 1.7). with a Bayesian algorithm, confirm Ross’ (1988) groupings based on morphological and phonological innovations, brack- 4 Following Ross (1988: 121); cf. Hooley (1971) and Z’Graggen eting the Papuan Tip, Meso-Melanesian and North New Guin(1975), who define the near-equivalent ‘Siassi Family,’ extending ea clusters strongly together (Simon Greenhill pers. comms from Karkar to the Huon. 2015; Malcolm Ross pers. comms 2015). 0 500 km Meso-Melanesian Cluster Madang North New Guinea Cluster Papuan Tip Cluster Figure 1.4. Austronesian language clusters within the West Melanesian Oceanic subgroup (adapted from Ross 1988). 5 Chapter . Introduction Austronesian Non-Austronesian Bel family 0 50 km Figure 1.5. Major language stock boundaries in Madang Province including the Bel language family (adapted from Z’Graggen 1975). Proto Oceanic Proto North New Guinea Proto Schouten Proto Huon Gulf Proto Ngero/Vitiaz (see Ross 1988) (see Ross 1988) Proto Ngero Proto Tuam (see Ross 1988) Proto Vitiaz Proto Bariai Korap (see Ross 1988) SW New Britain Proto Mengen (see Ross 1988) (see Ross 1988) Proto Bel Mindiri-Western Bel Eastern Bel Western Bel Bing-Wab Bilbil-Megiar-Takia Megiar-Takia Takia Bilbil Gedaged Mindiri Matugar Ham Bing Wab Megiar Roinji Nenaya Sio Tami Mangap Barim Kilenge Lukep Maleu Malasanga Singorakai Figure 1.6. Bel languages branching from North New Guinea Cluster (adapted from Ross 1988, 2008). Some informants referred specifically to Hankow Reef, Many were less clear on the location or said it was behind situated between Bagbag Island and Crown Island along Kranket Island or Yabob Island, closer to the Madang coast. the Bismarck Volcanic Arc, as the former Yomba Island. Others mention three islands close together that sank – 6       · .  Kinemba, Yomba, and Kerata – while others again say these were all clan areas on one larger island. It definitely was not the same place as Arop/Long Island and its destruction occurred prior to the major Arop/Long eruption in the mid-1600s. The Arop/Long eruption was the most significant volcanic event around New Guinea’s northeast coast within the last millennium, effecting both coastal and highland populations and coming to be known as the ‘time of darkness’ (see Chapter 2). The Bel people, at that time already based around coastal Madang (Mennis 1978, 2006a: 5–13, 2006b), recall stories of Arop/Long Island erupting and the subsequent ‘time of darkness’ in which the ash fell for three straight days, landing on gardens and house roofs. People were afraid that their houses would collapse so had to frequently brush the ash away, and there was a food shortage as many gardens were spoiled. 100,000 years (Silver et al. 2009), and no Yomba ash layer has so far been identified in stratigraphic profiles as it has for the Arop/Long eruption (Blong et al. 2018; Johnson 2013). This conflicts with many of the oral testimonies, which suggest that Yomba’s sinking was associated with an eruption and subsequent ash fall. Smaller hydromagmatic eruptions caused by superheated groundwater, along with frequent earthquakes around Madang, were proposed to have caused lateral collapse of the reef causing tsunami, similar to that which affected Ritter Island in 1888 (Mennis 2006a). However, this does not align with marine mapping in 2004 which showed no signs of a major gravitational collapse (Simon Day pers. comm. 2015). The mapping did show evidence that the reef was subsiding and Hankow may have formerly supported one or more smaller islands resting on the coral base. Ongoing research is aiming to solve the Yomba conundrum by linking the oral histories, In the present volume ‘Yomba’ will be used as a heuris- to their likely geological correlates (Mennis & Day in prep.). tic to refer to a possible ancestral homeland occupied somewhere beyond Madang, prior to the movement of All of these strands of evidence provide a good working Bel speakers onto the mainland. This acknowledges the framework, but it remains for archaeology to put forward histories attested to by local Bel groups, while also paying testable hypotheses that critically examine population hisattention to the complexity and contradictions inherent in tory along the northeast coast and untangle how, and perthe oral history records (see Damm 2005 and Mason 2000 haps why, the complex production and exchange networks for methodological discussions). around Madang emerged. Geology Ceramic technology Although Hankow Reef has been argued to be the rem- We can begin to model the events immediately leading to nants of Yomba Island (Mennis 2006b), sonar survey at Bel occupation around Madang by putting forward a series the reef demonstrates it has not erupted within the last of scenarios, which can be tested through archaeological Island size Large Island type Volcanic Non-volcanic Before Arop Time of disaster 10 Generations ago Location Unknown * 9 Unknown Unknown 8 Unknown Beyond Yabob/Kranket Vitiaz Arc Same as Arop? Unknown No Escape Escaped to mainland or islands Island language Bel Island pottery Unknown Potters Decent from Island Unknown Non-potters Unknown Another destroyed island Yes 0% 50% No Unknown 100% Figure 1.7. Oral testimonies of 23 Bel speakers regarding Yomba Island. *near Madang (adapted from Mennis 2006a: 10). 7 Chapter . Introduction excavation, radiometric dating, and ceramic analysis. Figure 1.8 illustrates four possibilities, relating Bel migration scenarios to their expected archaeological signatures: a) if there was a catastrophic, single migration of Bel speakers from a ‘Yomba’ homeland we would perhaps expect to find a handful of pots salvaged from Yomba and an abrupt start to locally-made pottery production in Madang, lining up chronologically with geological and oral history evidence. This would then be followed by a gradual change to local ceramic manufacture over the centuries; b) if there was trade between Yomba and Madang prior to that Bel migration, we would expect exotically made Madang style pots, with an abrupt change to local production at Yomba’s disappearance; c) alternatively, if there was a gradual diaspora to Madang, we would expect to see locally produced Madang style vessels prior to Yomba’s disappearance; and d) if this diaspora was characterised by high levels of ceramic transfer between the two potting centres, we would expect both locally and exotically produced Madang style pots around Madang itself, prior to Yomba’s disappearance. More broadly, other archaeologists have previously attempted to understand the movement of Austronesianspeakers onto the north and northeast coast of New Guinea, alongside transformations to maritime trade networks. Most notably, Lilley (1988a) described the dramatic increase in pottery production and exchange around the Vitiaz Strait in the recent pre-colonial past, leading up to ethnographic contact. He also suggested the archaeological evidence supports a generally westward push of Single migration narrative- no interaction No Madang-style pottery present Madang-style pottery present Local manufacture Bel on Yomba Island Bel around Madang Yomba disappears Single migration narrative- interaction No Madang-style Madang-style pottery present Exotic manufacture Local manufacture Bel on Yomba Island Bel around Madang Yomba disappears Diasporic narrative- no interaction No Madang-style Madang-style pottery present Local manufacture Bel on Yomba Island Bel around Madang Yomba disappears Diasporic narrative- interaction No Madang-style Madang-style pottery present Exotic/local manufacture Local manufacture Bel on Yomba Island Bel around Madang Yomba disappears Figure 1.8. Four timelines illustrating possible Bel movement from a Yomba homeland to Madang with their associated archaeological signatures. 8       · .  Austronesian-speaking groups from New Britain back onto northeast New Guinea, in accordance with linguistic reconstructions (Lilley 1999). However, this is contested by other archaeologists who see the emergence of the first red-slipped ceramics in the area resulting from a generally eastward expansion of potters along the north coast. Terrell & Welsch (1997) initially speculated that apart from tentative traces of Lapita pottery, with links ultimately to the Bismarck Archipelago, the earliest evidence for the spread of pottery traditions on the Sepik north coast occurred around 2000 years ago and arrived from the west. These early Sepik ceramics were seen to have strong affinities with Metal Age red-slipped pottery from the Moluccas and West Papua. Moreover, in an examination of modern ceramic manufacture around New Guinea, Pétrequin and Pétrequin (1999, 2006) also suggest that the origins of the Madang traditions were certainly from the west (i.e. Island Southeast Asia). The Pétrequins cluster together the Bel pottery production sequences into a larger group: the technique of paddle and anvil with the preassembly of a rim-neck preform. These industries require high levels of skill and an extended learning process, especially for larger vessels. Other groups that use these techniques occur in south and central Maluku (Indonesia), and on the north New Guinea coast at Sapij and Tumleo Island. Clearly this contradicts the narrative described above based on language, oral history, and geology (Fig. 1.9). To examine how the Madang potting traditions articulate with these various models of population movement and technological change, we require a more detailed understanding of how technological practices are intertwined with social boundaries and group interaction, and, moreover, what these processes might look like in the archaeological record. Theoretical foundations The production, use, and exchange of objects such as earthenware pots can represent fairly mundane, habitualised activities on the day-to-day scale. However, over a lifetime they become central processes that shape an individual’s cognition, physique, and sociality: they help make people who they are (Dobres 2010; Gosden 2008; Ingold 1993a, 2000: 406–419; Latour 1993; Malafouris 2010, 2013; Sofaer 2006). As will be argued in the following section, these activities generate and transform not only the material world itself, but the manufacturers, consumers, and traders that inhabit these worlds (Gaffney 2019; Gosden 2012). Perhaps more importantly for understanding diachronic change, the reproduction of these activities across generations constitute technological processes, which can become diagnostic of production groups and their unique genealogies of practice. Aspects of production To examine how ceramic manufacture originated and transformed through time along the northeast coast, we need to understand some foundations of craft production and knowledge generation. Fundamentally, it is through habitual material engagement that the parameters of human thought, movement, and action are established (Inoue 2006; Sterne 2003). These parameters, internalised and expressed as bodily dispositions, were originally described by Mauss (1934) as ‘techniques of the body.’ When these dispositions are dictated by purpose, formed through on-going processes of closely related technical activities (Bloch 1991), they are more often labelled as ‘tacit knowl- 0 AITAPE Terrell and Welsch 1997 Petrequin and Petrequin 1999 500 km WEWAK Koil Manam B I S M A R C K S EA Karkar Bagabag Hankow Reef Ross 1988 Lilley 1999 MADANG ASTROLABE BAY Arop Ritter NEW GUINEA VITIAZ STRAIT Umboi N E W B R I TA I N Siassi Islands HUON PENINSULA LAE HUON GULF 0 100 km Figure 1.9. Two competing scenarios for the movement of potting communities and Austronesian languages onto the northeast coast. 9 Chapter . Introduction edge’ (Polanyi 1966), ‘generative know-how’ (Lemonnier In this sense, technology is not inert but always in flux 1992), or ‘motor habits’ (Arnold 1998; Crown 2001; Minar (Roux 2003). This thinking derives from the French school 2001; Minar & Crown 2001). Here, the term ‘embodied of Techniques et Culture that describes technology as the knowledge’ is used throughout the book as a gloss for the study of techniques, which can be codified as the chaîne variable ways to describe knowledge generated by routines opératoire (operation sequence): the step-by-step technical of habitual action, and expressed as rhythm, movement, choices and actions that underlie the procurement, manuand dexterity (Jørgensen 2013; Sofaer & Budden 2013). facture, distribution, use, and discard of material culture (Lemonnier 1986, 1992, 1993; Leroi-Gourhan 1943, 1945). Embodied knowledge is difficult to codify and share be- By recording the chaîne opératoire, whether from partween individuals through speech or depiction and must ticipant observation or archaeological analysis, it allows be experienced to be acquired. It is also knowledge that us to sketch out how embodied knowledge is generated increases with degrees of engagement and is very difficult amongst individual craftspeople, and how the techniques to unlearn (Stark & Longacre 1993). In this way, expert of production come to form historically contingent techcraftspeople are often disposed to think with, rather than nological processes, unique to specific social groups. The about, their body and tools (Ingold 2000: 157–171, 406– dramatic increase in Anglophone scholarship examin419). For instance, as new movements or objects are first ing production as a dynamic technological process over encountered in the learning process they are conspicuous the last three decades is testament to the traction of this to the user and consciously considered, but after degrees approach (e.g. Bar-Yosef & van Peer 2009; Chazan 2009; of habitual engagement they become invisible extensions Chilton 1998; Conneller 2008; Edmunds 1990; Hitchcock of the self, except upon key moments of reflection (see & Bartram Jr. 1998; Jeffra 2015; Jones 2002; Knappett 2005a; Heidegger’s 1927 concepts of ‘availableness’ and ‘occurrent- Kreiter et al. 2014; Schlanger 1990, 1994; Stark 1998b; Stark ness’). Central to the expression of embodied knowledge is et al. 2000; Walls 2016; for summaries see Lemmonier this concurrent interplay between conscious intentionality 2013; Loney 2000). and routinised action. Aspects of exchange The production groups in which craftspeople live, understood here to operate within larger communities of prac- Sharing technical knowledge is one form of exchange tice that share common ways of doing and making things that occurs within and between social groups. This par(following Lave & Wenger 1991), create culturally specific ticipatory learning is foundational to the community of boundaries within which to experience the social and practice, because novices and experts facilitate ongoing material world (Gosden 1994; Keller & Keller 1996; Ma- correspondence by attuning themselves to shared ways hias 1993). As individuals actively engage with embodied of learning. Over time, embodied knowledge in the form knowledge, they attach themselves to, or distance them- of habitualised technical activities can be shared across selves from, particular social forms in these communities the generations. There are several modes of participaof practice (Dobres & Hoffman 1994; Kohring 2013; Shil- tory learning: ‘vertical learning’, involving the creation of ling 2005: 187). In this way, the production process is not a knowledge between parents and children; ‘oblique learning’ closed system, but interrelated with, and inseparable from, involving adults not immediately related to the child (e.g. other technological and social practices within the com- extended kin or unrelated adults); and ‘horizontal learning,’ munity, be they material, political, ideological, economic, which occurs between people of similar age and standing or symbolic (Dobres 2000). Embodied knowledge can within the same production group, or between different then distinguish specific social groupings within broader groups (Hosfield 2009: 46). These modes of learning are communities of practice. usually engaged through processes of teaching and pedagogy, or observation and imitation. This knowledge, despite being differentiated between social groups, would be unchanging were it not for people’s In the case of crafts such as pottery making, Shennan ability to creatively respond to technical challenges and (2002) asserts that learning usually occurs vertically or opportunities (Chomsky 1982; D’Agostino 1984; Daley obliquely, rather than horizontally (see also Shennan & 1982). Such creativity – the human capacity for sensory Steele 1999). In a similar way, most studies that try to unimprovisation and imagination – produces variation, and derstand the distribution of Pacific pottery take the ungradually modifies the boundaries of the social aesthetic stated view that specific motifs were shared vertically by and the technological practice (Gaffney and Zmigrod production groups, and hence can be used to trace group forthcoming). In recent archaeological literature embod- lineages. Conversely, more recent evolutionary approaches ied knowledge has been projected onto both the bodily have argued that the transmission of Lapita design was timescales of the single craftsperson, as well as deeper horizontal, passed between island groups, through unin archaeological time, interpreted in the framework of biased or biased forms of social replication (Cochrane practice theory (Dobres 2000) and process philosophy & Lipo 2010; cf. Summerhayes 2001b). Although this (Gosden and Malafouris 2015), to understand how an in- literature working at the macro-scale has emphasised dividual’s productive activities can generatively shape and how ‘transmission’ processes can lead to the proliferatransform technological processes through time. tion of ideas and cultural elements (e.g. Cochrane 2002; Cochrane et al. 2013), the present study tries to tease apart 10       · .  the specific mechanisms of how technical elements may be shared across space and time, and understands knowledge exchange in the context of socialisation and enskillment (see Crown 2001; Huntley 2006; Minar & Crown 2001; Van Keuren 2006). In this way, at the micro-scale, the exchange of decorating practices probably involved an intricate and socially modulated arrangement of vertical, oblique, and horizontal learning. Ethnographic case studies, for instance, demonstrate that modes of learning are situated and change as the student moves through the society, grows older, generates embodied knowledge, and particularly, as is often the case in Melanesia, as the individual moves between social groups within the community of practice (e.g. Hewlett et al. 2011; Jordan & Shennan 2003; Tehrani & Collard 2009). how these interactions have changed through the centuries. It is important to distinguish these different processes of interaction in the archaeological record, be it the sharing of cultural knowledge or the exchange of tangible objects, because it has implications for the organisation of the community of practice and the extent of interconnectedness between past social groups. As the quote from Ingold (2011) at the beginning of this chapter states, we must follow the materials through their life-histories and learn the movements that energise and transform these materials if we are to understand something of how technological processes can inform us about transformations to society through time. More tangible forms of exchange, involving objects for objects, also occur between individuals and groups. These exchanges materialise social bonds, reciprocity, and the logic of the economy, and so shared practices of object exchange are essential to the community of practice (Thomas 2009). These practices represent another aspect of the technological process – another link in the chaîne opératoire – and therefore it is not simply the formation of abstracted social relations, but the materiality of exchange itself, that circumscribes bodily dispositions and cognitive frameworks. Research objectives Objectives and approach Having begun to untangle how the processes of production and exchange take place in society we are better equipped to examine the core research concerns of this book: the history of the Bel people and how their production and exchange network emerged. However, it is often more difficult for an archaeologist to investigate why (as opposed to how) technological changes took place in the past owing to the multitude of factors affecting internal/ external change and variation, including gender dynamics (Kramer 1997; Mahias 1993), social identity (Gallay 2007; Knappett (2011) categorises material exchanges as taking Gosselain 2008), or community fragmentation (Scheans place at different scales: within micro-networks (within 1977). Despite these challenges, and because archaeology the social group), meso-networks (locally), and macro- is both a theory-building and narrative-building discipline, networks (regionally). Exchange objects themselves the explicit objectives of this volume are twofold. tend to acquire, produce, and shed meaning as they pass through these different scales of network (Gosden & Mar- The monograph aims first to explore how processes of potshall 1999; Clarke & Torrence 2011). Although Malinows- tery production and exchange emerged amongst the Bel in ki’s (1922) study of kula demonstrated that some objects of the pre-colonial past. This will involve describing the maexchange are inalienable from their own life histories and terials and techniques of production, and the movements the life histories of their previous owners (see also Weiner of finished products. Using an approach derived from the 1992), Aswani and Sheppard (2003) show that trade items French school of technology, the chaîne opératoire, the socan be viewed as inalienable gifts or alienable commodi- cial groups producing pots within broader communities ties (following Mauss 1925) depending on the specific so- of practice will be systematically distinguished, along with cial context of the exchange and the relationship between the nature of exchange between groups. The chaîne opératrading partners. toire is an effective method for addressing technological processes and the generation of embodied knowledge in In this way, like production technology, exchange net- the past (Knappett 2011). In this way, we can build up a works are far from static; they morph and mutate through picture of how the Madang network emerged, and how time as social relationships are tended to, cultivated, and changes to production and exchange occurred over time. severed (Strathern 1996). In this vein, Knappett (2005b: 66) notes that exchange networks are only temporary organi- Second, the study is designed to describe the specific sesational forms, which are produced by the on-going flow quence of events leading up to ethnographically docuof materials and the movement of traders. Although, these mented production and exchange networks (i.e. why trade links can regulate broader socio-political relation- technological changes took place). This will investigate the ships, while nodes of interaction, whether in the house- origins of the Bel on the northeast coast of New Guinea: hold, at the specified trading point, or across the landscape, a crucial area of human population movements into and often acquire fixed locations and meanings (Ingold 1993b), around Melanesia. This seeks to place the communities these systems are all prone to gradual modification across of practice around Madang in the broader pre-colonial intergenerational time. context. The current research is concerned with the timing of Bel occupation around Madang and the direction In archaeology, examining the materials of exchange pro- of migrations (from east or west; ‘Yomba’ or somewhere vides the physical evidence for these past interactions, and else). This will involve investigating the local distribution 11 Chapter . Introduction of former settlement areas, the nature of material culture remains, and the timing of trade and exchange between different communities. This requires the provisional construction of a temporal framework, based on chronometric, environmental, linguistic, historical, and material culture evidence, which can then be used as an interpretive foundation to address Bel culture history. The monograph is composed of eleven chapters. It progresses in three stages to achieve the objectives outlined above. Because production and exchange are historically contingent processes, the first stage necessarily presents the unique local and regional context. Chapter 2 introduces the physical landscape of the study area, providing a background to major resource zones available for production, and dramatic events, which punctuated Bel culture history. Chapter 3 outlines the ethnographic and oral history evidence for production and exchange along the northeast coast, providing an ethnohistorical analogue to trace earlier processes of production and distribution back in time. Chapter 4 describes ethnographic fieldwork I carried out to record modern pottery production among the Bel communities. Ethnography can be a central interpretive tool for investigating past technologies and provides another modern analogue for the past, a detailed ceramic chaîne opératoire to be compared to pre-colonial objects. Chapter 5 is a review of all available archaeological literature dealing with Madang ceramics. This presents what is known about raw material, form, decoration, and the timing of Madang production & exchange around northeast New Guinea and identifies major gaps in our knowledge. Empirical investigations of ceramic production and exchange are then presented in the second stage, in an attempt to delineate production groups operating within larger communities of practice. Chapter 6 describes new archaeological investigations around Madang, giving context to the pottery artefacts examined in the volume. Survey around the Madang area is discussed, and the excavation of two sites – Nunguri and Tilu – is described in detail with especial reference to their stratigraphy, dating, and material remains. Chapters 7, 8, and 9 (supplemented by online Appendices A and B) outline the technological analyses of pottery from the field investigations. The technological classification of pottery is completed systematically in three parts (one part per chapter) in order to target specific aspects of the technological process. First, Chapter 7 examines pottery-forming processes. This develops the theory and methodology behind technological analysis and outlines the results of the analysis. This classification sorts specific sherds into groups, based on the techniques used to form them. Chapter 8 then outlines the method and results behind the second stage of classification, which focuses on raw material procurement and finished pot distribution. This uses geochemical techniques to subdivide formal artefact classes into similar provenience groups based on clay and mineral tempers. Chapter 9 compares formal and geochemical groups with decoration and surface treatment. This chapter then summarises the pottery results to establish whether specific raw materials are as12 sociated with specific forming or decorating techniques, contributing to specific technological/stylistic groupings. The third stage of the monograph sees these archaeological analyses discussed with reference to models of Melanesian production and exchange (e.g. Allen 1985; Harding 1967; Lilley 2017), information on Bel culture history, and technological process theory. Chapter 10 is a discussion of the Madang ceramic classification in the context of process theory. The nature of pottery production and technological knowledge is examined during different chrono-stratigraphic periods. The Madang results are then compared to other available datasets around northeast New Guinea to distinguish how exchange changed through time. Chapter 11 is the concluding chapter that ties together the local context of landscape, language, oral history, ethnography, and archaeology in an attempt to distil a cohesive Bel culture history. This chapter addresses the crucial themes introduced in Chapter 1 about eastern/western population movements and the story of Yomba and concludes with some broader implications for Melanesian anthropology and archaeology. Chapter 2. The Archipelago of Contented People er Straits into the Bismarck Archipelago. To the north lies the long stretch of the Sepik north coast and the estuaries of the Sepik and Ramu Rivers. We found the mainland opposite Bilbil very fertile and mainly flat. Surrounded by low ranges of hills, it is covered by tropical forest. — Otto Finsch (1888)1 The Madang coast is a dynamic and changing landscape, both in terms of natural transformations and cultural processes. This chapter describes the physical landscape of the study area as people inhabited and experienced it in the pre-colonial past. In this way, the place specificity of Madang is described, with an especial reference to those factors that facilitated production and exchange. Physical aspects of Madang’s environment, land systems, and geology are explored, as these factors have pivotal effects on resource procurement, manufacture, and distribution. Major geophysical processes including tectonic and volcanic events are also presented, as they punctuate Bel social memory and may be essential to understanding something of the society’s pre-colonial culture history. Physical geography of Madang Local setting Madang lies on the northeast coast of the large island of New Guinea, adjacent to the Bismarck Sea in the South Pacific (Fig. 2.1). The coral reef and lagoon system that fringes the shoreline is the largest on the north coast. In this area, there is a dichotomy between the expansive mainland on the one hand and an archipelago of small offshore islands around the lagoon on the other, described as the archipelago of contented people by early explorers. To the south, Astrolabe Bay forms the physical and social connection between Madang and the Rai Coast, which the locals nowadays divide into the Nambawan Rai Coast and the Nambatu Rai Coast.2 The Rai Coast as a whole extends from Erima Harbour in the west, to the Huon Peninsula in the east. Further out to sea lies a string of island volcanoes, which act as stepping-stones across the Vitiaz and Dampi- The coastal region of Madang District runs from Sek Island in the north to Balima village in the south, forming the western coast of Astrolabe Bay. Alignment is approximately north-south, between latitude 5° 00′ and 5° 20′ south of the equator. In the centre, the modern town of Madang lies on the small Schering Peninsula, which forms the southern edge of Madang Lagoon, a large reef system of bays, islands, and the mouths of rivers draining the nearby hills. The lagoon is an even 30–40 m deep but there are numerous shallow patch reefs and coral rubble islands that support fringing reefs, all encompassed by a large barrier reef on the seaward side (Jebb & Lowry 1995). Five major passages allow for access through the barrier reef into the Bismarck Sea. The coastal plains inland of the lagoon border steep foothill areas immediately to the west and the Adelbert Range lies beyond that. To the south of the lagoon are the mouths of the Gum and Gogol rivers, and four offshore islands: Yomba3 (also called Mareg), Yabob, Urembu, and Bilbil (Fig. 2.2–2.3). Raw material zones Land systems This study area lies across several major land systems, characterised by different soils and underlying clays suitable for pottery making (Fig. 2.4). The poorly drained Madang Land System, which forms an almost continuous stretch of land 1–2 km wide running from Madang town in the south to Hansa Bay, adjacent to Manam Island, in the north consists of raised coral platforms, beach ridges, and minor lagoon fills (Haantjens et al. 1976: 22). Near the coast, shallow soils overlie coral bedrock, and on the inland side, young, fine-textured alluvial soils predomi3 Not to be confused with the legendary Yomba Island, the ancestral homeland of many groups on the Madang coast today (see Chapter 1). The name ‘Yomba’ to refer to the island north of Yabob (and also an area on the Schering Peninsula) is a later translocation. 1 Translated by Christiane Harding in: ‘Archipelago of the Contented People? Madang (Friedrich-Wilhelmshafen) in 1884,’ edited by Mary Mennis, 1996. 2 Referring to the ‘first’ and ‘second’ Rai Coast. 13 Chapter 2. The Archipelago of Contented People 0 500 km AITAPE Schouten Group WEWAK Manam Karkar Bagabag MADANG ASTROLABE BAY NEW GUINEA BISMARCK SEA Hankow Reef Crown Arop Tolokiwa Umboi VITIAZ STRAIT Sakar Ritter NE W BRITA IN Siassi Islands HUON PENINSULA LAE HUON GULF 0 100 km Figure 2.1. Madang on the northeast coast of New Guinea. Dashed blue line marks plate boundary. Figure 2.2. Aerial view looking east over coastal plains to Yabob Island, Urembu, and Bilbil Island (2015). 14       · .  145°50’ Bidup Midibur ID AWAN Admosin Island Sek Island Alexishafen Donip Kananam Tabad Island Malmal Island NAZ AWAN Malmal Sinup Island Baitabag Wanad Island Tabad Island Riwo Island Riwo DAM AWAN Kamba Nagada Nobonob Masas Island 5° 10’ Tab Island Paeowa Island Siar Siar Island AWAN BIZIWAN Kranket Island Bilia Island Bilia Mis GOD AWAN Schering Peninsula Ponim Sissiak Madang Yabob-Down-Below Yabob-Up-Top Yahil Hilu Hudini Yomba Island Yabob Island Urembu Island BISMARCK SEA Ord Bilbil Aguru Bahor So Sien Ohuru Bilbil Island Umium Ornuru 5° 20’ Malaga Dogia Maraga Hook Balima 0 145°50’ 5 MADANG PROVINCE km Figure 2.3. Coastal Madang District, northeast New Guinea. 15 Astrolabe Bay Chapter 2. The Archipelago of Contented People 145°50’ 5°10’ Madang BISMARCK SEA 5°20’ 145°50’ 0 5 km Madang Astrolabe Nubia Papul MADANG PROVINCE Gal Figure 2.4. Land systems of the Madang District (Haantjens et al. 1976). 16 Astrolabe Bay       · .  nate but lack substantial clay deposits. Shingle and sandy beaches and small mud flats dominate the coastal margins. To the south is the Nubia Land System with broad beach ridges occurring close to the mouth of the Gogol River. On older inland ridges, organic topsoils overlie clay deposits, which in turn overlie sandy alluvial soils (Haantjens 1976: Table 10; Haantjens et al. 1976: 50). The southernmost area comprises the Astrolabe Land System, formed by the alluvial plains of the Gogol and Nuru Rivers. This land system consists of colluvial aprons on the inland side and alluvial soils overlying good clay deposits, with minor reef corals and marine sands on the coastal side (Haantjens et al. 1976: 46). At the mouth of the Gogol itself is the minor Papul Land System with clayey alluvial soils. Inland, the mountainous country of the Adelbert Ranges, with weathered red-brown clay soils, forms the Gal Land System. of Madang are typified by these black volcanic minerals, while the offshore islands feature white sands derived from coral and limestone foundations. Geophysical processes Tectonic activity A series of geophysical processes have structured the development of these land systems and underlying geological zones around Madang and reverberate in Bel oral tradition. Most influentially in both contexts, the Madang coastline is generally emergent, defined by the long-term uplift of well-preserved reef deposits forming the coastal ranges (Haantjens et al. 1976: 21; Robbins 1976: 14). Similar Quaternary reef exposures are evident on the Huon Peninsula, east of Madang. Average uplift rates of 3.5 mm per year have persisted there for the past 300,000 years Geology (Chappell 1974; Chappell & Polach 1991; Chappell et al. The geology of coastal Madang is relatively young and 1996). However, around Madang intermittent subsidence most of its development took place during the mid–late caused by large earthquakes has been proposed to account Tertiary and in the Pleistocene (Fig. 2.5; Reiner and Mab- for the lack of massive coastal terraces found on the Huon butt 1976: 73). This underlying geology structures the dis- (Tudhope et al. 2000). tribution of clay and temper sands used for pottery along with raw stone materials. Structurally, the region is part of Analysis of the exposed Holocene reef deposits along the the Palaeogene Volcanic Arc (Bain 1973), which accounts coast suggests a series of uplift and possible subsidence for most of the physiographic features of the area (Loffler events against a backdrop of gradually falling sea levels 1977: 7). Madang Lagoon lies in the Pleistocene–Holocene (Tudhope et al. 2000: Fig. 5). These coseismic land moveWandokai Limestone, a coral limestone plateau, which ments have caused changes to Late Holocene relative sea extends from the northernmost extent of the lagoon to levels around Madang to be more rapid than eustatic sea the confluence of the Gogol and Nuru Rivers in the south level change. About 3000 years ago, uplift caused reef (Robinson et al. 1976: 3). This formation comprises massive terraces to rise by over 4.5 m, while possible subsidence crudely bedded and cavernous biocalcirudite, calcarenite, caused reefs to descend about 1.5 m at 2400 years ago and calcilutite, and mudstone, with subordinate arenite clasts, more than 0.5 m about 1200 years ago. Most recently, furand conglomerates, most of which are unsuitable for lithic ther uplift caused raised terrace elevations of 2–3 m somemanufacture but offer the potential for calcareous sand time in the last 1000 years (Morgan et al. 2005; Tudhope et tempers along the water’s edge. al. 2000). Recent research along Madang Lagoon suggests that the latest uplift event can be dated to about 550 BP by More suitable stone resources for artefact manufacture marine radiocarbon.4 The oldest dated geological deposits are located inland from the lagoon, where several rivers on offshore islands are only about 3000 years old, suggestdrain through the Tertiary deposits of the Finisterre Vol- ing that the initial uplift of ~4.5 m was the event which canics, with basalt and andesite flow breccia, tuffaceous first raised the reef systems significantly above sea level greywacke, tuff, agglomerate, argillite, and minor lime- (Day 2012; Mennis & Day in prep; Morgan et al. 2005). stones. The Meiro and Gum Rivers drain through the Gusap Argillite, with red and green cherty argillite, chert The seismic events responsible for these past land movebeds, greywacke, siltstone, subordinate basalt and andesite ments resulted from the rapid convergence of the Pacific breccia, dolerite dykes, and cherty limestone. South of the and Australian plates (Baldwin et al. 2012). This converlagoon is a recent Holocene alluvial deposit which extends gence continues today, so Madang is host to large earthfrom just north of the Gum River mouth to the Rai Coast quakes and volcanic eruptions. The ranges west of Main the southeast. These deposits are formed by the floods dang are particularly unstable even during relatively small from major rivers draining the Adelbert and Finisterre earthquakes (Pain 1972), which are a significant factor in Ranges and commonly comprise gravel, sand, silt, and landform development (Loffler 1977: 161). In 1970, a viomud, along with good clay deposits. The Gogol River and its tributaries drain inland deposits of the Kabenau Beds, 4 Note that U-series dating is needed in order to calibrate the which extend to the Sek River, containing well-bedded calmarine reservoir effect on coral derived C14 dates, especially careous clasts, siltstone and mudstone, conglomerate, and because upwelling deep water along the Rai Coast mixes with Madang waters and there is a very thin surface layer of brackminor limestone and lignite. Sand mineral grains around these river mouths are predominantly magnetite admixed ish water outflowing from the Gogol River (Simon Day pers. with volcanic pyroxene, olivine, and amphiboles (Robincomm. 2015). The ~550 BP date uses a DR of 0, so any variation son et al. 1976: 20). The beaches along the mainland, south to the modeled marine reservoir will be problematic. 17 Chapter 2. The Archipelago of Contented People 145°50’ 5° 10’ Madang BISMARCK SEA 5° 20’ 145°50’ 0 5 km Wandokai Limestone Gusap Argillite Alluvium Finisterre Volcanics MADANG PROVINCE Kabenau Beds Figure 2.5. Geological zones around Madang District (Robinson et al. 1973). 18 Astrolabe Bay       · .  lent magnitude-7.0 earthquake in the Madang area led to Summary landslides that removed over 60% of vegetation in some catchments (Johns 1986). An earthquake of magnitude 6.0 Throughout the Late Holocene, the Madang coast has been or above strikes Madang about every three years (Ever- characterised by dramatic physical change. Over time, the ingham 1975). landscape mosaic would have been reimagined through shifting cultivation, burning, hunting, collecting, erosion, and deposition. This gradual landscape change over the Volcanic activity longue durée, not always observable on the bodily timeThe Bismarck Volcanic Arc, comprising island volcanoes scale, was punctuated by dramatic events that had prosuch as Manam, Karkar, Bagabag, Lotin/Crown, Arop/ found impact on social groups over comparatively short Long, Umboi, and Ritter, lies to the north and east of Ma- time periods. These particularly include volcanic erupdang at the boundary of the Pacific and Australian plates. tions (with associated ash fall and tsunami) and earthThe islands are made up of thick, extensively reworked quakes (with associated uplift, subsidence, collapse, sea pyroclastics with basalt and andesite rocks (Johnson et al. level shifts, and tidal waves). The impacts extend beyond 1972). Many of these volcanoes were active in the human the initial force, also comprising aftereffects and recovpast and remain active today (Jebb and Lowry 1995), and ery time. Such events perforate the social memory of the Karkar and Manam continue to pose a high risk for people coastal people: the Yabob are still shown the level that the on the coast (Davies 2012). Arop/Long tsunami came up to on their island (Mennis 1981b: 23), while the Bilbil point out the tree into which a In the last millennium, the most significant eruption event man and his canoe were thrown during a tsunami; when around New Guinea’s northeast coast was that of Arop/ he climbed down, the ground was littered with fresh fish Long. The dating of the last major eruption is problematic, (Mennis 1981b: 17). but following several lines of evidence, both Blong (1982; Blong et al. 1982) and Polach (1981) prefer a date of the mid The landscape also expedited or inhibited local producseventeenth century, and Haberle (1998) further narrows tion. It is important to note the environmental disparity it to 305–270 cal. BP (i.e. AD 1645–1680). This eruption is between the Madang Lagoon area and that south of the estimated to have had a value of six on the volcanic ex- Schering Peninsula near Yabob and Bilbil Islands. Those plosivity index, on par with Krakatoa, and had significant groups living adjacent to the alluvial Astrolabe or Nubia effects on the New Guinea region. The associated ash fall, Land Systems would have had more direct access to potforming the Tabito Tephra and recorded through oral his- tery producing clay than those living around the coral tories in the Highlands as the Tudak (Time of Darkness), derived Madang Lagoon area. A variety of different temprevented crop growth, caused starvation, and collapsed per sands both volcanic and calcareous would have been houses. In some areas closer to the coast, large rocks fell abundant particularly around river mouths and coastal from the sky, killing people and animals (Blong 1982: Table margins. On the other hand, quality stone material for the 18). On Arop/Long itself, most if not all of the macrobiota production of flaked tools or ground axes would be found were eradicated; only those humans who fled to the main- further inland near the Finisterre Volcanics and the Gusap land survived (Thornton 2001). Whether this eruption had Argillite, or even further afield on the Rai Coast. significant impacts on trade networks and communications is not known. However, we do know that it was a This chapter has described the transformative physical central event in people’s recent past and it appears as an landscape around Madang, with a focus on land systems important story around New Guinea, including Madang. and geological zones, which will be important for teasing apart the procurement processes of Bel potters and the As uplift around the Madang coast is usually coupled with distribution of trade ceramics discussed in following chapoffshore subsidence, seismic and volcanic activity can also ters. This chapter also emphasised the dramatic tectonic result in significant tsunami hazards. Historical documen- activity that has typified the coast, having real and lasttation for tsunami events in the Bismarck Sea is abundant ing effects on Bel culture history that will be revisited in (Morgan et al. 2005). For instance, in AD 1888 a tidal wave Chapter 11. The next chapter focuses on the dynamic social originating from Ritter Island resulted in up to 3000 processes which occurred in Madang’s pre-colonial past. deaths (Paris et al. 2014). The Ritter tsunami was caused by the lateral collapse of the island, which sent deadly 8 m high waves onto the mainland (Ward & Day 2003). Other significant tsunamis occurred in AD 1857, originating near Umboi Island, and in the mid seventeenth century AD, associated with the Arop/Long eruption. The recent ~550 BP and ~3000 BP uplift around Madang are also associated with significant tsunami events and waves may have broken at over 2 m in height around the lagoon’s barrier reef (Mennis & Day in prep.; Morgan et al. 2005). 19 Chapter 3. Bel Production and Exchange Mager 1952; Mennis 1977). Multiple clans often cluster within the physical and census boundaries of a single modern village or island, but in the pre-colonial period the term ‘village’ may suggest a degree of centralisation that did not exist (Gosden 1989: 52; Hogbin and WedgeProduction and exchange are central cultural processes wood 1953: 253).2 Land rights to specific places are usually operating within Bel social groups and between the Bel regulated by genealogical relationships. Belonging to a and their trade friends. This chapter provides a new syn- specific clan entails rights to utilise certain land for huntthesis about how these processes played out in the recent ing, fishing, collecting, production, and social gatherings pre-colonial past, surveying available ethnographic, eth- for exchange. Although every clan has a great-man who nohistorical, and oral history evidence regarding Bel pro- is looked to as a leader, a council of senior males decides duction and exchange as it occurred from just prior to land matters (Morauta 1972: 12). European contact through to WWII, when coastal trade networks were severely disrupted by the conflict. Although In the time before Europeans, each clan presided over their these processes have diverged and co-modified over the own sleeping quarters, often housing extended families past c. 150 years of European contact, colonialism, and (Fig. 3.1–3.2), along with the men’s house (Gedaged = post-independence change, examining their nature at the dazem, Bilbil = darem). In many clan areas, houses sat on end of the pre-colonial period allows us to develop a valu- the ground, but on Bilbil Island, owing to its stony ground, able analogue for earlier times, and track some of these households sat on stilts made from tree stumps. These processes back into the archaeological record. stumps were brought over from the mainland and lashed together with vine (Mennis 1981a: 38, 62). The darem was a large, ornately decorated structure (Fig. 3.3). Otto Finch People of the coast described the building as having a roof extending all the way to the ground and a partially open front supported Settlement by a 6 m high, carved pole called the aimaka. Within the In Madang District, there are those who identify as be- darem were musical instruments, shields, pig jaws, trade longing to the coast and emphasise the seascape and long- valuables, and places for men to sleep (Finsch 1888). distance trading voyages in their stories (Suwa 2005). They refer to themselves collectively as the Bel people and speak Today, coastal Bel settlements are primarily established Austronesian languages. This marks a deliberate distinc- along the shoreline or on nearby islands. There are several tion from the ‘other’; the autochthonous ‘bush people’ who important areas of habitation that act as nodes in a netlive in the Madang hinterlands between the sea and the work of interrelationships along the coastline: Sek, Riwo, Adelbert Ranges, and who assert ‘we were here first, they Malmal, Siar, Bilia, Kranket, Yabob, and Bilbil (see Chapter came later.’ The Bel people, however, share a collective 2, Fig. 2.3). These places identify as ‘Bel’ on a broad level origin linguistically and culturally. These connections are but have developed unique identities of their own, with consciously reaffirmed through the knowledge of a com- slight variations in myth, language, oral traditions, and mon tongue, intermarriage, shared rituals, and the production and exchange of material culture. 2 The impression from the oral historical accounts collected by There were many things that the Bilbils could buy with their pots. The Bilbils worked to get all these things. — Pall Tagari of Bilbil (1978)1 Amongst the Bel, settlements are organised by named groups of patrilineal decent (hereafter called clans), known as temaleng in Gedaged or tamalen in Bilbil (cf. 1 Quoted in Mennis (1981b: 54). 20 Mennis (1980b, 1981a, 1981b) is that, prior to the colonial period, the Bel clans did not group themselves as villages but as a collection of clans, differing in geographic proximity to one another. The village names may have been applied by Europeans and have been retroactively imposed on the past by the informants.       · .  Figure 3.1. Stilted houses on Bilbil Island, illustrated by Nikolai Miklouho-Maclay 1872 (Miklouho-Maclay 1982: 253). Figure 3.2. A stilted house used as a school on Kranket Island. Photo: Lajos Biro (scan courtesy of M. Mennis). 21 Chapter . Bel Production and Exchange living around the coast or inland. Other groups such as the Bilbil and Yabob represent groups of clans all claiming to have originated on Yomba Island. Aspects of Bel production It is in the contexts of this physical and social organisation that the processes of pre-colonial production and exchange operated. In the past, many Bel clans could not meet their subsistence needs through tending their own land. Therefore, the production of pottery was both an economic necessity and a desirable means to maintain social bonds and assert primacy along the coast. This was completed alongside the local manufacture of canoes, which were essential to distribute pots and return home with trade goods. Food Bel subsistence in the past was bounded by two central concerns: the planting and harvesting of several key crop species, and the supplementation of the former by regular and strategically timed trading expeditions to acquire food. Tuberous, starchy plants dominated the diet (Table 3.1), but local production was not sufficient to match the Bel population’s carrying capacity. Women would tend to gardens of yam and taro, often spending all day there. Some islands such as Siar and Kranket had their own gardens, while others like Bilbil had plots on the mainland (Mennis 1981a: 72). On offshore islands, canarium and coconut trees were grown sparsely, supplementing the highFigure 3.3. A darem (spirit house) on Bilbil Island (Finsch in-carbohydrate root crops (Mennis 1981a: 38). These prac1888: 75). tices of gardening and agroforestry helped to structure Bel temporality as important seasons were defined by the action of planting or harvest. The yam season in June–Aueconomic specialisations. Many modern villages, such as gust coincided with the rising of the constellation Pleiades, Riwo, represent an amalgamation of clans with different marking the New Year (Mennis 1981a: 72, 2006a: 17). At this origin stories; some citing Yomba Island as their homeland, time, people came out at night to watch the stars, the posiothers claiming to have descended from groups previously tions of which indicated that it was time to uproot. In cel- Table 3.1. Important subsistence crops on the northeast coast. Bilbil name* Gedaged name** English name Taxon Procurement sources mangi magi Taro Colocasia esculenta/ Alocasia sp. Mainland Madang Karkar Island Rai Coast dabel dabel Yam Dioscorea alata Mainland Madang Karkar Island Rai Coast rengrangg zegazag Sago Metroxylon sagu Mainland Madang Gogol River valley – – Banana Musa sp. Mainland Madang – – Canarium Canarium sp. Mainland Madang Offshore islands Karkar Island Rai Coast niu niu Coconut Cocos nucifera Mainland Madang Offshore islands ge ge Pandan Pandanus sp. Mainland Madang * Mennis (1977); ** Mager (1952) 22       · .  and the food shared out evenly to each family. During a feast the women would serve the men and take a small share back to the house for themselves and the children. If a meal were to take place in the darem, women would bring the food to the entrance, and the men would take The Bel also kept small numbers of domesticates for con- it inside. Once inside, the food had to stay there; it was sumption: pigs, dogs, and chickens (Table 3.2; Mennis forbidden for scraps to be taken back to residences (Men1981a: 72). Pigs were a status food; the central component nis 1981b: 33). As eating utensils, shell and cassowary bone of feasts, bride prices, and prestige hunting. On Bilbil Is- were fashioned into spoons and spatulas, while coconut land, stone walls were constructed as pig fences in the Gad husks were used as bowls by children and old people clan area (Mennis 1981a: 67). Dogs and chickens were also (Mennis 1981b: 34; Fig. 3.6). kept as pets and for food. Smaller animals were hunted and collected. Flying foxes were killed with bows and ar- Pottery rows (Mennis 1980b: 72). Birds were hunted by constructing small hides called saleb in trees, from where people Pottery production, known as vai in Bilbil, was practiced would then wait to ambush the perching animals by hand amongst several Bel clans – those on Yabob and Bilbil Isor using slings (Mennis 1980b: 82). Reef fish were caught lands (Fig. 3.7–3.8) – in part because local gardening was with baskets or with nets called malala or raj (Fig. 3.5), not sufficient to meet subsistence needs. The pots prowhile larger pelagic fish such as tuna and sharks were tar- duced could be used for local consumption or regional geted with spears and arrows (Mennis 1980b: 85, 1981a: 72). exchange. These ceramics were produced using the padWomen and children would collect shellfish, which were dle and anvil technique: one of the two major producabundant around the mangrove swamps of the coast and tion methods used in New Guinea, alongside the coiling around the sandy intertidal zone of offshore islands (Men- method (May & Tuckson 2000). Among modern potting nis 1980b: 55). groups, paddle and anvil production tends to be restricted to the coast, is often associated with manual tempering, Eating was communal but strongly gendered. Plates and and has strong correlations with Austronesian languages, pots would be lined up on a palm or banana leaf mat such as Bel (May & Tuckson 2000). ebration, people would wash their hands and bathe in the sea (Mennis 1981b: 6). The yams would be prepared and stored in a dedicated building, the bad, creating a supply that lasted the community several months (Fig. 3.4). Figure 3.4. Women at Bilbil village preparing newly harvested yams for storage (2014). 23 Chapter . Bel Production and Exchange Table 3.2. Important subsistence animal species on the northeast coast. Bilbil name* Gedaged name** English name Taxon Procurement sources bor boz Pig Sus scrofa Mainland Madang Offshore islands Karkar Island Rai Coast – – Chicken Gallus gallus Mainland Madang Offshore islands guan guan Dog Canis familiaris Mainland Madang Offshore islands deb deb Fish various Madang coast Open sea rot/wonggazau zot/wonggazau Shellfish various Madang coast kodor kodoz Possum Phalanger sp. Mainland Madang Karkar Island Rai Coast melabo melabo Flying fox unknown Mainland Madang Offshore islands Karkar Island Rai Coast kuje/buntannes kuje/buntannes Birds various Mainland Madang Offshore islands Karkar Island Rai Coast * Mennis (1977); ** Mager (1952) Figure 3.5. Fishing nets, called raj in Bilbil. Bongu village, Astrolabe Bay (Biro 1899). Miklouho-Maclay, first observing pottery making on Bilbil Island in March 1872, saw that every family group engaged with manufacture. Under the roof of every house stood rows of finished and partly finished pots, all approximately the same form but varying in size (Sentinella 1975: 131). The Russian was visiting during a time of mass pottery pro24 duction, with many pots being made in anticipation of the times of hunger and extensive trading trips along the Rai Coast in May and June (Mennis 2006a: 18). MiklouhoMaclay recorded the production sequence of these pots as follows (Nikolai Miklouho-Maclay, 1872, in Sentinella 1975: 131):       · .  The implements used for the manufacture of pots are limited to two or three small boards and a pair of round stones, somewhat flattened on both sides. At first, with the aid of the small piece of flat wood, the upper rim is made from clay, which is then left to dry in the sun. When it had hardened somewhat, the rest of the sides of the pot are added bit by bit and smoothed out. The correct shape is given to the pot by holding it on the knees, the woman inserting her left hand with a round or flat stone in the pot, holding it against the internal surface of the wall and with the right hand striking on the corresponding place on the exterior with a flat piece of wood, evening out at the same time the surface and the thickness of the pot. When the pot is ready, it is first dried in the sun and then baked on a layer of brushwood covered with leaves and then sticks, etc. After stacking the pots in several rows one on top of the other, and covering the whole pile with light brush, it is set alight. pots, which could then be exchanged. On Bilbil and Yabob islands, pottery continued to be produced, as clay was accessible from the adjacent mainland. However, in several traditions, and all of those recounted to me, the Bilbil and Yabob assert that people were living on their islands prior to the arrival of pottery production. There is consensus among the potters that the culture hero Honpain arrived at Yabob Island making the first pots and teaching the Yabob women the skill (Mennis 1981a: 6, 1981b: 65). Honpain was a light-skinned migrant from the stars who eventually climbed back up to the heavens with her son, following a disagreement. The place of Honpain’s first manufacture in the centre of Yabob Island is still remembered. According to oral tradition, pottery production was later brought to Bilbil, and then Mindiri, from Yabob. As Mennis (2006b) notes, this does not preclude the possibility that Yomba Islanders migrated to Bilbil and Yabob with pot-making, prior to its destruction. Amongst the Bel, pottery making was, and continues to be, a protected practice over which the Yabob and Bilbil communities hold domain. Production is an exclusively Bel oral traditions are almost unanimous that the art female occupation and although women from various of pot making was imported from the survivors of the clans along the archipelago frequently marry into other Yomba disaster (Mennis 2006a: 30; see Chapter 1). In some clans, only those living at Yabob or Bilbil may learn to pot. places, such as Kranket Island, the practice disappeared For instance, if a Yabob or Bilbil potter married into anwithin one generation of the Bel’s arrival as there were no other village, she would be prohibited from continuing to appropriate clay sources on the island, and those on the make pots (Kakubayashi 1978). Most Bilbil women can mainland could not be accessed. At Kranket, which later pot and girls start learning at about four to six years old if specialised in building canoe hulls instead of pots, there born into the village. Women from other areas who marry are stories of potters who married into the island, spend- into Bilbil can learn to pot if they choose. Conventionally ing short periods on Yabob or Bilbil in order to produce these women derive from the Rai Coast, Karkar, or other Bel groups, but, more recently, Highland women are also marrying into the village and are training as potters. The potters assert that potting is not a difficult skill to learn if one has teachers to observe. This guarded practice acts as a means to regulate the relations amongst the Bel, limiting the number of production centres and thereby producing set rules for the roles that each group maintains in order to perpetuate exchange in the Madang area. This pattern was established prior to European contact. For the Bilbil and Yabob, the control of specialised pot production was essential to maintaining trade and exchange partners, thus ensuring a steady supply of food. As both Yabob and Bilbil Islands had limited arable land, the inhabitants supposedly specialised in making pots, which could then be exchanged for subsistence crops, among other things. This is paralleled in other areas with limited access to productive land (Curtis 1962; Groves 1960; Kramer 1985). The manufacture of pots reinforces a gendered division of labour: women collect clay and temper, produce pots, and tend to gardens. Men produce canoes, sell the pots and fish, and, in the past, went on trading voyages. Scheduling commitments mean that, today, potters might only produce pots one day per week, in which time a skilled potter could make five large pots. On other days (except Sundays), women will tend to their gardens, cook meals, Figure 3.7. Bilbil Island potters observed by Otto Finsch (1888). 25 Chapter . Bel Production and Exchange and look after their children, or more recently, work in Madang town. without much energy to people on Karkar Island, in Astrolabe Bay, and along the Rai Coast. Today, small canoes are used for fishing and short journeys, but motorboats have all but replaced canoes for long distance travel. DeCanoes spite this shift, the Bilbil own and maintain one ‘traditional’ Canoe production was just as important as pottery mak- canoe that is used for welcoming incoming cruise ships. ing to the Bel, as it enabled them to extend their mobility hundreds of kilometres to the north, south, and east. In In the recent past, the Bel produced two types of large this way pots could be transported relatively quickly and trading canoe for long distance travel: a double-masted D703 D705 E67039 FE564 46100 E64578 Figure 3.8. A sample of Bel pottery: D703, a small bodi and D705, a large you-bodi, collected by Miklouho-Maclay at Bilbil Island, 1870s (Macleay Museum); FE564, a large bodi cooking pot dated before 1912 (Te Papa); E67039, a modern magob tourist pot, and E64578 a recent three spouted you-bodi collected by Margaret Tuckson 1975 (Australian Museum); 46100, a standard bodi pot, 1973 (Auckland Museum). 26       · .  Balangut (Fig. 3.9 a–b, d),3 and the slightly smaller but more common Lalong (Fig. 3.9 c), with one mast (Mennis 2011: 14). Mennis (1980a, 1982, 2011) has documented the complete production sequence for building the Bel Lalong. Bush material was collected with the help of the inland groups: vines for lashings were gathered and dried out, the bark of the dim tree, later soaked and scraped, was collected to make putty for caulking the hull (Mennis 1980b: 10), and large roots would be cut and fashioned into the ornate breakwater (Mennis 2011: 28). To produce the large hulls, the men would burn the base of tree to make it easy to cut and then use an axe. A group of men would pull the timber to the beach and float over to their island (Mennis 1980b: 24). Once on the island the tree would soak in water for a week and then the men would begin hollowing out the hull (Mennis 1980b: 25). Stone axes and planes would be used to smooth the side of the hulls and shells used to shine the hull (Mennis 1980b: 26). and Yabob to produce hulls in exchange for pots and other items. The Bilbil and Yabob would then build the structures on top of the hulls that housed the sailors and their produce during trading trips (Mennis 1980b: 24). Magic Magic was a central technological component in reproducing processes of production. All initiated Bel men would use some degree of magic in their everyday routine. The knowledge of how to produce magic was entangled with the technical knowledge of canoe building and pottery distribution. For instance, secret words would be recited to allow one to hollow a canoe correctly or to sail without harm (Hannemann 1944: 5). Certain Bel groups were said to possess the knowledge of specific magic; the Bilbil and Yabob could not hull canoes because they did not possess the magic with which to carry it out. Although not well documented in oral traditions and ethnographic acThe division of labour by gender and clan groupings ex- counts4 there are also indications that pottery production tended to canoe production. The men from the lagoon was bound up in magic. The red slipped vessels of the Bel groups – the Riwo, Kranket, Sek, and Siar – specialised in canoe hull production and would be asked by the Bilbil 4 Probably owing to a gender bias in the collection of this data. Few women have been questioned about the belief systems behind pottery, and, moreover, they may not be permitted to divulge such information. 3 The term Palangut has previously been used, but Mennis (2014) now suggests that Balangut is a more accurate spelling. a) b) c) d) Figure 3.9. Bel canoes from the Madang coast: a) Balangut canoe at Yabob in the 1930s (Mennis 2011); b) Balangut canoe in Astrolabe Bay, 1905 (Bethke 1905, in May & Tuckson 2000: 162); c) Lalong canoe in Astrolabe Bay in the 1880s (Finsch 1888: 84); d) Balangut canoe, constructed in 2013 for independence day celebration, ashore at Bilbil village (2014). 27 Chapter . Bel Production and Exchange were said to be like young men painted with red during their initiation (Mennis 2014: 141), while the anvil stones used in pot production had magical properties necessary for driving out spirits from the raw clay. The source of these stones on the Rai Coast is perhaps the home of positive ancestral spirits (Tuckson 1966). As such, magic may have been an integral tool in the pot production chain. to a journey overseas, while the dadeng is the act of exchanging goods (Mager 1952). At European contact, this maritime exchange system was one of the most extensive on the northeast coast and allowed the Bel to trade goods up to 160 km east and west (Bogrodi 1959: 270). Oral traditions suggest mobility was even more expansive in the precolonial past, with the Bilbil visiting Arop/Long Island and Siassi (Hanneman 1944: 10),5 while the Yabob got as far as Bel magic revolved around animism and ancestor worship. Singorokai (Mennis 1981a: 28). The Rai Coast was a major It invoked the strength of the tibud (nature spirits) and trade zone, which attracted the attentions of the Bel tradthe meziab (ancestral spirits) in everyday routine. Tibud ers, as well as the communities of Arop, Sio, and Siassi, and were spirits that inhabited the natural realm: every tree, the hinterland people inland from the Rai Coast (Fig. 3.10). each point of land, the sea, and animals. Developing a fa- In this way, the people of the narrow coastal strip were miliarity with the local landscape required one to learn key mediators of long-distance trade and redistribution the tibud for different objects and places, and to negotiate (Kakubayashi 1981). As noted earlier, the coast is divided these supernatural agents in daily life. The meziab were in two by its occupants: the westerly Nambawan Rai Coast spirits of the ancestors, compiled into a singular social from Mingming to Saidor, and the easterly Nambatu Rai unit with the authority to protect and punish. Different Coast from Saidor to Gali.6 The Bel visited the Nambawan meziab existed for each clan line. Totems sacred to clans Rai Coast on day trips, leaving Bilbil in the morning and were hung above the darem (as can be seen in Fig. 3.3) returning by nightfall, but longer trips right along the Rai as well as on each trading canoe (Mennis 1980b: 27). The Coast in May–July allowed the Bel to stop at multiple setmeziab would customarily reside in the darem, but they tlements. On other shorter trading trips, the Bel would would accompany their canoe during trading expeditions travel to Bongu and nearby settlements in Astrolabe Bay, to preside over their kin (Mager 1952). The darem seems or voyage north to Karkar Island via Sek Island and Matuto have allowed the council of initiated males to control ka on the coast. and reproduce the meziab belief system. Modes of exchange A secret language was used in every stage of canoe building: from sharpening axes, to the hollowing of logs, at- The nature of exchange in pre-colonial Melanesia is a detaching the storage compartment, and putting on the sail bated topic. Since the systemic and functionalist perspec(Mennis 1980b: 24). For instance, the Bel would pray to the tives of Malinowski (1922) and Mauss (1925) there has tibud inside a tree before cutting it down to use for canoe been a tendency to distinguish ceremonial gift exchange production. Aufinger (1942) described two types of secret from commodity exchange and barter. However, because language used by the Bel: one of metaphor and innuendo; there was no quantitative exchange rate in many areas of and another of magic words – some borrowed from other Melanesia, some researchers have viewed all forms of prelanguages, others invented – with which people would colonial exchange as forms of gift exchange (e.g. Strathern speak to exclude the uninitiated from conversation. The 1988, 1992) involving inalienable objects (see Weiner 1992). latter was originally created so that sailors could speak From this standpoint, the gift was a metaphorical extenfreely while on the canoe. Evil water tibud were familiar sion of the giver, which carried the identity of previous with Bel language and would listen in on conversations. owners with it, forming integral components in meaningThis could prove detrimental to the trading trip. making systems of social relations. Others, however, view In this context, pots and canoes must be seen as inter- 5 There is substantial variation in the oral accounts about how twined with the broader social processes of belief, ritual, far east the Bel sailed. Oral histories collected by Mary Menand magic. Without specific magic being enacted and nis from Madang state that the Bel traveled as far as Sio or specific ceremonies taking place, the canoe could not take even Siassi, but those collected by Kakubayashi (1981) from form or complete trading voyages (Mennis 2011: 11). the Rai Coast state the Bel only reached Malamalai. However, Aspects of Bel exchange The production processes described above are inexorably bound up with the distribution, redistribution, and exchange of goods around the coast. The following section describes the social and technological basis for Bel exchange systems in the recent colonial and pre-colonial past. Trade systems The Bel have two words to describe their trading system: the waing and the dadeng (Mennis 2014). The waing refers 28 this is reconcilable if Hannemann (1944: 10) is accurate, as he states that some particularly courageous Bel made Siassi trade friends at Sio, which allowed them to venture further east than would usually be safe territory. This was likely a rare but notable occurrence. There are similar rare accounts of the Siassi traveling to Madang (Aufinger 1950; Mennis 1981b: 25, 95) 6 The Rai Coast is sometimes described as stretching from Astrolabe Bay in the west to Kwama River near Sio in the east (cf. Harding 1967: 8; Lilley 1986: 14), but Kakubayashi (1981) raises the point that Harding has misinterpreted Miklouho-Maclay’s description of the village of Talyata, which was the eastern extent of the trade voyage conducted by Maclay and his Bilbil friend, Kain. 500 Material culture km Pots Karkar Island Wooden bowls Bows and arrows Sarang Mal (bark cloth) Bagbag Island Megiar Mugil Canoe hulls Drums Matuka Mortar and pestle Garu Raw materials Rempi Budup Sek Nobanob Gogol Amele Erima BISM A R CK SEA Riwo Siar Kranket MADANG Jomba Yabob Bilbil Black paint Bush materials Imports: pots, canoes armbands, headrests Tolokiwa Island Arop/Long Exports: pigs, dogs, pig tusks, dog teeth, drums, obsidian A S T R O L A B E B AY Bongu Red paint Crown Island Obsidian Food Island Sago Rimba Umboi Island Bibi Kulilau Mindiri Dein Singor Biliau Mengu Segi Lamtus Orinma Wood Taro & yam Yam Pigs Wab V IT IAZ S T R AIT Canarium nuts Other Tobacco Malamalai To Siassi Gali Imports: pots Roinji Singorokai Malasanga Kiori 0 50 km Buai Sio Gitua Exports: pig tusks, drums, plates, black paint, canoes, stone axes Settlements Long trade routes Short trade routes       · .  29 Figure 3.10. Coastal trade networks in northeast New Guinea (adapted from P. J. Edwards in Mennis 2006a: 54–55). 0 Chapter . Bel Production and Exchange a disjuncture between gift exchange, based on positive or neutral reciprocity, and commodity exchange based on the premise of negative reciprocity and keeping one’s trade partners unaware. Gell (1992b) asserts that barter existed in the pre-colonial past because exchange was often focused on acquiring highly valuable material possessions in exchange for less valuable ones. Gorlich (1998a, 1998b) takes this further and suggests that barter existed in the past but must be understood as dialectic between gift and commodity exchanges with social meaning created by uncertainty about the object’s availability and risk of non-transactions or personal harm. Modern Bel informants certainly stress that commodity exchange existed in the past, directly paralleling the money economy around Madang today. This is particularly apparent in Pall’s quotation at the beginning of this chapter. However, it would be difficult to determine if this is accurate, or if the modern situation has retroactively influenced perspectives on the past. Accounts from the Rai Coast also suggest that there was a simple but accepted exchange rate for different objects in different contexts (Kabuyashi 1981). kinship term to denote closeness. Hannemann (1944: 10) describes several other terms in the friendship system: a kadoi was a highly esteemed friend often within the same clan or language group, with whom pigs and other valuables would be exchanged; tizan were trade friends whose village lay across the sea and new tizan could be acquired; matau represents the system of mutual aid, which would sometimes appeal to large groups, much like the modern wantok system.8 7 Dinau is a Tok Pisin word for debt. This word seems to be used by informants in the oral histories to refer to an equivalent Bel system of expectation and obligation. 8 Wantok is a Tok Pisin term meaning ‘one talk’ and denotes a person with whom one has a strong social bond, often deriving from the same language group, clan, or family. New Guinea’s northeast coast, along which Bel traders voyaged, was a linguistically diverse place, today comprising about fifty distinct languages. The Bel traders could communicate with many of their trade friends and interaction between these groups was made possible because at least some individuals in each group were multilingual (see Hogbin 1947 for a similar example in the Huon Gulf). As on the Sepik north coast, boys were left for short periods of time in the host communities of their family’s trade friends to learn their language (Welsch et al. 1992). This was used to maintain trade relationships in other villages over multiple generations. Moreover, in order to learn the The Bel use several words for their different modes of ex- nature of trading from a young age, boys from different change, which seem to primarily describe gift exchange. Bel groups would go with their fathers alongside the Bilbil Ziang is a gift or a donation (Mager 1952: 299). This re- and Yabob for trade voyages (Mennis 1980b: 17). fers to an item given in generalised and positive reciprocity. Dadom are gifts without expectation of return, made The different Bel languages around Madang (Gedaged in reconciliation. A wou in Gedaged or vou in Bilbil is a and Bilbil) were mutually intelligible and the non-Bel lannon-physical gift, but which carries the rights to repro- guages on the mainland and around Astrolabe Bay were duce some physical aspect of culture (Mager 1952: 350). For spoken by many Bel (see Sentinella 1975). To the north on instance, the vou might include the right to create a trade Karkar and the adjacent coast, the Bel traded only with the item in a certain style, perform a dance or song, or plant Takia, not the non-Austronesian speaking Waskia, and so certain foods. Apart from the Bilbil and Yabob, Bel clans limited their interactions to other Bel speakers (Mennis are not permitted to produce pots because they have not 1980b: 76). There were some Bel words in common with received the vou. Suzuzum refers to a gift given to some- the Tuam languages of the Siassi Islands, which were also body holding a feast (Mager 1952: 298). This is a form of part of the Ngero/Vitiaz linkage, but the languages were delayed reciprocity and a return gift is later expected. A not mutually intelligible (Mennis 1981b: 97). wodeng in Gedaged or odeng in Bilbil is another form of return gift, which is given to a trade friend in exchange for Sailing something else (e.g. betelnut, tobacco) (Mager 1952: 348). The dinau7 was a system used by the Bel, whereby if a The Bel traders were adept sailors who had developed group did not have pots to give at the time of transaction, a distinct lore regarding various environmental factors they could owe a set number of pots (Mennis 1981b: 59). along the coast, which largely informed the timing and This was a form of delayed payment that would have ben- extent of their voyages. The Bel used the stars to navigate efited many of the Bel clans during times of hunger. Those during the night and noted the different points of land groups who did not believe in the dinau were not consid- and river mouths during the day (Mennis 1981a: 15). The ered good trade friends (Mennis 1981b: 59). This system winds, which were all named (Fig. 3.11), were a ‘consuming operated reciprocally between Bel groups and inland ‘bush interest’ of the Bel (Mennis 2006a: 67) and their strength, people.’ direction, and timing were well understood. The Yawarti wind blew down from Karkar Island while the Dadau blew from Amele, an inland village northwest of Madang. These Trade friends winds were active during the wet season in January and Although not necessarily related by blood or marriage, February and were used to take the Bel to Rimba, Kul, and trade friends along the coast were considered the same Singor. Dadau Dere was also active during January and as brothers (see Hogbin 1964: 49). This is reflected by the February but was prone to changing direction. The Dolo Gedaged term for trade friend, gai, which is a general Yawarti, blowing from Siador in May–July, was used by 30       · .  1981a: 41). Weather likon could use spells to turn the rain into sunshine or quieten a stormy sea. They used to predict the winds for the next day and sang out to the tibud, inhabiting each point of land along the coast to ensure safe passage. Alternatively, rain spells were recited after dry periods to encourage crop growth. In this instance, the likon would strip the bark of a thin tree, then take the sword shaped leaves of a particular shrub along with some tree branches, binding them together with the oily bark. He then put this in the nearest fresh flowing river Magic and rubbed it with stones, simultaneously reciting a spell Just as magic pervaded the everyday production of ob- inherited from his father (Aufinger 1939). jects, so too did it form an integral part of Bel exchange and was a key component of the waing. Within each Bel Ceremonial displays were also necessary for the trade group there were specialised practitioners of magic named voyage to commence. Upon departure, the ‘father’ of the likon. Likon could recite spells to heal sores and cure sick- canoe would swim out with the canoe and fight it with ness, improve gardens, and settle conflicts. They possessed a branch. The young men would stand on the platform, magic stones, described as rounded black rocks, to help decorated with feathers and mal (bark cloth loincloths), carry out their tasks (Mager 1952). These men were par- while the leader would stand in the cabin, drumming the ticularly necessary in controlling the weather during trad- vongu drum, speaking magic, and blowing on the conch ing trips: they had power to start, change, and calm the shell (Mennis 1980b: 44). winds (Hannemann 1939). It appears that the likon was an exclusively male role, and a hereditary skill passed down Pots, canoes and magic were bound together even during from father to son, whereby children would be selected voyaging. For instance, broken potsherds would always and trained in different lore (Mennis 2006a: 40). The be carried in a basket inside the canoe, and if the canoe head likon presided over subordinates who held differ- was in rough weather the men would take the sherds out ent elemental domains. These lesser likon have taken on a and heat them over a fire before throwing them into the mythical status, reportedly being strange little beings that sea to calm the waves and any malignant tibud (Mennis would hide away if anyone entered their house (Mennis 1980b: 44). the Bel to return from the Rai Coast. However, Karag, the brother of Dolo Yawarti, was regarded with apprehension by the sailors and was a dangerous wind on which to travel. There were particular dangers associated with sailing on the wrong winds at the wrong times, especially if they blew sailors into enemy territory, but such challenges were carefully managed, alongside concerns regarding subsistence needs, and the placement of trade friends. N Yawarti Dadau Der Dolo Yawarti Karag Yawan Figure 3.11. Wind chart showing major sailing winds used by Bel traders (adapted from Mennis 2006a: 66). 31 Chapter . Bel Production and Exchange mals, brought back to Madang strapped to logs (Mennis 1981b: 19). While on trading expeditions the young men Food was an essential goal of the dadeng, required to sup- would go hunting in the bush with their trade friends port the Bel community’s large populations. The acquisi- for possums and feral pigs (Mennis 1980b: 27, 1981b: 26). tion of food through trade was both informal and strate- Betelnut would be shared in small quantities or purchased gic; localised and long distance. Throughout the year, the in bulk on the return journey (Mennis 1981b: 55). Bel on offshore islands would travel to the Madang coast to exchange pots, smoked fish, and other desirable items Material culture for crops, pigs, and bush material with the mainlanders. The most frequently traded crop was taro, but sago, banana, Items of material culture were active in reproducing and pandan, and yams were also exchanged. The mainlanders shaping exchange, interaction and social meaning along used numerous other plants for medicine, decoration, poi- the coast. The Bel, and particularly the Bilbil and Yabob, son, and magic but it is unknown if the Bel engaged with were remarked on by early Europeans for their material the same plants for these purposes (Miklouho-Maclay wealth (Table 3.4), compared to nearby groups who did 1885; Petir et al. 1997). The bush people would light fires not participate in the dadeng (Finsch 1888, quoted in Menon the beach to signal that they wanted to trade, or market nis 1996: 15). days would be timed by the counting down of coconut fronds; each day a frond would be cut off until it was the Objects used in a dadeng were called jabaz in Gedaged or day of trade (Mennis 1981b: 79). jambar in Bilbil. Bulk articles were also commonly known as su, because they were supported by su (supports) durThe time of local planting embedded the ground with new ing transport. Trades of bulk goods could be enhanced growth but marked the beginning of the time of hunger by samum, rare things such as beads or grass skirts to (Table 3.3). Small amounts of sago, taro, and sugarcane make an exchange of food or pots more desirable (Meger might still be collected locally, and trade with the adja- 1952: 128). So, although the fundamental goal of the larger cent mainland would assist subsistence to a degree, but scale dadeng was to shift pots and food between groups for to fully support their large populations the offshore Bel, subsistence maintenance, smaller objects were essential to particularly the Bilbil and Yabob, had to turn elsewhere these movements. to acquire food (Mennis 1981b: 23). In these times, the Bel travelled along the Rai Coast to acquire pigs, smoked fish, Although the Bel were recognised as specialist potters red sugar cane, and root crops (Mennis 1981b: 24). They along the northeast coast, they also traded for other pots would also go north to Karkar to get pigs and canarium made by different production groups. The nearest major nuts. Pigs could either be purchased as meat or as live ani- production centres were located inland, in the Gogol River Food Table 3.3. Annual planting and trading cycle (adapted from Mennis 2006a: 68, 2014: 18). Month Season Planting Winds Trading June Dry Harvest yams Night: Dadau Day: Dolo Yawarti Rai Coast (long trading trip) July Dry Harvest yams Night: Dadau Day: Dolo Yawarti Rai Coast (long trading trip) August Dry Day: Karag Late afternoon: Yawarti Dangerous sailing Bogati, Saidor, Mindiri in afternoon September Dry Day: Karag Late afternoon: Yawarti Dangerous sailing Bogati, Saidor, Mindiri in afternoon October In between Doldrums Small scale trading November Wet Hungry times Day: Dadau or Yawarti Afternoon: Dadau Dere December Wet Hungry times Day: Dadau Afternoon: Yawan Not ideal for trading January Wet Harvest small taro Night: Yawan Day: Dadau and Yawarti Afternoon: Dadau Dere Small scale trading inland February Wet Plant yams Night: Yawan Day: Dadau and Yawarti Afternoon: Dadau Dere Rai Coast (short trading trip) March Wet Hungry times Early morning: Yawan Day: Dadau Dere Rai Coast (short trading trip) Doldrums Sek and Rimba Harvest large taro Night: Dadau Day: Dolo Yawarti Rai Coast (long trading trip) April In between May Dry 32       · .  Table 3.4. Trade items acquired by Bel according to oral testimonies and word lists. Bilbil name* Gedaged name** English name Manufacture location Place of acquisition tabar zagat bizag tabaz zagat bizag Arrows Inland villages Bonga Malamalai Sel vol wol Bows Inland villages Bonga Masamalai Sel – palapal – – sawar kabu palapal mameng kidiai sawaz Spears Karkar Karkar sapam sapam Round shields – – madin dimui Long shields Karkar Karkar – – Slings Inland villages – – – Bone knives – – unding uding Bone needles – – – – adiu makak liwon Stone axe/adzes – Rai Coast Karkar – – Shell axe/adzes – – pat – Flaked stone – Rai Coast jaling jaling Obsidian Bismarcks Arop/Long Island Umboi Island – sinai Coiled pottery Gogol Valley Inland villages bodi bod Paddle & anvil pottery Sio/Gitua Mindiri Sio/Gitua Rai Coast kunap sauang kunap suang Mortar & pestles Karkar Karkar – duaz (?) Round wooden bowls Rai hinterlands Numbawan Rai Coast – suaz (?) Long wooden bowls Rai hinterlands Numbatu Rai Coast sikor sikoz Turtleshell armbands – – – papeng Turtleshell nose ornaments – – – – Shell armbands – – paspas – Woven armbands Karkar Karkar ari azi Armlets – – betambet bedabed Dog teeth necklaces Arop/Long Arop/Long kolekole kalekale Dog teeth breast-pieces Arop/Long Arop/Long pezamat paramat Pig tusk necklaces Siassi Arop/Long Rai Coast Arop/Long kamatui gamun matui gamun Fish palatial bones – – bul bul Shell ornaments – – bililik bililik Hip ornaments – – semarek mal Loincloths Rai Coast Rai Coast – nai Grass skirts – – si si Combs – – bem bem Red ochre/paint Karkar hills Karkar mumu mum Black paint Rai Coast (Sel) Rai Coast – nanon Red dye Karkar Madang mainland Karkar Madang mainland nong dadir nong nainal nalog Black dye Siassi Rai hinterlands Rai Coast jang jang Yellow dye – – dim sol*** – Pig bone tools – – kau kau Lime – – – kau adug Lime containers – – 33 Chapter . Bel Production and Exchange Table 3.4 continued. Bilbil name* Gedaged name** English name Manufacture location Place of acquisition asi as Timber Inland villages Inland villages – – Bush material Inland villages Inland villages vongu dugwag Hourglass drums Karkar Arop/Long Karkar Arop/Long do do Slit gongs Rai Coast Rai Coast kongu kag Bottle gourds – – aul aul Fishooks – – raj aza kakang Fishnets – – * Mennis (1977); ** Mager (1952); *** carved pig bone used to insert putty into the holes in canoes. area with markets at Gonua, Boi and Atu. These pots were produced by male potters using coiling methods (see Fig. 3.12), and although many acknowledged them to be less durable than Bel pots, the coiled ones were considered better to cook sago in, as they gave the crop a sweeter taste (Mennis 1981b: 57). Along the Rai Coast at Mindiri another Bel group produced red-slipped pots broadly in the same style as the Bilbil and Yabob pots; however, a different, black clay source was used (Mennis 1981b: 25). There were also some technical differences; for instance, Bilbil pots heated food quicker than Mindiri pots (Mennis 1981b: 24). Further along the coast, at Sigawa Island (Sio), at Nambariwa on the adjacent coast, and at Gitua there was another distinct potting tradition (May & Tuckson 2000: 148). The Sio/Gitua potters continue to make pottery today and use paddles and anvils during the forming stages. The pots are morphologically very similar to Bel pots; they are globular with decoration restricted to above the shoulder (Lilley 1986: 194). However, technologically the Sio/Gitua pots are distinct in that they do not contain temper (May & Tuckson 2000: 150). Stone axes were traded in primarily from the Rai Coast and less commonly Karkar Island (Mennis 1981a: 58). It appears that axes were not produced by the Bel around Madang owing to a lack of suitable raw material and all were imported (similar to Fig. 3.13). Stone axes collected from Astrolabe Bay have previously been described as diorite-porphyry (Kunst & Buchdruckerei 1886: 9), which suggests the raw material was procured from the hills behind the Rai Coast. However, further north, near Aitape, stone axes were differentiated by their users based on the stone’s sharpness and the positioning of the haft rather than the type of stone or the shape (Biro 1899: 181). the Bismarcks (Gaffney and Summerhayes 2019), was not revealed to them. Shell tools were also used in everyday routine and traded in from afar. Miklouho-Maclay described the use of simple unmodified shell (Cardium sp.) scrapers called arur with serrated edges used for scraping coconut (Sentinella 1975: 86). Small shell adzes made from Tridacna sp. may also have been used as taro peelers (Egloff 1975: 34). Bone tools were common around the coast. MiklouhoMaclay describes dongan which were pig-bone knifes, along with a variety of other sharpened bones (especially cassowary) used as knifes, levers, lancets, or awls (Sentinella 1975). Bone needles called uding were also made from thin flying fox bones. Dim sol were carved pig bones with the specific function for inserting putty to caulk holes in canoes. This simple tool was integral to canoe building and maintenance. The Bel exchanged pots for a variety of weapons, and they were not hesitant to use them for retribution or to maintain dominance over pot production and trade voyaging. The mainlanders around Madang in particular, and those on Karkar Island and the Rai Coast, were major suppliers of various spears, bows and arrows, and large battle shields, which were round or oblong, with raised carvings and coloured with paint (Fig. 3.14). Many other wooden items figured substantially in exchange around the coast. Hardwood items such as pestles and mortars were imported from Karkar Island (Fig. 3.15), while wooden bowls were procured from the Rai Coast. These bowls were produced on Tami Island, the Siassi Islands, and in the Rai hinterlands, behind the coastal Today’s older generations around Madang never used villages (Harding 1967: 38; Schmitz 1955: 306). Round flaked stone tools, but their origins and functions have wooden bowls were characteristically obtained from the been passed down as oral traditions. According to my in- Nambawan Rai Coast, while elongate wooden bowls were terlocutors around Madang, flakeable sedimentary stone typically procured from the Nambatu Rai Coast or Siassi was imported primarily from the Rai Coast. Obsidian was (Fig 3.16) (Christensen 1975; Kakubayashi 1981). also imported and the Bel today cite Arop/Long Island as the origin. Mager’s (1952: 123) dictionary specifies Rook Additionally, Hagen (1899: 257) suggested on the basis of Island (Umboi) as the source, which suggests that the Bel stylistic similarities between wooden totems, known as knew that obsidian derived from the east, towards the Vi- silum in Gedaged or sulum in Bilbil, that there was a comtiaz Strait, but its ultimate source, now known to be in mon production centre at Bongu in Astrolabe Bay (Fig. 34       · .  CG1281 46143 46145 46144 CG1282 46111 0 10 cm Figure 3.12. A sample of inland Madang pottery: 46145- large cooking pot purchased at Guman village by S. Bulmer, 1967; 46144- large cooking pot made at Guman; 46143- cooking pot, unknown provenience; 46111- large cooking pot purchased by G. Johnston at Begasain (all Auckland Museum); CG1281- small cooking pot from Gonua village; CG1282- large cooking pot, unknown provenience (both Te Papa museum) Figure 3.13. Stone axe from the Huon, collected by Richard Neuhauss, #VI 29995 (Kulturbesitz Ethnologisches Museum, Berlin). 35 Chapter . Bel Production and Exchange 3.17).9 However, owing to small variations in form and the fact that each clan or family group had specific names and meaning attached to the objects, it is more likely that each group produced their own (Bodrogi 1959: 54). Carved animals, known as mazoz in Gedaged and maror in Bilbil, were used as emblems by each clan and attached above the darem and on canoe masts. These were certainly made on Kranket Island and in the lagoon area, but it is not known if they were then exchanged with the Bilbil speakers to the south, or if they manufactured their own. Much of the Bel’s clothing and ornamentation was also imported (Mennis 1977), and such adornments readily distinguished them from other groups along the northeast coast (Sentinella 1975: 36). Miklouho-Maclay described his first encounters with the Bel while in residence near Garagassi in Astrolabe Bay. Their hair was carefully coloured by red ochre with foreheads, noses, and backs painted in the same colour (Sentinella 1975: 40). Loincloths known as mal were worn around the waist, obtained from the Rai Coast. Armbands made from turtle shell (Fig. 3.18), marine shell, and plant products were worn, especially during dances and initiations (Mennis 1980b: 30). According to modern Bel groups, shell ornaments were not locally produced but were trade items from Siassi, Arop/Long Is9 Silum are described as telum by Miklouho-Maclay who uses the Bongu term, but ‘s’ and ‘t’ may be rather interchangeable with regard to their Russian and English transliteration (Sentinella 1975: 349). Figure 3.14. Jangar, an elder of Bilbil Island with shield from Karkar Island. Lajos Biro (scan courtesy of M. Mennis). 00 10 10 cm cm Figure 3.15. A pestle and mortar for crushing sago, Karkar Island, German New Guinea (D25.2432, Otago Museum). 36       · .  00 10 10 cm cm Figure 3.16. Two wooden bowls from Astrolabe Bay. A round bowl (D24.3049) typical of the Numbawan Rai Coast and an elongate bowl (D24.3048) typical of Siassi. (Otago Museum). land, and the Huon. Eligible men and women wore small woven armlets produced by cooking and dying the bark of a climbing palm (Calamus sp.) (Mager 1952: 9). Necklaces, headbands, and hip decorations were produced from shell, fishbone, pig and dog teeth (Fig. 3.22; Sentinella 1975: 81). Pig tusks, known as pezamat in Gedaged or paramat in Bilbil, were bound in pairs as men’s festive ornaments. The most impressive and valuable of these pig-tusk decorations, in which each tusk came together in a perfect circle, could be exchanged for a whole pig and were used as brideprices (Mager 1952: 247). In early encounters, MiklouhoMaclay attempted to trade a steel knife for a paramat, but the Bilbil men did not wish to part with it, despite desiring the knife very much (Sentinella 1975: 41). Guinea, was sudden and astonishing (Burton 1999: 14). On the 20th September 1871, Nikolai Miklouho-Maclay, the Russian scientist-explorer, landed at Astrolabe Bay with his two servants (Fig. 3.20). The first encounters with local people that he describes conjure up fear, fascination, intimidation, and excitement (Sentinella 1975: 15). His encounters were generally made favourable by tapping into the existing gift exchange systems that operated amongst the Bel and nearby groups. The Bel were particularly fond of his gifts of metal nails. Later, he would gain favour by healing many of their sores while they were visiting for a singsing10 (Sentinella 1975: 82). The Russian therefore figures substantially in many Bel oral traditions and legends. However, ten years later in 1881, Otto Finsch, an ornithologist, laid the foundations for the Neu Guinea Kompagnie to establish plantations around Madang (Fig. 3.21), and by 1884 the German standard flew above a new colony (Mennis 2006a: 123). Finschafen, on the Huon Peninsula, was the Neu Guinea Kompagnie’s initial base of opera- Negotiating colonial disruptions The previous sections have attempted to approximate Bel production and exchange just prior to the colonial period. However, much of what exists today is also bound up with significant events and alterations that took place in the late 19th and early 20th centuries. Contact with Europe- 10 A singsing (Tok Pisin) is a gathering or festival involving ans around the northeast coast, as in most places in New dance and music. 37 Chapter . Bel Production and Exchange tions in the late 1800s, but this was moved to FriedrichWilhelmshafen (now Madang) in 1892 (Molnar-Bagley 1993: 18). Although many Europeans subsequently set out for Madang under the auspices of science and ethnology, communication difficulties between the Germans and the Bel, alongside German disregard for local customs and land ownership, led to a number of issues. For instance, Lutheran missionaries revealed secrets to the uninitiated and burned objects such as the meziab flutes or sacred stones, in an attempt to convert the people to Christianity, forcibly imposing the Western dogma and cosmology on the indigenous belief systems (Lawrence 1956; Mennis 1980b: 45). The magic spells from Yabob recorded by Aufinger (1939) only a few decades after significant European contact were already being forgotten, and with them the technical skills to produce canoes and control the weather to undertake safe trading expeditions. On two occasions the Bel were forcibly moved from their clan areas by the Germans, which caused a severe disjuncture and inhibited customs of potting and canoe making (Mennis 1980b: 68). This first occurred in 1904, after a local revolt. The leaders of the revolt were blindfolded and shot to demonstrate German superiority (Mennis 1980b: 31). This was despite intra and inter-communal violence being outlawed under the German administration, partially as a Christianising effort, and partly to offer a measure of personal safety for the missionaries. The Siars were then sent to Mindiri, the Yabob to Yeimas, the Biliau to Suit, and the Kranket to Wab (Mennis 1981a: 6–9). The Bilbil were not banished, but their elders were sent to Baining in east New Britain (Mennis 1981a: 9). Mennis (2014: 67) suggests, quite plausibly, that the revolt was instigated by the Bilbil when the Germans had implemented a new monetary system, challenging the Bilbil’s trade monopoly. Figure 3.17. Wooden ancestor figure, Bilbil Island, collected by Otto Finsch, #VI9336 (Kulturbesitz Ethnologisches Museum, Berlin). 00 55 cm cm O50.174 O50.174 O50.173 O50.173 O50.172 O50.172 Figure 3.18. Turtle shell armbands from Astrolabe Bay (Oldman Collection, Otago Museum). 38       · .  Figure 3.19. Trade valuables from Arop/Long Island: a) pig tusk necklace; b) cowry shell armband; c) dog teeth necklace (photos: M. Mennis 2014). Figure 3.20. Nikolai Miklouho-Maclay meets local people in Astrolabe Bay (Source: Russian Academy). 39 Chapter . Bel Production and Exchange Figure 3.21. The people of Riwo Island (also called Ruo) with Paul Boether (left) and Bergmann (right) during German administration. Photo: Lajos Biro (scan courtesy of M. Mennis). The Bilbil asserted on the basis of oral traditions that the pots were their money (and livelihood) and any attempt to undermine this understanding faced retribution. Many canoes rotted during their elder’s absence in Baining and on their return the Bilia, Kranket and Siar made only small canoes rather than large trading ships. The second relocation, onto the Rai Coast, occurred some years later when rumours of another revolt began to circle. After returning from the Rai Coast, the Bilbil relocated permanently to the adjacent mainland, using the island as a place to collect canarium nuts and coconuts, or to rest during fishing excursions (Mennis 1981a: 39). The Germans later established Malay and Chinese communities around Madang and sometimes Chinese junks would accompany Balangut on trade voyages. The Yabob even purchased a Chinese junk, as they were more durable than Bel canoes, and the Bilbil ended up renting three of these craft. This relegated the lagoon Bel to minor partners in trade expeditions as they were no longer required to produce canoe hulls (Mennis 1981a: 13, 1981b: 11). The Chinese were also employed to deforest the mainland around Madang (Mennis 1980b: 79–80), which exacerbated this situation as suitable trees for canoe hulls became scarce. A shift to selling pots for money that was used to buy crops 40 and even boats further reconfigured the local gift (and possibly barter) economy. Colonial introductions of new foods from Europe and Asia also saw a dramatic shift in the Bel diet over the last 130 years. The Germans brought pineapple, pawpaw, watermelons, and corn seeds, and the locals were instructed to plant and manage these new crops (Mennis 1981b: 1). Rice from Asia in particular has become a major carbohydrate contributor in Madang town. In many areas outside Madang town, however, pre-colonial crops such as yam continue to structure the local yearly cycle, and, along with sago and taro, contribute significantly to the diet. Finally, global conflicts during the early 20th century shifted the balance of power around the northeast coast of New Guinea. During the First World War, Australia came to occupy the former German colony and it was later administered as a League of Nations mandate territory by Australia (Sinclair 2005). During the Second World War when Japanese forces occupied the area, the Bel groups could not complete trading trips along the Rai Coast as it had become too dangerous (Mennis 1980b: 3). In particular, in 1942 the Japanese destroyed a number of trading canoes for firewood and used paddles for carrying cargo       · .  (Mennis 1981b: 54). Moreover, many coastal groups were encouraged by Australian propaganda airdrops to relocate inland in order to avoid bombings, while many of those living on offshore islands were moved to the coast by the Japanese (Mennis 1980b: 3, 6). In this way the geographical pockets which the Bel inhabit today do not accurately reflect pre-war distributions. Summary This chapter has surveyed existing information regarding pre-colonial Bel production and exchange, drawing on oral traditions, ethnography, early explorers’ accounts, and museum collections. The emphasis has been on thoroughly describing the production and exchange of material culture during the early contact and colonial periods, but with the finding that: 1) the nature of material production and exchange is entangled with subsistence strategies, settlement patterns, sailing technology, magic, and the physical environment, and 2) a series of European administrations have altered the trajectories of production and exchange, and particularly how these processes are remembered in oral history, recorded in ethnography, and collected for museums. The next chapter will focus its analytical lens onto the pottery production process itself, the key component in Bel production and exchange, and the subject of the present archaeological investigation. 41 Chapter 4. Modern Potting Communities edge of how to pot has been lost. Despite this, the sacred area where the culture hero Honpain made the first pots is remembered. At this place she taught the Yabob people to pot, and from there the potters then took the skill on to Bilbil Island. It is easy to learn to make pots, if you have the right teacher. — Sentie Noah of Bilbil (2015) Modern pottery production and exchange remains an important activity among the Bel groups, particularly at Yabob and Bilbil where manufacture still takes place. This chapter is an ethnographic study of this pottery making, known locally as vai. The vai was documented with interpretations of the production sequence informed by interviews with the Madang potters and elders living in these villages. The chaîne opératoire, following resource procurement, production, use, reuse, and discard of contemporary Bel ceramics is presented in detail, using ethnographic observations made in 2014 and 2015 while visiting Yabob and Bilbil villages. This provides an essential modern analogue for the archaeological pottery of Madang (see Fewster 2006; Roux 2007). Later in the book, the chaînes opératoires of archaeological pottery will be compared with these modern iterations to identify technological changes in production and exchange from the pre-colonial to the modern. Production continued into the 21st century only at Yabobup-top (Fig. 4.1). At the time of fieldwork only three older women at Yabob-up-top village knew how to produce Madang style pots using the paddle and anvil method: Yuk, Yamoi, and Yeyeg and only one of these potters, Yeyeg of the Nob clan, was still practicing, albeit very seldom. Some younger women previously made ceramics but their potter’s wheel had rusted and they were unable to make pots by hand moulding. These potters and their families were enthusiastic for researchers and tourists to record and spread word of their pottery, which they lamented to be a dying art as the young Yabob girls were uninterested in learning the art from their grandmothers. Bilbil pottery The Bilbil people live in a large village composed of four clans, which stretches along the coast several kilometres south of Madang. At European contact the Bilbil were living and producing pots offshore, but as punishment for a planned revolt against the Germans in 1904, people were displaced from Bilbil Island to the mainland (Mennis 1981b: 4). Many were shot and houses were burned. Later, under Australian administration following WWI, some Bilbil people returned to their island home to continue potting, but this did not last. Today, Bilbil Island is uninhabited, but each clan retains claim to their ancestral space there. Pottery is now produced at Bilbil village on the mainland where production is thriving (Fig. 4.2). Many women in the village produce pots on a regular basis. There is a sense of economic and social empowerment among the Bilbil potters, in contrast to the disempowerment felt at Yabob at the loss of their customary practice. The potters at Bilbil actively engage in the same practice as their ancestors, and as with other aspects of their material culture (see Chapter 3), pot production is important to represent their past in a highly visual manner. Production groups Yabob pottery The homeland of the Yabob people is Yabob Island, but there are now Yabob communities not only on the island itself but also on Yomba Island,1 and at two locations on the mainland: Yabob-up-top, located on a hill that overlooks the sea, and Yabob-down-below (also known as Morelan Hamlet), situated next to a beach looking out to Yomba Island. The Yabob Islanders, who were making pottery at European contact, were forced to abandon their homes during Japanese occupation in World War II and resettlement of the island did not begin until after Papua New Guinea’s independence in 1975 (Suwa 2005). On Yabob Island the importance of potting to their ancestors is recollected through oral traditions, although the knowl1 Not to be confused with Yomba Island, the legendary homeland of the Bel people (see Chapter 1). 42       · .  Figure 4.1. Yabob-up-top potting village (photo: M. Mennis, 2014). Figure 4.2. Bilbil potting village. A view from offshore (photo: M. Mennis, 2014). style pottery manufacture (May & Tuckson 2000: 163–164), but according to the Bilbil villagers who travel to that area regularly, this village is no longer producing. This means that Bilbil village is now the only extant production centre of Madang style ceramics, and during the course of the 19th–21st centuries, the number of these centres has reduced from perhaps five (or more) to just one. Defunct production centres Finsch (1888) noted at the end of the nineteenth century that Yomba and Bilia Island women also made pots, but if this is accurate, production has since ceased. Mindiri village (formerly Medize or Mediseh) on the Rai Coast has been recorded as the most distant place of Madang 43 Chapter . Modern Potting Communities larger clods of about 20–30 cm being selected (Fig. 4.4). The collection was then carried back to the village in a bilum3 and banana leaves were placed over the clay source to prevent it from drying out. Procurement Clay The Bel use a variety of clay sources, collectively referred to as ganal or rar, to produce pots. At Yabob, the potters use six distinct clay types, which are differentiated by colour. One of these, the ‘red’ clay, is close by the village, while the others require a slightly longer walk. At the red clay source, situated in a modern kaukau (sweet potato) garden, the young boys2 extracted the clay by using a metal digging stick to cut through the topsoil and expose the underlying clay deposit (Fig. 4.3). The clay was then extracted by hand and inspected by the elder potters for appropriate texture and lack of inclusions with As at Yabob, the Bilbil potters differentiate clays primarily by colour. The Bilbil potters select from only two clay sources, about one kilometre southwest of their village. This consists of a ‘yellow’ clay named rarmand, next to a small creek (Fig. 4.5) and a ‘black’ clay, rarsaran, that is dug out of the ground near a garden patch, only about 50 m away (Fig. 4.6). The yellow clay is lighter in colour than the black, but both have similar fine, clayey textures. The clay sources are covered with leaves to keep them from drying out while not in use. These sources have been in use for at least three generations as the modern potters say that their grandmothers were using the same clays. Both sources are considered equally good and are used to make the complete suite of vessel forms. Both clays can sometimes even be combined to make a single paste. 2 The boys extracting the clay were probably for demonstration only. In the 1970s the women exclusively dug for the clay in large groups (Mennis 2006a: 79-80), but Mager (1952: 341) does mention that men would also gather clay and help with clay preparation. The men certainly accompanied women to the mainland in pre-colonial times; however, it is not known how much involvement they had with procurement. In other areas of New Guinea the production process of paddle and anvil wares is dominated by women, but in inland areas of Madang labour is sometimes shared (May & Tuckson 2000: 18). During collection, the clay is dug out of the ground by hand and rolled into small balls, before being carried back to Bilbil in a bilum. There it is compacted into larger balls 3 String bag Figure 4.3. Extracting the ‘Red’ clay near Yabob-up-top (2014). 44       · .  Figure 4.4. Yeyeg (left) selects appropriate clay from the ‘Red’ source near Yabob-up-top (photo: M. Mennis, 2014). Figure 4.5. Dorcas rolls the ‘Yellow’ clay into a portable ball, near Bilbil village (photo: J. Field 2014). 45 Chapter . Modern Potting Communities and stored in a cool area under a house (Fig. 4.7). These clay balls then facilitate the production of pots for several weeks or months. During my study, clay samples were collected from the six sources near Yabob-up-top and the two clay sources near Bilbil village (Table 4.1). According to the potters, these represent all of the clay types in current use. I visited Bilbil – and Yabob , but at the request of the Yabob people, Yabob – were not visited and samples were collected on my behalf. At Yabob and Bilbil the emic classifiers of clay type by colour correlate to etic classifications derived from chemical characterisation. An energy dispersive x-ray fluorescence analysis at the XRF laboratory, Department of Anthropology and Archaeology, University of Otago, successfully delineated separate chemical signatures for the modern clay sources. Clay from each source was separately pulverised in a mortar and wet with distilled water. Each clay type was then moulded into three approximately 1 cm3 briquettes and left to dry at 30°C for two days. For each clay type, one briquette was fired in a muffle furnace for one hour at 700°C, one at 1000°C, and one was left unfired (Table 4.2). Ethnographic observations in the Massim suggest open firing, Figure 4.6. The ‘Black’ clay source near Bilbil village (2014). Figure 4.7. Balls of clay stored after collection, Bilbil village (2014) 46       · .  Table 4.1. Clay sources in use by Bilbil and Yabob potters, Madang. Clay code Bel name English name* Munsell code** Source location Bilbil 1 rarmand Yellow 10YR 5/6 yellowish brown E362533, N9415371 Bilbil 2 rarsaran Black 2.5Y 4/2 dark grayish brown E362483, N9415368 Yabob 1 Unknown Red 10YR 4/6 dark yellowish brown E364915, N9419816 Yabob 2 Unknown – 2.5Y 4/3 olive brown Unknown Yabob 3 Unknown – 2.5Y 3/1 very dark gray Unknown Yabob 4 Unknown – 2.5YR 4/8 red Unknown Yabob 5 Unknown – 2.5Y 4/2 olive brown Unknown Yabob 6 Unknown – 5Y 6/2 light olive gray Unknown * used by the potters to describe clay colour; ** wet, unfired clay (Munsell Soil Colour Chart, 2009) Table 4.2. Colour* of Bilbil and Yabob clays: unfired, 700°C and 1000°C. Clay code Unfired (dried) 700°C 1000°C 10YR 6/6 brownish yellow 5YR 4/6 yellowish red 2.5YR 5/6 red 2.5Y 5/3 light olive brown 5YR 4/6 yellowish red 5YR 5/8 yellowish red 10YR 4/6 dark yellowish brown 10R 3/1 dark reddish gray 2.5YR 3/1 dark reddish gray 2.5Y 4/2 dark grayish brown 5YR 4/6 yellowish red 2.5YR 5/6 red 2.5Y 3/1 very dark gray 5YR 4/4 reddish brown 5YR 5/6 yellowish red 2.5YR 4/6 red 10R 4/6 red 2.5YR 4/6 red 2.5Y 4/3 olive brown 2.5YR 4/6 red 2.5YR 5/6 red 2.5Y 6/4 light yellowish brown 5YR 5/6 yellowish red 5YR 6/8 reddish yellow Bilbil 1 Bilbil 2 Yabob 1 Yabob 2 Yabob 3 Yabob 4 Yabob 5 Yabob 6 *(Munsell Soil Colour Chart, 2009) 47 Chapter . Modern Potting Communities similar to that used around Madang, may reach between 700°C to <900°C (Lauer 1972). unfired. Importantly, the two Bilbil clays separate from the Yabob clays suggesting that location within the clay alluvium may play a part in the chemical variations observed. Each briquette was then shot for 300 seconds with XRF (Bruker AXS Tracer III–v portable XRF) using a vacuum setting with no filter (15kVper channel, filament ADC = 45μA) to measure isotopes of ten light elements (Na, Mg, Si, Ti, Al, Fe, Mn, Ca, K, P) and again with green filter settings (40kV per channel, filament ADC = 24μA, filter = 12milAl+1milTi+6milCu) for ten heavy elements (Mn, Fe, Zn, Ga, Th, Rb, Sr, Y, Zr, Nb). XRF was considered appropriate to characterise the chemical differences between clays owing to the scarcity of natural mineral inclusions that cause interference in tempered ceramics. Slip Madang style ceramics are noted for their bright red appearance, caused by the application of a slip (a suspension of clay particles in water). Although the earliest redslipped ceramics in New Guinea are Lapita, the Yabob and Bilbil potters are the only modern industries in the region that use the technique. There are two sources of clay used to produce slips: one near Yabob-up-top village, and another closer to Bilbil, at Hudini village. These clays are called main. Although the sources of these slips were not visited, samples were collected on my behalf. Experimentation with each slip showed that they were hard and not ideal for pot production, but when mixed with water could be applied and manipulated easily. Upon firing, the slip would transform from a dull orange-red paste to a bright red coating. Discriminant function analysis of the calibrated chemical data plotted eight distinct clay groups in line with the eight Yabob and Bilbil clay sources (Fig. 4.8). The effect of the firing process was minimal and the group centroid for each clay type separated clearly from others regardless of whether the sample was fired at 700°C, 1000°C, or left Source 30 Bilbil 1 Bilbil 2 Yabob 1 20 Yabob 2 Yabob 3 Yabob 4 Bilbil 2 Yabob 4 Yabob 5 10 Function 2 Yabob 6 Yabob 6 Group Centroid Yabob 5 0 Yabob 2 Bilbil 1 Yabob 1 -10 Yabob 3 -20 -30 -30 -20 -10 0 10 20 30 Function 1 Figure 4.8. Canonical discriminant functions (1 and 2) of clay samples from Bilbil and Yabob. Circles indicate each sample (unfired, 700°C, 1000°C) and squares indicate the group centroid of those three samples. 48       · .  Table 4.3. Ethnographic sand tempers collected from Bilbil and Yabob areas, Madang. Temper Both the Yabob and Bilbil potters manually temper their clays with mineral inclusions before vessel forming. Potters currently select black beach sand for temper because it is available immediately adjacent to their villages on the mainland (see Chapter 2). At Yabob-up-top, potters must descend down a steep track from the cliff to reach the beach and collect temper. At Bilbil village, sand is collected from the shoreline, only 10–20 m from many potters’ houses. Before, when people lived offshore, the potters say white beach sand was used from the islands, as it was more accessible. The Bilbil potters consider both white and black sand to be of ‘one kind’ as they are equally good for production. This is in contrast to observations in other areas of New Guinea where potters preferentially select black sand (May and Tuckson 2000). In order to produce a fine and malleable paste, the sand must be dried before mixing with the clay. Crushed coral, shell, or ceramic are not recognised as temper by the modern potters. Location Colour Inclusions Bilbil village Black Lithic fragments, clinopyroxene, calcareous (rare) Yabob-up-top Black Lithic fragments, clinopyroxene comprised of clinopyroxene and lithic fragments, and the Bilbil sand also contains rare calcareous grains. Modern potters at Bilbil and Yabob preferentially select production resources that are immediately available. This minimises labour requirements and reduces risk. Importantly, although the clays separate out chemically, they are not very far apart geographically. This is illustrated most clearly by the close proximity of the Bilbil yellow clay to the chemically distinct Bilbil black (Fig. 4.9). The Bilbil clays are within about 300 m of the village and the slip is only about 1 km away. Although the exact locations of the Yabob clays and slip are unknown, the one known source of red clay falls in the same pattern as the Bilbil raw materials, being very close to the village (Fig. 4.10). For the purposes of describing the Bel clays, the term ‘source’ will refer to an individual pit dug for the extraction of raw materials, which generally seem to be chemically distinct. When viewed in the geological context (see Chapter 2), the Bel Temper samples were collected from the two beaches/ river mouths used by the Bilbil and Yabob potters (Table 4.3). This included black beach sand from Bilbil and Yabob villages. Petrographic analysis for each sand grain assemblage was completed by James Scott at the Department of Geology, University of Otago, and revealed a relatively homogenous range of minerals. Each temper is primarily Hudini slip Black clay Yellow clay Bilbil Village Black sand temper 0 200 m Figure 4.9. Local ceramic resources used by modern Bilbil potters. 49 Chapter . Modern Potting Communities Red clay Yabob-up-top 0 300 Black sand temper m Figure 4.10. Local ceramic resources used by modern Yabob potters. Note that the source of Yabob clays 2–6 and the source of Yabob slip could not be recorded. clay sources seem to be restricted to the alluvial deposit bordering the coastline, south of Schering Peninsula and modern Madang town (Fig. 4.11). Importantly, the resource spaces of the Yabob and Bilbil do not seem to overlap as they are restricted to about 1 km. Kranket Is. Madang Production 3km Emic typologies 1km The Madang potters classify their ceramics based on a few significant attributes, such as size and rim form (Table 4.4). The Bilbil potters differentiate four different vessel types. The bodi is the most common cooking pot, with a globular body, everted rim and shoulder carination. Variant, and less common cooking pots include the magob, which is similar to the bodi but lacking an everted rim, and the tangeng, identical to bodi except much larger in size. The youbodi (pronounced yo-body) is very rarely produced. It is a large, globular water pot consisting of one to three small orifices with everted rims (spouts) at the top of the vessel. At Yabob-up-top in 1994, Mennis (2014: 141) observed only two pot types still in production: the bodi and the nombu (an equivalent to the Bilbil you-bodi). Twenty years later, in 2014, only the bodi was still in production, albeit extremely rarely. Although there is variation between individual potter’s techniques, the potters all work within the recognised technological framework for producing one of these distinct forms. Yabob-up-top Yabob Is. Bilbil village 1km Bilbil Is. 3km 0 5 km Alluvium Clay Kabenau Beds Slip Figure 4.11. Range to ceramic resources used by modern Bel potters near Madang. Note that the source of Yabob clays 2–6 and the source of Yabob slip could not be recorded. 50       · .  Table 4.4. Emic pottery classification at Bilbil and Yabob. Bilbil village Bel name Emic classifier Example Primary function Everted rim Cooking magob Direct rim Cooking tangeng Large Cooking you-bodi Small orifice(s) Water storage bodi Yabob-up-top Bel name bodi nombu Emic classifier Example Primary function Everted rim Cooking Small orifice(s) Water storage 51 Chapter . Modern Potting Communities Within each pot type, potters can distinguish technical variations and even identify individual potters. For instance, some potters make slightly longer everted rims, others, short everted rims. However, only women who pot can tell these differences, along with differences in quality; non-potters usually just recognise the object as a Madang pot. Similarly, Yabob and Mindiri pots are slightly different from Bilbil pots. The Yabob/Mindiri pots are often not as smooth and polished between the carination and the rim as the Bilbil pots. Further, over the past few decades, the Bilbil pots are being made smaller and smaller to cater to tourists who do not want to travel encumbered with large bodi, while larger forms are still sold for cooking on the Rai Coast and around Madang. The tourists also more often buy pots that have a decorative flair such as with additional spouts, geometric perforations, and inscriptions noting the place of manufacture. The way the potters produce pots has altered, and a wider range of experimentation within the general ‘Madang style’ is encouraged. That is, so long as someone can tell from the bright red slip or the general shape that the pot is ‘Madang,’ the potters feel free to experiment with decoration and form where possible. We might assume that similar consumer selection would have affected the morphology of vessels produced in the pre-colonial period. However, the significant difference is that nowadays people visit Bilbil and select what to purchase or exchange for, while in the past the Bilbil would select what to take to others as they loaded their canoes for trading voyages. Cognitively, the Bel potters divide the vessel into several structural units. This includes the bodi, referring to the body and the pot as a whole, the bodi aen, which equates to the rim, running around the opening, and the bodi avan, which refers to the orifice itself, especially used to discriminate the size. The Mindiri pots are said to have much smaller bodi avan than Bilbil vessels, for instance. The change of vessel morphology is entangled with the desires and expectations of the market along with recent historical endeavours to monopolise on production. In the 1960s a Danish potter set up a pottery workshop at Yabobup-top and encouraged the women to use a potter’s wheel and kiln firing. When this potter left, the Yabob reverted to traditional pot making (Mennis 2006a: 277). However, this disruption seems to have resulted in changes to vessel form, especially the accentuation of the shoulder carination. Precise differences between modern and pre-colonial production processes are, as yet, unclear. Clay preparation When production is to take place, clay is broken off the larger nodules and prepared for use. Large rock inclusions and roots are extracted by hand but smaller inclusions with the teeth. Clay is mixed with sand temper and pounded with an anvil on a bark mat until a few centimetres thick (Fig. 4.12). This mixture is called isol, where the correct mix is established not by set ratios but by feel and experience: the mix that does not stick to the hand Figure 4.12. Yeyeg mixes clay and temper at Yabob-up-top (2014). 52       · .  and can be rolled easily is preferred. It is then soaked in a trough or old canoe hull until saturated. The aqueous clay is then removed from the water, kneaded, and placed under a house in thick rolls to drain for about a week, after which it is ready for forming. drive the tibud (spirits) out of the clay. Fingers are used to smooth the interior of the rim, holding the pot from the inside, whilst simultaneously using a paddle, known as a hunuah (~30 cm long, rectangular piece of wood) to form the outer neck and lip. Then, placing an anvil inside the vessel, the outside is beaten into shape with a longer paddle (~40 cm half-cylinder). The next step uses a shorter, flatter paddle (~20 cm long, thick rectangle) to flatten and smooth the outside of the pot. At the end of this stage the shoulder carination begins to form about one third of the way down the vessel (Fig. 4.14). In all three stages, potters frequently touch the paddles with fresh water to moisten the clay. Roughing out and forming Vessel forming of a bodi pot was observed at Bilbil village (with Dorcas Kana) and at Yabob-up-top (with Yeyeg Kola). Although there is variation between each potter’s techniques, the overall method was observed to be consistent. The rim is moulded first. The clay body, isol, is compacted into a sphere about the size of a coconut, and rotated, using the thumb to create the orifice, in a process called bodi badi (Fig. 4.13). This rotation is rapid and gives the impression of a wheel-throwing technique. Fresh water is used sparingly to moisten the clay when required; saltwater is never used. Once the rim preform is roughly moulded it can be left for a day or two to harden slightly. Decoration After the final forming stages, the pot is further smoothed with a paddle to produce a good working surface between the neck and carination. The potter decorates this surface using a bamboo tool (~15 cm long, pointed at each end) to create impressions and incisions. First, to decorate the circumference of the neck, the bamboo is held about 5 cm from the point, between the thumb and forefinger, as the potter presses the tool into the clay at a steep angle (Fig. 4.15; Method 1 in Fig. 4.16). This decoration is described as bodi duam because the technique leaves balls of raised clay under the rim, just like the fruit of the duam tree. Secondary working forms the vessel body. Pati, or anvils are used to pound a concavity through the rim preform, shaping out the body. These anvils are ellipsoid river pebbles from the Rai Coast which are valued objects passed down from mother to daughter in law, teacher to apprentice. They are said to contain ancestral spirits, which Figure 4.13. Yeyeg moulds rim preform at Yabob-up-top (2014). 53 Chapter . Modern Potting Communities Figure 4.14. Yeyeg using a short, thick paddle to form shoulder carination, Yabob-up-top (2014). Figure 4.15. Yeyeg demonstrates decorating the neck with impressions, Yabob-up-top (2014). Note the potter is exhibiting the technique on an already slipped and fired pot. 54       · .  Method 1. Impressions around the base of the neck Method 2. Incisions on the vessel shoulder Figure 4.16. A common decoration applied by modern Bilbil and Yabob potters involving incision and impressions. Next, the potter holds the bamboo with a pinching grip about 1 cm above the point, making gash incisions in linear and semi-circular arrangements around the pot, just above the carinated shoulder (Method 2 in Fig. 4.16). The pot, having taken form and now decorated, is then left for two days to a week, until leather hard (Fig. 4.17). Once a pot has received its decorative additions it is considered bodi jumusek. nesses. Once all the cracks have been repaired, the pot is transferred to the large fire. The size of this large fire is dependent on the number of pots in the batch. An enclosure of dry palm wood is built up, one on top of another, with pots being placed in the centre. When the fire is hot, Kunai grass is placed on top of the pots until they turn red with the heat, after which, the fire is allowed to cool. Once fired, the vessel is considered bodi ibul damanginau and has become as a young man, just like the young men who emerge from the fires of initiation decorated with red paint. Decorations on recent Bilbil pots are visibly different from archaeological examples that people find on the ground, and the potters are aware of these differences. The stories regarding pre-colonial decorations have not been passed down. The modern decorations of double incised semi circles placed just above the shoulder carination seem to be arbitrary and do not have any particular significance to the potters. The potters cannot remember when this modern design originated, but it has been in use for at least a few generations. In recent years tourist decorations have also come into fashion. Certain designs that seemed to please tourists, or in some instances were even requested by tourists, have now been added to the design system. Slipping After the decorating stage is complete, the slip, main, is applied. The main is mixed with fresh water and applied by hand across the whole vessel and inside a small portion of the rim and neck (Fig. 4.18). Beginning as an orange-red colour, it is left to dry, and only upon firing does the slip take on its brilliant red characteristics. Firing The firing process is two-stage. First, a group of pots are placed amidst a small, low temperature fire, made from a few sticks of soft wood or pitpit.4 This small firing hardens the slip and tests the pot to see if it will crack when exposed to higher temperatures. Those pots with cracks at the end of the first firing are removed and put in fresh water to remodel, and smooth out any apparent weak- Figure 4.17. Kasare Dadau of Murpat Clan holds a decorated magob pot awaiting slipping and firing, Bilbil village (2015). 4 A type of wild sugarcane/cane grass. 55 Chapter . Modern Potting Communities Finishing techniques To complete the production sequence, the pots are later polished. Sago paste is mixed with water over a fire to produce a semi-transparent, glue-like substance (Fig. 4.19). This is then applied to the fired pot using a banana leaf or cloth (Fig. 4.20). The paste acts as a varnish and gives the red pots a lustrous shine. The paste dries on the pot within a few minutes, which is then ready for distribution or use. Distribution Although the Yabob no longer distribute pottery, finished pots are sold or exchanged at Bilbil village. The pre-colonial arrangement saw the Yabob and Bilbil people sail to other groups along the coast to distribute pottery (see Chapter 3). According to some potters interviewed, prior to European contact the Bilbil people traded mostly south to the Rai Coast and Astrolabe Bay, not north towards Yabob. In pre-colonial times, two or three pots could be traded with certain trade-friends for a lot of food or even pigs. Pots could also be used to pay bride prices with nonpotting groups. Distribution of pottery has changed significantly over the past century. Now that motorboats are common, people come from various places, especially the Rai Coast, to acquire the pots at Bilbil Village. Pots can be bought or exchanged for other items or food. They can also feature in brideprices to other villages. Tourists visiting Madang or on cruise ships also come to Bilbil to buy pots for good prices: a bodi or tourist pot would cost approximately 15– 30 PNG kina. Bilbil has styled itself as a tourist village and Figure 4.18. Dorkas Kana applying slip to unfired vessels (photo: M. Mennis 2014). Figure 4.19. The sago paste ready to be applied to the unpolished vessel, Bilbil village (2015). 56       · .  as tour buses park outside the village, potters lay out their wares in front of them for purchase (Fig. 4.21). In recent years, Bilbil potters would sometimes take their wares into Madang town centre, or even to Lae or Port Moresby for sale (Fig. 4.22). Within the last decade, the geographic mobility of Bilbil pots has extended even further, as the pots collected by tourists are sold online for inflated profits with foreigners acting as middlemen (Fig. 4.23). Consumption At Bilbil, despite metal vessels becoming more popular, a lot of women still cook with clay pots, which are said to infuse the food with a better flavour than metal. Other nearby villages also cook with pottery but less frequently. The bodi is the most common cooking pot used for boiling rice, taro, banana, yam, sweet potato, and other vegetables, along with pig and chicken. This is used for everyday home cooking. The magob, a less common cooking pot, is used for the same function as the bodi. Tangeng are used for boiling the same foods as bodi and magob, but typically at larger gatherings owing to their increased size. Pig, in particular, is cooked for feasts in tangeng. In the event of such feasts, supports in the form of two split logs called langalang are placed parallel and slightly apart on the ground, along which multiple pots can be placed for cooking. The you-bodi is used specifically for water collection and storage. The role of this vessel type seems to have diminished more than cooking vessels, perhaps owing to the introduction of more durable containers. Prior to the colonial period, the potters say that these pots had exactly Figure 4.20. Sentie Noah of Murpat Clan polishes a fired bodi using sago paste (2015). Figure 4.21. An assortment of standard bodi and tangeng, alongside tourist pots, laid out for sale at Bilbil village. Some have small pricetags attached (photo: M. Mennis, 2014). 57 Chapter . Modern Potting Communities Figure 4.22. Dorkas Kana sells her pots in Port Moresby (Source: emtv.co.pg). Figure 4.23. Tourist pots for sale online (Ebay.com: search “bilbil pot” 2015). the same uses, excepting the introduction of rice and other overseas imports. Leaves, known as uzak, can be used to line the insides of pots to stop the food burning. A herb called bilikuk can be used for the same function but also adds spice to the food. Other leaves such as bidifun would cover the mouths of pots, which would then be covered with a coconut shell, called a bekuten, to boil. pots it is considered fully broken and it is replaced for a less damaged su. Sometimes these broken vessels will be filled with soil and planted around the village as flowerpots (Fig. 4.25), but it is not clear when this tradition started. If either an in-use pot or a su is dropped on the ground and breaks into pieces, it is left as is, or is swept onto the rubbish pile (depending on where the pot has been dropped). Reuse and discard Summary Madang pots are reasonably durable and can last several years while in use, but they are still prone to breakage from heat fracture during cooking or impacts. When a pot is damaged it becomes a su. This is turned upside down and arranged in groups of three to form a tripod stand, supporting in-use bodi, magob, or tangeng whilst cooking (Fig. 4.24). The fire is placed in between the three su, which heats the cooking pots from below. When the su is further broken to the point that it can no longer support in-use Vai: a Madang style chaîne opératoire The vai represents the unique Bel production process from procurement to vessel forming and finishing. Conventionally, men would accompany women to collect the clays while women would prepare the raw materials, form the pots, and finish the vessels. In the past, men would then have the primary role in the distribution of pots (see Chapter 3), but recently this role has shifted to both men 58       · .  Figure 4.24. Three su (broken pots) are arranged to support an in use tangeng cooking vessel. The fire is placed in between the su, below the cooking vessel. Bilbil village (2015). Figure 4.25. Several bodi being reused as flower pots, Bilbil village (2014). 59 Chapter . Modern Potting Communities ability of metal cooking vessels, pot-making persists as a significant factor in community strategies of re-presenting their past to others and re-affirming their own identities as skilful pot-makers. It is not the finished pots but rather the technological processes which carry meaning, reframing the ‘traditional’ in order to negotiate the fluid nature of the modern pottery market (cf. Colombo Dougoud 2005; Graburn 1976, 1979). In recent years, production has diversified as a result of the tourist-driven market economy, but ‘traditional’ pots, of a style said to derive from the pre-colonial period, are still manufactured and exchanged regionally and locally (Gaffney 2018). This chapter has recorded the chaînes opératoires of these ‘traditional’ pots, providing the necessary analogue to trace these lines back in time. The remainder of this volume focuses on changes to the vai through time by investigating pre-colonial ceramic chaînes opératoires, including their archaeological context, forming and decorating techniques, raw material constituents, and distribution. This begins in the next chapter, which surveys existing archaeological evidence for Madang ceramic production and exchange and identifies important differences between the modern and pre-colonial times. and women or solely women. During the forming process women use specific techniques and specialised tools at different stages to consciously produce different attributes of the finished pot. Figure 4.26 depicts Yeyeg with examples of each stage in the forming sequence: from the rim preform (left) to the finished pot (right). These represent the emic links in the production sequence, identified as necessary to produce a finished bodi pot (see also Fig. 4.27). The chaîne opératoire for the taneng, magob, and you-bodi were not observed; however, we could expect that the tangeng sequence would be identical to the bodi, except much larger in scale, while the magob and you-bodi sequences would require a different rim forming technique. These embodying technological processes both reproduce and maintain the practice of pottery production, as well as illuminate the patterns of exchange and discard which are observable in the archaeological record through the study of potsherds. Although production and distribution is no longer the economic necessity as it was in the past (Mennis 2014), and consumption is ever diminishing owing to the wide avail- Figure 4.26. A display of the vessel forming sequence, Yabob-up-top (2014). 60       · .  Material Rim forming Procurement Process Clay Temper Fresh water Slip Soft wood Hard wood Sago paste Continued Continued Continued Continued Continued isol bodi badi rim preform hollowing Primary forming rim shaping body forming carination forming Decorating bodi duam secondary decoration bodi jumusek D R Y I N G Continued Reworking after small ﬁring Figure 4.27. A chaîne opératoire of a modern Madang style bodi pot (continued over). 61 Chapter 4. Modern Potting Communities Continued Slipping main applied Fresh water D R Y I N Slip Soft wood Hard wood G small ﬁre reworking Firing big ﬁre Finishing bodi ibul damanginau polish Distribution ﬁnished bodi trade/exchange Use breakage cooking Reuse damage Discard su storage/ﬂower pot breakage potsherds Figure 4.27 (continued). A chaîne opératoire of a modern Madang style bodi pot. 62 Sago paste Chapter 5. Traces of the Past the conclusion that further work was required to refine the ceramic sequence and settlement patterns in the area. The sequence of site occupation in the coastal Madang area remains to be determined. — Brian Egloff (1975) Little has changed since Egloff ’s statement over 40 years ago – anthropologists still know little about the pre-colonial occupation, production and exchange along the Madang coast. Our knowledge of Madang culture history and archaeology stems from a handful of provisional investigations along with larger scale operations outside of the area. However, all of these studies have drawn attention to the striking consistencies, but also subtle differences, between the archaeological and the ethnographic pictures. On that basis, this chapter surveys the current state of knowledge regarding pre-colonial production and exchange around Madang, with a particular focus on previously excavated ceramic assemblages. This follows on from the discussion of recent production and exchange presented in chapters 3 and 4 by extending the scope deeper in time. First, the chapter examines archaeological investigations near Madang itself in order to assess the nature of pre-colonial ceramic production. Secondly, it looks at the wider New Guinea area where Madang style pottery has been identified, in order to demonstrate the extent and influence that Madang’s material culture had in exchange at a regional level as these ceramics passed through different social networks. Lastly, it draws together chronological evidence for Madang ceramic manufacture to establish a tentative time depth for the style and the origins of Bel potting around the northeast coast. Allen collected 120 rims and decorated sherds from the disturbed deposit of a road cutting near Bilbil (PNG site code: JAE). Stratified pottery was noted to 40 cm deep in the cut face of what would have been an old village mound. The site was assumed to have ‘no great antiquity’ but probably predated European occupation. The pottery examined was all red slipped or blackened from the firing process and related in fabric to the modern Madang industries. Excluding ornamental pots designed for tourism, Allen observed only one form of vessel being produced at Bilbil: the bodi, a globular cooking pot with everted rim and carinated shoulder, decorated with geometric incision before slipping (see Chapter 4). However, from the collected sherds, he established four types (A–D) based on rim form (Table 5.1), decorated not just with incision but also applique, suggesting much greater variation in pottery design features in the past. Allen’s Type A and B are similar to the modern bodi and tangeneg, while Type C is related to the magob in rim form. Type D is unlike those pots produced ethnographically (see Chapter 4). Just a few years later, Egloff attempted to construct a rudimentary culture history for the area as part of a broader ethnographic study (Egloff 1973). He recorded 17 sites along the coast and on islands near Madang, six on Karkar Island, and one at Sarong village (Table 5.2; Fig. 5.1). Pottery directly ancestral to modern Madang pottery was found at 19 of these sites. A single piece of obsidian collected from Yabob was sourced to Talasea, West New Archaeological anthropology around Britain (Bird et al. 1981) demonstrating interaction with Madang trade links in the Vitiaz Strait. Surface collections or small Published archaeological investigations into pre-European test pits were implemented at most sites, but at Tilu (JCA), social life around Madang itself are limited. In 1969 Jim a clan area of Malmal village, excavation of two large anAllen, then at the Department of Anthropology and Soci- thropogenic mounds was undertaken. ology, University of Papua New Guinea, collected artefacts near Bilbil village (Allen 1971), and in 1973 and 1974 Brian In 1973, Egloff dug a 1 × 1 m test pit into Tilu, Mound A, Egloff, then at the Papua New Guinea National Museum to 120 cm deep, and in 1974 he returned to dig a 1 × 2 m and Art Gallery, conducted non-intensive survey and ex- unit into Mound A to 75 cm deep, and Mound B to 110 cm cavation at various locations from Karkar Island in the deep (Fig. 5.2). The deposits lacked defined stratigraphy, north to Bilbil Island in the south (Egloff 1975). These apart from occasional yellow and grey ash lenses, so arbistudies led to tentative attempts at artefact typology with trary 20 cm spits were excavated until dense coral gravel 63 Chapter . Traces of the Past Table 5.1. Allen’s (1971) ceramic types based on a surface collection near Bilbil village. No. of sherds % of collected Variation in angle of eversion and rim crosssection. Most rims straight, but some curve upward and others have grooves on the interior surface of the upper rim. 44 52% Everted rim with squared lip Variation in angle of eversion. External face of the rim bevelled to produce a square section. 6 7% C Collared Most commonly small and rounded collar. Collar attached separately after completion of body. 33 39% D Other Wide rims above a groove on the external face where the body joins the rim. 2 2% Type Classifier Description A Everted rim B Example was reached at the base of each pit. These ash lenses may relate to repeated eruptions of nearby Karkar, Arop/Long, or possibly the Yomba Island referred to in Bel oral history (cf. Chapter 2). Radiocarbon determinations indicate Tilu was occupied sometime between ~300 and 700 years ago (Table 5.3), and pottery present from the basal layers would suggest that the Madang style ceramics were already well established by this time. The conventional age of 550 BP at Tilu has been said to conflict with oral traditions regarding Yomba’s disappearance (e.g. Mennis 2006a: 78); however, it should be noted that the 2σ range for each of Egloff ’s dates do not specifically contradict the genealogical chronologies, and the lines of evidence overlap at ~300 years ago, assuming no complications in generational length or elision. 64 The subsequent ceramic analysis of collected material described significant formal variation within, and beyond, the four major classes already established by Allen (1971). In particular, incurving rim forms and complex outcurving-incurving forms were added to the known rim types (Table 5.4). Egloff ’s description of rim forms highlights the variation present in Madang style ceramics but does not correlate with the emic classification of pots as Allen’s typology does, based as the latter was on a few key classifiers. An attribute analysis was used to suggest that formal types were essentially represented by decorative techniques (Table 5.5). Body decoration was associated with a suite of attributes including rim form, lip decoration, interior rim notching, and red slipping. Further, these decorative types were spatially restricted. For instance, punctate sherds       · .  Table 5.2. Summary of sites recorded near Madang by Egloff (1975). Site code* Local name Location Description of site Work completed Material culture recorded Madang sherds JBA Tadilon Bilbil Island (SE coast) Natural shallow midden Test pit excavated Potsherds Yes JBB Luahor Bilbil Island (E coast) Number of middens extending to cliff 1 × 1m test pit excavated to 90 cm. Potsherds, obsidian, tapa beater, mollusc Yes JBC — Bilbil Island (N coast) Large midden being eroded by sea Collection from eroding face ~2m deep Potsherds Yes JBD — Bilbil Island (W coast) Historic habitation area Surface collection Potsherds Yes JBE Idwan Sek Island (W coast) Beach Surface collection Potsherds Yes JBF — Migajpanad Island — Surface collection Potsherds Yes JBG Manip Yabob Island (centre of Is.) Historic habitation area 0.5 × 1m test pit excavated to 105 cm Potsherds, shellfish Yes JBK Sarong Sarong village Modern village Surface collection Potsherds No JBL Kabai Karkar Island (hills on SE side of Is.) Hill-top garden Surface collection Potsherds Yes JBM Madip Karkar Island (hills on SE side of Is.) Coconut plantation on hill crest Surface collection Potsherds Yes JBN Wabef Karkar Island (S side of Is.) Pig root, pre-German habitation Surface collection Potsherds Yes JBO Babu Karkar Island (S side of Is.) Pig root, pre-German habitation area Surface collection Potsherds Yes JBP Kamalong Karkar Island (S side of Is.) Historic habitation area Surface collection Potsherds Yes JBQ Wasmak Karkar Island (S side of Is.) Garden, historic habitation area Surface collection Potsherds No JBR Vidar Vidar Plantation worker’s housing Surface collection Potsherds Yes JBS — St. Fidelis Seminary Seminary Excavated to 40 cm. Potsherds, mollusc Yes JBT — Nagoda Plantation Coral mine, preGerman habitation Surface collection Potsherds Yes JBU — Tabad Island Modern house on raised mound Test pit excavated to 70 cm. Potsherds Yes JBV — Riwo Island (NE of Is.) Modern village Surface collection Potsherds Yes JBW Malamal Udou Island near Malmal village (SW side) Midden 0.5 × 0.5m test pit excavated to 65cm Potsherds Yes JBX — Malmal Island Midden (in village pig sty) Surface collection Potsherds Yes JBY — Malmal village Midden next to copra drying building Surface collection Potsherds Yes JBZ — Yabob Island** (SW coast) Rise behind the beach Test pit excavated to 90 cm Potsherds Yes JCA Tilu Malmal village Two large midden mounds One 1 ×1m and two 1 × 2 m test pits excavated to varying depths Potsherds, obsidian, bone, shell artefacts, shellfish, taro peeler, human remains Yes *recorded in the Niugini Archaeological Survey, Papua New Guinea National Museum and Art Gallery. ** Yabob Island (cf. site record form in Papua New Guinea National Museum and Art Gallery), contra. Egloff (1975), which incorrectly states Bilbil Island. were never found with impressed sherds. These restric- The investigations by Allen (1971) and Egloff (1975) were tions could relate to selectivity in exchange, activity areas exploratory and coarse grained in nature. Limitations on on site, or temporal changes. The material composition sample sizes and a lack of direct dating precluded conof the sherds was not considered by Egloff, but he noted struction of a local cultural sequence. However, the dethat compositional variation might ‘delineate significant scriptions of pot forms, decorations, and attributes laid the cultural features’ (Egloff 1975: 17). foundations for comparative studies further afield. 65 Chapter . Traces of the Past 0 20 0 km 30 MN m Karkar Island JBN-JBQ 1m 2m JBL; JBM JBK Sarong 1974 1×2m 3m Mound B 2m 1m JBS JBR JBE JBW JBF JBY; JCA JBX JBU; JBV JBT Mound A 1974 1×2m 2m 3m Madang 1973 1×1m JBG; JBZ JBA-JBD Figure 5.2. Previous excavations at Tilu (JCA) site (Egloff 1975: map 2). Figure 5.1. Archaeological sites around coastal Madang, examined by Egloff (1975: Map 1). Table 5.3. Radiocarbon dates for Tilu Mound A and Mound B (Egloff, 1975). Laboratory code Location Excavation unit Excavation level* Depth below surface (cm) Conventional age BP** Date range cal. BP (2σ)*** GX3561 Mound A 1 × 1 (1973) 5 and 6 80 & 100–120 550 ± 115 314–678 GX3633 Mound A 1 × 2 (1974) 4 75 540 ± 110 315–672 GX3632 Mound B 1 × 2 (1974) 6 100 & 105 555 ± 125 302–697 * GX3561 and GX3632 are combined charcoal samples from varying levels and depths. ** as reported by Egloff (1975). *** δ¹³C correction not originally made. Here an estimate of -25.0+-2 is applied to Egloff’s C14 age for charcoal before calibration (Stuiver et al. 2005: Chapter 5, Table 1). lands in New Britain (Gosden & Webb 1994; Lilley 1991; Summerhayes 2001a) (Fig. 5.3). Ancestral Madang exchange networks At European contact, the Yabob and Bilbil Island potters around Madang were the major pottery providers for the northeast coast and linked with the Vitiaz Strait trade networks. The distribution of archaeological sites containing Madang style potsherds indicates a similar scenario was operational in the past as well. Ancestral Madang pots perforated a number of distinct socio-economic fields: north to Karkar Island (Egloff 1975), south into the Central Highland ranges (Coutts 1967; Hughes 1977; White 1972), and east into Arop/Long Island (Egloff & Specht 1982), the Sio/Gitua area (Lilley 1986), the Siassi Islands (Lilley & Specht 2007), the Tami Islands off the Huon Peninsula (Abramson 1969), and as far as the Arawe and Kove Is66 North As mentioned in Section 5.1, Egloff (1975) recorded Madang style pottery reaching as far north as Karkar Island. Although these finds are undated, some of the Karkar sherds were decorated with linear appliqué indicating they are probably pre-contact in age, given this attribute never occurs on colonial or modern pots (Egloff 1975: Appendix 1). Further Madang style sherds were recovered from 2018 excavations at Kulili on the north side of Karkar, and Krabat and Ngus on the south side of the island (Gaffney et al. 2018b). The villages on the southeastern half of Karkar       · .  Table 5.4. Variation in Madang rim forms identified by Egloff (1975) Type Example Allen (1971) type Type Example Allen (1971) type 1 A 12 Absent 2 A 13 Absent 3 A 14 Absent 4 A 15 C 5 A 16 Absent 6 B? 17 A/B? 7 A 18 A/C? 8 Absent 19 D 9 A/B? 20 Absent 10 Absent 21 C 11 A/C? 22 Absent Table 5.5. Madang ceramic decorative techniques, identified by Egloff (1975). Technique Description Carved paddle impressed Slight angular or straight impressions made with a carved paddle. Nubble applique Heavy blobs of red slip applied to the pot surface. Incised Thin incisions applied after slipping. Linear applique Thin lines of red slip applied to the pot surface. Punctation Large shallow punctations arranged in broad geometric patterns. Plain Not decorated, but red slip may be present. 67 Chapter . Traces of the Past 0 500 km Karkar Island Madang Kove Islands Arop/Long Island NEW GUINEA Sio Gitua NEW BRITAIN Siassi Islands Arawe Islands Kafiavana rockshelter Aibura cave Tami Islands 0 100 km Figure 5.3. Location map showing sites in the northeast New Guinea region where Ancestral Madang sherds have been reported. Island belong to the Bel language group, speaking Takia, in the Bel sub-family of languages (see Chapter 1). For these Karkar Islanders, the exchange of pots up from coastal Madang would have played an important role in maintaining links with the Bel speakers to the south. It is notable, however, that no Madang sherds were recorded at Sarong village, another Bel group on the mainland adjacent to Karkar, where distinctive, deeply incised Sarong-style potteries instead dominated (Egloff 1975: Fig. 1). This suggests that Sarong and nearby areas along the northeast coast lay outside the interactive network that was clearly operative between Karkar and Madang, which conflicts with oral traditions. No Madang sherds have yet been reported from the Sepik north coast. South Madang pottery may have reached the Central Highlands by 800 years ago. A 3.5–5 mm thick hand-made ware excavated from a horizon dated to 770 ± 110 BP (GaK-622)1 at Aibura cave (NAE) in the Eastern Highlands probably derives from the Madang coast (White 1972: 62). Two sherds, in which beach sand with an Upper Tertiary volcanic base was used for temper, contained well-sorted fresh angular pyroxene, feldspar, and rounded basalt tempers. One sherd also contained marine shell fragments. This 1 A number of determinations generated by the Gakushuin radiocarbon dating laboratory prior to GaK-7000 have been questioned as being too old or too young (see Spriggs 1989). It is not known how this effects the Aibura sequence. was petrographically consistent with a reference sample from Yabob (Key 1973). Similar sherds were also found on the surface at Kafiavana rockshelter (NBZ) further west (White 1972: 95). By some ethnographic accounts (Coutts 1967; Hughes 1977), Madang pots were traded along the Ramu River, up into the Upper Ramu plains in the Eastern Highlands and then dispersed into Chimbu Province and beyond. The importance of ‘hinterland middlemen’ in the transfer of ceramics into the Highlands in the pre-colonial past is supported by surface finds of Madang style pottery at three find spots – Muli, Wul, and Wapain – near the Gogol River, about 15 km inland from the Madang coast (Gaffney et al. 2018c). In other recordings (Keil 1973), pottery reached the Benabena in the Eastern Highlands through the Rawa, who obtained pots from the northern Finnistere Ranges, where in turn people obtained them from traders on the Rai Coast: a process of down the line trade over several hundred kilometres of rugged country and involving numerous middlemen (Fig. 5.4). Despite pottery being of not much consequence to most areas of the Highlands, where people prefer to use more durable cooking and storage devices (Hughes 1977: 116; May and Tuckson 2000: 154), in areas where pots were manufactured such as the Agarabi in the Eastern Highlands, these Madang trade pots were often prized over others (Coutts 1967: 487). Bulmer (1977, 1985, 2007) ascribed one sherd at the site of Wañelek (JAO) in the Kaironk Valley to the Madang style owing to external red slipping, evidence of paddle form- 68       · .  established dates from the coast, including within the Madang area itself, suggesting the sherd does not necessarily represent Ancestral Madang. Excavations of other ceramic-bearing sites in the Kaironk and Simbai valleys, led by Glenn Summerhayes and Judith Field, are beginning to clarify this problem East Figure 5.4. Madang pot traded into Megabo, Eastern Highlands, shortly after WWII. Decorated with gash-punctations around the neck (Photo: D. Keil, 1970s). ing, and incision. Geochemical and petrographic analyses suggest that this sherd was produced at a source somewhere on the northeast coast, but not necessarily with the same manufacturing origin as the modern industries owing to the presence of orthopyroxene, absent in comparative samples from Bilbil and Yabob Islands (Gaffney et al. 2015, 2016). Importantly this sherd derives from a secure posthole feature with lower fills dating to 2865 ± 90 BP (I-6859) and overlying deposits dating to 2840 ± 90 BP (I-6861) (Bulmer 1977: Table 8.1). This is inconsistent with A series of investigations in the Vitiaz Strait throughout the 1970s–1980s illuminated much about the extent and age of Madang style potteries east of Madang itself. A multidisciplinary project on Arop/Long Island aimed to describe recolonisation of an island environment following the major seventeenth century eruption attested in oral traditions (Specht et al. 1982). The archaeological component, completed in 1973 by Brian Egloff and Jim Specht, investigated material recovered from five archaeological sites on the island: JCB (Biliau), JCC (Bara), JAB (Poin Bare), JCT (Kairu Point), and JCW (Patauru) (Egloff & Specht 1982). The pottery found at these sites was all imported, deriving from five clay types (A–E), and could be divided into four distinct style groups (I–IV). The majority of sherds collected were classified as ‘Style Group I’ and considered ‘Ancestral Madang.’ Sherds of this style were found at each of the five sites. This style was formed only from Clays A–B, and consisted of Rim Profiles 1, 3, and 4 (Table 5.6), with thin direct and everted rim courses and interior rim notching. These were decorated with appliqué, incision, and punctuation, and covered by a red slip. Appliqué and incision never occurred on the same sherd, however. Style Group I is associated with a mixed charcoal date of 350 BP (ANU-1307) from JAB/B. Table 5.6. Arop/Long Island style groups I & IV rim forms (Egloff & Specht 1982). Style group I Rim form 1 Rim form 3 Rim form 4 Rim form 7a Rim form 8 Style group IV Rim form 7 69 Rim form 8a Chapter . Traces of the Past Style IV, formed only of Clay B and comprising Rim Forms 7–8 (Table 5.6) is not considered to be of the Madang style by the excavators, but the rim forms certainly fall within Allen’s (1971) Type B and C and seem to correlate with many of the forms identified by Egloff (1975; see Table 5.4). There is a date of 1040 BP (ANU-1308) that may be associated with a single Style Group IV sherd from JCB/A. However, the excavators warn that the charcoal date may relate to the soil matrix, not the sherd itself, presumably owing to assumed disturbance (Egloff & Specht 1982: 430). Thermoluminescence dating on the same sherd (Sample 1027) indicated it was at least 360 years old. Madang pottery penetrated further east, probably through down the line trade along the Huon Peninsula and into the Siassi Islands. Extensive archaeological investigations in the Vitiaz Strait by Ian Lilley in the 1980s illuminated more about the age and extent of Madang ceramics than any other study to date (Lilley 1986). This work demonstrated that Ancestral Madang pots were moved to Sio (KBP and KBQ) and Gitua (KBZ) on the Huon Peninsula, and later the Siassi Islands (KLJ) sometime after about 1300–1000 years ago (Lilley 2004a, 2017). The earliest dated context for Madang pottery anywhere is at KBQ site around Sio where a shell date (ANU-4607) from 260 cm deep gave a radiocarbon date of about 1500 years old. A charcoal sample The Arop/Long Island material awaits thorough stylistic (ANU-4337) from the same site also returned a date of 1300 and compositional analysis; however, from the evidence years old. These dates are significantly older than anything presented, the majority of sherds do derive from the Ma- recorded in the Madang area, lending weight to the idea dang potting tradition. These pots certainly reached the is- that the Madang potters originated somewhere outside of land’s shores by at least 350 years ago, although a re-exami- Madang, such as the Yomba Island cited in oral histories nation of the local sequence and refined radiocarbon dates (see Chapter 1). Based on a similarity in mineral incluwould be required to determine just how and when the sions between Ancestral Madang sherds and Ancestral importance of Madang wares changed on the island over Sio sherds, Specht, Lilley, and Dickinson (2006) have also time. Although oral traditions in Madang do not mention suggested a Huon source for Ancestral Madang pottery. Arop/Long being directly part of the Madang trade networks, modern recipients of the pots on Arop/Long state It was not until later in time that Madang pottery acquired that the objects derive from the Rai Coast, suggesting a characteristics of high-volume production and exchange, down the line exchange (Fig. 5.5). This form of exchange reaching its greatest extent, crossing the Vitiaz Strait and may have operated in the distant past as well. On the other filtering into the Bismarck Archipelago (Lilley 2004a). Mahand, the island also had strong connections to the east dang sherds have been found at two sites (FPD and FPF) and seems to be placed at the crossroads of the Madang on Kalapiai, in the Kove Islands off the north coast of New and the Vitiaz Strait trade networks. Obsidian sourcing, Britain, dating to almost 800 years old (Lilley 1991). These for instance, showed that all of the Arop/Long specimens sherds occur in contexts also containing recent Sio-style derived from Talasea in West New Britain (Egloff & Specht sherds. Similarly, in the Arawe Islands off the south coast 1982: 432). The Vitiaz Strait presents another possible route of New Britain, Madang sherds have been recovered from of entry for ceramics into Arop/Long Island. contexts of about 900 years old (Gosden & Webb 1994). Figure 5.5. Woman and children on Arop/Long Island with Madang trade pot used for cooking (photo: M. Mennis 2014). 70       · .  On the four Tami Islands, off the eastern tip of the Huon Peninsula, preliminary research has described various imported potteries collected from the surface, some of which appear to be Madang style (Abramson 1969). Abramson’s Types IB and IE with nubble and linear appliqué certainly appear to be Madang, while Type IC with horizontal relief bands may also derive from Madang (Abramson 1969: plate IA, b, c, f, h, i, j, IB, e). This represents the furthest extent southeast that Madang sherds have been recorded; however, the finds are as yet undated. Specht 2007). It is hypothesised that Type X and Madang ceramics share a common technological origin, ultimately from Palau, owing to the mutual presence of grog tempering (Specht, et al. 2006). This possible Micronesian connection should not be casually dismissed. As Specht, Lilley, and Dickinson (2006) note, kava drinking, which, in north New Guinea, is only practiced in Madang and Manus is thought to have originated in Vanuatu and later spread via West Polynesia into Micronesia, and thence to New Guinea. An interesting further connection between Manus and Madang, is the mutual production and use of multispouted water vessels, otherwise unique in New Guinea (see Chapter 4; Egloff 1977: 82-83; May & Tuckson 2000: 11). Lilley (1999) compares the linguistic evidence of east-west population movement from West New Britain to the New Guinea north coast (see Chapter 1) with the archaeological evidence, suggesting that the two modes of evidence In an attempt to delineate significant stylistic indicators align. If he is correct, then the predecessors of the Madang on the pots themselves, Lilley (1986) revised the proviindustries have an origin somewhere in the Vitiaz Arc and, sional classification systems established by Allen (1971) ultimately, the Bismarck Archipelago. and Egloff (1975). He avoided an absolutely typological approach and instead focused on designating three analytical Lilley (1988a) argues that the ethnographically observed classes, based on distinct lip attributes (Table 5.7). Class 1 levels of production and trade intensity originated only rims include everted and vertical rims with rounded lips. in the last few hundred years, when the movement of Ma- Class 2 rims have bevelled exterior faces and pointed lips. dang pots was concurrent with the movement of Sio and Class 3 rims have flat lips. Although it is not explicit why Type X ceramics. Type X pottery may have been in pro- lip attributes were chosen to delineate each class, the efduction somewhere on the eastern Huon from about 1000 fect was to streamline the method of analysis – Egloff ’s 22 years ago until about 500 years ago (Lilley 1988b; Lilley & types were reduced to three major classes. Table 5.7. Lilley’s (1986) Madang style rim classes based on sherds excavated from the Vitiaz Strait. Class Classifier Description Example Allen equivalent type Egloff equivalent type A, C, D 1, 2, 3, 7, 8, 10, 12, 13, 15, 16, 17, 19 61% 5, 6, 9, 11, 14, 18 31% 4, 20, 21, 22 8% 1 Round lip ‘…short, thick, vertical to everted forms with round lips.’ 2 Pointed lip ‘similar in most respects to…Class 1 but have a more pointed lip’ B, C 3 Flat lip ‘…distinctive flat lips…’ C 71 % of classified Chapter . Traces of the Past Significant chronological indicators were not identified through Lilley’s analysis. Contrary to Egloff (1975), Lilley (1986: 210) states that there is no clear association between rim form and decoration. However, in addition to the decorative types described by Egloff (1975), Lilley also identified ‘obliquely-incised appliqué,’ although this was only identified on 1% of decorated sherds (Lilley 1986: 206). Attributes delineating time or place within the Madang style ceramics remain to be found. Chronologies To summarise the research presented above, Table 5.8 compiles the published radiocarbon dates in contexts associated with Madang style potsherds around New Guinea. Here, terrestrial dates are calibrated using IntCal13 (Reimer et al. 2013). A northern hemisphere curve is preferred as each site lies only about 4° to 6° south of the equator. Shell samples are calibrated using the Marine13 curve (Reimer Table 5.8. Published radiocarbon dates in association with excavated Madang sherds. Site Unit Layer Depth (cm) Lab # Material Conventional age Calibrated date (2σ) Reference Paligmete, Pililo Island, Arawe Islands, West New Britain province FNY 2 1 (base) 100 ANU-4982 Charcoal 900 ± 140 BP* 562–1173 cal BP Gosden and Webb 1994 Kove Islands, West New Britain province FPD 1 5 – Beta-26264 Charcoal 350 ± 60 BP 302–507 cal BP FPF 4 5 – Beta-26268 Shell 750 ± 50 BP 290–460 cal BP Summerhayes 2001a Arup/Long Island, Vitiaz Strait, Madang province JAB/B JCB JCC – – – ANU-1307 Charcoal 350 ± 70 BP 286–518 cal BP – Biliau beds – ANU-1308 Charcoal 1040 ± 80 BP 787–1174 cal BP – Mud flow – ANU-1309 Charcoal 470 ± 240 BP 0–900 cal BP Charcoal 360 ± 100 BP 0–622 cal BP Egloff and Specht 1982 Sigwa Island, Sio, Huon Peninsula, Morobe province KBP I K 75–80 NSW-86 Egloff and Specht 1982 Sio, Huon Peninsula, Morobe province 57 ANU-4970 Shell 940 ± 80 BP 423–665 cal BP 115 ANU-4332 Charcoal 670 ± 60 BP 540–699 cal BP 152 ANU-4330 Charcoal 340 ± 90 BP 0–537 cal BP 175 ANU-4329 Charcoal 300 ± 100 BP 0–525 cal BP 232 ANU-4606 Charcoal 510 ± 60 BP 466–654 cal BP 4 260 ANU-4607 Shell 1500 ± 70 BP 913–1216 cal BP 1 20 ANU-4335 Charcoal Modern – 2 62 ANU-4334 Charcoal 400 ± 90 BP 156–627 cal BP 99 ANU-4336 Charcoal 950 ± 70 BP 705–980 cal BP 3 109 ANU-4337 Charcoal 1290 ± 100 BP 980–1368 cal BP 123 ANU-4338 Charcoal 1160 ± 90 BP 930–1274 cal BP 2 I KBQ 3 II Lilley 1986 Malai Island, Siassi Islands, Vitiaz Strait, Morobe province TP – 35 ANU-3821 Shell 680 ± 70 BP 145–471 cal BP – 64 ANU-3820 Shell 800 ± 70 BP 291–529 cal BP – 75 ANU-3819 Shell 540 ± 70 BP 0–285 cal BP – 127 ANU-3822 Shell 680 ± 70 BP 145–471 cal BP – 155 ANU-3800 Shell 990 ± 70 BP 480–676 cal BP – 191 ANU-3801 Shell 740 ± 70 BP 264–496 cal BP – ANU-4344 Charcoal 180 ± 100 BP 0–437 cal BP 67 ANU-4341 Charcoal Modern – 75 ANU-4333 Charcoal Modern – KLJ I II 2 83 ANU-4342 Charcoal Modern – 121 ANU-4343 Charcoal Modern – 191 ANU-4345 Charcoal Modern – 232 ANU-4346 Charcoal 270 ± 170 BP 0–540 cal BP 1 17 ANU-4339 Charcoal Modern – 5 182 ANU-4340 Charcoal Modern – 770 ± 110 BP 551–920 cal BP Lilley 1986 Aibura cave, Central Highlands, Eastern Highlands province NAE VII 4 – GaK-622** Charcoal *The association between the basal date and the Madang pottery is not stated in detail. **Note problems associated with Gakushuin radiocarbon lab results (cf. Spriggs 1989). 72 White 1972       · .  et al. 2013) with a ΔR offset of 0.2 However, as ΔR values for the southwest Pacific are variable (Petchey et al. 2004), any irregularities in the marine reservoir between locations may skew interpretations of chronology. It is clear that most dates cluster within the last millennium before 2 Following Lilley & Specht 2007, in preference to the –400 as present, but several dates may extend the origins of Masuggested for the Huon Peninsula by Chappell & Polach 1991, dang style pottery to about 1300 cal. BP. Figure 5.6 presents working before the modeled -400 was built into calibration radiocarbon distributions based only on charcoal samples programs. Other reports on ΔR from modern (post 1950 AD) from Table 5.8 (to eliminate any concerns of variable and samples suggest wildly variable values from the Sepik north unknown ΔR values). This illustrates that although most coast to the Vitiaz Strait (McGregor et al. 2008; Petchey & Ulm dates fall within, or after, the 2σ range of dates from Tilu 2012); however, because the nature of these variations in the (GX3561, GX3633, and GX3632) in Madang Lagoon, some pre-colonial past are unknown, a ΔR of 0 is still the most con- contexts pre-date this, namely: JCB on Arop/Long Island, servative approach. and the lower deposits of KBQ site at Sio. OxCal v4.2.4 Bronk Ramsey (2013); r:5 IntCal13 atmospheric curve (Reimer et al. 2013) Arawes (FNY) ANU-4982 Kove (FPD) Beta-26264 Arop (JAB/B) ANU-1307 Arop (JCB) ANU-1308 Arop (JCC) ANU-1309 Sio (KBP) NSW-86 ANU-4332 ANU-4330 ANU-4329 ANU-4606 Sio (KBQ) ANU-4334 ANU-4336 ANU-4337 ANU-4338 ANU-4344 Siassi (KLJ) ANU-4346 Aibura (NAE) Gak-622 GX3561 Tilu (JCA) GX3633 GX3632 2500 2000 1500 1000 Calibrated date (calBP) 500 0 Figure 5.6. Radiocarbon distributions with 2σ confidence, based on wood charcoal in association with excavated Madang sherds. 73 Chapter . Traces of the Past Madang style ceramics occur in more places around New Guinea contemporaneous with habitation at Tilu. Modern or very recent dates at sites such as those at Sio and Siassi in association with abundant Madang sherds demonstrate that the movement of pots was accelerating in quantity and extent within the last few centuries before European contact on the coast. These pots probably fed into the already established Vitiaz Strait trade networks via the Rai Coast. Madang style pots, however, reached Arop/ Long in the Vitiaz Strait between 800 and 1200 years ago, and Sio on the Huon Peninsula between about 1000–1300 years ago. This hints at the trading of Madang ceramics significant distances east, prior to occupation at Tilu. Lilley (2004a) suggests that this is indicative of a divide between intensive production of recent Madang style pots in Astrolabe Bay, and less intensive production of Ancestral Madang style pottery somewhere in New Guinea, possibly at Madang, possibly elsewhere. The origins of Madang style ceramic production, however, do not appear to be older than 1300 years old. Summary This chapter has surveyed all available archaeological information regarding the pre-colonial ceramics of Madang. At present, the evidence suggests Bel ceramic production and exchange was occurring around the northeast coast prior to the first evidence for these ceramics around Madang. It also suggests that there was a recent upwelling of production and exchange intensity leading to the ethnographic period, but that the ceramic technology itself was subtly different from that observed around Madang today. Despite this research, the history of the Bel potters remains elusive and speculative. Do their origins lie on a now destroyed Yomba Island in the Vitiaz Arc, do they lie further east on the mainland, sharing decent with Type X pottery, or are they closer to modern Madang township representing a practice unbroken from the contemporary Yabob and Bilbil potters? The archaeological investigations presented in the following chapters will address these questions in an attempt to refine chronology and classification, and to fill these key gaps in our knowledge regarding Bel culture history, production, and exchange. 74 Chapter 6. Archaeological Investigations The island of Bilbil is full of stones. If you dug there for one day you would feel like dying. — Damun of Yabob (1977)1 Madang area survey Several island and coastal locations were first targeted for ground survey to assess the potential for surviving archaeTo build a picture of Madang ceramic technology in the ological deposits. This survey was selective and provisional pre-colonial past, new archaeological fieldwork was com- in nature, with locations being selected based on both acpleted around Madang District, shedding light on this cessibility and promising descriptions in site record forms much-neglected area of the New Guinea coast. The aims completed by Brian Egloff during his 1973 fieldwork. The of the fieldwork were to search for material relating to 1) areas chosen were Kranket Island, Siar Island, and Malmal the initial occupation by Austronesian-speaking peoples village in the Madang lagoon, and Yabob Island and Bilbil in the Late Holocene, and 2) changing production and Island south of Madang town (Fig. 6.1). All of the islands exchange in these ancestral societies, leading to the eth- visited comprise uplifted coral on their northeast flanks, nographic present. As has been introduced in Chapter with small sandy beaches on the southwest, protected 1, Madang is strategically located to assess pre-colonial from the wave pounding and abrasion of the Bismarck movements of peoples and ideas from west to east, along Sea. Furthermore, all of the areas surveyed were occupied a coastline where it is possible that Austronesian-speakers by speakers of Bel languages and considered themselves passed, during their movements some 3400–3300 years Bel people, with familial and cultural affinities to each of ago (Terrell & Welsh 1997; White 2014), and east to west the other sites. in subsequent journeys out of the Bismarck Archipelago along the northeast coast of New Guinea (Lilley 1999; Ross Kranket Island 1988). Furthermore, as the heart of Bel ceramic manufacture and ethnographic trading, coastal Madang is central Kranket Island (JLH), the largest in Madang harbour, lies to understanding how pre-colonial production and ex- 200 m northeast of Schering Peninsula and the modern change networks developed. Madang town (Fig. 6.2). The island is over 2 km long with a large saltwater lagoon dominating the centre. Sandy soil Because little was known of these movements and sub- overlies a thick coral bed, which is uplifted about 3.5 m sequent production and exchange along this part of New above high tide level. Arable land lies mostly on the southGuinea’s coastline, fresh archaeological fieldwork was ern and western portions of the island. Kranket is home considered a priority. This research represents the first ar- to seven clans living in two villages, most occupants of chaeological investigations along the Madang coast since which subsist on fishing or by working on the mainland, preliminary reports in the mid-1970s (Allen 1971; Egloff but some horticulture is also practiced where the land is 1975). Specifically, surface survey of several key ancestral good. The western side of the island was investigated on areas in the local Madang area was undertaken, followed foot and pottery surface collections taken from beside by systematic excavations at two clan areas: Tilu at Malmal a new school where upturned garden rubbish had been village within the Madang Lagoon, and Nunguri on Bilbil swept (E0368588, N9424500)2. Pottery was observed in Island, just south of Madang town. This chapter describes only a few other gardens, but according to the local people, in detail the procedures of this survey and excavation along with the results of radiocarbon determinations and 2 GPS coordinates given in text are as they relate to the AGD66 a broad summary of the material assemblages collected datum and the Madang topographic map (Ed. 1.AAS, Series to refine the culture history for the area. This provides the T683) published by the National Mapping Bureau 2002. Digiimportant archaeological context of the ceramic assemtalGlobe images retrieved from Google Earth display these blages used to explore production and exchange in this coordinates taken from a Garmin Montana 650t GPS recorder monograph. with a correction made for the WGS84 datum and entered under into Google Earth in UTM mode. 1 Quoted in Mennis 1981a: 28. 75 Chapter . Archaeological Investigations 145 O 50’ ON Malmal NG LA GO 5 O 10’ MA DA Siar Island Kranket Island Madang Yabob Island Bilbil Island 5 O 20’ BISMARCK SEA 145 O 50’ 0 5 MADANG PROVINCE km Figure 6.1. Archaeological places investigated during the 2014 field season. 76 Astrolabe Bay       · .  Siar Island Exposed coral Beach Kranket Island n oo g La Boat landings Exposed coral Surface collection New school site Schering Peninsula (Madang) 0 500 m Figure 6.2. Aerial photograph of Kranket Island (DigitalGlobe 2015). about 30–40 years ago the whole island was littered with broken pottery. Since then, the practice of sweeping rubbish out to sea and the disuse of pottery in the 1980s mean there are now only a few places where potsherds are still visible on the surface. The island was circled by boat but there were no extant shoreline deposits visible. Siar Island Siar Island (JAD) has strong familial connections to Kranket. The Siar and Kranket people both speak the Gedaged language of Bel, and Siar was formerly used as a burial site for all nearby islands. The island is approximately circular, just under 300 m wide and 400 m long. The southwestern side is dominated by a low-lying, sandy shoreline suitable 0 100 m Exposed coral Beach Surface collection Figure 6.3. Aerial photograph of Siar Island (DigitalGlobe 2015). 77 Chapter . Archaeological Investigations for canoe landings, while the eastern side is badly eroded by wave action, leaving only coral, uplifted several metres (Fig. 6.3). Pottery surface samples were collected near the sandy beach on the west side of the island (E0367454, N9426375) (Fig. 6.4a). The elders on the island still occasionally used pottery to cook with, but not often as metal has all but replaced clay. Possible mounded archaeological deposits were observed amongst the gardens in the interior of the island, but these were not extensively investigated. Yabob Island Yabob Island (Fig. 6.5), 800 m offshore and several kilometres south of Madang town was the home to the Yabob potters at European contact, but pottery is no longer made on the island. It is about 300 m wide and roughly circular. In the centre of the island there is an anthropogenic mound (JBG), approximately 3 m in diameter, where, according to oral traditions, the first ever pots were made by Honpain Figure 6.4. Survey of the Madang District, 2014: a) Siar Island beach site looking back towards Kranket Island in the south; b) a historical German coral wall next to where several lithic artefacts were found, Yabob Island; c) a selection of siliceous (foreground) and obsidian (background) lithic artefacts collected by local children on Yabob Island; d) a view from Tilu, Mound A, northeast towards Malmal Island; e) section of eroding midden, west side of Bilbil Island. Thick ashy soil overlies exposed coral bedrock; f) fresh water springs forming Jacob’s Well, Bilbil Island (Photos: D. Gaffney 2014). 78       · .  Inlet Coral wall Mound Exposed coral Beach 0 100 m Figure 6.5. Aerial photograph of Yabob Island (DigitalGlobe 2015). (see Chapter 3). A sample of pottery was collected midway up the mound from a pig rooting (E0365427, N9418754). At a sandy inlet on the north coast of the island (E0365443, N9418845) a sherd was collected, with a clay and design dissimilar to those at the mound along with two pieces of water-rolled obsidian, and a charcoal sample from a thin soil eroding out of a bank. Obsidian, slate, and a stone axe fragment were found near a small mound by an early twentieth century German coral wall in the centre of the island (E0365335, N9418754) (Fig. 6.4b). Numerous obsid- ian and other siliceous lithics, enthusiastically collected by the local children at the find locations noted, were also accepted as samples (Fig. 6.4c). The Yabob Islanders say they used to get their obsidian from West New Britain. Malmal village Malmal village is situated 12 km north of Madang township (Fig. 6.6). Tilu (JCA) clan area, previously excavated by Egloff (1975), consists of two elongated mounds ~20 m Malmal Island Malmal Udou Mound B Tilu Malmal village Mound A 0 200 m Figure 6.6. Aerial photograph of Malmal area showing Tilu site (DigitalGlobe 2015). 79 Chapter . Archaeological Investigations long, demarking the north and south of the hamlet. These two central mounds (A and B) remain a central and wellpreserved component of the site, which overlooks the Madang Lagoon to the north and east, and is landlocked to the south (Fig. 6.4d). A sandy canoe harbour lies about 20 m north of Mound A, allowing access from Malmal Island, several hundred metres to the northeast. Land access to the site is via a dirt road from the south, or through Malmal village proper, 200 m to the east. Given sea levels were around 3 m higher in the late Holocene (Tudhope et al. 2000), Tilu would have probably been an island during occupation about 550 BP (Egloff 1975). A sample of sherds was collected from the ground surface. Posthole backfill from recent construction of a structure, just north of Mound A, indicated potsherds, obsidian and fire ash were present to some considerable depth. At the time of survey, Tilu was used for coconut and betelnut production, with plantations around most of the area. Bilbil Island Bilbil Island (Fig. 6.7) is a 600 m long coral island, located 9 km south of Madang, and 1.8 km east of Bilbil potting village on the mainland. At the time of survey, the island was covered with vegetation including fruit trees and systematically planted coconut trees, making the interior of the island very cool and shaded. A small sandy beach on the southwest side of the island is the only point of canoe access. No cultural material was observed in this area; however, a modern fresh water well was present at the north end of the shore. An eroding midden about 50 m north of this beach had been observed by Mary Mennis and several Bilbil people many years earlier (Fig. 6.4e). The exposure that we observed was over 2 m high, with the top metre being a volcanically derived soil containing unidentified long bones and pottery. Below this soil was coral bedrock exposed through uplift and extensive wave action. In the lowest levels of the coral no in situ artefacts were observed, although a few water-rolled sherds were present, mixed with coral gravel, below the exposure. In the interior, several parallel mounds lay in the northern two thirds of the island suggesting the location of former clan areas. Just 20 m north of the eroding midden were two high linear mounds (1–2) on the western side of the island both about 1.5 m high and covered with potsherds and stone flaking debris. Further east, two other parallel mounds (3–4) were located. Pottery and lithic remains, including an axe fragment were found, some of which was brought to the surface by extensive crab bioturbation. At the south of the island lay a series of freshwater springs named ‘Jacob’s Well,’ used by the Bilbil Islanders during German occupation and possibly earlier (Fig. 6.4f). The Bilbil villagers unanimously claim that the high point on the island, just east of Jacob’s Well, is the former location of Miklouho-Maclay’s house from 1872. There were no material remains to suggest this was the case, but a stack of concrete bricks inscribed from 1910 was present at the top of the hill (E0364710, N9413987), probably the remnants of the rest house the German Navy built in that location. The Eroding midden Nunguri clan area Exposed coral Modern well Beach Historic German remains Jacob’s well 0 150 m Figure 6.7. Aerial photograph of Bilbil Island (DigitalGlobe 2015). 80       · .  inscr. ‘STrommye’? inscr. ‘1910’ inscr. ‘S M S’ 36 cm SIDE PLAN 23cm 17cm Figure 6.8. Inscribed German brick, highest point of Bilbil Island. inscription ‘SMS’ is likely the designation of an Imperial German Navy ship, Seiner Majestät Schiff meaning “His Majesty’s Ship” (Fig. 6.8; see Mennis 1981a: 38). obsidian from sources further afield. The single axe fragment collected at Yabob is small and does not preserve any diagnostic landmarks but is made from a black volcanic rock. The obsidian and other stone assemblages have been analysed separately at the University of Otago and the reSurface collections sults are reported in Gaffney and Summerhayes (2019). The main aim of the surface collection was to recover a The collection of surface finds during the survey was unsmall number of diagnostic potsherds from the broader systematic and does not accurately reflect the distribution study area to compare both in style and fabric with those of different artefact classes in the study area, but rather from excavated contexts to build a summary picture of gives an idea of locations likely to preserve deep chronoceramic variation around Madang. The artefacts collected logical sequences. Significant undisturbed deposits were from the surface survey were suggestive of pre-colonial only observed at three places: Yabob Island, Bilbil Island, habitation in each area (Table 6.1). This included decorated and at Tilu, Malmal Village. At each of these areas, large and undecorated pottery sherds, predominantly connect- mounds were present indicating the locations of previous ed to the Madang style (see Lilley 1986: 193), with red slip, sustained settlement. appliqué, and incised decoration, and everted rims. These pottery finds allude to the dominance of the style in local Excavation procedures production and exchange systems. The pottery is examined further in Chapters 7–9. Following the surface survey, two areas were selected for excavation owing to their potential for stratified arBoth obsidian and sedimentary flaked lithics were also chaeological deposits relating to the recent Madang cecollected and suggest the use of heterogeneous local and ramic style: Nunguri clan area (Mound 1) on Bilbil Island, regional rock types from the mainland, supplemented by a former place of pottery production, and Tilu clan area Table. 6.1. Summary of surface collections made during 2014 survey. Site code Place name JLH Kranket Island JAD Siar Island JBG JCA JBD Total Pottery Axe fragments Obsidian Non-obsidian lithics Total 10 0 0 0 10 6 0 0 0 6 Yabob Island 7 1 28 10 46 Tilu, Malmal 86 0 1 0 87 Bilbil Island 0 0 0 1 1 109 1 29 11 150 81 Chapter . Archaeological Investigations (Mound A) at Malmal village, the pottery consuming village where Egloff (1975) had previously dug. At each location the area was mapped and one 1 × 1 m unit was laid out along the cardinal axes and cleared of surface vegetation. A datum was set up nearby each unit. Excavations were placed at the top of the anthropogenic mounds in order to maximise the depth of archaeological deposit and potential chronological sequence. At both sites there were deep deposits with few visible stratigraphic boundaries, indicating a rapid accumulation of sediments. Arbitrary 10 cm and 20 cm excavation spits were excavated with hand trowels or shovels. Soil samples were collected for each spit. Where possible, charcoal was sampled in situ and without being handled, with depth measurements taken. All material was passed through 4 mm sieves and sorted into primary classes in the field before being bagged and shipped back to the University of Otago, Department of Archaeology and Anthropology. The material is currently stored in the department, on temporary loan from the National Museum and Art Gallery. All stone artefacts were bagged individually, while other artefact classes were bulk bagged. The excavation depths were limited at each site by safety concerns and the acceptance that the lowest deposits were culturally sterile. At both sites the local communities were actively involved in the fieldwork and discussions about the material findings. Ethnographic and historical information gathered from village elders during the course of the excavations allowed for more contextualised interpretations of these areas. Nunguri excavations Nunguri, Bilbil Island Nunguri clan area is centred around an anthropogenic mound on the western interior of Bilbil Island, just back from the eroding midden running along the western side of the island (E0364656, N9414189) (Fig. 6.9–6.10). This mound was selected for excavation in preference to other mounds on the island owing to the scarcity of tree roots and the proximity to the eroding section, which would give a good indication of subsurface deposits within the mound. The groundcover consisted of leaf litter from overhanging trees, but very little other vegetation. Nunguri (formerly Nigeri, Nguri, or Ngur) was the clan area of Kain, friend of Miklouho-Maclay (Mennis 1981b: 37). Kain and others were supposedly buried under their houses in this area, but many of the bones have been swept out to sea (Mennis 1981a: 54). Excavation Excavation of a 1 × 1 m unit (Test Pit 1) on Nunguri mound consisted of seventeen spits over two stratigraphic layers (Fig. 6.11–6.12), taken to a maximum depth of 2.53 m below surface. Spits 1–10 were controlled 10 cm units, while Spits 11–14 were increased to 20 cm using a shovel owing to time constraints, and Spits 15–17 were sectioned to 0.5 × 1 m in the south end of the unit. By Spit 16 at ~2.35 m below surface there was very little pottery present, some small shell, and no obsidian, and by Spit 17 there was no cultural material except a few tiny sherds that were probably intrusive Nunguri Beach Figure 6.9. Looking east at Bilbil with beach at south and Nunguri site in the centre (2014). 82       · .  4m cliff (coral) MN MN Mound 3 Mound 1 TP1 Mound 2 Covered vegetation 1 TP Open vegetation Mound 4 Track to beach 0 14m cliff (coral) 50 m Large hill 0 150 m Figure 6.10. Plan of excavations at Nunguri clan area, Bilbil Island (plan: G. Summerhayes; aerial photograph: B. Mennis). from above. Digging was stopped owing to safety concerns at depth in loose sediments and although bedrock was not reached, the deposit was convincingly sterile. through a very dark brown ashy soil (10YR 2/2), to a sandy light brown soil (2.5Y 3/2), mixing with Layer 2 below. Tree roots, coral gravel and boulders were common. From Spit 2–13 grey and white ash was mixed usually without definite feature boundaries, suggesting numerous single and Stratigraphy multiple use fires. This had disturbed the layer, especially Layer : is represented by Spits 1–15 and consists of a thick in the northwest corner of the unit, but this was difficult soil layer, transitioning, albeit without definite strati- to control for during excavation. A possible rubbish pit of graphic boundaries, from loamy black topsoil (5YR 2.5/1) loosely packed ashy soil (Feature 1) was present in the NE Figure 6.11. Excavations at Test Pit 1, Nunguri site, Bilbil Island (Photo: D. Gaffney 2014). 83 Chapter . Archaeological Investigations North wall D1 East wall South wall West wall Topsoil: 5YR 2.5/1 Black 1 2 Soil: 10YR 2/2 Very dark brown 3 4 5 6 Layer 1 7 8 9 10 11 12 Ashy soil: 5Y 8/1 13 14 Sandy soil: 2.5Y 3/2 Layer 2 15 Sand: 16 17 Soil Sand Fire ash Coral gravel # 1m Excavation spit Coral Potsherd Tree root Figure 6.12. Stratigraphy of Test Pit 1, Nunguri site, Bilbil Island corner of the unit in Spits 2–3, heavily mixed with pottery and coral gravel. Alternatively the presence of large, loosely distributed coral boulders between Spits 2–9 may suggest the edge of a deflated coral wall, similar to that observed on Yabob Island (see above), but the test pit was too small to ascertain if this was the case. From Spits 11–16, mixing of soil and sand suggests a transitional period between sandy beach and habitation deposit over many years. Chronology Six charcoal samples from Nunguri, Test Pit 1, were submitted to the Australian Institute of Nuclear Science and Engineering (AINSE), at Lucas Heights (Fink et al. 2004), for radiocarbon dating (Table 6.2; Fig. 6.13–6.14). Samples that had been collected in situ and with known depth measurements were selected preferentially. It was not possible to identify charcoal samples to species owing to a lack Layer : is pale yellow beach sand mixed with coral gravel, of local reference collections. Dates were calibrated usrepresented by Spits 16–17 (~2.15 m–2.5 m below surface). ing IntCal13, which has been used for samples from other This layer represents the original beach deposit prior to areas of New Guinea and the Western Pacific (Petchey et significant uplift and before significant habitation at the al. 2014; Shaw 2014). The radiocarbon dates suggest two mound. The base of excavation was laid with tarpaulin major phases of site activity spanning a period of about before the site was backfilled completely. 500–600 years. Table 6.2. Radiocarbon determinations from Test Pit 1, Nunguri site, Bilbil Island. Lab code Sample type Spit Depth below Datum 1 (cm) δ (¹³C) per mil Conventional radiocarbon age (BP) Calibrated range (2σ) (cal. BP)* OZS559 Wood charcoal 3 40–60 –26.2 ± 0.1 200 ± 25 0–299 OZS558 Wood charcoal 5 71 –26.5 ± 0.3 210 ± 25 0–304 OZS557 Seed charcoal 6 80–90 –30.9 ± 0.2 Modern — OZS556 Wood charcoal 11 131 –24.6 ± 0.1 495 ± 20 508–539 OZS555 Wood charcoal 13 170–190 –25.9 ± 0.1 530 ± 20 515–623 OZS554 Wood charcoal 14 207 –26.4 ± 0.1 540 ± 20 520–626 *Intcal13 atmospheric curve (Reimer et al. 2013); Oxcal 4.2.4 84       · .  Initial site occupation occurred around 500–600 years ago. Samples OZS556, OZS555, and OZS554 are statistically identical and suggest the earliest phase of occupation comprised permanent settlement and a rapid build-up of sediment. This occupation spanned at least Spits 11–14 from 1.21 m to 1.97 m below surface. The median age increases with depth, which may indicate that there is a stratigraphically secure sequence from Spit 11 down. The second phase, from at least Spit 1–6, dates to within the last few centuries. It is difficult to assess the true age of samples OZS558 and OZS559, which cover plateau regions of the calibration curve, extending the calendar age ranges (Blackwell et al. 2006; Lilley & Specht 2007). The ‘Modern’ result for OZS557 in Spit 6 refers to the fact that the corrected age is within the last 200 years. This sample has no inbuilt age, being from a carbonised seed pod (Canarium sp.), and the fact that the overlying samples OZS558 and OZS559 date to ≥200 BP suggest there is a small amount of inbuilt age on these wood charcoal fragments. Significantly, the upper part of the deposit is not older than about 300 years, while the lower portion is older than 500 years old. It is not clear if there was a break in settlement at around 400 years ago, but dating carbon from Spits 7–10 would resolve this question. Excavated material N S 200 ± 25 BP OZS558 71cm Spit 5 210 ± 25 BP OZS557 Modern 40–60cm Spit 3 80–90cm Spit 6 I OZS556 495 ± 20 BP OZS555 530 ± 20 BP 131cm Spit 11 170–190cm Spit 13 OZS554 207cm Spit 14 East face All material was individually washed, dried at 20°C, and sorted at the University of Otago. At Nunguri, pottery was the dominant artefact class (Table 6.3–6.4). Potsherds were divided into plain body sherds, decorated body sherds, rims, necks, and bases. All artefacts were re-bagged individually and catalogued, apart from plain body sherds, which were bulk bagged, counted and weighed. Midden shell, lithic flakes, animal bone, possible pottery anvils (identified as such by the Bilbil people assisting with excavation), pigments (used for decoration and marking measurements in producing canoes), shell beads, carved shell armbands, fire-cracked rocks (mumu3 stones), and OZS559 540 ± 20 BP II 1m Figure 6.13. Charcoal samples submitted to AINSE for dating, Nunguri, Bilbil Island. shell adzes with associated debitage, were also present 3 Mumu (Tok Pisin) are earth ovens dug into the ground and heated by oven stones. OxCal v4.2.4 Bronk Ramsey (2013); r:5 IntCal13 atmospheric curve (Reimer et al. 2013) OZS559 OZS558 OZS556 OZS555 OZS554 900 800 700 600 500 400 Calibrated date (calBP) 300 200 100 0 Figure 6.14. Plot of calibrated date distributions with 1σ, 2σ, and median age from Test Pit 1, Nunguri site, Bilbil Island (Oxcal 4.2). 85 3 4 225.46 5 161.25 6 60.50 105.00 7 8 9 10 79.25 90.25 80.00 94.15 11 12 13 14 180.60 116.00 173.85 140.20 15 2316.83 10750.85 10904.25 7702.74 2772.73 5310.61 3095.43 2943.49 2605.27 2674.76 4034.94 2861.64 3406.01 4208.35 2119.53 Pottery 16 17 Total 128.40 111.21 40.15 2060.76 40.93 10.61 67758.97 Obsidian 5.64 25.12 6.54 15.81 3.99 13.67 5.47 9.20 10.12 4.78 8.81 7.68 5.43 10.40 4.90 0.00 0.00 137.56 Non-obsidian lithic 0.34 24.95 10.85 9.63 14.07 11.41 34.34 1.89 23.60 2.85 12.23 1.77 6.11 180.23 0.00 0.00 0.00 334.27 Animal remains 0.00 16.50 9.14 79.90 13.80 11.73 5.49 6.74 14.94 14.34 12.06 1.51 1.25 3.97 0.00 0.00 0.00 191.37 Human remains 0.32 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.55 4.80 0.64 0.00 0.00 0.00 0.00 0.00 0.00 6.31 54.54 72.60 8284.19 188.05 940.80 887.15 718.30 248.40 493.80 474.40 409.85 398.15 367.95 785.55 476.75 720.90 598.20 448.80 Shell adzes Shell midden 0.00 0.00 53.15 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 5.95 0.00 0.00 0.00 59.10 Shell ornaments 0.00 1.49 0.00 2.79 5.27 0.00 2.33 2.46 0.00 0.36 4.71 1.35 2.06 0.00 0.00 0.00 0.00 22.82 Shell debitage* 4.05 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 4.05 Fire-cracked rocks 0.00 1056.54 101.68 38.25 384.39 615.81 0.00 0.00 221.12 26.43 428.59 0.00 76.58 44.94 0.00 0.00 0.00 2994.33 Pottery anvil (?) 0.00 222.18 0.00 0.00 0.00 202.90 0.00 0.00 0.00 0.00 0.00 0.00 322.11 0.00 0.00 0.00 928.98 0.00 107.42 0.00 0.00 0.00 14.06 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 121.48 Pigment 0.00 0.00 0.00 0.00 0.00 1.33 0.00 0.00 0.00 4.33 0.00 0.00 2.22 0.00 0.00 0.00 0.00 7.88 9 10 * Preliminary inspection- numbers expected to increase following thorough inspection of shell midden. Spit Pottery Obsidian 1 2 3 4 5 6 7 8 11 12 13 14 15 16 17 Total 1613 5493 4875 2755 1062 3465 1736 1938 1485 1580 2752 1795 2466 2412 1288 36 16 36,767 16 44 17 38 10 25 13 18 16 14 15 11 13 24 9 0 0 283 Non-obsidian lithics 2 10 4 6 7 6 2 2 6 3 5 2 6 6 0 0 0 67 Shell adzes 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 2 Shell ornaments 0 1 0 1 1 0 1 1 0 1 1 1 2 0 0 0 0 9 Shell debitage* 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 Fire-cracked rocks 0 10 3 2 2 7 0 0 3 2 2 0 1 3 0 0 0 35 Pottery anvils (?) 1 0 1 0 0 0 1 0 0 0 0 0 0 1 0 0 0 4 Grinders 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 2 Pigment 0 0 0 0 0 1 0 0 0 1 0 0 1 0 0 0 0 3 Total 1632 5560 4901 2802 1082 3505 1753 1959 1510 1601 2775 1809 2489 2447 1297 36 16 37,173 * Preliminary inspection- numbers expected to increase following thorough inspection of shell midden. Table 6.4. Material by number (n) recovered from Test Pit 1, Nunguri site, Bilbil Island. 181.79 Grinders Chapter . Archaeological Investigations 2 224.05 Table 6.3. Material by weight (g) recovered from Test Pit 1, Nunguri site, Bilbil Island. 1 50.44 As an indicator of environmental change around the site, the number of water-rolled potsherds increased gradually in Layer 1 down to Spit 15 and then dramatically in Layer 2 (Table 6.5, Fig. 6.19). This suggests that in the past the clan area was much closer to the sea, probably owing to uplift events within the last 500 years (Morgan et al. 2005). Spits 11–15, which date to the first phase of occupation, around 500–600 years ago, contained significantly more waterrolled sherds than Spits 1–6, which represent more recent habitation. Spits 16–17, consisting of mixed soil and beach sand, would presumably have originally been close to sea level, accounting for the high proportion of water-rolled sherds in the layer. The presence of minor water-rolled sherds in the upper deposit may be related to redeposition in the deposit or occasional significant wave action, given Nunguri’s proximity to the eroding cliff. 86 throughout Layer 1 (Fig. 6.15–6.16). Sparse human remains are present in Spit 1 and in Spits 9–11 (see Gaffney et al. 2018a). Layer 2 contained only small fragments of pottery and unmodified shell midden. Towards the lower spits of the deposit, the mass of pottery (g) gradually decreased, relative to the mass of sieved deposit (g), while the mass of shell midden (g) was constant (Fig. 6.17). Figure 6.18 shows that obsidian (g) and non-obsidian (g) flaked lithic material was also relatively consistent throughout Layer 1, although the small sample size and presence of a single large core in Spit 14 effects the variance in the non-obsidian line. Spit Sieved deposit (kg)       · .  a b c d e f g 0 10 cm h i Figure 6.15. Stone artefacts from Test Pit 1, Nunguri site, Bilbil Island. Pigment: a. N-5351-Spit 6; b. N-5353-Spit 13; c. N-5352Spit 10. Boring tool/grinder: d. N-5312-Spit 6; Grindstone: e. N-5354-Spit 2. Pottery anvils (?): f. N-5359-Spit 14; g. N-5357-Spit 3 (broken). Fire-cracked rock: h. N-5367-Spit 2; i. N-5375-Spit 5. 87 Chapter . Archaeological Investigations a b c d e 0 10 cm f g h i j Figure 6.16. Shell artefacts from Test Pit 1, Nunguri site, Bilbil Island. Shell adze fragments: a. N-5504-Spit 3 (blade end); b. N-5517-Spit 14 (butt end). Shell beads: c. N-5516a-Spit 13; d. 5516b-Spit 13; e. N-5513-Spit 10. Shell armbands: f. N-5507-Spit 5; g. N-5512-Spit 8; h. N-5502-Spit 2; i. N-5506-Spit 4; j. N-5511-Spit 7. 88       · .  0.06 Pottery Obsidian 0.05 Density of artefacts Lithics Shell midden 0.04 Shell adzes 0.03 Shell ornaments Shell debitage 0.02 Fire-cracked rocks Pottery anvil (?) 0.01 Grinders 0 1 2 3 4 5 6 7 8 Pigment 9 10 11 12 13 14 15 16 17 Spit Figure 6.17. Density of artefact types to sieved deposit (g), Test Pit 1, Nunguri site, Bilbil Is. 0.0014 Density of artefacts 0.0012 0.001 0.0008 Obsidian 0.0006 Lithics 0.0004 0.0002 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Spit Percentage water-rolled Figure 6.18. Density of obsidian and non-obsidian lithics to sieved deposit (g), Test Pit 1, Nunguri site, Bilbil Island. 20 18 16 14 12 10 8 6 4 2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Spit Figure 6.19. Water-rolled pottery sherds, Test Pit 1, Nunguri site, Bilbil Island. 89 16 17 Chapter . Archaeological Investigations Table. 6.5. Water-rolled pottery sherds, Test Pit 1, Nunguri site, Bilbil Island. Spit 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Total Water-rolled potsherds (n) 0 16 21 15 8 39 12 27 22 0 93 55 51 45 50 4 3 461 % of total sherds 0.00 0.29 0.43 0.54 0.75 1.13 0.69 1.39 1.48 0.00 3.38 3.06 2.07 1.87 3.88 11.11 18.75 1.25 Tilu excavations Tilu, Malmal village Excavation was undertaken at Tilu (JCA), Malmal Village, in an attempt to clarify Egloff ’s original dating for the site and to produce a finer resolution examination of material culture change through time. This change would then be compared to the sequences present at Nunguri, Bilbil Island. Mound A was selected as the focus for excavation owing to the large quantity of surface artefacts in the area (Fig. 6.20–6.21). Excavation A 1 × 1 m excavation (Unit 1) was undertaken at Mound A, 2.88 m above the high tide mark (E0367062, N9433104) and was dug twelve spits to a maximum depth of 2.5 m below surface, where the water table was reached and coral bedrock began to inhibit excavation, indicating the level of the natural coral uplift (Fig. 6.20a-c). Spits 1–8 began as 10 cm thick contexts but were increased to 20 cm from Spit 9. Owing to time constraints, Spit 12 was a ~100 cm thick and almost entirely culturally sterile unit. Two dis- Figure 6.20. Excavations at Tilu site, 2014: a) A view south towards Mound A, Tilu clan area, Malmal village (photo: G. Summerhayes 2014); b) Unit 1 excavations, Mound A, Tilu (photo: G. Summerhayes 2014); c) Teppsy Beni and Feli excavate Unit 1 at Tilu (photo: D. Gaffney 2014); d) Elias (centre) from Tilu clan area digs Shovel Pit 1, while Herman Mandui (left) records (photo: D. Gaffney 2014). 90       · .  0 tinct stratigraphic layers were identified in the excavation, very similar to those at Nunguri. Downslope from Mound A, a separate 1 × 1 m shovel pit (Shovel Pit 1) was dug unsystematically with a crow bar/metal digging stick to assess the horizontal extent of the deposit (Fig. 6.20d). This pit was dug to about 120 cm deep, where the water table was reached. The same two major layers of stratigraphy were observed in Shovel Pit 1 as in Unit 1. Collections made from this shovel pit were limited to diagnostic potsherds and lithic artefacts, hand picked from unsieved deposit. MN 50 m Mound B Stratigraphy Layer : comprised Unit 1, Spits 1–10, and was ~130 cm thick, with a gradual change from loamy dark grey soil (5Y 3/1) to ashy dark grey soil (5Y 4/1, 5Y 2.5/2) but without clear divisions in stratigraphy (Fig. 6.22–6.23). The layer contained coral boulders and gravel and was interwoven with roots from nearby trees. Feature 1 (Spits 5–6) was a scatter of charcoal, oven stones, and burnt pottery with undefined boundaries, about 0.6 m below surface, indicated a fire. A thin and undefined grey ashy lens (10YR 6/1) was present (Spits 9–10) 1.1 m below surface and seems to represent a single use fire. Pottery and shell midden were abundant in Layer 1, along with substantial amounts of Shovel Pit Test Pit 1 Mound A Figure 6.21. Plan of excavations at Tilu clan area, Malmal village (G. Summerhayes 2014). East wall D1 South wall 1 Topsoil: 5Y 3/1 2 3 Soil: 5Y 4/1 4 Layer 1 5 Soil: 5Y 2.5/2 6 7 8 9 Ash lens: 10YR 6/1 10 Transitional sandy soil: 10YR 4/1 Layer 2 11 12 Sand: 2.5Y 7/4 1m Soil # Sand Excavation spit Fire ash Coral Potsherd Coral gravel Tree root Shell Figure 6.22. Stratigraphy of Unit 1, Tilu site, Malmal village. 91 Chapter . Archaeological Investigations North wall MN ~130cm Brown soil ~110cm White sand Coral bedrock 1m Figure 6.23. North wall of completed Unit 1, Tilu site, Malmal village. obsidian. In contrast to Nunguri on Bilbil Island, almost no non-obsidian stone flaking debris was recovered at Tilu. Disarticulated human and animal bone was present but uncommon. A small amount of mixing between layers was present in Layer 1 from about 1.1 m below surface but otherwise the layers were distinctly separated. Layer : comprised of Spits 11–12, from about 1.3 m below surface to base of excavations, was a pale yellow sand (2.5Y 7/4), representing the former beach, assuming increased relative sea levels in the past owing to uplift. Basal deposits were culturally sterile with no significant indicators of habitation. The quantity of pottery drastically reduced in Layer 2, no obsidian was found, and most of the shell was non-cultural, as indicated by several large bivalves still in articulation. Pottery was not present below about 1.7 m below surface. N S OZS552 630 ± 30 BP OZS551 78 cm Spit 6 595 ± 20 BP OZS550 585 ± 20 BP 37cm Spit 4 87 cm Spit 7 I OZS549 925 ± 20 BP OZS548 675 ± 20 BP 120–140 cm Spit 10 140–150 cm Spit 11 II Chronology OZS547/OZS546 580 ± 20 BP Seven wood charcoal samples from Tilu, Unit 1, were sub150–250 cm Spit 12 570 ± 20 BP mitted to AINSE for radiocarbon dating (Fig. 6.24–6.25; Table 6.6). The same sampling procedures were applied as for the Nunguri samples. The dates provide a robust chronology in order to interpret site occupation. Samples OZS552, OZS551, OZS550, OZS548, OZS547, and OZS546 are statistically identical and suggest a single phase of site ocEast face cupation about 550–650 years ago, with rapid sediment build up. This is consistent with the three dates already 1m published for the site (GX3561, GX3633, and GX3632), but the new dates provide much more precision in error rang- Figure 6.24. Charcoal samples submitted to AINSE for dating, es. Occupation at Tilu is contemporary with, or slightly Tilu, Malmal village. 92       · .  OxCal v4.2.4 Bronk Ramsey (2013); r:5 IntCal13 atmospheric curve (Reimer et al. 2013) OZS552 OZS551 OZS550 OZS549 OZS548 OZS547 OZS546 1100 1000 900 800 700 Calibrated date (calBP) 600 500 Figure 6.25. Plot of calibrated date distributions with 1σ, 2σ, and median age from Unit 1, Tilu, Malmal village (Oxcal 4.2). Table 6.6. Radiocarbon determinations from Unit 1, Tilu site, Malmal village. Lab code Sample type Spit Depth below Datum 1 (cm) δ (¹³C) per mil Conventional radiocarbon age (BP) Calibrated range (2σ) (cal. BP)* OZS552 OZS551 Wood charcoal 4 37 –28.1 ± 0.1 630 ± 30 552–664 Wood charcoal 6 78 –25.9 ± 0.1 595 ± 20 543–647 OZS550 Wood charcoal 7 87 –26.5 ± 0.1 585 ± 20 540–644 OZS549 Wood charcoal 10 120–140 –24.1 ± 0.1 925 ± 20 791–915 OZS548 Wood charcoal 11 140–150 –27.7 ± 0.1 675 ± 20 564–674 OZS547 Wood charcoal 12 150–250 –24.5 ± 0.1 580 ± 20 538–641 OZS546 Wood charcoal 12 150–250 –24.4 ± 0.1 570 ± 20 535–637 * Intcal13 atmospheric curve (Reimer et al. 2013); Oxcal 4.2.4 earlier than, the first occupation phase at Nunguri. The slightly older date, OZS549, is anomalous given it is bracketed by 550–650 year old dates in what was observed to be an undisturbed part of the deposit. It is possible that the older date represents inbuilt age on the wood or an earlier occupation in the area that has been redeposited. Through the spits, pottery (g) steadily became less frequent, relative to sieved deposit (g), while shell midden (g) became much more dense relative to deposit (g) from Spits 7–9, becoming sparse in Layer 2, Spits 11–12 (Fig. 6.26). This demonstrates that the majority of material disposed in the area was shellfish food waste. Excavated material Stone material was predominantly fire-cracked rock and obsidian flaked artefacts. Down to Spit 10, fire-cracked rock, the result of oven stones being used to heat mumu, were made of water-rolled basalt cobbles with large pyroxene phenocryst inclusions, indicating they were likely sourced from local streams. Today, mumu stones are collected from the Tagog stream nearby to the site. There were no possible pottery anvils recovered, perhaps suggesting that Tilu was at no point a local production centre like Bilbil. Obsidian at Tilu (mean weight = 0.37g) was generally smaller than those collected at Nunguri (mean weight = 0.54g), and much fewer in number. All of these obsidian pieces originally derive from the Talasea source in West New Britiain (see Gaffney and Summerhayes 2018). Sedimentary flaked lithics, like those at Nunguri, were rare. All material was cleaned and catalogued in the same manner as the Nunguri material. At Tilu Unit 1, the dominant material recovered by weight was midden shell (Table 6.7). Pottery was the most significant artefact by number and was of a similar style to that on Bilbil Island (Table 6.8). However, the mean weight of potsherds relative to sieved deposit was much less at Tilu, (=0.01 kg of ceramic per 1 kg of sediment), compared to Nunguri (=0.03 kg ceramic per 1 kg of sediment). Obsidian and sparse sedimentary lithics, human and animal bone, and shell artefacts were present in minor amounts throughout Layer 1. In Layer 2 the only artefacts present were very fragmentary potsherds and a single drilled shell ornament. In contrast to Nunguri, only two potsherds from Tilu showed evidence of water-rolling, suggesting the site was less affected by the Animal bone recovered from Layer 1 includes pig and posnearby shoreline. sibly dog remains with clear cutmarks suggestive of butch93 Chapter . Archaeological Investigations Table 6.7. Material by weight (g) recovered from Unit 1, Tilu site, Malmal village. Spit Sieved deposit (kg) Pottery 1 2 3 4 5 6 7 8 9 10 11 12 119.00 98.00 108.00 117.00 75.50 116.00 128.00 88.00 146.00 117.00 84.00 594.00 1791.00 Total 3088.01 2568.00 2506.29 1928.61 1146.29 2075.38 1634.58 867.22 1878.67 657.27 56.58 19.45 18426.35 Obsidian 6.99 3.00 2.49 2.68 2.87 4.17 2.73 1.74 2.21 2.40 0.00 0.00 31.28 Non-obsidian lithic 0.00 4.93 19.87 0.00 0.00 0.00 0.00 0.19 0.61 0.00 0.00 0.00 25.60 Animal remains 14.28 59.36 34.76 9.39 12.50 23.07 6.76 8.14 51.87 16.51 17.60 0.00 254.24 Human remains 0.00 0.00 35.90 5.28 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 41.18 25384.85 1682.20 1569.25 1726.55 2326.77 1429.53 1657.45 4578.00 2221.30 6255.85 1403.57 275.98 258.40 Shell adzes Shell midden 0.00 0.00 0.00 7.06 0.00 0.00 0.00 0.00 37.32 0.00 0.00 0.00 44.38 Shell ornaments 0.00 22.79 0.00 16.54 0.00 0.18 0.00 4.73 0.00 0.00 0.00 0.08 44.32 Shell debitage* Fire-cracked rocks 0.00 0.00 0.00 0.00 0.00 0.00 0.00 7.22 0.00 0.00 0.00 0.00 7.22 26.33 57.70 14.57 3.03 163.92 0.00 733.26 15.27 268.20 226.23 0.00 0.00 1508.51 * Preliminary inspection- numbers expected to increase following thorough inspection of shell midden. Table 6.8. Material by number (n) recovered from Unit 1, Tilu site, Malmal village. Spit Pottery 1 2 3 4 5 6 7 8 9 10 11 12 Total 1568 1255 1173 1126 764 1377 785 470 819 352 40 21 9750 24 8 5 9 8 17 4 4 7 5 0 0 91 Obsidian Non-obsidian lithics 0 1 1 0 0 0 0 1 2 0 0 0 5 Shell adzes 0 0 0 1 0 0 0 0 1 0 0 0 2 Shell ornaments 0 1 0 3 0 1 0 1 0 0 0 1 7 Shell debitage* 0 0 0 0 0 0 0 1 0 0 0 0 1 Fire-cracked rocks Total 4 5 1 1 3 0 4 2 6 1 0 0 27 1596 1270 1180 1140 775 1395 793 479 835 358 40 22 9883 * Preliminary inspection – numbers expected to increase following thorough inspection of shell midden. 0.045 0.04 Density of artefacts 0.035 Pottery 0.03 Obsidian Lithics 0.025 Shell midden 0.02 Shell adzes Shell ornaments 0.015 Shell debitage 0.01 Fire-cracked rocks 0.005 0 1 2 3 4 5 6 7 8 9 10 11 12 Spit Figure 6.26. Density of artefact types to sieved deposit (g), Unit 1, Tilu site, Malmal village. ery, as well as reef fish, birds, and reptiles (see Gaffney et al. 2020). A similar range of shell midden and artefacts was present at Tilu as at Nunguri, including shell adzes and ornaments (Fig. 6.27). Human remains from Tilu were rare and include two mandibular fragments from Spit 3 and two disarticulated teeth from Spit 4 (see Gaffney et al. 2018a). 94       · .  a b d c 0 5 cm e f g h Figure 6.27. Shell artefacts from Unit 1, Tilu site, Malmal village. Shell adzes: a. T-1509-Spit 4 (butt end); b. T-1528-Spit 9. Shell core debitage: c. T-1523-Spit 8. Shell beads: d. T-1533-Spit 12; e. T-1513a-Spit 4; f. T-1513b-Spit 4; g. T-1512-Spit 4; h. T-1524-Spit 8. Potsherds and fire-cracked rocks were the only material types recovered from Tilu Shovel Pit 1 (Table 6.9). The pottery recovered was identical to that excavated at Unit 1 apart from a unique ring base collected 60 cm below the surface. Table 6.9. Material by number (n) and weight (g) recovered from Shovel Pit 1, Tilu site, Malmal village. Depth below surface (cm) Potsherds (n) Weight (g) Fire-cracked rock (n) 45 22 362.16 0 0.00 Summary 60 18 344.31 1 126.91 70 9 34.13 1 284.60 The initial fieldwork aims were partially addressed by the 2014 investigations. No evidence was located for Lapita age settlements around the Madang coast that could shed light on the directions of migration around New Guinea 80 1 78.07 0 0.00 95 Weight (g) 120 6 31.85 0 0.00 Total 56 850.52 2 411.51 Chapter 6. Archaeological Investigations by early Austronesian-speaking potters. Neither was there any evidence for immediately post-Lapita occupation, even directly after 1300 BP, which Lilley and Specht (2007) identify as the date that Ancestral Madang ceramics materialise in the Vitiaz Strait. As people have been active on the northeast coast since the Late Pleistocene (Groube et al. 1986), this lack of older deposits on the coast is likely owed to the dynamic physical landscape (see Chapter 2). The Madang coast would be unlikely to preserve places relating to early occupation, especially within the lagoon area and offshore islands, which are only 3000 years old at the earliest. Subsequent disturbance to deposits through uplift, earthquakes, or tsunami would also bias the distribution of sites along the coast. Lilley (1986: 31), for example, notes that the Ritter Island tsunami disturbed coastlines on north and east Umboi Island so much that it was often not possible to locate prehistoric occupation areas. The ceramic styles observed both in survey and excavations around Madang were almost exclusively red-slipped, paddle and anvil made, manually tempered wares that appear to be directly ancestral to the modern Madang style potteries produced at Bilbil and Yabob. Excavations carried out at both Tilu and Nunguri recovered sufficient artefactual evidence in stratigraphic position to construct a concise and fine resolution sequence for the Madang area in the last half millennium before present. Radiocarbon dating at both sites revealed occupation spanning the last 550–650 years, associated with a full ‘Austronesian-style’ assemblage of pottery, obsidian flaked stone, polished axe/ adzes, shell adzes, shell ornaments, and pig and dog bone. The following chapters present the analyses carried out on pottery recovered from the 2014 surface survey and excavations to investigate changes to ceramic production and exchange leading up to the ethnographic present. 96 Chapter 7. Pre-Colonial Potting I: Production Knowledge is only rumour until it lives in the muscle. – New Guinea proverb The next three chapters describe the analysis of Madang ceramics and the pre-colonial chaîne opératoire, including production technology (Chapter 7), procurement and distribution (Chapter 8), and decorating procedures (Chapter 9). These chapters attempt to systematically classify the pottery assemblages presented in Chapter 6, whilst addressing important research aims at each step. This chapter is designed to assess the forming stage undertaken by precolonial Madang potters and has two specific aims. The first is to describe the complete diversity in pre-colonial forming technologies, which gives clues to the number of production groups operating in the past, and if they were all working within a broader community of practice (i.e. producing the same types of pot using the same techniques). Second, the study seeks to describe minor variations to pottery forming over time, which may give insight into the nature of innovation and interaction between production groups. From a culture history perspective, particular variations in the forming technology may also be limited to specific timeframes and so be useful as chrono-stratigraphic markers. and surface survey is treated separately to describe ceramic technology on-site. These results will then be synthesised in the discussion of production and exchange in Chapter 10. Methodological issues Style and production groups The concept of style has dominated New Guinea ceramic analyses, echoing the style–function debates of the latter 20th century (see Plog 1995; Pollock 1983; Sackett 1977; Wobst 1977), and inspired by studies on highly decorative Lapita pots (e.g. Anson 1986). Such stylistic approaches have been popular in Near Oceania owing to the centrality of prehistoric exchange, group mobility, and interaction in explaining culture change (e.g. Cochrane & Lipo 2010; Green 1996, Irwin & Holdaway 1996; Lilley 2004a, 2004b; Summerhayes 2001b; Terrell & Welsch 1990). The definition of style in New Guinea has largely been derived from Specht (1969: 64), who suggested it is the intuitive ‘content and appearance’ of a pottery industry or industries distributed across time and space. Lilley (1986: 160) takes this further and notes that intra-community variation in style will be negligible compared to inter-community variThese specific aims will contribute to addressing the ations. This is important because the ‘style-groups’ of each broader research objectives presented in Chapter 1. For production community are then thought to be so intuiinstance, if there was a recent migration of Bel potters to tively distinct that they could not be mistaken for another, Madang, entering into an existing manufacturing centre, either by an archaeologist or an indigenous consumer. we might expect to see abrupt changes to some aspects For instance, in ‘Madang style’ ceramics, the most obvious of the production sequence as the newcomers introduced technical element is the brilliant red slipping, which imtheir own techniques of forming vessels. However, if the mediately distinguishes it as ‘Madang.’ In short, the more earliest pottery around Madang represents the initial communicative aspects of technology such as surface arrival of ‘Yombans,’ we would expect pottery to arrive decoration, shape, and visual style are assumed to give around Madang suddenly, followed by gradual innova- insights into the ethno-linguistic group identities of the tions over time. producers. The implicit assumption here is that intuitive styles can be matched to specific communities of practice. To address these aims, firstly, methodological issues pertaining to ceramic classification are discussed and the The concept of isochrestic style can elucidate these basic technological approach to ceramics is described. The assumptions. Isochrestic style is the range of shapes and analytical procedures of the Madang formal analysis are forms which could be used, with near-equal efficacy, for a then presented, including the nature of classification and specific function (Sackett 1986). For instance, a cooking attribute analysis. Lastly, the results of these formal analy- pot may be globular like that of the Bel or long and tuses are presented. Pottery from each excavated assemblage bular like those of inland Madang groups (see Chapter 3). 97 Chapter . Pre-Colonial Potting I: Production Within a community, potters will be aware of only a few isochrestic styles and may choose to produce even fewer, largely dictated by the technological practices they work within (Sackett 1990). Lechtman (1977) notes that it is then the techniques used to produce decoration and form that are stylistic as opposed to the object itself. In this way, the individual’s decisions during manufacture can be thought to have style, rooted in social practice and in technological choice (Gosselain 1992; Lemonnier 1992: 90; Latour & Lemonnier 1994). So, two pots could be remarkably similar in form, as many recent New Guinea ceramics are, but represent distinct isochrestic variation in technological choice. These technological choices are not necessarily discursive or even reflected upon by the individual but represent the specific technological trajectories of different communities among a myriad of possible options (Lemonnier 1993, 2013; Levi-Strauss 1960: 16). Often these choices are arbitrary or even illogical to outsiders, but are influenced by historically contingent social preferences, beliefs, representations, aesthetics, and ideas (Dobres & Hoffman 1994). This is to say that style is technological; technology is stylistic. lenback & Schiffer 2010; Schiffer 1995; Skibo & Schiffer 2008). The conceptual advantage of the chaîne opératoire, however, is that it preferences technological and social processes over transformations to the artefact (Knappett 2012; Lemonnier 1992). Objects are very rarely intended to become socially inert at the end of the production process, nor does technology, as a mode of engagement between people and things, cease to acquire or create meaning during distribution, use, and discard (Pelegrin 1990; Sellet 1993; Sillar & Tite 2000). A consumer of pottery, for instance, engages embodied knowledge during cooking or while storing the pot in such a way that it will not break. As such, these aspects of material engagement can be viewed as necessary and meaningful links in technological processes. For this reason, more pragmatic approaches have expanded the chaîne opératoire beyond the realms of making things to include using, unmaking, and remaking (Naji 2009). Although technological processes represent a continuum, for analytical purposes the chaîne opératoire can then be subdivided into phases of procurement, production, distribution, consumption, alteration, re-use, and discard (Fig. 7.1; Gaffney 2019). Technology as methodology The phases of a chaîne opératoire can be deconstructed further. Here, a ‘technical element’ is the smallest analytical unit of study, formed by the interplay between a gesture, an intention, and matter (modified from Lemonnier 1992: 31). As technical elements are repeated, through habitual material engagement, the boundaries between gesture, intentionality, and matter become blurred, creating embodied knowledge (see Malafouris 2008). The chaîne opératoire is an arrangement of technical elements in order. The community specific sequences of these elements are known as ‘technical syntaxes,’ which are, of course, regulated by broader social and technological conventions. These distinctions are important because technical elements are more fluid in terms of sharing (i.e. transmission) and are often passed horizontally within or between communities of practice, while syntaxes are more conservative and are likely to be passed from parent to child, teacher to pupil (Apel 2008; Apel & Darmark 2007). Because decorative elements or minor technical variations are less reliant on extended syntaxes, they are more readily shared across social boundaries than vessel forming techniques (Arnold 1985, 1998; Mayor 2005). Thus, individual elements may be widely distributed, while syntaxes are usually specific to production groups. We now return more explicitly to the concepts of embodied knowledge and technological process to clarify how archaeological observations can be used to investigate precolonial production. As outlined in Chapter 1, technology is a dynamic process. Technology materialises, and is the materialisation of, embodied knowledge; it comprises the processes operating between the boundaries of enculturated habit and the technical choices of individuals and social groups (Lemonnier 1992; Pigeot 1990). Therefore, technology – a study of techniques – can be used as the methodological link to investigate everyday routine and human enculturation into specific communities of practice (Knappett 2005a, 2011: 98). The methodological apparatus developed by the French school to investigate technological choice throughout the production process is known as the chaîne opératoire (Haudricourt 1968; Leroi-Gourhan 1943, 1945). A chaîne opératoire describes the sequence of operations, or techniques, which brings a raw material from an unmodified state to a manufactured state (Cresswell 1976: 6), or, more broadly, the series of operations involved in any transformation of matter by (usually human) beings (Lemmonier 1992). Individual chaînes opératoires act as idiosyncratic strands of broader technological processes, specific to communities. They emphasise every stage in an artefact’s production, extending archaeological enquiry from the single object to longer processes, both on the bodily time scale and over the longue durée, enhancing the archaeologist’s ability to interpret past technological practices and how they have changed over time (Martinon Torres 2002). The chaîne opératoire is complementary to the North American behavioural chain, which focuses on not only production, but the entire life history of an artefact (Hol98 Variation within technical elements illustrates the range of what constitutes socially appropriate forms within the community of practice (Lemonnier 1983: 17; Mahias 1993). By examining specific variations in the technical syntax, however, different social/production groups may be identified (Roux & Courty 2005). On the other hand, elements in the syntax that are invariant may be seen as necessary and cannot be modified without jeopardising the chain for either social or physical reasons (van der Leeuw 1993).       · .  technological process individual chaîne opératoire Procurement Clay extraction technical element Collection of temper Transport of raw materials Storage of raw materials intention gesture matter Production Mixing of raw materials Vessel forming engagement Surface treatment Firing embodied knowledge Consumption Distribution Finishing Storage of finished vessel Transport of finished vessel Trade/exchange of vessel Mixing of foodstuffs Cooking/heating Cleaning Storage Discard Re-use Alteration Damage technical syntax Reworking Addition of new material Secondary use Storage Breakage Disposal Figure 7.1. Analytical terminology of a ceramic chaîne opératoire (after Gaffney 2019). 99 Chapter . Pre-Colonial Potting I: Production Approaches to New Guinea pottery classification tions rather than an objective reality (Adams & Adams 2007: 5; Bedford 2006a: 73; Hill & Evans 1972). In such a way, Lilley’s (1986) classification of Madang ceramics was appropriately used as an analytical device to order large amounts of information, rather than an attempt to reconstruct emic type-systems (see Chapter 5). Further, in no way are such analyses free from value judgments or subjectivity (Keighley 1973; contra Sokal & Sneath 1963 [influential for Egloff 1979 and Irwin 1985]). However, the clear explication of how attributes are measured and described does much to alleviate concerns of intra-observer variation and makes inter-observer comparisons more accessible and accurate (Whittaker et al. 1998). Conventionally, the study of archaeological ceramics in New Guinea has attempted formal classification. Such approaches stem, particularly, from foundational North American classification systems (Gifford 1960; Rouse 1951, 1960, 1965; Shepard 1963, 1965; Willey 1945), which advocated the use of polythetic attribute analyses to group artefacts. Ceramic attributes, first used in the New Guinea area by Specht (1969) and described as the smallest unit of analysis that an artefact can be reduced to (e.g. lip profile, rim course, orifice diameter), have become fundamental to pottery classification. This emphasis on classification derives from the real need to establish historical sequences in new areas of archaeological research. Without such re- Technological classification gional sequences, based on pottery attributes and absolute dating, interregional comparisons would be difficult. The methods behind New Guinea ceramic classification certainly do not need to be reinvented, but methodologiThe analytical procedures behind classification have been cal justifications linking higher-order theory and low-level variable and assemblage-specific. However, the practice attributes are here repositioned to focus on the technoof constructing categories to sort artefacts into, creating logical (and social) processes of production as opposed mutually exclusive units for further analysis and interpre- to style and classic typology. As Pétrequin and Pétrequin tation, has been similar. In this volume, the term ‘classifi- (1999) point out, examining the subtleties of pottery techcation’ denotes the most general kind of categorisation of nology delineate past production communities in a more artefacts into matching sets, with the possibility of contrast systematic manner than stylistic approaches to shape and and similarity to other sets, while ‘typology’ is a specific decoration. These archaeologists are the only ones in New kind of classification designed to sort limited groups of Guinea to attempt a typology of ceramic technological artefacts into mutually exclusive categories (following traditions, comparing ethnographic and archaeological Adams 1988; Adams & Adams 2007). In this sense, nearly chaînes opératoires around the New Guinea region to exall ‘classifications’ in circum-New Guinea ceramic analysis, amine broad movements and interactions of past comvariously referred to as types, styles, classes, groups, forms, munities. The Pétrequins’ typology is hierarchical, sorting and wares, are typological in some way. ceramic traditions with increasing levels of exclusiveness based upon significant technical elements and their synOften, intuitive typological sorting allows an assemblage taxes in the technological process. to be subdivided into more manageable units before further sorting or classifying based on specific attributes Consequently, social approaches to technology are not (see Bulmer 1978; Lilley 1986; Terrell & Schechter 2011). necessarily divorced from typology, but the conceptual Other times, types will first be defined based on specific procedures behind classification are different. To identify geometries and rim attributes, before classification and a chaîne opératoire through pottery artefacts, a hierarchicorrelation with further attributes such as decoration, cal typology based on three successive strategies of eight metric data, or fabric groups (see Garling 2007; Hogg steps is required (Fig. 7.2). Firstly, formal classification 2011; Kaplan 1976; Lauer 1970a; Rhoads 1980; Skelly 2014; distinguishes clusters of attributes pertaining specifically Specht 1969; Summerhayes 1996, 2000; Vanderwal 1973; to vessel forming to demonstrate elements and syntaxes Wickler 2001). Others again have hierarchically sorted ar- diagnostic of different chaîne opératoire and technological tefacts into more or less inclusive units of analysis based processes. This establishes technical classes and their varion specific attributes (Egloff 1971, 1979; Irwin 1977, 1985). ants based on production techniques, which should deHierarchical classifications (or ‘taxonomies’) can then lineate specific communities of practice. Secondly, sherds be arranged by cluster/network analyses to form types belonging to different technical classes/variants are then (Renfrew & Strerud 1969; Sokal & Sneath 1963). However, sorted into distinct fabric groupings and further classified such approaches have been cautioned against as not all by a geochemical examination of tempers and clays (Roux attributes have equal weighting in the classification sys- & Courty 2005; see also Summerhayes 2000). The result tem (Christiansen & Read 1977; Dunn & Everitt 1982; Read of this hierarchical procedure are techno-compositional 1989). Lastly, a more recent study has avoided typology and classes that are then, thirdly, compared with morphologiclassified sherds based on stratigraphy before undertaking cal variations and decorative elements to establish techan attribute analysis (Shaw 2014), but this is unique in New no-style groups, diagnostic of specific production groups Guinea archaeology. within the broader communities of practice (see Roux 2011). This chapter describes the results of the first stage It should be acknowledged that typologies and classifica- of this hierarchical classification (steps 1–4), aiming to tions need not have fixed, empirical meaning; they are identify specific technological processes and the variation pragmatic constructs to help order and interpret observa- within them during the production phase. 100       · .  Ceramic assemblage Intuitive sorting Provisional style groups Hierarchical sorting based on technical syntaxes Technical classes Classification based on variation in technical elements Technical variants Sorting based on ocular microscopy Techno-fabric groups cross comparison with mophological and decorative attributes Classification based on mineralogial identification Techno-mineralogical groups Grouping based on clay composition Techno-compositional groups Techno-style groups Figure 7.2. Procedures of a technological classification (adapted from Roux 2011). The procedures examined in this chapter are highlighted yellow. globular pots common around coastal Madang are described as ‘Madang style.’ Intra-group stylistic variations in the technological process (e.g. between clan lines, or The Madang classification between different potters within a clan) are assumed to be To classify the Madang assemblages, all sherds were first minor compared to inter-group variation (e.g. between sorted intuitively based on internal and external surface Madang style and exotics). ‘Exotic’ non-Madang styles features which would be common to the entirety of the were present but extremely rare at both Tilu and Nunguri vessel, such as forming method and surface treatments to and in surface collections. For this reason, the emphasis the exterior. This was a necessary step to deal with the of this chapter is on the forming technologies of Madang large volume of artefacts, and separated sherds into pro- style ceramics, with exotics only mentioned in passing. visional style groups. For consistency in the classification of northeast New Guinea pottery, the broad style group- After describing broad style groupings, the sherds were ings assigned by Lilley (1986: 160) are retained. So, the red- then sorted by portion (Fig. 7.3). Rim sherds were conslipped, manually tempered, paddle and anvil constructed, sidered those that displayed a lip. Necks were those that Method 101 Chapter . Pre-Colonial Potting I: Production Orifice Lip Rim axis End point Rim Inflection point Neck Shoulder Corner point Carination Central vertical axis Refraction axis Corner point Body Base End point Ring base Figure 7.3. Anatomical landmarks of a Madang style pot. displayed a corner point and an abrupt change to vessel with an inverted neck, fashioned either in the body formcourse between the body and lip. Shoulder carinations ing stage or during preforming. Class 2 vessels followed a were those that displayed a corner point and an abrupt similar chaîne opératoire to Class 1, but a paddle was used change to vessel course between the body and shoulder. to bevel the exterior of the rim, probably during the body Ring bases were those with a distinct circular moulding forming stage. Class 3 rims are distinct from Classes 1, 5, acting as a stand. Plain body and base sherds were not and 2 in having a very thick internal collar and neck, with further considered in the analysis as all appeared to belong a body thickness abruptly thinner than the collar. This into the Madang style group and did not provide significant ternal collar is formed by the gentle hitting of the anvil on information about the production process. the interior corner point of the rim during body forming (see Tuckson 1966). Lastly, Class 4 is distinguished from The Madang style rims and necks were then hierarchically Classes 1, 2, 3, and 5 by the possible absence of a rim presorted into ‘technical classes,’ based on diagnostic techni- form and interior smoothing using an anvil up to the lip. cal syntaxes in the chaîne opératoire (Fig. 7.4). Again, for This results in thin bodies with restricted necks and no consistency with established literature, the numbering of abrupt change in incurving vessel course. It is worth notthese classes follows Lilley (1986) rather than hierarchical ing that Classes 1, 2, and 3 all represent bodi or tangeng closeness pertaining to the chaîne opératoire. Rim Classes within the current Bel type-system (see Chapter 4), despite 1–3 are retained, with minor alterations to their descriptors, following different technical syntaxes. Class 4 is equivalent and Classes 4 and 5 are added, as they were not repre- to the magob and Class 5 is equivalent to single spouted sented in the Vitiaz Strait assemblages, but are described you-bodi. by Egloff (1975) around Madang. Here, Class 1 represents vessels that likely began with an everted rim preform, with Technical attributes the body subsequently being shaped using an anvil up to the interior corner point where the body meets the rim. Following hierarchical sorting of Madang style sherds Class 5 vessels followed the same chaîne opératoire, but into major technical classes, attribute analysis was used to 102       · .  Ceramic assemblage Exotics Madang-style Key technical elements: Thickened neck and rim Anvil used to form collar in interior of vessel Hand moulding to produce everted rim Technical class 1 Globular body Manually tempered No rim preform (?) Rim preform Anvil used to shape body interior Unmodified everted rim Key technical elements: Red-slip Paddle and anvil Anvil used to shape body interior Hand moulding to produce inverted rim Paddling to bevel external rim Technical class 2 Unmodified inverted rim Technical class 5 Unmodified everted rim Unmodified incurving rim Technical class 3 Technical class 4 Figure 7.4. Classificatory procedures to distinguish Madang style technical classes based on chaîne opératoire. investigate variations in specific technical elements and to define ‘technical variants.’ For instance, minute differences in rim or lip shapes give insight into the individual potter’s gestures and intentionality on the clay – that is, embodied knowledge – and levels of enculturation into the production group (Kohring 2013). Importantly, owing to isochrestic variation, not all possible combinations of attributes will be represented. Those variants that are present will represent the extent of technological choice, while those not represented will indicate the boundaries of technological processes. By examining the variation within specific attributes, we can also describe which elements had tight social regulation in production and those with more relaxed controls. Moreover, specific technical elements may be spatially or temporally linked to specific production communities, which can then be used as chrono-stratigraphic indicators of technological change. entered into the Madang pottery database comprise both discrete (I–VI) and continuous (VII–XI) variables and are laid out as follows: I. Rim direction describes the general relationship between the rim and body shape (Summerhayes 2000: 35). Correct rim direction can be determined by placing each rim under a horizontal surface and moving the sherds at different orientations until the lip contacts the horizontal plane most completely (see Rice 1987: 222; Summerhayes 2000: 35). This allows the rim direction of very fragmentary sherds, only preserving the lip, to be determined. Note that direction was not recorded for rims that did not preserve enough circumference to be confidently orientated. Four rim directions were observed in the Madang assemblages: everted, inverted, incurving, and direct. Everted rims are those with an interior corner point (C.P.) and an abrupt change in vessel course forming a distinct rim Rims, necks, shoulder carinations, and ring bases (hereaf- and neck that projects away from the central vertical axis ter ‘formal sherds’) were examined to investigate variation (Fig. 7.5). Inverted rims are those with an interior corner of technical elements within different technical classes. point (C.P.) and an abrupt change in vessel course forming Rim sherds provide the most information about vessel a distinct rim and neck that projects towards the central form (Poulsen 1987: 870), and so the majority of selected vertical axis. Incurving rims are recognised by the presattributes concern rim morphology. The formal attributes ence of an inflection point (I.P.) and no corner point, con103 Chapter . Pre-Colonial Potting I: Production Rim direction C.P. C.P. Everted C.P. I.P. I.P. Incurving Inverted Direct Rim course Straight Convex Concave Sigmoid Sigmoid (convex-concave) (concave-convex) Rim profile Parallel Convergent Convergent Divergent Divergent Lenticular Lenticular (gradual) (abrupt) (gradual) (abrupt) (gradual) (abrupt) Lip profile Round Pointed Flat Flat (round edge) (sharp edge) Extra lip features Symmetrically Asymmetrically Asymmetrically thickened thickened interior thickened exterior Grooved interior Grooved exterior Grooved External sub-rim groove Ungrooved Grooved Grooved (with ridge) Figure 7.5. Discrete technical attributes recorded in the ceramic database. 104       · .  touring towards the central vertical axis. Direct rims, only VIII. Rim length, measured with callipers on everted and observed on exotics, are straight with no evidence of cor- inverted rims only, was taken as the distance between the ner or inflection points, but may project parallel, towards, lip and the neck running through the rim axis. or away from the central vertical axis. IX. Rim orientation was measured by an adjustable proII. Rim course is defined as the general line through the tractor attached to a horizontal surface (see method used rim axis, between the lip and body (see Irwin 1985: 105). by Skelly 2014: 104), recording the angle between the cenA rim course is either straight or curved (Fig. 7.5). If it is tral vertical axis and the rim axis (Irwin 1985: Fig. 43). Note curved, it is either concave or convex relative to the central that positive orientation is not always possible on fragvertical axis, or it is sigmoid (convex–concave or concave– mentary samples. Here, orientation for those rims with convex). This course is often defined by how the potter < 5% circumference was not recorded. spins the rim in her hands while preforming or moulding the body. X. Rim inclination was measured on everted and inverted rims by goniometer, taken as the angle between the rim III. Rim profile is the cross-section of a rim, representing axis and the body course. the relationship between the interior and exterior surfaces of the rim walls (Frankel et al. 1994: 15; Summerhayes XI. Diameter was calculated for both rims and ring-bas2000: 36). Seven rim profiles were observed in the Madang es. Orifice diameter is the maximum width of the vessel assemblages (Fig. 7.5): parallel, with interior and exterior mouth from rim axis to rim axis. This measure is obtained surfaces of constant thickness; convergent (gradual or by orientating the lip correctly (as above) on a diameterabrupt), with walls tapering towards the lip; divergent recoding sheet which has concentric circles marked at 1 cm (gradual or abrupt), with thickening towards the lip; and intervals (see Rice 1987: 223; Skelly 2014: 105; Summerhayes lenticular (gradual or abrupt), with the interior and exte- 2000: 36). Orifice diameter can indicate primary function rior surfaces diverging and then tapering towards the lip. (e.g. bodi cooking pots have larger orifices than you-bodi water pots) or technological variation between communiIV. Lip profile is the cross section of the lip at the end ties (e.g. modern Bilbil bodi usually have wider mouths point of the rim (Summerhayes 2000: 36). Four lip profiles than Mindiri bodi). Ring-base diameter is the maximum were identified: round, pointed, flat (round edge), and flat width of a circular stand, placed on the same diameter(sharp edge) (Fig. 7.5.). recording sheet. V. Extra lip features are a category for additional morphological attributes that are not covered by the lip profile category. This includes elements that are largely inconsequential to the forming process but can describe the peculiar movements of the individual potter as they moulded the lip. Lips can be symmetrically thickened, asymmetrically thickened on the interior surface, asymmetrically thickened on the exterior surface, grooved (along the circumference of the lip), grooved around the exterior of the lip, or grooved around the interior. In the Madang style, these variations appear to be caused primarily by the pinching of the clay and the position of the fingers while forming the preform or holding it while paddling the body. Assemblage composition In total, 46,682 pottery sherds were recovered in the 2014 field season and are examined in the current study. This includes 46,573 sherds from excavated contexts and 109 from surface survey (Table 7.1). Prior to laboratory processing, starch and phytolith residue samples were extracted from 44 Tilu sherds and 12 Nunguri sherds. Micro-botanical analysis by Judith Field, Sindy Luu, and Adelle Coster indicate starchy foods were being cooked in some of the pots at Tilu (Gaffney et al. 2020). All ceramics were then individually washed through 2 mm sieves, cleaned with brushes, and dried at 20°C in the Archaeology Laboratories, University of Otago. Plain body sherds were bulk VI. External sub-rim grooves are produced by deliberate bagged and weighed, while decorated and formal sherds paddling to form a groove or sometimes a groove with as- were individually bagged and assigned catalogue numbers. sociated ridge, below the rim and neck (Fig. 7.5). These are The decorated and formal assemblages were analysed and rare and only observable on especially well preserved rims. later reexamined to develop a clear idea of the variation that existed within the assemblage and to ensure internal VII. Thickness was measured at three distinct points with consistency throughout the analysis. Vernier callipers. Rim thickness, recorded on all incurving rims, everted and inverted rims of sufficient preservation, At both Tilu and Nunguri, the ratio of sherd portions is was considered the maximum distance between the inte- consistent throughout Layer 1. However, in Layer 2 the rior and exterior rim walls, perpendicular to the rim axis small sample sizes are problematic resulting in few for(Fig. 7.6). Neck thickness, only recorded on everted and mal and decorative sherds being represented (Fig. 7.7–7.8). inverted rims and necks, was the distance between exterior Generally, however, both excavated assemblages produced and interior surfaces immediately below the rim, perpen- comparable proportions of rims, necks, decorated body dicular to the body course. Body thickness, measured on sherds, and plain body sherds for analysis. Plain sherds decorated body sherds only, was the maximum distance dominate the assemblages, comprising 90% of the total between exterior and interior surfaces (Fig. 7.6). sherd count (n = 41,826), while decorated sherds consti105 Chapter . Pre-Colonial Potting I: Production c) d) a) Rim thickness b) Neck thickness c) Rim length d) Rim orientation e) Rim inclination a) e) b) Orifice diameter Body thickness Base diameter Figure 7.6. Metric technological attributes recorded in the ceramic database. tute 8% (n = 3637), rims constitute 2% (n = 1090), and necks, bases, and carinations < 1%. Rims were visually compared and refitted during processing and analysis to give a minimum number of vessels (MNV) per spit (see Goodby 1998; Specht 1969: 72). These values constitute all sherds assignable to a single vessel and can be used to assess whether artefacts are in primary deposition and to reduce concerns of taphonomic bias, caused by differential breakage patterns, when quantitatively comparing potsherds and associated attributes through time (Table 7.2). At Tilu, refitting determined that there was lit- tle redeposition and vertical disturbance. Although rim sherds from the same vessel were collected from base of Spit 1 and top of Spit 2, one sherd in Spit 9 refitted with two from Spit 10, and one from Spit 1 refitted with another from Spit 4, all other sherds refitted with others from their own spit. In the case of multiple sherds refitting from different spits, the MNV count was assigned to the spit most frequently represented. In the case that two sherds from different spits rejoined, the MNV count was assigned to the lower spit. Owing to the large number of surface sherds around Tilu, redeposition was considered more plausible than downward movement through the deposit. At Nun- 106       · .  Table 7.1. Madang ceramic assemblages by number (#) and weight (g). Rim Site Nunguri Test Pit 1 Spit # R. base g # Carination Decorated body g # g # g Plain body # Total g # g 22 314.85 5 33.54 – – – – 95 282.84 1491 1685.60 1613 2316.83 2 149 2132.81 17 124.24 – – – – 425 1490.31 4902 7003.50 5493 10750.85 3 135 1929.10 14 76.55 – – – – 403 1378.69 4323 7519.91 4875 10904.25 4 92 1843.99 10 49.80 – – – – 251 989.98 2402 4818.97 2755 7702.74 5 37 468.06 9 76.64 – – – – 103 424.26 913 1803.77 1062 2772.73 6 64 539.87 11 86.64 – – – – 227 656.16 3163 4027.94 3465 5310.61 7 49 404.78 4 30.76 – – – – 158 484.44 1525 2175.45 1736 3095.43 8 29 239.09 2 25.58 – – – – 140 460.55 1767 2218.27 1938 2943.49 9 35 375.04 7 36.62 – – – – 107 376.59 1336 1817.02 1485 2605.27 10 39 424.35 1 1.78 – – – – 116 417.55 1424 1831.08 1580 2674.76 11 53 676.31 3 22.50 – – – – 191 512.04 2505 2824.09 2752 4034.94 12 43 391.57 5 20.16 – – – – 154 505.43 1593 1944.48 1795 2861.64 13 40 313.63 6 44.50 – – – – 219 795.23 2201 2252.65 2466 3406.01 14 47 744.88 5 34.81 – – – – 239 763.36 2121 2665.30 2412 4208.35 15 46 595.67 4 10.51 – – – – 120 335.34 1118 1178.01 1288 2119.53 16 1 7.27 – – – – – – – – 35 33.66 36 40.93 17 – – – – – – – – – – 16 10.61 16 10.61 881 11401.27 103 674.63 – – – – 2948 9872.77 32835 45810.31 36767 67758.98 1 25 333.72 3 13.80 – – – – 131 602.69 1409 2137.80 1568 3088.01 2 30 348.61 4 53.12 – – – – 93 480.81 1128 1685.46 1255 2568.00 3 15 288.58 5 28.4 – – – – 100 428.87 1053 1760.44 1173 2506.29 4 14 191.48 1 5.55 – – – – 74 334.34 1037 1397.24 1126 1928.61 5 6 70.18 2 2.81 – – – – 40 119.32 716 953.98 764 1146.29 6 15 267.52 3 23.67 – – 1 1.21 66 215.80 1292 1567.18 1377 2075.38 7 23 306.99 2 11.16 – – – – 54 242.63 706 1073.80 785 1634.58 8 5 126.60 3 15.08 – – – – 14 70.07 448 655.47 470 867.22 9 24 426.05 1 0.65 – – – – 69 289.39 725 1162.58 819 1878.67 10 5 56.82 – – – – – – 21 106.24 326 494.21 352 657.27 11 – – – – – – – – 1 1.87 39 54.71 40 56.58 12 – – – – – – – – 4 4.72 17 14.73 21 19.45 162 2416.55 24 154.24 – – 1 1.21 667 2896.75 8896 12957.60 9750 18426.35 45 cm 8 170.72 – – – – – – 3 28.59 11 162.85 22 362.16 60 cm 5 174.58 – – 1 29.6 – – 2 12.42 10 127.71 18 344.31 70 cm – – – – – – – – – – 9 34.13 9 34.13 80 cm 1 78.07 – – – – – – – – – – 1 78.07 120 cm 1 15.83 – – – – – – – – 5 16.02 6 31.85 15 439.20 – – 1 29.6 – – 5 41.01 35 340.71 56 850.52 215.41 Sub-total Tilu Shovel Pit 1 Neck g 1 Sub-total Tilu Unit 1 # Sub-total Kranket Is. Surf. 7 170.20 – – – – – – 3 45.21 – – 10 Siar Is. Surf. 4 68.66 – – – – – – 1 23.49 1 4.23 6 96.38 Yabob Is. Surf. 6 324.23 – – – – – – 1 15.42 – – 7 339.65 Malmal Surf. 15 430.22 – – – – – – 12 182.94 59 314.92 86 928.08 32 993.31 – – – – – – 17 267.06 60 319.15 109 1579.52 1090 15250.33 127 828.87 1 29.6 1 1.21 3637 13077.59 41826 59427.77 46682 88615.37 Sub-total Total guri, fewer sherds could be accurately refitted suggesting This section can be used in conjunction with Appendix A, more horizontal displacement, whether cultural or natural. which illustrates a wider range of technical variation in No Nunguri rims refitted with those from different spits. the assemblages. The next section describes the results of the attribute analysis with regard to forming techniques. This first examines formal sherds from Nunguri, Tilu, and surface survey followed by a summary comparison at the end of the section. 107 Chapter . Pre-Colonial Potting I: Production Table 7.2. Rims by number of sherds and minimum number of vessels. Site Nunguri Test Pit 1 Spit 22 21 2 149 141 3 135 129 4 92 88 5 37 35 6 64 60 7 49 40 8 29 26 9 35 31 10 39 33 11 53 39 12 43 30 13 40 30 14 47 36 15 46 37 16 1 1 17 – – 881 777 1 25 16 2 30 23 3 15 12 4 14 12 5 6 6 6 15 12 7 23 15 8 5 4 9 24 13 10 5 4 11 – – 12 Test Pit 1 In the Nunguri assemblage, Madang style rims are abundant through Layer 1 (Spit 1–15). A single rim is present in Layer 2 (Spit 16–17). Overall, Class 1 rims are the most frequent technical class present, followed by Class 2, 4, 3, and 5 rims (Table 7.3), but the relative percentages of these classes changes through the excavation spits (Fig. 7.9). A single ‘exotic,’ non-Madang style rim form was identified at Nunguri, in Spit 15. Different technical classes are more common in different spits, and therefore at different times in the past. A Pearson’s Chi-squared Test, using a Monte-Carlo simulation owing to low cell counts for Class 5 rims, supports these results showing that there is a significant association between excavation spit and technical class (X2 = 145 df = 60 P < 0.01). A crosstabulation of these test results shows that the observed value deviates from the expected value in several instances where adjusted Pearson residuals are >2 or < –2 (see Agresti 2002: 81). Of particular note, observed Class 4 rims are substantially fewer than expected in Spit 2 (adj. residual = –3), and in Spit 4 (adj. residual = –2.1), towards the top to the excavated sequence. Towards the base of the sequence in Spit 11 (adj. residual = 2.8) and Spit 13 (adj. residual = 2.4), Class 4 rims are substantially more common than expected. On the other hand, observed Class 3 rims are substantially fewer than expected in Spits 14 and 15 (adj. residuals = –2.3), and more common than expected in Spits 4–6 (adj. residuals = 3.5). This suggests that Class 4 rims were more frequently produced and con- Table. 7.3. Technical classes represented at Nunguri, Test Pit 1 by MNV and excavation spit. – – 162 117 45 cm 8 7 Spit 60 cm 5 3 1 Sub-total Tilu Shovel Pit 1 Minimum number of vessels* 1 Sub-total Tilu Unit 1 Number of rim sherds Results: Nunguri Technical class 70 cm – – 2 80 cm 1 1 3 120 cm Sub-total 1 1 4 15 12 5 Cal. BP 1 2 8 3 7 3 4 5 1 2 Exotic – Total 21 78 28 21 12 2 – 141 0–299 73 18 13 22 3 – 129 41 16 21 8 2 – 88 0–304 14 5 11 5 – – 35 Kranket Is. Surf. 7 7 6 24 2 16 16 2 – 60 Siar Is. Surf. 4 4 7 20 6 5 8 1 – 40 Yabob Is. Surf. 6 6 8 17 3 – 6 – – 26 Malmal Surf. 15 15 9 14 13 2 2 – – 31 32 32 10 17 8 1 7 – – 33 1090 938 Sub-total Total 11 508–539 12 * Calculated by refitting and visual inspection. Note that rims that do not preserve enough information to be classified into a distinct technical class are also excluded from this count. 13 2 13 1 – 39 10 1 8 – – 30 13 515–623 15 5 – 10 – – 30 14 520–626 18 8 – 10 – – 36 15 23 9 – 4 – 1 37 16 1 – – – – – 1 17 – – – – – – – 384 151 96 132 13 1 777 Total 108 10 11       · .  100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 Decorated Plain 12 13 14 15 16 17 Spit Rims Necks Figure 7.7. Portions of ceramic vessels represented at Nunguri, Test Pit 1, by number. 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 Spit Rim Neck Carination Decorated Plain Figure 7.8. Portions of ceramic vessels represented at Tilu, Unit 1, by number. sumed earlier in the sequence ~500 cal. BP, while Class 3 rims are more frequent later in time, between 0–300 cal. BP. Nunguri rim attributes vary by technical class. All Class 1, 2, and 3 vessels have everted rims, with an interior corner point (Fig. 7.10–7.12) while Class 5 rims are inverted (Fig. 7.14), and Class 4 vessels have incurving rims, with an inflection point but no corner point (Fig. 7.13). Class 1 rims display the most variation in rim form attributes and are similar to Class 2 rims, while Class 3, 4, and 5 rims display less variation. Nunguri Class 1 rim courses usually follow a straight line, but can also be convex, flaring outward or, occasionally, concave. There are four rare instances of Class 1 rims with sigmoid rim courses, but these comprise less than 1% of Class 1 rims (Table 7.4). Over half of the observed Nunguri Class 1 rim profiles are parallel, but gradually convergent, divergent, and lenticular forms are also common, while abruptly convergent, divergent, and lenticular forms are rare. Class 1 lip profiles are generally round, but can also be flat or pointed, and extra lip features are usually absent although grooved, grooved interior, grooved exterior, 109 Chapter . Pre-Colonial Potting I: Production 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Spit Class 1 Class 2 Class 3 Class 4 Class 5 Figure 7.9. Percentage of Madang style technical classes at Nunguri, Test Pit 1 by excavation spit. symmetrically thickened, and asymmetrically thickened exterior lip features also occur owing to irregularities in potter’s handling of the rim (Fig. 7.15–7.16). ally flat, but can sometimes be pointed or round, while extra features are almost always absent, although asymmetrical thickening can occur. Nunguri Class 2 rims display a similar variety of rim courses to Class 1, usually being straight, less often being convex and concave, and, rarely, sigmoid. Rim profiles are more often abruptly convergent or lenticular, owing to external paddling to create a bevel, but can sometimes be parallel, with the paddling producing a distinct change in vessel course at the bevel. Less often, Class 2 rim profiles are gradually convergent, divergent, and lenticular, or abruptly divergent. Lip profiles are usually round but can sometimes be pointed, or flat with a round edge, while extra lip features are usually lacking, although several instances of grooving and asymmetric thickening to the exterior were observed. Fewer Class 5 rims were deposited at Nunguri (n = 13). Those observed have straight, convex, or concave rim courses, with parallel, lenticular, divergent or convergent profiles, flat or rounded lips, and are lacking extra lip features apart from one instance of asymmetric thickening to the exterior. Sub-rim grooves and grooves with associated ridges are common on Class 5 rims, with 60% (n = 6) displaying this technical attribute (Table 7.4). These grooves are rare on Class 1 and 2 rims, being present on < 5% of the sherds. No examples of Class 3 and 4 rims with these grooves are present. Stratigraphically, the earliest example of this attribute comes from a single Class 1 rim in Spit 11, which displays a sub-rim groove with ridge, but most examples are present from Spit 1–7 (n = 15). Class 3 rims almost always follow a straight rim course owing to the collar being extended as the potter ran an anvil or hand around the inside of the neck. However, rare examples follow a convex or concave course. Their rim Based on these discrete formal attributes, an attempt was profiles are usually divergent (abrupt or gradual) but can made to demonstrate that specific combinations of techalso be parallel. Lips are usually flat but sometimes round, nical elements were more associated with each technical and usually asymmetrically thickened on the exterior. class. A multiple correspondence analysis (MCA) was used to investigate underlying structures in this data. MCA can Class 4 rims follow a concave or straight rim course. Note analyse the pattern of relationships of several categorical that many of those observed to be straight almost certainly dependent variables (Abdi & Valentin 2007), and is broadderive from vessels with an incurving and concave rim ly synonymous with optimal scaling, appropriate scoring, course, but too little of the rim survives to assign it as con- and homogeneity analysis. MCA can then be viewed as a cave. Three examples are convex towards the lip. Profiles generalisation of principal component analysis for catare usually gradually divergent or parallel but can also be egorical data. Such an approach is borrowed from other gradually convergent or abruptly divergent. Lips are usu- social sciences, which have attempted to relate different 110       · .  N-142 Spit 2 N-163 Spit 2 N-642 Spit 3 N-692 Spit 3 N-2011 Spit 7 N-2738 Spit 11 N-3405/3407/3454 Spit 14 N-3702 Spit 15 N-3866 Spit 16 0 5 cm Figure 7.10. Class 1 Madang style rims at Nunguri, Test Pit 1. 111 Chapter . Pre-Colonial Potting I: Production N-132 Spit 2 N-150 Spit 2 N-693 Spit 3 N-1556 Spit 5 N-1546 Spit 5 N-2689 Spit 11 N-2698/2699/2709 Spit 11 N-3701 Spit 15 N-3699 Spit 15 0 5 cm Figure 7.11. Class 2 Madang style rims at Nunguri, Test Pit 1. 112       · .  N-124 Spit 2 N-125 Spit 2 N-651 Spit 3 N-1192 Spit 4 N-1999 Spit 6 N-2009 Spit 7 N-2385 Spit 9 N-2702 Spit 11 0 5 cm Figure 7.12. Class 3 Madang style rims at Nunguri, Test Pit 1. 113 Chapter . Pre-Colonial Potting I: Production N-134 Spit 2 N-165 Spit 2 N-635 Spit 2 N-2224 Spit 8 N-2223 Spit 8 N-2405 Spit 9 N-3415/3452 Spit 14 N-3802 Spit 15 0 5 cm Figure 7.13. Class 4 Madang style rims at Nunguri, Test Pit 1. 114       · .  N-169 Spit 2 N-136 Spit 2 N-662 Spit 3 N-660 Spit 3 N-1700 Spit 6 N-1711 Spit 6 N-2007 Spit 7 N-2713 Spit 11 0 5 cm Figure 7.14. Class 5 Madang style rims at Nunguri, Test Pit 1. social attributes across space or time (e.g. Bourdieu 1979, 1988, 1998). Figure 7.17 plots an MCA with dependent variables (rim course, rim direction, rim profile, lip profile, lip feature, and technical class) using the ‘FactoMineR’ package developed by Le and colleagues (2008). This applied the MCA algorithm to a ‘Burt table’ – a symmetrical matrix of all two-way cross-tabulations between the variables – to show that multiple technical elements are related to the forming of each technical class. In the plot (bottom left), Classes 1, 2, and 5 group closely together and overlap, while Class 3 and 4 group apart and tend not to overlap. Class 3 rims, for instance, are usually everted, divergent abrupt with asymmetrically thickened exteriors and flat round lips. As these attributes represent sections in the overall 115 technological process (see above, Fig 7.1), similarities in multiple elements hint at similar manufacturing sequences. In this way the forming sequence of Class 1, 2, and 5 rims can be argued to be more similar than the sequence for Class 3 or Class 4 rims, which follow different technical processes. It is also important that no cluster coded by technical class separates into two distinct clusters, which would justify separation into technical variants based on different technical processes. Several continuous attributes were measured within these classes. Rim thickness varies by technical class at Nunguri. Classes 1, 2, 3, and 5 overlap at one standard deviation (SD) from the mean (Table 7.5), while Class 4 overlaps only with Chapter . Pre-Colonial Potting I: Production Table. 7.4. Rim and lip forms by technical class at Nunguri, Test Pit 1 (note some rims not represented where attributes could not be confidently recorded). Class 1 n Class 2 % n Class 3 % n Class 4 % n Class 5 % n % Rim course Straight 248 68 83 56 87 91 59 47 8 62 Convex 93 25 52 35 7 7 3 2 2 15 Concave 21 6 10 7 2 2 65 51 3 23 Sigmoid (convex-concave) 2 <1 3 2 – – – – – – Sigmoid (concave-convex) 2 <1 1 <1 – – – – – – 46 Rim profile 204 56 17 12 2 2 63 49 6 Convergent abrupt Parallel 4 1 61 41 – – – – – – Convergent gradual 96 26 11 7 – – 1 1 1 8 23 Divergent abrupt 4 1 1 1 38 40 1 1 3 Divergent gradual 42 11 3 2 49 52 63 49 1 8 Lenticular abrupt 5 1 52 35 – – – – 2 15 Lenticular gradual 13 4 3 2 – – – – – – Lip profile 336 88 115 78 16 17 7 5 10 77 Flat (round edge) Round 23 6 6 4 78 82 13 10 3 23 Flat (sharp edge) 2 1 – – 1 1 111 84 – – 19 5 26 18 – – 1 <1 – – 92 Pointed Extra lip features 334 89 134 93 15 16 121 93 12 Grooved Absent 6 2 – – 3 3 – – – – Grooved exterior 3 <1 2 1 – – – – – – Grooved interior 12 3 7 5 1 1 – – – – Symmetrically thickened 14 4 – – 1 1 1 1 – – Asymmetrically thickened exterior 8 2 2 1 75 79 8 6 1 8 Asymmetrically thickened interior – – – – – – – – – – Sub-rim groove 167 97 71 95 81 100 131 100 4 40 Grooved Absent 2 1 1 1 – – – – 2 20 Grooved with ridge 4 2 3 4 – – – – 4 40 Table. 7.5. Average rim and neck thickness (mm) by technical class at Nunguri, Test Pit 1. Class 1 Class 2 Class 3 Class 4 Class 5 11.06 15.53 6.41 10.07 Rim thickness Mean 9.50 SD 1.86 1.85 4.24 1.76 2.07 CV 20% 17% 27% 27% 21% 7.79 10.22 N.A. 8.20 Neck thickness Mean 7.47 SD 1.82 1.67 2.09 N.A. 1.73 CV 24% 21% 20% N.A. 21% a Monte-Carlo simulation owing to small sample sizes in Class 5, shows that there are significant differences in rim thickness between technical classes (Table 7.6). The effect size (η2) of 49.79% shows that almost half of the variation in rim thickness is accounted for by technical class. A post hoc Kruskal-Wallis test run between each class (e.g. Class 1–Class 2, Class 1–Class 3, Class 1–Class 4, etc.) shows that significant differences exist between each class except Classes 2 and 5 (Table 7.7). Table. 7.6. Kruskal-Wallis test of rim thickness (mm) by technical class at Nunguri, Test Pit 1. Class 1 Class 2 Class 1 and 5 at 1 SD. Generally, Class 3 rims are the thickNo. of cases 376 149 est, followed by Class 1, 2, and 5 rims, while Class 4 rims Mean rank 352.67 496.65 are much thinner (Fig. 7.18). A Kruskal-Wallis test, which is a non-parametric one-way analysis of variance, using X² = 377.372, df = 4, η² = 49.79%, 116 Class 3 Class 4 Class 5 094 127 013 646.66 114.38 500.08 P<0.01       · .  Rim profile 45 40 40 35 35 30 30 25 25 20 20 15 15 10 10 5 5 0 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 0 Class 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 45 45 40 40 35 35 30 30 25 25 20 20 15 15 10 10 5 5 0 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 0 Class 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 45 45 40 40 35 35 30 30 25 25 20 20 15 15 10 10 MNV MNV 1 5 5 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 0 Class 3 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 45 45 40 40 35 35 30 30 25 25 20 20 15 15 10 10 MNV MNV MNV MNV 1 5 5 Class 4 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 16 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 45 45 40 40 35 35 30 30 25 25 20 20 15 15 10 10 MNV MNV MNV MNV Rim course 45 5 5 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Class 5 0 1 2 3 4 5 6 7 8 Spit 9 10 11 12 13 14 15 16 Spit Parallel Straight Convex Concave Convergent abrupt Convergent gradual Sigmoid (concave-convex) Divergent abrupt Divergent gradual Sigmoid (convex-concave) Lenticular abrupt Lenticular gradual Figure 7.15. Rim course and profile by technical class at Nunguri, Test Pit 1 (note some rims not represented where attributes could not be confidently recorded). Table. 7.7. Post hoc test of rim thickness (mm) comparing individual technical classes at Nunguri. Technical class 1–2 1–3 1–4 1–5 2–3 2–4 2–5 3–4 3–5 N 525 470 503 389 243 276 162 221 107 4–5 140 X² 69.505 149.218 167.684 8.492 69.785 175.701 0.008 155.116 12.280 31.483 df 1 1 1 1 1 1 1 1 1 1 P <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 0.93 <0.01 <0.01 <0.01 η² 13.24% 31.81% 33.40% 2.19% 28.84% 63.89% 0.00% 70.51% 11.58% 22.65% 117 Chapter . Pre-Colonial Potting I: Production Extra lip features 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Class 2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Class 3 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 70 70 60 60 50 50 40 40 30 30 20 20 MNV MNV 1 10 10 Class 4 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 16 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Class 5 MNV MNV MNV MNV Class 1 MNV MNV 1 MNV MNV Lip profile 0 1 2 3 4 5 6 Spit 7 8 9 10 11 12 13 14 15 16 Spit Absent Round Flat (round edge) Pointed Flat (sharp edge) Grooved exterior Asymmetrically thickened exterior Grooved interior Asymmetrically thickened interior Grooved Symmetrically thickened Figure 7.16. Lip profile and extra lip features by technical class at Nunguri, Test Pit 1 (note some rims not represented where attributes could not be confidently recorded). 118       · .  -0.5 0.0 Rim course 0.5 1.0 1.5 2.0 Rim direction Rim profile 2.0 1.5 1.0 0.5 -0.5 Straight Convex Concave Concave-convex Convex-concave Everted Class Parallel Convergent abrupt Convergent gradual Divergent abrupt Divergent gradual Lenticular abrupt Lenticular gradual Incurving Lip feature Lip profile 2.0 1.5 1.0 0.5 0.0 -0.5 -0.5 0.0 0.5 1.0 1.5 2.0 Class 1 Class 2 Class 3 Class 4 Class 5 Asymmetrically thickened exterior Absent Grooved interior Grooved exterior Symetrically thickened Grooved -0.5 0.0 0.5 Round Flat (round-edge) 1.0 Figure 7.17. MCA of discrete formal attributes for Madang style rims, Nunguri, Test Pit 1. 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 1 2 3 Class 4 5 Figure 7.18. Rim thickness by technical class at Nunguri, Test Pit 1 119 1.5 Pointed Flat (sharp-edge) Dimension 1 (14.98%) Rim thickness (mm) Dimension 2 (12.64%) 0.0 2.0 Chapter . Pre-Colonial Potting I: Production The coefficient of variation (CV) expresses dispersion of data as a percentage and is used to describe standardisation in pot form, with lower percentages being more standardised, and higher percentages being less standardised (Arnold 1991; Longacre 1999; Masson & Rosenswig 2005). At Nunguri, Class 1, 2, and 5 rim thicknesses show similar CV at c. 20%, while Class 3 and 4 are higher at 27%, suggestive of less standardisation in rim thickness (Table 7.5). Ethnographic observations from elsewhere in the world have equated 5–10% CV with full time craft specialists with tight social regulations on form, while ≥15% has been equated to part time producers with embodied knowledge that is less frequently engaged (Benco 1988; Longarcre et al. 1988). project away from the central vertical axis at the lip. Class 3 rims are everted and their orientations form a continuum. Most Class 3 rims project inwards, towards the central vertical axis (70%, n = 24, mean = 37°), although some project slightly outwards (30%, n = 56, mean = 22°). Class 4 rims are always incurving, with rims projecting towards the central vertical axis. Class 5 rims are inverted and always project towards the axis (Table 7.8). Figure 7.21 illustrates the range in rim orientation and direction of each technical class. The mean rim orientation and inclination of Class 1, 2, and 3 rims are similar, while Class 4 and 5 rims are generally higher in their orientation (Table 7.9). The CV of orientation for all Classes is more than for thickness1 while inclination is less and similar to Rim shape can be assessed by plotting different combina- thickness. Figure 7.22 plots orientation against inclination tions of attributes. Figure 7.19. charts rim thickness against showing a general trend for inclination to decrease as orineck thickness showing that rim thickness increases with entation increases. neck thickness in a linear manner. Classes 1, 2, and 5 overlap, while Class 3 sherds generally have much thicker rims Orifice diameter also varies between technical classes. and slightly thicker necks. Figure 7.20 plots rim length Class 1, 2, 3, and 4 all overlap at 1 SD from the mean (Table against rim thickness showing that Class 1, 2, and 5 overlap, 7.10). Class 1 rims are between 5 cm–20 cm wide with a while Class 3 are thicker and shorter. mean and mode of 14 cm. Figure 7.23, which plots MNV Rim direction, orientation, and inclination also vary by technical class. Class 1 and 2 rims are everted and always 1 Note, there is no test of significance for differences between CV 18 16 Neck thickness (mm) 14 12 10 8 6 4 2 0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 Rim thickness (mm) Class 1 Class 2 Class 3 Class 5 Figure 7.19. Relationship between rim and neck thickness by technical class at Nunguri (note some rims not represented where attributes could not be confidently recorded). 120       · .  18 16 Rim thickness (mm) 14 12 10 8 6 4 2 0 5 0 10 15 20 25 30 Rim length (mm) Class 1 Class 2 Class 3 Class 5 Figure 7.20. Relationship between rim length and thickness by technical class at Nunguri, Test Pit 1 (note some rims not represented where attributes could not be confidently recorded). Table. 7.8. Rim direction (MNV) by technical class and excavation spit at Nunguri, Test Pit 1. Spit Class 1 Class 2 Class 3 Class 4 Class 5 1 6 0 5 0 – – – – – – 2 61 0 20 0 3 14 0 11 0 2 3 56 0 17 0 4 9 0 17 0 3 4 33 0 11 0 6 12 0 5 0 1 5 11 0 5 0 4 5 0 2 – – 6 19 0 2 0 4 10 0 12 0 2 1 7 19 0 5 0 2 3 0 7 0 8 13 0 2 0 – – 0 6 – – 9 13 0 9 0 0 1 0 2 – – 10 14 0 8 0 0 1 0 5 – – 11 10 0 11 0 1 0 0 10 0 1 12 8 0 9 0 0 1 0 6 – – 13 15 0 5 0 – – 0 7 – – 14 15 0 8 0 – – 0 10 – – 15 22 0 9 0 – – 0 3 – – 16 1 0 – – – – – – – – 17 – – – – – – – – – – 316 0 126 0 24 56 0 103 0 10 Total 121 Chapter . Pre-Colonial Potting I: Production 1o 2o 3o 9o 25o 55o 80o 75o Class 1 Class 2 78o Class 3 85o Class 4 Class 5 Figure 7.21. Minimum and maximum rim orientation by technical class at Nunguri, Test Pit 1. Table 7.9. Average rim orientation and inclination (°) by technical class, Nunguri, Test Pit 1. Class 1 Class 2 Class 3 Class 4 Table 7.10. Average orifice diameter (cm) by technical class, Nunguri, Test Pit 1. Class 5 Orientation Mean 32 Mean 31 33 43 Class 1 Class 2 Class 3 Class 4 Class 5 14 14 14 15 10 57 SD 2 2 2 3 2 CV 17% 16% 16% 19% 20% SD 13 12 20 15 22 CV 41% 39% 61% 35% 39% 134 130 N.A. 160 Inclination Mean 121 SD 22 22 22 N.A. 3 CV 18% 16% 17% N.A. 2% wider rims (8–20 cm) may be equivalent to modern bodi and tangeng. No examples of Class 2, 3, or 4 rims smaller than 9 cm are present, and their distributions in Figure 7.23 are unimodal, overlapping with the larger Class 1 distribution. by orifice diameter, suggests the Class 1 distribution may be bimodal. If this is correct, small Class 1 rims (5–6 cm) might represent the spouts of you-bodi water pots, while A Kruskal-Wallis test using a Monte-Carlo simulation shows that there are significant differences between the means of each class with regard to orifice diameter (Table 180 160 140 Rim inclination° 120 100 80 60 40 20 0 0 10 20 30 40 50 60 70 80 90 Rim orientation° Class 1 Class 2 Class 3 Class 5 Figure 7.22. Relationship between rim orientation & inclination by class, Nunguri, Test Pit 1 (note some rims not represented where attributes could not be confidently recorded). 122       · .  60 50 MNV 40 30 20 10 0 5 6 7 8 9 10 Class 1 11 12 13 14 15 16 Orifice diameter (cm) Class 2 Class 3 Class 4 17 18 19 20 21 22 Class 5 Figure 7.23. Vessel orifice diameter by technical class, Nunguri, Test Pit 1. 7.11). Mood’s test confirms these results showing there are significant differences between the medians of the classes (Table 7.12). The effect size (η2), however, is small showing that only 7.49% of orifice variation is accounted for by technical class. A post hoc Kruskal-Wallis test indicates that there are significant differences in orifice diameters between each class, except Class 1 and 3 and Class 2 and 3 (α = 0.05) (Table 7.13). Exotics A single exotic rim form was identified in the Nunguri assemblage, in Spit 15 (Fig. 7.24). Although similar to Madang style Class 4 rims, this example lacks the red-slip on the interior and exterior surfaces and is formed from a distinct fabric (see Chapter 8). The rim derives from a restricted, incurving vessel with parallel profile, flat sharp edged lip with asymmetrical thickening on the exterior. Table 7.11. Kruskal-Wallis test of orifice diameter by technical class at Nunguri, Test Pit 1. Class 1 Class 2 Class 3 Class 4 No. of cases 269 114 71 73 Class 5 10 Mean rank 250.49 286.14 261.45 345.08 69.65 X² = 40.167, df = 4, η² = 7.49%, P< 0.01 Table 7.12. Mood’s test of orifice diameter by technical class at Nunguri, Test Pit 1. Class 1 Class 2 Class 3 Class 4 Class 5 269 114 71 73 10 Diameter > median 91 52 29 44 1 Diameter ≤ median 178 62 42 29 9 Number of cases Median = 14, Χ² = 21.928, df = 4, P<0.01 123 Chapter . Pre-Colonial Potting I: Production Table 7.13. Post hoc test of orifice diameter comparing individual technical classes at Nunguri. Technical class 1–2 1–3 1–4 1–5 2–3 2–4 2–5 3–4 3–5 4–5 N 383 340 342 279 185 187 124 144 81 83 X² 4.798 0.239 21.656 15.504 1.080 7.857 17.199 10.407 15.678 17.325 df 1 1 1 1 1 1 1 1 1 1 P 0.03 0.63 <0.01 <0.01 0.30 <0.01 <0.01 0.01 <0.01 <0.01 η² 1.26% 0.07% 6.35% 5.56% 0.06% 4.20% 13.87% 7.23% 19.36% 20.87% N-3712 Spit 15 0 5 cm Figure 7.24. Exotic rim from Nunguri, Test Pit 1. A chi-squared test under the same conditions as above (see pg. 212) shows that at Tilu there is no significant correlation between spit and class (X2 = 32, df = 36, P = 0.584). A Unit 1 correspondence analysis confirms this lack of correlation At Tilu, Unit 1, Madang style rims dominate the formal (X2 = 32.136, inertia = 0.279, P = 0.908). However, crosstabuassemblage in Layer 1 (Spits 1–10). Class 1 rims are most lation shows that the observed Class 3, 4, and 5 rims are common, with Class 2 being next most frequent (Table 7.14; substantially more than expected in every instance (adj. Fig. 7.25). Class 3 and 4 rims are rare and only present in or residuals = ≥2), owing to their absence in most spits, espeabove Spit 2, and only one Class 5 rim is present, in Spit 4. cially below Spit 4. Although the sample sizes are small, it Two exotic rim forms were also recovered, in Spit 2 and is possible that Classes 3 and 4 in particular are chronoSpit 4. No formal sherds were found in Layer 2 (Spits 11–12). logically more recent than Classes 1 and 2. Results: Tilu Table 7.14. Technical classes represented at Tilu, Unit 1 by MNV and excavation spit. Technical class Spit Cal. BP 1 1 2 3 4 5 Exotic Total 10 3 – 3 – – 16 2 16 4 1 1 – 1 23 3 10 2 – – – – 12 4 552–664 5 10 – – – 1 1 12 5 1 – – – – 6 6 543–647 10 2 – – – – 12 7 540–644 11 4 – – – – 15 3 1 – – – – 4 8 10 3 – – – – 13 10 9 791–915 4 – – – – – 4 11 564–674 – – – – – – – 12 535–641 – – – – – – – 89 20 1 4 1 2 117 Total A range of rim forms is presented in Figures 7.26–7.28. Tilu Class 1 rims display the most variation in rim form, with Classes 2–5 showing less variation, in part due to unequal sample sizes (Table 7.15). Although Class 1 rims are most commonly straight, they can also be convex or, less often, concave; their rim profiles are usually parallel, but can also be lenticular, convergent, or divergent (Fig. 7.29). Class 1 lip profiles are almost always round but can be flat with round edges or pointed. Usually, extra lip features are absent on Class 1 rims but, occasionally, lips are symmetrically and asymmetrically thickened or grooved around the lip or the exterior of the lip (Fig. 7.30). Class 2 rims are most commonly straight, but can also be concave, and, rarely, convex. The concave rim course has seemingly been produced by the potter running her fingers around the inside of the rim wall while paddling the exterior to produce the bevel. Class 2 rims are always abruptly lenticular or abruptly convergent owing to this external bevelling. Their lips are round and occasionally display interior grooving. 124       · .  100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 Spit Class 1 Class 2 Class 3 Class 4 Class 5 Figure 7.25. Percentage of Madang style technical classes at Tilu, Unit 1 by excavation spit. The sole Class 3 rim from Tilu, Unit 1 has a straight rim course with a gradual lenticular profile. Its lip is flat with round edges and asymmetrically thickened on the exterior. It is worth noting that this rim (T-168) is identical to N-2702 from Nunguri in Spit 11, one of two that are the earliest Class 3 rims from that site. Class 4 rims are straight or concave, gradually divergent, thickening slightly towards the lip. All Class 4 rims exhibit internal beveling produced by running the finger or anvil around the interior lip. As such the lip is flat with either sharp or round edges, and never grooved or thickened. The single Class 5 rim at Tilu has a concave rim course with abruptly lenticular profile. Its lip profile is round and lacking extra features. and flat sharp lips. This implies that the technical process used to produce Class 4 vessels is more different than that required to produce Class 1, 2, 3, and 5 vessels. These MCA plots give a general appraisal of technical relatedness between classes and illustrate the relatedness of different technical elements. However, it can only be viewed as an appraisal, given the percentage of data variation explained by the first dimension (X axis) is 33.49%, and only 17.72% on the second dimension (Y axis). This is much less than would be acceptable for a principal component analysis using continuous data, but it is not unexpected for a MCA. Rim thickness in the Tilu, Unit 1 assemblage was assessed by technical class. Class 1, 2, 4, and 5 rims all overlap at one standard deviation (SD) from the mean (Table 7.16). No examples of sigmoid or abruptly divergent rims are However, the single Class 3 rim sherd is much thicker and present at Tilu, Unit 1. Sub-rim grooves are rare in the as- does not fall within the total range of thicknesses for Class semblage, only observed on two Class 1 rims and single 1, 2, 4, vessels (Fig. 7.32). The coefficient of variation (CV) Class 2 and Class 5 rims. There are no examples of sub-rim is similar for Class 1 and 2 at c. 15%, suggestive of a relagrooves with associated ridges. tively standardised thickness for these two classes, while Class 4 has a higher CV resulting from a small sample As per the Nunguri discrete attributes, an MCA was applied size. A Kruskal-Wallis test, using a Monte-Carlo simuto the Tilu data to demonstrate that specific technical ele- lation, shows that there is significant variation in mean ments were repeated and more associated with different rim thickness between Classes 1 and 2 (α = 0.05, X2 = 4.501, technical classes. This used the same FactoMineR package df = 1, η2 = 7.76%, P = 0.031). Class 3–5 rims were not tested (Le et al. 2008). The Tilu data plot slightly differently to owing to small sample sizes. the Nunguri set, in part owing to smaller sample sizes for Classes 3, 4, and 5. Here, the Class 2 and 5 rims plot closely Figure 7.33 describes the shape of the rim by plotting rim together (Fig. 7.31, bottom left), while Class 1 plots slightly and neck thicknesses. This shows that Class 1, 2, and 5 apart. The Class 3 rim plots closer to Class 1 than at Nun- overlap, and neck thickness increases with rim thickness, guri and Class 4 clusters further from the others owing although the rim is generally thicker than the neck. Class to its tendency to have gradually divergent rim profiles 3 rims are thicker than most Class 1, 2, and 5 examples 125 Chapter . Pre-Colonial Potting I: Production T-1/17/286 Spit 1 T-165 Spit 2 T-549/551 Spit 6 T-642 Spit 7 T-722 Spit 8 T-759/843/844 Spit 9 T-841 Spit 9 T-842 Spit 10 T-871 Spit 10 0 5 cm Figure 7.26. Class 1 Madang style rims at Tilu, Unit 1. 126       · .  T-2 Spit 1 T-6 Spit 1 T-164/166/296 Spit 2 T-295/407 Spit 3 T-647/648 Spit 7 T-725 Spit 8 T-755 Spit 9 T-749 Spit 9 T-747 Spit 9 0 5 cm Figure 7.27. Class 2 Madang style rims at Tilu, Unit 1. 127 Chapter . Pre-Colonial Potting I: Production Class 3 T-168 Spit 2 Class 4 T-5 Spit 1 T-136 Spit 1 (orientation approximate) T-173 Spit 2 Class 5 T-416 Spit 4 0 5 cm Figure 7.28. Class 3–5 Madang style rims at Tilu, Unit 1. 128       · .  Rim profile 12 12 10 10 8 8 6 6 4 4 2 2 0 2 3 4 5 6 7 8 9 10 3 4 5 6 7 8 9 10 12 10 8 8 6 6 4 4 2 2 2 3 4 5 6 7 8 9 10 Class 2 0 1 2 3 4 5 6 7 8 9 10 12 12 10 10 8 8 6 6 4 4 2 2 0 1 2 3 4 5 6 7 8 9 10 Class 3 0 1 2 3 4 5 6 7 8 9 10 12 10 10 8 8 6 6 4 4 2 2 1 2 3 4 5 6 7 8 9 10 Class 4 0 1 2 3 4 5 6 7 8 9 10 12 10 10 8 8 6 6 4 4 2 2 Class 5 1 2 3 4 5 6 7 8 9 10 MNV 12 0 MNV 12 0 MNV MNV 2 10 1 MNV 0 1 12 0 MNV Class 1 MNV MNV 1 MNV MNV Rim course 0 1 2 3 4 5 Spit 6 7 8 9 10 Spit Parallel Convergent abrupt Straight Convex Concave Convergent gradual Divergent abrupt Divergent gradual Lenticular abrupt Lenticular gradual Figure 7.29. Rim course and profile by technical class at Tilu, Unit 1 (note some rims not represented where attributes could not be confidently recorded). 129 Chapter . Pre-Colonial Potting I: Production Extra lip features 16 14 14 12 12 10 10 8 8 6 6 4 4 2 2 0 2 3 4 5 6 7 8 9 10 Class 1 0 1 2 3 4 5 6 7 8 9 10 16 16 14 14 12 12 10 10 8 8 6 6 4 4 2 2 0 2 3 4 5 6 7 8 9 10 Class 2 0 1 2 3 4 5 6 7 8 9 10 16 16 14 14 12 12 10 10 8 8 6 6 4 4 2 MNV MNV 1 2 0 1 2 3 4 5 6 7 8 9 10 Class 3 0 1 2 3 4 5 6 7 8 9 10 16 16 14 14 12 12 10 10 8 8 6 6 4 4 2 MNV MNV MNV MNV 1 2 0 1 2 3 4 5 6 7 8 9 10 Class 4 0 1 2 3 4 5 6 7 8 9 10 16 16 14 14 12 12 10 10 8 8 6 6 4 4 2 MNV MNV MNV MNV Lip profile 16 2 0 1 2 3 4 5 6 7 8 9 10 Class 5 0 1 2 3 4 Spit 5 6 7 8 9 10 Spit Absent Round Flat (round edge) Pointed Flat (sharp edge) Grooved exterior Asymmetrically thickened exterior Grooved interior Asymmetrically thickened interior Grooved Symmetrically thickened Figure 7.30. Lip profile and extra lip features by technical class at Tilu, Unit 1 (note some rims not represented where attributes could not be confidently recorded). 130       · .  Table 7.15. Rim and lip forms by technical class at Tilu, Unit 1 (note some rims not represented where attributes could not be confidently recorded). Class 1 Class 2 Class 3 Class 4 Class 5 n % n % n % n % n % Straight 48 60 10 56 1 100 2 67 – – Convex 19 24 2 11 – – – – – – Concave 13 16 6 33 – – 1 33 1 100 Rim course Rim profile Parallel 51 61 – – – – – – – – Convergent abrupt 3 4 6 33 – – – – – – Convergent gradual 7 8 – – – – – – – – Divergent abrupt – – – – – – – – – – Divergent gradual 5 6 – – – – 3 100 – – Lenticular abrupt 4 5 12 67 – – – – 1 100 Lenticular gradual 14 17 – – 1 100 – – – – 80 90 19 100 – – – – 1 100 Flat (round edge) 7 8 – – 1 100 1 33 – – Flat (sharp edge) – – – – – – 2 67 – – Pointed 2 2 – – – – – – – – Lip profile Round Extra lip features Absent 56 64 16 80 – – 3 100 1 100 Grooved 1 1 – – – – – – – – Grooved exterior 6 7 – – – – – – – – Grooved interior – – 4 20 – – – – – – Symmetrically thickened 8 9 – – – – – – – – Asymmetrically thickened exterior 14 16 – – 1 100 – – – – Asymmetrically thickened interior 3 3 – – – – – – – – 57 97 13 93 1 100 3 100 1 100 Grooved 2 3 1 7 – – – – – – Grooved with ridge – – – – – – – – – – Sub–rim groove Absent Table 7.16. Average rim and neck thickness (mm) by technical class at Tilu, Unit 1. Class 1 Class 2 Class 3 Class 4 Class 5 10.51 14.64 10.40 9.99 Rim orientation values are similar for Class 1–5 rims, overlapping at 1 SD (Table 7.17), but Classes 1–3 project away from the central vertical axis, while Classes 4–5 project towards it. Here the CV for Class 1, 2, and 4 rims are much greater than for rim thickness indicating more variation in orientation angle. Figure 7.35 plots rim inclination against Rim thickness Mean 9.87 SD 1.67 1.58 – 3.40 – CV 17% 15% – 33% – 7.43 9.99 N.A. 6.51 Table 7.17. Average rim orientation and inclination (°) by technical class, Tilu, Unit 1. Neck thickness Mean 6.98 SD 1.28 1.10 – N.A. – CV 18% 15% – N.A. – Class 1 Class 2 Class 3 Class 4 Class 5 31 21 25 35 Orientation Mean (Class 4 is not considered as neck thickness could not be measured). Figure 7.34 further describes rim shape, plotting rim length against thickness. This demonstrates that, again, Class 1, 2, and 5 overlap, while the Class 3 rim is distinctly thicker and shorter. 131 28 SD 11 14 – 21 – CV 41% 44% – 85% – 125 104 N.A. 159 Inclination Mean 120 SD 15 14 – N.A. – CV 13% 11% – N.A. – Chapter . Pre-Colonial Potting I: Production 0 1 Rim course 2 3 4 Rim direction Rim profile 1 Dimension 2 (11.93%) 0 -1 Straight Convex Concave Everted Class Incurving Parallel Convergent gradual Lenticular abrupt Lip feature Convergent abrupt Divergent gradual Lenticular gradual Lip profile 1 0 -1 0 1 2 3 4 Class 1 Class 2 Class 3 Class 4 Class 5 0 Absent Asymmetrically thickened exterior Asymmetrically thickened interior Grooved interior Grooved exterior Symetrically thickened 1 2 Round Flat (round-edge) Dimension 1 (16.41%) Figure 7.31. MCA of discrete formal attributes for Madang style rims, Tilu, Unit 1. 20 18 16 Rim thickness (mm) 14 12 10 8 6 4 2 0 1 2 3 Class 4 Figure 7.32. Rim thickness by technical class at Tilu, Unit 1. 132 5 3 4 Pointed Flat (sharp-edge)       · .  12 Neck thickness (mm) 10 8 6 4 2 0 0 2 4 6 8 10 12 14 16 Rim thickness (mm) Class 1 Class 2 Class 3 Class 5 Figure 7.33. Relationship between rim and neck thickness by technical class at Tilu, Unit 1 (note some rims not represented where attributes could not be confidently recorded). 16 14 Rim thickness (mm) 12 10 8 6 4 2 0 0 5 10 15 20 25 30 Rim length (mm) Class 1 Class 2 Class 3 Class 5 Figure 7.34. Relationship between rim length and thickness by technical class at Tilu, Unit 1 (note some rims not represented where attributes could not be confidently recorded). 133 Chapter . Pre-Colonial Potting I: Production 180 160 Rim inclination° 140 120 100 80 60 40 20 0 0 10 20 30 40 50 60 Rim orientation° Class 1 Class 2 Class 3 Class 5 Figure 7.35. Relationship between rim orientation and inclination by technical class, Tilu, Unit 1 (note some rims not represented where attributes could not be confidently recorded) orientation, showing Classes 1, 2, 3, and 5 clustering. The Table 7.18. Average orifice diameter (cm) by technical class, chart also shows that there is a general trend for rim incliTilu, Unit 1. nation to decrease as orientation increases. Orifice diameter also shows variation between technical classes. Class 1 orifices are 9–18 cm wide with a peak between 12–16 cm wide and mean of 14 cm (Table 7.18). This appears to follow a unimodal distribution with no clear subgroupings (Fig. 7.36). Class 2 rims are between 11–21 cm wide, with a mean of 15 cm. These also appear to follow a normal distribution, but the Class 2 samples with orifices 19–21 cm wide may either represent the tail of a unimodal distribution, or a bimodal distribution. A Kruskal-Wallis test using a Monte-Carlo simulation suggests that there are no significant differences between the means of Class 1 and 2 orifice diameters (X2 = 3.075, df = 1, η2 = 5.30%, P = 0.083).2 However, Mood’s test suggests that there is a significant difference between median orifice diameters for these two classes (Median = 14, X2 = 5.102, df = 1, P = 0.042). Sample sizes of Class 3, 4, and 5 rims were difficult to assess confidently owing to small sample sizes and so were not tested; however, the individual Class 3 and 5 rims are smaller than most Class 1 and 2 rims. Class 1 Class 2 Class 3 Class 4 Class 5 14 15 11 13 10 SD 2 3 – 1 – CV 15% 18% – 8% – Mean Exotics Two exotic rims were recovered from Tilu, Unit 1 (Fig. 7.37). Each appear to be derived from distinct technical traditions. T-188, from Spit 2, is a direct rim, projecting towards the central vertical axis, with straight rim course, parallel profile, and round lip. It has an orifice diameter of 10 cm and could represent a small restricted cup, or part of a longer vessel. T-410, from Spit 4, is a direct rim with straight rim course, parallel profile, and flat, sharp-edged lip with grooving. It has a larger orifice diameter of 18 cm and appears to be from an open bowl. Beyond this, neither sherd preserves significant evidence of manufacturing technique. Shovel Pit 1 2 Note that as Kruskal-Wallis tests assume similar distributions of data, a non-parametric Levene’s test was applied to From Shovel Pit 1 at Tilu a small selection of rims were determine that the distributions were not statistically different collected from various depths below surface (Table 7.19). (F = 1.393, df = 4, P = 0.248). These are all assignable to Madang style technical Classes 1, 134       · .  14 12 MNV 10 8 6 4 2 0 9 10 11 12 13 Class 1 14 15 16 Orifice diameter (cm) Class 2 Class 3 17 18 Class 4 19 20 21 Class 5 Fig. 7.36. Vessel orifice diameter by technical class, Tilu, Unit 1. T-188 Spit 2 T-410 Spit 4 0 5 cm Figure 7.37. Exotic rims from Tilu, Unit 1. 2, or 3. One instance of a Class 3 rim is present from 60 cm Table 7.19. Technical classes represented in Tilu, Shovel Pit below surface, which is slightly deeper (>20 cm) than that 1 by MNV and depth. from Unit 1. These rims were excluded from the analytical Technical class Unit 1 sample owing to an uncertain chronology. No exDepth 1 2 3 4 5 Exotic Total otic rim forms were recovered, but a distinct ring base was 45 cm 6 1 – – – – 7 found from 60 cm below surface (Fig. 7.38). It is not assignable to the Madang style. This sherd has a base diameter of 60 cm 2 – 1 – – – 3 4 cm, standing the body about 0.5 cm off the ground. 70 cm – – – – – – – 135 80 cm – 1 – – – – 120 cm 1 – – – – – 1 1 Total 9 2 1 – – – 12 Chapter . Pre-Colonial Potting I: Production 0 5 cm 5 0 cm Figure 7.38. Ring base from 60 cm below surface, Tilu, Shovel Pit 1. Table 7.20. Technical classes represented in surface collections by MNV and location. Results: Surface survey In total, 32 rims were collected from surface survey around Madang. This includes seven from Kranket Island, four from Siar Island, six from Yabob Island and 15 from Tilu at Malmal Village (Table 7.20). Almost all (n = 31) of these rims fit firmly within the Madang style, being assignable to Classes 1–5, but one sherd, from a sandy inlet on the north side of Yabob Island (see Chapter 6) is distinct and appears to derive from a different technical tradition (Fig. 7.39). This sherd (Y-5) is red-slipped on the external surface and appears to come from an outcurving bowl with concave rim course, gradually divergent rim profile and round lip, similar to the exotic rim (T-188) from Tilu, Spit 2. 136 Technical class Site 1 2 3 4 5 Exotic Total Kranket Is. 5 0 2 0 0 0 7 4 Siar Is. 1 1 1 0 1 0 Yabob Is. 2 2 0 1 0 1 6 Malmal 9 4 1 0 1 0 15 17 7 4 1 2 1 32 Total       · .  K-4 Surface find, Kranket Island (Class 1) K-2 Surface find, Kranket Island (Class 1) K-3 Surface find, Kranket Island (Class 1) S-1 Surface find, Siar Island (Class 2) K-1 Surface find, Kranket Island (Class 3) S-2 Surface find, Siar Island (Class 3) Y-1 Surface find, Yabob Island (Class 4) Y-5 Surface find, Yabob Island (Exotic?) S-3 Surface find, Siar Island (Class 5) 0 5 cm Figure 7.39. Rims collected from surface survey, Madang 2014. 137 Chapter . Pre-Colonial Potting I: Production Summary This chapter has described the diversity of pre-colonial ceramic chaînes opératoires to start to distinguish the production groups working within broader communities of practice. The formal sherds represented at Nunguri, Tilu, and from surface survey show that about 96% of the ceramic vessels consumed locally are assignable to the Madang style. This Madang style is a technological tradition, composed of a number of distinct technological classes. Class 1 and 2 are almost identical in shape except for an additional bevelling stage in forming. Class 3 vessels are thicker, especially around the rim and produced by an extra stage of paddling around the interior of the rim. This is a procedure used by modern Bel potters. Class 4 vessels are incurving and lack a neck, similar to modern Magob, but where the Magob’s rim profile diverges towards the lip, Class 4 rims are often parallel. It is not clear whether pre-colonial Class 4 vessels began with a rim-preform like modern Madang pots. Class 5 vessels are inverted and probably represent water-pots with restricted necks. As the MCA plots hint at (see Fig. 7.17 & Fig. 7.31), Class 1, 2, and 5 rims represent more similar technical processes than do Class 3 and 4 rims. This may indicate that Class 1, 2, and 5 were produced in the same production groups, while Class 3, and 4 derive from related but different groups. Although there were five technical classes in pre-colonial Madang style pottery which were repeated in production through time, within each class there was a lot of variation in form. This variation gives insight into innovations within production groups and demonstrates a great deal of leniency for individual potters to create difference within more regulated technological processes. It is also relevant that Class 3, 4, and 5 occur later in the sequence at Tilu and Nunguri, perhaps reflecting changing preferences in forming technology across generations. It may also represent interaction and knowledge exchange between production groups. The next chapter will continue the ceramic classification by examining procurement & distribution. This work attempts to distinguish the number of production groups making Madang style pots, including the source of their raw materials and the distribution of their pots around the Madang area. The results will further tease apart some possibilities raised in this chapter, such as if certain chaînes opératoires were used by specific production groups within a broader community of practice, or if multiple groups engaged with multiple chaînes opératoires, and if Classes 3, 4, and 5 represent introductions from distinct production groups or internal innovations. 138 Chapter 8. Pre-Colonial Potting II: Procurement and Distribution studies, supplemented with ethnographic observations of modern potting around Madang and reference samples of known clay and temper sources (see Chapter 4). Next, the method of sampling, chemical analysis using scanning electron microscope (SEM), and statistical treatment of Geochemical characterisation of ceramic fabrics has the the data are presented. Finally, the results of the chemical potential to delineate changes to raw material procure- compositional analysis are described in detail, including ment and discriminate between local and exotic products. macroscopic fabric analysis and chemical analyses of minThis chapter is the second stage of the technological analy- eral inclusions and clays. This work illustrates how changsis laid out in Chapter 7, and uses geochemical techniques es in technological processes and exchange relationships to address questions about procurement and distribution. are reflected in the archaeological ceramics. The analysis of these stages of the chaîne opératoire is a necessary step to assess changes to production and ex- Methodological issues change over time. Foundations The study of the chemical composition of Madang ceramics has two specific aims: (1) to establish the incipience of Amongst the Bel, the manufacture of pots involves several local production and subsequent adjustments to techno- essential materials. That includes clay, non-plastic minerals, logical organisation in terms of changing resource pro- slip, water, and firing fuel, along with several tools used to curement, and (2) to assess how changes occurred in local procure these materials: the digging stick for extraction, baand regional pottery distribution over the last 500–600 nana leaves to lie the raw clay on, string bags for transport, years. These aims feed into broader research objectives in- and so on. Two of these constituent parts – clay and nonvestigating the history of the Bel around the mainland of plastic minerals – can be examined mineralogically and New Guinea and subsequent movements, interactions, and chemically to classify archaeologically useful groupings. exchanges. For instance, if there was a single migration of Bel speakers to the coast we would expect the earliest pot- The first of these, potting clay, consists of reconstituted tery to be predominantly made from local raw materials, aluminium silicates, formed from the decomposition of but if there was more gradualistic diaspora and interaction non-plastic minerals (Guggenheim & Martin 1995). Genbetween an insular homeland and the New Guinea main- erally, clay is derived from common rock types, which land we might expect to see local and exotic raw materials break down into different mineral grains such as feldspars, used to produce Madang style pots, followed by the re- spinels, micas, quartz, amphiboles, and pyroxenes (Deer placement and eventual complete cessation of exotic ma- et al. 1992: 354). When these grains are sufficiently weathterials. Following initial Bel settlement, we can also explore ered into particles less than 2µm in size, they are defined if forming techniques became associated with specific raw as pure clay minerals: illite, kaolinite, smectite, or chlorite material zones used by certain production groups, or if (Velde & Meunier 2008: 3). Naturally occurring clays are multiple Bel groups, using different raw materials, engaged usually combinations of these pure clay minerals. More with various forming techniques. resilient grains are often not completely broken down but remain in the clay as non-plastic inclusions or are redeTo address these aims, the chapter first presents the meth- posited through wind and water action in the sediment odological links to assess procurement and distribution (Arnold 1974). through chemical compositional analysis. This is based on technological theory and previous ethnoarchaeological For clay to be workable it cannot be overly plastic. Although most potting communities in New Guinea do not need to alter their clays (May & Tuckson 2000: 22), the 1 Quoted in Mennis (1981b: 79). [The Bilbil] would come to get clay for their pots. They would beach their canoes and fill them with the clay. – Nagida of So village (1978)1 139 Chapter . Pre-Colonial Potting II: Procurement and Distribution Bel potters manually add temper owing to the paucity of naturally-occuring non-plastic inclusions in their clays. As presented in Chapter 4, modern Bel potters use local black beach and river-mouth sands as tempers. Manual tempering can also make the ceramic more resistant to thermal shock during initial firing and cooking by reducing shrinkage and can prevent the clay cracking when the water content evaporates (Rye 1981: 16). Summerhayes 1987). Since then, geochemical studies of pottery have come into the mainstream, now being seen as a necessary step in any analysis, especially as analytical machines become more widely accessible (e.g. Cochrane & Neff 2006; Eckert & James 2011; Gaffney et al. 2015, 2016, 2019; Golitko 2011; Golitko & Terrell 2012; Hogg 2007; Leclerc 2015; Shaw 2014; Shaw et al. 2016; Summerhayes 1997, 2000; Sutton 2016; Sutton et al. 2016; Teele 2011; Vilgalys & Summerhayes 2016). Despite this progress, theoPacific temper sands are broadly defined as terrigenous, retical developments have not kept pace with advances in deriving from exposed bedrock, or calcareous, which rep- materials science (Summerhayes 2015). resents calcium carbonate debris from marine reef systems (Dickinson 2006: 3). Terrigenous tempers are particularly The general aim of characterisation studies in New Guinea useful in describing source locations, forming diagnostic archaeology has been to distinguish the geographic orcomponents of their original bedrock derivations (Dick- ganisation of pottery production and distribution and to inson 1998). Although calcareous tempers present more of measure how this organisation changed through time. As a problem owing to the uniformity of reef deposits, their noted above, this includes the number of production cenpresence in a pottery fabric is usually a good indicator of tres, the number of sources or resource zones, the proxmanual tempering because of their relative absence in clay imity of production groups to these sources, the proximdeposits. The firing temperatures reported for the New ity of the consumer to the production group, and so on. Guinea region, no higher than 1000°C, do not usually af- These factors are usually then built into broader models of fect the mineralogy of terregenous inclusions (Dickinson changing procurement strategies, resulting from changes 2006: 4). to mobility or restrictions to territory, and changing distribution patterns, resulting from the reorganisation of Although grog tempering is a technique unknown to mod- production and exchange networks. However, some charern Madang potters, it has been reported in archaeological acterisation studies lack bridging concepts between obMadang style ceramics (Specht et al. 2006). Grog consti- served chemical data and higher-order theories and modtutes small grains of crushed pottery reused in the pro- els. That is, there is a lack of middle-range theory in the duction process, or small grains of fired and crushed clay form of experimentation or analogy (Arnold 2015). How, specifically designed to temper clays (Vincent 1988: 88). for instance might we interpret a reduction in the number of clay sources being used at a certain time, or a change Thus, within the clay matrix of the archaeological ceramic, from calcareous to terrigenous tempers? non-plastic mineral or rock grains may be present as naturally occurring fragments from the underlying geology Technology as methodology or may have been manually added as a temper. Conventionally in archaeology the two lines of evidence, clay and There is not necessarily an inherent link between technotemper, are compared to known reference material and logical choice, production and consumption groups, and geological information to ‘triangulate’ the probable manu- mineralogical/compositional variability, so one must demfacturing areas of different groups (Arnold 1980; Bishop et onstrate, rather than assume, that the physical properties al. 1982). This is the basic aim of such studies and allows of the archaeological potsherd have social significance important variables to be understood: the distance from (Arnold 1998, 2000; Summerhayes 2015). In Madang, as production group to source or resource zone; the number elsewhere in the Indo-Pacific (see Aronson et al. 1994; of sources used by production groups; and the distance Stark et al. 2000), the technological decision-making unfrom one group’s sources to other production groups or derlying the selection of raw materials has been shown to settlements. be equally affected by socio-political factors such as personal and social relationships or range of mobility, as it is by technical factors such as clay workability (see Chapter Geochemical analyses and New Guinea ceramics 4; Mennis 2006a: 81). Similar social concerns are in play as To better understand production and distribution, ceramic finished products are then distributed from production technologists have long advocated for the supplementa- groups to consumers. tion of formal and stylistic analyses with detailed studies of clay and temper (Arnold 1985; Rice 1982: 51; Rye But is it possible to distinguish these relationships through 1981; Rye & Allen 1980; Shepard 1980: 320). At the start of the chemical analysis of pottery? By asking these types of the 21st century, Summerhayes (2000: 30) noted that in questions we begin to problematise the assumptions that the Pacific, although temper analyses had become com- underlie geochemical analysis, which is used so frequently monplace in the Pacific particularly through the masterly to understand past technologies. Here, the methodological work of the late Bill Dickinson (e.g. Dickinson 1998, 2006; framework of the chaîne opératoire informed by ethnoDickinson & Shutler Jr. 1971), clay compositional stud- graphic observation is used to bridge the raw chemical ies were largely lacking (exceptions being Ambrose 1992, data and the broader concerns of this monograph regard1993; Anson 1986, 1999; Irwin 1985; Lilley 1986; Hunt 1989; ing the nature of production groups in pre-colonial Ma140       · .  dang and the technological processes operating within and between these groups. More simply, this chapter addresses how pottery procurement and distribution was organised in the past, and the possible reasons for specific technological choices within a myriad of possibilities (following Tite 1999). and tempers are archaeologically visible as technical attributes: the mineralogical inclusions and chemical signatures of the potsherd. Other learned syntaxes, such as how much clay to extract, how to dig it from the ground, and the manner of transport are not necessarily visible as technical attributes. Procurement The Bilbil potters collect clay from two chemically distinct pits within close proximity of each other (see Chapter 4). As an alluvial clay underlies much of Bilbil land, it is apparent that clay selection is constrained not by the natural occurrence of the raw material. However, the distance from the village to the source, kin relations with the landowners, and the location of good quality gardening plots all appear to guide decision-making. In contrast, Yabob potters used six clay sources. It is possible that some of these clays were used for wheel thrown pottery when the industry changed briefly in the 1970s and it is not known how many sources were in use by either Bilbil or Yabob prior to European disruptions to pot production. However, it is worth noting that the discrepancy in the number of clay sources in use does not relate to range of territory or restrictions on access to raw material, but may relate to the intensity of collection and the standardisation of procurement (Bilbil nowadays pot more frequently than Yabob), or other factors. It is also important to note that changes to chemical pastes or slightly different tempers do not necessarily imply distinct changes to the organisation of procurement (Arnold 2000) We return, again, to the concepts of technological process and embodied knowledge. Very few studies have attempted to integrate similar theoretical concerns and ceramic compositional analysis (exceptions being Jones 2002, 2004; Sillar & Tite 2000; Stark et al. 2000; Tite 1999, 2008). It is clear that procurement is one technological phase of the ceramic chaîne opératoire and as such must be composed of a series of socially specific techniques and material engagements. The social and cognitive processes informing which clay and temper sources the potters select should therefore be produced and reproduced in a similar way to those processes guiding the forming stage. As a technological process, raw material procurement involves the interplay between intention, gesture, and matter. The regular engagement with this interplay – the routine of collecting clay and temper – required the repeated movement from the household to the raw material, the digging and extraction of raw materials, and the transport of the material back to the household for forming. These actions all represent technical elements and syntaxes that are informed by the learning process and become embodied through habitual action. Such routines would have produced a specific kind of embodied and technological knowledge, a procedural knowledge that only became apparent when inadequate materials were encountered, the source was exhausted, or a new source was sought. Although the emic selection of raw materials is partly based on physical properties such as colour and workability (Arnold 1971), the specific place (or source) for procurement of these materials is as much a part of the technological choices made (Jones 2002: 87). The materials and the source location are enmeshed in culturally significant concepts of place. In this way, the choices to procure different raw materials from different locations are not only women-made transformations of the local landscape (Arnold 1985: 17), but socially constituted and culturally meaningful decisions related to the potters’ experience of becoming, learning, and remembering, within the local landscape. Around Madang, young Bel girls observe and assist their mothers, aunts, grandmothers and other experienced potters revisiting the same trusted deposits. The Bilbil clay sources, for instance, have been in use for three generations at least. The experienced potters also instruct, discursively or otherwise, which properties to be aware of when selecting the raw material. Many of these technical syntaxes such as the repeated collection of the same clays Distribution As mentioned in Chapter 7, technological processes do not become inert after the procurement and production stages (Pelegrin 1990; Sellet 1993; Sillar & Tite 2000). Rather, ceramic technology continues to create meaningful human engagements throughout distribution, consumption and discard. However, as technologies pass through different stages in space and time, and objects pass through different social groups and communities of practice, they acquire and create different meaning (Knappet 2005a, 2005b). In the case of the Bel, ceramic production would not exist without the socially and economically enticing stages of distribution and consumption. It is in these stages that locally produced ceramics can be exchanged for food or a variety of objects, providing the necessary excuse to maintain social bonds and remain dominant on the playing field (see Welsch & Terrell 1998). Owing to the reciprocal dependency of trade friends that characterises New Guinea exchange (Mauss 1925; Thomas 1991: 14), including the Madang network (Mennis 2006a), the identification of exotic ceramics implies meaningful interrelations with other social groups, which were presumably maintained, if possible, through time. The ways in which humans interact with ceramics throughout distribution and consumption can also be viewed as embodied processes, produced through habitual engagements. In the case of Bel distribution, technical syntaxes would incorporate the loading of pots into balungut 141 Chapter . Pre-Colonial Potting II: Procurement and Distribution canoes, the storage of these pots on the balungut, stacked up in pyramids, the sprinkling of broken pots into the sea to quell malignant spirits, and the technical knowledge of sailing. The technological processes of ceramics and canoes become more conspicuously entwined at this point. Importantly, the distribution phase of the Bel ceramic chaîne opératoire moves from the female domain to the male domain. Many of these technical choices relating to distribution cannot be assessed through compositional analysis. One set of choices that can be assessed, however, relates to the groups receiving and distributing goods locally around Madang, along with possible exotic materials being introduced into the area. Lilley’s (1986) work in particular has already demonstrated that Bel ceramics were being exchanged all along the northeast coast prior to Europeans. The current study, limited to several sites around Madang, cannot assess the distribution of Bel ceramics to regional groups, but it can give insight into other types of ceramics being exchanged into, and consumed by, the Bel groups. In this sense the research questions pertaining to distribution are: what was the organisation of local Bel distribution, and, perhaps also, why were certain types of pots distributed to different groups. sitionally as the resource spaces of each group are unlikely to overlap. His model assumes that the resources used by specific production groups are distributed as patches. In the case of local exchange, however, it will only be possible to distinguish when resource spaces of different production communities do not overlap. There are interpretive problems that arise if these spaces do overlap. Figure 8.1. illustrates four scenarios of local production in which trade and exchange will be convoluted owing to an overlap in resource spaces. In each of these scenarios, both production communities could feature pots with the same raw material signatures. Ethnographic data suggest that the Bilbil and Yabob procurement zones do not overlap and that they should be chemically distinguishable (see Chapter 4). However, in the past, resource spaces and the number of production groups may have been different. Method The Madang classification To address these questions about procurement and distribution, the Madang ceramics will be further classified. A ceramic assemblage can be viewed as the product of one Arnold (1980) makes the point that regional and sub-re- or more production groups (Roux & Courty 2005) opergional exchange should be possible to distinguish compo- ating within broader communities of practice. The aim a) b) c) d) procurement raw material exchange production group Figure 8.1. Several scenarios in which local exchange is difficult to chemically distinguish. 142       · .  of the present classification is to identify those specific production groups who were using distinct raw materials within the broader community practice of forming and decorating pots using similar chaînes opératoires. Most compositional studies attempt to distinguish interregional variation in raw materials to infer distribution. It is less common for such studies to attempt to distinguish procurement and distribution at the level of the local production group. However, some studies on coarseware ceramics have been successful at delineating local variation (see Stark et al. 2000). sified based on geochemical assessments of each fabric’s inclusions, and, lastly, techno-compositional groups are created by a geochemical and statistical analysis of the clay matrix itself. By examining clay and temper sources and paste preparation techniques, these technical classes indicative of specific forming processes may be subdivided to delineate specific production groups engaging with these technological processes. The specific analytical procedures are described below. Here, the classification of Madang style ceramics continues the hierarchical typology outlined in Chapter 7 (adapted from Roux 2011). This approach seeks to identify ceramic chaînes opératoires through eight successive procedures (Fig. 8.2). This chapter deals specifically with procedures 5–7 and will distinguish techno-compositional groupings within the formal technical classes (1–5) assigned in Chapter 6. First, techno-fabric groups are provisionally sorted optically, then, techno-mineralogical groupings are clas- All formal and decorated sherds were first examined optically under 40x magnification for mineral inclusions to allow for primary sorting into different techno-fabric groups. This techno-fabric study was designed to give a general appraisal of the fabrics only, and allow for preliminary classification, facilitating further geochemical analysis. All Madang style sherds comprised similar or the same types of inclusions, but in varying proportions. Allocation into groups was based on visual estimates of the mineral Macroscopic fabric analysis Ceramic assemblage Intuitive sorting Provisional style groups Hierarchical sorting based on technical syntaxes Technical classes Classification based on variation in technical elements Technical variants Sorting based on ocular microscopy Techno-fabric groups cross comparison with mophological and decorative attributes Classification based on mineralogial identification Techno-mineralogical groups Grouping based on clay composition Techno-compositional groups Techno-style groups Figure 8.2. Procedures of a technological classification (adapted from Roux 2011). Those procedures examined in this chapter are highlighted yellow. 143 Chapter . Pre-Colonial Potting II: Procurement and Distribution Table 8.1. Macro-tempers of Madang style ceramics. Macro-fabric group Dominant inclusions Possible minor inclusions 1- Ferromagnesium Fe oxides/pyroxene Quartz, feldspars, coral, shell, grog, lithic 2- Light Quartz/feldspars Fe oxides, pyroxene, coral, shell, grog, lithic 3- Calcareous Coral/shell Quartz, feldspars, Fe oxides, pyroxene, grog, lithic 4- Grog Grog Quartz, feldspars, Fe oxides, pyroxene, coral, shell, lithic grain types of greatest abundance in each sherd. A total of four macro-tempers were identified (Table 8.1). Clay colour was also recorded for all rim sherds, which likely reflects both geochemical variation in clay sources, occasional mixing of pastes, and differential firing temperatures (see Chapter 4, Table 4.1–4.2). This assigned rims into several informal categories: orange, red, brown, yellow, and grey. To reduce intra-observer error in the analysis, each of the informal categories was assigned a Munsell colour reference, which would then be used as a control (orange = 5YR 5/6; red = 2.5YR 4/6; brown = 5YR 4/6; yellow = 10YR 6/6; grey = 10YR 6/4). Black and grey surface alteration can be caused by firing, cooking, or re-use as su. Under oxidising firing conditions, discolouration of the potsherd’s core in cross-section is primarily owing to the presence of organic material (Rye & Allen 1980). Both surface and core blackening were recorded as presence or absence attributes. stratified sampling procedures, using decreasing orders of inclusiveness: site, spit, vessel portion, technical class, techno-fabric grouping and clay colour. This sampling procedure applies to those stratified sherds assigned to the ‘Madang style’ at Nunguri and Tilu only. Overall, 100 Madang style sherds from Nunguri and Tilu were sampled from excavated contexts to provide a robust temporal dimension to raw material procurement. Owing to the small proportion of sherds being selected compared to the total recovered in excavation, comprehensive random sampling was not considered viable as it insufficiently represented rare, but potentially important, technical classes. The guidance of selective sampling was preferred as it ensured that the widest range of fabrics and forms was selected (see Summerhayes 1987: 138; Wilson 1978). An equal number of sherds were selected from each excavated site: 50 from Nunguri and 50 from Tilu (Table 8.2). These numbers allow for a comparable investigation of geochemical variability at both sites. Five sample spits were chosen in each excavation. These spits were chosen deliberately to preference those with associated radiocarSampling procedures for geochemical analysis bon age determinations, while also selecting from approxiTo further establish groupings, a total of 121 sherds was mately regular depth intervals. Ten sherds were selected selected for geochemical analysis. It is necessary to expli- from each of those five spits. The sampling was limited to cate clearly the method of selection, from which interpre- rim sherds only, which formed a discrete population from tations of chemical and mineralogical variability in each which to select. This approach allowed for comparisons assemblage will be based. Sherds were chosen by selective between composition and technical class. Sampling re- Table 8.2. Madang style sherds selected from Nunguri and Tilu for geochemical analysis. Number of sampled sherds with MNV present in each spit in brackets. Nunguri Spit 2 5 Class 1 Class 2 Class 3 Class 4 Class 5 Total 2 (78) 2 (28) 2 (21) 2 (12) 2 (2) 10 (141) (210 BP) 3 (14) 2 (5) 3 (11) 2 (5) – 10 (35) 11 (495 BP) 2 (10) 3 (13) 1 (2) 3 (13) 1 (1) 10 (39) 13 (530 BP) 4 (15) 3 (5) – 3 (10) – 10 (30) 15 4 (23) 3 (9) – 3 (4) – 10 (36) 15 (140) 13 (60) 6 (34) 13 (44) 3 (3) 50 (281) Total Tilu 2 5 (16) 3 (4) 1 (1) 1 (1) – 10 (22) 4 (630 BP) 9 (10) – – – 1 (1) 10 (11) 6 (595 BP) 8 (10) 2 (2) – – – 10 (12) 7 (585 BP) 7 (11) 3 (4) – – – 10 (15) 8 (10) 2 (3) – – – 10 (13) 37 (57) 10 (13) 1 (1) 1 (1) 1 (1) 50 (73) 9 Total 144       · .  strictions occurred in the basal deposits of Nunguri, Layer 2, Spits 16–17, and Tilu, Layer 2, Spit 12, in which there were fewer than ten rim sherds each. These spits were not sampled from. Rim selection was a broadly proportional representation of each technical class in each spit. Thus, if there were twice as many Class 1 rims present in a particular spit, then twice as many would be sampled. However, where very few examples of a particular class were present, they were purposively sampled so the complete diversity of rim forms could be included. Within each technical class, examples of each macro-temper and clay colour were selected to account for as much variability as possible. The minimum number of vessels (MNV) was calculated through metrical comparisons and refitting to ensure each rim sampled was from a different vessel (see Chapter 7). analysed to give a modern baseline of raw materials in use around coastal Madang (see Chapter 4). Further, two sand samples from basal deposits at Nunguri (Test Pit 1, Spit 17) and Tilu (Unit 1, Spit 12) were also examined to demonstrate the immediately available sands at the two sites. By establishing a chemical and mineralogical ‘fingerprint’ for the Madang area before then examining artefactual material, the procedures in this chapter follow the ideal approach laid out by Earle and Erickson (1977). Geochemical characterisation A geochemical analysis of the sampled sherds was undertaken to describe techno-mineralogical and technocompositional groups. Quantitative chemical analysis of non-plastic minerals and the clay matrix was completed To provide a spatial dimension to clay and temper vari- on a Zeiss Sigma Field Emission Gun Scanning Electron ability in the Madang area, 13 rim sherds were also sam- Microscope (SEM) in the Otago Centre for Electron Mipled from surface collections at Kranket, Siar, and Yabob croscopy (OCEM), using an XMax20 Silicon Drift Energy islands. This represents over 80% of surface-collected rims Dispersive Spectrometer (EDS) detector and AZtec acquifrom these islands (Table 8.3). sition and processing software from Oxford Instruments. The SEM produces high-resolution images and detects All eight sherds identified as possible ‘exotics’ in the formal contrasts between areas with different elemental composianalysis (see Chapter 7) were sampled in total to provide tions and surface topographies. The EDS allows spot-point insight into the variety of fabrics being imported into Ma- analysis and determines the chemical composition of a dang in the past. This includes one rim from Nunguri Test point or zone on the sample to define which elements are Pit 1, two rims and three decorated body sherds from Tilu present in that location (Froh 2004). Clay and mineral inUnit 1, a ring base from Tilu Shovel Pit 1, and one rim from clusions can then be separately analysed and considered as Yabob Island (Table 8.4). two different but complementary datasets. The spot point ability of the SEM is used in preference to other methods Samples of ethnographic clays from Yabob (1–6) and Bilbil such as NAA, PIXE-PIGME, LA-ICP-MS, and XRF which (1–2), fired at 700°C and 1000°C, and several local beach produce chemical noise and erroneous values when clay sands (Bilbil Island, Bilbil Village, Yabob Village) were also and mineral inclusions are undifferentiated (Zuluaga et al. 2011). The SEM therefore has the added benefit of allowing quantitative analysis of the heterogeneity in prehistoric Table 8.3. Madang style rim sherds from surface collections ceramics (Brissaud et al. 1985). for geochemical analysis. To produce scanning electron micrographs, the SEM bomProvenance Class 1 Class 2 Class 3 Class 4 Class 5 Total bards the sample with a focused electron beam; in this Kranket Is. 5 – 2 – – 7 instance with 15kV Extra High Tension (EHT) acceleratSiar Is. 1 – – – 1 2 ing voltage. A working distance of 8mm optimised the Yabob Is. 2 1 – 1 – 4 production of high-quality images. This distance, from Total 8 1 2 1 1 13 the polepiece to the sample, affects the scanning angle of the electron beam, and hence depth of field and resolution. When the electron beam interacts with the sample, it dislodges backscattered electrons (energies between the incident beam and 50eV) and secondary electrons (enTable 8.4. Exotic sherds from excavation and surface ergies less than 50eV) that are used for image formation collection for geochemical analysis. (Reed 2005). Further, the interaction between the electron Decorated beam and nuclei produce x-ray photons with characterProvenance Spit Rim body Base Total istic wavelengths, which are used for elemental analysis. 2 1 1 2 When the EDS detector absorbs an x-ray photon, a voltage 4 1 1 pulse is produced with peaks proportional to the x-ray Tilu, Unit 1 5 1 1 energy received. These pulses are digitised and stored in 6 1 1 a multichannel counter that produces a spectrum of all Tilu, Shovel Pit 1 60 cm 1 1 pulse heights simultaneously (Froh 2004). Nunguri, Test Pit 1 Yabob Island Total 15 1 Surf. 1 4 1 1 3 1 8 To measure x-ray energy accurately, the target must be set at exactly 90° to the beam. Thus, all samples must be completely flat and polished with no surface irregulari145 Chapter . Pre-Colonial Potting II: Procurement and Distribution ties (Reed 1975: 177). All sherds were first comprehensively photographed, drawn and recorded before any destructive analysis took place. Samples were prepared by cutting the selected sherds on an 8-inch diamond blade and impregnating them with epoxy resin (Hillquist Thin-section Epoxy A and B) to form briquettes, which were polished to 1 micron, ultrasonicated in distilled water for 5 minutes, and cleaned with ethanol. These briquettes were then coated with carbon (using an Emitech K575X Peltier-cooled High Resolution Sputter Coater and Emitech 250x Carbon Attachment), which acts as a thin (10 nm) conductive layer to ensure an electrical path to ground. This prevents the sample ‘charging’ under the electron beam and producing an overexposed image. Carbon coating is preferred owing to its low atomic weight and minimal effect on x-ray intensities (Summerhayes 2000: 38). changes. Light metals such as Fe, Mg and Al can be affected by increased firing conditions. The water used to pot may add S, Mn, Cl, Br, and Fe (Allen & Rye 1982). Furthermore, several elements (Na, K, Mg, Ca, Ba) can leech out of or enrich into ceramics owing to different post-depositional conditions (Freeth 1967; Golitko et al. 2012). Although none of these alterations is usually sufficient to significantly alter chemical groupings, they must be borne in mind to address potential variation in the data. Factors that do significantly affect paste variability are usually natural geological variations in clay sources (Arnold 2000). Within homogenous geological zones, micro-variations in weathering, climate, and parent strata will produce a chemically heterogeneous resource zone (see Chapter 2, Fig. 2.5). The eight elements were converted to oxide weight, normalised, and exported. The spectra outputs provided the raw chemical data, which were then Two areas of each sample were selected for image cap- statistically transformed to allow for interpretation of clay ture and elemental analysis: one electron micrograph at compositional variability. 100x magnification for the non-plastic mineral inclusions analysis; and one at 2000x for clay matrix analysis. This Statistical treatment of the clay chemical data approach attempted to target areas of the sherd that were oxidised, but the elemental analysis, along with the pXRF The central aim of statistical manipulations of the clay data groupings presented in Chapter 4, indicate that there is no is to investigate underlying patterns and create groupings significant differentiation of results between the oxidised that make mathematical, chemical, and archaeological and non-oxidised areas of clay. Two different methods of sense (Beiber et al. 1976; Harbottle 1982; Summerhayes analysis were used: (1) A spot-point analysis of the clay 2000: 39). Here Hierarchical Cluster Analysis (HCA) and matrix which collected elemental data at five points per Principal Component Analysis (PCA), performed with view, and (2) map-scanning, which obtained x-ray data for SPSS ver.21, were used to construct techno-compositional an entire 100x micrograph rather than particular points. groupings from the clay data. These approaches have previMap-scans display the number of x-rays detected from ously been effective in distinguishing chemical groupings specific energy ranges as phases, represented by colour- in Pacific ceramic studies (e.g. Gaffney et al. 2015; Garling coded pixels. These images are useful in visually discrimi- 2007; Shaw 2014; Summerhayes 1987, 2000). However, it is nating similar inclusions in the clay matrix. important to justify clearly the choice of clustering analyses used. Many methods will generate clusters even when Each spot-point collected data for 10 seconds with a pro- applied to random data (Dubes & Jain 1979), and some cessing time of 3 seconds, using pulse pile-up correction groupings can reflect the structure of the clustering algoto eliminate peak overlap issues. In map-scanning, peak rithm rather than patterns in the chemical data itself. On overlaps were eliminated using ‘TruMap’; an algorithm that basis, multiple manipulations were undertaken and that runs in real time to automate peak deconvolution. The cross-compared to ensure internal consistency. TruMap was run for two cycles (10 minutes), with a pixel dwell time of 300µs and process time of three seconds. HCA attempts to group points in multidimensional space Chemical spectra were then collected from these map- and investigate underlying structures in the clay chemiscan phases and compared to the chemical composition cal data (Dubes & Jain 1976), with the premise that clays of known reference samples to identify the mineral (using of similar derivations will form clusters (Summerhayes Deer et al. 2013). 2000: 40). Agglomerative HCA methods first plot all of the samples as points in multidimensional space (the number For clay analysis, the spectra were limited to eight key ele- of dimensions being equal to the number of elemental ments: Na, Mg, Al, Si, K, Ca, Ti, and Fe. Similar studies (e.g. variables). These individual points represent single clusHogg 2007; Summerhayes 1987, 2000) have found these ters that are then merged with the next closest point in lighter elements more useful than attempting to assess space to form more inclusive clusters. Pairs of clusters are trace elements using EDS. Two elements, Cl and C, were successively merged as one moves up the hierarchy (Johnexcluded from the spectrum. Cl was excluded because it son 1967). The results of hierarchical clustering are then was the major constituent of the epoxy, which might be a presented as dendrograms. cause of contamination. C was excluded because the samples were coated in this element. Here, two agglomerative methods – the Group Average method and Ward’s method – were chosen to group samOther elements are sometimes excluded from clay spectra. ples into clusters. The Group Average method defines This is because several cultural factors can affect paste vari- distance between groups as the average of the distances ability: firing, mixing with water, and post-depositional between every pair of points within each group (Everitt 146       · .  1977: 15). In this way group centroids (averages) of specific samples and groups can be calculated. Ward’s (1963) method is different in that at each stage of agglomeration it merges only those two points the union of which results in the minimum increase in the error sum of squares. A third assumption is sampling adequacy. Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy (MSA), which is the sum of partial correlations relative to the sum of correlations (Kaiser 1970; Kaiser & Rice 1974), was run on the entire dataset, and for each individual variable. To ensure reliable results, the sum of partial correlations Both Ward’s and Group Average were clustered using Eu- should be small compared to the sum of the correlations, clidean distance: indicating that correlations are restricted and that there is clustering among only a few variables. A value of >0.5 is n generally considered acceptable, while values closer to 1 d = |x–y|= |x –y |2 are ideal (Cerny & Kaiser 1977; Dziuban & Shirkey 1974). 1 i i ∑ i= d= distance; x= point a; y= point b The measure of distance will affect the shape of clusters as the points may be closer or further apart depending on the distance metric used. In 3D space Euclidean distance is the shortest distance between two points, but in >3D the generalised equation above still satisfactorily describes distance, especially when applied to quantitative variables such as chemical data (Manly 1986: 106). PCA is a common multivariate technique that reduces a large number of variables to a smaller number of artificial variables (Manly 1986) and allows a visual appraisal of the clusters formed by HCA. The PCA calculates orthogonal linear combinations of the auto-scaled variables (in this case each element oxide weight) by using a correlation matrix based on the maximum variance criterion (Baxter 1994: 66). These linear combinations are referred to as principal component scores and each successive component (principal component 1, 2, 3… n) attempts to account for the maximum variance not accounted for by previous components. By plotting the regression scores of principal components 1, 2, and 3 in 2-D and 3-D space most of the total variance should be displayed. ‘Loadings’ are the coefficients of these principal component scores. The original variables are multiplied by these loadings to give new values, which can also be plotted to describe the correlation between each variable and the principal components (Bruno et al. 2000). PCA must meet four assumptions to be considered valid. One assumption is that the data are normally distributed. Here, the unstandardised clay chemical values for the eight selected element oxide weights were first logarithm base10 transformed (log10x), which approximates the normal distribution (following Bishop & Neff 1989). It also reduces the effect that larger values (such as Si) would have in dictating groupings. Lastly, the PCA assumes that the dataset is suitable for dimension reduction, meaning that there should be equal variances among the variables for them to be reduced to components. A correlation matrix was calculated to determine that each component was significantly correlated, indicating that the ‘correlation matrix’ was an ‘identity matrix’. An identity matrix is a nxn diagonal matrix with ones in the diagonal position and zeros elsewhere. Bartlett’s Test of Sphericity was similarly used to distinguish if the variables were from a population of equal variances and whether these variables were significantly (α=≤0.05) correlated (Bartlett 1937; Sokal & Rohlf 1969). The assignment of chemical groups is a very subjective process based on a visual appraisal of the PCA in threedimensions. These groups are often lumped or split depending upon data quality and research interests. Here the term techno-compositional group is used as a heuristic label to describe these artificial groupings. Results: ethnographic samples The following sections present the results of the Madang mineralogical and compositional analyses. First, the ethnographic baseline is established by presenting the resources used by modern Bilbil and Yabob potters. Then, the pre-colonial pottery results are presented by site. The optical observations of macro-tempers and clay colour are described to give a broad overview of fabrics present in the assemblages. This subdivides technical classes based on rim form into several techno-fabric groupings. Second, the section presents the chemical results of non-plastic mineral grains. This clarifies techno-fabric groups, distinguishing specific techno-mineralogical groups. Third, the chemical analysis of the ceramic matrix is presented. Specific groupings are distinguished within each technomineralogical group to designate techno-compositional groups. The information summarised here can be used in combination with Appendix B, which presents illustrations of the mineralogical variation in selected sherds and clay samples. Another assumption of PCA is that there are no significant outliers, which can have a disproportionate influence on clustering results. Here, outliers were identified by computing the Mahalanobis distance (see Beier & Mommsen Inclusions and tempers 1994). For each sherd sample (5 spectra), those spectra with significantly (P < 0.01) different Mahalanobis dis- The ethnoarchaeological clay sources (Bilbil 1–2, and Yabob tances were identified as outliers and discarded. 1–6) contain varying amounts of natural mineral inclusions with different grain sizes, both of which affect the texture of the clay and its workability. Generally, the Bilbil 147 Chapter . Pre-Colonial Potting II: Procurement and Distribution clays are coarser, containing larger and more numerous inclusions of quartz. Bilbil 1 contains quartz grains up to ~400µm and Bilbil 2 contains quartz up to ~500 µm. Rare augite, epidote, and albite are present in Bilbil 1 as smaller grains (~70–250 µm), along with minor Fe and Ti oxides (~25 µm). Bilbil 2 contains minor amphibole grains (Table 8.5). By contrast, the Yabob clays are relatively fine grained with minor quartz inclusions ~50–300 µm in size. Many Yabob sources also contain ilmenite or Fe and Ti oxides (~20–30 µm), along with rare epidotic, lithic and feldspathic fragments (Table 8.6). All Yabob clays except Yabob 3 contain the same distinct iron-rich rock fragments. Table 8.5. SEM results of mineral inclusions naturally present in Bilbil clay sources. Source 1 Firing (°C) 700 2 1000 700 1000 X X X X Coral Calcareous Shell Indeterminate Augite Clinopyroxenes X Ferroan augite Pigeonite Indeterminate An examination of temper sands shows that the Bilbil Island beach sand is almost entirely calcareous, formed from the underlying reef coral. It is almost identical to sand excavated from basal desposits at Tilu and Nunguri and lends further support to the suggestion that both sites were uplifted coral islands at first occupation. There are also rare grains of clinopyroxene in each sample, and the Nunguri basal sand contains small foraminifera fossils. The Yabob and Bilbil village beach sands are distinct as they derive from mainland beaches and river mouths. Volcanic grains, particularly lithic fragments and augite dominate both sands. The Yabob sand also contains amphibole and ilmenite, while the Bilbil sand contains occasional calcareous grains. It is clear that a range of minerals will be present in the pre-colonial Madang pottery if it is formed from similar constituents, namely quartz deriving from clay deposits, and lithics and pyroxenes from mainland black beach sands. If sands from offshore islands were used instead of mainland sands, calcareous grains will be almost exclusively present. Amphiboles, ilmenite, and calcareous grains may be instructive to assign sherds to different technomineralogical groups owing to their variable presence in local clays and sands from different geological zones and environments. Orthopyroxenes Anorthite Bytownite Plagioclase Labradorite Andesine Oligoclase Albite X X Orthoclase Alkali feldspars Anorthoclase Sanidine Anthophyllite Magnesio hornblende Ferrogedrite Gedrite Amphiboles Ferroglaucophane Edenite Kaersutite Tschermakite Tremolite Inderminate Epidote Anorthite X Quartz Quartz X X Ulvospinel Magnesiochromite Indeterminate Non-silicates Clay composition Chemically, the ethnographic clay sources tend to separate the Bilbil and Yabob resource zones. There is greater variation within each source (i.e. the Euclidean distance between samples fired at 700°C and 1000°C) compared with the preliminary pXRF analysis using heavier elements (see Chapter 4). However, the general structure of the data is the same. The two Bilbil clay sources separate out from the Yabob sources, and Yabob 2, 5, and 6 cluster together, while Yabob 1, 3, and 4 cluster away. Note that although the HCA clusters Yabob 5 (1000°C) with the Bilbil clays, the PCA separates Yabob 5 from the Bilbil samples (Fig. 8.3–8.5). Enstatite Ti oxide X Fe oxide X Ilmenite Cu oxide Rock Mica? Other Humite? Titanite Iron rich silicate (garnet?) Grog Grog? Organic Plant fibres 148       · .  Table 8.6. SEM results of mineral inclusions naturally present in Yabob clay sources. Source 1 Firing (°C) Calcareous 700 2 1000 700 3 1000 700 4 1000 5 6 700 1000 700 1000 700 1000 Coral Shell Indeterminate Clinopyroxenes Augite Ferroan augite Pigeonite Indeterminate Orthopyroxenes Enstatite Plagioclase Anorthite Bytownite Labradorite Andesine Oligoclase Albite Alkali feldspars X Orthoclase X Anorthoclase Sanidine Amphiboles Anthophyllite X Magnesio hornblende Ferrogedrite Gedrite Ferroglaucophane Edenite Kaersutite Tschermakite Tremolite Inderminate Epidote Anorthite Quartz Quartz Non-silicates Ulvospinel X X X X X X X X X X X X X X X X X X X X X Magnesiochromite Indeterminate Ti oxide X Fe oxide X Ilmenite X Cu oxide Other Rock X X X X Mica? Humite? Titanite Iron rich silicate (garnet?) Grog Grog? Organic Plant fibres 149 X X Chapter . Pre-Colonial Potting II: Procurement and Distribution 0 5 10 15 20 25 Yabob 4 (700) Yabob 4 (1000) Yabob 6 (700) Yabob 6 (1000) Yabob 2 (700) Yabob 5 (700) Yabob 2 (1000) Yabob 3 (700) Bilbil 2 (1000) Bilbil 1 (700) Bilbil 2 (700) Yabob 5 (1000) Bilbil 1 (1000) Yabob 1 (1000) Yabob 3 (1000) Yabob 1 (700) Figure 8.3. Ward’s HCA of eight ethnographic clay sources. 0 5 10 15 20 Yabob 4 (700) Yabob 4 (1000) Yabob 6 (700) Yabob 6 (1000) Yabob 2 (700) Yabob 5 (700) Yabob 2 (1000) Yabob 3 (700) Bilbil 2 (1000) Bilbil 1 (700) Bilbil 2 (700) Yabob 5 (1000) Bilbil 1 (1000) Yabob 1 (1000) Yabob 3 (1000) Yabob 1 (700) Figure 8.4. Group average HCA of eight ethnographic clay sources. 150 25       · .  Fe 1.0 3.00000 0.5 Ti Ca Principal component 2 (21.5%) 0.0 K 2.00000 Na Mg Al -0.5 Yabob 5 Si -1.0 1.00000 -1.0 Bilbil 1 -0.5 0.0 0.5 1.0 Yabob 1 .00000 Yabob 2 Yabob 3 Bilbil 2 Yabob 6 -1.00000 Yabob 4 -2.00000 -2.00000 -1.00000 .00000 1.00000 2.00000 3.00000 4.00000 Principal component 1 (43.8%) Yabob Bilbil Figure 8.5. PCA showing clay compositional data from eight ethnographic clay sources. Class 3 rims are composed of calcareous-dominant tempers than expected based on the sample size. This is because Class 3 rims are more abundant in the upper spits, which contain more calcareous sherds in general. Results: Nunguri Nunguri ceramics were formed using locally available raw materials. The majority of sherds were tempered with black beach sand and probably produced using clays from the Bilbil Village area. Techno-fabric groups Throughout Nunguri’s occupation, potters used black ferromagnesium beach sands in preference to ‘light’ (quartz and feldspathic; see Summerhayes 2000), calcareous or grog tempers. However, proportionally, the abundance of light tempers decreases while calcareous tempers increases (Fig. 8.6–8.7). This may reflect a trend towards using immediately available beach sands from Bilbil Island. Grog is rare, and it is often difficult to discern whether small grains of clay are deliberate grog inclusions or accidental incorporations of clay grains from the pottery working floors. Unlike the tempers of Type X ceramics, no Nunguri sherds contained grog temper alone. Overall, there are statistically significant associations between technical-classes and the fabric groups (Pearson’s Chi-Squared using Monte Carlo simulation: X2=52.519, df=12, P<0.01). Each class was formed predominantly using ferromagnesium tempers, with light, calcareous, and grog tempers contributing less (Fig. 8.8). Statistically, more When Spits 1–5 (~0–300 cal. BP) are examined separately there are clear correlations between technical class and fabric (Pearson’s Chi-Squared using Monte Carlo simulation: X2=50.156, df=12, P<0.01). During this period, Class 2 rims are more frequently than expected tempered with Fe-Mg grains (adj. residual= 3.2) and less frequently with calcareous inclusions (adj. residual= –2.6). Class 3 rims are less frequently tempered with Fe-Mg (adj. residual= –5.2) and more frequently with light (adj. residual= 3.5) or calcareous inclusions (adj. residual= 3.9). Class 5 rims show a similar pattern and are less frequently Fe-Mg tempered than expected (adj. residual= –2.8) but more frequently tempered with calcareous grains (adj. residual= 3.0). Provisionally this implies that during this time Class 2 rims were unlikely to be tempered with island beach sands, but Class 3 and 5 were more likely, though mainland sands were still preferred. When Spits 11–16 (~500–600 cal. BP) are considered separately there is no clear correlation between technical class and fabric (Pearson’s Chi-Squared using Monte Carlo simulation: X2=12.369, df=12, P=0.417). This may suggest a trend towards diverging procurement strategies associated with different forming techniques through time. 151 Chapter . Pre-Colonial Potting II: Procurement and Distribution tempers are usually non-diagnostic, but some coral grains were observed. This suggests that the calcareous tempers Tables 8.7–8.11 summarise the identification of mineral derive from the local beach rather than being actively inclusions by technical class and techno-fabric in five manufactured from crushed shell as has been observed excavation spits at Nunguri. This confirms the prelimi- elsewhere in the Pacific (e.g. Rye 1976; Takayama & Shutnary sorting of techno-fabric groups. All sherds contain ler Jr. 1978). Epidote is present in 44% of samples (n=22) augite, and nearly all sherds contain quartz. Calcareous and amphibole is present in 36% of samples (n=18). Rock Techno-mineralogical groups 160 140 Number of rim sherds 120 100 80 60 40 20 0 1 2 3 4 5 1. Fe-Mg 6 7 8 Spit 2. Light 10 9 11 3. Calcareous 12 13 14 15 16 4. Grog Figure 8.6. Nunguri rim sherd techno-fabric groupings by excavation spit. 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 13 Spit 1. Fe-Mg 2. Light 3. Calcareous 4. Grog Figure 8.7. Nunguri decorated body sherd techno-fabric groupings by excavation spit. 152 14 15       · .  400 350 Number of rim sherds 300 250 200 150 100 50 0 2 1 1. Fe-Mg 3 Technical class 2. Light 3. Calcareous 4 5 4. Grog Figure 8.8. Nunguri rim sherd techno-fabric groupings by technical class. grains, some clearly volcanic, were also common. All of these mineral inclusions potentially derive from local raw clays or temper additives. Techno-compositional groups The Nunguri clay chemical data suggest that most sampled sherds were derived from similar clay deposits. Figures 8.9 and 8.10 display the dendrograms produced using two different HCA methods. No clear patterning can be distinguished to subdivide techno-mineralogical groupings into distinct techno-compositional groupings. Figures 8.11–8.13 show the PCA plots of these compositional data relative to excavation spit, technical class, and techno-fabric groups. There is substantial overlap between each of these variables, suggesting that similar clay deposits were being used in the production of pots throughout Nunguri’s occupation. These pots were usually tempered with black beach sands but were sometimes tempered with calcareous sands from offshore islands or light sands from an unknown location. This may represent multiple production groups using the same clays but different tempers, or the same production group making do with different tempers owing to variations in availability. The use of calcareous tempers usually requires the addition of salt rather than fresh water to prevent cracking (Rye 1976). This indicates either that different production groups had developed slightly different procurement and production choices, or that a single production group was adept at engaging with various chaînes opératoires during procurement and production. 153 Figure 8.14 shows a PCA plot of the Nunguri sherds alongside the modern clay samples collected from Yabob and Bilbil villages (see Chapter 4). The sherds predominantly cluster close to the Bilbil Village sources and provisionally suggests that potters were digging clay pits around the same area from c.500 years ago. Note that the outlier is sherd N-188, which tentatively was produced using different clay source, or was mixed from local clays in a different way. Chapter . Pre-Colonial Potting II: Procurement and Distribution Table 8.7. SEM results of mineral inclusions in Nunguri, Spit 2 samples. Sample N-131 N-158 N-132 N-174 N-126 N-127 N-188 N-635 N-136 N-169 Class 1 1 2 2 3 3 4 4 5 5 Fabric 1 4 1 3 1 1 2 1 3 1 Coral Calcareous Shell Indeterminate Augite Clinopyroxenes X X X X X X X Ferroan augite X X X X X X X X X Pigeonite X X Indeterminate Orthopyroxenes Enstatite X X X Anorthite X Bytownite Plagioclase Labradorite X X X X Andesine X X X X X X Oligoclase Albite X X X Orthoclase Alkali feldspars X X X Anorthoclase X Sanidine X X X Anthophyllite Magnesio hornblende Ferrogedrite Gedrite Amphiboles X X X Ferroglaucophane Edenite Kaersutite X X Tschermakite Tremolite Inderminate Epidote Anorthite Quartz Quartz X X X Ulvospinel X X Magnesiochromite X X X X X X X X X X X X X X Indeterminate Non-silicates Ti oxide Fe oxide X Ilmenite Cu oxide Rock X X X X Mica? Other Humite? Titanite Iron rich silicate (garnet?) Grog Grog? Organic Plant fibres X 154 X X X X X       · .  Table 8.8. SEM results of mineral inclusions in Nunguri, Spit 5 samples. Sample N-1557 N-1548 N-1561 N-1556 N-1581 N-1551 N-1547 N-1558 N-1574 N-1562 Class 1 1 1 2 2 3 3 3 4 4 Fabric 1 1 3 1 1 3 2 1 3 1 X X X X X X X X X X X X Coral Calcareous Shell Indeterminate Augite Clinopyroxenes X X X X Ferroan augite Pigeonite Indeterminate Orthopyroxenes Enstatite X Anorthite X Bytownite Plagioclase X X Labradorite X X Andesine Oligoclase X Albite X X X Orthoclase Alkali feldspars X X X X X X X X X Anorthoclase X X X X X X X X X Sanidine Anthophyllite Magnesio hornblende Ferrogedrite Gedrite Amphiboles Ferroglaucophane Edenite Kaersutite X Tschermakite Tremolite Inderminate Epidote Anorthite Quartz Quartz X X X X X X X X Ulvospinel Magnesiochromite X X X X X X Indeterminate Non-silicates Ti oxide Fe oxide Ilmenite X X X X X X Cu oxide Rock X X X X X Mica? Other Humite? Titanite Iron rich silicate (garnet?) Grog Grog? Organic Plant fibres X X 155 X X Chapter . Pre-Colonial Potting II: Procurement and Distribution Table 8.9. SEM results of mineral inclusions in Nunguri, Spit 11 samples. Sample N-2711 N-2703 N-2689 N-2698 N-2707 N-2702 N-2690 N-2692 N-2729 N-2713 Class 1 1 2 2 2 3 4 4 4 5 Fabric 3 1 1 3 1 1 1 2 3 2 Coral Calcareous Clinopyroxenes X X X X Shell Indeterminate X Augite X X X X X X X X X X Ferroan augite Pigeonite X X Indeterminate Orthopyroxenes Enstatite X Anorthite Plagioclase Bytownite X X Labradorite X X Andesine X X X X X X X X X X X Oligoclase Albite X X X X X X X Orthoclase Alkali feldspars Anorthoclase X Sanidine X X X Anthophyllite Magnesio hornblende Ferrogedrite Gedrite Amphiboles Ferroglaucophane Edenite Kaersutite Tschermakite X Tremolite X Inderminate Epidote Anorthite X Quartz Quartz X X X X X Ulvospinel X X X X X X X X X Magnesiochromite Indeterminate Non-silicates Ti oxide Fe oxide Ilmenite X X X X Cu oxide Rock X Mica? Other Humite? Titanite Iron rich silicate (garnet?) Grog Grog? Organic Plant fibres X 156 X X X X       · .  Table 8.10. SEM results of mineral inclusions in Nunguri, Spit 13 samples. Sample N-3162 N-3141 N-3146 N-3153 N-3139 N-3142 N-3148 N-3159 N-3155 N-3140 Class 1 1 1 1 2 2 2 4 4 4 Fabric 1 2 1 1 1 1 1 2 1 1 X X X X X X X X X X X X X X Coral Calcareous Shell Indeterminate Augite Clinopyroxenes Ferroan augite Pigeonite X Indeterminate Orthopyroxenes Plagioclase Enstatite Anorthite X Bytownite X X X Labradorite X Andesine X X Oligoclase X X X X X Albite X X Orthoclase Alkali feldspars Anorthoclase X Sanidine X X X Anthophyllite Magnesio hornblende Ferrogedrite Gedrite Amphiboles X X X Ferroglaucophane Edenite Kaersutite Tschermakite Tremolite Inderminate Epidote Anorthite Quartz Quartz X X X X X X X X X X Ulvospinel Magnesiochromite X Indeterminate Non-silicates Ti oxide Fe oxide Ilmenite X X Cu oxide Rock X X X X X X X Mica? Other Humite? Titanite Iron rich silicate (garnet?) X Grog Grog? X Organic Plant fibres 157 X X X X Chapter . Pre-Colonial Potting II: Procurement and Distribution Table 8.11. SEM results of mineral inclusions in Nunguri, Spit 15 samples. Sample N-3811 N-3810 N-3698 N-3696 N-3711 N-3701 N-3699 N-3802 N-3806 Class 1 1 1 1 2 2 2 4 4 N-3798 4 Fabric 1 1 1 1 1 1 2 1 1 2 X X X X X X X X Coral Calcareous Clinopyroxenes Shell Indeterminate X Augite X X X Ferroan augite Pigeonite X Indeterminate Orthopyroxenes Enstatite Anorthite Bytownite Plagioclase X Labradorite X X X X Andesine Oligoclase Albite Alkali feldspars X X Orthoclase X X X X Anorthoclase X X Sanidine X Anthophyllite X Magnesio hornblende Ferrogedrite X Gedrite Amphiboles X Ferroglaucophane Edenite Kaersutite X X Tschermakite Tremolite Inderminate X Epidote Anorthite X Quartz Quartz X X n* Ulvospinel X X X X X X X X X X X X X X X Magnesiochromite X X Indeterminate Non-silicates Ti oxide Fe oxide Ilmenite X X X Cu oxide Rock X X X X X X X Mica? Other Humite? Titanite Iron rich silicate (garnet?) Grog Grog? Organic Plant fibres X X *n = small grains naturally occurring in the clay 158       · .  Spit Class Fabric Sample 13 4 FeMg N3140 13 1 FeMg N3146 13 1 Light N3141 2 3 FeMg N126 2 2 FeMg N132 15 2 FeMg N3711 5 3 Calc N1551 15 1 FeMg N3810 15 4 FeMg N3806 15 2 Light N3699 5 1 FeMg N1557 11 1 FeMg N2703 5 2 FeMg N1581 15 4 Light N3798 11 4 FeMg N2690 2 5 Calc N136 2 3 FeMg N127 15 4 FeMg N3802 5 3 Light N1547 5 3 FeMg N1558 15 1 FeMg N3698 13 2 FeMg N3142 2 5 FeMg N169 11 4 Light N2692 13 2 FeMg N3139 2 4 FeMg N635 15 2 FeMg N3701 11 2 FeMg N2689 11 4 Calc N2729 13 4 FeMg N3155 13 4 Light N3159 5 1 Calc N1561 5 4 FeMg N1562 11 2 Calc N2698 5 1 FeMg N1548 5 4 Calc N1574 2 2 Calc N174 13 1 FeMg N3153 2 1 FeMg N131 15 1 FeMg N3696 2 4 Light N188 5 2 FeMg N1556 11 3 FeMg N2702 2 1 Grog N158 11 1 Calc N2711 15 1 FeMg N3811 11 2 FeMg N2707 11 5 Light N2713 13 1 FeMg N3162 0 5 10 15 Figure 8.9. Ward’s HCA of Nunguri sherds. 159 20 25 Chapter . Pre-Colonial Potting II: Procurement and Distribution Spit Class Fabric Sample 13 4 FeMg N3140 13 1 FeMg N3146 5 1 Calc N1561 13 1 Light N3141 5 3 Calc N1551 15 1 FeMg N3810 2 4 FeMg N635 15 2 FeMg N3701 11 2 FeMg N2689 11 4 Calc N2729 13 4 FeMg N3155 13 4 Light N3159 5 4 FeMg N1562 5 1 FeMg N1557 11 1 FeMg N2703 5 3 FeMg N1558 5 2 FeMg N1581 15 4 Light N3798 11 4 FeMg N2690 2 5 Light N136 15 1 FeMg N3698 2 3 FeMg N127 15 4 FeMg N3802 5 3 Light N1547 2 5 FeMg N169 11 4 Light N2692 13 2 FeMg N3139 11 2 Calc N2698 2 3 FeMg N126 2 2 FeMg N132 15 2 FeMg N3711 15 4 FeMg N3806 15 2 Light N3699 5 2 FeMg N1556 11 3 FeMg N2702 13 2 FeMg N3142 13 1 FeMg N3153 5 1 FeMg N1548 5 4 Calc N1574 2 2 Calc N174 2 1 FeMg N131 15 1 FeMg N3696 11 1 Calc N2711 15 1 FeMg N3811 11 2 FeMg N2707 11 5 Light N2713 13 1 FeMg N3162 2 1 Grog N158 2 4 Light N188 0 5 10 15 Figure 8.10. Group average HCA of Nunguri sherds. 160 20 25       · .  1.0 Ca 3.0 0.5 Fe K Na Si 0.0 Mg Ti 2.0 Principal component 2 (16.4%) -0.5 Al -1.0 1.0 -1.0 -0.5 0.0 0.5 1.0 0.0 -1.0 -2.0 -3.0 -2.0 0.0 2.0 4.0 Principal component 1 (57.7%) Spit 2 Spit 5 Spit 11 Spit 13 Spit 15 Figure 8.11. PCA showing clay compositional data from Nunguri sherds by excavation spit. 1.0 Ca 3.0 0.5 Fe K Na Si 0.0 Mg Ti 2.0 Principal component 2 (16.4%) -0.5 Al -1.0 1.0 -1.0 -0.5 0.0 0.5 1.0 0.0 -1.0 -2.0 -3.0 -2.0 0.0 2.0 4.0 Principal component 1 (57.7%) Class 1 Class 2 Class 3 Class 4 Class 5 Figure 8.12. PCA showing clay compositional data from Nunguri sherds by technical class. 161 Chapter . Pre-Colonial Potting II: Procurement and Distribution 1.0 Ca 3.0 0.5 Fe K Na Si 0.0 Mg Ti 2.0 Principal component 2 (16.4%) -0.5 Al -1.0 1.0 -1.0 -0.5 0.0 0.5 1.0 0.0 -1.0 -2.0 -3.0 -2.0 0.0 2.0 4.0 Principal component 1 (57.7%) Fe-Mg Light Calcareous Grog Figure 8.13. PCA showing clay compositional data from Nunguri sherds by fabrics. 1.0 Fe Ti 2.0 Mg 0.5 Ca K 0.0 Na 1.0 Al Si Principal component 2 (23.0%) -0.5 0.0 -1.0 -1.0 -0.5 0.0 0.5 -1.0 -2.0 -3.0 -4.0 -5.0 -4.0 -2.0 0.0 2.0 4.0 Principal component 1 (47.0%) Yabob Bilbil Nunguri Figure 8.14. PCA showing clay compositional data of Nunguri sherds and ethnographic clays. 162 1.0       · .  Results: Tilu Techno-mineralogical groups Compositionally, the Tilu sherds are almost identical to those from Nunguri. In general, Tilu pots were usually tempered with black beach sand and produced using local clays, probably from the Bilbil area. The mineral inclusions present in Tilu rim sherds are consistent with those observed in the Nunguri sample. Tables 8.12–8.16 present the descriptive results of this analysis. Augite was observed in all but one sample (T-660). All sherds contained quartz and all but two sherds contained some type of plagioclase. Epidote is present in 32% of sampled rims (n=16) and amphibole is present in 22% of rims (n=11). An iron-rich silicate (possibly a garnet) is present in only 6% of samples (n=3), much less common than in Nunguri rims, of which 26% contain the same mineral (n=13). During the course of analysis, the rims from lower spits were observed to more frequently contain calcareous inclusions, especially in Spit 9. There is a significant correlation between excavation spit and presence of calcareous inclusions (Pearson’s Chi-Squared using Monte Carlo simulation: X2=13.33, df=4, P=0.006), and these inclusions are particularly more common than expected in Spit 9 (adj. residual= 3.1). One of these Spit 9 samples contains identifiable coral grains (T-762) and it is likely that all of the calcareous grains present in the Tilu samples represent coral detritus from offshore islands. For this reason, three sherds from Spit 10 were selected for supplementary mineralogical analysis. The small sample from Spit 10 demonstrates the same trend as two of the three sherds contain calcareous grains (Table 8.17). Techno-fabric groups Like at Nunguri, the sherds at Tilu predominantly contain black ferromagnesium sand grains (Fig. 8.15–8.16). Calcareous and light dominated tempers are present but form a minor contribution to the Tilu assemblage. There is no clear trend towards using calcareous tempers later in time (as at Nunguri), and, in fact, in the earliest spits (8–10) calcareous tempers are more prominent. Only one sherd (T298) contains obvious grog that has used broken sherds for temper. However, Fe-Mg and light mineral inclusions are also present in this sherd. Again, no sherds are exclusively tempered with grog in the manner of Type X. There are no statistically significant associations between technical-classes and the fabric groups at Tilu (Pearson’s Chi-Squared using Monte Carlo simulation: X2=3.167, df=12, P=0.862). Class 1 was formed predominantly using ferromagnesium tempers, with light, calcareous, and grog tempers contributing less (Fig. 8.17). Class 2 rims are also dominantly composed of ferromagnesium tempers with some instances of calcareous and grog tempering. Classes 3, 4, and 5 are only tempered with ferromagnesium tempers. However, as Classes 2–5 are comprised of a much smaller sample than Class 1, the absence of examples of certain temper types is not substantially less than expected given the number of cases (based on a cross tabulation of the Chi-Squared results). Techno-compositional groups At Tilu, there seems to have been a single resource zone for the procurement of potting clay. Figures 8.18–8.19 show the HCA results relative to excavation spit, class, and fabric. As at Nunguri, there is no clear patterning to suggest that specific techno-mineralogical groups can be subdivided 25 Number of rim sherds 20 15 10 5 0 1 2 3 1. Fe-Mg 4 2. Light 5 Spit 6 3. Calcareous 7 8 4. Grog Figure 8.15. Tilu rim sherd techno-fabric groupings by excavation spit. 163 9 10 Chapter . Pre-Colonial Potting II: Procurement and Distribution into different techno-compositional groupings. The PCA There are no ethnographic or oral history accounts of the plots in Figures 8.20–8.22 coded by spit, class, and fabric Malmal (Tilu) people producing pottery. The results of the also show substantial overlap within each variable and geochemical analysis suggest that many of the Tilu pots suggest a similar manner of procurement to the Nunguri were imported from pottery producers to the south – parsherds. Figure 8.23 shows that most of the Tilu sherds ticularly those using clays from around modern Bilbil Vilgroup with the ethnographic Bilbil clay sources rather lage. than the Yabob sources. The single notable outlier is T-843 from Spit 10, which may have been produced using a clay from the Yabob area, or another currently unknown place. 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 2 1 3 4 5 6 7 8 9 10 11 12 Spit 1. Fe-Mg 2. Light 3. Calcareous 4. Grog Figure 8.16. Tilu decorated body sherd techno-fabric groupings by excavation spit. 100 90 Number of rim sherds 80 70 60 50 40 30 20 10 0 1 2 1. Fe-Mg 3 Technical class 2. Light 3. Calcareous 4 4. Grog Figure 8.17. Tilu rim sherd techno-fabric groupings by technical class. 164 5       · .  Table 8.12. SEM results of mineral inclusions in Tilu, Spit 2 samples. Sample T-167 T-180 T-160 T-163 T-174 T-162 T-177 T-164 T-168 T-173 Class 1 1 1 1 1 2 2 2 3 4 Fabric 1 1 1 3 2 1 1 1 1 1 X X X X Coral Calcareous Shell Indeterminate Augite Clinopyroxenes X X X X X X X X Ferroan augite Pigeonite Indeterminate Orthopyroxenes Enstatite Anorthite X X Bytownite Plagioclase Labradorite X X X X Andesine X Oligoclase Albite X X X Orthoclase Alkali feldspars X X X X X Anorthoclase X X Sanidine Anthophyllite X Magnesio hornblende Ferrogedrite X Gedrite Amphiboles X X Ferroglaucophane Edenite Kaersutite X Tschermakite Tremolite Inderminate Epidote Anorthite Quartz Quartz X X X X X X X X X X X X Ulvospinel Magnesiochromite X X Indeterminate Non-silicates Ti oxide Fe oxide X Ilmenite X X X X X X X Cu oxide Rock X X X Mica? Other Humite? Titanite Iron rich silicate (garnet?) Grog Grog? Organic Plant fibres X 165 X X X X Chapter . Pre-Colonial Potting II: Procurement and Distribution Table 8.13. SEM results of mineral inclusions in Tilu, Spit 4 samples. Sample T-409 T-411 T-412 T-413 T-414 T-420 T-417 T-418 T-419 T-416 Class 1 1 1 1 1 1 1 1 1 5 Fabric 1 1 1 1 1 1 2 1 1 1 X X X X X X X X X X Coral Calcareous Shell Indeterminate Augite Clinopyroxenes Ferroan augite Pigeonite Indeterminate Orthopyroxenes Enstatite X Anorthite X Bytownite Plagioclase Labradorite X X X X X X Andesine Oligoclase X Albite Alkali feldspars X X X X X Orthoclase X Anorthoclase X X X X X X X Sanidine Anthophyllite Magnesio hornblende Ferrogedrite Gedrite Amphiboles Ferroglaucophane Edenite Kaersutite Tschermakite Tremolite Inderminate Epidote Anorthite Quartz Quartz X X X X X X X X X X X X X X X X X X X X X X X X X X Ulvospinel Magnesiochromite Indeterminate Non-silicates Ti oxide Fe oxide Ilmenite X X Cu oxide Rock X X Mica? Other Humite? X Titanite Iron rich silicate (garnet?) Grog Grog? Organic Plant fibres X 166       · .  Table 8.14. SEM results of mineral inclusions in Tilu, Spit 6 samples. Sample T-556 T-549 T-559 T-552 T-550 T-553 T-555 T-639 T-554 T-558 Class 1 1 1 1 1 1 1 1 2 2 Fabric 1 1 1 1 1 1 1 1 1 1 Indeterminate X X X X Augite X X X X X X X X Coral Calcareous Clinopyroxenes Shell X Pigeonite Indeterminate Orthopyroxenes X Ferroan augite X X Enstatite Anorthite X X X Bytownite Plagioclase X Labradorite X X X Andesine Oligoclase X Albite X X X X X X X Orthoclase Alkali feldspars X Anorthoclase X X X X Sanidine X Anthophyllite Magnesio hornblende Ferrogedrite Gedrite Amphiboles Ferroglaucophane Edenite Kaersutite Tschermakite Tremolite Inderminate Epidote Anorthite Quartz Quartz X Ulvospinel X X X X X X X X X X X X X X X X Magnesiochromite Indeterminate Non-silicates Ti oxide Fe oxide X Ilmenite X X Cu oxide Rock X Mica? Other Humite? X X Titanite Iron rich silicate (garnet?) Grog Grog? Organic Plant fibres 167 X Chapter . Pre-Colonial Potting II: Procurement and Distribution Table 8.15. SEM results of mineral inclusions in Tilu, Spit 7 samples. Sample T-644 T-653 T-660 T-650 T-643 T-642 T-662 T-648 T-641 T-652 Class 1 1 1 1 1 1 1 2 2 2 Fabric 1 1 2 1 1 1 1 1 1 4 X X X X X X X X Coral Calcareous Shell Indeterminate Augite Clinopyroxenes X X X Ferroan augite Pigeonite Indeterminate Orthopyroxenes Enstatite X X Anorthite Bytownite Plagioclase X X Labradorite X X X X X X X X Andesine Oligoclase X Albite X X Orthoclase Alkali feldspars X X X X X Anorthoclase Sanidine X Anthophyllite Magnesio hornblende Ferrogedrite Gedrite Amphiboles Ferroglaucophane Edenite Kaersutite X X X X X Tschermakite Tremolite Inderminate Epidote Anorthite Quartz Quartz X X Ulvospinel X X X X X X X X X X X Magnesiochromite Indeterminate Non-silicates X X X Ti oxide Fe oxide X X X Ilmenite X X X X X X X Cu oxide Rock X X X X Mica? Other Humite? Titanite Iron rich silicate (garnet?) Grog Grog? Organic Plant fibres X 168 X       · .  Table 8.16. SEM results of mineral inclusions in Tilu, Spit 9 samples. Sample T-757 T-752 T-753 T-748 T-751 T-840 T-841 T-762 T-747 T-749 Class 1 1 1 1 1 1 1 1 2 2 Fabric 1 1 1 3 1 1 1 3 1 1 Coral Calcareous X Shell Indeterminate Clinopyroxenes Augite X Ferroan augite X X X X X X X X X X X X X X X X X X X X X X X X Pigeonite Indeterminate Orthopyroxenes Enstatite X Anorthite X Bytownite Plagioclase X X X X Labradorite X X Andesine Oligoclase Albite X X Orthoclase Alkali feldspars X X X X X X X Anorthoclase X X X X Sanidine X X X X Anthophyllite Magnesio hornblende Ferrogedrite Gedrite Amphiboles X X Ferroglaucophane Edenite Kaersutite X Tschermakite Tremolite Inderminate X Epidote Anorthite X Quartz Quartz X X X X X X X X X X X X Ulvospinel X X X Magnesiochromite Indeterminate Non-silicates X Ti oxide Fe oxide X Ilmenite X X X X X X X X X Cu oxide Rock X X X X X X Mica? Other Humite? Titanite Iron rich silicate (garnet?) X Grog Grog? X Organic Plant fibres 169 X X Chapter . Pre-Colonial Potting II: Procurement and Distribution Table 8.17. SEM results of mineral inclusions in Tilu, Spit 10 samples. Sample T-843 T-842 T-871 Class 1 1 2 Fabric 1 3 1 X X X X X X Coral Calcareous Shell Indeterminate Augite Clinopyroxenes X Ferroan augite Pigeonite Indeterminate Orthopyroxenes Enstatite X Anorthite X Bytownite Plagioclase Labradorite Andesine X Oligoclase Albite X Orthoclase Alkali feldspars X Anorthoclase X Sanidine Anthophyllite Magnesio hornblende Ferrogedrite Gedrite Amphiboles X Ferroglaucophane Edenite Kaersutite Tschermakite Tremolite Inderminate Epidote Quartz Anorthite Quartz X n* X Ulvospinel Magnesiochromite Indeterminate Non-silicates Ti oxide Fe oxide X X Ilmenite Cu oxide Rock X X X Mica? Other Humite? Titanite X Iron rich silicate (garnet?) Grog Grog? Organic Plant fibres X X *n = small grains naturally occurring in the clay 170       · .  Spit Class Fabric Sample T417 T639 T420 T553 T177 T555 T162 T550 T556 T840 T413 T552 T554 T751 T418 T662 T642 T747 T164 T163 T180 T167 T168 T412 T841 T748 T643 T416 T559 T752 T411 T419 T414 T749 T757 T641 T753 T174 T160 T842 T652 T660 T650 T549 T558 T173 T653 T648 T762 T409 T871 T843 T644 4 1 Light 6 1 FeMg 4 1 FeMg 6 1 FeMg 2 2 FeMg 6 1 FeMg 2 2 FeMg 6 1 FeMg 6 1 FeMg 9 1 FeMg 4 1 FeMg 6 1 FeMg 6 2 FeMg 9 1 FeMg 4 1 FeMg 7 1 FeMg 7 1 FeMg 9 2 FeMg 2 2 FeMg 2 1 Calc 2 1 FeMg 2 1 FeMg 2 3 FeMg 4 1 FeMg 9 1 FeMg 9 1 Calc 7 1 FeMg 4 5 FeMg 6 1 FeMg 9 1 FeMg 4 1 FeMg 4 1 FeMg 4 1 FeMg 9 2 FeMg 9 1 FeMg 7 2 FeMg 9 1 FeMg 2 1 Light 2 1 FeMg 10 1 Calc 7 2 Grog 7 1 Light 7 1 FeMg 6 1 FeMg 6 2 FeMg 2 4 FeMg 7 1 FeMg 7 2 FeMg 9 1 Calc 4 1 FeMg 10 2 FeMg 10 1 FeMg 7 1 FeMg 0 5 10 15 Figure 8.18. Ward’s HCA of Tilu sherds. 171 20 25 Chapter . Pre-Colonial Potting II: Procurement and Distribution Spit Class Fabric Sample 0 4 1 Light 6 1 FeMg 4 1 FeMg 6 1 FeMg 6 1 FeMg 9 1 FeMg 4 1 FeMg 6 1 FeMg 4 1 FeMg 2 2 FeMg 2 1 Calc 2 1 FeMg 2 1 FeMg 2 3 FeMg 7 1 FeMg 2 2 FeMg 6 1 FeMg 6 2 FeMg 9 1 FeMg 7 1 FeMg 9 2 FeMg 6 1 FeMg 6 2 FeMg 2 4 FeMg 7 1 FeMg 7 2 FeMg 9 1 Calc 4 1 FeMg 10 2 FeMg 2 1 FeMg 10 1 Calc 7 2 Grog 7 1 Light 9 2 FeMg 9 1 FeMg 7 2 FeMg 9 1 FeMg 7 1 FeMg 4 1 FeMg 2 1 Light 9 1 Calc 9 1 FeMg 2 2 FeMg 6 1 FeMg 7 1 FeMg 4 5 FeMg 6 1 FeMg 9 1 FeMg 4 1 FeMg 4 1 FeMg 4 1 FeMg 10 1 FeMg 7 1 FeMg 5 10 15 T417 T639 T420 T553 T556 T840 T413 T552 T418 T164 T163 T180 T167 T168 T662 T177 T555 T554 T751 T642 T747 T549 T558 T173 T653 T648 T762 T409 T871 T160 T842 T652 T660 T749 T757 T641 T753 T650 T412 T174 T748 T841 T162 T550 T643 T416 T559 T752 T419 T411 T414 T843 T644 Figure 8.19. Group average HCA of Tilu sherds. 172 20 25       · .  1.0 4.0 Fe 0.5 Ca 0.0 Mg K Ti Al 3.0 Na Principal component 2 (27.6%) -0.5 Si -1.0 2.0 -1.0 -0.5 0.0 0.5 1.0 1.0 0.0 -1.0 -2.0 -3.0 -2.0 0.0 2.0 4.0 Principal component 1 (52.3%) Spit 2 Spit 4 Spit 6 Spit 7 Spit 9 Spit 10 Figure 8.20. PCA showing clay compositional data from Tilu sherds by excavation spit. 1.0 4.0 Fe 0.5 Ca 0.0 Al Mg K Ti 3.0 Na Principal component 2 (27.6%) -0.5 Si -1.0 2.0 -1.0 -0.5 0.0 0.5 1.0 0.0 -1.0 -2.0 -3.0 -2.0 0.0 2.0 4.0 Principal component 1 (52.3%) Class 1 Class 2 Class 3 Class 4 Class 5 Figure 8.21. PCA showing clay compositional data from Tilu sherds by technical class. 173 1.0 Chapter . Pre-Colonial Potting II: Procurement and Distribution 1.0 4.0 Fe 0.5 Ca 0.0 Mg K Ti Al 3.0 Na Principal component 2 (27.6%) -0.5 Si -1.0 2.0 -1.0 -0.5 0.0 0.5 1.0 1.0 0.0 -1.0 -2.0 -3.0 -2.0 0.0 2.0 4.0 Principal component 1 (52.3%) Fe-Mg Light Calcareous Grog Figure 8.22. PCA showing clay compositional data from Tilu sherds by fabrics. Fe 1.0 0.5 4.0 Ti 0.0 Principal component 2 (21.7%) 3.0 Ca Mg K Na Al -0.5 Si -1.0 -1.0 2.0 -0.5 0.0 0.5 1.0 1.0 0.0 -1.0 -2.0 -3.0 -4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 Principal component 1 (39.8%) Yabob Bilbil Tilu Figure 8.23. PCA showing clay compositional data of Tilu sherds and ethnographic clays 174 3.0       · .  Results: surface survey Sherds collected during surface survey are consistent in fabric and geochemistry with Madang style sherds from Tilu and Nunguri. Techno-fabric groups As per the excavated contexts at Nunguri and Tilu, surfacecollected rims from the 2014 Madang survey are also primarily tempered with ferromagnesium grains (Table 8.18). Light dominant tempers are less common, while calcareous and grog tempers are rare. class and clay, or collection location and clay. However, it is notable that the two light tempered sherds (K-2 & K-7) group together in the HCA and cluster away from the FeMg tempered sherds in the PCA. When the survey sherds are compared to ethnographic clay sources (Fig. 8.29), the majority of sherds cluster with the Bilbil sources, like the Tilu and Nunguri samples. However, two sherds (K-2 & K-7) group away from any of the reference sources suggesting that the specimens may have been produced using different clays. As these sherds were collected from an exposed beach, post-depositional alteration is another possibility for this anomaly. Table 8.18. Techno-fabric groups of survey collected sherds. Techno-mineralogical groups The range of specific minerals in the survey sample is consistent with the suite of minerals identified at Nunguri and Tilu (Table 8.19). Every analysed rim contains augite and quartz. Plagioclase, epidote, amphibole, and ilmenite are also common. The clay data illustrated as dendrograms in Figures 8.24– 8.25 and as PCA plots in Figures 8.26–8.28 are limited owing to the small sample size. Most of the sherds overlap, and there are no clear associations between technical 3 FeMg K1 2 FeMg Y4 5 FeMg S3 2 FeMg Y3 1 FeMg K3 1 FeMg S4 4 FeMg Y1 1 FeMg K5 1 FeMg K6 1 FeMg Y6 1 FeMg K4 1 Light K2 3 Light K7 1. Fe-Mg 2. Light 3. Calcareous 4. Grog rims 5 2 – – decorated 1 2 – – rims 4 1 – – decorated – – 1 – rims 3 1 – – decorated – – – 1 Kranket Island Yabob Island Techno-compositional groups Class Fabric Sample Techno-fabric group 5 10 Siar Island 15 Figure 8.24. Ward’s HCA of surface survey sherds. 175 20 25 Chapter . Pre-Colonial Potting II: Procurement and Distribution Table 8.19. SEM results of mineral inclusions in surface survey samples: Kranket Island, Siar Island, and Yabob Island. Sample K-1 K-2 K-3 K-4 K-5 K-6 K-7 S-3 S-4 Y-1 Y-3 Y-4 Y-6 Class 3 1 1 1 1 1 3 5 1 4 2 2 1 Fabric 1 2 1 1 1 1 2 1 1 1 1 1 1 X X X X X X X X Coral Calcareous Clinopyroxenes Shell Indeterminate X Augite X X X X X X X Ferroan augite Pigeonite Indeterminate Orthopyroxenes Plagioclase Enstatite X Anorthite X Bytownite X Labradorite X X X X X X X X X X X X X X X Andesine Oligoclase Albite X X X X X X X X Orthoclase Alkali feldspars Anorthoclase X X Sanidine X X X Anthophyllite X Magnesio hornblende Ferrogedrite Gedrite Amphiboles Ferroglaucophane Edenite Kaersutite X X Tschermakite Tremolite Inderminate Epidote Anorthite Quartz Quartz X X X Ulvospinel Non-silicates X X X X X X Magnesiochromite X Indeterminate X X X X X X X X X X X X X X X X X Ti oxide Fe oxide Ilmenite X X X X X X X X X X X X X X X Cu oxide Rock X X X X X X Mica? Other Humite? X X Titanite X Iron rich silicate (garnet?) Grog Grog? Organic Plant fibres X 176 X       · .  5 Class Fabric Sample 3 FeMg K1 2 FeMg Y4 5 FeMg S3 2 FeMg Y3 1 FeMg S4 4 FeMg Y1 1 FeMg K5 1 FeMg K6 1 FeMg Y6 1 FeMg K4 1 FeMg K3 1 Light K2 3 Light K7 10 15 20 25 Figure 8.25. Group average HCA of surface survey sherds. 1.0 Ca 2.0 Ti 0.5 Na Al 0.0 K Principal component 2 (28.4%) 1.0 Fe -0.5 Si Mg -1.0 -1.0 -0.5 0.0 0.5 0.0 -1.0 -2.0 -3.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 Principal component 1 (37.4%) Kranket Island Siar Island Yabob Island Figure 8.26. PCA showing clay compositional data from surface survey sherds by location. 177 1.0 Chapter . Pre-Colonial Potting II: Procurement and Distribution 1.0 Ca 2.0 Ti 0.5 Al Na 0.0 K Principal component 2 (28.4%) 1.0 Fe -0.5 Si Mg -1.0 -1.0 -0.5 0.0 0.5 1.0 0.0 -1.0 -2.0 -3.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 Principal component 1 (37.4%) Class 1 Class 2 Class 3 Class 4 Class 5 Figure 8.27. PCA showing clay compositional data from surface survey sherds by technical class. 1.0 Ca 2.0 Ti 0.5 Na Al 0.0 K Principal component 2 (28.4%) 1.0 Fe -0.5 Si Mg -1.0 -1.0 -0.5 0.0 0.5 0.0 -1.0 -2.0 -3.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 Principal component 1 (37.4%) Fe-Mg Light Figure 8.28. PCA showing clay compositional data from surface survey sherds by fabric. 178 1.0       · .  4.0 Fe 1.0 0.5 Mg Ti 3.0 K Na 0.0 Ca Al Si Principal component 2 (17.3%) -0.5 2.0 -1.0 -1.0 -0.5 0.0 0.5 1.0 1.0 0.0 -1.0 -2.0 -3.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 Principal component 1 (42.7%) Yabob clay Bilbil clay Kranket Island Siar Island Yabob Island Figure 8.29. PCA showing clay compositional data of survey sherds and ethnographic clays. Results: exotics Many of the sherds identified as ‘exotic’ based on form and decoration are distinct from the Madang style sherds. Exotics were preliminarily examined under low powered magnification to identify grain types but they were not sorted into techno-fabric groups like those following the Madang style classification system. They were examined under SEM to distinguish mineral inclusions and clay chemistry. Mineralogy The single ‘exotic’ sherd from Nunguri (N-3712) contains similar minerals to Madang style sherds, namely augite, quartz, plagioclase and potassium feldspar, and spinels (Table 8.20). It is distinguished as an exotic based on formal differences and a lack of red slip, but it may alternatively represent a similar potting tradition in the Madang or northeast coast area owing to the similarities in tempers. At Tilu, several sherds are distinctly exotic and ‘non-Madang’ based on their mineral components. T-188, T-230, T-410, and T-604 for instance contain sparse inclusions and are potentially non-tempered – the inclusions instead representing natural minerals in the clay source. Although T-188 and T-230 contain similar types of non-plastics to the Madang style rims, the small size of the quartz grains (~20 µm) indicates they are natural inclusions in the clay, dissimilar to the Madang style rims which contain much larger quartz grains. T-410 and T-604 also contain a similar range of mineral inclusions, but lack augite, which is almost always present in Madang style ceramics. T-520 and T-913 appear to be manually tempered owing to the large grain size of their inclusions, consistent with manually tempered Madang style ceramics. T-520 contains a range of minerals, consistent with the Madang style, but it is still considered to be non-Madang owing to inconsistencies in the decoration technique: it exhibits deep linear incision. This may represent another technical tradition in the Madang or northeast coast area, distinct from the Madang style. T-913, the unique ring-base, also appears to be manually tempered (or derived from a very coarse-grained clay source). The inclusions in the sherd are almost exclusively quartz, with some plagioclase and rare, but very small (< 50 µm), ilmentie and Fe-oxide grains. The possible exotic sherd from Yabob Island (Y-5) contains minerals within the range of the Madang style suite, including augite, plagioclase, amphibole, quartz and ilmenite. Although the rim form does not fall within Technical Classes 1–5, it is red slipped. As such the sherd remains an unknown. It may represent a unique form of rim (or even 179 Chapter . Pre-Colonial Potting II: Procurement and Distribution Table 8.20. SEM results of mineral inclusions in exotic samples from Tilu, Nunguri, and Yabob. Sample Designation Spit T-188 T-230 T-410 T-520 T-604 T-913 N-3712 Y-5 Rim Body Rim Body Body Ring-base Rim Rim 2 2 4 5 6 SP 60 cm 15 Surf. X X X X Coral Calcareous Shell Indeterminate Augite Clinopyroxenes X Ferroan augite Pigeonite Indeterminate Orthopyroxenes Enstatite Anorthite X Bytownite Plagioclase Labradorite X X X X X X Andesine Oligoclase X Albite X X Orthoclase Alkali feldspars X X X X X X X X Anorthoclase Sanidine X Anthophyllite Magnesio hornblende Ferrogedrite Gedrite Amphiboles X Ferroglaucophane Edenite Kaersutite X Tschermakite Tremolite Inderminate Epidote Anorthite Quartz Quartz X X X X X n* n X X X X X X Ulvospinel Magnesiochromite X Indeterminate Non-silicates Ti oxide Fe oxide X Ilmenite X X X X X X X Cu oxide Rock Mica? Other Humite? Titanite Iron rich silicate (garnet?) Grog Grog? Organic Plant fibres *n = small grains naturally occurring in the clay 180 X       · .  a stand) in the Madang style or it may be derived from a separate technical tradition but utilising similar tempering sands. To resolve this, a larger sample of similar sherds is required, perhaps from future excavations on Yabob Island. the very earliest ceramics in the Madang style found in this study appear to be produced from locally available materials, supportive of a recent migration of Bel potters to the Madang coast with no evidence for previous trading from other source zones. Overall, the PCA plots indicate no clear patterning relative to place of raw material collection, technical class, or fabric The clay chemical dataset for exotic sherds is again limited group, which suggests one or more production groups usby a small sample size. Figures 8.30–8.31 show HCA den- ing local raw materials were making pots with a range of drograms of these data, while Figures 8.32–8.33 show PCA forming processes. The same sources appear to have been plots of the same data, firstly focusing only on the sherds in use for the last 500–600 years. However, there is a tentathemselves, and then in comparison to ethnographic clay tive partitioning of Nunguri and Tilu samples in Figure sources from around Madang. Interestingly, all of the ex- 8.34. Elsewhere in the world, chemical clusters have shifted otic sherds cluster towards the Bilbil clay sources. This gradually within one potter’s lifetime either through the may suggest that some of the ‘exotic’ sherds were pro- use of different raw materials or differential mixing of the duced using clays from the Bilbil area and may in fact be same raw materials (Arnold 2015). The incomplete separalocal but deriving from distinct forming traditions. tion of Nunguri and Tilu clusters may hint at similar minor changes to the location of clay pits over time, no more than a few generations. These differences are not expected to be Summary substantial in terms of distance to the source but may relate The macroscopic and geochemical analysis of a sample to source exhaustion or relationships with landowners. of Madang sherds and clay sources was successful in distinguishing local raw material procurement and distribu- In the next chapter, the results of the geochemical analysis tion. The results show that there is little clear evidence for will be compared with forming and decorating techniques, the import of non-local ceramics into the area. Figures which will complete the technological classification. Chap8.34–8.36 show exotic sherds grouping within the range of ter 9 will examine procedure eight in this classification locally produced Madang style pots. In each plot, K-7 and scheme, in an attempt to clarify techno-style groupings N-188 are clear outliers, but both fall within the range of within the Madang style (following Roux 2011), distinMadang style technical groupings. These outliers are pos- guishing the number of production groups working within sibly derived from different clay sources. However, post- the broader community of practice. depositional alterations may also be at play. Importantly, Clay composition Designation Sample Rim T410 Rim Y5 Rim N3712 Body T230 Rim T188 Body T520 Body T604 Ring base T913 5 10 15 Figure 8.30. Ward’s HCA of exotic sherds. 181 20 25 Chapter . Pre-Colonial Potting II: Procurement and Distribution 5 Designation Sample Rim T410 Rim Y5 Rim N3712 Rim T188 Body T230 Body T520 Ring base T913 Body T604 10 15 20 25 Figure 8.31. Group average HCA of exotic sherds. 1.0 Si K 2.0 Na 0.5 N-3712 Fe Principal component 2 (31.2%) 0.0 1.0 Ti Ca -0.5 Mg T-520 Al -1.0 -1.0 0.0 0.5 1.0 Y-5 T-410 T-913 0.0 -0.5 T-188 T-230 -1.0 T-604 -2.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 Principal component 1 (39.3%) Tilu Nunguri Yabob Island Figure 8.32. PCA showing clay compositional data from exotic sherds by location. 182 4.0       · .  1.0 Si 2.0 0.5 Al K 0.0 Ca Na Mg -0.5 Fe -1.0 -1.0 -0.5 0.0 0.5 1.0 0.0 -1.0 -2.0 -3.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 Principal component 1 (47.3%) Yabob clay Bilbil clay Tilu Yabob Island Nunguri Figure 8.33. PCA showing clay compositional data from exotic sherds and ethnographic clays. 1.0 Mg 4.0 0.5 K Fe Si 0.0 Ti Na -0.5 Principal component 2 (20.3%) Principal component 2 (20.2%) 1.0 Ti Al Ca 2.0 -1.0 -1.0 -0.5 0.0 0.5 1.0 0.0 -2.0 -4.0 -4.0 -2.0 0.0 2.0 4.0 6.0 8.0 Principal component 1 (30.5%) Nunguri Tilu Kranket Island Yabob Island Siar Island Exotics Figure 8.34. PCA showing clay compositional data from all sherds by location. 183 5.0 Chapter . Pre-Colonial Potting II: Procurement and Distribution 1.0 Mg 4.0 0.5 K Fe Si 0.0 Ti Na -0.5 Al Principal component 2 (20.3%) Ca 2.0 -1.0 -1.0 -0.5 0.0 0.5 1.0 0.0 -2.0 -4.0 -4.0 -2.0 0.0 2.0 4.0 6.0 8.0 Principal component 1 (30.5%) Class 1 Class 2 Class 3 Class 4 Class 5 Exotic Figure 8.35. PCA showing clay compositional data from all sherds by class. 1.0 Mg 4.0 0.5 K Fe Si 0.0 Ti Na -0.5 Al Principal component 2 (20.3%) Ca 2.0 -1.0 -1.0 -0.5 0.0 0.5 1.0 0.0 -2.0 -4.0 -4.0 -2.0 0.0 2.0 4.0 6.0 8.0 Principal component 1 (30.5%) Fe-Mg Light Calcareous Grog Exotic Figure 8.36. PCA showing clay compositional data from all sherds by fabric. 184 Chapter 9. Pre-Colonial Potting III: Decorating Generally, a decorated piece…is preferred to one in its plain state, because grass-roots art pleases the eye, regales the mind, and reminds the viewer of …traditional concepts and ways of life. — E.F. Hannemann, Madang (1968)1 concept of style. As presented in Chapter 7, style is often acknowledged as the ‘content and appearance’ of a pot, including form and decoration (see Specht 1969: 64). However, unlike form, specific decorations are often viewed as important for understanding something of the maker’s social identity and symbology (e.g. Allen 1984a; Rathje 1978). As such, decoration has often been used to distinguish chronologically restricted ‘styles’ (or synonymous terms) to construct regional sequences of social and technological change. The issue of decoration in particular has featured prominently in Lapita and post-Lapita sequences of the Bismarck Archipelago and abroad (e.g. Bedford & Clark 2001; Spriggs 2004; Summerhayes 2001c), in various ceramic phases and periods of the Massim (e.g. Bickler 1998; Egloff 1979; Lauer 1974), and in the Lapita, Early Papuan Pottery (EPP), and post-EPP sequences on the south coast of New Guinea (e.g. Allen 2016; Allen et al. 2011, Bulmer 1982; David et al. 2011; Skelly 2014; Summerhayes & Allen 2007). These approaches tend to be culture historical and invoke, often quite justifiably, diffusionist or migrationist paradigms. Decoration is the final important aspect of technology for identifying group interaction and knowledge exchange. This chapter describes the decorating processes used by pre-colonial Madang potters in an attempt to clarify the groupings derived in the previous chapters. The specific aims of this chapter are: (1) to examine the decorating processes used by Madang style potters over the past 500–600 years, and (2) compare decorating techniques with compositional and forming techniques to distinguish if specific production groups used distinct decorations. These aims tie in to broader concerns about how technological change, innovation, and knowledge acquisition took place in the community of practice. From a culture history perspective, particular decorative processes may be peculiar to certain timeframes, acting as useful relative dating devices when comparing different ceramic assemblages around the northeast coast. Lapita decoration has received the most attention owing to its highly ornate decorative systems. Some studies have To address these concerns, methodological issues pertain- focused on the tools used to produce decorations (e.g. ing to the analysis of decoration are discussed and a social Ambrose 2007; Bedford 2006b; Best 2002: 47) or on the and technological approach is described. The analytical intersections of decorative processes with other technoprocedures of the Madang decorative analysis are then logical processes such as tattooing (e.g. Green 1979; Kirch presented, including the nature of classification and attrib- 1997: 131). Other studies have explored the range of motifs ute analysis. Lastly, the results of the decorative analysis and their chronological and geographical distributions are presented and compared to formal and compositional (e.g. Anson 1983, 1986, 2000; Chiu 2015; Sand 2007), the attributes. Pottery from each excavated assemblage and skill and standardisation of the decorations (e.g. Clark surface survey are treated separately, to describe pottery 2007; Clark & Murray 2006; Hogg 2011), or the social technology on-site. This will be tied together in the discus- meaning behind specific motifs (e.g. Chiu 2005, 2007; sion of production and exchange in the next chapter. Schechter & Terrell 2009). Lapita scholars have also long been aware that pottery decoration needs to be investigated from a structural rather than a purely descriptive Methodological issues perspective (Mead 1975; Sharp 1988; Leblanc 2016). Decoration and New Guinea ceramics Decoration on recent New Guinea ceramics has received Generally, decorative analyses of New Guinea-area ce- less serious attention (outside of several doctoral disserramics have been, implicitly or explicitly, guided by the tations mentioned in chapters 7 and 8 and several papers by Bulmer 1970, 1971, 1985). Superficially, however, some decorative arrangements seem to be restricted to specific 1 In Hannemann (1969: 2), Grass-Roots Art in New Guinea. 185 Chapter . Pre-Colonial Potting III: Decorating places, while others bear distinct similarities that can be seen from the Sepik coast to the North Solomons, especially with regard to simple incised and applied techniques (see Schurig 1930). Technology as methodology So, pottery decorations, like vessel morphology, can closely resemble each other over large geographical distances and this is particularly the case in New Guinea (Petrequin & Petrequin 1999). Could this be because of interaction, common descent or convergent innovations? To evaluate such differences and similarities systematically, the concept of isochrestic variation (following Lemonnier 1992; Sackett 1986) is again useful. Similar decorative forms may in fact represent distinct technological choices pertaining to the tools employed, the gestures used, and the intention of the designers (i.e. technical elements). Within the framework of embodied knowledge, the boundaries between these tools, gestures, and intentions become increasingly blurred over time, increasing the durability of the technological process, and as multiple technical elements are patterned and interlinked they become technical syntaxes. The forms of the decorations in themselves represent the physical result of such processes. Singular technical elements and short syntaxes may be widely distributed within and between production groups owing to interaction and imitation (Mayor 2005; see Chapter 7, Fig. 7.1); however, it is assumed that more complex operations, involving extended syntaxes, will be less widely distributed and primarily passed on through learning and descent. ingly subtle and unconscious to the potter. Describing the changing distribution of technical elements and syntaxes will then be an important proxy for determining the nature of cultural learning and sharing between production groups over time. Although the processes of knowledge acquisition between generations and across social groups are universal occurrences, these processes are not invariant. Rather, they are locally and historically situated and different contexts of learning will affect the nature of knowledge acquisition and the distribution of technical elements (see Roux 2007). Ethnoarchaeological studies have shown that social learning is not straightforward. For instance, amongst the Kalinga in the Philippines, learning primarily occurs vertically, mother-to-daughter, but there is also a strong horizontal impact from peers of a similar age (Graves 1981; Longacre 2008). Further, ethnographic studies of social learning demonstrate that throughout a person’s lifetime the nature of learning can change, from being predominantly vertical in early stages of life, to being primarily horizontal in later stages (e.g. Hewlett et al. 2011; Tehrani & Collard 2009). The sharing of technical knowledge between production groups is not necessarily dictated only by geographic distance (contra Welsch et al. 1992), but also by kinship, gender, social relations, environment, belief, and language affinities. Sassaman and Rudolphi (2001) demonstrate an instance of this for some North American pottery traditions. As the potters in question were probably female, and assuming a similar kinship structure and exogamous marThis point is important. Although technological syntaxes ital system to today, then women would have commonly are more durable strands of materiality, those technologi- participated in at least two production communities in cal processes with shorter syntaxes (e.g. modern Madang their life. First, in their natal communities, they began to decoration) are usually more prone to visible innovation learn pottery making from mothers, aunts, grandmothresulting in variation to technical elements than longer ers, and sisters. Second, they would have continued to pot syntaxes such as vessel forming (Hosfield 2009: 53). To in their marital communities. The movement of women dabble in evolutionary speculation, decoration may also potters between different communities would have been a be more likely to be innovative because the specific form source of technical variation, introducing ideas from one of decoration is unlikely to affect the success of the form- group to the next relatively quickly, especially as these ing and firing phase, necessary for the production group’s women began to teach their daughters and others in their wellbeing and ability to maintain social reciprocity. How- marital communities. ever, this does not clarify how or why innovations occur. Another source of variation is related to skill level and To further investigate the interplay between technology, degrees of embodiment within the given technological knowledge, and innovation we return here to the learning process. Ethnoarchaeological studies elsewhere in the process and modes of social learning. The nature of learn- world suggest that variation will occur as very experienced ing has been shown to affect the durability and transfer potters consciously innovate and novices make mistakes of specific technical syntaxes and elements. For instance, (Dietler & Herbich 1998; Roux 1990; Roux et al. 1995). a recent review (Hosfield 2009) found a significant as- Whether these variations were adopted, tolerated, or counsociation between vertical learning and conservatism in teracted would have depended upon the social regulations technical knowledge, while horizontal learning between important to maintaining the processes of decorating. members of the same or different production groups was more often associated with innovation and change. As It is tempting to apply similar frameworks to Madang. ToGosselain (1992) points out, gestural innovation is unlikely day, sharing of technical knowledge is primarily vertical to arise during vertical learning processes as the parent or or oblique for Bel girls born into a potting village, but for teacher will correct movements that are considered mis- those who marry in from non-potting villages, the process takes, and there is less likelihood of major variations as is different, being primarily horizontal. Those women who these ‘correct’ procedures become embodied and increas- marry into potting groups and begin to learn the tech186       · .  niques would introduce variation and innovation, drawing upon accumulated gestural knowledge and memories from other communities and technological processes. Conversely, the movement of experienced Bel potters into other production groups would introduce another source of innovation and difference but might also prevent divergent strands of technological process developing, maintaining similarities between groups. In the past, oblique and vertical learning may have been the major way in which potters learnt vessel forming and decorating techniques. If this were the case, gestures and tools may have been used in very consistent ways over the past 600 years. Individual technical elements may have changed rapidly owing to people’s capacities to improvise, especially at the ends of the skill spectrum, but if there was a similar marriage and kin structure to today we would expect these innovations to be widespread amongst the community of practice. Method The Madang Classification To address the organisation of production groups and the application of minor technical elements within this community of practice, this chapter will complete the hierarchical classification, which has featured throughout chapters 7 and 8. The previous chapters have distinguished groupings based on procurement and forming processes. This final stage of the technological classification will compare techno-compositional groupings (along with technical classes, techno-fabric/mineralogical groups) with minor technical variants – in this case decorating techniques. This analysis will distinguish whether particular decorations are associated with particular production groups or with particular forming techniques, used in multiple production groups. Doing so will establish techno-style groups and emphasise micro-technical traditions within the local Madang area if they exist (see Fig. 9.1). Ceramic assemblage Intuitive sorting Provisional style groups Hierarchical sorting based on technical syntaxes Technical classes Classification based on variation in technical elements Technical variants Sorting based on ocular microscopy Techno-fabric groups cross comparison with mophological and decorative attributes Classification based on mineralogial identification Techno-mineralogical groups Grouping based on clay composition Techno-compositional groups Techno-style groups Figure 9.1. Procedures of a technological classification (adapted from Roux 2011). The procedures examined in this chapter are highlighted yellow. 187 Chapter . Pre-Colonial Potting III: Decorating 3. ‘Impression’ is formed by pressing a tool into the clay, perpendicular or at an angle to the vessel surface, but Single decorative forms are easily shared through group not moving it laterally (Skelly 2014: 97). Impressions are interaction and so are potentially poor indicators of more usually shallow and imprint the form of the tool used durable strands of cultural and technological similarities to produce them. ‘Punctation’ is a form of impression or difference. However, the structures behind decoration caused by pushing a sharp object deeply into the clay, involving more extended syntaxes, including the combibut not so far as to perforate the interior surface (Rice nations and placement of attributes, are probably more 1987: 145). Egloff (1975) used the term ‘inner rim notchdiagnostic of production communities (Arnold 1983, 1984; ing’ to describe impressions along the interior corner Friedrich 1970). The description of surface decoration point. Here, this category is subsumed by the impression should therefore describe 1) the application methods (i.e. and punctation category (although Egloff ’s category will gestures/tools) used, 2) the configurations of these techbe considered later in the results). nical elements (i.e. intentional use of specific shapes and patterns), and 3) the location of these configurations on 4. ‘Paddle impression’ is here a separate method of imthe vessel (i.e. the intentional use of specific space). The pression owing to the different tools required and the presence and absence of different configurations on difdifferent gestures used. Paddle impressions may be creferent areas of the pot are guided by social rules. Therefore, ated through the use of specialised carved paddles for specific technical elements and their configurations may surface decoration or plain paddles may be used with be spatially or temporally restricted such that they can be different gestures to produce distinct linear impressions. used as chronological indicators of specific production groups. Consequently, the following decorative attributes Body sherds were designated as ‘plain’ if they did not diswere recorded on all sherds exhibiting surface alteration. play any decoration, while formal sherds were conservatively designated as ‘plain’ only if a significant proportion I. Application method refers to the way in which decora- of the neck and shoulder were preserved without decoration is applied to the surface of a pot. In the Madang as- tion. semblages, four broad application methods were identified, each with several sub-methods: II. The decorative motifs of the Madang style have not previously been reported in detail. Here, a hierarchical 1. Appliqué is created by the addition of plastic clay or slip approach is used to investigate the structure of decoratto the surface of a vessel using a tool or by hand (Ir- ing and the patterning of different application methods. win 1985: 109). In the Madang style, appliqué appears to Design elements are the smallest self-contained attributes have been applied with a sectioned semi-circular piece under analysis, directly resulting from a single technical of bamboo or similar tool. This broad group can be sub- element (e.g. one deliberate scrape of the tool) (see Chiu divided based on the gestures used to apply the clay/slip. and Sand 2005; Siorat 1990). Design configurations are ar‘Nubble appliqué’ is formed by the application of small, rangements of the same design elements to fit a spatial discrete dots onto the vessel. ‘Linear appliqué’ is the ap- dimension on the pot (Friedrich 1970). Design motifs, on plication of thick or thin lines of clay/slip. ‘Appliqué the other hand, combine two or more design configuravariant’ is a category to describe the method of applying tions or design elements (this is here synonymous with lines of appliqué, which are then scraped away at regular Friedrich’s ‘secondary element use’). Motifs are cognitively intervals to produce nubbins (see Egloff 1975: 3). Some different from configurations, as they require the repetiappliqué lines are not scraped but deliberately incised tion and patterning of multiple technical elements. The (‘incised linear appliqué’), while others are impressed range of motifs exhibited describes which combinations of with the side or flat end of the tool (‘impressed linear technical elements are considered correct within the speappliqué’). If preservation is only sufficient to determine cific social and temporal practice of decorating. A number the presence of appliqué and not the specific type, then of configurations (Fig. 9.2–9.3) and motifs (Fig. 9.4–9.6) the sample is designated ‘appliqué (?).’ was identified, which were coded numerically. Technical attributes 2. Incision is the cutting of the vessel surface, pulling the tool parallel or at an angle to the vessel surface (Orton & Hughes 2013: 89). All incision in the Madang assemblages was completed prior to slipping and firing using a bamboo or wooden tool. Incision can be subdivided based on the specific gestures used in application. ‘Linear incision’ is a freehand technique, which uses a sharp tool to produce long thin lines. ‘Groove-incision’ is completed with a broader, blunter tool or part of the tool, which creates shallow and wide grooves (Rice 1987: 146). ‘Gash-incisions’ are produced by cutting controlled, short (<5 mm) and sometimes deep dashes into the clay (Irwin 1985: 109). III. The location of decoration describes which design elements were publicly viewed and those that were only seen upon closer inspection or in use (Skelly 2014: 100). This has important implications for examining the communicative and more readily imitated elements of design. Moreover, within different communities specific motifs or configurations might be appropriate on some parts of the vessel but not others. Decorative location was recorded on formal sherds, divided into discrete areas including lip, inside lip, outside lip, interior corner point, neck, sub-rim groove, sub-rim groove (ridge), and body/shoulder (Fig. 9.7). All decorated non-rims were designated as ‘body/ shoulder.’ 188       · .  Impression Nubble appliqué LA10 NA1 NA2 Im1 Im2 LA11 Im3 NA3 LA12 Im4 NA4 LA13 Punctation Pu1 NA5 LA13 Pu2 Appliqué variant Pu3 NA6 Linear appliqué AV1 Paddle impression PI1 LA1 AV2 LA2 PI2 LA3 Incised linear appliqué PI3 LA4 InLA1 PI4 LA5 InLA2 PI5 Impressed linear appliqué PI6 LA6 ImLA1 PI7 LA7 PI8 ImLA2 PI9 LA8 LA9 ImLA3 Figure 9.2. Design configurations represented in the Madang style. 189 PI10 Chapter . Pre-Colonial Potting III: Decorating Linear incision LI1 LI16 GI16 LI2 LI17 GI17 LI3 Gash incision GI18 GI1 GI19 GI2 LI4 GI20 GI3 LI5 GI21 GI4 GI22 LI6 GI5 LI7 GI23 GI6 LI8 GI7 LI9 GI24 GI8 GI25 GI9 LI10 GI26 GI10 LI11 Groove incision GI11 GrI1 LI12 GI12 LI13 GI13 GrI2 LI14 GrI3 GI14 GI15 LI15 Figure 9.3. Design configurations represented in the Madang style (cont.). 190 GrI4       · .  Appliqué bands Appliqué intersects M1 M23 M11 M31 M24 M2 M12 M32 M25 M3 M26 M13 M33 M4 M27 M14 M34 Combination M5 incision M15 M35 M16 M28 M6 Curved appliqué bands M17 M7 M36 M18 M19 M29 M8 Appliqué end points M20 M9 M10 M21 M22 M30 Figure 9.4. Appliqué design motifs represented in the Madang style. 191 Chapter . Pre-Colonial Potting III: Decorating Linear/gash incised bands M50 M37 M63 M51 M38 M39 M64 M52 M75 M76 M65 M53 M77 M54 M66 M40 M78 M67 M55 M41 M79 M56 M68 M42 M80 M57 M43 M69 M58 M81 M44 M59 M70 M45 Incised intersects M82 M71 M83 M46 M60 M47 M72 M61 M84 M48 M73 M49 M74 M62 Figure 9.5. Linear-gash incised design motifs represented in the Madang style. 192 M85       · .  Groove incised bands Paddle impressions M86 M92 M87 M97 Impressed bands M93 M88 Combination M94 linear incision M98 impression M89 linear incision M95 punctation M90 groove incision impression M91 M99 M96 Figure 9.6. Groove incised, impressed and paddle impressed design motifs represented in the Madang style. IV. Slipping is an important stage in a pot’s production and initiation into the social world in Madang style ceramics (see Chapter 4). Slip was recorded as present or absent for both the interior and exterior surfaces of all formal and decorated sherds. The extent to which slip is applied to the interior surface in particular may be diagnostic of specific technical conventions. Results: Nunguri The following sections describe the results of the decorative analysis of body and rim sherds from Nunguri, Tilu, and surface survey around Madang. Nunguri ceramics were decorated using several different application methods (Table 9.1). In Layer 1 (Spits 1–15), appliqué (Fig. 9.8– f e b e d c a) Body/shoulder f g h g b) Sub-rim groove (ridge) c) Sub-rim groove d) neck a e) outside lip f) lip a g ) inside lip h) interior corner point Figure 9.7. Locations of decoration on rim sherds. 193 Chapter . Pre-Colonial Potting III: Decorating Table 9.1. Application methods represented at Nunguri, Test Pit 1 by excavation spit.* Excavation spit Application method 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Total Appliqué Nubble appliqué 8 67 54 27 18 30 28 26 28 38 63 49 76 87 44 643 Linear appliqué 29 175 192 132 58 112 63 45 47 32 62 42 61 97 38 1185 Appliqué variant 5 11 21 13 5 10 10 9 3 4 2 3 9 15 11 131 Incised linear app. – 2 3 – 1 – – – – – – 1 3 – – 10 Impressed linear app. – 15 17 10 4 6 3 4 – – – 2 – 1 2 64 Appliqué (?) 7 48 43 45 12 32 18 16 14 22 38 35 47 56 22 455 49 318 330 227 98 190 122 100 92 96 165 132 196 256 117 2488 Linear incision 35 95 61 37 15 45 42 34 24 29 43 37 35 17 12 561 Gash incision 15 44 49 22 4 19 23 22 14 11 21 19 16 20 4 303 3 16 12 4 1 2 2 – 2 3 1 – – – 1 47 53 155 122 63 20 66 67 56 40 43 65 56 51 37 17 911 Impression 13 92 74 55 21 35 21 16 21 14 11 11 15 20 26 445 Punctation – 5 4 4 1 – 1 1 – – 2 1 – 3 3 25 13 97 78 59 22 35 22 17 21 14 13 12 15 23 29 470 Subtotal Incision Groove incision Subtotal Impression Subtotal Paddle impression Paddle impression Total 12 18 9 14 2 6 1 4 – 3 3 1 – 3 9 85 127 588 539 363 142 297 212 117 153 156 246 201 262 319 172 3954 * Note, the counts in this table refer to all observable cases of the application method, on both body and formal sherds. Thus, some sherds may preserve evidence of >1 application method. 9.10), incision (Fig. 9.11–9.14), impression, and paddle impression (Fig. 9.15) were used to decorate the exterior body or shoulder, while punctation and impression in particular were used to decorate other locations. No decorated sherds were recovered from Layer 2 (Spits 16–17). 13 (adj. residual = 3.9), 14 (adj. residual = 7.2), and 15 (adj. residual = 3.7). Conversely, incision is more common than expected in spits 1 (adj. residual = 4.9), 2 (adj. residual = 8.4), 7 (adj. residual = 2.9), and 8 (adj. residual = 2.4), and less than expected in spits 4 (adj. residual = –2.3), 5 (adj. residual = –2.6), 11 (adj. residual = –6.4), 14 (adj. residual = –5.2), Within each application method, several sub-methods are and 15 (adj. residual = –3.6). Paddle impression and imclearly distinguishable. For instance, appliqué was usually pression also both became more common over time. Cases applied as a line or nubbin, but was also applied and subse- of paddle impression are substantially more than expected quently impressed, inscribed, or scraped (appliqué variant). in spits 1 (adj. residual = 5.8), 2 (adj. residual = 3.3), 4 (adj. Incision was usually inscribed in long freehand lines or residual = 2.4), and 15 (adj. residual = 2.8), and less than exsmall gashes, and rarely as thicker grooves. Impression was pected in spits 9 (adj. residual = -2), 13 (adj. residual = –2.7). particularly prominent on the interior corner point, while Impression is substantially more common than expected punctation was less common and usually limited to the lip. in spits 2 (adj. residual = 4.8), 3 (adj. residual = 4.1), and 6 A substantial number of plain sherds was recovered from (adj. residual = 3.7), and less than expected in spits 10 (adj. Nunguri (n = 32,835), but only eight rims were complete residual = –2.4), 11 (adj. residual = –2.3), 12 (adj. residuenough (including lip, rim, and body/shoulder portions) al = -2.5), 13 (adj. residual = –3.1), 14 (adj. residual = –3.8), to be assigned as ‘plain’ with any certainty. and 15 (adj. residual = –2.8). The proportion of each application method changes significantly over time relative to excavation spit (Pearson Chi-Squared test, X2 = 405.606, df = 42, P<0.01). Figure 9.16 shows the general trend of this change, with appliqué becoming less common over time and incision, paddle impression, and impression becoming more common. Cross tabulation of the chi-squared results show that appliqué is substantially less common than statistically expected in spit 2 (adj. residual = –6.3), spit 3 (adj. residual = –11.1), and spit 8 (adj. residual = –2.6), but more common than expected in spits 5 (adj. residual = 2.3), 11 (adj. residual = 7.1), The relative proportion of sub-methods also changes through the excavation spits, indicating that the technique of decorating with each method changed. The sub-method of appliqué is significantly associated with excavation spit (X2 = 240.175, df = 70, P<0.01). Importantly, impressed appliqué and linear appliqué both become more common over time, while nubble appliqué becomes less common. The sub-methods of incision (X2 = 46.454, df = 28, P = 0.016), and impression (X2 = 40.133, df = 14, P = <001), are also statistically significantly related to individual excavation spit, but there are no clear linear trends over time. 194       · .  N-1113 Spit 3 M2 N-2278 Spit 8 M2 N-2685 Spit 10 M1 N-2082 Spit 7 M3 N-2075 Spit 7 M4 N-495 Spit 2 M2 N-3122 Spit 12 M6 N-983 Spit 3 M7 N-2080 Spit 7 M11 N-1510 Spit 4 M11 0 5 cm Figure 9.8. Examples of appliqué decoration on Madang style body sherds at Nunguri. 195 Chapter . Pre-Colonial Potting III: Decorating N-2084 Spit 7 M25 N-826 Spit 3 M24 N-572 Spit 2 M11/M22 N-987 Spit 3 M28 N-398 Spit 2 M27 N-1061 Spit 3 M27 N-1301 Spit 4 M32 N-1644 Spit 5 M32 N-1321 Spit 4 M30 N-1620 Spit 5 M35 N-3841 Spit 15 M32 0 5 cm N-1912 Spit 6 M36 Figure 9.9. Examples of appliqué decoration on Madang style body sherds at Nunguri (cont.). 196       · .  N-912 Spit 3 M12 N-273 Spit 2 M12 N-93 Spit 1 M11 N-573 Spit 2 M13 N-1018 Spit 3 M13 N-1661 Spit 5 M12 N-1319 Spit 4 M15 N-2198 Spit 7 M14 N-446 Spit 2 M16 N-317 Spit 2 M18 0 5 cm Figure 9.10. Examples of appliqué decoration on Madang style body sherds at Nunguri (cont.). 197 Chapter . Pre-Colonial Potting III: Decorating N-1406 Spit 4 M38 N-459 Spit 2 M37 N-3042 Spit 12 M41 N-541 Spit 2 M42 N-490 Spit 2 M42 N-1386 Spit 4 M42 N-1786 Spit 6 M43 N-2359 Spit 8 M42 N-2054 Spit 7 M43 N-3794 Spit 15 M46 N-1678 Spit 5 M44 0 N-828 Spit 3 M46 5 cm Figure 9.11. Examples of incised decoration on Madang style body sherds at Nunguri. 198       · .  N-266 Spit 2 M46 N-476 Spit 2 M48 N-2379 Spit 8 M47 N-473 Spit 2 M46 N-1346 Spit 4 M53 N-1119 Spit 3 M51 N-1033 Spit 3 M53 N-2916 Spit 11 M49 N-38 Spit 1 M56 N-452 Spit 2 M56 0 5 cm N-2269 Spit 8 M58 Figure 9.12. Examples of incised decoration on Madang style body sherds at Nunguri (cont.). 199 Chapter . Pre-Colonial Potting III: Decorating N-802 Spit 3 M64 N-2102 Spit 7 M61 N-432 Spit 2 M63 N-2574 Spit 10 M68 N-2666 Spit 10 M64 N-1674 Spit 5 M64 N-3633 Spit 14 M71 N-2783 Spit 11 M73 N-1135 Spit 3 M71 N-2864 Spit 11 M76 N-1064 Spit 3 M74 N-827 Spit 3 M76 0 5 cm N-397 Spit 2 M79 N-451 Spit 2 M78 Figure 9.13. Examples of incised decoration on Madang style body sherds at Nunguri (cont.). 200       · .  N-3094 Spit 12 M78 N-2742 Spit 11 M80 N-2740 Spit 11 M81 N-2193 Spit 7 M82 N-3081 Spit 12 M83 N-2513 Spit 9 M84 N-293 Spit 2 M85 N-1087 Spit 3 M86 N-2652 Spit 10 M87 0 5 cm N-891 Spit 3 M91 N-1011 Spit 3 M90 Figure 9.14. Examples of incised decoration on Madang style body sherds at Nunguri (cont.). 201 Chapter . Pre-Colonial Potting III: Decorating N-36 Spit 1 M95 N-37 Spit 1 M99 N-577/591 Spit 2 M96 0 5 cm Figure 9.15. Examples of paddle impression on Madang style body sherds at Nunguri. Nunguri decorated sherds differ significantly by body Table 9.2. Average body thickness (mm) by application thickness relative to application method (Table 9.2–9.4). method at Nunguri. However, the effect size (η2) of 1.3% suggests that decoraAppliqué Incision Impression Paddle impression tion method does not account for much of the variation Mean 4.82 4.90 4.19 4.61 in body thickness. A post hoc Kruskal-Wallis test (Table SD 1.65 1.56 1.40 1.71 9.3) shows that there are statistically significant differences between almost all of the decorative methods: between CV 34% 32% 34% 37% the thicknesses of appliqué and incised sherds (X2 = 4.430, df = 1, P = 0.04), appliqué and impressed sherds (X2 = 26.857, df = 1, P<0.01), incised and impressed sherds (X2 = 34.874, df = 1, P = 0.01), incised and paddle impressed sherds Table 9.3. Kruskal-Wallis test of body thickness (mm) by (X2 = 3.881, df = 1, P = 0.49), and between impressed and application method at Nunguri. paddle impressed sherds (X2 = 6.485, df = 1, P<0.01). The Paddle difference between appliqué and paddle impressed sherds Appliqué Incision Impression impression is not statistically significant (X2 = 1.208, df = 1, P = 0.27). No. of cases 1992 621 153 73 Mean rank 1424.62 1504.33 1066.68 1316.99 A wide range of decorative configurations is apparent on Madang style sherds from Nunguri (Table 9.5–9.7). Nub- X² =36.217, df = 3, η² =1.3%, P<0.01 202       · .  100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Spit Appliqué Incision Impression Paddle impression 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Spit Impressed linear appliqué Nubble appliqué Incised linear appliqué Linear appliqué Appliqué variant 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 14 15 Spit Linear incision Groove incision Gash incision 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 13 Spit Impression Punctation Figure 9.16. Percentage of application methods represented on the external body/shoulder of sherds at Nunguri, Test Pit 1 by excavation spit. 203 Chapter . Pre-Colonial Potting III: Decorating Table 9.4. Mood’s test of body thickness (mm) by application method at Nunguri. Appliqué Incision Impression Number of cases 1992 621 153 Paddle impression 73 Thickness > median 1006 338 43 32 Thickness ≤ median 986 283 110 41 Median = 4.45, X² = 35.521, df = 3, P < 0.01 ble appliqué was usually applied as single (NA1) or double (NA2) bands, while linear appliqué was usually applied as straight lines (LA1) or diagonal dashes (LA4). Single straight lines of scraped (AV1) and impressed (ImLA1) appliqué were also commonly applied. Linear incision was usually inscribed as freehand lines (LI1), along with a wider variety of small gash arrangements. Paddle impressions usually consist of multiple lines (PI1) or chevrons (PI3). Simple impressions (Im1 and Im3) were sometimes made onto the body/shoulder of the pot but were used frequently to decorate the interior corner point. Punctation (Pu3) was sometimes used to decorate the lip and, rarely, the body/shoulder. sels at Nunguri. Punctation was mainly used to punch holes around the top of the lip, but shallower impressions, and rare gash incisions were also made (Table 9.7). There appears to be a general trend over time towards leaving the lip undecorated as Spit 14 and Spit 15 contain substantially more cases of lip decoration than do Spits 1–13 (Figure 9.17). The outside of the lip was also, very occasionally, decorated with incision, impression/punctation, and appliqué. These decorations are rare and there is no clear trend over time related to this decorative zone. The inside lip was only occasionally decorated on incurving vessels and never on everted/inverted vessels. Impressions made around the interior corner point of Decoration is almost exclusively limited to the body/ Nunguri pots are common (Table 9.7). These impressions shoulder. Lip decoration is only present on 3% of all ves- vary slightly in shape and size, reflective of the tool being Table 9.5. Appliqué configurations represented at Nunguri, Test Pit 1 by location. Body Lip Outside lip Inside lip Sub-rim groove Sub-rim groove ridge NA1 303 – – – – – Interior corner point – NA2 96 – – – – – – NA3 1 – – – – – – NA4 – – – – – – – NA5 – – – – – – – NA6 3 – – – – – – LA1 441 – – – – – – LA2 3 – – – – – – – LA3 12 – – – – – LA4 252 – 1 – – – – LA5 25 – – – – – – LA6 9 – – – – – – LA7 16 – – – – – – LA8 2 – – – – – – LA9 17 – 1 – – – – LA10 7 – – – – – – LA11 1 – – – – – – LA12 1 – – – – – – LA13 2 – – – – – – – LA14 2 – – – – – AV1 118 – – – – – – AV2 8 – – – – – – InLA1 8 – – – – – – InLA2 1 – – – – – – ImLA1 62 – – – – – – ImLA2 2 – – – – – – ImLA3 1 – – – – – – 204       · .  Table 9.6. Incised configurations represented at Nunguri, Test Pit 1 by location. Body Lip Outside lip Inside lip Sub-rim groove Sub-rim groove ridge LI1 318 – – – – – Interior corner point – LI2 21 – 1 – 1 – – LI3 6 – – – – – – LI4 1 – – – – – – LI5 – – – – – – – LI6 – – – – 1 – – LI7 3 – – – – – – LI8 14 – – – – – – LI9 3 – – – – – – LI10 2 – – – – – – LI11 – – – – 1 – – LI12 7 – – – – – – LI13 2 – – – – – – LI14 2 – – – – – – LI15 9 – – – – – – LI16 17 – – – – – – LI17 1 – – – – – – GI1 16 – – – – – – GI2 27 – – – – – – GI3 92 2 2 – 1 – – GI4 8 – – – – – – GI5 3 – – – – – – GI6 15 – – – – – – GI7 23 – – – – – – GI8 10 – – – – – – GI9 57 – – – – – – GI10 4 – – – – – – GI11 1 – – – – – – GI12 15 – – – – – – GI13 1 – – – – – – GI14 5 – – – – – – GI15 14 – – – – – – GI16 1 – – – – – – GI17 1 – – – – – – GI18 1 – – – – – – GI19 3 – – – – – – GI20 2 – – – – – – GI21 – – – – – – – GI22 2 – – – – – – GI23 2 – – – – – – GI24 1 – – – – – – GI25 1 – – – – – – GI26 – – – – – – – GrI1 36 – – – – – – GrI2 2 – – – – – – GrI3 2 – – – – – – GrI4 1 – – – – – – 205 Chapter . Pre-Colonial Potting III: Decorating Table 9.7. Impressed and paddle impressed configurations represented at Nunguri, Test Pit 1 by location. Body Lip Outside lip Inside lip Sub-rim groove Sub-rim groove ridge Interior corner point Im1 59 4 1 4 – – 272 Im2 2 – – – – – – Im3 30 – – – – – – Im4 1 – – – – – – Pu1 – 1 1 – – – – Pu2 3 – – – – – – Pu3 1 19 1 – – – – PI1 5 – – – – – – PI2 17 – – – – – – PI3 10 – – – – – – PI4 3 – – – – – – PI5 3 – – – – – – PI6 – – – – – – – PI7 3 – – – – – – PI8 2 – – – – – – PI9 – – – – – – – PI10 1 – – – – – – 16% 14% 12% 10% 8% 6% 4% 2% 0% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Spit Figure 9.17. Percentage of Nunguri rims with decorated lips by excavation spit. used and the gestures of the potter, but they are all classified here as Im1, a shallow impression, probably produced using the side of a wooden tool. As described by Egloff (1975), ‘inner rim notching’ refers to impressions made around the circumference of the interior corner point of everted/inverted rims or on the inner lip of incurving and direct rims. Inner rim notching (combining Im1 on interior corner point of Class 1, 2, 3, and 5 rims, and Im1 on the inner lips of Class 4 rims) was present on just under half of all Nunguri rims (47%, n = 276). There is a significant association between inner rim notching and excavation spit (X2 = 46.595, df = 15, P<0.01) and it is possible that inner rim notching became less common over time. However, although Figure 9.18 illustrates a generally negative trend over time, the high variance between each spit means this interpretation is tentative. Numerous motifs are present in the Nunguri assemblage (Table 9.8). Appliqué motifs account for 57% of the total observed cases, while incision accounts for 41%, and impressed and paddle impressed motifs account for less than 1% each. Combinations of two or more application methods to form motifs are extremely rare and also represent less than 1% of the observed cases. Notably, M11 appliqué motif, with double linear band flanked by short linear diagonals, is the most common motif represented at the site. This is particularly the case in Spits 1–9. In Spits 10–15, M1 appliqué motif, with a single linear band flanked by nub- 206       · .  90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Spit Figure 9.18. Percentage of Nunguri rims with inner rim notching by excavation spit. Table 9.8. Motifs represented at Nunguri, Test Pit 1 by excavation spit. Appliqué M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 M12 M13 M14 M15 M16 M17 M18 M21 M22 M24 M25 M27 M28 M29 M30 M31 M32 M34 M35 M36 Subtotal 1 2 3 4 5 6 3 – – – – – – – – – 2 – 1 – – – – – – – – – – – – – – – – – – 6 8 – – 1 1 – – – – 1 11 1 2 – 1 1 – 2 – 1 1 – 2 – 1 – – – 1 – – 35 5 2 – – – – 1 – – – 24 2 2 – – – – 1 1 – 1 – 2 1 – – – – – – – 42 3 – – – – – – 1 – – 20 – 1 – 1 – 1 2 – – – – – – – 1 – 1 – – – 31 6 1 – – – – – – – – 5 3 – – – – – – – – – – – – – – – 2 – 1 – 18 1 1 – – – – – – – – 21 – – – – 1 – 3 – – – – – – – – – – – – 1 28 Excavation spit 7 8 9 1 1 1 1 – – – – – – 5 1 – 1 1 – – – – – 1 1 – – – – – – – – – 14 207 2 2 – – – – – – – – 5 – – – – – – – – – – – – – – – – – – – – 9 4 2 – – – – – – – – 8 – – – – – – – – – – – – – – – – – – – – 14 10 11 12 13 14 15 Total 3 – – 1 – – – – – – – – – – – – – – – – – – – – – – – – – – – 4 4 – – – – – – 1 – – 4 – – – – – – – – – – – – – – – – 1 – – – 10 6 – – – – 3 – – – – 3 – – – – – – – – – – – – – – – – – – – – 12 7 2 – – – – – – – – 3 – – – – – – – – – – – – – – – 1 – – – – 13 15 3 – 2 – – – 1 1 – 7 – – – – – – – – – – – – – – – – – – – – 29 3 – – – – – – – – – 6 – – – – – – – – – 1 – – – – – – 1 – – – 11 71 14 1 5 1 3 1 3 1 10 124 7 6 1 3 2 1 8 1 1 4 1 4 1 1 1 1 5 1 1 1 285 Chapter . Pre-Colonial Potting III: Decorating Table 9.8 (continued). Motifs represented at Nunguri, Test Pit 1 by excavation spit. Excavation spit 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Total M37 – 1 M38 – 1 2 – – – – – 1 – – – – – 1 – – – – – 4 – 1 – – – – – – M41 – – – – – – 3 – – – – – 2 – – – M42 3 8 7 4 – 2 3 6 8 1 1 2 5 3 – 1 52 M43 – – – – M44 – – – – – 2 4 – – – 1 – 4 – – 11 1 – – – – – – – – – – M45 – – – – 1 – – – – 1 – – – – – – M46 – 4 5 1 2 – 5 2 3 2 3 – 1 2 2 1 32 M47 – – – 2 – – – 1 – – 1 – – 1 – 5 M48 – 1 M49 – 1 – – – – – – – – – – – – – 1 1 1 – 1 1 – 1 – 2 2 1 – – 11 M51 – – M52 – – 2 – – – – – – 1 – 1 – – – 4 – – – – 1 – – – – – – – – 1 M53 – M54 – 1 3 1 – – – – – – 2 – 2 1 1 11 1 – – – – – – – – – – – – – M55 1 – 1 – – – – – 1 – – – – – – – 2 M56 3 3 – – – – – – – – – – – – – 6 M57 – – – – – 1 – – – – – – – – – 1 M58 – – – – – – – 1 – – – – – – – 1 M59 – – – 1 – – – – – – – – – – – 1 M60 – – – – – – – – – – – – – 1 – 1 M61 – – – – – – 1 – – – – – – – – 1 M63 – 2 1 1 – – – 3 – 1 1 – – – – 9 M64 2 1 1 – 1 – 1 – 2 2 – – 1 – 1 12 M65 – – 1 – – – – – – – – – – – – 1 M66 – – – – – 1 1 – – – – – – 1 – 3 M67 1 – – – – – – – – – – – – – – 1 M68 – – – – – – – – – 2 – – – – – 2 M69 – 1 – – – – – – – – – – – – – 1 M71 – – 1 – – – – – – 1 – 1 – 1 – 4 M72 – – 1 – – – – – – – – – – – – 1 M73 – – – – – – – – – – 1 – – – – 1 M74 – – – – – – 1 – – – – – – – – 1 M76 – – 1 – – – – – – – 1 – – – – 2 M77 – 1 – – – – – – – – – – – – – 1 M78 – 1 – – – 1 – – – – – 1 – – – 3 M79 – – – – – – – 1 – – – – – 2 – 3 M80 – – – – – – – – – – 2 – – – – 2 M81 – – – – – – – – – – 1 – – – – 1 M82 – – – – – – 1 – – – – – – – – 1 M83 – – – – – – – – – – – 1 – – – 1 M84 – – – – – – – – 1 – – – – – – 1 M85 – 1 – – – – – – – – – – – – – 1 M86 – – 1 – – – – – – – – – – – – 1 Incision M87 – – – – – – – – – 1 – – – – – 1 Subtotal 9 29 27 13 2 14 19 18 9 13 14 14 13 9 4 207 208       · .  Table 9.8 (continued). Motifs represented at Nunguri, Test Pit 1 by excavation spit. Excavation spit 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Total M88 1 M89 – – – – – – – – – – – – – – – 1 – 1 – – – – – – – – – – – – M90 1 – – – – – – – – – – – – – 1 – 1 M91 – – 1 – – – – – – – – – – – – 1 Subtotal 1 – 2 – – – – – – – – – – 1 – 4 Impression Paddle impression M95 1 – – – – – – – – – – – – – – 1 M96 – 2 – – – – – – – – – – – – – 2 M97 – – – – – 1 – – – – – – – – – 1 M99 1 – – – – – – – – – – – – – – 1 Subtotal 2 2 – – – 1 – – – – – – – – – 5 18 66 71 44 20 43 33 27 23 17 24 26 26 39 15 501 Total bins, is generally more common. This suggests that there was a gradual shift away from using linear and nubble appliqué combinations towards motifs exclusively using linear appliqué. Various incised motifs are also common at Nunguri, particularly M42 with linear bands flanked by continuous gash incised bands, and M46 with linear bands flanked by continuous double-gash incised bands. Interestingly, the structural procedures of applying long bands on a pot’s external surface, which are then flanked by smaller configurations, is a common characteristic shared by the application methods. Nunguri slip is usually dark reddish brown (2.5YR 3/6, 5YR 3/6) or reddish brown (2.5YR 4/6) but can also be very dark brown (7.5YR 2/3) and very dark reddish brown (5YR 2/3) owing to varying firing conditions. Slip is often associated with fine striations on the external surface, the largest of which resemble deliberate incisions. These are usually only visible under low-level magnification and more likely resultant from the tool used to apply the slip or contact with the bush material used in firing. Slip is present on the external surface of 98% (n = 2874) of decorated body sherds and 93% (n = 717) of rims. However, this slip is only present internally on 8% (n = 218) of decorated sherds and 75% (n = 573) of rims, suggesting that it was applied around the entire exterior surface but only around the inside of the rim and sometimes the inside the top of the body/shoulder. Those sherds without slip appear to be the result of post-depositional processes, but this remains to be demonstrated. To delineate technical variation within the broader groupings based on form, mineralogy, and clay chemistry, these decorative attributes will now be cross-compared with technical class, techno-fabric/mineralogical group, and techno-compositional group. This procedure establishes techno-style groupings, which will allow the interplay between production groups and communities of practice to be addressed. Firstly, Figure 9.19 shows the association between decorative method and technical class. Appliqué 40 35 Number of cases 30 25 20 15 10 5 0 1 2 3 4 5 Class Appliqué Impression Incision Paddle impression Figure 9.19. Cases of application method on body/shoulder by technical class at Nunguri. 209 Chapter . Pre-Colonial Potting III: Decorating and incision were used to decorate Classes 1–5, paddle impression was used to decorate both Class 3 and 4, and impression was used to decorate Classes 1, 3, and 4. Although the figure below suggests that there are differences between how the vessel classes were decorated, a Pearson’s Chi Squared Test (with Monte-Carlo simulation owing to low cell counts) shows that there is no significant association between technical class and decorative method (X2 = 14.886, df = 12, P = 0.23). As only one clear clay chemical cluster was delineated among the Nunguri rims, it is not possible to compare and test different techno-compositional groupings against decorative elements. It is possible, however, to code the PCA outputs of the clay chemical data based on decoration, in an attempt to tease apart the single techno-compositional group. Despite this, Figures 9.21–9.23 show that decorative method, the presence of lip decoration, and the presence of inner rim notching do not clearly discriminate subgroupings based on clay chemical data. There is, however, a statistically significant association between the presence of lip decoration and technical class (X2 = 13.017, df = 4, P = 0.01). Cross-tabulation shows that Class 2 vessels in particular are more likely than expected to display lip decorations (adj. residual = 3.5). Moreover, albeit tentatively, a significant association probably exists between technical class and the presence of slip (X2 = 21.199, df = 8, P = 0.04), as Class 1 vessels display slip substantially less often than expected (adj. residual = –3.7) and Class 3 (adj. residual = 3) and Class 4 (adj. residual = 2.2) display slip substantially more often than expected (note that this could be because Class 3 and 4 rims are more recent and so less affected by observed wave action; see Chapter 6). Lastly, there is also a significant association between vessel class and inner rim notching (X2 = 154.419, df = 4, P < 0.01), as Class 1 (adj. residual = 5.9) and Class 2 (adj. residual = 5.7) display the attribute substantially more than expected, and Class 5 (adj. residual = -2.4) and particularly Class 4 (adj. residual = -11.4) display the attribute less often. There is no correlation between techno-fabric group and decoration method (X2 = 4.062, df = 9, P = 0.79) (Fig. 9.20). Nor is there a significant association between fabric groups and lip decoration (X2 = 4.646, df = 3, P = 0.20), fabric groups and red slip (X2 = 9.794, df = 6, P = 0.13), or fabric and inner rim notching (X2 = 3.515, df = 3, P = 0.33). To finalise the classification, numerous techno-style groups can be distinguished confidently within the Nunguri assemblage, shown in orange in Table 9.9, while many more, shown in blue, may have existed but were not identified owing to the small sample size of decorated rims within each class and fabric. The 31 techno-style groups are the result of slight variations in the chaîne opératoire during the procurement, forming and decorating phases. These results suggest there was a flexible approach to many minor forming and decorating techniques being produced by multiple production groups, thus forming multiple technical classes. Because the Madang assemblages contain large quantities of locally produced material which all fits within the same broad ‘style group,’ just how these technostyle groups fit into the production groups and community of practice is not straightforward. The implications of this classification will be further examined in the following chapter. Results: Tilu In the Tilu assemblage, a similar set of application methods were used to decorate Madang style ceramics. Throughout Layer 1 (Spits 1–10), appliqué (Fig. 9.24–9.25), incision (Fig. 9.26), and paddle impression (Fig. 9.27) are common on 100 90 Percentage of cases 80 70 60 50 40 30 20 10 0 1. Fe-Mg 2. Light 3. Calc. 4. Grog Fabric Appliqué Impression Incision Paddle impression Figure 9.20. Cases of application method on body/shoulder by techno-fabric at Nunguri. 210       · .  3.0 Principal component 2 (16.4%) 2.0 1.0 0.0 -1.0 -2.0 -3.0 -2.0 0.0 2.0 4.0 Principal component 1 (57.7%) Appliqué Incision Paddle impression Plain Unknown Figure 9.21. PCA showing clay chemical data of Nunguri sherds coded by body/shoulder application method. 3.0 Principal component 2 (16.4%) 2.0 1.0 0.0 -1.0 -2.0 -3.0 -2.0 0.0 2.0 4.0 Principal component 1 (57.7%) Lip punctation Lip impression Absent Figure 9.22. PCA showing clay chemical data of Nunguri sherds coded by lip decoration. 211 Chapter . Pre-Colonial Potting III: Decorating 3.0 Principal component 2 (16.4%) 2.0 1.0 0.0 -1.0 -2.0 -3.0 -2.0 0.0 2.0 4.0 Principal component 1 (57.7%) Present Absent Unknown Figure 9.23. PCA showing clay chemical data of Nunguri sherds coded by inner rim notching. Table 9.9. Techno-style groups within the Madang style at Nunguri. Orange cells indicate observed correlations, blue cells were not observed. Class 1 FeMg (Local clay) Light (Local clay) Calcareous (Local clay) Grog (Local clay) 2 3 4 5 Appliqué Appliqué Appliqué Appliqué Appliqué Incision Incision Incision Incision Incision Impression Impression Impression Impression Impression Paddle imp. Paddle imp. Paddle imp. Paddle imp. Paddle imp. Appliqué Appliqué Appliqué Appliqué Appliqué Incision Incision Incision Incision Incision Impression Impression Impression Impression Impression Paddle imp. Paddle imp. Paddle imp. Paddle imp. Paddle imp. Appliqué Appliqué Appliqué Appliqué Appliqué Incision Incision Incision Incision Incision Impression Impression Impression Impression Impression Paddle imp. Paddle imp. Paddle imp. Paddle imp. Paddle imp. Appliqué Appliqué Appliqué Appliqué Appliqué Incision Incision Incision Incision Incision Impression Impression Impression Impression Impression Paddle imp. Paddle imp. Paddle imp. Paddle imp. Paddle imp. 212       · .  the body, with instances of impression/punctation only occurring on the lip, inner lip, or interior corner point. Sample sizes in Layer 2 (Spits 11–12) restrict interpretations as only five decorated body sherds were recovered, displaying six discrete cases of appliqué decoration (Table 9.10). Within each application method, several sub-methods are observable. Most commonly, appliqué was applied as a line or a nubbin (Fig. 9.28), but, less often, lines of appliqué were subsequently impressed, incised, or scraped. Incision was usually produced as lines or gashes and rarely as grooves. Impressions on the lip are usually punctations which punch deeply into the clay, while those impressions on the interior corner point and inner lip are often shallower impressions. Unlike Nunguri, no impressions were produced on the body/shoulder. Although numerous plain body sherds were recovered (n = 8896), only two rims were complete enough to be assigned as ‘plain.’ The relative proportions of each application method change significantly through the excavation spits (Pearson Chi-squared test, X2 = 92.405, df = 33, P<0.01). However, this is owed to large amounts of variation in Spit 7 (adj. residual = -5.1). Paddle impression is the reverse, being substantially more common than expected in Spit 7 (adj. residual = 5). Unlike at Nunguri, there is no convincing change in application method over time. Also unlike Nunguri, the relative proportion of sub-methods within each application method does not change significantly through the spits for appliqué (X2 = 72.927, df = 55, P = 0.106), incision (X2 = 19.023, df = 16, P = 0.265), or impression (X2 = 1.224, df = 6, P = 1) (see Fig. 9.16). This suggests that the process of decorating with appliqué, incision, impression and paddle impression remained relatively consistent over time. Body thickness of Tilu’s decorated sherds differs significantly by application method (Table 9.11–9.13). A post hoc Kruskal-Wallis test (Table 9.12) shows there is no statistically significant difference between mean thickness of appliqué and incised sherds (X2 = 1.563, df = 1, P = 0.217) but there is between appliqué and paddle impressed (X2 = 32.475, df = 1, P<0.01), and incised and paddle impressed (X2 = 20.141, df = 1, P<0.01). This may relate to the body being further thinned during the decorative paddling stage. The decorative configurations, used in isolation or to form motifs, are limited with a smaller range expressed than in the Nunguri assemblage. There is, however, a substantially larger range of configurations represented in any one spit at Tilu than the limited configurations applied to modern bodi. In the past, nubble appliqué was applied as single (NA1) or double (NA2) lines, and linear appliqué as single straight bands (LA1), diagonal dashes (LA4), or waves (LA9) (Table 9.14). Several paddle decorations are visible but the most common are simple linear marks (PI2). Table 9.10. Decorative application methods represented at Tilu, Unit 1 by excavation spit.* Excavation spit Application method 1 2 3 4 5 6 7 8 9 10 11 12 Total 247 Appliqué Nubble appliqué 63 42 41 30 19 23 15 7 25 7 1 1 Linear appliqué 55 33 54 22 10 12 10 6 16 9 1 1 41 Appliqué variant 5 2 6 6 4 2 – – 2 1 – – 28 Incised linear app. – 3 – 1 – 1 – – – – – – 5 Impressed linear app. – – – – – – 1 – – – – – 1 Appliqué (?) 5 11 10 6 1 7 1 1 7 3 – 2 54 128 91 111 65 34 45 27 14 50 20 2 4 376 Linear incision 11 6 2 8 2 4 4 1 3 – – – 41 Gash incision 5 5 4 1 2 4 3 1 – – – – 25 Groove incision – – 2 1 – – – – – – – – 3 16 11 8 10 4 8 7 2 3 – – – 69 Impression** 10 12 10 9 2 6 10 3 7 1 – – 69 Punctation*** 1 – – – 1 1 2 – 5 2 – – 12 11 12 10 9 3 7 12 3 12 3 – – 81 Subtotal Incision Subtotal Impression Subtotal Paddle impression Paddle impression Total 24 19 19 18 10 22 30 6 27 7 – – 182 179 133 148 102 51 82 76 25 92 30 2 4 924 * Note, the counts referred to in this table refer to all observable cases of the application method, on both body and formal sherds. Thus, some sherds may preserve evidence of >1 application method. ** Note, most cases of impression were produced on the interior corner point. Only two instances occur on the inner lip and one instance occurs on the lip. No instances occur on the body. *** Note, all cases of punctation occur on the lip. 213 Chapter . Pre-Colonial Potting III: Decorating T-220 Spit 2 M1 T-774 Spit 9 M1 T-29 Spit 1 M1 T-387 Spit 3 M1 T-43 Spit 1 M2 T-193 Spit 2 M4 T-510 Spit 5 M4 0 T-214 Spit 2 M2 T-39 Spit 1 M8 5 cm Figure 9.24. Examples of appliqué decoration on Madang style body sherds at Tilu. 214       · .  T-44 Spit 1 M11 T-213 Spit 2 M11 T-125 Spit 1 M11 T-350 Spit 3 M19 T-139 Spit 1 M26 T-34 Spit 1 M32 T-229 Spit 2 M32 T-31 Spit 1 M32 0 T-309 Spit 3 M33 5 cm Figure 9.25. Examples of appliqué decoration on Madang style body sherds at Tilu (cont.). 215 Chapter . Pre-Colonial Potting III: Decorating T-253 Spit 2 M47 T-211 Spit 2 M50 T-150 Spit 3 M53 T-614 Spit 6 M58 T-48 Spit 1 M57 T-472 Spit 4 M64 T-243 Spit 2 M59 T-46 Spit 1 M85 0 5 cm Figure 9.26. Examples of incised decoration on Madang style body sherds at Tilu. 216       · .  T-696 Spit 7 M92 T-583 Spit 6 M92 T-731 Spit 8 M93 T-572 Spit 6 M94 T-860 Spit 10 M92 T-469 Spit 4 M98 0 5 cm Figure 9.27. Examples of paddle impressed decoration on Madang style body sherds at Tilu. Table 9.11. Average body thickness (mm) by application method at Tilu, Unit 1. Table 9.12. Kruskal-Wallis test of body thickness (mm) by application method at Tilu, Unit 1. Appliqué Incision Paddle impression Appliqué Incision Mean 4.82 5.13 4.14 No. of cases 570 64 173 SD 1.50 1.61 1.47 Mean rank 425.67 464.00 310.40 CV 31% 31% 36% X² = 37.063, df = 2, η² = 4.6%, P<0.01 Table 9.13. Mood’s test of body thickness (mm) by application method at Tilu, Unit 1. Appliqué Incision Paddle impression No. of cases 570 64 173 Thickness > median 312 37 54 Thickness ≤ median 258 27 119 Median = 4.43, X² = 31.099, df = 2, P< 0.01 217 Paddle impression Chapter . Pre-Colonial Potting III: Decorating 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 Spit Appliqué Incision Paddle impression 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 Spit Incised linear appliqué Impressed linear appliqué Nubble appliqué Linear appliqué Appliqué variant 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 Linear incision 5 6 7 Spit Groove incision 8 9 10 11 12 Gash incision Figure 9.28. Percentage of application methods represented on the external body/shoulder of sherds at Tilu, Unit 1 by excavation spit. 218       · .  Chevrons (PI3) and ‘toothed’ linear paddle marks (PI4) are also common (Table 9.16). Typical linear incisions are the single linear band (LI1) and chevrons (LI15), and common gash incisions include a variety of precise repetitive arrangements (GI1-5). Groove incisions are uncommon (Table 9.15). tion between excavation spit and presence of lip decoration (X2 = 38.282, df = 9, P<0.01) and this is particularly the case in Spit 9 (adj. residual = 4.8) and 10 (adj. residual = 3.5), in which counts of lip decoration are substantially more than expected. The two examples from Spit 10 are decorated with Pu1; small, deep, circular punctations produced with a needle-like point, dotted regularly around the cirMost of these decorations were applied to the external cumference of the lip. The decorated lips from Spit 1–9 are body of the pot (see Table 9.14–9.16); however, it is not decorated with Pu3, a much broader punctation using the possible to distinguish if the decorations were restricted side of a sharp tool, and Im3, probably produced using to the shoulder as per modern Madang pots owing to the the same or similar tool as Pu3 but leaving only a shallow lack of carinations. As above, appliqué, paddle impression, impression around the lip. and incision were all exclusively used on the shoulder/ body, while punctation and impression were all exclu- Impressions around the interior corner point are comsively used to decorate the lip and interior corner point in mon on Tilu rims (Table 9.16). As at Nunguri, these are a simple and probably quick manner. No instances exist all classified here as Im1. Inner rim notching was present of punctation around the neck as per modern decoration. on most Tilu rims (78%, n = 69) and notably absent on far fewer (22%, n = 20; note some instances could not be obLip decoration at Tilu is more common than at Nunguri. served where the interior corner point was not preserved). Although configurations on the lip are only represented on This is in contrast to Nunguri, where only half of all rims 10% of Tilu vessels, it is clear that this technical element displayed inner rim notching, suggesting it became less was more common during Tilu’s early occupation, in the common over time. However, unlike lip decoration, there deeper excavation spits. Figure 9.29 illustrates the percent- is no statistically significant association between inner age of rims with punctate or impressed lips relative to the rim notching and excavation spit at Tilu (X2 = 17.921, df = 9, MNV in each spit. There is a statistically significant associa- P = 0.31). Table 9.14. Appliqué configurations represented at Tilu, Unit 1 by location. Body Lip Outside lip Inside lip Sub–rim groove Sub–rim groove ridge Interior corner point NA1 137 – – – – – – NA2 51 – – – – – – NA3 – – – – – – – NA4 1 – – – – – – NA5 2 – – – – – – NA6 3 – – – – – – LA1 133 – – – – – – LA2 – – – – – – – LA3 8 – – – – – – LA4 26 – – – – – – LA5 2 – – – – – – LA6 8 – – – – – – LA7 1 – – – – – – LA8 – – – – – – – LA9 18 – – – – – – LA10 – – – – – – – LA11 – – – – – – – LA12 – – – – – – – LA13 – – – – – – – LA14 – – – – – – – AV1 24 – – – – – – AV2 2 – – – – – – InLA1 2 – – – – – – InLA2 – – – – – – – ImLA1 1 – – – – – – ImLA2 – – – – – – – 219 Chapter . Pre-Colonial Potting III: Decorating Table 9.15. Incised configurations represented at Tilu, Unit 1 by location. Body Lip Outside lip Inside lip Sub–rim groove Sub–rim groove ridge Interior corner point LI1 10 – – – – – – LI2 7 – – – – – – LI3 1 – – – – – – LI4 – – – – – – – LI5 2 – – – – – – LI6 – – – – – – – LI7 – – – – – – – LI8 – – – – – – – LI9 – – – – – – – LI10 – – – – – – – LI11 – – – – – – – LI12 1 – – – – – – LI13 2 – – – – – – LI14 – – – – – – – LI15 10 – – – – – – LI16 2 – – – – – – LI17 – – – – – – – GI1 4 – – – – – – GI2 8 – – – – – – GI3 5 – – – – – – GI4 3 – – – – – – GI5 5 – – – – – – GI6 – – – – – – – GI7 – – – – – – – GI8 – – – – – – – GI9 2 – – – – – – GI10 – – – – – – – GI11 – – – – – – – GI12 – – – – – – – GI13 – – – – – – – GI14 – – – – – – – GI15 – – – – – – – GI16 – – – – – – – GI17 – – – – – – – GI18 – – – – – – – GI19 – – – – – – – GI20 – – – – – – – GI21 – – – – – – – GI22 – – – – – – – GI23 – – – – – – – GI24 – – – – – – – GI25 – – – – – – – GI26 – – – – – – – GrI1 2 – – – – – – GrI2 – – – – – – – GrI3 – – – – – – – GrI4 – – – – – – – 220       · .  Table 9.16. Impressed and paddle impressed configurations represented at Tilu, Unit 1 by location. Body Lip Outside lip Inside lip Sub–rim groove Sub–rim groove ridge Interior corner point Im1 – – – 2 – – 67 Im2 – – – – – – – Im3 – 1 – – – – – Im4 – – – – – – – Pu1 – 2 – – – – – Pu2 – – – – – – – Pu3 – 9 – – – – – PI1 7 – – – – – – PI2 34 – – – – – – PI3 14 – – – – – – PI4 17 – – – – – – PI5 7 – – – – – – PI6 2 – – – – – – PI7 1 – – – – – – PI8 7 – – – – – – PI9 1 – – – – – – PI10 – – – – – – – 1 2 60% 50% 40% 30% 20% 10% 0% 3 4 5 6 7 8 9 10 Spit Figure 9.29. Percentage of Tilu rims with decorated lips by excavation spit. At Tilu, 21 different motifs were recorded; again far fewer than in the Nunguri assemblage. There is, however, still a substantially larger range of motifs represented at Tilu than the single design seen on modern bodi. Appliqué motifs comprise 84% of all observed cases, and incision and paddle impression, 8% each. This is notably different from Nunguri, where appliqué motifs account for only 57% and incised motifs account for 41%. No examples of impressed/ punctate motifs were observed, nor were any combinations of two or more different decorative methods used to produce motifs. The M1 appliqué motif, a single linear band flanked by nubble bands, constitutes almost half of the total (n = 67, 48.55%) while all other motifs are com- paratively minor (Table 9.17). This picture contrasts with Nunguri, where M11 is generally more common than M1, except in lower deposits. This pattern supports the idea that the M1 motif was more common earlier in time, and that the Tilu sequence overlaps with the Nunguri sequence, but also stretches slightly further back in time. Tilu slip cannot be visually differentiated from Nunguri slip. It is usually dark reddish brown (2.5YR 3/6, 2.5YR 3/4, 5YR 3/6) or reddish brown (2.5YR 4/6) but can also be very dark brown (7.5YR 2/3) owing to varying firing conditions. Slip is externally present on 95% (n = 628) of decorated body sherds and 90% (n = 103) of rims at Tilu. Similar to 221 Chapter . Pre-Colonial Potting III: Decorating Table 9.17. Motifs represented at Tilu, Unit 1 by excavation spit. Excavation spit Application method 1 2 3 4 5 6 7 8 9 10 11 12 Total Appliqué M1 16 13 17 6 3 2 4 – 4 1 1 – 67 M2 2 1 3 2 1 – – – – 1 – – 10 M4 – 2 1 1 2 – – 2 1 1 – – 10 M8 1 – – – – 1 – – – – – – 2 18 M11 3 3 6 2 1 1 – 1 1 – – – M19 – – 1 – – – – – – – – – 1 M20 – – – – – – – 1 – – – – 1 M26 1 – – – – – – – – – – – 1 M32 3 1 1 – – – – – – – – – 5 M33 – – 1 – – – – – – – – – 1 26 20 30 11 7 4 4 4 6 3 1 – 116 – 1 – – – – – – – – – – 1 M50 – 1 – – – – – – – – – – 1 M53 1 – – – – 1 – – – – – – 2 M57 2 – – – – – – – – – – – 2 M58 – – – – – 1 – – – – – – 1 M59 – 1 – – – – – – – – – – 1 M64 – – – 1 – – – – – – – – 1 M75 – – – – – – 1 – – – – – 1 Subtotal Incision M47 M85 1 – – – – – – – – – – – 1 4 3 – 1 – 2 1 – – – – – 11 M92 – – – – – 1 1 – – 1 – – 3 M93 – – – – – – – 1 – – – – 1 M94 – – – – – 1 – – – – – – 1 M98 – – 1 1 1 1 2 – – – – – 6 – – 1 1 1 3 3 1 – 1 – – 11 30 23 31 13 8 9 8 5 6 4 1 – 138 Subtotal Paddle impression Subtotal Total the Nunguri sherds, slip is only internally present on 7% P = 0.161), or between the presence of slip and technical (n = 48) of decorated body sherds and 79% (n = 90) of rims, class (X2 = 5.868, df = 4, P = 0.268). There is also no signifiindicating that it was applied around the exterior but not cant association between techno-fabric groups and the far into the interior. decorative method used (X2 = 4.679, df = 9, P = 0.786). This is illustrated in Figure 9.31, which shows appliqué as the There appears to be no statistically significant association dominant decorative method for each fabric, with paddle between technical class and application method (Pearson’s impression being of secondary importance, and incision Chi-Squared test X2 = 5.559, df = 8, P = 0.679). Figure 9.30, being uncommon. which has excluded cases of lip decoration and inner rim notching, shows that Class 1 rims display cases of appliqué, As above, it is difficult to compare the techno-composiincision, and paddle impression, while Class 2 rims display tional groupings with decorative methods, but the PCA all but incision. Classes 3, 4, and 5 all display instances data are here coded by decorative attributes to tease apart of appliqué, the same kind as that applied to Class 1 and the single techno-compositional group. Despite this, Fig2 vessels. Small sample sizes make further interpretation ures 9.32–9.34 show that decorative method, the presence difficult. of lip decoration, and the presence of inner rim notching do not clearly discriminate subgroupings based on clay Additionally, unlike Nunguri, there is no statistically sig- chemical data. This pattern is consistent with Nunguri and nificant association between lip decoration and techni- suggests that those production groups that were using the cal class (X2 = 2.694, df = 4, P = 0.383), the presence of in- Bilbil-area clay sources were using various decorative techner rim notching and technical class (X2 = 6.647, df = 4, niques interchangeably. 222       · .  30 Number of cases 25 20 15 10 5 0 1 2 3 4 5 Class Appliqué Incision Paddle impression Figure 9.30. Cases of application method on the body/shoulder by technical class at Tilu, Unit 1. 100 90 Percentage of cases 80 70 60 50 40 30 20 10 0 1. Fe-Mg 2. Light 3. Calc. 4. Grog Fabric Appliqué Incision Paddle impression Figure 9.31. Percentage cases of application method on the body/shoulder used by techno-fabric at Tilu, Unit 1. Table 9.18 summarises the Tilu techno-style groupings, representative of variations in the chaîne opératoire during the procurement, forming and decorating phases. Ten variations have been confidently determined but many more are likely to have been produced as is the case at Nunguri with a larger sample size. These techno-style groups arose within different production groups in the local Madang area. This local context makes interpreting the organisation of production groups within the broader community of practice much more difficult than at a regional level. The specific implications of these results in the local Madang setting will therefore be elaborated on in the following chapter. Results: surface survey The decorated sherds collected during surface survey are limited (Table 9.19), and the following is an overview rather than an in-depth analysis. The collection includes four decorated rims and three decorated body sherds from Kranket, one decorated rim and one decorated body from Siar, and five decorated rims and one decorated body from Yabob. Decorative configurations of individual decorative elements are also limited, but within the range of Nunguri and Tilu decorations (Table 9.20). Most of the decorative 223 Chapter . Pre-Colonial Potting III: Decorating 4.0 Principal component 2 (27.6%) 3.0 2.0 1.0 0.0 -1.0 -2.0 -3.0 -2.0 0.0 2.0 4.0 Principal component 1 (52.3%) Appliqué Incision Paddle impression Plain Unknown Figure 9.32. PCA showing clay chemical data of Tilu sherds coded by body/shoulder application method. 4.0 Principal component 2 (27.6%) 3.0 2.0 1.0 0.0 -1.0 -2.0 -3.0 -2.0 0.0 2.0 4.0 Principal component 1 (52.3%) Lip punctation Lip impression Absent Figure 9.33. PCA showing clay chemical data of Tilu sherds coded by lip decoration. 224       · .  4.0 Principal component 2 (27.6%) 3.0 2.0 1.0 0.0 -1.0 -2.0 -3.0 -2.0 0.0 2.0 4.0 Principal component 1 (52.3%) Present Absent Unknown Figure 9.34. PCA showing clay chemical data of Tilu sherds coded by inner rim notching. Table 9.18. Techno-style groups within the Madang style at Tilu. Orange cells indicate observed correlations, blue cells were not observed. Class 1 Appliqué FeMg (Local clay) Light (Local clay) Calcareous (Local clay) Grog (Local clay) 2 Appliqué 3 Appliqué 4 Appliqué 5 Appliqué Incision Incision Incision Incision Incision Impression Impression Impression Impression Impression Paddle imp. Paddle imp. Paddle imp. Paddle imp. Paddle imp. Appliqué Appliqué Appliqué Appliqué Appliqué Incision Incision Incision Incision Incision Impression Impression Impression Impression Impression Paddle imp. Paddle imp. Paddle imp. Paddle imp. Paddle imp. Appliqué Appliqué Appliqué Appliqué Appliqué Incision Incision Incision Incision Incision Impression Impression Impression Impression Impression Paddle imp. Paddle imp. Paddle imp. Paddle imp. Paddle imp. Appliqué Appliqué Appliqué Appliqué Appliqué Incision Incision Incision Incision Incision Impression Impression Impression Impression Impression Paddle imp. Paddle imp. Paddle imp. Paddle imp. Paddle imp. 225 Chapter . Pre-Colonial Potting III: Decorating Table 9.19. Decorative application methods represented from surface collections.* Application method Kranket Island Siar Island Yabob Island Total Nubble appliqué – 1 – 1 Linear appliqué 7 – 2 9 Appliqué variant – – 2 2 Incised linear app. – – – – Impressed linear app. – – – – Appliqué Appliqué (?) – – – – Subtotal 7 1 4 12 Incision Linear incision 2 – 1 3 Gash incision 1 1 – 2 Groove incision – – – – Subtotal 3 1 1 5 2 1 2 5 Impression Impression Punctation – – 1 1 Subtotal 2 1 3 6 – – – – 12 3 8 23 Paddle impression Paddle impression Total * Note, the counts in this table refer to all observable cases of the application method, on both body and formal sherds. Thus, some sherds may preserve evidence of >1 application method. Table 9.20. Decorative configurations represented in surface collections by location. Body Kranket Island LA1 1 LA4 2 LA7 2 LA10 1 LI2 1 LI12 1 GI11 1 Im1 – Siar Island NA1 1 GI25 1 Im1 – Yabob Island LA4 2 AV1 1 LI11 1 Pu3 – Im1 – Lip Outside lip Inside lip Neck Sub-rim groove Sub-rim groove ridge Interior corner point – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – 2 – – – – – – – – – – – – – – – – – – – – 1 – – – – – – 1 – – – – – – – – – – – 1 – – – – – – – – – – – – – – – 2 226       · .  configurations are limited to the body, but one instance of Results: exotics appliqué variant (AV1) is present on the outside lip of a Class 4 incurving rim. Interestingly, this surface collection Of the 12 ‘exotic’ sherds excavated and collected around contains the only instance of a sherd with neck decoration, Madang, 11 retain decorative elements. At Tilu Unit 1, sevwhich is similar to modern Bel pots in being punctation en body sherds with decoration atypical of the Madang produced around the circumference of the neck. This is style are present (Fig. 9.37). This includes: T-230, a thick present on Y-6, a large Class 1 sherd, which suggests either (8.82 mm), deeply grooved and impressed sherd from that the configuration was in use prior to the cessation of Spit 2; T-520, a thinner (6.29 mm), deeply incised sherd Yabob Island potting or that the pot was traded in later. from Spit 5, and five thick, deeply incised and punctate Only two motifs are present in the surface collection (M11 impressed sherds from Spit 6 which likely derive from the and M70), both from Kranket Island. Inner rim notching same vessel. One of the exotic rims from Tilu Unit 1 (Ton the interior corner point is present on five sherds (31%). 188) is also decorated with punctation, while the ring-base (T-913) from Tilu Shovel Pit 1 is decorated with impresThe sample sizes of decorated surface collections are too sions around the circumference of the base and on the small to effectively delineate groupings when compared body of the vessel. The single exotic sherd from Nunguri to vessel class and fabric. However, as above, the PCA plots (N-3712) is a rim decorated with bands of gash incision on showing techno-compositional data can be coded by dif- the body and punctation along the lip. The possible exotic ferent decorative elements. Below, the techno-composi- from Yabob (Y-5) is decorated with punctations along the tional grouping is coded first by decorative method and outside lip (see Chapter 7). The decorations produced on then by presence of inner rim notching. These figures these ‘exotic’ sherds are similar in form to Madang style (9.35–9.36) show similar results to Nunguri and Tilu with decorations: straight freehand incisions bordered by dino clear association between clay and decoration. agonals, lines of gash incisions, rows of punctations. However, the gestures and tools used, along with the placement on the vessels are clearly different. 2.0 Principal component 2 (28.4%) 1.0 0.0 -1.0 -2.0 -3.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 Principal component 1 (37.4%) Appliqué Incision Unknown Figure 9.35. PCA showing clay chemical data of survey sherds coded by body/shoulder application method. 227 Chapter . Pre-Colonial Potting III: Decorating 2.0 Principal component 2 (28.4%) 1.0 0.0 -1.0 -2.0 -3.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 Principal component 1 (37.4%) Present Absent Figure 9.36. PCA showing clay chemical data of survey sherds coded by inner rim notching. T-230 Spit 2 T-520 Spit 5 T-581 Spit 6 T-578 Spit 6 T-579 Spit 6 T-604 Spit 6 0 5 cm Figure 9.37. Exotic decorated body sherds from Tilu, Unit 1. 228 T-580 Spit 6       · .  Summary The results of the decorative analysis examining minor technical elements suggests that the range of Madang style Technical Classes 1–5 were decorated with various decorations, as opposed to particular classes or fabrics being decorated with specific techniques, which might indicate specific production groups making specific technical classes but using similar raw materials. Rather, the results suggest that there is no reason to assign a single chaîne opératoire to different production groups. This supports the results of the fabric analysis in Chapter 8. More likely, around Madang in the past, like today, one or more production groups were producing a wide range of vessel classes. However, unlike today, these vessels were decorated with a range of different decorative forms using a range of decorative techniques. With regard to using technical elements as useful chronostratigraphic indicators, the presence of lip decoration (impression and particularly punctation) is related to earlier iterations of the local Madang style. It is expected that this technical element will be present on many of the ceramics from pre-650 BP deposits further east. Paddle impression is also more likely to be associated with earlier pots, being much more common at Tilu, while manual impression using a point or blunt end on the body/shoulder is more likely to be late. Punctation applied around the neck and small gash incisions arranged as per modern Bel designs are not present on any of the pre-colonial pots excavated from Nunguri and Tilu. However, an instance of neck punctation is present from a surface find at Yabob and on the you-bodi collected by Miklouho-Maclay (see Chapter 3). This chapter has completed the ceramic technological analysis undertaken to create a hierarchical classification system to delineate techno-style groups. These technostyle groups represent variations to the chaîne opératoire within various production groups, which may be able to be discriminated based on techno-fabric/mineralogical groupings. Various production groups seem to have been producing a series of technical classes and continued interaction, imitation, and knowledge exchange between these groups constitute a broader community of practice: a set of social and technological processes which left behind broadly groupable ceramics: the Madang style. The next chapter is a discussion of these results and will explore production and exchange in the local Madang context, and the nature of changing technological processes leading up to ethnographic times. 229 Chapter 10. Materialising Ancestral Madang Every sufficiently advanced technology is indistinguishable from magic. — Arthur C. Clarke (1973)1 Procurement To materialise Madang’s ancestral potters, this chapter ties together the analytical lines running throughout the book. By doing so the chapter addresses one of the major research objectives: to examine the changing processes of production and exchange leading up to the ethnographically observed Madang exchange networks. This is done in two progressive stages. First, a discussion of the precolonial pottery making process is necessary to describe the nature of past production groups and communities of practice around Madang. This will integrate procurement, forming, and decorating processes into a holistic examination of the chaîne opératoire. These observations, based on the ceramic classification presented in Chapters 7–9, are tied to embodied knowledge as a process by which materials, people, and techniques change and interact over time. Second, this chapter will examine aspects of pre-colonial exchange around Madang and the northeast coast. This process begins with meso-networks on the local level, integrating information from the ceramic analyses, and later expands to discuss macro-networks operating at the regional level (see Chapter 1), through comparisons with pottery assemblages along the Madang coast, in Sepik, the Bismarck Archipelago, and the New Guinea Highlands. This approach will further explore embodiment, and the sharing of knowledge across space and time by building the current study into a regional network of object exchange. The investigation of the procurement phase aimed to establish the timing of local production and subsequent adjustments to technological organisation in terms of changing resource procurement. The results presented in Chapter 8 indicate that the procurement of raw materials for potting in the pre-colonial past probably followed a similar set of processes to today. Clay, sand temper, slip, firing fuel, and water were all required to make a pot. Women would have visited the mainland to collect clays, deriving particularly from the Holocene alluvium of the Astrolabe Land System, which characterises the area around modern Bilbil village. There, organic topsoil overlies thick and easily accessible clay deposits, formed by the colluvial aprons of the Gum and Gogol rivers, which flow out to the Bismarck Sea to the north and south of Bilbil, respectively. These clay sources may have been located under and beside garden plots or near streams where the deposits are conspicuously exposed. The specific techniques behind selection, extraction, and transport are all unknown, but it is plausible that wooden digging sticks were used to extract the material and string bags were employed in a similar manner to today, to carry balls of clay back to the households. The geochemical results demonstrate that the clays used to produce Madang style pots in the past were more similar to modern Bilbil potting clays than to the modern Yabob clays. Based on the PCA and HCA results, one sherd (N2713 from Spit 11) possibly clusters closer to modern Yabob clay sources than the Bilbil sources. Another Nunguri sherd (N-188 from Spit 2) and two Kranket Island sherds Vai: aspects of pre-colonial production (K-2 and K-7) group away from both the Bilbil and Yabob This section will explore the nature of pottery production sources, and might represent a third procurement zone, around Madang in the pre-colonial past. It will first sum- such as Mindiri to the south (Fig. 10.1). The ethnographic marise the various chaînes opératoires (or vai in the Bilbil clay sources used at Mindiri could not be obtained in the language) observed in the ceramic analyses, and how these course of this study but would contribute greatly to indiffer across time and space. Using this information, pro- terpretations about procurement and exchange along the duction groups working within broader communities of northeast coast. Importantly, even the oldest rims in the practice will be systematically distinguished. sample, from Tilu Unit 1, Spit 10, cluster with the modern Bilbil clay sources and show that the very earliest Madang style pottery found in the area so far, dating to 550–650 1 Clarke’s Third Law, in: ‘Profiles of the future: an inquiry into cal. BP, was probably locally produced and not imported. the limits of the possible.’ 230       · .  Malmal Siar Island Kranket Island ? Yabob Island Yabob Island N-2713 Bilbil Island ? N-188 K-2 K-7 Mindiri (?) Figure 10.1. Clay procurement zones around Madang including the major procurement hub around Bilbil and possible minor sources at Yabob and Mindiri. 231 Chapter . Materialising Ancestral Madang The sherds marked as ‘exotic,’ based on the initial stage of intuitive sorting, overlap with Madang style ceramics in their clay chemical composition but differ in their tempers (see below). As mentioned in Chapter 8, this is perhaps because these ‘exotic’ pots derive from production groups using clays from the same Astrolabe Land System, near Bilbil. This is possible given distinct coil-made pots are today produced from similar clays in the Gogol River Valley on the same land system. An alternative explanation is that the data quality obtained using the SEM-EDS was not sufficient to separate minor variations in local or sub-regional clay sources.2 An alternative approach such as LA-ICP-MS might provide a useful and complementary dataset to discriminate sources based on trace elements. A variety of sand types were used to temper the extracted clay, including black beach sand, white (‘light’) beach sand, coral beach sand, and possible grog. Black sands, dominated by clinopyroxenes, feldspars, Fe/Ti oxides, quartz, and volcanolithic fragments seem to have been preferred by the potters and resulted in harder vessels.3 White beach sands comprised of primarily feldspathic and quartz grains, along with minor pyroxenes and other dark minerals, were also used, but much less often. Calcareous coral sand, deriving from uplifted offshore islands, was used with similar infrequence. Lastly, a temper type identified as grog is also present, albeit very rarely. This is similar to that identified in previous petrographic analyses of Madang style pottery (Specht et al. 2006) in containing substantial clay grains or broken sherds. However, because no sherds were exclusively tempered with grog, also containing substantial locally available white and black beach sands, it is not clear whether grog was used deliberately as a tempering agent like Type X ceramics or if ‘grog’ represents accidental inclusions from the reuse of workshop/house floors. The diversity of tempers used to produce Madang style pots may indicate multiple production groups in operation, each using different tempers. For instance, there could have been one group using black beach sand from a river mouth south of Madang Lagoon, one group using white sand from a mainland beach around the Lagoon, one group using calcareous sand from an offshore island, and one using grog. However, because the clays are all chemically similar, it is possible that a single or small number of production groups were using different tempers. This would echo the statements of modern Bel potters, who do not distinguish between sand types. For instance, perhaps the favoured black sand was collected during trips to the mainland to procure clay, and then taken back to offshore islands. When the supplies of black sand ran out, potters could have collected sand from the immediate coral beach rather than paddling back to the mainland to get more black sand. This would especially be the case if only a small batch of pots were being made, during the off-season in August–January. The fabric analysis at Nunguri suggests a tendency to temper with coral beach sand more frequently through time, indicating an increased reliance on offshore materials (e.g. Bilbil Island, Yabob Island, Yomba Island, Urembu Island, or the Madang Lagoon islands). Importantly, all of the tempers identified in Madang style ceramics are consistent with locally available grains and do not resemble those from further east towards the Vitiaz Strait (Gaffney et al. 2019), north towards the Sepik (Summerhayes pers. comm.), or inland, towards the Madang Highlands (Gaffney et al. 2015). Moreover, the local procurement of materials around Madang does not appear to have become more standardised over time as in other areas of New Guinea (Chynoweth 2015; Hogg 2007; Summerhayes 2000; Sutton et al. 2016; Vilgalys & Summerhayes 2016). Rather, from the outset, Madang’s pottery-makers were accustomed to using relatively standardised clay sources (i.e. a number of clay pits within a small geographic range) but were flexible in their use of tempering agents. ‘Exotic’ sherds are more easily distinguished by their tempers (Fig. 10.2). Many lack the clinopyroxene augite (e.g. T-913 ring base), which is characteristic of Madang style pots produced around the Madang archipelago. Many others appear not to be tempered at all and the only inclusions present are probably natural fragments from the clay deposit (e.g. T-188, T-230, T-410, T-604). Although the clay sources may have been similar compared with most Madang style pots, the choice of tempers, and the choice of whether to temper at all, distinguish different production groups. The ethnographic data in Chapter 4 demonstrate that the number of clay sources in use does not seem to correlate to the residential mobility of the potters but is linked to political, social, and technological concerns. Six distinct clay sources were used at the modern Yabob potting village, while at Bilbil only two sources were targeted, despite both groups being equally sedentary and with similar land tenure patterns. Rather, the number of clay sources may represent the standardisation and regularity of procurement processes (see Arnold 2000). Bilbil potters continue to practice pottery making relatively frequently and in a standardised manner, while at the time of this research Yabob potters produced ceramics very infrequently and with far fewer practitioners. The standardised selection of potting clay near Bilbil over c. 600 years does suggest there may have been social regulation of these clay deposits determining who could access the clay to produce pots. It may also reflect which land parcels were accessible to the Bel and their alliances with different groups on the mainland. 2 Note, Lilley’s (1986: 230) sourcing clustered Bilbil clay with Tuvaltae clay. Tuvaltae/Korak is a small potting tradition just north of Sarong on the mainland, immediately west of Karkar. 3 Based on optical observations, although more thorough tech- The continuity of raw material collecting, broadly unbronological analyses would be required to demonstrate this as- ken and maintained over the course of about 600 years attests to the durability of the Bel potters’ procurement sertion. 232 Light fabric (augite absent)       · .  Fe oxide Rock Quartz Ilmenite Labradorite Albite T-913 Anorthite Labradorite Ilmenite Quartz (natural inclusion) Fe oxide Augite T-188 Augite Quartz (natural inclusion) Albite Amphibole Labradorite Non-tempered Labradorite T-230 Epidote Albite Ilmenite Amphibole T-410 Quartz Orthoclase Quartz T-604 Figure 10.2. Different procurement choices represented by ‘exotic’ sherd tempers. 233 Chapter . Materialising Ancestral Madang phase. Raw materials are essential to making things work. Because the Bel pottery production process hinged on making the raw material work and because production was essential to exchange and the procurement of subsistence crops and valuables, tampering with procurement strategies would have been risky and unnecessary. Forming The study of the Madang style forming phase aimed to describe the range of pre-colonial pottery production technologies around Madang and to describe the minor adjustments to these technologies over time. The results presented in Chapter 7 show that the forming stages of pottery manufacture in the pre-colonial past were immediately ancestral to the processes used around Madang today. Despite this, there are several important technical modifications that have taken place within these technological processes, over the past c. 600 years. Pre-colonial Madang style pots were all produced using the paddle and anvil method and were made using five broadly discernible technical sequences (Madang style Classes 1–5). These sequences represent distinct repetitions in chaînes opératoires, which may have been distinguished by the pre-colonial potters themselves (i.e. emic types), or several sequences may have been described as the same ‘type’ based on primary function or size. Four of these sequences (Classes 1, 2, 3, and 5) appear to have been made using a rim preform, given the abrupt reduction in body thickness just below the neck. Class 4 pots are different: incurving with no neck. If they were produced in a similar fashion to modern magob, which are also incurving, then they were made with a rim preform, but an anvil rather than the potter’s thumb would have been used to open the orifice. women potters were most likely also the cooks, they would have used su (recycled bodi employed as cooking stands) in everyday routine. These rims, made to be thicker and flatter, would have formed sturdier supports for cooking when turned upside down compared to high-angled, thinrimmed Class 1 vessels. By extension, the practice of using su may be around 500–600 years old. Additionally, in both the Nunguri and Tilu assemblages, there is a lack of Class 5 rims, which seems to mimic the rarity of you-bodi water pots in modern Madang groups. However, these become more common over time. On the other hand, Class 4 rims become less frequent in the assemblages and are not represented in the suite of ethnographically observed pots today or those in museum collections.4 These Class 4 pots are morphologically more similar to Pila pots produced today along the northeast coast of New Guinea, c. 12 km northwest of Madang, than they are to modern magob pots. Magob are distinct from other Madang pots in the method of producing the rim preform, using a stone to pound open the orifice. This technique is also used to produce pottery at several Pila villages. It is possible then that the magob, and by extension Class 4 pre-colonial pots, were introduced by the ancestors of Pila potters, who married in to Bel potting groups and that Class 4 rims represent an earlier iteration of the modern magob. Vice versa, it is also possible that Pila pots stem from an offshoot of Bel potting traditions around Madang, or that the two traditions have emerged from a common ancestoral tradition. The seriation of pot classes at Nunguri and Tilu are illustrated in Figures 10.3 and 10.4. When overlain, based on known radiocarbon dates from excavated contexts at both sites, the trends are consistent, supporting the radiocarbon Through time, specific chaînes opératoires were engaged determinations that indicate Tilu was occupied slightly more often and indicate gradual trends towards making earlier than Nunguri, and the later deposits at Tilu probthings differently (Fig. 10.3–10.4). Importantly, each form- ably equate to early deposits at Nunguri (Fig. 10.5). Such a ing sequence was enacted with similar clays and a variety chronological sequence is not the end goal of this monoof tempers, suggesting that each production group em- graph, but it provides the chronometric resolution needed ployed multiple forming methods, as they do today. The to then address the tempo with which technological prochanges to the way people were making pots in the past cesses changed and morphed over time. are consistent with the stratigraphic sequences at Nunguri and Tilu and associated radiometric dates (see Chapter 6). The substantial variability of each pot’s chaîne opératoire For instance, Class 3 pots, representing an additional set within these broader technological classes suggests there of techniques for smoothing the interior of the lip to pro- was a high degree of creative freedom in the production duce thick and relatively flat rims, become more frequent phase. For instance, Class 1 pots ranged in orifice diameter over time, beginning around 500–600 years ago (Nunguri from 5–20 cm, with no real suggestion that these can be Spit 12 and Tilu Spit 2) but becoming increasingly frequent easily subdivided into different sub-groups although the within the last 300 years (Nunguri Spits 1–7). Interestingly, size may have guided the eventual use in cooking or storthe early Class 3 rims at Nunguri are identical to the sole age. Class 1 rims also followed a variety of profiles and Class 3 rim found in the upper deposits at Tilu. Today all courses, with a similar variety of lip profiles and extra lip of the observed bodi pots are made using this technique. features. There is little indication that minor technical This demonstrates a gradual inclination towards producing bodi in this manner, as opposed to producing them 4 I examined ethnographic Madang pots from several key muwith thinner rims (Class 1), or rims that are smoothed and seums in Australasia, including the extensive Tuckson collecpaddled on the exterior of the lip (Class 2). tions in the Australian Museum, and two early examples from The innovation of Class 3 pots with flatter rims may be tangled up with consumption and reuse patterns. As the 234 the Macleay Museum, University of Sydney. It is possible that such forms exist in museums in America or Europe, for instance in Berlin or Budapest.       · .  Spit 1 -./00#&# -./00#%# Spit 16 Class 1 Class 2 Class 3 0 Class 4 Class 5 100 % Figure 10.3. Seriation of Madang style technical classes at Nunguri, Test Pit 1. Spit 1 Spit 10 Class 1 Class 3 Class 2 0 Class 4 Class 5 100 % Figure 10.4. Seriation of Madang style technical classes at Tilu, Unit 1 variants tracked a pattern of gradual change. Rather, most technical elements continually fluctuated, suggesting that although the broad technical sequences (Classes 1–5) were socially regulated, experimentation and flexibility within these sequences was relatively ubiquitous. Decorating The analysis of the decorating phase presented in Chapter 9 was designed to clarify the groupings identified through the procurement and forming stages. This aimed to confirm the number of production groups working within various communities of practice. Moreover, because minor technical elements more readily lend themselves to investigating small-scale changes to technology (Mayor 235 Chapter . Materialising Ancestral Madang 200 BP 0-299 cal. BP OZS558 OZS557 210 BP 0-304 cal. BP Modern OZS556 495 BP 508-539 cal. BP 530 BP 515-623 cal. BP 540 BP 520-626 cal. BP OZS552 630 BP 552-664 cal. BP OZS551 595 BP 543-647 cal. BP OZS550 585 BP 540-644 cal. BP OZS549 OZS548 OZS547 OZS546 925 BP 675 BP 580 BP 570 BP 791-915 cal. BP 564-674 cal. BP 538-641 cal. BP 535-637 cal. BP Tilu sequence OZS555 OZS554 Nunguri sequence OZS559 Class 1 Class 2 Class 3 0 Class 4 Class 5 100 % Figure 10.5. Comparison of Nunguri and Tilu seriations with associated absolute dates. Note the approximate overlap between the sequences. 2005), this analysis aimed to distinguish how technological change and knowledge acquisition took place locally over the past c. 600 years. There are indications that there was gradual change to how pots were decorated over the past six centuries. Appliqué on the body/shoulder was more common earlier in time, and less common later in time, although these decoraThe results revealed that there was a series of established tive methods are certainly not diagnostic of early and late conventions which seem to have been cognitively con- periods of pottery manufacture. Rim decoration, on the structed and shaped the nature of decorations. First, pot- other hand, was particularly common early in time and ters almost never combined multiple application methods. does appear to be diagnostic of early Madang style pottery Only four instances of combination motifs were observed (c. 650–550 cal. BP). at Nunguri and no instances were seen at Tilu. Secondly, the potters tended to apply long lines flanked by smaller It is clear that specific decorations did not map directly elements. These lines would demark specific locations on to specific production groups as different decorative the pot and structure the finer decorative methods. Most methods and motifs were used interchangeably across decoration was made on the body/shoulder and this was different raw materials and different technical classes. It usually applied as appliqué or incision. This is different to is likely that decorative techniques, owing to their short minor decorations around the rim, neck, and lip of the and uncomplicated technical syntaxes, were observed and pot. For instance, decoration on the interior corner point learned horizontally across production groups, along with was always impressed. Decoration on the lip was always being taught vertically through generations. This process punctation/impression. These rules were almost always may account for similarities in decoration across the techconformed to, so we can infer that when the potter did not nical groupings. This assertion contrasts with Egloff (1975), follow the rules it was for a specific reason. Perhaps such who suggests that decorative methods are akin to ‘types’ instances represent women from other villages that were and are the most useful indicator of different production newly learning the rules of potting, unaware of specific groups. He also stated that specific rim morphologies conventions; perhaps they were masters of the craft who are correlated with decorative methods (Egloff 1975: 6), held authority and the vou5 to openly experiment with though without statistical demonstration. The contexts complex designs (see Roux 1990; Roux et al. 1995). These of each rim count at Egloff 's sites are also not clear, nor decorative regulations are in stark contrast to vessel form- whether any kind of refitting or MNV calculation was done, ing, within which minor technical elements were fluid and which would affect the frequency of decorated rims where flexible. sample sizes are low. 5 A form of socially regulated intellectual property; see Chapter 3. A Pearson’s chi-squared test with Monte Carlo simulation owing to low cell counts confirms that, based on 236       · .  Egloff ’s data (in Egloff 1975: Table 1), there are significant correlations between rim forms and decorative methods (X2 = 349.998, df = 100, P < 0.01), even when unknowns and undecorated sherds are removed (X2 = 241.189, df = 54, P < 0.01). To elaborate, Table 10.1 shows a cross tabulation of Egloff ’s rim forms with decorative method, with ad- Table 10.1. Cross-tabulation of chi-squared results showing correspondence of decorative method to Egloff ’s (1975) rim type. Blue indicates adjusted residuals greater than two. Red indicates adjusted residuals less than minus two. Appliqué Incision 1 –1.8 2.1 2.5 –3.0 2 –1.1 –0.5 1.9 1.1 3 0.9 –1.4 0.7 –0.1 4 0.3 0.2 –0.5 –0.5 5 0.7 –1.9 2.8 –1.1 6 2.5 –1.2 –1.5 –1.4 7 –0.2 0.7 –0.4 –0.4 8 1.2 –0.9 –0.4 –0.4 9 3.6 –1.5 –2.2 –2.0 10 – – – – 11 1.7 –0.7 –1.0 –0.9 12 –1.6 –0.9 4.8 –0.4 13 – – – – 14 –3.3 4.6 –0.8 –0.8 15 3.4 –2.5 –1.2 –1.1 16 – – – – 17 –1.2 1.6 –0.3 –0.3 18 0.1 0.6 –0.7 –0.6 19 –1.5 2.6 –0.9 –0.8 20 –1.2 1.6 –0.3 –0.3 21 –0.3 –0.1 –0.6 1.4 22 –4.9 0.1 –1.6 11.1 237 Paddle Punctation Chapter . Materialising Ancestral Madang justed residual values of greater than two, indicating substantially more counts than expected, marked blue while those with less than negative two, indicating substantially fewer counts than expected, are red. The results indicate that many rim forms were more or less frequently decorated with appliqué, incision, paddle impression, or punctation. For instance, Rim Form 22 is frequently decorated with punctation, and much less often with appliqué. This form is consistent with modern Madang style rims (Class 3), which are decorated with punctation around the neck. blage except for one surface find (Y-6). This would support the assertion that decoration can, to some degree, be used as a relative temporal signature, just as forming techniques gradually shifted from Class 1 to Class 3 being dominant. Gaffney Egloff (1975) We can integrate the Madang style decorating techniques into broader learning schemes within the production groups. For instance, the introduction of impression as a decorating method at Nunguri, to supplement appliqué, incision, and paddle impression, probably resulted from vertical learning, perhaps with potters marrying in from This pattern is counter to the data presented in this volume, another potting tradition, who then continued that methwhich suggest no significant differences between the tech- od and taught it to their daughters. This is because the use nical classes (X2 = 13.680, df = 12, P = 0.315), despite similar of tools to impress rather than incise or apply would have sample sizes (Egloff = 291, Gaffney = 213, both with un- required subtly different gestures to those learned by potknowns and plain sherds removed). Table 10.2 compares ters accustomed to using paddling, appliqué, or incision. the decorative data produced in this volume (compiling Alternatively, independent innovation by one or more potTilu, Nunguri, and surface collections) with Egloff ’s (1975) ters within the production group may have occurred and data, again using a cross-tabulation of chi-squared results later been increasingly adopted by others observing that and adjusted residuals. This time, Egloff ’s rim forms have potter or her pots. been condensed into Madang style Classes 1–5. The table shows that there are many notable correlations between In pre-colonial Madang the symbolic meaning of decoraform and decoration in Egloff ’s assemblage, whereas in the tions is not apparent. As outlined in Chapter 4, amongst 2014 assemblage there is only one correlation, with Class 4 modern Bel potters there are three stages of decoration rims less frequently displaying impression/punctation. It is with varying importance. First, the zone of decoration is unclear why Egloff ’s assemblage produces notable associa- delimited to the shoulder and neck of the pots (this rule tions and the 2014 assemblage does not, but it may relate to is usually only ignored if producing a pot for the tourist Egloff ’s assemblage representing a greater geographic and trade). Second, small incisions are usually made around temporal range. It does appear that Egloff ’s Rim Form 22, the shoulder. Finally, the potter makes deep impressions which is well represented (n = 70), is a modern iteration of around the neck. However, these aspects of decoration are a Class 3 vessel that is not represented in the 2014 assem- considered by the Bel to be arbitrary matters of personal preference and without discursive meaning. This is similar to other examples of pottery production internationally Table 10.2. Cross-tabulation of chi-squared results show- (e.g. Stanislawski 1977 studying the Hopi, or Gosselain 1992 ing correspondence of decorative method to Class 1–5 rims. studying the Bafia), where production technique is conEgloff ’s assemblage above, Gaffney’s assemblage below. Blue sidered important but decorations are seemingly arbitrary. indicates adjusted residuals greater than two. red indicates This is not to say that specific decorative forms did not adjusted residuals less than minus two. have specific meanings in the past (although these are not known by the modern Bel). For instance, in the early 20th Appliqué Incision Paddle Punctation century, Hanneman (1969) collected numerous designs from Bel artefacts, particularly wooden ones. For many 1 –4.8 2.0 –1.2 2.8 of these geometric designs the makers or users could ascribe the meaning and symbolism behind the decoration. 2 –1.4 2.3 –0.8 –1.5 These decorative elements were often personal to the de7.0 –3.2 0.6 –1.6 3 signer and represented important events or chance observations. Small incisions might indicate cassowary tracks 4 0.1 –0.6 0.6 –0.7 if produced by one person, or mouse tracks if produced by another; one design might indicate coconut fronds, 5 –0.8 –0.9 –1.5 2.6 while another similar one might indicate fish spines. It is particularly striking that many of the incised designs Appliqué Incision Paddle Imp/Punct executed on pre-colonial pots are remarkably similar in form to those produced on the colonial period wooden 1.9 –0.2 –1.9 0.8 1 implements described by Hanneman (1969). 2 –1.3 1.0 –0.5 1.0 3 –0.5 1.3 0.1 –0.8 4 1.9 0.2 –0.3 –2.5 5 0.3 0.6 –0.4 –0.8 Given the hints towards an intertwined ritual and bodily significance of pottery production (see Chapters 3–4), we could expand ideas about these rules surrounding decoration further. Oral traditions suggest pots in Madang must be slipped prior to firing, and it is through firing that pots become ‘as a man’; red like the red paint worn by initiated 238       · .  Bel males. This is reinforced in the Bel language, referring to the slip as the pot’s skin (Yeyeg, quoted in Christensen 1975: 86). This stage is an integral and invariant procedure in the production process and without it the pots could not fulfil their function. Just as the red slip represents the red pigment and paint used in initiation ceremonies, perhaps we can tentatively speculate that incisions and appliqué marks on the surface of pots were also symbolic. Perhaps incision marked the incision of young men and the appliqué the scarification. It is worth stating that many scarification designs amongst Sepik groups are identical to those produced on historical pottery and wooden bowls (see designs in Schechter 2011; Terrell 2011), while the correlations between pottery decoration and tattooing have long been cited (Kirch 1997: 131). The magic of the chaîne opératoire We can expand this discussion about ritual and assert that magic was another significant and essential tool in the production sequence. Such intangible and archaeologically invisible materials are often overlooked. However, as presented in Chapter 3, magic was a pervasive undercurrent of life for the pre-colonial Bel, until the Lutheran Church instigated the destruction of magical items and the dilution of magical knowledge systems (Mennis 2006a: 138). Although no magic was discussed during the 2014/2015 interviews with Bel potters, previous researchers have noted the importance of magic in the production both of pots and of the canoes used to transport these pots (Aufinger 1942; Hannemann 1944; May & Tuckson 2000; Mennis 2006a, 2011, 2014; Tuckson 1966) and oral histories frequently indicate production technologies were restricted and required magic (Mennis 1980b, 1981a, 1981b). The Bilbil and Yabob held the vou – technological expertise (also referred to as magic) – to produce and distribute pottery, giving these groups an elevated status on the northeast coast.6 role in activating the chaîne opératoire and materialising ancestral Madang pots. Production groups and communities of practice From the foregoing discussion of pottery procurement, forming, and decorating, we can build pre-colonial pottery manufacture into a tentative series of chaînes opératoires (Fig. 10.6). Although many processes remain archaeological enigmas, simply because they do not leave lasting traces on potsherds or do not survive in the archaeological record, this description of pottery manufacture forms the backbone of interpretation and allows for comparison with other pottery manufacturing groups, extant or defunct. The systematic technological classification presented in Chapters 7–9, and summarised above indicates that in the pre-colonial past one or more production groups were in operation around Madang, collecting similar local raw materials and producing the same variety of potteries with standard techniques. Each production group made pots, following five distinct forming sequences, but there was a large degree of freedom in how the individual pots took form. Decorations were also interchangeable within each raw material and forming sequence, and between production groups, despite following set rules. I posit that these production group/s were operating within a closely related community of practice (see Lave & Wenger 1991; Wenger 1998). Within this community of practice, there was considerable mobility, movements of potters, and transfer of ideas, horizontally and vertically, which maintained technical variability within the community but prevented divergent threads clearly emerging in different production groups. Communities of practice are here seen as overlapping networks of production groups, which share technical traditions (Eckert et al. 2015) – ways of doing and making. These technical traditions are reshaped through learning and sharing across generations (Stark 1998b) and can be During the forming stage, potters would use anvils with identified by examining technical processes and the repetimagical properties to pound malignant tibud7 out of the tion of technical elements in the chaîne opératoire (Eckert clay (Tuckson 1966). In a similar way, men would recite 2008). As introduced in Chapter 1 and expanded in Chapmagical words to appease these spirits when chopping ters 7–9, learning and knowledge acquisition are processes down trees to make canoe hulls or while pulling these logs whereby motor skills and abstract thinking are intertwined through the forest (Mennis 1980b: 24). It follows that it was at every stage of making: learning the properties of raw the raw material that contained malignant tibud, which materials; the use of tools; the craft lexicon; the symbolism needed to be appeased or driven out to safely carry on behind specific shapes, functions, and decorations; and with daily routine. It is not unreasonable to suggest that the envisaged vessel forms. Within the production group, in digging for clay and slip, and collecting temper sands, practitioners first participate through peripheral learning firing material, and fresh water, Bel woman would have ne- such as the collection of clays or the preparation of paste, gotiated locally abundant tibud through the use of magical but progressively graduate into more complex tasks and words or materials. Perhaps then, magic played a central more demanding bodily requirements. Although considerable time is spent learning the movements and materials, 6 Note that on the south Papuan coast, pots were also imbued production entails broader enmeshment with the social world and further time is spent becoming part of the with magic (May & Tuckson 2000: 60). 7 Spirits, souls of the dead, ancestors, or sprites who inhabited group, developing savoir-faire (know how), and making the living world. Everything which could not be explained was a real Bel potter. In some modern studies of apprenticeattributed to tibud, which could cause sickness, natural disas- ship, the ‘technical’ component of the production process ters, and accidents (summarised from Mager 1952; see Chapter is negligible compared to emotional and social labour involved in becoming part of a group (Wallaert 2013). Thus, 3, pg. 53). 239 Chapter . Materialising Ancestral Madang learning within this community of practice resulted in the account for occasional sources of innovation and the dispecialised knowledge of raw material locations, forming verse technical variation within each, owing to differing techniques and firing temperatures, but also a range of im- skill levels, different desires to learn, and different lengths plicit social values such as what to wear, how to act, and of time spent as a member of the community of practice. when to perform specific tasks (Wendrich 2013). This leads on to the second process. Females learning to Through technical traditions shared by the community pot can be compared to the male initiation ceremony. Alof practice, different modes of identity and personhood though gradual and without lavish rituals, the learning were created (Dobres 2000) and social boundaries were process produces knowledge that is controlled and reguformed (Stark 1998b; Wallaert 2013). Those who produced lated through the vou; a magic protocol that keeps technipots within the group were, through their habitual engage- cal knowledge secret within the community of practice. ment with the process of production, seen as different The similarities in clay chemistry, indicating that the area from those who distributed the pots or those who gar- around Bilbil was the locus of clay procurement, suggest dened. Moreover, on the community level those groups that the production groups producing Madang style pots who produced and distributed pots were regarded with were not geographically disparate, unless raw clay was distinct elements of difference, compared with those who being traded long distances. Although Mindiri can be inprimarily produced food for consumption or those who cluded within the community of practice, around Madang made wooden bowls. In the case of the Bel groups, this itself production and consumption of pots was almost exfact can be observed in oral accounts of Bel’s unique place clusively local and it seems there were geographic and soas pottery makers and traders in the broader northeast cial limitations in the past, as there are today, concerning coast’s social landscape (see Mennis 1980b, 1981a, 1981b). which groups could pot. So, although we could expand To paraphrase Childe (1936), in Madang woman makes the boundaries of the community of practice to include herself, and her social group’s place in the broader social the canoe builders, those who transported the pots, and environment. so forth, those groups specifically engaging with pottery manufacture were tightly regulated. Based on the relatively Among the modern Bel potters, the women who learnt to restricted clay sources used and the general conformity pot as young girls are often the most proficient manufac- to an overall Madang style aesthetic, it is likely that the turers, likely because the pottery learning process has be- process of vou, preventing females who married into other come intertwined with other core cognitive processes (see clans along the coast then producing their own pots, has Piaget 1972; Sternberg 1983, 1999). Although vessel forming a long history. is usually a conservative practice, this is especially the case in household production where manufacture takes place Dadeng: aspects of pre-colonial exchange behind closed doors (Arnold 1998). For the Bel, who often pot communally, we might expect less conservatism but In addition to following processes of making, we can also more standardisation in form as there is more communi- examine how objects become bound up in the life histories cation, discursive and non-discursive, about how others and social interactions of the people who transported, exare producing pots. However, the archaeological pottery changed, used, and valued them (Ingold 2011). It is through suggests otherwise, that there was substantial variation these lines of entanglement that lives are lived. Despite this, within broadly conservative but gradually changing tech- we should also recognise that entanglements have their nological processes. boundaries. Fluidity has its limits, stoppages, asymmetries, and consolidations and these are important to describing Two processes operating within the community of practice, the local and historical contingencies of exchange in the which perhaps explains the interplay between conserva- area (Knappett 2011; Strathern 1996; Thomas 2019). tism and change, may be a dialectic between 1) marriage, and 2) the vou. If marriage along the Madang archipelago The community of practice – a network of groups along was primarily exogamous, as it is today, then it would the coast, and to some extent extending inland, with have encouraged substantial residential mobility over the shared ways of doing and making things – was essential lifetime of the potter, with Bel potters learning to pot in to the maintenance and reconfiguration of these exchange one clan group but teaching their daughters in another. Si- processes. This section will describe aspects of pre-colomultaneously, Bel woman from non-potting clans or non- nial exchange around Madang and examine how, and perBel woman would have married into production groups haps why, changes to pottery distribution occurred over and learned to pot as adults in these marital groups. The last 500–600 years. This will compare the results of the movement of inexperienced women into potting groups present volume with the wider archaeological literature as well as the movement of experienced potters within regarding the northeast coast at 1) the local level, and 2) these groups would have been a major source of innova- the regional level. tion and variation within the technical tradition. The high degree of interaction within the community of practice Local distribution over the generational timespan would account for consistent gestures and tools being used, following specific The community of practice operated within local and reand vertically taught chaînes opératoires, but would also gional networks, but had locational nodes, at the clay de240       · .  Access to materials Extraction Magic (?) Transport Local ‘Yabob’ clay (?) Local ‘Bilbil’ clay Black (Fe/Mg) sand temper White (light) sand temper Calcareous sand temper Grog temper (?) Mixing of paste Mixing of paste Mixing of paste Mixing of paste Rim preform No rim preform (?) Paddle and anvil forming Paddle and anvil forming Magic (?) Anvil used to shape body interior Hand molding to produce everted rim Unmodiﬁed everted rim Continued Anvil used to form collar in interior of vessel Anvil used to shape body interior Hand molding to produce inverted rim Paddling to bevel external rim Unmodiﬁed inverted rim Unmodiﬁed everted rim Unmodiﬁed incurving rim Continued Continued Continued Continued Figure 10.6. Chaînes opératoires of pre-colonial Madang style potteries. 241 Chapter . Materialising Ancestral Madang Continued Continued Continued Continued Continued Paddling to bevel external rim Unmodiﬁed inverted rim Unmodiﬁed everted rim Unmodiﬁed incurving rim Various decoration Various decoration Various decoration Various decoration Application of slip Application of slip Application of slip Application of slip Application of slip Firing Firing Firing Firing Firing Technical class 1 Technical class 2 Technical class 5 Technical class 3 Technical class 4 Distribution Distribution Distribution Distribution Distribution Use in cooking (See Gaﬀney et al. 2020) Use in cooking (See Gaﬀney et al. 2020) Use as water storage? Use in cooking? Use in cooking? Breakage Breakage Breakage Breakage Breakage Re-use? Re-use? Re-use? Re-use? Re-use? Discard Discard Discard Discard Discard Unmodiﬁed everted rim Incision Appliqué Impression Paddle impression Figure 10.6 continued. Chaînes opératoires of pre-colonial Madang style potteries. 242       · .  posit, the household, the workshop, the canoe, and across the landscape (Ingold 2010). On the local scale, oral traditions and ethnographic observations suggest exchange took place, often informally, within Bel settlements, along the island chain between Bel groups, and on the mainland between Bel and inland clans (see Lawrence 1964: 27; Mennis 1980b, 1981a, 1981b; Sentinella 1975). At ethnographic contact, Bel groups were primarily living on offshore islands and in coastal localities around Madang Lagoon (Kranket, Siar, Riwo, Malmal, and Sek) or on the four small islands extending southwards down Astrolabe Bay (Bilbil, and three islands used by the Yabob). To establish where Bel groups were living deeper in the pre-colonial past, how they were interacting, and with whom they were exchanging, it is useful to synthesise the available local archaeological evidence. fulfilled the functions of Class 4 and 5 vessels, which were possibly intended as water storage and boiling pots, just as modern Highlands populations do not often use clay pots owing to different cooking methods (Hughes 1977). As above, Egloff ’s excavations in 1973–1974 were geographically extensive, working locally from Sek Island in the north to Bilbil Island in the south. Each of the 17 sites Egloff recorded bore distinct red-slipped Madang style potsherds, suggesting that Bel potting groups were exchanging objects around the Lagoon and onto the mainland in the pre-colonial past. These sites include Sek Island, Migajpanad Island, Yabob Island, Bilbil Island, Nagada, Vidar, St Fieldis (near Alexishafen), Tabad Island, Riwo Island, Malmal village, Malmal Udou, and Malmal Island. To refine the chronology of exchange, Egloff obtained three radiocarbon determinations from Malmal village (Tilu) indicating occupation by c. 550 BP, calibrated to c. 300–700 years ago at two sigma (see Chapter 5). Egloff ’s dates were primarily derived from combined samples of charcoal from different depths, using conventional dating techniques that produced a substantial error range of 400 years. The 2014 excavations and subsequent AMS radiocarbon dating of single wood-charcoal samples from both Nunguri and Tilu have refined this initial date of ceramic occupation around Madang. Although the 2014 dates broadly confirm the original interpretations of Egloff (1975), who suggested Madang style pottery production and exchange along the local coastline was very recent, the earliest known evidence for occupation can now more precisely be situated around 550–650 cal. BP. As presented in Chapter 5, previous archaeological work was completed by Jim Allen (1971) and Brian Egloff (1973, 1975) around Madang. However, the excavations were limited in extent and the subsequent analysis of ceramic material was cursory. Nonetheless, these assemblages provide comparative datasets from a variety of locations around coastal Madang. The range of pot forms and decorations presented by Allen and Egloff are consistent with those described from excavated contexts at Nunguri and Tilu, along with surface collections from Kranket, Yabob, and Siar. Allen’s description of sherds from the JAE site fits within the present Madang style classification (Classes 1–5) and suggests a very recent pre-colonial occupation on the mainland by non-Bel groups, trading with Bel potters for food, bush material, tools, as bride prices, gifts, compensation, and so on.8 It is notable that Tilu in the north contains subtly older deposits than at Bilbil Island, despite Bilbil featuring as a Although Allen used a different classification system, the locus of pre-colonial pottery production in oral accounts. JAE rims were grouped by morphology and forming tech- Deposits from other mounds on Bilbil Island might yet nology, making the results broadly comparable to the pre- yield slightly older dates, in line with Tilu. Alternatively, sent Madang style Classes 1–5. Following my classification pottery production at Bilbil Island may indeed be slightly system, 54% of Allen’s sherds are attributable to Class 1, 7% younger than the initial production and exchange around to Class 2, 40% to Class 3, and 0% to Classes 4 and 5. The Madang, with the first potters living on Yabob Island, as nearest approximation of this distribution is from the up- the Honpain story states (see Chapter 3 and 6), and later per deposits at Nunguri, especially Spit 5, which contains dispersing to Bilbil, closer to the favoured clay sources. 40% Class 1, 14% Class 2, 31% Class 3, 14% Class 4, and 0% Class 5. This deposit is associated with a date of 210 ± 25 BP It is not certain from either the 1973–4 or 2014 excavations (OZS558; see Chapter 6), calibrated to either modern (15–0 at Tilu, whether the initial ceramic occupation at 550–650 cal. BP), or very recent in the pre-colonial past (304–267 cal. BP represents the exchange between Bel potters near cal. BP and 215–145 cal. BP) at two standard deviations. The Bilbil/Yabob and other Bel groups living at Malmal (Tilu), high number of Class 3 rims at JAE suggests recent pre- or if those living at Malmal (Tilu) were non-Bel speakers colonial occupation (c. 300–150 years ago), while the lack and less closely related, both culturally and genealogically, of shoulder carinations, along with the presence of appli- to the pottery makers. As Tilu was plausibly an island at qué, indicates the deposit is not modern. The absence of initial occupation it may be more suggestive of a Bel settleClass 4 and 5 rims may result from a small sample size ment pattern, as the Bel identify as ‘people of the coast/sea,’ (n = 85). It may also represent cultural differences in the and occupy many of the offshore islands. The sequence mainlanders’ practices of storage and food preparation. suggests no disjuncture in material culture, implying that Perhaps other items such as wooden bowls or bamboo Tilu has been occupied by a lineage of people with similar exchange, consumption and discard practices throughout 8 This is following oral traditions that suggest Bel groups did the past few hundred years. The working interpretation is not live near the JAE site or on the mainland around modern therefore that the early deposits at Tilu represent part of Bilbil village until the German administration in the late 19th the initial Bel settlement in the Madang area. century (see Chapter 3 and 4, also Mennis 1980b, 1981a, 1981b). 243 Chapter . Materialising Ancestral Madang Neither Egloff nor Allen describe any non-Madang sherds in their local assemblages.9 Fabric and clay composition were not examined by either scholar, so it is impossible to state unequivocally that exotic or non-Madang style pots were absent from the assemblages. It is readily apparent though that Madang style pots dominated local exchange and consumption. This is consistent with the 2014 investigations, which only recovered 12 ‘exotic’ sherds, equating to under 0.0003% of the total. sherds were recovered, and may indicate a social boundary in the direct exchange of pots north towards Hansa Bay. As yet, there is no evidence that Madang style pots were exchanged as far as Aitape on the Sepik coast, or vice versa, but the superficial similarities between the Madang style and the Aitape sequence are striking, as pointed out by Lilley (2002: 88). Madang style ceramics were certainly traded further east, where it appears the majority of pre-colonial and colonial Local exchange patterns are often difficult to tease apart period exchange was directed. Madang style sherds have where resource procurement zones overlap (Arnold 1980). been found along the Rai Coast, at Sio and Gitua on the However, the comparison between the material described Huon Peninsula, in the Vitiaz Strait, the Huon Gulf, and above with summary collections from the 1970s suggests into New Britain (Abramson 1969; Egloff & Specht 1982; local exchange of pots (known as Dadeng in Bilbil) be- Gaffney et al. 2019; Gosden & Webb 1994; Lilley 1986, 1991; tween Bel groups was probably occurring from c. 650–550 Lilley & Specht 2007; Summerhayes 2001a). Lilley (1986) years ago, while exchange with mainland groups was al- provides the only substantial description of a Madang most certainly occurring by c. 300–150 years ago, prior to style assemblage outside of Madang itself and he describes European contact. sherds from dated deposits at Sio (KBQ) and on Malai Island (KLJ) in the Siassi group. Regional distribution Of the 129 classifiable Madang style rims identified in the The discussion will now turn to regional exchange net- Vitiaz Strait sites, Lilley discriminates three classes, from works. Like the island of Tumleo off the Sepik, or Yule which this volume’s classification is ultimately derived. Island, Motupore, and Mailu off the south Papuan coast, Class 1 rims, everted with round lips, represent 61% of which could not produce enough food to support their the assemblages; Class 2 rims, with pointed lips, represent populations (see Irwin 1991; Summerhayes & Allen 2007; 31%; Class 3 rims, with flat lips, represent only 8% (LilTerrell et al. 2011), the Madang offshore islanders had to ley 1986: 195). Interestingly, Class 4 and 5 vessels are not turn their gaze outwards to maintain subsistence (Men- present, perhaps suggesting consumer choice in the exnis 2014). At ethnographic contact, the Madang exchange change of pots with different functions, as at JAE on the network extended several hundred kilometres along the Madang mainland (see above). There may not have been northeast coast, overlapping with the Siassi network, the any demand, as wooden bowls were produced along the Sio, the Arop, the Tami and so on, to obtain crops and Rai Coast and in the Siassi and Tami islands. domestic animals, along with desired material culture (Harding 1967). To investigate the extent of this network The full range of Madang style decorative methods is repin the pre-colonial past, the analytical lens is now cast resented in the Vitiaz Strait assemblages: appliqué (72%), more broadly onto the northeast New Guinea region. incision and punctation (21.5%), paddle impression (0.3%), Comparisons will be drawn between the local Madang and incised/impressed appliqué (6.1%) (Lilley 1986: 206). pottery assemblages and those described elsewhere in the Overall, appliqué was common on Class 1 (33%) and 2 archaeological literature, in an attempt to establish how vessels (37%), but incision was more common on Class 3 pots were exchanged through the region. The distribution vessels (60%) (Lilley 1986: 209), which mirrors the results of Madang style pots outside of Madang itself suggests from the 2014 sample, along with Egloff ’s (1975) Rim Form that local and regional exchange was subtly different. 22. A cross tabulation of chi-squared results (Lilley 1986: Table 6.3), however, indicates that based on the sample size To the north, Madang style pots were transported at least no statistically-meaningful association between decoraas far as Karkar Island, probably to the Takia-speaking tive method and rim form is necessarily present (X2 = 7.607, Bel, who today occupy the southern half of the landform. df = 4, P = 0.109, using Monte Carlo simulation), further Egloff (1975) excavated six sites on the island (JBL–JBQ), supporting the 2014 results (contra Egloff 1975). Lilley’s all of which contained red-slipped Madang style sherds, results have been converted to display adjusted residuals consistent with those described in this monograph. Many in Table 10.3 and can be compared to Table 10.2 above. of them were decorated with linear appliqué, suggesting that this exchange occurred prior to the colonial pe- As per the 2014 sample, Lilley noted only one carinated riod. Recent excavations and dating on Karkar confirm shoulder, in contrast to modern Madang vessels, further the pre-colonial presence of the Madang style in the last demonstrating that almost all pre-colonial Madang style half-millennium before present (Gaffney et al. 2018). At pots were more spherical in shape (e.g. Fig. 3.8, especially Sarong (JBK), on the adjacent mainland, no Madang style pot D705). Inner rim notching was present on 66% of Lilley’s Class 1 rims, 72% of Class 2 rims, and 80% of Class 9 Egloff (1975: Figure 1) describes Sarong-style sherds, which 3 rims, which is more akin to the Tilu assemblage (78%) are deeply incised, but these appear to only be present further than Nunguri (47%), supporting a generally earlier date for much of this exchange. north at Sarong and on Karkar. 244       · .  Table 10.3. Cross-tabulation of chi-squared results showing correspondence of decorative method and rim class (based on Lilley 1986: Table 6.3). Blue indicates adjusted residuals greater than two. Appliqué Incision Plain 1 0.0 1.0 –1.6 2 –1.3 –0.1 2.5 3 1.2 –0.9 –0.7 is testament to the flexibility of the procurement stage. Of particular interest, however, are three anomalous sherds. Table 10.4 shows Lilley’s Paste Groups 1 and 18 do not fit with the range of fabrics observed around Madang, as they distinctively lack pyroxene. Lilley had difficulty in ‘sourcing’ these pastes to the nearest ethnographic clay because while most of his Madang style sherds grouped with the Bilbil clay sources, so too did many Lapita and Sio sherds (Lilley 1986: 244–245). It is possible that some Madang style sherds would not group with the Bilbil sources in clay or fabric, especially if they date to before initial Bel occupation around Madang, perhaps suggesting potting initially occurred somewhere else in the region. On the site level, at Malai (KLJ) in the Siassi Islands, Madang sherds are present alongside Type X and Sio styles and formed a substantial part of the assemblage (30%). The site dates to very recently, mostly within the last 200 years and definitely within the last 550 years, making it a good comparison to Nunguri. Throughout the sequence, across two excavation pits, Madang sherds become more common, especially by Layer 2/3, which likely dates to very recently in the pre-colonial past (Lilley 1986: 297). Class 1 and 2 rims are present throughout the sequence, contributing 53% and 30% of the Madang rims respectively.10 Curiously, Class 3 rims are absent, which on the basis of the analyses in the previous chapters, strongly suggests the deposit is older than c. 200 BP (c. 0–300 cal. BP), despite Lilley’s numerous modern dates on wood-charcoal in Layer 2 (see Chapter 5). No trend over time was observed in the number of different rims being deposited, but appliqué Lilley (1986) also provides the only other examination of Madang style fabric and composition, using thin-section petrography, supplemented by emission spectrography and XRD. His analysis of non-plastic inclusions was, in some ways, more thorough than mine, describing also the relative abundance of different minerals along with mineral texture. Nonetheless, the results are broadly comparable, and demonstrate pottery exchange between Madang and further east. Supporting the results presented in Chapter 8, Lilley (1986: 229) notes a high diversity of pastes in use by Madang style potters (Table 10.4), some even overlapping with the Sio style. Madang style pastes can be grouped into half of the nineteen paste groups that Lilley identified. Most sherds fall into the range of fabrics described around Madang and comprise various proportions of Fe-Mg (primarily pyroxene), light (plagioclase and quartz), and calcareous (probably coral rather than shell) inclusions. This 10 Note, 17% of Madang style rims could not be assigned to a class. Table 10.4. Comparison of Gaffney’s fabric and Lilley’s paste group designation.* Gaffney fabric Calcareous Description Calcareous dominant with minor Fe-Mg (Px)/light (Pl+Qtz) inclusions Light dominant (Qtz+Pl) with minor Fe-Mg (Px)/calc. inclusions Light Light dominant (Qtz and/ or Pl) with minor Fe-Mg (Px) inclusions Fe-Mg Fe-Mg (Px) dominant with minor light (Pl and/or Qtz) and calc. inclusions Anomalous Calc. dominant with minor light (Pl or Qtz) inclusions. Fe-Mg (Px) absent Lilley paste group Sherd type Classification 2 Rim Class 3 2 Decorated body Appliqué 2 Decorated body Appliqué 2 Decorated body Appliqué 2 Plain body Slipped 16 Rim Class 1 16 Decorated body Incision 16 Decorated body Paddle impressed 16 Plain body Slipped 19 Decorated body Appliqué 19 Plain body Slipped 15 Decorated body Incised linear appliqué 12 Decorated body Incision 8 Plain body Slipped 11 Rim Class 1 18 Decorated body Incision 1 Decorated body Incision 1 Decorated body Incised linear appliqué * Numbers and designations extrapolated from Lilley (1986: Fig. 7.6–7.7.) 245 Chapter . Materialising Ancestral Madang became less common through the layers, which is consist- quence is most consistent with the Madang chronologies ent with the trend at Nunguri. The petrology of Lilley’s and may provide the best support for a pre-700 year old Madang style material was considered too heterogenous ancestral Madang style. Madang sherds in Lilley’s Pit II to delineate any meaningful trends through time. Layer 4 (older than 900 cal. BP) are all Class 1, similar to the earliest deposits at Tilu. At Sio (KBQ), Madang is again found alongside Type X and Sio style, but the Madang style only makes up 13% of the From these analyses, Lilley (1986: 340; 1988) was the first to total sherd count (Lilley 1986: 307). It is interesting that al- formulate a phase-based model for pre-colonial exchange though the Sio assemblage has a large Madang component, along New Guinea’s northeast coast. His (1986: 341) Phase no Sio pots were observed in the Madang assemblages I 1 denotes Lapita exchange in the Vitiaz Strait c. 2500– analysed. This could be owed to lack of interest in reciev- 2800 BP and will not be discussed further. In Lilley’s iling pots on the part of the Bel traders, who desired food lustration of Phase 2, sometime before c. 550–600 BP and and exotic material culture. possibly up to 1600 BP, Ancestral Madang pottery characterised by incision is shown arriving in Sio from the MaTwo test pits were excavated at Sio (KBQ), but there are dang coast. However, the 2014 excavations around Madang inconsistencies in the ceramic sequences, with Pit I sug- suggest the earliest evidence for pottery making along that gesting Class 3 was present early in time, at least c. 900 coast is only c. 650 years old. Although any interpretations years ago, and Pit II suggesting Class 3 was more recent, should remain tentative, given the possibility that future c. 550 years ago at the earliest (Lilley 1986: 336).11 Lil- research could uncover earlier evidence for ceramic manley (1986: 334) formulates two hypotheses to reconcile ufacture in the area, if Madang style pots were traded in these discrepancies: 1) his radiocarbon dates do not ac- to Sio at this early time, then they may have been coming curately describe the age of associated pottery, implying from somewhere else. disturbance, mixing, redeposition, or vertical movement through the deposits, or 2), there is spatial variation in Lilley Phase 3 is dated from c. 550–600 BP to c. 300–350 BP the deposition of pottery between the pits. Based on the and sees increasing numbers of Madang style pots being Madang style seriation proposed above (Fig. 10.5), hypoth- imported, used and deposited in the Huon/Vitiaz Strait. esis 2) is plausible if both of the deposits containing the This occurs just after initial occupation and pottery manuMadang style Class 3 rims date solely to around 0–600 facture around Madang and is simultaneous with the burst years ago. There is likely to be some bias in the Pit I as- of local pottery exchange between Bel groups around Masemblage meaning Class 3 rims are not present in upper dang. Phase 4, from c. 300–350 BP to less than 200 BP is deposits. Lilley (1986: 333) suggests this could be caused by subdivided into two sub-phases. The first sub-phase sees the redeposition of older beach sands mixed with sherds Madang sherds continuing to play a minor part in regional onto Layer 1. Any Class 3 rims present in deeper deposits assemblages. During the next sub-phase, leading up to Euand in association with older dates can be thus regarded ropean contact, greater numbers of both Madang and Sio with skepticism. This is only problematic for Pit I, where pots were being traded around the Huon/Vitiaz Strait area. a small number of Class 3 rims is present.12 The Pit II se- Phase 4 is contemporaneous with an increase in the number of sherds being deposited locally at the Madang sites, 11 Pit I consists of six layers, containing 28 Madang rims and 795 and the first suggested evidence of exchange with non-Bel decorated sherds (Lilley 1986: 327). Class 1 (53%) is present in groups around Madang. Layers 1–4, Class 2 (25%) present only in Layers 2–3, and Class 3 (11%) present in Layers 2/3 (mixed) and Layer 4. Pit I Layer Having discussed local and regional exchange, Figure 10.7 2 is dated to 670 ± 60 BP (ANU-4332), calibrated to 539–699 illustrates tentative chronologies for Madang style ceramic cal. BP. However, there appears to be very minor inversion in exchange along the northeast coast, adapting Lilley’s four Layer 3, dated to 1) 340 ± 90 BP (ANU-4330), 267–535 cal. BP phase sequence on the basis of new evidence. This new (92.3%); 2) 300 ± 100 BP (ANU-4329), 250–521 cal. BP (77.5%); model incorporates the results of the present volume with 3) and 510 ± 60 BP (ANU-4606), 467–653 cal. BP (100%). Layer the wider archaeological literature. The chronological 4 is dated to 1500 ± 70 BP on shell (ANU-4607), calibrated to timeframes are only indicative and are simply designed to 913–1216 cal. BP (100%), but containing very low densities of show changes to the network over time. The first period, Madang style sherds (see Lilley and Specht 2007). Pit II con- before c. 700 years ago indicates the possible exchange of sists of four layers containing 21 Madang rims and 34 deco- pots to Arop/Long Island and Sio, but from somewhere rated sherds. Class 1 (43%) is present from Layer 1–3, Class 2 other than the Madang Coast – perhaps ‘Yomba Island,’ (10%) is only present in Layer 2, and Class 3 (33%) is present which will be discussed in the next section. Then, with a only in Layers 1–2. Pit II Layer 1 is modern. Layer 2 is dated shift of potters to the Madang Coast just after 700 years to 400 ± 90 BP (ANU-4334), 284–558 cal. BP (97.7%). Layer 3 is ago, there was an expansion of the Madang exchange netdated to 950±BP (ANU-4336), 723–979 cal. BP (99.4%) at the work, delivering to Sio and Arop, but also feeding into latest and 1290 ± 100 BP (ANU-4337), 978–1362 cal BP (100%) local exchange networks. Through Sio, down-the-line exchange probably accounts for the Madang style in the at the earliest. 12 It is not clear from Lilley (1986) how many Madang Class 3 Arawes in southwest New Britain and possibly in the Tami vessels are present in Pit I Layer 4, but the number must be Islands. It is around this time that Madang pots also make it to Aibura Cave in the New Guinea Highlands, through fewer than four (the total number of Class 3 rims in the pit). 246       · .  inland down-the-line exchange perhaps using river valleys. Then, between around 550–200 years ago there was a further reconfiguration of the network, with pots making their way in to the Siassi group, and then further on to Kove in New Britain. Lastly, in the period just before European contact, the export of pots to Sio and the Siassi group dramatically increased and oral traditions tell us that pots were certainly being moved up into Karkar to the north. With this basic framework in mind, we now return to the concepts of embodied knowledge and communities of practice. Exchange, embodied knowledge, and communities of practice Approaches to exchange and trade voyaging must diverge from simply tracking the lines connecting A to B and instead tease apart how people forged and experienced these links (Scheldman 2011). In this way we can begin to locate creativity, learning, and embodiment, not only on the micro-scale, within the production group, but also across the social landscape between trade friends. To integrate the foregoing discussion of ceramic exchange more thoroughly with embodied knowledge, it is important to note that, like production, exchange must operate within the logic of local and regional communities of practice. As we have seen from the distribution of Madang potterybearing sites, the pre-colonial communities of practice around northeast New Guinea extended between groups living along the coast, on offshore islands, and in the foothills. These groups shared ways of coming together, hosting guests, and exchanging objects (see Thomas 2009). The community of practice that facilitated Bel object exchange was not necessarily the same as the community of practice formed and experienced by the Bel women who were making the pots (see above). On these overseas trade voyages, men were making their own communities of practice, subtly different from those formed around Madang itself, by reinforcing existing relationships and creating new threads in their social networks. Through processes of routine action – learning the art of sailing – often beginning from a young age (Mennis 1980b: 17), men would have generated and reshaped their bodily knowledge. Pushing the balangut out to sea, tethering the sails, and understanding the winds and the stars, were all part of this technological process (Mennis 2011), making knowledge and know-how at the interface of bodies, techniques, and materials (Lemmonier 1992). This process also involved learning the magic to negotiate local tibud at different points of land and in the water around them (see above). While on trade voyages in the recent pre-colonial period, men would recite secret words to evade the water tibud and would throw heated potsherds into the waves to quell them (Mennis 1980b: 44). It was at this point that two material lines – pottery and canoes – were woven together integrally. HANSA BAY HANSA BAY Hankow Reef Hankow Reef MADANG MADANG ? Arop Arop VITIAZ STRAIT VITIAZ STRAIT N E W B R I TA I N Sio N E W B R I TA I N Sio Kafiavana Aibura HUON PENINSULA HUON PEN. Arawe Is. ? Tami Is. HUON GULF HUON GULF 0 100 km Before c. 650 years ago c. 650–550 years ago HANSA BAY HANSA BAY Karkar Is. MADANG ? Karkar Is. Hankow Reef Hankow Reef MADANG Kove Is. Arop VITIAZ STRAIT N E W B R I TA I N Sio HUON PEN. N E W B R I TA I N Sio Siassi Is. Kafiavana Aibura Kove Is. Arop VITIAZ STRAIT Siassi Is. Kafiavana Aibura Arawe Is. HUON PEN. ? Tami Is. Arawe Is. ? Tami Is. HUON GULF HUON GULF c. 550–200 years ago c. 200–100 years ago Figure 10.7. Diagrammatic model showing pre-colonial exchange of Madang style pots around the northeast coast. 247 Chapter . Materialising Ancestral Madang The community of practice also required the interaction of different people for social relationships to be made and remade. Although learning the correct number of pots to exchange for the correct number of taro or wooden bowls was an important set of skills, embodying the finer points of the Dadeng involved a broader set of social regulations: knowing when to arrive at different settlements, who to speak with, how to dress, who to offer betelnut to, and so on. This was particularly the case as traders attended to obligations including expectations of gifting, delayed returns, and debt (dinau) between trade friends. Learning the Dadeng would have been further complicated by the mutability of things, which can require different modes of exchange for the self-same object, depending on specific social contexts (Robbins & Akin 1999; Thomas 1991: 100; Walter & Green 2011: 19). The potsherds described in this monograph probably represent the discard of similar pots acquired through various modes of exchange: for example, direct, mass exchange for food or other material culture, small-scale down-the-line trade and commodity exchange involving middlemen, and gifting during ceremonies or rituals, bride prices, compensations, and reconciliatory acts. Summary To conclude this chapter on pre-colonial production and exchange, we have seen that on present evidence, locally produced Madang style ceramics started to be made suddenly at around 600–500 years ago, representing one or more production groups operating within a broader community of practice. Within this community of practice there seems to have been substantial interaction and knowledge exchange, but bound by tight regulations governing the procurement and production of pots. These pots fed into local exchange networks, along with larger regional systems, which seem to have become more extensive and intensive over time, leading up to the ethnographically observed situation in the late 19th century. By following the materials, learning the movements, and drawing the lines (Ingold 2011), we see that they are all inexorably woven together. However, all of this interaction – the binding up of bodies, minds, gestures, pots, canoes, food, magic, spirits, people, communities, and landscapes – can become bewildering. It is rather in the asymmetries of these interactions, and at the limits of the community of practice, that social boundaries are made, delimiting and expanding people’s place within their social world. We have begun to chart these asymmetries by describing the production groups who made the Madang style ceramics and illustrating the transformative exchange networks proposed above (Fig. 10.7). The next chapter will outline possible explanations for these asymmetries from a culture history perspective, with the aim of reconciling some discrepancies in linguistics, oral tradition, landscape, and material culture. 248 Chapter 11. Bel Culture History of New Guinea. Later interpretations by Terrell and colleagues (2011), however, imply the precursors of the Aitape ceramic sequence are less clear. Conversely, Lilley (1999) suggested that culturally related groups, such as those of the Sepik and Vitiaz Strait, are just as likely to produce dissimilar pottery as they are similar pottery and saw the arOne of the key objectives of this monograph has been to chaeological evidence and Ross’ (1988) linguistic model as describe the history of the Bel potters and the sequence coalescent. I would build upon this and state that culturof events leading up to the ethnographically documented ally related groups are capable of producing pots which on societies around New Guinea’s northeast coast, and Ma- the surface appear distinct but are closely related in terms dang in particular. The previous chapters have focused on of their chaînes opératoires. As introduced in the previous how technological changes occurred through time and chapters, this is because the extended technical syntaxes the evolving processes of production and exchange along required to form pots are generated by long periods of the marginal coastal strip and offshore islands. This final learning and enskilment and eventually become durable chapter will try to identify which events caused these strands of the technological process, even when there is changes. Some core culture history concerns put forward variability in surface features that result from innovations in Chapter 1 regarding Bel origins, migrations, and inter- to minor technical elements and short syntaxes. Equally relatedness, will be teased apart. This brings language, oral though, non-related groups are capable of producing pots tradition, landscape, and material culture back together in that closely resemble one another, but do not share deep an attempt to derive a rudimentary narrative for change technological commonalities, potentially owing to ongoing through time. The volume will conclude by considering maritime interaction and interrelated shifts in aesthetics the implications of this narrative for Pacific anthropol- leading to similarities in surface technical elements along ogy and archaeology and will describe future prospects the coast. This all raises important questions about the for research to further fill gaps in our understanding of Madang style, which, on the surface, closely resembles the Melanesian production and exchange. later sequence of Sepik ceramics, but is dissimilar to both the early Sepik and Vitiaz Strait ceramics despite the Bel languages deriving from the Proto Ngero/Vitiaz Linkage Debates in language, landscape, history, along with the Austronesian languages of the northeast archaeology coast and the Vitiaz Strait. From which direction did the Madang potting traditions arrive, and were they coupled in Austronesian-speaking migrants (east or west)? a simple manner with the movement of the Bel languages? In 1988, Malcolm Ross suggested that the North New Guinea cluster of Austronesian languages had resulted This volume has thoroughly described the pre-colonial from east–west migrations of communities from West ceramic material culture of coastal Madang District, reNew Britain (see Fig. 1.9). However, Terrell and Welsch porting on the objects themselves, along with the broader (1997) argued that the post-Lapita ceramics around Aitape technological processes of production and exchange. It is along the Sepik coast of New Guinea, in particular the clear that various technical elements allow Madang style Sumalo ware, were more consistent with the movement of sherds to be grouped apart from Lapita, Type Y, Type X, potting traditions from further west, in Island Southeast and Sio sherds, which represent distinct technological traAsia. These ceramics were seen to have more in common ditions active at various times around the northeast coast. with Metal Age pottery from the Moluccas than post-Lap- Surficially, these styles would be expected to be related to ita ceramics of the Vitiaz Strait and the northeast coast Madang if there was an east–west movement of potting groups, unless of course these groups chose to deliberately differentiate their products. Even if this was the case, how1 Quoted in Mennis (1981b: 33). The Melanesians were strong people, the Papuans moved back and the Melanesians took their place… The Melanesians came to all the islands from Bilbil to Sek — Pall of Tagari of Bilbil (1976)1 249 Chapter . Bel Culture History ever, the technological processes underlying production are expected to be derivative and closely related. In support of this, strands of the Lapita technological process such as incision, red-slipping, and paddling (see Kirch & Yen 1982: 192–193) have clearly been reiterated or reinvented at various times, surviving through to the Madang potters, some 2500 years later. Other traditions such as Type X and Sio preserve different technical elements, but owing to overlapping chronologies and distinct production signatures, cannot be antecedent or derivative of the Madang style. The working interpretation is that, along with Type X and Sio style, the Madang style is distantly derivative of Lapita and post-Lapita potting traditions around northeast New Guinea. It is notable that many of the vessel forms in the Aiser tradition mimic Madang style Class 4 vessels, the precursor to the modern Bel magob. As suggested in Chapter 10, it is hypothesised that these are linked to Pila pottery producers around Hansa Bay, but the nuances of these interrelations are not known with certainty. May and Tuckson (2000: 175, 302, 310) state that the technique of opening out the orifice of the clay roughout with a stone anvil, as used for magob, is the same amongst Pila potters, as well as Korak speaking potters at Tuvaltae (adjacent to Karkar Island), Kaiep potters near Wewak, and Tumleo Island potters near Aitape itself. A more remotely similar technique involving a stone anvil opening out the orifice of the roughout after a long clay preparation sequence is evident amongst Sio potters2 (May and Tuckson 2000: 150). Further northwest, along the Sepik coast, various ceramic In addition to these similarities, the technique of preformwares have been described (Dickinson 2011; Golitko 2011; ing the rim of restricted vessels such as the Madang style Golitko & Terrell 2012; Schechter 2011; Terrell 2010; Terrell Class 1, 2, 3, and 5 pots, as well as modern Korak and Tum& Schechter 2007, 2011; Terrell & Welsch 1997). This work leo Island varieties (Pétrequin and Pétrequin 2006: 340), provides a dated sequence of four distinct ceramic tradi- is suggestive of historical connections and imply these tions present around Aitape: Nyapin ware (2000–1500 BP), technological processes may be closely related. It is quite Sumalo ware (lasting somewhere between five and 270 plausible that the Aiser ware followed a similar sequence years, starting 1200 or 1400 BP), Aiser ware (1000–500 BP), involving a stone anvil to open the roughout, with a disand Wain ware (more recent than Aiser ware but without tinct stage of rim preforming prior to vessel manufacture absolute dates) (Jones 2011; see Fig. 11.1). Like the Madang using paddle and anvil. style, Terrell sees these traditions as ultimately deriving from Lapita, which had its formative period in the Bis- Decorations on Aiser sherds comprise diagonal incimarck Archipelago (Terrell & Schechter 2007; Terrell et sions and punctations, appliqué nubbins and bands, and al. 2011). It seems that Lapita potters later moved onto punctate-appliqué bands, which are not represented in the the north coast as there is evidence for dentate pottery earlier wares. These incisions and appliqués are arranged production in the form of two poorly-provenienced sur- into similar motifs to the Madang style, with long douface finds around Aitape (Swadling 1981: 66, 1988: 19; Ter- ble bands being flanked by smaller decorative elements. rell & Welsch 1997: 558). Provisional conclusions made by Common forms include configurations similar to LI12, Golitko (2011; Golitko & Terrell 2012) suggest that with few LI15, NA1, NA3, and motifs M28 and M63 (compare this exceptions these four ceramic wares, along with the north volume Chapter 9, with Terrell & Schechter 2011b: Fig. coast Lapita, were made locally. 7.9–10). However, the zonation of decorations is different, with Aiser rims commonly being decorated on the lip and The four Sepik wares share certain technical similarities inner lip, while Madang decorations tend to be restricted with plainware Lapita, and with the Madang style. The to the body/shoulder. Moreover, Aiser decorations display Nyapin, Sumalo, and Aiser wares were almost always red- combinations of incised, applied, and impressed appliqué slipped, but in terms of vessel forms, the Nyapin ware and decorations, whereas Madang sherds rarely combine decothe Sumalo ware do not closely resemble the more recent rative methods. Madang style. Nyapin ware consists of some carinated vessels decorated with incised lines or wavy bands, stick It is interesting that appliqué appears in the Aitape sepunctations, and shell-edge impressions. Sumalo ware quence broadly contemporaneously with, or just before, consists of carinated bowls with everted and concave rims, appliqué occurs around Madang. The chemical analyses and shallow platters (Terrell & Schechter 2011b). Such ves- from Aitape and Madang suggest that Aiser and Madang sel forms, along with these minor technical elements, are represent distinct, locally made styles, but the technical not common in the Madang style. There are greater af- similarities speak of common decent, with divergent trafinities to the Aiser ware, however. This ware consists of ditions emerging vertically through generations of geothin walled, round bottomed pots with simple, everted, graphic separation. As the Pétrequins (1999) point out in unnotched rims, and incurving bowls. Like the Madang their examination of modern production sequences, the style, there are no shallow bowls or platters in the Aiser method of manufacture on Tumleo Island near Aitape is tradition. The Aiser production process is difficult to establish based on the available data; it is unclear whether 2 Sio pots are produced with an elaborate clay preparation stage the Aiser pots were produced using a paddle and anvil, which involves forming the paste into a cup and leaving it to hand moulding, coil, or slab construction. However, based dry overnight before then collapsing that clay into an upsideon the thickness of the body sherds (~0.3–0.6 cm), it is down cone from which the vessel is then shaped. The Korak likely that they were constructed using a paddle and anvil and Sio-Gitua pottery production sequences share distinct technique, similar to Madang style pots. similarities (May and Tuckson 2000: 150). 250       · .  BC/AD cal. BP Madang-style Type X Sio-style Aitape Type Y Lapita 2000 1900 1800 1700 1600 1500 1400 1300 1200 1100 1000 900 800 700 600 500 400 300 200 100 1 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 0 100 200 ? 300 400 500 600 ? 700 800 900 ? 1000 1100 ? 1200 1300 1400 1500 1600 1700 1800 1900 2000 ? 2100 2200 2300 ? 2400 2500 2600 2700 2800 2900 3000 3100 3200 3300 3400 3500 Figure 11.1. Duration of major ceramic traditions around the northeast coast of New Guinea. 251 Chapter . Bel Culture History remarkably similar to that at Bilbil/Yabob, probably deriving from common descent, although they suggest this is owed to an eastward spread of potters from the Moluccas. If, as Terrell and colleagues (2011) suggest, the Aitape sequence represents an unbroken series of traditions leading from Lapita to the present day, it could imply that the Madang style does derive proximally from further west as an offshoot of the Aiser tradition. Alternatively, if the forming sequence of Aiser pots is shown to be dissimilar to the Madang style, it would imply that the commonalities in decoration, slip, and vessel form, are surficial and might represent the horizontal communication and mimicry of designs across a north/northeast coast community of practice and exchange. However, if Aiser and Madang styles were produced using similar chaînes opératoires, dissimilar to the Sumalo ware, it would support a second push from the east onto the north coast by a separate Austronesian-speaking potting group. The superficial decorative and formal similarities, although tantalising, are at present unable to reconcile these different scenarios. long argued that this island was real and once located at Hankow Reef in the Bismarck Sea. The dredging of a submarine volcano just west of Hankow Reef revealed a base of rhyolites and dacites. Usable clays for pottery manufacture are likely to have existed there and, we can infer, also on Hankow (Mennis 2006b). The disappearance of this island was probably not caused by a volcanic eruption as some oral histories attest (Johnson 2013), but, tentatively, could have resulted from a tsunami caused by tectonic activity (Mennis 2006b). Although one approach would be to distil the various oral histories into a unifying narrative, in which the Belspeakers arrived in a single migration event with strong environmental push forces, this might ignore much of the historical complexity that characterised Bel migration. It is perhaps more plausible to view these oral traditions as descriptions of processes lasting several generations or more. For instance, it is possible that people hailing from ‘Yomba’ had populated areas of the Madang coast prior to the island’s destruction. Alternatively, if Yomba was located elsewhere, it may be a collective concept to denote common origins, which would have been important to maintaining social unity and identity throughout intergenerational movements to the Madang coast. In this way, these events can be seen as diasporic processes (see Lilley 2004b), whereby migration flows would have linked the new occupants to their ancestral homeland. Additionally, many of the Bel villages report that they did not derive from Yomba but from other vanished islands. For example, the Siar say their ancestors were already in Madang when Yomba sank, but when Siasawan Point broke, the people living there relocated to Siar Island. As such the oral testimonies may preserve communal memories of various tsunami, seismic, and uplift/subsidence events, which typified the area in the Late Holocene (see Chapter 2) and indeed still do today. Within the Madang Lagoon area and along the coast, there would have been migration flows and interactions of cultures prior to any such Yomba disaster. The new Bel arrivals from ‘Yomba’ may have tapped into an already dynamic and changing social landscape. Despite these limitations, the distribution of similar technical syntaxes involved in rim preforming suggests that this stage of production characterised Late Lapita and post-Lapita ceramic manufacture in the Bismarck Archipelago, before potting groups moved westwards onto New Guinea and its outlying islands, in association with North New Guinea cluster speech communities, as well as southwards to the Massim and south coast of Papua, where Yule Island and Boera pottery is made in a similar fashion, in association with Papuan Tip cluster speech communities (Pétrequin and Pétrequin 2006: 341). However, it is unclear whether these syntaxes are related to similar rim preforming in the Aru Islands and other parts of Island Southeast Asia, where the potters speak non-Oceanic Austronesian languages. The similarities between technical elements and short syntaxes along the stretch of northeast New Guinea between Aitape and Sio also imply that the series of staged population movements by North New Guinea cluster language speakers onto the Sepik coast and later into the Vitiaz Strait and eventually Madang, were characterised by ongoing social networks that facilitated interaction and knowledge exchange. We can investigate these ideas further using the archaeological data presented in this volume. Returning to the four scenarios introduced in Figure 1.8 of Chapter 1, if there Yomba Island had been a single sudden migration of Bel speakers from Following western migration from West New Britain, or Yomba we would expect to see an abrupt start to locallyperhaps, an eastern movement from north New Guinea, made pottery production in Madang, perhaps alongside some oral narratives suggest Bel speakers based them- a handful of exotic pots salvaged from Yomba, lining up selves on an ancestral homeland, often referred to as chronologically with geological evidence and some oral Yomba Island (Mennis 1980b, 1981a, 1981b). At ‘Yomba’ – histories. This would then be followed by a gradual change whether a now-submerged island in the Bismarck Sea, a to ceramic manufacture over the centuries. If there was series of clan areas, or something much more historically trade between Yomba and Madang prior to any such Bel complex – the inhabitants probably spoke Proto-Bel (see migration, we would expect exotically made Madang style Ross 2009) and produced pottery and canoes, trading with pots, with an abrupt change to local production at Yomba’s Arop/Long Island and Sio. The most common Bel origin disappearance. Alternatively, if there was a gradual diasstory tells of Yomba then sinking into the sea and its in- pora to Madang, we would expect to see locally produced habitants fleeing to the Madang coast and other places in Madang style vessels prior to Yomba’s disappearance. If northeast New Guinea (Mennis 1978). Mary Mennis has this diaspora was characterised by high levels of ceramic 252       · .  transfer between the two potting centres, we would expect both locally and exotically produced Madang style pots around Madang, prior to Yomba’s disappearance. such an explanation correspond with the archaeological and geological evidence, however, it would then require us to presuppose an unbroken sequence of ten generations, during which reproduction occurred only later in The current chrono-stratigraphic and ceramic results in- life (between the ages of 40–60). This would align the bedicate there was an abrupt start to the Madang style along ginning of the genealogies with around c. AD 1400–1600, a the coast, involving local production from about 600 BP, time prior to Arop/Long Island’s major eruption. However, calibrated to 550–650 years ago. This corresponds very there is an element of post hoc speculation here – making closely with geological evidence presented in Chapter 2, the genealogical, geological, and archaeological narratives which suggests the last major tectonic uplift in the area line up. occurred at 550 BP and could have caused a tsunami, along with the destruction of offshore islands. However, for now, At present we cannot establish precisely which events unthis association must be regarded with caution, as the folded leading up to the end of ‘Yomba,’ and what led the geological dating used marine-derived radiocarbon, and Bel to choose Madang as their new home. Most parsimothe local ΔR correction is completely unknown (Simon niously, however, the genealogies have likely been elided Day pers. comm. 2015), meaning that the geological and or telescoped and the broad picture that the archaeologiarchaeological evidence may not line up very closely at all. cal, oral historical, and geological evidence paints is one of a single relocation of Bel potters to the coast over a Genealogies collected by Mennis (2006a) suggest that the short time span. It is unclear whether or not the Bel homefirst Bel arrived at Madang around eight to ten generations land continued to exist for a time after the settlement of ago, which conventionally (based on average generational Madang, but there is little to no archaeological evidence cycles of 25 years; see Kirch 2018) would date to about 250 for ongoing exchange with an ancestral Yomba. Another years ago. Taken at face value, this is about 300 years later possibility is that the Bel who lived at Yomba procured than initial ceramic occupation and would support a grad- their clay from the Bilbil area, travelling by canoe over one ual diaspora to Madang prior to Yomba’s disappearance. hundred kilometres, like the Amphlett Islanders do in the However, there are a number of ways in which genealogies Massim (Lauer 1970b). This is a less common, but not imcan be remembered, forgotten, revised, and transformed possible, situation and would be archaeologically invisible. as they are passed down (Linnekin 1997: 13; Thompson 2017: 204). For instance, it is possible that a process of eli- Adaptations to the coast sion may have occurred in the genealogies, whereby the middle portion of the sequence is poorly remembered, The Bel who left Yomba to the Madang coast, whether all and identical names are conflated (see Mercer 1979: 140). at once or over several generations, seem to have preferenSimilarly, when some members of the sequence are forgot- tially settled in uninhabited areas or those marginal zones ten, the rest of the ancestors are recalled as being more re- with less resistance to intrusion. One potential scenario cent than they were, in an effect known as ‘telescoping’ due would see any uplift and associated tsunami that may have to memory error (Sypher et al. 1994). The same process destroyed ‘Yomba’ then displacing existing populations on has been observed on Arop/Long, where genealogies sug- islands near Madang too, making the Yomba Islanders’ regest the previously discussed eruption happened c.1850 AD, location easier. Oral traditions say some people escaped which is about 200 years too late according to geological the island on canoes or makeshift rafts and bunches of dating (Mennis 2006b, quoting Tom Harding and Eldon coconuts, others who were visiting the mainland or other Ball). Additionally, if the purpose of the genealogy is to islands returned to find their home no more (Mennis reaffirm the claims of different clans, explain the political 2006b). The stony ground on the offshore islands around landscape, or legitimise alliances, then the specific length Madang was not productive for horticulture compared of that sequence may be secondary to the names that it to the mainland. Some tried to live on several other isrecites (Linnekin 1997: 14). lands before finally deciding on new offshore settlements. Mennis (2006b) argues that others lost the art of potting Another overlooked factor is the assumed generational as there was no clay on their islands and they could not length; some authors suggest the conventional span of 20 access sources on the mainland. Furthermore, particular to 30 years may not hold in this area of New Guinea. In clans may have quickly placed embargoes on clay sources one instance, Mennis (1980b: 17) recorded that Bek from to regulate production. Those in Madang Lagoon instead Riwo was born in about 1900, but his father saw Miklouho- specialised in canoe hull production for trade. The arMaclay in 1871, so even if Bek’s father was only ten years chaeological evidence attests to this social regulation of old when he observed the Russian, he would have been raw material sources and, perhaps, limitations on the clans about 40 when he fathered Bek. In another more tentative who were permitted to pot. Certainly, the new geographic instance on Ritter Island, Simon Day interviewed a man distribution of the Bel speakers, from a single homeland to whose father claimed to have witnessed the 1888 lateral a mosaic spread across a coastal archipelago, would have collapse. The father supposedly watched the event when required new approaches to resource acquisition, local he was about twenty years old and fathered his son when trade and exchange, craft specialisation, and collaboration. he was about 70 (Simon Day pers. comm. 2015). To make 253 Chapter . Bel Culture History future research. Figure 11.2 illustrates this sequence alongside major external events in the Bel social memory such as the end of Yomba, the eruption of Arop/Long Island, The local sequence the arrival of Europeans, and the Pacific War. These later Having outlined aspects of Bel culture history, integrat- historic time periods are derived from Mennis (2006a) ing the current archaeological evidence, it is now possible and Sinclair (2005). to build a general narrative of long-term cultural change around coastal Madang District. The following sequence Yomba Period (pre AD 1300), defined by Bel ceramic prois designed as an interpretive framework to be tested with duction on Yomba Island (or other ancestral homelands). Towards a Bel culture history BC/AD cal. BP Post-Independence Period 2000 0 Late Australian Administration Early Australian Administration German Colonial Period European Contact Period Pacific War 1900 100 Miklouho-Maclay arrives 1800 Late Pre-colonial Madang Period 200 1700 300 Arop/Long erupts Gradual internal innovation and (?) introduction of new techniques from NE coast 1600 400 Early Pre-colonial Madang Period 1500 500 1400 600 Yomba dissappears 1300 700 1200 800 ? 1100 Yomba Period 900 1000 1000 900 1100 800 1200 700 Figure 11.2. Diagrammatic representation of proposed Bel culture history sequence around Madang. 254       · .  Possible ceramic evidence present at Sio and Arop/Long Island. Madang style: red-slipped, applied/incised, paddle and anvil ware. Early Pre-Colonial Madang Period (c. AD 1300–1700(?)), defined by Bel ceramic production around Madang. Possible specialisation of craft production associated with dispersal of groups along mainland coast. Gradual increases in production and exchange intensity. Madang style: redslipped, applied/incised, paddle and anvil ware. Class 1 and 2 common but scarcity/absence of Class 3 vessels. Late Pre-Colonial Madang Period (c. AD 1700(?)–1871), defined by Bel ceramic production around Madang. Notable increases in production and exchange intensity both locally and onto Rai Coast/Huon Peninsula. Madang style: red-slipped, incised/applied, paddle and anvil ware. Class 1, 2, 3, and 4 vessels common. European Contact Period (AD 1871–1884), defined by Bel ceramic production around Madang. The Bel, and particularly Bilbil, people’s material affluence noted by early European explorers. Madang style: red-slipped, incised/applied, paddle and anvil ware, Class 3 vessels common. Conclusions As introduced in Chapter 1, longstanding models describing transformations of production and exchange in Melanesia attest to increasing specialisation and diversification after the gradual breakdown of larger regional networks in the Lapita period (Allen 1985). Ethnographic accounts (e.g. Strathern 1988) suggest that, over time, social groups and individuals would have fostered elements of difference in their material culture, ways of making things, and ways of doing things. The local Madang model for change proposed above fits well with this trend of increasing intensity in processes of production and exchange leading up to, and continuing through, ethnographic contact. As presented in Chapter 10, this is also consistent with Lilley’s (1988, 2017) observations for the Siassi link in the Vitiaz super-system, where low-intensity production and exchange of Lapita ceramics was followed by a ceramic downturn. It should be noted, however, that based on the distinct similarities in forming and slipping of Ancestral Madang and plainware Lapita, it is unlikely that ceramic manufacture ceased altogether in the area. Later, fragmented pottery production and exchange networks then re-emerged along the northeast coast with increased specialisation and intensity. The fact that this trend seems to have occurred in various places in Island Melanesia at a similar time implies that these networks were not closed systems but part of a wider regional complex of interaction. German Colonial Period (AD 1884–1914), defined by continuing Bel ceramic production around Madang. Colonial process causes disruptions to settlements and pre-colonial technologies. Bilbil and Yabob forcibly moved to mainland. Changes to trade and exchange with introduction of the The Madang results also mirror Irwin’s (1985), Bickler’s money economy. Madang style: red-slipped, incised, pad- (1997) and Sutton’s (2016) studies of the Papuan south dle and anvil ware, Class 3 vessels common. coast, where, upon arrival of new settlers, there was an archaeologically-instantaneous adaptation to local reEarly Australian Administration (AD 1914–1941), defined sources and materials such as clay and temper sands. Imby continuing Bel production around Madang but ex- mediately, local pottery production boomed. Evidence change diminishes and halts during wartime. Madang from elsewhere along the northeast coast (Lilley 1986) and style: red-slipped, incised, paddle and anvil ware, Class 3 the Vitiaz Strait (Egloff & Specht 1982) suggests that the vessels common. distribution of Madang pots continued and even increased in amplitude after the Bel speakers moved to the coast. An Late Australian Administration (AD 1945–1975), defined by outstanding question remains: was there a common incontinuing Bel production around Madang but exchange centive for insular potting groups around New Guinea to has been completely reconfigured owing to war. Attempts move closer to the coast and coastal resources? Despite at disruptions to conventional pottery production in 1970s, being historically contingent, similarities can certainly be introducing kilns and wheel throwing. Increases in tourist drawn between the experiences of small potting groups trade. Madang style: red-slipped, incised, paddle and an- who migrated to the coast, often inhabiting coastal islands vil/wheel thrown/hand moulded wares, bases and stands without much arable land, but specialising in, and intensiintroduced, flowerpots introduced, dominance of Class 3 fying, ceramic manufacture and exchange. vessels. The pre-colonial Bel production and exchange may also Post-independence Period (AD 1975–current), defined by challenge conventional interpretations that ‘segmentary’ continuing Bel production around Madang, particularly economies were organised according to stasis and equilibat Bilbil, with Yabob and Mindiri becoming defunct. Ex- rium, lacking flux, and without strong internal processes change to Rai coast is now almost completely absent, with driving material capital and profit. Around coastal Madang, a focus on local tourist trade. Wheel throwing rejected in specific groups, genders, clans, and individuals worked tofavour of paddle and anvil technique. Red-slipped, incised wards differing specialisations. Clans could control certain and perforated, paddle and anvil/hand moulded ware, bas- magic and intellectual property, essential to production es and stands, flowerpots, handwritten incision/signatures, and exchange: some clans and genders controlled canoedominance of Class 3 vessels. hull production, others controlled pottery manufacture, and others still controlled pottery distribution. Did the Bilbil begin to capitalise on a privileged position as the 255 Chapter . Bel Culture History pot makers and distributors around Madang and the northeast coast, which allowed them unequal access to exotic resources? This certainly seems to have been the case in Harding’s (1967: 139; see also Sahlins 1972: 382–383) description of the Siassi traders who made material gains by controlling the movement of objects and managing expectations of value. Both Nikolai Miklouho-Maclay and Otto Finsch noticed the Bilbil people’s disproportionate access to exotic materials when they arrived in the late 1800s. Seen through a Western capitalist lens, the Bilbil and other subsistence traders like the Siassi were understood as thrifty and calculating, leading to Sahlins’ (1963) classic distinction between Melanesian free-enterprising capitalists and Polynesian feudalists. These coastal groups lacked the landesque capital which provided mainland populations with the potential to generate, transform, and control both food and interpersonal alliances. Amongst the Bilbil, however, these potentialities for both growing produce and making social relationships moved to the clay pit, the potter’s hands, and the red-slipped vessels that were then mobilised around the coast on trade voyages. This volume suggests that embargos on this way of life and trends towards increasing monopolisation of production and exchange can be traced back to the Early Pre-Colonial Madang Period, just after the Bel relocated to Madang. from contexts older than c. 550–650 years old. This would demonstrate, based on forming techniques, whether or not they are consistent with the Pre-Colonial Madang Period ceramics, and would assess the chemical composition of the pots relative to the current study. If the chemical composition is similar or the same it might imply that Bel potters travelled to Madang to procure clay, or that Madang potting traditions are in fact older than c. 550–650 years on the coast and that earliest sites have not yet been located. If the chemical signatures were different, it would be the first strong evidence for local manufacture on Yomba (or an equivalent, geographically distinct, ancestral homeland). The main inhibitor for refining these narratives is a lack of fieldwork. To form a more secure local sequence, excavations could be undertaken on Siar Island, Yabob Island, Yomba Island, Bilia, Riwo, and Sek, along with the Madang coastal strip. To look for earlier ceramic deposits along the Madang coast, one would need to excavate inland, on the hills and former beachlines behind modern Bilbil village, away from the recent coral uplifts, which date predominantly to within the last 500 years, and seemingly no more than 3000 years at the most (Morgan et al. 2005). Research on Arop/Long Island, Crown, and Bagabag Islands would fill in another piece of the puzzle, investigating linkages between Madang and the Vitiaz Strait. Because these isFrom a methodological and theoretical perspective, this lands lie closest to Hankow Reef, the possible location of volume has attempted a new approach to Pacific ceramic Yomba Island, they would be the best places to investigate analysis. It provides an alternative approach to style, re- Yomba Period Madang style ceramics. Further, coverage of centring technological process rather than morphological Mindiri and the Rai Coast would investigate a crucial link typology. This allows for a systematic delineation of past between Madang, Sio and the Vitiaz network. production groups, working within broader communities of practice. Untangling how the processes of pottery The Tilu assemblage is basically identical to the Nunguri production and distribution operated will aid archaeo- assemblage in material terms, containing domesticated logical interpretations that suggest forming technology is animal bones, shellfish, pottery, obsidian and sedimentary non-conservative and easily transmitted across large geo- stone tools, and shell adzes and ornaments, which might graphic distances, between non-related groups. be described as a quintessentially ‘Austronesian-style’ assemblage. However, this observation gets to the very heart The Madang results have also shown that individual tech- of whether material remains map to ‘culture’ effectively, so nical elements or very short syntaxes, such as decorations, that archaeologists can assign ethno-linguistic groups to are prone to change and disappearance over short periods, sites and assemblages. Without investigating sites in the perhaps only a few hundred years. Minor technical form- area known to be occupied by non-Bel speakers we caning elements, such as inner-rim paddling on Class 3 vessels not test these assumptions. With regard to anthropological can also arise in similar time frames, whether through in- concerns about learning, language, and communities of ternal innovations by skilled potters or introductions from practice, these theoretical and methodological assumpother potting groups. It has been important to approach tions need to be further explored to refine our understandthe ceramic assemblages from a technological perspec- ing of the human past in New Guinea, especially in the tive, as on the surface the Madang style appears similar to time leading up to ethnographic contact, but certainly many of the globular pots that populate the New Guinea with implications for Lapita and post-Lapita settlements coast, as Otto Finsch observed over 100 years ago (Finsch around the Pacific generally. 1888, translated in Mennis 1996: 29). Endnote Based on the first systematic archaeological investigations around Madang, dealing thoroughly with two ceramic se- As presented at the start of this book, following Tim Inquences from the coastal area, the monograph has also gold’s (2011) advocation, we must follow the materials, presented archaeological narratives that describe recent learn the movements, and draw the lines. That is, in the Bel culture history, and the development of their pottery study of material culture we must examine the process of production and trade. To further test the narratives pro- making, rather than exclusively the made objects. These duced from these assemblages, the study would need to lines of technological process were not severed at Eurobe expanded, to analyse Madang style sherds excavated pean contact. They continue today. 256       · .  In modern Bel culture, many aspects of the ‘traditional’ are re-enactments of pre-colonial life as opposed to unbroken continuations of these technological processes: from dance and music, to settlement patterns and canoe making. In recent years there has been a deliberate attempt to re-establish collective Bel identity following German colonial disruptions and WWII and, later, fuelled by Papua New Guinea’s independence (Suwa 2005). It is not simply the finished objects that create collective identity but these vital technological processes, created at the interface of minds, bodies, and materials, which unify producers, distributors, and consumers in modern communities of practice. ally produce bodi well into her nineties. But pottery making has changed over the past century. Far from being a transgression on ‘traditional’ technological processes, recent Madang style tourist pottery reflects local, creative responses to new opportunities (Fig. 11.3). These pots carry the message across the globe that the community exists, that it is different, and can proudly and uniquely produce objects that not only reflect the vibrancy of contemporary kastom,3 but also of ancestral traditions, making brilliant red, globular clay pots as the tumbuna4 did in the past. Today, only Bilbil village remains to continue the legacy of the Madang style into the future; maintaining potting processes, which have been reimagined and recreated over the past 650 years or more. Pottery making alone has continued broadly unbroken and has been resistant to many concurrent social changes. Yeyeg of Yabob-up-top was considered a master of the 3 Kastom (Tok Pisin) can roughly be translated as tradition and practice, but there are subtle differences between this craft and potted for most of her life. Her long-established term and the English ‘custom,’ as shown by Keesing (1982, practice of paddle and anvil techniques outlasted modern 1993) and Lindstrom (2008). introductions of the potter’s wheel that have come and gone during her lifetime, and she continued to occasion- 4 Tumbuna (Tok Pisin) are the ancestors. Figure 11.3. Modern Bilbil pot designed for tourists. Inscribed: Bilbil village, Madang Province, Papua New Guinea (Personal collection, gifted to research team; photograph: Les O’Neill 2016). 257 Bibliography Bibliography Abdi, H. and Valentin, D., 2007. Multiple correspondence analysis. In Salkind, N. (ed.), Encyclopedia of Measurement and Statistics. pp. 651–657. Abramson, J., 1969. A preliminary archaeological survey of the Tami Islands, Territory of New Guinea. Archaeology and Physical Anthropology in Oceania, 4 (2): 83–90. Adams, W.Y., 1988. Archaeological classification: theory versus practice. Antiquity, 62 (234): 40–56. Adams, W.Y. and Adams, E.W., 2007. Archaeological Typology and Practical Reality: a Dialectical Approach to Artifact Classification and Sorting. Cambridge: Cambridge University Press. Agresti, A., 2002. Categorical Data Analysis, 2nd Ed. New York: Wiley. Allen, J. 1971. A pottery collection from the Madang District T.P.N.G. Records of the Papua New Guinea Museum, 1 (2): 1–5. Allen, J., 1982. Pre-contact trade in Papua New Guinea. In May, R.J. and Nelson, H. (eds.), Melanesia: Beyond Diversity, v.1. Canberra: Research School of Pacific Studies, Australian National University. pp. 193–205. Allen, J., 1984a. In search of the Lapita Homeland: reconstructing the prehistory of the Bismarck Archipelago. The Journal of Pacific History, 19 (4): 186–201. Allen, J., 1984b. Pots and poor princes: a multidimensional approach to the role of pottery trading in coastal Papua. In van der Leeuw, S., and Pritchard, A. (eds.), The Many Dimensions of Pottery: Ceramics in Archaeology and Anthropology. Amsterdam: University of Amsterdam. pp. 406–73. Allen, J., 1985. Comments on complexity and trade: a view from Melanesia. Archaeology in Oceania, 20: 49–57. Allen, J., 2016. Rephrasing Sue Bulmer’s six-style Port Moresby pottery sequence. Archaeology in Oceania, 51 (S1): 33–40. Allen, J., 2017. Excavations on Motupore Island, Central District, Papua New Guinea. University of Otago, Working Papers in Anthropology 4. Dunedin: Department of Anthropology and Archaeology, University of Otago. Allen, J. and Rye, O.S., 1982. The importance of being earnest in archaeological investigations of prehistoric trade in Papua. In Dutton, T. (ed.), The Hiri in History: Further Aspects of Long Distance Motu Trade in Central Papua. Canberra: Australian National University, Development Studies Centre. pp. 99–115. Allen, J., Summerhayes, G.R., Mandui, H. and Leavesley, M., 2011. New data from Oposisi: implications for the Early Papuan Pottery phase. Journal of Pacific Archaeology, 2 (1): 69–81. Ambrose, W.R., 1992. Clays and sands in Melanesian pottery analysis. In Galipaud, J-C. (ed.), Poterie Lapita et Peuplement. Noumea: Orstom. pp. 169–176. Ambrose, W.R., 1993. Pottery raw materials: source recognition in the Manus Islands. In Spriggs, M. (ed.), A Community of Culture: The People and Prehistory of the Pacific. Canberra: Department of Prehistory, Research School of Pacific Studies, Australian National University. pp. 206–217. Ambrose, W., 2007. The implements of Lapita ceramic stamped ornamentation. In Bedford, S., Sand, S., Connaughton, S.P. (eds.), Oceanic Explorations: Lapita and Western Pacific Settlement, Terra Australis 26. Canberra: ANU E Press. pp. 213–222. Anson, D., 1983. Lapita Pottery of the Bismarck Archipelago and its Affinities. Unpublished PhD thesis. Sydney: University of Sydney. Anson, D., 1986. Lapita pottery of the Bismarck Archipelago and its affinities. Archaeology in Oceania, 21 (3): 157–165. Anson, D., 1999. Compositional analyses of dentate-stamped Lapita and nail-incised and applied relief pottery from Watom Island. In Galipaud, J-C. (ed.), Le Pacifique de 5000 à 2000 Avant le Présent : Suppléments à l’Histoire d’une Colonization/ The Pacific from 5000 to 2000 BP : Colonisation and Transformations. Paris: IRD. pp. 85–103. Anson, D., 2000. Reber-Rakival dentate-Stamped motifs: documentation and comparative implications. Journal of New Zealand Archaeology, 20: 95–118. Apel, J., 2008. Knowledge, know-how and raw material-the production of Late Neolithic flint daggers in Scandinavia. Journal of Archaeological Method and Theory, 15 (1): 91–111. Apel, J. and Darmark, K., 2007. Den flathuggna pilspetsens fylogeni. Mellansvenskt stenhantverk ur ett kulturevolutionistiskt perspektiv. In Stenback, N. (ed.), Stenålder i Uppland. Uppdragsarkeologi och eftertanke. Stockholm: Riksantikvarieämbetet. pp. 31–65. Arnold, D.E., 1971. Ethnomineralogy of Ticul, Yucatan potters: etics and emics. American Antiquity, 36 (1): 20–40. Arnold, D.E., 1974. Principles of paste analysis: a preliminary formulation. Journal of the Steward Anthropological Society, 6 (1): 33–47. Arnold, D.E., 1980. Localized exchange: an ethnoarchaeological perspective. In Fry, R.E. (ed.), Models and Methods in Regional Exchange. Washington: Society for American Archaeology. pp. 147–150. Arnold, D.E., 1983. Design structure and community organization in Quinua, Peru. In Washburn, D.K. (ed.), Structure and Cognition in Art. Cambridge: Cambridge University Press. pp. 56–73. Arnold, D.E., 1984. Social interaction and ceramic design: community-wide correlates in Quinua, Peru. In Rice, P. (ed.), Pots and Potters: Current Approaches in Ceramic Archaeology, Monograph XXIV Institute of Archaeology University of California, Los Angeles. Los Angeles: Cotsen Institute of Archaeology. pp. 133–161. Arnold, D.E., 1985. Ceramic Theory and Cultural Process. Cambridge: Cambridge University Press. Arnold, D.E., 1998. Andean ceramic technology: an ethnoarchaeological perspective. MASCA Research Papers in Science and Archaeology, 15: 353–367. Arnold, D.E., 2000. Does the standardization of ceramic pastes really mean specialization? Journal of Archaeological Method and Theory, 7 (4): 333–375. Arnold, D.E., 2015. Raw material selection, landscape, engagement, and paste recipes: insights from ethnoarchaeology. Paper presented at Matieres å Penser: raw materials acquisition and processing in early Neolithic pottery production, University of Namur, Belgium, 29 May, 2015. Arnold III, P.J., 1991. Dimensional standardization and production scale in Mesoamerican ceramics. Latin American Antiquity, 2 (4): 363–370. 258       · .  Aronson, M., Skibo, J.M. and Stark, M.T., 1994. Production and use technologies in Kalinga pottery. In Longacre, W.A. and Skibo, J.M. (eds.), Kalinga Ethnoarchaeology: Expanding Archaeological Method and Theory. Washington, DC: Smithsonian Institution Press. pp. 83–112. Aswani, S., Sheppard, P., 2003. The archaeology and ethnohistory of exchange in precolonial and colonial Roviana: gifts, commodities, and inalienable possessions. Current Anthropology, 44 (S5): S51-S78. Aufinger, P.A., 1939. Wetterzauber auf den Yabob-Inseln in Neuguinea. Anthropos, 34 (1): 277–291. Aufinger, P.A. 1942. Die Geheimsprachen auf den kleinen Inseln bei Madang in Neuguinea. Anthropos, 37/40 (4–6): 629–646. Aufinger, P.A., 1950. Mythological Fragments from the Small Islands near Madang (New Guinea). Anthropos, 45 (4–6): 779– 786. Bain, J.H.C., 1973. A Summary of the Main Structural Elements of Papua New Guinea. Canberra: Bureau of Mineral Resources. Baldwin, S.L., Fitzgerald, P.G. and Webb, L.E., 2012. Tectonics of the New Guinea region. Annual Review of Earth and Planetary Sciences, 40: 495–520. Bar-Yosef, O. and Van Peer, P., 2009. The chaîne opératoire approach in Middle Paleolithic archaeology. Current Anthropology, 50 (1): 103–131. Bartlett, M.S., 1937. Properties of sufficiency and statistical tests. Proceedings of the Royal Statistical Society, Series A 160: 268– 282. Baxter, M.J., 1994. Exploratory Multivariate Analysis in Archaeology. Edinburgh: Edinburgh University Press. Bayliss-Smith, T.P. and Hviding, E., 2015. Landesque capital as an alternative to food storage in Melanesia: irrigated taro terraces in New Georgia, Solomon Islands. Environmental Archaeology, 20 (4): 425–436. Beasley, H.G., 1936. Notes on red feather money from Santa Cruz Group New Hebrides. The Journal of the Royal Anthropological Institute of Great Britain and Ireland, 66: 379–391. Beaumont, P., O’Connor, S., Leclerc, M. and Aplin, K., 2019. Diversity in early New Guinea pottery traditions. Journal of Pacific Archaeology, 10 (1): 15–32. Bedford, S., 2006a. Pieces of the Vanuatu Puzzle: Archaeology of the North, South and Centre, Terra Australis Vol. 23. Canberra: ANU E Press. Bedford, S., 2006b. The Pacific’s earliest painted pottery: an added layer of intrigue to the Lapita debate and beyond. Antiquity, 80 (309): 544–557. Bedford, S. and Clark, G.R., 2001. The rise and rise of the incised and applied relief tradition: a review and reassessment. In Clark, G.R., Anderson, A. Vunidilo, T. (eds.), The Archaeology of Lapita Dispersal in Oceania: Papers from the Forth Lapita Conference, June 2000, Canberra, Australia. Canberra: Pandanus Books. Beier, T. and Mommsen, H., 1994. Modified Mahalanobis filters for grouping pottery by chemical composition. Archaeometry, 36 (2): 287–306. Benco, N.L., 1988. Morphological standardization-an approach to the study of craft specialization. In Kolb, C.C. and Lackey, L.M. (eds.), A Pot for all Reasons-Ceramic Ecology Revisited. Philadelphia: Temple University Press. pp. 57–72. Best, S., 2002. Lapita: A View From the East. Auckland: New Zealand Archaeological Association. Bickler, S.H., 1997. Early pottery exchange along the south coast of Papua New Guinea. Archaeology in Oceania, 32 (2): 151–162. Bickler, S.H., 1998. Eating Stone and Dying: Archaeological Survey on Woodlark Island, Milne Bay Province, Papua New Guinea. Unpublished PhD thesis. Charlottesville, VA: University of Virginia. Bieber, A.M. Jr., Brooks, D.W., Harbottle, G. and Sayre, E.V., 1976. Application of multivariate techniques to analytical data on Aegean ceramics, Archaeometry, 18: 59–74. Bird, J.R., Ambrose, W.R., Russell, L.H. and Scott, M.D., 1981. The Characterisation of Melanesian Obsidian Sources and Artefacts Using the Proton-Induced Gamma Ray Emission (PIGME) Technique. Report No. E/510. Lucas Heights, NSW: Australian Atomic Energy Commission. Biro, L., 1899. Nemet Uj-Guineai (Berlinhafeni) Neprajzi Gyujteseinek Leiro Jegyzeke. Budapest: Hungarian National Museum. Bishop, R.L. and Neff, H., 1989. Compositional data analysis in archaeology. In Allen, R.O. (ed.), Archaeological Chemistry IV. Washington: American Chemical Society. pp. 57–86. Bishop, R.L., Rands, R.L. and Holley, G.R., 1982. Ceramic compositional analysis in archaeological perspective. In Schiffer, M.B. (ed.), Advances in Archaeological Method and Theory, 5. New York: Academic Press, Inc. pp. 275–330. Blackwell, R.R., Buck, C.E. and Reimer, P.J., 2006. Important features of the new radiocarbon calibration curves. Quaternary Science Reviews 25: 408–413. Bloch, M., 1991. Language, anthropology and cognitive science. Man, 26 (2): 183–198. Blong, R.J. 1982. The Time of Darkness: Local Legends and Volcanic Reality in Papua New Guinea. Canberra: Australian National University Press. Blong, R.J., Pain, C.F. and McKee, C.O., 1982. Long Island, Papua New Guinea: aspects of land forms and tephrostratigraphy. Records of the Australian Museum 34 (7): 419–426. Blong, R., Fallon, S., Wood, R., McKee, C., Chen, K., Magill, C. and Barter, P., 2018. Significance and timing of the mid-17thcentury eruption of Long Island, Papua New Guinea. The Holocene, 28 (4): 529–544. Blust, R., 2013. The Austronesian Languages. Canberra: Asia-Pacific Linguistics. Bodrogi, T., 1959. New Guinean Style Provinces: The Style Province “Astrolabe Bay”. In Bolgar, T. and Bodrogi, L. (eds.), Opuscula Ethnologica Memoriae Ludovici Biró Sacra. Budapest: Akad, Kiadó. pp. 39–99. Bourdieu, P., 1979. La Distinction: Critique Sociale du Jugement. Paris: Editions de Minuit. Bourdieu, P., 1988. Homo Academicus. Stanford, CA: Stanford University Press. Bourdieu, P., 1998. The State of Nobility: Elite Schools in the Field of Power. Stanford: Stanford University Press. Brissaud, I., Houdayer, A., Jehanno, C. and Sabir, A., 1985. Pottery analysis using PIXE, neutron activation, scanning electron microscopy and X-ray fluorescence techniques. Journal of Radioanalytical and Nuclear Chemistry, 89 (2): 473–486. Bruno, P., Caselli, M., Curri, M.L., Genga, A., Striccoli, R. and Traini, A., 2000. Chemical characterisation of ancient pottery from south of Italy by Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES): statistical multivariate analysis of data. Analytica Chimica Acta, 410 (1): 193–202. 259 Bibliography Bulmer, S., 1977. Between the mountain and the plain: prehistoric settlement and environment in the Kaironk Valley. In Winslow, J.H. (ed.), The Melanesian Environment. Canberra: ANU Press. pp. 61–73 Bulmer, S., 1970. The historic and prehistoric pottery of New Guinea, with remarks on the external relationships of the pottery of the Port Moresby region. Paper delivered to ANZAAS, August 1970. Bulmer, S., 1971. Pottery variation and change in New Guinea and its implications for the reconstruction of prehistoric settlement. Unpublished typescript. Bulmer Collection. Bulmer, S., 1978. Prehistoric Culture Change in the Port Moresby Region. Unpublished PhD thesis. Port Moresby: University of Papua New Guinea. Bulmer, S., 1979. Prehistoric ecology and economy in the Port Moresby region. New Zealand Journal of Archaeology, 1: 5–27. Bulmer, S., 1982. West of Bootless Inlet: archaeological evidence for prehistoric trade in the Port Moresby area and the origins of the Hiri. In Dutton, T. (ed.), The Hiri in History: Further Aspects of Long Distance Motu Trade in Central Papua. Canberra: Australian National University, Development Studies Centre. pp. 117–130. Bulmer, S., 1985. Papuan pottery- an archaeological consideration. Indo-Pacific Prehistory Bulletin, 6: 123–132. Bulmer, S., 2007. Pottery styles at Wañelek, Papua New Guinea. In Anderson, A., Green, K. and Leach, F. (eds.), Vastly Ingenious: the Archaeology of Pacific Material Culture in Honour of Janet M. Davidson. Dunedin: University of Otago Press. pp. 277–290. Burton, J., 1999. The Yaganon People of the Rai Coast: Ethnography and social Mapping study. Report prepared for the Ramu Nickel Project, Highlands Pacific Ltd. Thursday Island, QLD, Australia. Cath-Garling, S., 2017. Evolutions or Revolutions? Interaction and Transformation at the ‘Transition’ in Island Melanesia. University of Otago Studies in Archaeology 27. Dunedin: Department of Anthropology & Archaeology, University of Otago. Cerny, C.A. and Kaiser, H.F., 1977. A study of a measure of sampling adequacy for factor-analytic correlation matrices. Multivariate Behavioral Research, 12 (1): 43–47. Chalmers, J., 1887. History and description of pottery trade. In Lindt, J.W. (ed.), Picturesque New Guinea. London: Longmans, Green, and Co. pp. 118–125. Chappell, J. 1974. Geology of coral terraces, Huon Peninsula, New Guinea: a study of Quaternary tectonic movements and sea-level changes. Geological Society of America Bulletin, 85 (4): 553–570. Chappell, J. and Polach, H. 1991. Post-glacial sea-level rise from a coral record at Huon Peninsula, Papua New Guinea. Nature, 349 (6305): 147–149. Chappell, J., Omura, A., Esat, T., McCulloch, M., Pandolfi, J., Ota, Y. and Pillans, B., 1996. Reconciliation of late Quaternary sea levels derived from coral terraces at Huon Peninsula with deep sea oxygen isotope records. Earth and Planetary Science Letters, 141 (1): 227–236. Chazan, M., 2009. Pattern and technology: why the chaîne opératoire matters. In Shea, J. and Lieberman, D. (eds.), Transitions in Prehistory: Essays in Honor of Ofer Bar-Yosef. Oxford: Oxbow. pp. 469–78. Chynoweth, M. 2015. Locating Lapita: Investigating the Origins of Pottery and Obsidian from Kasasinabwana Shell Midden in the Massim, Papua New Guinea. Unpublished BA Hons dissertation. Dunedin: University of Otago. Chynoweth, M., Summerhayes, G.R., Ford, A. and Negishi, Y., 2020. Lapita on Wari Island: what’s the problem? Asian Perspectives, 59 (1): 100–116. Childe, V.G., 1936. Man Makes Himself. London: Watts. Chilton, E., 1998. The cultural origin of technical choice: unraveling Algonquian and Iroquoian ceramic traditions in the Northeast. In Stark, M. (ed.), The Archaeology of Social Boundaries. Washington, DC: Smithsonian Institution Press. pp. 132–360. Chiu, S., 2005. Meanings of a Lapita face: materialized memory in ancient house societies. Taiwan Journal of Anthropology, 3 (1): 1–47. Chiu, S., 2007. Detailed analysis of Lapita face motifs: case studies from Reef/Santa Cruz Lapita sites and New Caledonia Lapita site 13A. In Bedford, S., Sand, S., Connaughton, S.P. (eds.), Oceanic Explorations: Lapita and Western Pacific Settlement, Terra Australis 26. Canberra: ANU E Press. pp. 241–264. Chiu, S., 2015. Where do we go from here? Social relatedness reflected by motif analysis. In Sand, C., Chiu, S. and Hogg, N. (eds.), The Lapita Cultural Complex in Time and Space: Expansion Routes, Chronologies and Typologies. Noumea: Institut d’archeologie de la Nouvelle-Caledonie et du Pacifique. pp. 185–206. Chiu, S. and Sand, C., 2005. Recording of the Lapita motifs: proposal for a complete recording method. Archaeology in New Zealand, 48 (2): 133–150. Chomsky, N., 1982. A note on the creative aspect of language use. The Philosophical Review, 91 (3): 423–434. Christensen, R. (ed.), 1975. Madang and Siassi: Traditional Art and Craft. Madang, PNG: Madang Teachers College. Christenson, A.L. and Read, D.W., 1977. Numerical taxonomy, R-mode factor analysis, and archaeological classification. American Antiquity 42 (2): 163–179. Clark, E. and Tsai, H.M., 2013. Islands: ecologically unequal exchange and landesque capital. In Hornborg, A., Clark, B., and Hermele, K. (eds.), Ecology and Power: Struggles Over Land and Material Resources in the Past, Present and Future. London: Routledge. pp. 52–67. Clark, G., 2003. Specialisation, standardisation and Lapita ceramics. In Bedford, S., Sand, S., Connaughton, S.P. (eds.), Oceanic Explorations: Lapita and Western Pacific Settlement, Terra Australis 26. Canberra: ANU E Press. pp. 289–299. Clark, G. and Murray, T., 2006. Decay characteristics of the eastern Lapita design system. Archaeology in Oceania, 41 (3): 107– 117. Clarke, A. and Torrence, R., 2011. Archaeology and the collection: tracing material relationships in colonial Papua from 1875 to 1925. Journal of Australian Studies, 35 (4): 433–448. Cochrane, E.E., 2002. Explaining the prehistory of ceramic technology on Waya Island, Fiji. Archaeology in Oceania, 37 (1): 37–50. Cochrane, E.E. and Lipo, C.P., 2010. Phylogenetic analyses of Lapita decoration do not support branching evolution or regional population structure during colonization of Remote Oceania. Philosophical Transactions of the Royal Society B: Biological Sciences, 365 (1559): 3889–3902. 260       · .  Cochrane, E.E. and Neff, H., 2006. Investigating compositional diversity among Fijian ceramics with laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS): implications for interaction studies on geologically similar islands. Journal of Archaeological Science, 33 (3): 378–390. Cochrane, E.E., Rieth, T.M. and Dickinson, W.R., 2013. Plainware ceramics from Sāmoa: Insights into ceramic chronology, cultural transmission, and selection among colonizing populations. Journal of Anthropological Archaeology, 32 (4): 499–510. Columbo Dougoud, R., 2005. How a Papua New Guinean artefact became traditional. In Otto, T. (ed.), Tradition and Agency: Tracing Cultural Continuity and Invention. Aarhus University Press. pp. 235–266. Conneller, C., 2008. Lithic technology and the chaîne opératoire. In Pollard, J. (ed.), Prehistoric Britain, Oxford: Blackwell. pp. 160–176. Cooper, M., 1971. Economic context of shell money production in Malaita. Oceania, 41 (4): 266–276. Counts, D., 1979. Taming the tiger: change and exchange in New Britain. In Force, R.W. and Bishop, B. (eds.), Persistence and Exchange: Papers from a Symposium on Ecological Problems of the Traditional Societies of the Pacific Region: XIV Pacific Science Congress, Khabarovsk, USSR, August-September, 1979. Honolulu, Hawaii: Pacific Science Association. pp. 51–58. Coutts, P.J.F., 1967. Pottery of Eastern New Guinea and Papua. Mankind, 6 (10): 482–488. Cresswell, R., 1976. Avant-propos. Techniques et culture, 1: 5–6. Crown, P.L., 2001. Learning to make pottery in the Prehispanic American Southwest. Journal of Anthropological Research, 57 (4): 451–469. Curtis, F., 1962. The utility pottery industry of Bailen, southern Spain. American Anthropologist, 64 (3): 486–503. D’Agostino, F., 1984. Chomsky on creativity. Synthese, 58 (1): 85–117. Daley, J., 1982. Design creativity and the understanding of objects. Design Studies 3 (3): 133–137. Damm, C., 2005. Archaeology, ethno-history and oral traditions: Approaches to the indigenous past. Norwegian Archaeological Review, 38 (2): 73–87. David, B., Geneste, J.M., Aplin, K., Delannoy, J.J., Araho, N., Clarkson, C., Connell, K., Haberle, S., Barker, B., Lamb, L. and Stanisic, J., 2010. The Emo site (OAC), Gulf Province, Papua New Guinea: resolving long-standing questions of antiquity and implications for the history of the ancestral hiri maritime trade. Australian Archaeology, 70 (1): 39–54. David, B., McNiven, I.J., Richards, T., Connaughton, S.P., Leavesley, M., Barker, B. and Rowe, C., 2011. Lapita sites in the central province of mainland Papua New Guinea. World Archaeology, 43 (4): 576–593. Davies, H.L. 2012. The geology of New Guinea- the cordilleran margin of the Australian continent. Episodes, 35 (1): 87–102. Day, S., 2012. Linking geological evidence of past tsunamis and volcanic eruptions with oral traditions of natural disasters from the Bismarck Sea Region, Papua New Guinea. Paper presented at the 34th International Geological Congress, Brisbane 5–10 August. Deer, W.A., Howie, R.A. and Zussman, J., 1992. An Introduction to the Rock-Forming Minerals. London: Longman. Dickinson, W.R., 1998. Petrographic temper provinces of prehistoric pottery in Oceania. Records of the Australian Museum 50 (3): 263–276. Dickinson, W.R., 2006. Temper sands in prehistoric Oceanian pottery: geotectonics, sedimentology, petrography, provenance. Geological Society of America Special Papers, 406, pp. 1–160. Dickinson, W.R., 2011. Petrography of coastal sands and prehistoric sherd tempers from the northwestern coast of Papua New Guinea. In Terrell, J., Schechter, E.M. (eds.), Exploring prehistory on the Sepik Coast of Papua New Guinea. Fieldiana 42. Chicago: Field Museum of Natural History. pp. 241– 249. Dickinson, W.R. and Shutler, R., 1971. Temper sands in prehistoric pottery of the Pacific Islands. Archaeology and Physical Anthropology in Oceania, 6 (3): 191–203. Dietler, M. and Herbich, I., 1998. Habitus, techniques, style: an integrated approach to the social understanding of material culture and boundaries. In Stark, M. (ed.), The Archaeology of Social Boundaries. Washington, DC: Smithsonian Institution Press. pp. 232–263. Dietler, M. and Herbich, I., 1989. Tich Matek: the technology of Luo pottery production and the definition of ceramic style. World Archaeology, 2 (1): 148–64. Dobres, M.A., 2000. Technology and Social Agency: Outlining a Practice Framework for Archaeology. Oxford: Wiley-Blackwell. Dobres, M.A., 2010. Archaeologies of technology. Cambridge Journal of Economics, 34 (1): 103–114. Dobres, M.A. and Hoffman, C.R., 1994. Social agency and the dynamics of prehistoric technology. Journal of Archaeological Method and Theory, 1 (3): 211–258. Dubes, R. and Jain, A.K., 1976. Clustering techniques: the user’s dilemma. Pattern Recognition, 8 (4): 247–260. Dubes, R. and Jain, A.K., 1979. Validity studies in clustering methodologies. Pattern Recognition, 11 (4): 235–254. Dunn, G. and Everitt, B., 1982. An Introduction to Mathematical Taxonomy. Cambridge: Cambridge University Press. Dutton, T. (ed.), 1982. The Hiri in History: Further Aspects of Long Distance Motu Trade in Central Papua. Canberra: Australian National University, Development Studies Centre. Dziuban, C.D. and Shirkey, E.C., 1974. When is a correlation matrix appropriate for factor analysis? Psychological Bulletin, 81: 358–361. Eckert, S.L., 2008 Pottery and Practice: the Expression of Identity at Pottery Mound and Hummingbird Pueblo. Albuquerque: University of New Mexico Press. Eckert, S.L. and James, W.D., 2011. Investigating the production and distribution of plain ware pottery in the Samoan archipelago with laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS). Journal of Archaeological Science, 38 (9): 2155–2170. Eckert, S.L., Schleher, K.L. and James, W.D., 2015. Communities of identity, communities of practice: understanding Santa Fe black-on-white pottery in the Española Basin of New Mexico. Journal of Archaeological Science, 63: 1–12. Edmonds, M., 1990. Description, understanding and the chaîne opératoire. Archaeological Review from Cambridge, 9 (1): 55– 70. Egloff, B.J., 1973. Report on fieldwork in the Madang District, 26 April, 1973 to 3 June, 1973. Port Moresby: Papua New Guinea Public Museum and Art Gallery. 261 Bibliography Egloff, B.J., 1975. Archaeological investigations in the coastal Madang area and on Eloaue Island of the St. Matthias Group. Papua New Guinea Museum and Art Gallery Records, 5: 15–31 Egloff, B.J., 1977. Pottery of Papua New Guinea: The National Collection. Port Moresby: Trustees of the Papua New Guinea National Museum and Art Gallery. Egloff, B., 1979. Recent Prehistory in Southeast Papua. Terra Australis 4. Canberra: Department of Prehistory, Research School of Pacific Studies, Australian National University. Egloff, B.J. and Specht, J., 1982. Long Island, Papua New Guinea: Aspects of the prehistory. Records of the Australian Museum, 34 (8): 427–446. Everingham, I.B., 1975. Seismological Report on the Madang Earthquake of 31 October 1970 and Aftershocks. Canberra: Australian Government Publishing Service. Everitt, B., 1977. Cluster Analysis. London: Heinemann Educational for the Social Science Research Council. Fewster, K.J., 2006. The potential of analogy in post-processual archaeologies: a case study from Basimane Ward, Serowe, Botswana. Journal of the Royal Anthropological Institute, 12 (1): 61–87. Fink D., Hotchkis, M., Hua, Q., Jacobsen, G., Smith, A.M., Zoppi, U., Child, D., Mifsud, C., van der Gaast, H., Williams, A., and Williams, M., 2004. The ANTARES AMS facility at ANSTO. NIM B, 223–224, 109–115. Finsch, O., 1888. Samoafahrten: Reisen in Kaiser Wilhelms-Land und Englisch-Neu-Guinea in Den Jahren 1884 u. 1885 an Bord des Deutschen Dampfers “Samoa” von Dr. Otto Finsch. Leipzig: Ferdinand Hirt & Sohn. Foley, R., 2000. The languages of New Guinea. Annual Review of Anthropology, 29: 357–404. Frankel, D., Thompson, K. and Vanderwal, R., 1994. Kerema and Kinomere. In Frankel, D. and Rhoads, J.W. (eds.), Archaeology of a Coastal Exchange System: Sites and Ceramics of the Papuan Gulf. Canberra: Australian National University. pp. 1–49. Friedrich, M.H., 1970. Design structure and social interaction: archaeological implications of an ethnographic analysis. American Antiquity, 35 (3): 332–343. Freeth, S.J., 1967. A chemical study of some Bronze Age pottery sherds. Archaeometry, 10 (1): 104–119. Froh, J., 2004. Archaeological ceramics studied by scanning electron microscopy. Hyperfine Interactions, 154 (1–4): 159–176. Gaffney, D., 2016. Materialising Ancestral Madang: Aspects of Production and Exchange along the Northeast Coast of New Guinea. Unpublished MA thesis. University of Otago. Gaffney, D. 2018. Maintenance and mutability amongst specialist potters on the northeast coast of New Guinea. Cambridge Archaeological Journal, 28 (2): 181–204. Gaffney, D. 2019. Technological process in Pre-colonial Melanesia. In Flexner, J. and Leclerc, M. (eds.), Archaeologies of Island Melanesia: Current Approaches to Landscapes, Exchange, and Practice. Terra Australis 51. ANU Press: Canberra. pp. 191–210. Gaffney, D, and Summerhayes, G.R., 2017. An Archaeology of Madang. University of Otago Working Papers in Anthropology, 5. Department of Anthropology and Archaeology, University of Otago: Dunedin, NZ. ISBN: 978-0-9922626-4-8. Gaffney, D., and Summerhayes, G.R. 2019. Coastal mobility and lithic supply lines in northeast New Guinea. Archaeological and Anthropological Sciences, 11 (6): 2849–2878. Gaffney, D., and Zmigrod, L. forthcoming. Historicising creativity: an interdisciplinary perspective between the social and natural sciences. In Porr, M., and Weidtman, N. (eds.) One World Anthropology and Beyond: a Multidisciplinary Engagement with the Work of Tim Ingold. Routledge: London. Gaffney, D., Summerhayes, G.R., Ford, A., Scott, J.M., Denham, T., Field, J. and Dickinson, W.R., 2015. Earliest pottery on New Guinea mainland reveals Austronesian influences in highland environments 3000 years ago, PloS one, 10 (9): p.e0134497. Gaffney, D., Ford, A. and Summerhayes, G.R., 2016. Sue Bulmer’s legacy in highland New Guinea: a re-examination of the Bulmer Collection and future directions. Archaeology in Oceania, 51 (S1): 23–32. Gaffney, D., Summerhayes, G.R., Mennis, M., Beni, T., Cook, A., Field, J., Jacobsen, G., Allen, F., Buckley, H., and Mandui, H. 2018a. Archaeological investigations into the origins of Bel trading groups around the Madang coast, northeast New Guinea. Journal of Island and Coastal Archaeology, 13 (4): 501–530. Gaffney, D., Summerhayes, G.R., Miamba, K., 2018b. Excavations on Karkar Island, February 2018. Unpublished field report. Gaffney, D., Summerhayes, G.R., and Miamba, K. 2018c. An archaeological survey of inland Madang, northeast Papua New Guinea. Archaeology in New Zealand 61 (3): 12–26. Gaffney, D., Summerhayes, G.R., and Mennis, M. 2019. A Lapita presence on Arop/Long Island, Vitiaz Strait, Papua New Guinea? In Bedford, S. and Spriggs, M. (eds.), Debating Lapita: Distribution, Chronology, Society, and Subsistence. Terra Australis 52. ANU Press: Canberra. Gaffney, D., Greig, K., Stoddart, D., Tromp, M., Luu, S., Field, J., Coster, A., and Summerhayes, G.R., 2020. Tropical foodways and exchange along the coastal margin of northeast New Guinea. Journal of Field Archaeology, 45: 498–511. Gallay, A., 2007.Les jarres de marriage décorées du delta intérieur du Niger (Mali): essai du delimitation archéologique d’un territoire ethnique. The Arkeotek Journal, 1 (1). Garling, S.J., 2007. Post-Lapita Evolutions or Revolutions? Interaction and Exchange in Island Melanesia: the View from the Tanga Islands. Unpublished PhD thesis: Canberra: Australian National University. Gell, A., 1992a. The technology of enchantment and the enchantment of technology. In Coote, J. and Shelton, A. (eds.), Anthropology, Art and Aesthetics. Oxford: Oxford University Press. pp. 40–63. Gell, A., 1992b. Inter-tribal commodity barter and reproductive gift-exchange in old Melanesia. In Humphrey, C. and HughJones, S. (eds.), Barter, Exchange and Value: an Anthropological Approach. Cambridge: Cambridge University Press. pp. 142–68. Gifford, J.C., 1960. The type-variety method of ceramic classification as an indicator of cultural phenomena. American Antiquity, 25 (3): 341–347. Golitko, M., 2011. Provenience investigations of ceramic and obsidian samples using laser ablation inductively coupled plasma mass spectrometry and portable X-Ray fluorescence. In Terrell, J., Schechter, E.M. (eds.), Exploring prehistory on the Sepik Coast of Papua New Guinea. Fieldiana 42. Chicago: Field Museum of Natural History. pp. 251–287. Golitko, M. and Terrell, J.E., 2012. Mapping prehistoric social 262       · .  fields on the Sepik coast of Papua New Guinea: ceramic compositional analysis using laser ablation-inductively coupled plasma-mass spectrometry. Journal of Archaeological Science, 39 (12): 3568–3580. Golitko, M., Dudgeon, J.V., Neff, H. and Terrell, J.E., 2012. Identification of post-depositional chemical alteration of ceramics from the north coast of Papua New Guinea (Sanduan Province) by Time-of-Flight–Laser Ablation–Inductively Coupled Plasma–Mass Spectrometry (ToF–LA–ICP–MS). Archaeometry, 54 (1): 80–100. Goodby, R.G., 1998. Technological patterning and social boundaries: ceramic variability in southern New England, AD 1000– 1675. In Stark, M. (ed.), The Archaeology of Social Boundaries. Washington, DC: Smithsonian Institution Press. pp. 161–182. Görlich, J., 1998a. Between war and peace: Gift exchange and commodity barter in the central and fringe highlands of Papua New Guinea. In Schweizer, T. and Douglas, W. (eds.), Kinship, Networks, and Exchange. Cambridge: Cambridge University Press. pp. 303–331. Görlich, J., 1998b. The construction of social meaning and material value: A note on trade in Melanesia. Oceania, 68 (4): 294– 301. Gosden, C., 1989. Prehistoric social landscapes of the Arawe Islands, West New Britain Province, Papua New Guinea. Archaeology in Oceania, 24 (2): 45–58. Gosden, C., 1994. Social Being and Time. Oxford: Basil Blackwell. Gosden, C., 2008. Social ontologies. Philosophical Transactions of the Royal Society B: Biological Sciences, 363 (1499): 2003–2010. Gosden, C., 2012. Magic, materials and matter: understanding different ontologies. In Maran, J., and Stockhammer, P.W. (eds.), Materiality and Social Practice: Transformative Capacities of Intercultural Encounters. Oxford: Oxbow Books. pp. 13–19. Gosden, C. and Malafouris, L., 2015. Process archaeology (PArch). World Archaeology, 47 (5): 701–717. Gosden, C. and Marshall, Y., 1999. The cultural biography of objects. World Archaeology, 31 (2): 169–178. Gosden, C. and Webb, J., 1994. The creation of a Papua New Guinean landscape: Archaeological and geomorphological evidence. Journal of Field Archaeology, 21 (1): 29–51. Gosselain, O.P., 1992. Technology and style: potters and pottery among Bafia of Cameroon. Man, 27 (3): 559–586. Gosselain, O., 2008. Mother Bella was not a Bella. Inherited and transformed traditions in Southwestern Niger. In Stark, M.T., Bowser, B.J. and Horne, L. (eds.), Cultural Transmission and Material Culture: Breaking Down Boundaries. Tucson, AZ: University of Arizona Press. pp. 150–177. Gorecki, P., Mabin, M. and Campbell, J., 1991. Archaeology and geomorphology of the Vanimo coast, Papua New Guinea: preliminary results. Archaeology in Oceania, 26 (3): 119–122. Graburn, N.H. (ed.), 1976. Ethnic and Tourist Arts: Cultural Expressions from the Fourth World. Berkeley: University of California Press. Graburn, N.H., 1979. New Directions in Contemporary Arts. Amsterdam: Amsterdam University Press. Graves, M.W., 1981. Ethnoarchaeology of Kalinga Ceramic Design. Unpublished PhD thesis. Tuscan: University of Arizona. Green, R.C., 1979. Early Lapita art from Polynesia and Island Melanesia: continuities in ceramic, barkcloth and tattoo decorations. In Mead, S.M. (ed.), Exploring the Visual Art of Oceania: Australia, Melanesia, Micronesia, Polynesia. Honolulu, Hawaii: University of Hawaii Press. pp. 13–31. Green, R.C., 1996. Prehistoric transfers of portable items during the Lapita horizon in Remote Oceania: a review. Bulletin of the Indo-Pacific Prehistory Association, 15: 119–130. Greenhill, S.J., Gray, R.D., 2009. Austronesian language phylogenies: myths and misconceptions about Bayesian computational methods. In Adelaar, A. and Pawley, A. (eds.), Austronesian Historical Linguistics and Culture History: a Festschrift for Robert Blust. Canberra: Pacific Linguistics. pp. 1–23. Gregory, C.A., 1982. Gifts and Commodities. London: Academic Press. Groube, L., Chappell, J., Muke, J. and Price, D., 1986. A 40,000 year-old human occupation site at Huon Peninsula, Papua New Guinea. Nature, 324: 453–455. Groves, M., 1960. Motu pottery. The Journal of the Polynesian Society, 69 (1): 2–22. Guggenheim, S. and Martin, R.T., 1995. Definition of clay and clay mineral: joint report of the AIPEA nomenclature and CMS nomenclature committees. Clays and Clay Minerals, 43 (2): 255–256. Haantjens, H.A., 1976. The soil families. In Robbins, R.G. (ed.), Lands of the Ramu-Madang Area, Papua New Guinea. Land Research Surveys No. 37. Melbourne: CSIRO. pp. 79–95. Haantjens, H.A., Reiner, E., Robbins, R.G., Saunders, J.C., 1976. Land systems. In Robbins, R.G. (ed.), Lands of the RamuMadang Area, Papua New Guinea. Land Research Surveys No. 37. Melbourne: CSIRO. pp. 18–57. Haberle S.G., 1998. Dating the evidence for agricultural change in the highlands of New Guinea: the last 2000 years. Australian Archaeology 47: 1–19. Hagen, B., 1899. Unter Den Papua’s: Beobachtungen Und Studien Uber Land Und Leute, Thier- Und Pflanzenwelt in KaiserWilhelmsland. Wiesbaden: C.W. Kreidel’s Verlag. Hannemann, E.F., 1939. Buk Tamolai Suteken. Madang, PNG: Lutheran Mission. Hannemann, E.F., 1944. Village Life and Social Change in Madang Society. Columbus, Ohio: Board of Foreign Missions, American Lutheran Church. Hannemann, E.F., 1969. Grass Roots Art of New Guinea. Sydney: Pacific Publications. Harbottle, G., 1982. Chemical characterization in archaeology. In Erikson, J.E. and Earle, T. (ed.), Contexts for Prehistoric Exchange. New York: Academic Press, Inc. pp. 13–51. Harding, T.G., 1967. Voyagers of the Vitiaz Strait: a Study of a New Guinea Trade System. Seattle: University of Washington Press. Haudricourt, A.G., 1968. La technologie culturelle. Essai de méthode. In Poirier, J. (ed.), Ethnologie Generale I. Paris: Gallimard. pp. 731–882. Heidegger, M., 1927 (1962). Being and Time. Trans. Macquarrie, J. and Robinson, E. San Francisco: Harper and Row. Hewlett, B.S., Fouts, H.N., Boyette, A.H. and Hewlett, B.L., 2011. Social learning among Congo Basin hunter–gatherers. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 366 (1567): 1168–1178. Hill, J.N. and Evans, R.K., 1972. A model for classification and typology. In Clarke, D.L., Models in Archaeology. London: Methuen. pp. 231–273. Hitchcock, R.K. and Bartram, L.E., 1998. Social boundaries, 263 Bibliography technical systems, and the use of space and technology in the Kalahari. In Stark, M. (ed.), The Archaeology of Social Boundaries. Washington, DC: Smithsonian Institution Press. pp. 12–49. Hogbin, H.I., 1947. Native trade around the Huon Gulf, northeastern New Guinea. The Journal of the Polynesian Society, 56 (3): 242–255. Hogbin, H.I. and Wedgwood, C., 1953. Local grouping in Melanesia. Oceania, 22: 58–76. Hogbin, H.I., 1964. Wogeo kinship terminology. Oceania, 34 (4): 308–309. Hogg, N., 2007. Settling Down: Mobility Patterns of Three MidLate Lapita Sites in the Anir Group, Papua New Guinea. Unpublished B.A. (Hons) dissertation. Dunedin: University of Otago. Hogg, N.W.S., 2011. Specialised Production of Early-Lapita Pottery: a Skill Analysis of Pottery from the Island of Emirau. Unpubliahed Master of Arts thesis. Dunedin: University of Otago. Hollenback, K.L. and Schiffer, M.B., 2010. Technology and material life. In Hicks, D. and Beaudry, M.C. (eds.), The Oxford Handbook of Material Culture Studies. Oxford: Oxford University Press. Hooley, B.A., 1971. The Austronesian languages of Morobe District, Papua New Guinea. Oceanic Linguistics, 10: 79–151. Hosfield, R., 2009. Modes of transmission and material culture patterns in craft skills. In Shennan, S. (ed.), Pattern and Process in Cultural Evolution. Berkeley: University of Canifornia Press. pp. 45–60. Hughes, I., 1977. New Guinea Stone Age Trade. The Geography and Ecology of Traffic in the Interior. Terra Australis 3. Canberra: Department of Prehistory, Research School of Pacific Studies, Australian National University. Hunt, T., 1989. Lapita Ceramic Exchange in the Mussau Islanda, Papua New Guinea. Unpublished PhD thesis. Washington: University of Washington. Huntley, D.L., 2006. From recipe to identity: exploring Zuni glaze ware communities of practice. In Habicht-Mauche, J.A, Eckert, S.L. and Huntley, D. (eds.), The Social Life of Pots: Glaze Wares and Cultural Dynamics in the Southwest, AD 1250– 1680. Tucson: University of Arizona Press. pp. 105–123. Ingold, T., 1993a. Tool-use, sociality and intelligence. In Gibson, K.R. and Ingold, T. (eds.), Tools, Language and Cognition in Human Evolution. Cambridge: Cambridge University Press. pp. 429–445. Ingold, T., 1993b. The temporality of the landscape. World archaeology, 25 (2): 152–174. Ingold, T., 2000. The Perception of the Environment: Essays on Livelihood, Dwelling and Skill. London: Routledge. Ingold, T., 2010. Bringing things back to life: creative entanglements in a world of materials. ESRC National Centre for Research Methods. Working Paper Series 05/10. Ingold, T., 2011. Introduction. In Ingold, T. (ed.), Redrawing Anthropology: Materials, Movements, Lines. Farnham: Ashgate Publishing, Ltd. Inoue, S., 2006. Embodied habitus. Theory, Culture & Society, 23 (2–3): 229–231. Irwin, G.J., 1977. The Emergence of Mailu as a Central Place in the Prehistory of Coastal Papua. Unpublished PhD thesis. Canberra: Australian National University. Irwin, G., 1985. The Emergence of Mailu: as a Central Place in Coastal Papuan Prehistory. Terra Australis 10. Canberra: Department of Prehistory, Research School of Pacific Studies, The Australian National University. Irwin, G. 1991. Themes in the prehistory of coastal Papua and the Massim. In Pawley, A. (ed.), Man and a Half: Essays in Pacific Anthropology and Ethnobiology in Honour of Ralph Bulmer. Auckland: The Polynesian Society. pp. 503–510. Irwin, G. and Holdaway, S., 1996. Colonisation, trade and exchange: from Papua to Lapita. In Davidson, J., Irwin, G., Leach, F., Pawley, A., Brown, D., Oceanic Culture History: Essays in Honour of Roger Green. Auckland: New Zealand Archaeological Association. pp. 225–235. Jebb, M.H.P., Lowry, J.K., 1995. Natural history of Madang Lagoon with an appendix to collecting localities. Records of the Australian Museum Supplement, 22: 1–24. Jeffra, C.D., 2015. Experimental approaches to archaeological ceramics: unifying disparate methodologies with the chaîne opératoire. Archaeological and Anthropological Sciences, 7 (1): 141–149. Johns, R.J., 1986. The instability of the tropical ecosystem in New Guinea. Blumea- Biodiversity, Evolution and Biogeography of Plants, 31 (2): 341–371. Johnson, R.W., 2013. Fire Mountains of the Islands: a History of Volcanic Eruptions and Disaster Management in Papua New Guinea and the Solomon Islands. Canberra: ANU E Press. Johnson, R.W., Taylor, G.A.M., Davies, R.A. and D’Addario, G.W., 1972. Geology and petrology of Quaternary volcanic islands off the north coast of New Guinea. Canberra: Bureau of Mineral Resources. Johnson, S.C., 1967. Hierarchical clustering schemes. Psychometrika, 2: 241–254. Jones, A., 2002. Archaeological Theory and Scientific Practice. Cambridge: Cambridge University Press. Jones, A., 2004. Archaeometry and materiality: materials-based analysis in theory and practice. Archaeometry, 46 (3): 327–338. Jones, M., 2011. Analysis of chronological parameters for the Aitape District pottery sequence. In Terrell, J., Schechter, E.M. (eds.), Exploring prehistory on the Sepik Coast of Papua New Guinea. Fieldiana 42. Chicago: Field Museum of Natural History. pp. 289–293. Jordan, P. and Shennan, S., 2003. Cultural transmission, language, and basketry traditions amongst the California Indians. Journal of Anthropological Archaeology, 22 (1): 42–74. Jørgensen, LB., 2013. Introduction to Part II: technology as practice. In Sørensen, M.L., and Rebay-Salisbury, K. (eds.), Embodied knowledge: Perspectives on Belief and Technology. Oxford: Oxbow. pp. 91–94. Kaiser, H.F., 1970. A second generation little jiffy. Psychometrika, 35 (4): 401–415. Kaiser, H.F., and Rice, J., 1974. Little jiffy, mark iv. Educational and Psychological Measurement, 34 (1): 111–117. Kakubayashi, F., 1978. ニューギニア・マダン周辺の土器作りとその経済的機能の研究. 民族學研究 / A study on the potterymaking and its economic function around Madang, Papua New Guinea, Minzokugaku-kenkyu: the Japanese Journal of Ethnology, 43 (2): 138–155. Kakubayashi, F. 1981. 生産と交易: ニューギニア・ライ海岸方面における伝統的経済活動の研究 / Production and trade: a study of traditional economic activities around the Rai Coast of 264       · .  New Guinea. Minzokugaku-kenkyu: the Japanese Journal of Ethnology, 46 (2): 173–188. Kaplan, S., 1976. Ethnological and biogeographical significance of pottery sherds from Nissan Island, Papua New Guinea. Fieldiana, 66 (3): 35–89. Keesing, R.M., 1982. Kastom in Melanesia: an overview. The Australian Journal of Anthropology, 13 (4): 304–308. Keesing, R.M., 1993, Kastom re-examined. Anthropological Forum, 6 (4): 587–596. Keighley, J., 1973. Some problems in the quantitative interpretation of ceramic data. In Renfrew. C. (ed.), The Explanation of Culture Change: Models in Prehistory, London. Pittsburgh: University of Pittsburgh Press. pp. 131–6. Keil, D., 1973. Pottery of the northern portion of the Bena Bena Census Division, Eastern Highlands District and of the NahuRawa Census Division, Madang District, with a note on the pottery manufacture in the Middle Ramu Valley, Madang District and notes on other sources of pottery in the Eastern Highlands District. Unpublished report, Northwestern University. Keller, C.M. and Keller, J.D., 1996. Cognition and Tool Use: The Blacksmith at Work. Cambridge: Cambridge University Press. Key, C.A., 1973. Pottery manufacturing techniques in Papua-New Guinea. Asian Perspectives, 14: 65–68. Kirch, P.V., 1991. Prehistoric exchange in western Melanesia. Annual Review of Anthropology, 20 (1): 141–165. Kirch, P.V. 1997. The Lapita Peoples. Oxford: Blackwell. Kirch, P.V., 2007. Hawaii as a model system for human ecodynamics. American Anthropologist, 109 (1): 8–26. Kirch, P.V. 2018. Voices on the wind, traces in the earth: Integrating oral narrative and archaeology in Polynesian history. Journal of the Polynesian Society, 127 (3): 275–306. Kirch, P.V. and Yen, D.E., 1982. Tikopia. The Prehistory and Ecology of a Polynesian Outlier. Bernice P. Bishop Museum Bulletin Honolulu, 238. Honolulu, Hawaii: Bishop Museum. Knappett, C., 2005a. Pottery. In Maschner, H.D.G. and Chippindale, C. (eds.), Handbook of Archaeological Methods. Lanham , MD: AltaMira Press. pp. 673–714. Knappett, C., 2005b. Thinking Through Material Culture: An Interdisciplinary Perspective. Philadelphia: University of Pennsylvania Press. Knappett, C., 2011. An Archaeology of Interaction: Network Perspectives on Material Culture and Society. Oxford: Oxford University Press. Kohring, S., 2013. Conceptual knowledge as technologically materialised: a case study of pottery production, consumption and community practice. In Sørensen, M.L., and RebaySalisbury, K. (eds.), Embodied Knowledge: Perspectives on Belief and Technology. Oxford: Oxbow. pp. 106–116. Kramer, C., 1985. Ceramic ethnoarchaeology. Annual Review of Anthropology, 14 (1): 77–102. Kramer, C., 1997. Pottery in Raiasthun: Ethnoarchaeology in Two Indian Cities. Washington, D.C: Smithsonian Institution Press. Kreiter, A., Riebe, D.J., Parkinson, W.A., Pető, Á., Tóth, M., Pánczél, P. and Bánffy, E., 2014. Unique in its chaîne opératoire, unique in its symbolism: undressing a figurine from the 6th Millennium BC Körös culture, Hungary. Journal of Archaeological Science, 44: 136–147. Kunst and Buchdruckerei (publishers), 1886. Catalog Der Ethnologischen Sammlung Der Neu Guinea Compagnie. Latour, B., 1993. Ethnography of a high-tech case. In Lemonnier, P. (ed.), Technological Choices: Transformation in Material Cultures Since the Neolithic. London: Routledge: 372–98. Latour, B. and Lemonnier, P., 1994. Introduction. In Latour, B. and Lemmonier, P. (eds.), De la préhistoire aux missiles balistiques: L’intelligence sociale des techniques. Paris: La Découverte. pp. 11–24. Lauer, P.K. 1970a. Pottery Traditions of the D’Entrecasteaux Islands of Papua. Unpublished PhD thesis. Canberra: Australian National University. Lauer, P.K., 1970b. Sailing with the Amphlett Islanders. The Journal of the Polynesian Society, 79 (4): 381–398. Lauer, P.K., 1972. A neglected aspect of New Guinea pottery technology: firing. Pottery in Australia, 11 (1): 7–16. Lauer, P.K. 1974. Pottery Traditions in the D’Entrecasteaux Island of Papua. Brisbane: Anthropology Museum: University of Queensland. Lave, J. and Wenger, E., 1991. Situated Learning: Legitimate Peripheral Participation. Cambridge: Cambridge University Press. Lawrence, P., 1956. Lutheran Mission influence on Madang societies. Oceania, 27 (2): 73–89. Lawrence, P., 1964. Road Belong Cargo: a Study of the Cargo Movement in the Southern Madang District, New Guinea. Manchester: Manchester University Press. Lê, S., Josse, J. and Husson, F., 2008. FactoMineR: an R package for multivariate analysis. Journal of statistical software, 25 (1): 1–18. Leach, J., 2003. Creative Land: Place and Procreation on the Rai Coast of Papua New Guinea. Oxford: Berghahn Books. Leach, J., 2012. Creativity as a social relation? In Biggs, S. (ed.), Remediating the Social, Electronic Literature as a Model for Creativity and Innovation in Practice. Edinburgh: ELMCIP. pp. 28–31. LeBlanc, K., 2016. A structural approach to ceramic design analysis: a pilot study of the “Eastern Lapita Province”. Journal of Archaeological Science: Reports, 6: 125–135. Lechtman, H., 1977. Style in technology: Some early thoughts. In Lechtman, H. and Merrill, R. (eds.), Material Culture: Styles, Organization, and Dynamics of Technology. St Paul: West Publishing Company. Leclerc, M., 2015. From Over the Sea or From the Island: Provenance Study of Ceramics from Vanuatu. Unpublished PhD thesis. Canberra: Australian national University. Lemonnier, P., 1983. L’étude des systèmes techniques, une urgence en technologie culturelle. Techniques & Culture, 54–55: 46–67. Lemonnier, P., 1986. The study of material culture today: toward an anthropology of technical systems. Journal of Anthropological Archaeology, 5 (2), pp. 147–186. Lemonnier, P., 1992. Elements for an Anthropology of Technology. Ann Arbor, Mich: Museum of Anthropology, University of Michigan. Lemonnier, P. (ed.), 1993. Technological Choices: Transformation in Material Cultures Since the Neolithic. London: Routledge. Lemonnier, P., 2013. Introduction. In Lemonnier, P. (ed.), Technological Choices: Transformation in Material Cultures Since the Neolithic, 2nd Edition. London: Routledge. pp. 1–35. Leroi-Gourhan, A., 1943. Evolution et Techniques I—L’Homme et la Matie`re. Paris: Albin Michel. 265 Bibliography Leroi-Gourhan, A., 1945. Evolution et Techniques II—Milieu et Techniques. , Paris: Albin Michel. Lévi-Strauss, C., 1973. Le champ de l’anthropologie. In LeviStrauss, C. (ed.), Anthropologie Structurale Deux. Chicago: University of Chicago Press. pp. 11–44. Lilley, I., 1986. Prehistoric Exchange in the Vitiaz Strait, Papua New Guinea. Unpublished PhD thesis. Canberra: Australian National University. Lilley, I., 1988a. Prehistoric Exchange Across the Vitiaz Straight, Papua New Guinea. Current Anthropology, 29 (3): 513–516. Lilley, I., 1988b. Type X: description and discussion of a prehistoric ceramic ware from northeastern Papua New Guinea. Bulletin of the Indo-Pacific Prehistory Association, 8: 90–100. Lilley, I., 1991. Lapita and Post-Lapita developments in the Vitiaz Straits-West New Britain Area. Bulletin of the Indo-Pacific Prehistory Association 11: 313–322. Lilley, I., 1999. Too good to be true? Post-Lapita scenarios for language and archaeology in West New Britain-North New Guinea. Bulletin of the Indo-Pacific Prehistory Association, 18: 25–34. Lilley, I., 2002. Lapita and Type Y pottery in the KLK site, Siassi, Papua New Guinea. In Bedford, S., Burley, D. and Sand, C. (eds.), Fifty Years in the Field, Papers in Honour of Richard Shutler. Auckland: New Zealand Archaeological Association. pp. 79–90. Lilley, I., 2004a. Trade and culture history across the Vitiaz Strait, Papua New Guinea: the emerging post-Lapita coastal sequence. Records of the Australian Museum, 29: 89–96. Lilley, I., 2004b. Diaspora and identity in archaeology: moving beyond the Black Atlantic. In Meskell, L. and Preucel, R.W. (eds.), A Companion to Social Archaeology. Oxford: Blackwell. pp. 287–312. Lilley, I., 2017. Melanesian maritime middlemen and pre-colonial glocalisation. In Hodos, T. (ed.), Routledge Handbook of Archaeology and Globalisation. London: Routledge. pp. 335–353. Lilley, I., and Specht, J., 2007. Revised dating of Type X pottery, Morobe Province. In Specht, J. and Attenbrow, V. (eds.), Archaeological Studies of the Middle and Late Holocene, Papua New Guinea, Technical Reports of the Australian Museum, 20: 217–226. Lindstrom, L., 2008, Melanesian kastom and its transformations. Anthropological Forum, 18 (2): 161–178. Linnekin, J., 1997. Contending approaches. In Denoon, D., Firth, S., Linnekin, Meleisea, M. and Nero, K. (eds.), The Cambridge History of the Pacific Islanders. Cambridge: Cambridge University Press. pp. 3–36. Loffler, E., 1977. Geomorphology of Papua New Guinea. Canberra: CSIRO. Loney, H.L., 2000. Society and technological control: a critical review of models of technological change in ceramic studies. American Antiquity, 65 (4): 646–668. Longacre, W.A., 1999. Standardization and specialization: what’s the link. In Skibo, J.M. and Feinman, G.M. (eds.), Pottery and People: a Dynamic Interaction, Salt Lake City: University of Utah Press. pp. 44–58. Longacre, W.A., 2008. Foreword: cultural transmissions and ceramic sociology. In Stark, M.T., Bowser, B.J. and Horne, L. (eds.), Cultural Transmission and Material Culture: Breaking Down Boundaries. Tucson, AZ: University of Arizona Press. pp. ix–xii. Longacre, W.A., Kvamme, K.L. and Kobayashi, M., 1988. Southwestern pottery standardization: an ethnoarchaeological view from the Philippines. The Kiva, 53 (2): 101–112. Mager, J.F., 1952. Gedaged-English Dictionary. Columbus, Ohio: Board of Foreign Missions of the American Lutheran Church. Mahias, M.C., 1993. Pottery techniques in India. In Lemonnier, P. (ed.), Technological Choices: Transformation in Material Cultures Since the Neolithic. London: Routledge. pp. 157–80. Malafouris, L., 2008. At the potter’s wheel: an argument for material agency. In Knappet, C and Malafouris, L. (eds.), Material agency: Towards a Non-Anthropocentric Approach. New York: Springer. pp. 19–36. Malafouris, L., 2010. Metaplasticity and the human becoming: principles of neuroarchaeology. Journal of Anthropological Sciences, 88 (4): 49–72. Malafouris, L., 2013. How Things Shape the Mind: a Theory of Material Engagement. Cambridge, MA: MIT Press. Malinowski, B., 1922. Argonauts of the Western Pacific: an Account of Native Enterprise and Adventure in the Archipelagos of Melanesian New Guinea. London: Routledge. Manly, B.F., 1986. Multivariate Statistics. London: Chapman and Hall. Martin, K., 2019. Big Men, ceremonial exchange and life-cycle events. In Hirsch, E., Rollason, W. (eds.), The Melanesian World. London: Routledge. pp. 375–388. Martinón-Torres, M., 2002. Chaîne opératoire: the concept and its applications within the study of technology. Gallaecia, 21: 29–44. Mason, R.J., 2000. Archaeology and native North American oral traditions. American Antiquity, 65 (2): 239–266. Masson, M.A. and Rosenswig, R.M., 2005. Production characteristics of postclassic Maya pottery from Caye Coco, northern Belize. Latin American Antiquity, 16 (4): 355–384. Mauss, M., 1925. The Gift. London: Cohen & West. Mauss, M., 1934. Les techniques du corps. Journal de Psychologie, 32 (3–4): 271–293. May, P. and Tuckson, M., 2000. The Traditional Pottery of Papua New Guinea. Honolulu, Hawaii: University of Hawaii Press. Mayor, A., 2005. Traditions céramiques et histoire du peuplement dans la Boucle du Niger (Mali) au temps des empires précoloniaux Unpublished PhD thesis. Geneva: University of Geneva. McGregor, H.V., Gagan, M.K., McCulloch, M.T., Hodge, E. and Mortimer, G., 2008. Mid-Holocene variability in the marine 14 C reservoir age for northern coastal Papua New Guinea. Quaternary Geochronology, 3 (3): 213–225. McSwain R., 1977. The Past and Future People: Tradition and Change on a New Guinea Island. Melbourne: Oxford University Press. Mead, S.M., 1975. The decorative system of the Lapita potters of Sigatoka, Fiji. In Mead, S.M., Birks, L., Birks, H. and Shaw, E. (eds.), The Lapita Pottery Style of Fiji and Its Associations. Wellington: Polynesian Society. pp. 19–43. Mennis, M.R., 1978. The existence of Yomba Island near Madang Fact or Fiction? Oral History, Institute of Papua New Guinea Studies, 6 (6). Mennis, M.R., 1980a. The first lalong canoe built for forty years, Bilbil village, Madang province. Oral History, Institute of Papua New Guinea Studies, viii(1). 266       · .  Mennis, M.R., 1980b. Oral testimonies from coastal Madang, part 1. Oral History, Institute of Papua New Guinea Studies, viii(10). Mennis, M.R., 1981a. Oral testimonies from coastal Madang, part 2. Oral History, Institute of Papua New Guinea Studies, ix(1). Mennis, M.R., 1981b. Oral testimonies from coastal Madang, part 3. Oral History, Institute of Papua New Guinea Studies, ix(2). Mennis, M.R., 1982. Austronesian Canoes of Astrolabe Bay. Unpublished MA thesis, University of Papua New Guinea: Port Moresby. Mennis, M.R. (ed.), 1996. Archipelago of Contented People? Madang (Freidrich-Wilhelmshafen) in 1884. Otto Finsch, Translated by C. Harding. Madang, PNG: Kristen Press. Mennis, M., 2006a. A Potted History of Madang: Traditional Culture and Change on the North Coast of New Guinea. Aspley, Australia: Lalong Enterprises. Mennis, M 2006b. Yomba Island (Hankow Reef), Atlantis of the South Pacific, fact or fiction? Unpublished report, Aspley, Australia. Mennis, M.R., 2011. Mariners of Madang and Austronesian Canoes of Astrolabe Bay, Papua New Guinea. Brisbane: Aboriginal and Torres Strait Islander Studies Unit, University of Queensland. Mennis, M.R., 2014. Sailing for Survival: a Comparative Report of the Trading Systems and Trading Canoes of the Bel People in the Madang Area and of the Motu People in the Port Moresby Area of Papua New Guinea. Dunedin: Department of Anthropology and Archaeology, University of Otago. Mennis, M.R. and Day, S., in prep. A critical examination of the oral tradition of Yomba Island, from the southern Bismarck Sea region of Papua New Guinea, using geological evidence. Mercer, P.M. 1979. Oral tradition in the Pacific: problems of interpretation. The Journal of Pacific History, 14 (3): 130–153. Miklouho-Maclay, N., 1885. List of plants in use by the natives of the Maclay-Coast, New Guinea. Proceedings of the Linnean Society of New South Wales 10 (3): 346–358. Minar, C.J., 2001. Motor skills and the learning process: the conservation of cordage final twist direction in communities of practice. Journal of Anthropological Research 57 (4): 381–405. Minar, C.J. and Crown, P.L., 2001. Learning and craft production: an introduction. Journal of Anthropological Research, 57 (4): 369–380. Molnar-Bagley, E., 1993. Lajos Biro and His Contributions to Early New Guinea Contact History. Unpublished Master of Arts thesis. Port Moresby: University of Papua New Guinea. Morgan, E., Day, S., Elemunop, J., Silver, E., Ward, S., and Hoffmann, G., 2005. Tsunami deposits related to volcanic island collapses in the southern Bismarck Sea. Poster presented at the AGU Fall Meeting, December. Morauta, L., 1974. Beyond the Village: Local Politics in Madang, Papua New Guinea. New York: Athlone Press. Naji, M., 2009. Gender and materiality in-the-making the manufacture of Sirwan femininities through weaving in southern Morocco. Journal of Material Culture, 14 (1): 47–73. Nunn, P.D. and Petchey, F., 2013. Bayesian re-evaluation of Lapita settlement in Fiji: radiocarbon analysis of the Lapita occupation at Bourewa and nearby sites on the Rove Peninsula, Viti Levu Island. Journal of Pacific Archaeology, 4: 21–34. Orton, C., Hughes, M. and Hughes, M., 2013. Pottery in Archaeology. Cambridge: Cambridge University Press. Pain, C.F., 1972. Characteristics and geomorphic effects of earthquake-initiated landslides in the Adelbert Range, Papua New Guinea. Engineering geology, 6 (4): 261–274. Parkinson, R., 1887. Im Bismarck-Archipel: Erlebnisse und Beobachtungen auf der Insel Neu-Pommern (Neu-Britannien). Leipzig: FA Brockhaus. Paris, R., Switzer, A.D., Belousova, M., Belousov, A., Ontowirjo, B., Whelley, P.L. and Ulvrova, M., 2014. Volcanic tsunami: a review of source mechanisms, past events and hazards in Southeast Asia (Indonesia, Philippines, Papua New Guinea). Natural Hazards, 70 (1): 447–470. Pawley, A., 2006. Madang Languages. In Brown, K. (ed.), Encyclopedia of Language and Linguistics (2nd edition). Amsterdam: Elsevier. pp. 429–432. Pawley, A. and Ross, M., 1993. Austronesian historical linguistics and culture history. Annual Review of Anthropology, 22: 425–459. Pelegrin, J., 1990. Prehistoric lithic technology: some aspects of research. Archaeological Review from Cambridge, 9 (1): 116– 125. Petchey, F. and Ulm, S., 2012. Marine reservoir variation in the Bismarck region: an evaluation of spatial and temporal change in ∆ R and R over the last 3000 years. Radiocarbon, 54: 45–58. Petchey, F., Phelan, M. and White, P.J., 2004. New ΔR for the southwest Pacific Ocean. Radiocarbon, 46 (2): 1005–1014. Petchey, F., Spriggs, M., Bedford, S., Valentin, F. & Buckley, H. 2014. Radiocarbon dating of burials from the Teouma Lapita cemetery, Efate, Vanuatu. Journal of Archaeological Science, 50: 227–42. Petir, A., Materem, D., Yapong, P., Mukarek, S., Okira, M. and Platts-Mills, T., 1997. Useful Plants of Salemben Village, Madang Province, Papua New Guinea. Madang, PNG: Kristen Press. Pétrequin, A.M. and Pétrequin, P., 1999. La poterie en NouvelleGuinée: savoir-faire et transmission des techniques. Journal de la Société des Océanistes, 108 (1): 71–101. Pétrequin, A.M. and Pétrequin, P., 2006. Objets de Pouvoir en Nouvelle-Guinée. Paris: Éditions de la Réunion des Musées Nationaux. Piaget, J., 1972. Intellectual evolution from adolescence to adulthood. Human Development, 15 (1): 1–12. Pigeot, N., 1990. Technical and social actors. Flintknapping specialists and apprentices at Magdalenian Etiolles. Archaeological Review from Cambridge, 9 (1): 126–141. Plog, S., 1995. Approaches to style. In Carr, C. and Neitzel, J.E. (eds.), Style, Society, and Person: Archaeological and Ethnological Perspectives. New York: Springer. pp. 369–387. Polach HA., 1981. Radiocarbon dating of Long Island and Tibito tephras. In Johnson R.W. (ed.), Cooke-Ravian Volume of Volcanological Papers, Geological Survey of Papua New Guinea Memoir 10. Port Moresby: Geological Survey of Papua New Guinea. pp. 108–113. Polanyi, M., 1966. The Tacit Dimension. Chicago: University of Chicago Press. Pollock, S., 1983. Style and information: an analysis of Susiana ceramics. Journal of Anthropological Archaeology, 2 (4): 354–390. Poulsen, J., 1987. Early Tongan Prehistory: the Laptia Period on Tongatapu and its Relationships. Terra Australis. Canberra: Department of Prehistory, Research School of Pacific 267 Bibliography Studies, Australian National University. Pycroft, A.T., 1935. Santa Cruz red feather-money—its manufacture and use. The Journal of the Polynesian Society, 44 (3): 173–183. Rathje, W.L., 1978. Melanesian and Australian exchange systems: a view from Mesoamerica. Mankind, 11 (3): 165–174. Read, D. W., 1989. Intuitive typology and automatic classification: divergence or full circle? Journal of Anthropological Archaeology, 8 (2): 158–188. Reed, S.J.B., 1975. Electron Microprobe Analysis. Cambridge: Cambridge University Press. Reed, S.J.B., 2005. Electron Microprobe Analysis and Scanning Electron Microscopy in Geology. Cambridge: Cambridge University Press. Reimer, P. J., Bard, E., Bayliss, A., Beck, J. W., Blackwell, P. G., Bronk Ramsey, C., Grootes, P. M., Guilderson, T. P., Haflidason, H., Hajdas, I., Hatt, C., Heaton, T. J., Hoffmann, D. L., Hogg, A. G., Hughen, K. A., Kaiser, K. F., Kromer, B., Manning, S. W., Niu, M., Reimer, R. W., Richards, D. A., Scott, E. M., Southon, J. R., Staff, R. A., Turney, C. S. M., & van der Plicht, J. 2013. IntCal13 and Marine13 Radiocarbon Age Calibration Curves 0–50,000 Years cal BP. Radiocarbon, 55 (4): 1869–1887. Reiner, E. and Mabbutt, J.A., 1976. Geomorphology. In Robbins, R.G. (ed.), Lands of the Ramu-Madang Area, Papua New Guinea. Land Research Surveys No. 37. Melbourne: CSIRO. pp. 71–78. Renfrew, C. and Sterud, G., 1969. Close-proximity analysis: a rapid method for the ordering of archaeological materials. American Antiquity, 34 (3): 265–277. Rhoads, J.W., 1980. Through a Glass Darkly: Present and Past Land Use of Papuan Sagopalm Users. Unpublished PhD thesis, Australian National University, Canberra. Rice, P.M., 1982. Pottery production, pottery classification, and the role of physicochemical analyses. In Olin, J.S. and Franklin, A.D. (ed.), Archaeological Ceramics. Washington, DC: Smithsonian Institution Press. pp. 47–56. Rice P.M., 1987. Pottery Analysis: a Sourcebook. Chicago: University of Chicago Press. Robbins, R.G., 1976. Summary description. In Robbins, R.G. (ed.). Lands of the Ramu-Madang Area, Papua New Guinea. Land Research Surveys No. 37. Melbourne: CSIRO. pp. 13–17. Robbins, J., and Akin, D., 1999. An introduction to Melanesian currencies - Agency, identity, and social reproduction. In Akin, D. and Robbins, J. (eds.), Money and Modernity: State and Local Currencies in Melanesia. Pittsburgh: University of Pittsburgh Press. pp. 1–40. Robinson, G.P., Jaques, A.L. and Brown, C.M., 1976. Madang, Papua New Guinea 1: 250,000 Geological Series-Map and Explanatory Notes. Canberra: Australian Government Publishing Service. Ross, M., 1988. Proto-Oceanic and the Austronesian Languages of Western Melanesia. Canberra: Pacific Linguistics. Ross, M., 2008. A history of metatypy in the Bel languages. Journal of Language Contact, 2: 149–164. Ross, M., 2009. Reconstructing the history of the Bel languages. Unpublished manuscript. Canberra: Research School of Pacific and Asian Studies, The Australian National University. Rouse, I., 1951. Areas and periods of culture in the Greater Antilles. Southwestern Journal of Anthropology, 7 (3): 248–265. Rouse, I., 1960. The classification of artifacts in archaeology. American Antiquity, 25 (3): 313–23. Rouse, I., 1965. Caribbean ceramics: a study in method and in theory. In Matson, F.R. (ed.), Ceramics and Man. Viking Fund Publications in Anthropology, 41. Chicago: Aldine Publishing Co. pp. 88–103. Roux, V., 1990. The psychological analysis of technical activities: a contribution to the study of craft specialisation. Archaeological Review from Cambridge, 9 (1): 142–153. Roux, V., 2003. A dynamic systems framework for studying technological change: application to the emergence of the potter’s wheel in the southern Levant. Journal of Archaeological Method and Theory, 10 (1): 1–30. Roux, V., 2007. Ethnoarchaeology: a non-historical science of reference necessary for interpreting the past. Journal of Archaeological Method and Theory, 14 (2): 153–178. Roux, V., 2011. Anthropological interpretation of ceramic assemblages: foundations and implementations of technological analysis. In Scarcella, S. (ed.), Archaeological Ceramics: a Review of Current Research. Oxford: Archaeopress. pp. 80–88. Roux, V. and Courty, M.A., 2005. Identifying social entities at a macro-regional level: Chalcolithic ceramics of South Levant as a case study. In Bosquet, D., Livingstone-Smith, A. and Martineau, R. (eds.), Pottery Making Processes: Reconstruction and Interpretation: Actes du XIVéme Congrés de l’UISPP. Université de Liége, Belgique, 2–8 Septembre 2001. pp. 201–214. Roux, V., Bril, B. and Dietrich, G., 1995. Skills and learning difficulties involved in stone knapping: the case of stone-bead knapping in Khambhat, India. World Archaeology, 27 (1): 63– 87. Rye, O.S., 1976. Keeping your temper under control: materials and the manufacture of Papuan pottery. Archaeology and Physical Anthropology in Oceania, 11 (2): 106–137. Rye, O.S., 1981. Pottery Technology: Principles and Reconstruction. Washington, DC: Taraxacum. Rye, O.J. and Allen, J., 1980. New approaches to Papuan pottery analysis. Journal de la Société des Océanistes, 36 (69): 305–314. Sackett, J.R., 1977. The meaning of style in archaeology: a general model. American Antiquity, 42 (3): 369–380. Sackett, J.R., 1986. Isochrestism and style: a clarification. Journal of Anthropological Archaeology, 5 (3): 266–277. Sackett, J.R., 1990. Style and ethnicity in archaeology: the case for isochrestism. In Conkey, M.W. and Hastorf, C.A. (eds.), The Uses of Style in Archaeology. Cambridge: Cambridge University Press. pp. 32–43 Sahlins, M., 1963. Poor man, rich man, big-man, chief: Political types in Melanesia and Polynesia. Comparative Studies in Society and History, 5 (3): 285–303. Sahlins, M., 1965. On the sociology of primitive exchange. In Banton, M. (ed.), The Relevance of Models for Social Anthropology. ASA Monographs 1. London: Tavistock. pp. 139–236. Sahlins, M., 1972. Stone Age Economics. London: Routledge. Salisbury, R.F., 1966. Politics and shell-money finance in New Britain. In Swartz, M.J., Turner, V.W., Tuden, A. (eds.), Political Anthropology. New Brunswick, NJ: Transaction Publishers. pp. 113–128. Sand, C., 2007. Looking at the big motifs: a typology of the central band decorations of the Lapita ceramic tradition of New Caledonia (southern Melanesia) and preliminary regional comparisons. In Bedford, S., Sand, C., and Connaughton, S.P. (eds.), Oceanic Explorations: Lapita and 268       · .  Western Pacific Settlement, Terra Australis 26. Canberra: ANU E Press. pp. 265–287. Sassaman, K.E. and Rudolphi, W., 2001. Communities of practice in the early pottery traditions of the American Southeast. Journal of Anthropological Research: 407–425. Saunders, J.C., 1976. Forest resources. In Robbins, R.G. (ed.), Lands of the Ramu-Madang Area, Papua New Guinea. Land Research Surveys No. 37. Melbourne: CSIRO. pp. 110–125.. Scheans, D.J., 1977. Filipino Market Potteries. Manila: National Museum of the Philippines. Schechter, E.M., 2011. Historic and modern pottery in the Aitape area. In Terrell, J., Schechter, E.M. (eds.), Exploring prehistory on the Sepik Coast of Papua New Guinea. Fieldiana 42. Chicago: Field Museum of Natural History. pp. 159–173. Schechter, E.M. and Terrell, J.E., 2009. The meaning and importance of the Lapita face motif. Archaeology in Oceania, 44 (2): 45–56. Scheldeman, G., 2011. Beyond A to B. In Ingold, T. (ed.), Redrawing Anthropology: Materials, Movements, Lines. Surrey, UK: Ashgate Publishing Ltd. Schiffer, M.B., 1995. Social theory and history in behavioral archaeology. In Skibo, J., Walker, W. and Nielsen, A. (eds.), Expanding Archaeology. Salt Lake City: University of Utah Press. pp. 22–35. Schlanger, N., 1990. Techniques as human action. Two perspectives. Archaeological Review from Cambridge, 9 (1): 18–26. Schlanger, N., 1994. Mindful technology: unleashing the chaîne opératoire for an archaeology of mind. In Renfrew, C. and Zubrow, E.B.W. (eds.), The Ancient Mind: Elements of Cognitive Archaeology. Cambridge: Cambridge University Press. pp. 143–151. Schmitz, C.A., 1955. Zur ethnographie der Huon-Halbinsel, Nordost-Neuguinea. Zeitschrift für Ethnologie, 2: 298–312. Schurig, M., 1930. Die Südseetöpferei. Leipzig: Druckerei der Werkgemeinschaft. Seligman, C.G., 1910. The Melanesians of British New Guinea. Cambridge: Cambridge University Press. Sellet, F., 1993. Chaîne opératoire; the concept and its applications. Lithic technology, 18 (1–2): 106–112. Sentinella, C.L., 1975. Mikloucho-Maclay: New Guinea Diaries, 1871–83. Mikouho-Maclay, trans. Sentinella, C.L. Madang, PNG: Kristen Press. Sharp, N.D., 1988. Style and substance: a reconsideration of the Lapita decorative system. In Kirch, P.V. (eds.), Archaeology of the Lapita Cultural Complex: a Critical Review. Washington: Thomas Burke Memorial Washington State Museum. pp. 61–81. Shaw, B., 2014. The Archaeology of Rossel Island, Massim, Papua New Guinea: Towards a Prehistory of the Louisiade Archipelago. Unpublished PhD thesis. Canberra: Australian National University. Shaw, B., 2016. The late prehistoric introduction of pottery to Rossel Island, Louisiade Archipelago, Papua New Guinea: insights into local social organisation and regional exchange in the Massim. Archaeology in Oceania, 51 (S1): 61–72. Shaw, B. and Dickinson, W.R., 2016. Excavation on Nimowa Island, Louisiade Archipelago, Papua New Guinea: insights into cultural practices and the development of exchange networks in the southern Massim region. The Journal of Island and Coastal Archaeology, 00: 1–30. Shaw, B., Leclerc, M., Dickinson, W., Spriggs, M. and Summerhayes, G.R., 2016. Identifying prehistoric trade networks in the Massim region, Papua New Guinea: evidence from petrographic and chemical compositional pottery analyses from Rossel and Nimowa Islands in the Louisiade Archipelago. Journal of Archaeological Science: Reports, 6: 518–535. Shaw, B., Coxe, S., Kewibu, V., Haro, J., Hull, E. and Hawkins, S., 2020. 2500-year cultural sequence in the Massim region of eastern Papua New Guinea reflects adaptive strategies to small islands and changing climate regimes since Lapita settlement. The Holocene. Shennan, S., 2002. Genes, Memes, and Human History: Darwinian Archaeology and Cultural Evolution. London: Thames & Hudson. Shennan, S. and Steele, J., 1999. Cultural learning in hominids: a behavioural ecological approach. In Box, H. and Gibson, K (eds.), Mammalian Social Learning, Symposia of the Zoological Society of London. Cambridge: Cambridge University Press. pp. 367–388. Shepard, A.O., 1963. Beginnings of Ceramic Industrialization: an Example from the Oaxaca Valley, Notes from a Ceramic Laboratory 2. Washington, DC: Carnegie Institution of Washington. Shepard, A.O. 1965. Ceramics for the Archaeologist. Washington: Carnegie Institute of Washington. Shepard, A.O., 1980. Ceramics for the Archaeologist. Washington: Carnegie Institute of Washington. Shilling, C., 2005. The Body in Culture, Technology and Society. London: Sage. Sillar, B. and Tite, M.S., 2000. The challenge of ‘technological choices’ for materials science approaches in archaeology. Archaeometry, 42 (1): 2–20. Silver, E., Day, S., Ward, S., Hoffmann, G., Llanes, P., Driscoll, N., Appelgate, B. and Saunders, S., 2009. Volcano collapse and tsunami generation in the Bismarck Volcanic Arc, Papua New Guinea. Journal of Volcanology and Geothermal Research, 186 (3): 210–222. Sinclair, J.P., 2005. Madang. Madang: Divine World University Press. Siorat, J.P., 1990. A technological analysis of Lapita pottery decoration. In Spriggs, M. (ed.), Lapita Design, Form and Composition, Proceedings of the Lapita Design Workshop, Canberra, Australia- December 1988. Canberra: Department of Prehistory, Research School of Pacific Studies, The Australian National University. pp. 59–82. Skelley, R., 2014. From Lapita to the Hiri: Archaeology of the Kouri Lowlands, Gulf of Papua, Papua New Guinea. Unpublished PhD thesis. Clayton: Monash University. Skelley, R., and David, B., 2017. Hiri: Archaeology of Long-distance Maritime Trade Along the South Coast of Papua New Guinea. Honolulu: University of Hawai’i Press. Skibo, J.M. and Schiffer, M., 2008. People and Things: a Behavioral Approach to Material Culture. New York: Springer. Sofaer, J.R., 2006. The Body as Material Culture: a Theoretical Osteoarchaeology Cambridge: Cambridge University Press. Sofaer, J. and Budden, S., 2013. Many hands make light work: potting and embodied knowledge at the Bronze Age tell in Százhalombatta, Hungary. In Sørensen, M.L., and RebaySalisbury, K. (eds.), Embodied Knowledge: Perspectives on Belief and Technology. Oxford: Oxbow. pp. 117–127. 269 Bibliography Sokal, R.R. and Rohlf, F.J., 1969. Bartlett’s test of homogeneity of variances. In Biometry: the Principles and Practice of Statistics in Biological Research. San Francisco: WH Freeman and Co. pp. 370–389 Sokal, R.R. and Sneath, P.H., 1973. Numerical Taxonomy: the Principles and Practice of Numerical Classification. San Francisco: Freeman and Co. Specht, J.R., 1969. Prehistoric and Modern Pottery Industries of Buka Island, TPNG. Unpublished PhD thesis. Canberra: Australian National University. Specht, J., Lilley, I. and Dickinson, W.R., 2006. Type X pottery, Morobe province, Papua New Guinea: Petrography and possible Micronesian relationships. Asian Perspectives, 45 (1): 24–47. Specht, J., Ball, E.E., Blong, R.J., Egloff, B.J., Hughes, I.M., McKee, C.O. and Pain, C.F., 1982. Long Island, Papua New Guinea: Introduction. Records of the Australian Museum, 34 (6): 407– 417. Spriggs, M., 1989. The dating of the Island Southeast Asian Neolithic: an attempt at chronometric hygiene and linguistic correlation. Antiquity, 63 (240): 587–613. Spriggs, M., 1995. Lapita culture and Austronesian prehistory in Oceania. In Bellwood, P., Fox, J. and Tryon, D. (eds.), The Austronesians: Historical and Comparative Perspectives. Canberra: ANU Press. Spriggs, M., 1997. The Island Melanesians. Cambridge, Mass: Blackwell. Spriggs, M., 2004. Is there life after Lapita, and do you remember the 60s? The post-Lapita sequences of the western Pacific. Records of the Australian Museum, 29: 139–144. Stanislawski, M.B. 1977. Ethnoarchaeology of Hopi and HopiTewa pottery making: styles of learning. In Ingersoll, D., Yellen, J.E. and Macdonald, W. (eds.), Experimental Archaeology. New York: Columbia University Press. pp. 378–408. Stark, M.T. (ed.), 1998a. The Archaeology of Social Boundaries. Washington, DC: Smithsonian Institution Press. Stark, M.T., 1998b. Technical choices and social boundaries in material culture patterning: an introduction. In Stark, M. (ed.), The Archaeology of Social Boundaries. Washington, DC: Smithsonian Institution Press. pp. 1–11. Stark, M.T. and Longacre, W.A., 1993. Kalinga Ceramics and New Technologies: An Ethnoarchaeological Perspective. In Kingery, W.D. (ed.), The Social and Cultural Contexts of New Ceramic Technologies. Waterville, Ohio: American Ceramics Society. pp. 1–32. Stark, M.T., Bishop, R.L. and Miksa, E., 2000. Ceramic technology and social boundaries: cultural practices in Kalinga clay selection and use. Journal of Archaeological Method and Theory, 7 (4): 295–331. Sternberg, R.J., 1983. Criteria for intellectual skills training. Educational Researcher, 12 (2): 6–26. Sternberg, R.J., 1999. Thinking Styles. Cambridge: Cambridge University Press. Sterne, J., 2003. Bourdieu, technique and technology. Cultural Studies, 17 (3–4): 367–389. Strathern, M., 1988. The Gender of the Gift: Problems with Women and Problems with Society in Melanesia. Berkeley: University of California Press. Strathern, M., 1992. Qualified value: the perspective of gift exchange. In Humphrey, C. and Hugh-Jones, S. (eds.), Barter, Exchange and Value: an Anthropological Approach. Cambridge: Cambridge University Press. pp. 169–91. Strathern, M., 1996. Cutting the network. Journal of the Royal Anthropological Institute, 2 (3): 517–535. Sutton, N.P., 2016. Pots on the Move? The Nature of an Early Papuan Pottery Assemblage from Oposisi, Yule Island, Papua New Guinea. Unpublished MA thesis. Dunedin: University of Otago. Sutton, N., Vilgalys, G., Summerhayes, G. and Ford, A., 2016. Revisiting the late prehistoric sequence of the Port Moresby region of Papua New Guinea: the continuing contribution of Susan Bulmer. Archaeology in Oceania, 51 (S1): 41–49. Summerhayes, G.R., 1987. Aspects of Melanesian Ceramics. Unpublished MA (Hons) thesis. Sydney: University of Sydney. Summerhayes, G.R., 1996. Interaction in Pacific Prehistory. Unpublished PhD thesis. Melbourne: La Trobe University. Summerhayes, G.R., 2000. Lapita Interaction. Terra Australis, 15. Canberra: Australian National University Press. Summerhayes, G.R., 2001a. Defining the chronology of Lapita in the Bismarck Archipelago. In Clark, G.R., Anderson, A.J. and Vunidilo, T. (eds.), The Archaeology of Lapita Dispersal in Oceania. Terra Australis 17. Canberra: Pandanus Books, Australian National University. pp. 25–38. Summerhayes, G.R. 2001b. Lapita in the far west: recent developments. Archaeology in Oceania, 36 (2), 53–63. Summerhayes, G.R., 2001c. Far western, western, and eastern Lapita: a re-evaluation. Asian Perspectives, 39 (1): 109–138. Summerhayes, G.R., 2007. Island Melanesian pasts: a view from archaeology. In Friedlaender, J (ed.), Genes, Language and Culture History in the Southwest Pacific. Oxford: Oxford University Press. pp. 10–35. Summerhayes, G.R., 2015. Archaeology of trade and exchange. In Wright, J.D. (ed.), International Encyclopedia of the Social & Behavioral Sciences. 2nd ed. Oxford: Elsevier. pp. 481–484. Summerhayes, G.R. and Allen, J., 2007. Lapita writ small? Revisiting the Austronesian colonisation of the Papuan south coast. In Bedford, S., Sand, C., and Connaughton, S.P. (eds.), Oceanic Explorations: Lapita and Western Pacific Settlement. Canberra: ANU E Press. Suwa, J.I., The abandonment of Yabob Island 1942–1975 and the memory of cultural continuity. In Evans, M. (ed.), Refereed Papers From the 1st International Small Island Cultures Conference, Kagoshima University Centre for the Pacific Islands, February 7th–10th 2005. Sydney: SICRI, Macquarie University. pp. 108–116. Swadling, P., 1981. Papua New Guinea’s Prehistory: an Introduction. Port Moresby: National Museum and Art Gallery of Papua New Guinea. Swadling, P., 1988. The Sepik-Ramu. Port Moresby: National Museum and Art Gallery of Papua New Guinea. Swadling, P., Chappell, J., Francis, G., Araho, N. and Ivuyo, B., 1989. A Late Quaternary inland sea and early pottery in Papua New Guinea. Archaeology in Oceania, 24 (3): 106–109. Sypher, H.E., Hummert, M.L., Williams, S.L. 1994. Social psychological aspects of the oral history interview. In McMahan, E.M. and Rogers, K.L (eds.) Interactive Oral History Interviewing. Routledge: New York. pp. 47–63. Takayama, J. and Shutler, R., 1978. Preliminary report of a pottery site on Fefan Island, Truk, central Caroline Islands. Archaeology and Physical Anthropology in Oceania, 13 (1): 1–9. 270       · .  Teele, B., 2011. Style vs. Substance: Understanding Prehistoric Samoan Pottery Production on the Island of ‘Upolu Using Stylistic and Chemical Techniques. Unpublished MA dissertation. Dunedin: University of Otago. Tehrani, J.J. and Collard, M., 2009. On the relationship between interindividual cultural transmission and population-level cultural diversity: a case study of weaving in Iranian tribal populations. Evolution and Human Behavior, 30 (4): 286–300. Terrell, J.E., 2010. Language and material culture on the Sepik Coast of Papua New Guinea: using social network analysis to simulate, graph, identify, and analyze social and cultural boundaries between communities. Journal of Island & Coastal Archaeology, 5 (1): 3–32. Terrell, J.E., 2011. Wooden platters and bowls in the ethnographic collections. In Terrell, J., Schechter, E.M. (eds.), Exploring Prehistory on the Sepik Coast of Papua New Guinea. Fieldiana 42. Chicago: Field Museum of Natural History. pp. 175– 195. Terrell, J.E. and Schechter, E.M., 2007. Deciphering the Lapita code: the Aitape ceramic sequence and late survival of the ‘Lapita Face’. Cambridge Archaeological Journal, 17 (01): 59–85. Terrell, J. and Schechter, E.M. (eds.), 2011a. Exploring prehistory on the Sepik Coast of Papua New Guinea. Fieldiana 42. Chicago: Field Museum of Natural History. Terrell, J.E. and Schechter, E.M., 2011b. Prehistoric pottery wares in the Aitape area. In Terrell, J., Schechter, E.M. (eds.), Exploring Prehistory on the Sepik Coast of Papua New Guinea. Fieldiana 42. Chicago: Field Museum of Natural History. pp. 87–157. Terrell, J. and Welsch, R.L., 1990. Trade networks, areal integration; and diversity along the north coast of New Guinea. Asian Perspectives, 29 (2): 155–165. Terrell, J.E. and Welsch, R.L., 1997. Lapita and the temporal geography of prehistory. Antiquity, 71 (273): 548–572. Terrell, J.E., Schechter, E.M. and Golitko, M., 2011. Conclusions. In Terrell, J., Schechter, E.M. (eds.), Exploring Prehistory on the Sepik Coast of Papua New Guinea. Fieldiana 42. Chicago: Field Museum of Natural History. pp. 295–303. Thomas, N., 1991. Entangled Objects: Exchange, Colonialism and Material Culture in the Pacific. Cambridge MA: Harvard University Press. Thomas, T., 2009. Communities of practice in the archaeological record of New Georgia, Rendova and Tetepare. In Sheppard, P., Thomas, T., and Summerhayes, G.R. (eds.), Lapita: Ancestors and Descendants. Auckland: New Zealand Archaeological Association. pp. 119–145. Thomas, T., 2019. Axes of entanglement in the New Georgia group, Solomon Islands. In Flexner, J. and Leclerc, M. (eds.) Archaeologies of Island Melanesia: Current Approaches to Landscapes, Exchange, and Practice. Terra Australis 51. Canberra: ANU Press. pp. 103–116. Thompson, P., 2017. The Voice of the Past: Oral History, 4th ed. Oxford: Oxford University Press. Thornton, I.W.B., 2001. Colonization of an island volcano, Long Island, Papua New Guinea, and an emergent island, Motmot, in its caldera lake. I. general introduction. Journal of Biogeography, 28 (12): 1299–1310. Tite, M.S., 1999. Pottery production, distribution, and consumption—the contribution of the physical sciences. Journal of Archaeological Method and Theory, 6 (3): 181–233. Tite, M.S., 2008. Ceramic production, provenance and use- a review. Archaeometry, 50 (2): 216–231. Tuckson, M., 1966. Pottery in New Guinea. Pottery in Australia, 5 (1): 9–16. Tudhope, A.W., Buddemeier, R.W., Chilcott, C.P., Berryman, K.R., Fautin, D.G., Jebb, M., Lipps, J.H., Pearce, R.G., Scoffin, T.P. and Shimmield, G.P., 2000 Alternating seismic uplift and subsidence in the late Holocene at Madang, Papua New Guinea: evidence from raised reefs. Journal of Geophysical Research, Solid Earth, 105 (B6): 13797–13807. Van der Leeuw, S., 1993. Giving the potter a choice. In Lemonnier, P. (ed.), Technological Choices: Transformation in Material Cultures Since the Neolithic. London: Routledge. pp. 238–288. Urwin, C., Hua, Q., Skelly, R.J. and Arifeae, H., 2018. The chronology of Popo, an ancestral village site in Orokolo Bay, Gulf Province, Papua New Guinea. Australian Archaeology, 84 (1): 90–97. Van Keuren, S., 2006. Decorating glaze-painted pottery in eastcentral Arizona. In Habicht-Mauche, J.A, Eckert, S.L. and Huntley, D. (eds.), The Social Life of Pots: Glaze Wares and Cultural Dynamics in the Southwest, AD 1250–1680. Tucson: University of Arizona Press. pp. 1250–1680. Vanderwal, R.L., 1973. Prehistoric Studies in Coastal Papua. Unpublished PhD thesis. Canberra: Australian National University. Velde, B.B. and Meunier, A., 2008. The Origin of Clay Minerals in Soils and Weathered Rocks. New York: Springer. Vilgalys, G. and Summerhayes, G.R., 2016. Do hiccups echo? Late Holocene interaction and ceramic production in southern Papua New Guinea. Asian Perspectives, 55 (1): 61–88. Vincent, B.A., 1988. Prehistoric Ceramics of Northeast Thailand with Special Reference to Ban Na Di. Oxford: BAR. Wallaert, H., 2013. Apprenticeship and the confirmation of social boundaries. In Wendrich, W. (ed.), Archaeology and Apprenticeship: Body Knowledge, Identity, and Communities of Practice. Tuscan: University of Arizona Press. pp. 20–42. Walls, M., 2016. Making as a didactic process: Situated cognition and the chaîne opératoire. Quaternary International, 405 (A): 21–30. Walter, R. and Green, R.C., 2011. Su’ena: Five Hundred Years of Interaction in the Eastern Triangle, Solomon Islands. University of Otago Studies in Prehistoric Anthropology, No 23. Dunedin: Department of Anthropology: University of Otago. Ward, J.H. Jr, 1963. Hierarchical grouping to optimize an objective function. Journal of the American Statistical Association, 58 (301): 236–244. Ward, S.N., and Day, S., 2003. Ritter Island volcano- lateral collapse and the tsunami of 1888. Geophysical Journal International, 154 (3), 891–902. Weiner, A.B., 1992. Inalienable Possessions: The Paradox of Keeping-While Giving. Berkeley: University of California Press. Welsch, R.L. and Terrell, J.E., 1998. Material culture, social fields, and social boundaries on the Sepik coast of New Guinea. In Stark, M. (ed.), The Archaeology of Social Boundaries. Washington, DC: Smithsonian Institution Press. pp. 50–77. Welsch, R.L., Terrell, J. and Nadolski, J.A., 1992. Language and culture on the north coast of New Guinea. American Anthropologist, 94 (3): 568–600. Wendrich, W., 2013. Archaeology and apprenticeship: body knowledge, identity, and communities of practice. In 271 Bibliography Wendrich, W. (ed.), Archaeology and Apprenticeship: Body Knowledge, Identity, and Communities of Practice. Tuscan: University of Arizona Press. pp. 1–19. Wenger, E., 1998. Communities of Practice: Learning, Meaning, and Identity. Cambridge: Cambridge university press. White, J.P., 1972. Ol Tumbuna: Archaeological Excavations in the Eastern Central Highlands, Papua New Guinea. Terra Australis 2. Canberra: Department of Prehistory, Research School of Pacific Studies, Australian National University. White, P., 2014. New Guinea. In Cochrane, E. and Hunt, T. (eds.), The Oxford Handbook of Prehistoric Oceania. Oxford: Oxford University Press. Whittaker, J.C., Caulkins, D. and Kamp, K.A., 1998. Evaluating consistency in typology and classification. Journal of Archaeological Method and Theory, 5 (2): 129–164. Wickler, S., 2001. The Prehistory of Buka: Stepping Stone Island in the Northern Solomons, Terra Australis, 16. Canberra: Australian National University Press. Widgren, M. and Håkansson, N.T., 2016. Landesque capital: what is the concept good for? In Håkansson, N.T. and Widgren, M. (eds.), Landesque Capital: The Historical Ecology of Enduring Landscape Modifications. London: Routledge. pp. 9–30. Willey, G.R., 1945. Horizon styles and pottery traditions in Peruvian archaeology. American Antiquity, 11 (1): 49–56. Wilson, A.L., 1978. Elemental analysis of pottery in the study of its provenance: a review. Journal of Archaeological Science, 5 (3): 219–236. Wobst, H.M., 1977. Stylistic behavior and information exchange. In Cleland, C.E. (ed.), For the Director: Research Essays in Honor of James B. Griffin. Ann Arbor, MI: Museum of Anthropology, University of Michigan. pp. 317–342. Woodford, C.M., 1908. Notes on the manufacture of the Malaita shell bead money of the Solomon group. Man, 8: 81–84. Wurm, S.A., 1983. Linguistic Prehistory in the New Guinea Area. Journal of Human Evolution, 12 (1): 25–35. Z’Graggen, J.A., 1971. Classificatory and typological studies in languages of the Madang District. Pacific Linguistics Series C, 19. Canberra: Australian National University. Z’Graggen, J.A., 1975. The Languages of the Madang District, Papua New Guinea. Canberra: Department of Linguistics, Research School of Pacific Studies, Australian National University. Z’Graggen, J.A., 1976, Austronesian languages: Madang Province. In Wurm, S. (ed.), New Guinea Area Languages and Language Study, Vol. 2: Austronesian Languages, Canberra: Pacific Linguistics. pp. 285–300. Zuluaga, M.C., Alonso-Olazabal, A., Murelaga, X. and Ortega, L.A., 2011. A comparison of scanning electron microscopy energy dispersive X-ray (SEM/EDX) and inductively coupled plasma optical emission spectrometry (ICP-OES) for provenance inferences of grog-tempered Bronze Age pottery. Microchemical Journal, 99 (2): 443–448. 272

RELATED PAPERS

RELATED TOPICS

Log In

Materialising Ancestral Madang: Pottery Production and Subsistence Trading on the Northeast Coast of New Guinea

Materialising Ancestral Madang: Pottery Production and Subsistence Trading on the Northeast Coast of New Guinea

Related Papers

RELATED PAPERS

RELATED TOPICS