Avaldsnes - A Sea-Kings' Manor in First-Millennium Western Scandinavia 9783110421088, 9783110425789

Table of contents :
Preface
Table of content
Abbreviations
Section A. Scholarly Background
1. Rethinking Avaldsnes and Kormt
2. Exploring Avaldsnes 1540–2005
3. Avaldsnes and Kormt in Old Norse Written Sources
4. The Avaldsnes Royal Manor Project’s Research Plan and Excavation Objectives
Section B. Excavation Results 2011–12
5 Excavations and Surveys 1985-2012
6. Site Periods and Key Contexts
7. The Prehistoric Settlement and Buildings
8 Prehistoric Agriculture
9. The Production Area
10. Two Iron Age Boathouses
11. A Late Iron Age Palisade Facing the Karmsund Strait
12. Grave Monuments at Avaldsnes
13. Political and Ritual Aspects of Cooking Pits
14. The High Medieval Royal Manor
15. The Post-Medieval Rectory
Section C Scientific Analyses 2011–12
16. Geophysical Surveys
17. Microstratigraphy (Soil Micromorphology and Microchemistry, Soil Chemistry, and Magnetic Susceptibility)
18. Geochemical analysis using portable X-ray fluorescence
19 Biological Remains
Section D Specialist Studies
20. Artefacts from the 2011–12 Excavations
21. Migration Period Pottery from Avaldsnes: A Study of Shards from Bucket-shaped Pots
22. The Flaghaug Burials
23. The Raised Stones
24. Avaldsnes, Kormt and Rogaland. A Toponymy and Landscape Survey
25. Depositional Traditions in Iron Age Kormt
26. Emerging Kingship in the 8th Century? New Datings of three Courtyard Sites in Rogaland
Section E Avaldsnes: a Sea-Kings’ Manor
27. Aristocratic Presence along the Karmsund Strait 2000 BC–AD 1368
28. The Warrior Manor
29. Sea Kings on the Norðvegr
References Appendices
References
Appendix I: The ARM Project Council, Advisory Group, Staff, and Authors
Appendix II: Radiocarbon dates

Citation preview

Avaldsnes – A Sea-Kings’ Manor in First-Millennium Western Scandinavia

Reallexikon der germanischen Altertumskunde – Ergänzungsbände

Herausgegeben von Sebastian Brather, Wilhelm Heizmann und Steffen Patzold

Band 104

  Avaldsnes – A Sea-Kings’ Manor in First-Millennium Western Scandinavia

Edited by Dagfinn Skre

ISBN 978-3-11-042578-9 e-ISBN (PDF) 978-3-11-042108-8 e-ISBN (EPUB) 978-3-11-042113-2 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress Bibliografische Information der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet unter http://dnb.dnb.de abrufbar © 2018 Walter de Gruyter GmbH, Berlin/Boston Satz: Dörlemann Satz GmbH & Co. KG, Lemförde Druck und Bindung: Hubert & Co. GmbH & Co. KG, Göttingen ♾ Gedruckt auf säurefreiem Papier Printed in Germany

Dagfinn Skre

Preface

Avaldsnes (Old Norse Ǫgvaldsnes – Ǫgvaldr’s headland), situated on the island of Kormt off the south-western coast of Norway, is mentioned frequently in the Old Norse literature as the setting for dramatic and decisive events involving the first Norwegian kings. The first King of Norway, Haraldr hárfagri, is supposedly buried in the vicinity. For these reasons, since the 16th century, the site has captivated the interest of artists, antiquarians, and historians (Skre, Chs. 2, 23). However, except for the rich finds in the Flaghaug grave mound excavated 1834–5, archaeologists have paid the site only limited attention. This changed following the accidental discovery of a subterranean passageway, excavated in 1985–6. The find raised considerable interest locally and among researchers at the regional Archaeological Museum in Stavanger. The increased popular interest in Avaldsnes’ history resonated among local politicians. Development of roads and housing in the Avaldsnes area prompted them to initiate an overall plan for the area, and Marit Synnøve Vea of Karmøy Municipality’s Culture Section was assigned the task of drafting the strategy. In 1993 she completed the report ‘Proposal on the Avaldsnes Project’ (Framlegg om Avaldsnesprosjektet). The report proposed as the plan’s aims three main goals: protection of cultural heritage, promotion of understanding of the past, and development of tourism. To those ends, the report called for construction of a visitors’ centre and establishment of a research programme to produce new knowledge regarding Avaldsnes. The latter would require the involvement of scholarly institutions – universities, museums, and colleges (Vea 1993:7–10). The plan was approved by the municipal council, and Vea was appointed director of the municipality’s Avaldsnes Project. The Archaeological Museum in Stavanger was designated as the project’s point of contact with the archaeological community. In 1993 professor of archaeology Bjørn Myhre became the museum’s director and main contact for the project. In 1994 the municipality invited the museum to produce a research plan for Avaldsnes. A group of 11 specialists – most employed at the museum, with research interests spanning the Stone Age to the Medieval Period – were assembled to design the plan, which was published the following year (Lillehammer 1995). Although the proposed four-year project was never launched as planned, several archaeological surveys were conducted by the Archaeological Museum in Stavanger in the following years (Bauer and Østmo, Ch. 5). Additionally, the municipality organised several geophysical surveys (Stamnes and Bauer, Ch. 16). These efforts were funded by the museum, Karmøy Municipality, and the local benefactor Sigurd Steen Aase. In 2004 the municipality decided to extend its contacts within the archaeological community. They invited a group of Scandinavian specialists to form a steering com-

VI 

 Dagfinn Skre

mittee together with representatives from the municipality and from the Archaeological Museum in Stavanger. The group assigned top priority to finding the Viking Age royal-manor farmyard. The museum conducted survey excavations for this purpose in 2005 and 2006; in the second year settlement remains from the Viking Period were identified under the car park just south of the St Óláfr’s Church (Bauer and Østmo, Figs. 5.1–2) Following the discovery, the municipality invited the editor of the present volume to organise an excavation and research project at the University of Oslo. There, the Avaldsnes Royal Manor (ARM) Project was first hosted by the Institute of Archaeology, Conservation Studies and History and since 2010 by the Museum of Cultural History. The project’s funding, in total 34 million NOK, was supplied by Karmøy Municipality, which in turn received considerable contributions from Rogaland County Council and from the regional businesses enumerated in this book’s colophon page. As these words are being written, preparations are underway for the excavation in the coming summer of the remains of a late 13th- or early 14th-century masonry building discovered in 2012, made possible by a NOK 5.4-millon grant from the Norwegian government. The ARM Project’s research and excavation plans (Skre, Ch. 4; Bauer and Østmo, Ch. 5) were developed in the 2007–9 pilot project, which also collected and analysed documentary and archaeological evidence. Excavation took place 2011–12 in close collaboration regarding facilities and logistics with the municipality, and regarding artefactual finds with the Archaeological Museum, which by that time had become a part of the University of Stavanger. Post-excavation work continued into early 2013, when the publication phase started. Two volumes are planned from the project in addition to several papers in academic journals. To my knowledge, Karmøy is the only municipality in Norway ever to set up an archaeological research agenda. They have demonstrated untiring dedication in realising the visionary ideas first articulated by Vea twenty-five years ago. Of the many who have contributed to making this possible, three were instrumental: the director of the municipality’s Avaldsnes Project Marit Synnøve Vea, the benefactor Sigurd Steen Aase, and Kjell Arvid Svendsen, mayor of Karmøy 1996–2011. It was my utter good fortune in 2004 to have met by chance this trio who devoted themselves to the task of generating funding and support for the ARM Project. They imposed but one expectation on the ARM Project’s results: an adherence to the highest professional standards, an inspiration throughout. The two subsequent mayors, Aase Simonsen (2011–15) and Jarle Nilsen (2015–present) have embraced the project with enthusiasm, generating funding for the publication phase and the coming excavation of the masonry building. A number of people employed in Karmøy Municipality and Karmøy Kulturopplevel­ ser – I cannot possibly mention all of them – have contributed in every possible way to the excavation’s success. Cooperation with the local representatives of the Church of Norway and the owner of the Avaldnes manor, Opplysningsvesenets fond, as well as

Preface 

 VII

their tenant, has been smooth and amiable. Support from Rogaland County Council, both financial and political, has been substantial. Throughout the excavations and publication phases, cooperation with the Directorate for Cultural Heritage (Riksantikvaren) and the Archaeological Museum, University of Stavanger, has been entirely positive and productive. The ARM Project has included a Project Council and an Advisory Board. Members of both bodies are listed in Appendix I; one of whom merits particular mention: Professor Knut Helle. Until he passed away in 2015 he offered strong support and substantial scholarly contributions to the project. Uncompromising in seeking quality, he was always open to discuss new ideas and perspectives. I dedicate this book to his memory. The research efforts presented in this volume enable the delineation of Avaldsnes’ history into three phases. During the first phase, sea kings used Avaldsnes as a base for dominating the islands and waters along the coastal sailing route. Around AD 900, one of the sea kings rose to become a land king, the first King of Norway, ushering in the second phase. The third phase began around AD 1400, when the manor became the rectory for the vicar of St Óláfr Church, as it has since remained. With the Karmøy Municipality’s heavy investments in research and dissemination at Avaldsnes, a fourth phase is immanent. The site’s history, significant far beyond the local, regional, and national levels, is reclaiming Ǫgvaldr’s headland. Dagfinn Skre Oslo, March 2017

Table of content Preface   V Abbreviations 

 XIII

Section A Scholarly Background Dagfinn Skre 1 Rethinking Avaldsnes and Kormt 

 3

Dagfinn Skre 2 Exploring Avaldsnes 1540–2005 

 11

Else Mundal 3 Avaldsnes and Kormt in Old Norse Written Sources  Dagfinn Skre 4 The Avaldsnes Royal Manor Project’s Research Plan and Excavation Objectives   53

Section B Excavation Results 2011–12 Egil Lindhart Bauer and Mari Arentz Østmo 5 Excavations and Surveys 1985–2012   65 Mari Arentz Østmo and Egil Lindhart Bauer 6 Site Periods and Key Contexts   83 Mari Arentz Østmo and Egil Lindhart Bauer 7 The Prehistoric Settlement and Buildings  Egil Lindhart Bauer and Mari Arentz Østmo 8 Prehistoric Agriculture   137 Mari Arentz Østmo 9 The Production Area 

 157

Egil Lindhart Bauer 10 Two Iron Age Boathouses 

 183

 103

 35

X 

 Table of content

Mari Arentz Østmo 11 A Late Iron Age Palisade Facing the Karmsund Strait 

 209

Mari Arentz Østmo and Egil Lindhart Bauer 12 Grave Monuments at Avaldsnes   227 Egil Lindhart Bauer 13 Political and Ritual Aspects of Cooking Pits  Egil Lindhart Bauer 14 The High Medieval Royal Manor  Egil Lindhart Bauer 15 The Post-Medieval Rectory 

 253

 277

 309

Section C Scientific Analyses 2011–12 Arne Anderson Stamnes and Egil Lindhart Bauer 16 Geophysical Surveys   327 Richard I. Macphail and Johan Linderholm 17 Microstratigraphy (Soil Micromorphology and Microchemistry, Soil Chemistry, and Magnetic Susceptibility)   379 Rebecca J. S. Cannell, Paul N. Cheetham and Kate Welham 18 Geochemical analysis using portable X-ray fluorescence 

 421

Rachel Ballantyne, Stella Macheridis, Emma Lightfoot, and Alice Williams 19 Biological Remains   455

Section D Specialist Studies Mari Arentz Østmo 20 Artefacts from the 2011–12 Excavations 

 513

Elna Siv Kristoffersen and Åsa Dahlin Hauken 21 Migration Period Pottery from Avaldsnes: A Study of Shards from Bucket-shaped Pots   527



 XI

Table of content 

Frans-Arne H. Stylegar and Håkon Reiersen 22 The Flaghaug Burials   551 Dagfinn Skre 23 The Raised Stones 

 639

Stefan Brink 24 Avaldsnes, Kormt and Rogaland. A Toponymy and Landscape Survey  Torun Zachrisson 25 Depositional Traditions in Iron Age Kormt 

 665

 687

Frode Iversen 26 Emerging Kingship in the 8th Century? New Datings of three Courtyard Sites in Rogaland   721

Section E Avaldsnes: a Sea-Kings’ Manor Dagfinn Skre 27 Aristocratic Presence along the Karmsund Strait 2000 BC–AD 1368  Dagfinn Skre 28 The Warrior Manor 

 749

 765

Dagfinn Skre 29 Sea Kings on the Norðvegr 

 781

References Appendices References   803 Appendix I: The ARM Project Council, Advisory Group, Staff, and Authors  Appendix II: Radiocarbon dates   867

 861

Abbreviations A-ID [number] = Inventory number in the Norwegian archaeological site database Askeladden: https://askeladden.ra.no/ AM, AmS = Museum of Archaeology, University of Stavanger ARM Project = Avaldsnes Royal Manor Project, Museum of Cultural History, University of Oslo B[number] = Inventory number, University Museum of Bergen Ber. RGK = Bericht der Römisch-Germanischen Kommission. Philipp von Zabern. Mainz am Rhein. Beta-[number] = AMS radiocarbon dating, Beta Analytic, Florida, USA C[number] = Inventory number, Museum of Cultural History, University of Oslo DI = Diplomatarium Islandicum, 1–16. 1857–1976. Bókmentafèlag. Kaupmannahöfn. [NB! Number (e.  g. DI 2:543) always indicates the document number, not page.] DN = Diplomatarium Norvegicum. Oldbreve til Kundskab om Norges indre og ydre Forhold, Sprog, Slægter, Sæder, Lovgivning og Rettergang i Middelalderen, 1–22. 1847–1995. Kristiania. [NB! Number (e.  g. DN I:543) always indicates the document number, not page.] FAO = World Soil Resources: www.fao.org/ag/agI/agII/wrb/soilres.stm#down. ID = A-ID ISKF = Instituttet for sammenlignende kulturforskning. Oslo. KHM = Museum of Cultural History, University of Oslo. KLNM = Kulturhistorisk Leksikon for Nordisk Middelalder fra vikingtid til reformasjonstid, 1–22. 1956–1978. Copenhagen. MCH = KHM NF = Nicolaysen 1862–6 NG = Oluf Rygh 1897–1924: Norske Gaardnavne, 1–17. Kristiania NGL = Norges gamle Love indtil 1387, 1–4. 1846–95. Kristiana. NGU = Geological Survey of Norway. NIKU = Norsk institutt for kulturminneforskning. NRJ = H.J.Huitfeldt-Kaas and A.O. Johnsen (eds.) 1885–1906: Norske Regnskaber og Jordebøger fra det 16de Aarhundrede (1514–1570), 1–4. Kjeldeskriftfondet. Kristiana. NSL = J. Sandnes and O. Stemshaug (eds.) 1997: Norsk stadnamnleksikon, 4th edn., Oslo. NTNU = Norwegian University of Science and Technology, Trondheim. OAS = Oslo arkeologiske serie, 2003–10. Unipub. Oslo. ON = Old Norse ONP = Ordliste over det norrønne prosasprog, http://onp.ku.dk/adgang_til_ordliste_etc/ordliste_ og_citater/ OPIA = Occasional Papers in Archaeology, 1989–. Societas Archaeologica Upsaliensis, Uppsala. R[number] = Illustration in Oluf Rygh 1885: Norske Oldsager. Cammermeyer. Kristiana. RGA = Reallexikon der germanischen Altertumskunde, 1–37. 2nd edn. 1973–2008. De Gruyter. ­Berlin/New York. RGA-E = Reallexikon der Germanischen Altertumskunde – Ergänzungsbände. De Gruyter. Berlin/New York and Berlin/Boston. RN = Regesta Norvegica. 1–10, 1989–2015. Riksarkivet, Kjeldeskriftavdelingen. Oslo. [NB! Number (e.  g. RN 2:543) always indicates the document number, not page.] S[number] = Inventory number, Museum of Archaeology, University of Stavanger SHM[number] = Inventory number, The Swedish History Museum, Stockholm T-[number] = Conventional radiocarbon datings, NTNU, Trondheim T[number] = Inventory number, NTNU, Museum of Natural History and Archaeology, Trondheim TRa-[number] = AMS radiocarbon datings, NTNU, Trondheim. TUa-[number] = AMS radiocarbon dating samples prepared at NTNU and dated at Uppsala University

XIV 

 Abbreviations

Ua-[number] = AMS radiocarbon datings, Uppsala University UMB = University Museum, University of Bergen UOÅ = Universitetets Oldsaksamling, Årbok WCD = The New International Webster’s Comprehensive Dictionary of the English Language. Merriam-Webster. Springfield, Mass.

Section A Scholarly Background

Dagfinn Skre

1 Rethinking Avaldsnes and Kormt Avaldsnes is among a select group of Scandinavian sites to feature repeatedly in the Old Norse literature (Fig. 1.1; Mundal, Ch. 3; Brink, Ch. 22). Upon closer examination, such sites tend to be rich in archaeological monuments and finds dating to the time periods referred to in the saga accounts and skaldic verses. Further back in time, fewer such sites show up, and more often the accounts take on the cast of legends. For instance, it is fairly easy to connect the urban topography revealed through more than 150 years of archaeological excavations in high-medieval Oslo with the sites, streets, and buildings mentioned in, for example, the early 13th-century King Sverris saga, written only a few years after the events. Less obvious is the correspondence between Snorri’s statements regarding the founding of Oslo by King Haraldr harðráði in the mid-11th century and the archaeology of the urban remains there dating back to around AD 1000. What can be made of the account in Egill Skallagrímsson’s saga of Egill’s sacking of Lund in Skåne in the mid-10th century, at which time the town Lund did not exist? Was ‘Lund’ at the time of Egill’s visit the name of the site that is now called Uppåkra, a huge 1st–10th-century aristocratic settlement some 4 kilometres south of the town (Andrén 1998)? Skíringssalr in Othere’s account from c. 890 is securely identified as the 9th-century town Kaupang in Vestfold (Skre 2007c), and Lejre mentioned repeatedly in Beowulf is evidently the site near Roskilde in Sjælland where several hall buildings from the 7th–10th centuries have been excavated (Christensen 2015). Uppsala, mentioned in Ynglingatal and by Saxo Grammaticus, is securely identified as Old Uppsala with its huge 7th-century mounds, several raised platforms with remains of hall buildings, and the recently discovered one-kilometre row of posts (Ljungkvist et al. 2011; Jörpeland et al. 2013). However, it remains an uncertain endeavour to establish a connection between these unique and highly impressive monuments and the row of twenty kings that according to Ynglingatal ruled there, the first of whom was purportedly a son of the god Frey. Although connections between monuments and more or less legendary accounts in sagas and poems necessarily remain obscure, they are the rule rather than the exception. Uppsala is but one example; others are the locations where Odin and his following, according to Snorri’s Ynglingasaga, resided on their way through Scandinavia to Uppsala, namely Fornsigtuna and Fyn. In Fornsigtuna two large hall buildings on raised platforms (Hedman 1991) have been found, and in Gudme on Fyn an early Iron Age aristocratic settlement with prestigious finds and large central hall building have been excavated (Nielsen et al. 1994). Thus, the written evidence should not be dismissed as untrustworthy, but rather be involved in careful attempts to connect the two types of evidence. Fruitful results have emerged from such undertakings (e.  g., regarding Uppsala, see Sundqvist 2002).

4 

 A: Scholarly Background

Fig. 1.1: Avaldsnes, Old Norse Ǫgvaldsnes (‘Ǫgvaldr’s headland’), seen towards the south-southeast. The archaeological evidence retrieved by the ARM Project 2011–12 is found near the St Óláfr Church, commissioned by King Hákon Hákonarson c. 1250. The sheltered sailing route along the western Scandinavian coast, the Norðvegr, here protected from the ocean in the west by the island of Kormt, Old Norse Kǫrmt (‘low protective wall’, to the right in the photo), runs just past the site. The route continues southwards past the conspicuous mountain seen in the distance. Its name, Bokn, meaning ‘sign, signal’ and related to the English ‘beacon’, most probably owes to its visible appearance in the flat coastal landscape (Skre, Ch. 29:782–4), which made for a useful navigational mark (Brink, Ch. 24:668).Photo: KIB media.

Avaldsnes is one of the sites connected in sagas and poems to prominent persons and their activities. Accounts that involve historical persons that occurred only a century or two before the time when they were written down can be treated as relatively historically reliable. For example, the tale of Ásbjǫrn selsbani (Mundal, Ch. 3), despite Snorri’s literary embellishment, appears to be based on the actual murder of one of King Óláfr inn helgi’s men at Avaldsnes committed by Ásbjǫrn, a relative of the prominent men Þórir hundr and Erlingr Skjálgsson. Detailed analyses of the texts are necessary to identify trustworthy elements (Mundal, Ch. 3; Skre, Ch. 27:761–4). Less trustworthy are those saga accounts that have a legendary or folkloric form, contain elements borrowed from other literary works, or occur in a less precisely defined distant past, apparently at least 5–6 centuries before they were written down. To this category belongs the tale of King Ǫgvaldr and his cow that always accompanied him – the story contains all the elements of a legend. As Mundal points out (Ch 3:45–6), traditions regarding significant persons of a

1 Skre: Rethinking Avaldsnes and Kormt 

 5

Åker Uppsala Avaldsnes

Helgö Skiringssal

Tissø Lejre

Uppåkra Sorte Muld

Gudme

0

300 km

Fig. 1.2: Prominent aristocratic sites in first-millennium Scandinavia mentioned in Chapters 1, 4, and 28. Illustration: I. T. Bøckman, MCH.

distant past are often ‘drawn’ to sites that are coincidentally known to have been prominent in a more recent past. Avaldsnes’ status as a royal manor from the 10th century onwards may have inspired saga writers of the 13th–14th centuries to set their stories about a more distant past at Avaldsnes. However, the likelihood of such tales reflecting a historical reality increases in light of the archaeological evidence of an aristocratic presence there from the 3rd century onwards, although not necessarily continuously (Skre, Ch. 27). Moreover, it appears that Avaldsnes had achieved a mythical status well before the 10th century (Mundal, Ch. 3:36–7; Skre, Ch. 28:777–8). As such, the sequence of causality in the explanation above may be inverted: the site’s mythical status, probably established in the site’s first heyday in the 3rd–4th centuries, may itself have drawn men of power and ambition to settle there and to bury their predecessors nearby.

6 

 A: Scholarly Background

K a

r m s u n

d

Avaldsnes

KORMT

0

5 km

Fig. 1.3: In this book, Avaldsnes and the surrounding land in northern Kormt and on both sides of the Karmsund Strait constitute the primary area of study. Illustration: I. T. Bøckman, MCH.



1 Skre: Rethinking Avaldsnes and Kormt 

 7

These are but some of the deliberations developed in this volume. Although Avaldsnes has more than 450 years of research history (Skre, Ch. 2), the multifaceted scholarly challenges and possibilities in reconsidering Avaldsnes and Kormt have been the primary impetus for initiating the Avaldsnes Royal Manor (ARM) Project. In bringing to light a substantial corpus of archaeological material from the site, initiating a wide scope of scholarly research efforts, and revitalising the existing archaeological and written evidence, the project seeks to produce new insights into the history of Avaldsnes and the land along the Karmsund Strait. This research is presented in the present volume. In the second volume, this knowledge will be used as a springboard to address some classic research questions in northern European history: the transformation through the first millennium AD of the Germanic tribal societies and the emergence of kingship. The details of the project’s objectives and research plan are described in Chapter 4.

1.1 The content of this volume – a guide to readers The book is divided in five sections. Section A (chapters 1–4) lays out the scholarly background for the research presented in the book. Section B (chapters 5–15) presents the results from the 2011–12 excavations in thematic chapters. Section C (chapters 16–19) presents scientific analyses from the excavations. Section D (chapters 20–26) presents specialist studies of relevant finds, sites, and place names from Avaldsnes, Kormt, and nearby. Section E (chapters 27–29) explores the research questions outlined in chapter 4 on the basis of the results from Sections A–D. Readers may derive an overview of the book’s content from the abstracts that introduce each chapter. Some readers will doubtless have special interest in specific chapters and sections. The general reader is advised to read section A ad libitum, and use chapter 6 as a key to excavation results and scientific analyses presented in sections B and C. The specialist studies in section D may also be read ad libitum, and are referred to in section E. Chapter 27 is based on the site chronology and main finds presented in chapter 6, while chapters 28–29 are more thematic. The volume is extensively cross-referenced. These references appear in the following format: (Østmo, Ch. 9:163), indicating author, chapter number, and page; and (Østmo, Fig. 9.4), indicating this specific figure occurring in chapter 9. Initial capitals (Ch., Fig., Tab.) indicate that these are cross-references within the volume; references to chapters, figures, and tables in other publications are not capitalised. All radiocarbon datings, both from 2011–12 and earlier campaigns, have been calibrated according to OxCal v4.2.3; they are listed with their respective calibration curves in Appendix II. When referred to in the text, datings are given in terms of the one sigma (68.2 % probability) unless otherwise stated. If the one sigma spans more than one time interval, only the start of the earliest and end of the most recent is indi-

8 

 A: Scholarly Background

ARCHAEOLOGICAL PERIODS

2000 BC

SITE PERIODS

NEOLITHIC

EARLY BRONZE AGE

1500 BC

I 1000 BC LATE BRONZE AGE

Early Iron Age

500 BC

PRE-ROMAN IRON AGE

AD

II

ROMAN IRON AGE

MIGRATION PERIOD

III 500 AD

Late Iron Age

MEROVINGIAN PERIOD

IV

VIKING AGE

1000 AD

MIDDLE AGES

V VI

1500 AD

VII

Fig. 1.4: This time line provides an overview of the standard chronological periods in west-Scandinavian archaeology (left) as well as the main chronological periods in the Avaldsnes site (Site Periods I–VII, right; see Østmo and Bauer, Ch. 6). Illustration: I. T. Bøckman, MCH.



1 Skre: Rethinking Avaldsnes and Kormt 

 9

cated. For example, for the dating Beta-304876 where the one sigma spans the two periods AD 214–61 and 280–326, this is written as AD 214–326. Place names are in generally written in their modern form and according to their native spelling, except where a name’s Old Norse forms are discussed. With one notable exception: the form Kormt is used throughout the book, despite the island’s current name of Karmøy. This is intended to avoid confusion with the modern Karmøy Municipality, which also encompasses other islands and a part of the mainland. Kormt corresponds to the island’s Old Norse name Kǫrmt, a form used in local speech into the 20th century. Names of Old Norse literary works are spelled according to conventions in the specialist disciplines, as are names of persons mentioned therein. Finally, the use throughout this book of Norðvegr (‘the route to the north’ or ‘the northern route’; Brink, Ch. 24:667) as the reconstructed Old Norse form of the coastal sailing route’s name is not meant to disregard the current debate on whether the original form could instead be Nórvegr (‘the narrow route’; e.  g. Myrvoll 2011)). In the context of this volume, the essential issue is that the route bore a name that became the name of the kingdom that was created around AD 900. The name’s original form and meaning, and indeed the sailing route, will be discussed in detail in the second Avaldsnes volume.

Book acknowledgements: Copyediting and language revision of the present book has been undertaken by Anthony Zannino. Illustrations have been managed by Ingvild Tinglum Bøckman, who also has produced most of them (see captions). Maps of Norway are used under licence from The Norwegian Mapping Authority (Kartverket). Topographical data for Europe are obtained from Natural Earth Data. LiDAR data from Avaldsnes are produced by Blom Geonatics AS.

Dagfinn Skre

2 Exploring Avaldsnes 1540–2005 Avaldsnes, Kormt, and the Karmsund Strait are frequently mentioned in the Old Norse written sources, often referred to as the residence and burial site of kings. The site has attracted the attention of scholars since the 16th century, first by the humanists who began to study the Old Norse texts; subsequently by historians and antiquarians, from the mid-19th century academic historians, and from the early 20th century joined by archaeologists. In this chapter, the significant contributions from this range of scholars are summarised. The literature on Avaldsnes tends to adopt one of two perspectives: some scholars focus their analysis on evidence from the site itself, while others situate the site within discussions of broader societal or political issues. Summarising scholarship of the first type, this chapter traces how various types of evidence became available at different times and how scholars have shifted in their assessment of the evidence. Discussions of the second type of scholarship identify the continuities and changes regarding the contexts in which Avaldsnes has been situated. One thread in particular has been winding its way through these 450 years of Avaldsnes research: the problem of why kings preferred to reside on the modestly fertile and windblown island of Kormt, rather than the lush densely populated regions further inland and along the fjords. The most significant shift in the scholarship is seen in the integration of Avaldsnes within the research into the rikssamlingen (‘the unification of the realm’). The unification process has a long research history, but one that before the early 20th century did not consider Avaldsnes’ location on the outer coast. In the 1990s the scope of this research shifted from a national, narrowly 9th–10th-century perspective to a regionally North European, long-term perspective. This literature review of Avaldsnes scholarship forms the foundation for the research strategy employed by the current research project, detailed in Chapters 4 and 5.

As the medieval literary evidence gradually became known in the 16th–19th centuries, scholars realised that Avaldsnes appeared to have been a kings’ seat for a millennium, that is, from the ‘heroic age’ until the death of Hákon Magnússon (1380), the last king of the medieval Norwegian kingdom. In the late 19th and 20th centuries, when archaeological finds and monuments could be identified and dated with increasing reliability, it became evident that prestigious sites and finds at Avaldsnes and Kormt were not limited to that millennium; rather, indicators of aristocratic presence extend back into the Bronze Age, and even to the late Stone Age, that is, over a period of more than 3,000 years. However, the history of scholarship entails not only accumulating evidence, but also discarding it. For instance, until the 1860s, the fornaldarsǫgur were considered by scholars as the primary evidence regarding the Iron and Viking Ages; since P.A. Munch (1810–63), however, these sagas have hardly been deemed worthy of mentioning by scholars of those periods. Conversely, the archaeological record, ranked by most early scholars at best as support for the written evidence, has now attained a voice of its own. Every scholar who has discussed Avaldsnes in a broader context has had to address

12 

 A: Scholarly Background

the fact that, over many centuries, kings preferred to reside on an island bordered by the open ocean to the west rather than the densely populated hinterland where agricultural yields were higher. Various theories were put forth during the first 300 years of Avaldsnes research; Johan Koren Christie’s paper (1842) citing the site’s position at the narrow Karmsund Strait, a bottleneck on the Norðvegr, the sheltered sailing route that connected the numerous dispersed settlement districts along the western Scandinavian coast and fjords, has apparently settled the question. However, the connection between Avaldsnes and the sea route has received different interpretations since Christie’s paper. Identifying and characterising these shifts over the 450 years of Avaldsnes scholarship is a central objective of this chapter. Thus, this chapter supplies one of the pillars that support the research efforts of the Avaldsnes Royal Manor Project; in particular, supplying the background for designing an adequate research strategy for the current project (Skre, Ch. 4). This chapter will not discuss every scholarly work that in some way has mentioned Avaldsnes, only research that discusses the Avaldsnes site, the monuments there, and their location along the Karmsund Strait. While this includes nearly all scholarly writings on Avaldsnes up to c. 1850, from the subsequent period only those that have produced significant insights or were typical of their time will be mentioned.

2.1 Renaissance and early modern historians, c. 1540–1711 2.1.1 Mattis Størssøn, Absalon Pederssøn Beyer, and Peder Claussøn Friis Mattis Størssøn’s (ca. 1500–69) Norske Kongers Krønicke oc bedrifft (‘The chronicles and achievements of Norwegian kings’), written in the 1540s (Jørgensen 1994:186–7) and printed in 1594, is for the most part a brief paraphrasing of Snorri Sturluson’s Heimskringla. Størssøn was a law officer of the crown (‘lagmann’), from c. 1533 in Agder on the southernmost tip of Norway, and from c. 1540 in Bergen. He held this office for the rest of his life (Sørlie 1962:viii–ix). As was common at the time, Størssøn based his work on written evidence; monu­ ments are mentioned only in passing or not at all. Exceptionally, though, Størssøn mentioned the mound north of the St Óláfr’s Church at Avaldsnes (1594:28):



2 Skre: Exploring Avaldsnes 1540–2005 

 13

King Haraldr hárfagri lived for three years after Eiríkr blóðøx had received the authority of the land, and King Haraldr died at Avaldsnes and was laid in the great mound north of the church.1

In his capacity as a royal official, Mattis Størssøn passed through Avaldsnes by the sea route on numerous occasions and would have seen the mound for himself (Østmo and Bauer, Ch. 12:231–5; Stylegar and Reiersen, Ch. 22). Størssøn was of course familiar with Snorri’s account of Haraldr’s mound and its location (Mundal, Ch. 3:37–8); Snorri must have been his source for identifying the mound as Haraldr’s. Further evidence of this period’s modest interest in historic sites and antiquities is to be found in the writings of another prominent figure: the priest Absalon Pederssøn Beyer (1528–75), who belonged to the same circle of historically interested humanists in Bergen. Beyer’s manuscript, Om Norgis Rige (‘About the Norwegian Realm’), written in the years 1567–70, was circulated, copied, and widely read before it was printed in 1781. His text on the early period is brief; the time before Haraldr hárfagri covers 3 pages (Beyer [1567–70] 1895:4–6). Beyer would have travelled numerous times past Avaldsnes on his voyages to Stavanger, Oslo, Copenhagen, and beyond, as is reflected in his references to Avaldsnes (Beyer [1567–70] 1895:107): In the same county lies Avaldsnes church, on the left-hand side as one sails into the Karmsund Strait; it has been one of the vey largest municipal churches in Norway, built at the expense of King Hákon, but now most of it is in decay.2

His view on the rulers prior to the establishment of the Kingdom of Norway is interesting (Beyer [1567–70] 1895:5): Before autocratic kings ruled in Norway by divine providence, there was a kind of secularly ruled state called aristocratia; that is, a regime of the best men […] How long this state prevailed one does not know, even though some promontory kings [neße konger] are listed with their years of rule.3

Although they may well have stayed overnight in the Avaldsnes harbour or vicarage, there is no explicit record of Størssøn or Beyer staying there. The first historian – since Snorri’s assumed visit – who can be said to have visited Avaldsnes with certainty was the priest Peder Claussøn Friis (1545–1614). Born in Egersund, Rogaland, and rector

1 “Konning Harald haarfager leffde trij aar epther att Erich blodøxe fich landitt att raade, och bleff koning Harald haarfager død paa Auellsnes och bleff lagtt ij then store høu norder fraa kircken.” 2 “Vdi samme stict ligger Auelsnes kircke paa den venstre haand naar mand kommer ind y Karmesund, som haffuer verred en aff de allerstörste herrets kircker y Norrig, oc ved konning Haagens bekostning opbygget, men nu er den störste part der aff forfalden.” 3 “För end eenvolds konger af gudz forsiun regerede vdi Norge, da var der en slags verdslig regimentis stat, som kaltis aristocratia, det er de beste mends regimente, […] Huor lenge denne stat haffuer verit oc varit, veed mand icke, endog at nogle neße konger opregnis vdi deris aars regimente.”

14 

 A: Scholarly Background

in Lista on the southernmost tip of Norway at the age of 21, Friis probably received his education at the bishop’s see in Stavanger, where he became a member of the chapter at age 30 and archdeacon before he was 45. Friis, a prominent cleric in his time, left his main legacy in topographical and historic studies. In his later years he produced the first extensive translation into Danish of Snorri’s Heimskringla and a few other kings’ sagas (Jørgensen 1994). His topographical works describe botanical and zoological occurrences in various districts, as well as historical sites, persons, and events (Jørgensen 1994; 2000). From at least 1590 he collected material for his main topographical work, Norrigis Bescrifuelse (‘Norway’s description’), completed in 1613 and published by Ole Worm (see below) with an elaborate title and some alterations and additions in 1632 (Storm 1881:lxvii–lxxix).4 In the following two centuries Friis’ Norrigis Bescrifuelse became immensely influential. Not only did it become the template for the topographical genre that flourished from the mid-18th century onwards; it was also paraphrased and translated in more or less all Norwegian historical works of the following century. For instance, the topographical introduction of Tormod Torfæus’ Historia rerum Norvegicarum from 1711 is more or less a translation of Friis’ book into Latin (1711:1:27–110; Storm 1881:lxxiii; Torfæus 2008:118–238). Of the several learned men in 16th- and early 17th-century Norway, writes Storm (1881:liv), but ‘one daresay that regarding command of the subject, skill in rendering, and store of knowledge, [Friis] stands above them all’5. Friis’ topographical interest is evident in the greater prominence he accords monuments and antiquities in his text. For example, his description of Avaldsnes parish in Norrigis Bescrifuelse (Friis [1613] 1881:324–5): Afueldsnæs, in bygone days called Ǫgvaldsnæs after Ǫgvaldr King who first lived and is buried there, and a cow was his God, who was laid in the mound and buried with him. Since then there lived Haraldr hárfagri and several kings after him at the same Augvadsnæs. […] There stands a small church called the King’s Chapel and is one of the four first churches built by K. Óláfr Tryggvason here in Norway. […] and this chapel now stands deserted and in disrepair, but close by is built a lovely great stone church.6

The information regarding the three kings’ connections to Avaldsnes would have come to Friis through his readings of Old Norse saga manuscripts. However, his description

4 A revised edition was published by Gustav Storm in 1881. 5 ”det tør dog siges, at han baade i beherskelse af Stoffet, i Skildringens Dygtighed og Kundskabsmasse rager frem over dem alle”. 6 “Afuelsnæs, kaldis fordum Augvaldsnæs, aff Aguald Konning, som der først bode, oc ligger der begrafuen, oc var en Koe hans Gud, huilcken oc bleff lagt i Højen oc begrafuit met hannem. Siden bode Harald den Haarfage oc flere Konger effter hannem paa samme Augvadsnæs. […] Der hos staar en liden Kircke, kaldis Kongens Capel, oc er en aff de 4 første Kirker som K. Olaff Trøggesøn lod bygge her i Norrige […] oc staar dette Kapel nu øde oc forfalden, men strax hos er bygt en skiøn oc stor Stenkircke.”



2 Skre: Exploring Avaldsnes 1540–2005 

 15

of the St Óláfr’s Church and what he calls a chapel is evidently first-hand.7 His education and later duties in Stavanger would have made him intimately familiar with the diocese. The rector at Avaldsnes at the time, Christopher Sigurdssøn, was also a member of the chapter (Skadberg 1950:128–9); it is probable that Friis visited him in the Avaldsnes rectory. A copy of Friis’ translation of the kings’ sagas was sent by the bishop to the Royal Historian Claus Lyschander, probably shortly after Friis’ death (Storm 1881:lxvii). In this way, it would have come to the attention of the king; perhaps for this reason, King Christian IV of Denmark-Norway (reign 1588–1648) visited Avaldsnes during his sojourn in Norway in 1627 (Skadberg 1950:20–1). Ole Worm, the king’s physician and a prominent antiquarian, was familiar with both of Friis’ two main works, the saga translation and Norrigis Bescrifuelse – as mentioned, it was he who had them printed in 1633 and 1632 respectively. Friis had not, however, mentioned what later proved to be Worm’s prime interest in Avaldsnes: a runic inscription on a raised stone (Skre, Ch. 23:640–4, Fig. 23.2). Friis’ writings brought to the attention of the Copenhagen intelligentsia the idea of Avaldsnes as a historic royal manor and site of prominent antiquities. Friis was the first since Snorri to acknowledge as much in writing. The following century’s authors of history and topography appear to have remained content with Friis’ work. Almost a hundred years were to pass from Friis’ completion of Norrigis Bescrifuelse before new antiquarian information on Avaldsnes was published. At that time additional Old Norse manuscripts had come to light, providing Tormod Torfæus (1636–1719) a broader base of evidence from which to draw for the writing of his four-volume Historia rerum Norvegicarum, a history of Norway from the earliest times to 1387.

2.1.2 Tormod Torfæus 1711: Historia rerum Norvegicarum Þormóður Torfason, or Tormod Torfæus (Fig. 2.1) as he came to call himself, was born just outside Reykjavik in Iceland; he graduated from Skálholt School in Iceland with excellent recommendations and enrolled at Copenhagen University at age 18. Only six years later, in 1660, he was appointed by King Fredrik III as translator of ancient Icelandic texts. Fredrik was greatly interested in Old Norse history and held Torfæus in his good graces. At the king’s behest, Torfæus travelled to Iceland to collect antique documents, and returned with important findings. He left this commission in 1664 to take up a position as a royal clerk in Stavanger, Rogaland. The following year he married there; through marriage he became the owner of the farm Stangeland at Kormt, where he lived the rest of his life, from 1682 as the royal Historiograph for

7 An alternative identification of the ruin that he identified as a royal chapel is presented in Ch. 14 (Bauer).

16 

 A: Scholarly Background

Fig. 2.1: Tormod Torfæus (1636–1719) portrayed in a posthumous engraving. Photo and owner: The National Library of Norway.

Norge (‘Historian of Norway’) and assessor at Copenhagen University, a position that de facto equalled a professorship. Torfæus employed secretaries at Stangeland and had them produce transcripts of several medieval books and documents, some of which later disappeared. The publications arrived at a steady pace from the historian at Stangeland. From 1696 to 1706 he published books on the history of the Faroes, the Orkneys, Greenland, and the Norse discovery of America, all based on saga accounts. However, his magnum opus that he had been working on for 30 years was the Historia rerum Norvegicarum, about 2,000 pages in four volumes, published in 1711. Eight years later, at the age of 83, Torfæus died in his home at Stangeland. He was buried, fittingly for a royal historian, in the chancel of the St Óláfr’s Church, built by King Hákon Hákonarson in the mid-13th century. His tombstone can still be seen there. This Historia brought the past of this peripheral country to the knowledge of European readers. However, Torfæus was regarded by some critics already in his own time, and by most after a few decades, as reading the sagas with too uncritical an eye. Only 60 years after Torfæus’ Historia was published, Gerhard Schøning, professor of history and eloquence and later to become Royal archivist, in his Norges Riiges Historie (‘History of the Norwegian realm’) described Torfæus’ work as merely a first step (1771, fortale p. 6–7): [Torfæus’ Historia] … does not deserve to be called a History of the Norwegian Realm, but rather a collection or magazine for that history’s further preparation. […] to bring what he has collected into the proper arrangement; to ascribe each event to its correct time and place; in short, to write from this the History of the Norwegian Realm, all this he has left to others. The present mono-



2 Skre: Exploring Avaldsnes 1540–2005 

 17

graph that I have taken on the task of writing should therefore in reality be the first to be called the History of the Norwegian Realm.8

The assessment of Torfæus as an uncritical compiler of evidence was reinforced in 1916 by Kristian Kålund in his publication of the letters between Torfæus and Árni Magnússon. Regarding Torfæus, Kålund (1916b:x–xi) stated: […] he should hardly be called a scholar; he had not undertaken thorough studies. […] His critique of the evidence was less developed and he shared the superstition of his time; omens and dreams were meaningful to him, as well as the determination of fate by means of horoscopes.9

A few years after Kålund’s work was published the then-leading historian of Norway, Halvdan Koht, wrote that Torfæus’ work was ‘a reproduction, lock, stock, and barrel, from all kinds of evidence, but without any critical assessment and without any other intention than writing a mere chronicle’10 (Koht 1929:38–40). As late as in 1995 the Latinist Inger Ekrem wrote that Torfæus ‘was well versed in Old Norse manuscripts, whose reliability he trusted blindly’ (1995:81). A more balanced critique than most since Schøning’s is A.O. Johnsen’s 1969 description of Torfæus’ work. In line with P.A. Munch, his assessment does not hold the work to modern standards but to those of Torfæus’ own time. Still, they found nothing of historic value in his writings. Only the last few years have seen more positive assessments of Torfæus’ work, in particular his methods, notably following the Historia’s first translation into Norwegian (Torfæus 2008). Although some knowledge of Latin was mandatory for historians until some 30–40 years ago, Torfæus’ vocabulary and nuances of expression surpass even the competence of modern specialists (Kraggerud 2008). The Latinist Vibeke Roggen (2003:93) writes, half-jokingly, that scholars who have expressed strong opinions on Torfæus’ Historia are probably more numerous than those who have actually read it. Most historians from Schøning onwards have overlooked the distinction that for Torfæus, retelling the sagas was not the same as vouching for their accuracy. In fact, his writings critically assessed the saga’s historic value. At the same time, aspects of his method deviated from that which was developed by Árni, Schøning, and subse-

8 “…fortiener ikke saa meget Navn af det Norske Riiges Historie, som heller af en Samling eller et Magazin til bemeldte Histories viidere Udarbeidelse. […] at bringe, hvad af ham var samlet, i behørig Sammenhæng; at henføre enhver Ting til sin rette Tiid og sit rette Sted, kort, at forfatte deraf det Norske Riiges Historie, det har han alt sammen overladt til andre at udføre. Nærvarende Skrivt, som jeg har foretaget mig at udarbeide, skulde altsaa blive den første, eegentlig saa kaldet, Norske Riigs-Historie.” 9 “…videnskabsmand kan han næppe kaldes, lige så lidt som han havde foretaget dybere gående studier […] Hans kritik var lidet udviklet, og han delte tidens overtro; varsler og drømme havde betydning for ham, så vel som horoskop-beregning til bestemmelse af skæbnen.” 10 “en gjenfortelling av rubb og stubb etter all slags kilder, men uten kritisk siktning og uten tanke på annet enn den rene krønike.”

18 

 A: Scholarly Background

quent historians. Rather than rejecting many stories outright as historical evidence, Torfæus insisted that most of them must contain some kernel of historic value; if not, why should generations of Icelanders passed them down orally and eventually had them written down? This crucial difference in understanding between Torfæus and subsequent historians resulted in a distinct contrast in their approach to the sagas. Rather than simply peeling away untrustworthy elements in search of historically credible pieces of information, Torfæus identified mythical and legendary elements and offered his interpretations (Roggen 2003:99–100; Jørgensen 2008). Apart from a passage that paraphrased Peder Claussøn Friis’ account (Torfæus 2008:173) the passages in Torfæus’ Historia that deal with Ǫgvaldr and Avaldsnes are based on Torfæus’ own observations as well as on the fornaldarsǫgur and the more historically rooted sagas written by Oddr Snorrason munkr and Snorri (see Mundal, Ch. 3). The site is first mentioned in his discussion of whether the ancient sagas are fictional or true. He proposes that grave monuments and raised stones may be studied for corroboration of the sagas: (Torfæus 1711, introduction; 2008:62): […] the Histories of Half the Hero and Olaf Tryggvason refer to two such stones, erected in memory of Ǫgvaldr Rogius and a cow, which he had worshipped during his lifetime, and which he ordered to be buried near him after he died. The stones stood in Kormti, where I myself live, on the priest’s estate of Ǫgvaldsness, and stood in our time on either side of the church, one of them may still be seen today, and they bear out some part of the ancient account; this makes it absolutely certain not only that this Ǫgvaldr existed but also that what is recorded about him in the recently discussed histories is true.11

Further details about the stones are given in a subsequent passage about Ǫgvaldr and the cow he worshiped, and their burial in the two mounds there (Torfæus 1711, introduction; 2008:360): Evidence of their tombs is found in two stone columns which could until recently be seen there, one rising thirty feet from the ground, and the other twenty-six feet. The first is still standing, but the second burned down in the year 1698, in the accidental fire that destroyed all the houses.12

11 “…memorant Halfi Herois et Olafi Tryggvini Regum Historiæ de binis talibus saxis, in Augvaldi Rogii, et vaccæ, qvam vivus ille coluerat, ac prope se post fata contumulari jussit, memoriam erectis: qvi qvidem, qvoniam in insulæ hujus Kormtis, ubi ipse habito, sacerdotali prædio Augvaldsnesia, ad suum qvisqve templi latus nostrâ ætate, et alter etiamnum conspicitur, et nihil non narrationi priscæ respondet, fidem indubiam faciunt, non modô Augvaldum illum extitisse, sed vera etiam esse ea, qvæ de ipso in memoratis nuper historiis leguntur.” 12 “…testes sepulchrorum duæ columnæ lapideæ, qvæ inibi nuper visebantur, qvarum alia a terra longitudinem triginta pedes, alia viginti sex erecta prominuit; prior etiamnum perstat, posterior anno millesimo sexcentesimo nonagesimo octavo, fortuitô incendio simul cum universis ædibus conflagravit.”



2 Skre: Exploring Avaldsnes 1540–2005 

 19

His third reference to the stones appears in his account of the story of Óðinn, disguised as an old man, telling the story about Ǫgvaldr and his sacred cow to King Óláfr Tryggvason (Torfæus 1711:vol. 2, book 9, chapter 17; 2010:93; Mundal, Ch. 3:37–8): An intact tombstone, raised in memory of the king, can still be seen undamaged. But on 14th August in the year 1698, when the whole of Avaldsnes rectory by accident burnt down, the cow’s tombstone, which could not withstand such heat, burst into smaller bits, except for a small base which still stands as an indication of its place. The remaining bits where reshaped and used as steps in the entryway to the church.13

Torfæus tells us that Ǫgvaldr was a fourth-generation descendent of Nor, the first king of Norway, and that he settled at Avaldsnes; that is why the farm was given his name (Torfæus 2008:360–1). He was killed in a battle with another king and entombed in the great mound north of the church. Based on his meticulous genealogy and chronology, Torfæus calculated that King Ǫgvaldr died c. AD 200, actually the very time to which the main grave in Flaghaug has been dated using modern methods (Stylegar and Reisersen, Ch. 22:574; Vea 2004). As Torfæus tells us, 700 years after Ǫgvaldr’s death, the landnámsman Finn the Rich from Stavanger was berthed in the Avaldsnes harbour before setting sail to Iceland. In response to Finn’s enquiries as to when King Ǫgvaldr after whom the farm was named had lived there, a voice from the mound answered with a verse (Torfæus 1711:vol. 1, book 4, ch. 5; 2008:360–1): He then heard a voice deep inside the mound intoning verses with the following sense.

A long time has elapsed since that pack of fierce dogs followed Hækling, sailed over the salty way of the salmon and headed for this place and then I became lord of this house (he meant the burial mound).14

By ‘the way of the salmon’ the voice meant ‘the sea’, Torfæus noted. He did not question the truth of the story, but discussed the name of Ǫgvaldr’s killer. Flateyjarbók named him Dixi, but Torfæus believed that the sited passage from Hálfs saga ok Hálfsrekka was correct in naming him Hækling. To substantiate this conclusion Torfæus

13 “Cippus monumento Regis adstructus integer adhuc visitur, is autem, qvi vaccæ erat, anno MDCXCVIII, die XIV Augusti, cum universæ ædes Augvaldznesiæ fortuito incendio conflagrarent, tanti caloris impatiens in minutiores partes dissiliit, præter exiguam basin, qvæ positus indicium superest, reliqvæ particulæ in usum graduum ædes intrantibus conversæ sunt.” 14 “…vocem intra tumuli penetralia rhythmum modulantem, hoc sensu excepit. Longum qvidem tempus elapsum est, postqvam Hæklingum secutus ferocium canum manipulus, salsam præternavigans salmonum viam, cursum huc direxerat, tunc hujus villæ (tumulum intellexit) dominus evasi.”

20 

 A: Scholarly Background

cited information that he had personally collected on site: “some mounds nearby even today take their name from Hækling, which the natives in their own dialect call ‘Køkling’, which adds credibility to this tradition”.15 In these passages Torfæus combines the evidence from the ancient manuscripts with his own on-site observations. Although some of his conclusions are untenable from a modern perspective, this combining of disparate types of information represented a methodical leap from earlier research practices. Moreover, several of his observations are valuable in themselves and have been used by modern scholars (e.  g., Skre, Ch. 23:642–5, 652–3).

2.2 Enlightenment and rationalism 1771–1862 2.2.1 Gerhard Schøning: Norges Riiges Historie, 1771 The breakthrough of rationalism in the 18th century brought an end to beliefs in the supernatural among the intelligentsia. Torfæus’ younger friend Árni Magnússon rejected such phenomena, as did Gerhard Schøning (Fig. 2.2) in his Norges Riiges Historie. For an example of this shift, compare Torfæus’s recounting of the story about the landnámsman Finn the Rich from Stavanger (above) with Schøning’s. The latter did not dispute the historicity of the information about Ǫgvaldr found in Hálfs saga ok Hálfsrekka and other manuscripts, but expressed scepticism that a voice could have been heard from within the mound. In a footnote he added (1771:282): One may believe that the above-mentioned Finn has heard the verse about Ǫgvaldr when he lay at Ǫgvaldsnæs; not sung from his mound, however, as Finn or others have since pretended, but from an aged skald living there.16

While not withdrawing from assessing the historicity of the account overall, the footnote text signalled the distance to Torfæus’ Historia, as Schøning announced in his introduction (above, pp. 16–17). However, apart from such rather symbolic gestures, Schøning did not bring to bear a systematically critical attitude to the substance of the saga evidence. He argued explicitly against the notion that sagas are untrustworthy legends and tales – a small number are fictitious, he agreed, but most are absolutely trustworthy; fictitious elements are easily distilled from the factitious parts (Schøning

15 “Colles propinqvi hodieqve ab Hæklingo, qvem propria dialecto Köklingum incolæ pronunciant, nomen sortiti fidem huic traditioni faciunt.” 16 “Man kan troe, at benævnte Finn har der faaet anførte Vers om Augvalld at høre, da han laae ved Augvaldsnæs, dog ei sunget i hans Høi, som Finn eller andre siden have foregivet, men af en gammel der boende Skalld.”



2 Skre: Exploring Avaldsnes 1540–2005 

 21

Fig. 2.2: Gerhard Schøning (1722–80) portrayed in a contemporary print. From Hansen 1886:301.

1771:fortale p. 4–5). Furthermore, the passages that concern Avaldsnes are obviously based on Torfæus’ work, which is cited by name (Schøning 1771:279–82). At the same time, as stated in his introduction, Schøning adopted a more analytic approach. In his time, Ǫgvaldr was not unique, Schøning (1771:181, 278–82, 447) maintained, but one of the seafaring pirates of that period; that is, the 3rd–4th centuries. He drew a distinction between kings who were land-based and those who preferred to sail the seas; Ǫgvaldr was initially one of the former, and subsequently became a sea pirate. As land king he defeated enemies and conquered land until his rule covered the entire region of Rogaland. Having done so, his position was bound to be challenged, Schøning wrote (1771:279–80): […] for which reason a residence in the islands was most suitable; more so since he found his greatest joy in war and piracy, on which he constantly ventured. The ancient writers have noted that on these undertakings he always brought a cow and that with two intentions: He drank her milk as an unusually powerful medicine or refreshment, and besides he cultivated or sacrificed to her as a deity. […] But neither Ǫgvaldr’s courage nor the alleged deity could secure him against enemy attacks, and from that followed a violent death.17

17 “…hvorover Boepæl paa Øerne var ham beqvemest; allerhelst da han for Resten fandt sin største Fornøielse i Krig og Søerøver-Toge, paa hvilke han iidelig sværmede omkring. De gamle have anmærket, at saavel paa desse Søerøver-Toge, som paa andre Reiser, førte han stedse en Koe med sig, og det i dobbelt Henseende: Hendes Mælk drak han, som en usædvanlig kraftig Lægedom eller Forfriskelse; og for Resten dørkede eller offrede han til hende, som en Guddom. […] Men hverken Augvalds eegen

22 

 A: Scholarly Background

Although he did not reject the historicity of the story about Ǫgvaldr’s cow, Schøning aimed his rationalistic critique at what he took to be unfounded superstition: this ‘alleged deity’ could not protect Ǫgvaldr. Schøning’s observations about sea-kings in the 3rd–4th centuries and the reasons for their settlement on islands are more interesting: firstly, that islands were more easily defended, and secondly, that Ǫgvaldr could more easily attain ‘joy in war and piracy’ from an island base. The cited passage is an apt example of how Schøning was more adept than Torfæus both in analysing the accumulated evidence, written and topographic, and in drawing general conclusions – definitely a leap in ambition and scope from Tomod’s research practise. This difference between them is profound; as the Icelandic philologist Véstein Ólason (2006:109) claims, Torfæus was less a historian and more a writer of Norwegian history in the tradition initiated by Sæmund and Ari in the early 12th century: In that sense we may claim that the heyday of the ancient Icelandic saga writing had, if not its final phase, at least an epilogue here in Karmøy.18

2.2.2 Antiquarians of the 18th and early 19th centuries In Schøning’s time and well into the 19th century topographic and antiquarian writings were highly popular. Several regions in Norway were described according to the standards of the topographic genre – as explicitly laid down by Peder Claussøn Friis in the early 17th century (above). In 1745, County Governor Bendix Christian de Fine published a topographic description of Stavanger County that contained information regarding the raised stone north of the church (Skre, Ch.  23:646). Additionally, de Fine briefly retold the stories from Heimskringla that were connected to Avaldsnes. One passage in particular demonstrates that he had reflected on the position of the royal manor. After stating that some 10th-century kings stayed there, he writes (de Fine [1745] 1952:45): Also, because of the site’s particularly convenient location, some of the subsequent kings have resided there, from whence they could set sail as they wished and be ready to ward off all hostile incursions into the country.19

Tapperhed, ei heller denne formeente Guddom, kunde sætte ham i Sikkerhed mod fiendtligt Angreeb, og derpaa fulgte voldsomme Endeligt.” 18 “I den forstand kan vi hevde at den gamle islandske historieskrivningens storhetstid fikk, om ikke sin avslutning da i hvert fall en epilog her på Karmøy.” 19 “Desforuden har der Resideret een og anden af de efter følgende Konger, for Stædets særdeeles beleilige Situation, hvorfra de kunde komme til Søes naar de vilde, og være færdige til at afværge alle Fientlige indfald i Landet.”



2 Skre: Exploring Avaldsnes 1540–2005 

 23

Probably, this assessment of Avaldsnes would have been inspired by the role royal naval bases played in warfare in de Fine’s own time; they were aimed at warding off external enemies. Bishop Peder Hansen’s account, published 1800, contained significantly more information on Avaldsnes. King Ǫgvaldr ruled over Rogaland in AD 316–3020 Hansen stated (1800:261). His text appears to have been based on Torfæus’ Historia and Schøning’s Norges Riiges Historie, but also incorporated pieces of information that he had obtained from a local priest or collected himself on the site (Hansen 1800:259). Hansen introduced one new piece of information that is not known from any other sources: digging in the cemetery, 16 paces south of the St Óláfr’s Church, he writes, revealed the foundations of an octagonal chapel, 34 ells in circumference (Bauer, Ch. 14:295–7). Hansen introduced a theme that came to be the main point of discussion in the following years; namely, in which mound was Ǫgvaldr entombed and which housed the remains of his cow? Hansen held that the southern Kjellerhaug was Ǫgvaldr’s and the northern Flaghaug his cow’s, each mound with an associated memorial stone (Hansen 1800:263–4; Skre, Ch. 23:646). The topographer Jens Kraft, who placed Ǫgvaldr in the 7th century, agreed with Hansen on the location of the cow’s mound, but placed Ǫgvaldr’s mound at Kongshaug, an elevated ridge about 150 metres west of the church (Kraft 1829:266–7; 1842a:124–5). Kraft (1829:224) also claimed that Skratteskjera, the skerries where Óláfr Tryggvason drowned some sorcerers in AD 997 (Mundal, Ch. 3:38–40), were “Fladeskjær” nearby. Both of Kraft’s propositions were rejected by Johan Koren Christie (1842:326–32), who served as tutor in the Avaldsnes rectory around 1840 and thus had detailed knowledge of the site (Østrem 2010:208–9). Regarding Ǫgvaldr’s grave Christie based his argument on the 1834–5 excavation of Flaghaug, the great mound north of the church (Stylegar and Reiersen, Ch 22). The excavation clearly demonstrated that this was a royal grave for a male person, and Christie concluded that it was Ǫgvaldr’s. The cow, he argued, was buried in the Kjellerhaug mound just south of the church. Regarding Skratteskjera, Christie argued that “Fladeskjær” was not flooded by the tide and thus would not have served the purpose of drowning the sorcerers. Likewise, the King’s men would have chosen the nearest convenient spot, which was “Pibeskjærret”, a skerry in the Avaldsnes harbour that was covered by two feet of water at high tide. This skerry would have ample room for a boat crew – even for two if they were all lying on their backs, Christie argued (1842:339–40). Christie also discussed the location of the Viking Age royal farmyard. He suggested that it should be sought south of the present rectory buildings. If it had been lying where the rectory buildings currently lie, he argued, there would not have been

20 His text says AD 1316–1330, but since he also writes “i det 4de Seculo” (‘in the 4th Century’), the years are obviously a misprint.

24 

 A: Scholarly Background

sufficient room for Erlingr Skjálgsson’s 1,500 men who, according to Heimskringla, surrounded King Óláfr inn helgi when he walked from the church to his chambers (Christie 1842:334; Mundal, Ch 3:40–3). Christie also argued that the octagonal remains described by Bishop Hansen must have been from the church that Óláfr Tryggvason built in 997. Avaldsnes would be an optimal location for convening for chieftains, Christie (1842:322–3) reasoned: Whether they came from the north […] or the south […] the Vikings travelled past the site. For beyond Kormt (Karmø) is the wild ocean, and the coast is shut off by an exceptionally dense and dangerous series of skerries with seething breakers, so clearly, sailing west of the island was not a preferred choice […] The site’s local advantages were noticed early on by the region’s petty kings, and far back in historic time we find Ǫgvaldsnes mentioned as a royal manor in the obscure legends of antiquity.21

Christie is the first to describe the natural conditions that led sailors to prefer the Karmsund Strait for sailing west of Kormt. Although not explicitly, Christie seems to consider the military advantage of holding Avaldsnes; the site allowed the exertion of power over sea travellers. The minutely detailed analyses of saga evidence, monuments, and landscape features undertaken by these antiquarians was in line with Torfæus’ combination of saga information with his own observations. Similarly, their profound confidence in the saga evidence resembles that of Torfæus and Schøning. Still, their painstaking argument and involvement of a wider scope of evidence marked a new leap in research practice. Also, their rational and thorough empirical argument, in particular Christie’s, pointed ahead to the academic historic research of the late 19th century.

2.3 The Age of Academic Exploration 1862–2005 In late 19th-century Avaldsnes research, as in historical scholarship in general, a shift took place from the antiquarian tradition to the academic. In this process monuments and sites became less important, while the approach to the written evidence became increasingly critical and refined. This latter development continued into the 20th century. Thus, just a few decades after the publication of Christie’s paper, his and his

21 “Herforbi droge Vikingerne, vad enten de fore nordfra […] eller søndenfra  … Ti udenom Kormt (Karmø) er det vilde Hav og Stranden indsluttet af en usædvanlig tæt og farlig Skjærkjede med frådende Brændinger, saa Seiladsen vestom Øen, som rimeligt, ikke gjerne valgtes […]. Stedets lokale Fordele vare da og tidlig bemærkede af Egnens Smaakonger, og langt opover den historiske Tid finde vi i Oldtidens omtaagede Sagn Augvaldsnæs nævnt som Kongsgaard.”



2 Skre: Exploring Avaldsnes 1540–2005 

 25

predecessors’ confidence in the sagas was brutally shaken, never to reappear among scholars. The two categories of sagas that mention Avaldsnes, the fornaldarsǫgur and the konungasǫgur (Kings’ sagas that deal with the period c. 850–1100, e.  g., Heimskringla) had rather different fates in this process. The former continued to be used as evidence for pre-850 history up to the 1860s, even by Norwegian academic historians such as Rudolf Keyser (1803–64) and P.A. Munch (1810–63), although based on a much more critical assessment than that of Torfæus, Schøning, and the antiquarians (Dahl 1990:70–5). This use of sagas by scholars rapidly faded in the late 19th century, partly due to a rather short thesis by the Danish historian Edwin Jessen, Undersøgelser til Nordisk Oldhistorie, published in 1862. In clear prose and with a polemic tone he dismissed the fornaldarsǫgur as useless in terms of historic evidence (Jessen 1862; Helle 2001:16; Krag 2006:91–2), due to internal contradictions and inconsistencies. If there is a historic core, there is no way of identifying it, he stated. The critical assessment of the konungasǫgur underwent further refinement over the course of the 19th century, in Norwegian scholarship most prominently by Gustav Storm (1845–1903; Helle 2001:18). While Keyser and Munch worked on solving contradictions and cleaning out mistakes and misunderstandings in the sagas, Storm (1869; 1873) introduced the notion that saga writers were authors in their own right and not mere compilers of oral tradition. Thus, Storm supplemented Keyser’s and Munch’s methods for distilling reliable evidence from the sagas with a path toward assessment of the sagas in light of the interests, personalities, and circumstances of their authors. In the early 20th century a thesis by the Swedish historian Lauritz Weibull proposed a radical break with the positive assessment of the source value of the konungasǫgur shared by the vast majority of scholars at the time. In Kritiska undersökningar i Nordens historia omkring år 1000 published in 1911 he rejected the main themes and grand lines in the saga accounts and asserted that their sole value lay in the scattered bits and pieces of reliable information that might be identified using the strict methods that he had developed. Although Weibull had lasting impact, his methods proved less suitable for actual research (Bagge 2014:595). His negative assessment of the historicity of the sagas was soon to be challenged by research into the nature of oral tradition. By applying insights from his own discipline to the sagas, the folklorist Knut Liestøl (1929) analysed the mechanisms by which historical events were transformed in the Icelandic oral tradition on which the written sagas were based. Liestøl, soon joined by the historian Halvdan Koht, believed it possible to use insights in the transformation of oral tradition to peel away fiction, and trust what was left (Helle 2001:21–4; Bagge 2014:581). The joining of these two scholars’ perspectives on the sagas, inspired in some aspects by Weibull, still forms the methodological backbone of Norwegian saga research, and has had significant international impact. For instance, the methods that Else Mundal applies in her analyses of the saga evidence in the present volume (Ch. 3; see also Skre, Ch. 27:761–4) are rooted in this tradition.

26 

 A: Scholarly Background

2.3.1 Post-Weibullian Avaldsnes research Rikssamlingen (‘The unification of the realm’) was a major topic in 20th-century Viking Age studies, and Avaldsnes was mentioned repeatedly.22 Johan Schreiner (1929:22–4) accords Avaldsnes a key role in royal control of western Norway in the 10th–13th centuries. He pointed to the ease with which sea traffic could be controlled in the narrow Karmsund Strait, and drew a line from the prominent Bronze Age finds through those from the early Iron Age to the two Viking Age ship graves. The long continuity of rich finds, he stated, can only be explained thus (Schreiner 1929:24):23 This major manor was a constant threat to sea traffic along the sailing route, a threat to chieftains further north, who depended on sailing the Karmsund Strait. The need to secure the coastal trade route was the basic factor for the political unification of Norway, or rather, the unification of the coastal regions by Haraldr hárfagri and subsequent monarchs.

Haraldr, supported by the earls of Hålogaland further north, conquered Avaldsnes to secure the sailing route (Schreiner 1929:31–2). Schreiner’s assessment was supported by several prominent historians, Andreas Holmsen and Claus Krag among them. According to Holmsen (1949:172) the prospering trade in the 9th century made safe sailing a prominent concern for chieftains along the coast. Haraldr created peace, he noted (1949:178); before Haraldr hárfagri settled at Avaldsnes and his four other manors along the sailing route, ‘the very worst among coastal pirates’ (“de aller verste kystrøverne”) had resided there. Continuing Schreiner’s and Holmsen’s line of reasoning, Per Hærnes (1997:89– 90) and Claus Krag (2000:46) pointed to Avaldsnes’ prominent position in the coastal regions that comprised the core of Haraldr hárfagri’s kingdom, Hordaland and Rogaland. Krag suggested, based on Opedal’s work (below), that when Haraldr took control of these regions they might already have been a chieftain’s realm. Trade and communication along the Norðvegr connected the coastal regions and made military control along the sailing route the key to Haraldr’s power (Krag 2000:50–2). While these historians did not conduct in-depth studies of Avaldsnes, Odd Nordland (1950) presents a thorough and detailed discussion of Avaldsnes’ position at a bottleneck on the Norðvegr. As an ethnologist and philologist, Nordland (1919–99) produced significant works on numerous topics, including Old Norse literature. His Avaldsnes paper was published in 1950, four years after he had taken his master

22 An overview of research is given in Andersen 1977:40–157. 23 ”Denne storgård betegnet en stadig trussel mot skibsfarten i leden, en trussel mot stormennene lenger nord, som var henvist til å seile gjennom Karmsundet. Behovet for å frede handelsveien langs kysten gav grunnlaget for Norges politiske samling, eller rettere kystlandskapene samling under Harald Hårfagre og de følgende landsstyrere.”



2 Skre: Exploring Avaldsnes 1540–2005 

 27

degree. He took his doctoral degree in 1956 and later became professor of Nordic culture and history of religion at the University of Oslo. Based mainly on philological and historic evidence, as well as in his detailed knowledge of the landscape and sailing conditions, Nordland pointed to the ease with which trade and traffic along the Norðvegr could be controlled from Avaldsnes. Applying the majority of methods from the Koht/Liestøl toolbox (summarised in Helle 2001:22–4), Nordland (1950:17–34) discussed in detail all passages in Heimskringla that relate to Avaldsnes, including the various versions and related sagas. In line with Storm and others, he emphasised Snorri’s creative authorship and attempted to identify the rationale behind his combinations of disparate pieces of information to compose coherent stories about events at Avaldsnes. Nordland (1950:34) concluded this discussion: Thus, we may suspect that Snorri’s account of events in the Karmsund Strait rest upon a quite weak historic basis. On the other side, however, we need to ask if he still may be right in the basic outline of his assessment of the Karmsund Strait and the role it played in history. – Although the foundation may be weak, the conclusions may be correct.24

To support the reasonability of what he deemed to be Snorri’s conjectures, Nordland (1950:34–6) referred to the prominent Bronze Age monuments near Avaldsnes, and to Shetelig’s assessment of the Flaghaug find (below, p. 28). He also pointed to the fact that Avaldsnes’ main qualities are military-strategic, not agrarian-economic. In contrast to Bø, the nearby farm to the north with the richest agricultural yield in Kormt, Avaldsnes has the best natural harbour facilities along the Karmsund Strait and lies on the spot with the best view of the sailing route north and south. He called Avaldsnes ‘the warrior manor’ (“krigargarden”, 1950:38). Nordland agreed with Professor in Old Norse Philology Magnus Olsen’s (1913) reading of Haraldskvæði, composed by the skald Þorbjǫrn Hornklofi c. AD 900. Olsen’s paper strengthened the common assumption among historians at the time, based in Haralds saga ins hárfagra (ch. 38) as well as Egils saga Skalla-Grímssonar (ch. 36), that Haraldr hárfagri had five royal manors, and that Avaldsnes was his main residence. Olsen argued that the first lines in stanza 5 in Haraldskvæði were misspelled in both versions of the saga-redaction Fagrskinna. The B version has: Kunna hugðak þik konung, þanns a kvinnum býr, dróttin Norðmanna (“I thought you knew the King, who lives at Kvinne, King of Northmen”), while the A version has “a kymnum” (Olsen 1913:66). Olsen argued that the writers of both manuscripts had misunderstood their sources, and that the correct phrase should be “i Kormtu (‘in

24 ”Det Snorre fortel om einskild-hendingar i Karmsundet, kan me difor mistenkja for å stå på eit temmelig veikt historisk grunnlag, men me lyt på den andre sida spyrja oss om han ikkje likevel kan ha rett i hovuddraga i den vurderinga han har av Karmsundet og i den rolla han let det spela i soga. – Om grunnlaget et tunt, kan slutningane vera rette.”

28 

 A: Scholarly Background

Kormt’)”; he concluded that Haraldr hárfagri’s “true residence, the first ‘capital’ of Norway, had been Avaldsnes in Karmsund” (Olsen 1913:72).25 Olsen (1913:70) reinforced his argument concerning Avaldsnes’ prominence by referring to the unanimous evidence of the sagas that Haraldr was buried “at Haug by the Karmsund Strait”26, across the strait from Avaldsnes. In the sagas he found several examples of prominent men being buried some distance from their residence. Nordland (1950:40) agreed with Olsen and added that the belief common in more recent times, that the dead would not haunt the living if they were separated by water, may have been held at the time. An additional indication of Avaldsnes’ early prominence, writes Olsen (1913:70) and Norland (1950:36), was the richly furnished grave in Flaghaug described by Haakon Shetelig (1912a:53–9; Stylegar and Reiersen, Ch.  22). Based on the three Roman vessels in the grave, and the concentration of such vessels in the close vicinity together with their distribution along the coast, Shetelig suggested that Avaldsnes was a centre of distribution of these and other objects as well as the new inhumation burial custom. He highlighted that this distribution centre was located at ‘one of the most important points along the coastal sailing route’27 (1912a:58–9). Both Olsen and Nordland commented on the fornaldarsǫgur. In the legend of King Ǫgvaldr and his cow, Olsen (1913:70) found an indication that Avaldsnes “probably played a prominent role as a royal seat also in prehistoric times”28. The only firm piece of evidence Nordland found in the fornaldarsǫgur’s Avaldsnes accounts was that Ǫgvaldr was the actual name of a person who gave his name to the manor, whereas the names of most persons in these stories are derived from place-names and are thus unhistorical. He interpreted Ǫgvaldr as he who ‘rules by fear’29 (Nordland 1950:42). Nordland points to the telling fact that some farms situated near narrow straits on sailing routes or at an important eið (‘isthmus’) across which people, cargoes, boats and sometimes ships could be hauled, bear names with a person’s name as prefix – Ǫgvaldsnes is but one example. Such names at isthmuses have the suffix -eið and those by straits have the suffix -nes (‘promontory’). Nordland (1950:47) commented: “The name conveys the sense of a distinct will by the route, a will of a kind that made people remember names of persons.”30

25 “Harald haarfagres egentlige residens, Norges første ‘hovedstad’, har været Avaldsnes i Karmsund.” 26 “á Haugum við Karmtsund” 27 “et av de viktigste punkter i hele kystleden”. 28 “sandsynligvis har spilt en fremtrædende rolle som kongssæte ogsaa i forhistorisk tid”. 29 “herskar med otte”. 30 ”Ein får ei viss kjensle av at ein i dette namnet møter ein vilje attmed leia, ein vilje av eit slag som fekk folk til å hugsa person-namn”.



2 Skre: Exploring Avaldsnes 1540–2005 

 29

Out of about 700 farm names with the suffix –nes, ten have personal names as a prefix; of these ten, Nordland found that the majority lie at strategic points along routes. At one of them, Sotenäs in Båhuslän, Nordland found the same legendary saga themes as at Avaldsnes. Sóti was defeated by King Óláfr inn helgi, and from his mound he acted among the living. Nordland suggested that the naming of nes after men indicate that they were neskonungar (‘promontory kings’). The word is known in writing already in the early 13th century, and Nordland (1950:49–53) believed that it denoted a genuine and accurate historic tradition. He connects this tradition to a term used by the two skalds Þorbjǫrn Hornklofi c. AD 900 and Eyvindr skáldaspillir in the mid-10th century: Holmrygir, meaning ‘Island Rugii’. Rugii is the tribal name that forms the prefix of the name of the region where Kormt is located; Rogaland. The Holmrygir, writes Nordland (1950:53–4), would be Rugii that resided on easily defendable islands along the sailing route. This gave them a strategic benefit that was used to extract income from travellers sailing past. Among the scholarly works on Avaldsnes discussed above, Nordland’s paper is the most exhaustive and of the widest scope; his method and assessment of the evidence is most closely in accordance with modern standards. Although previous scholars, such as Christie, had mentioned Avaldsnes’ position on the sailing route along the Norðvegr, Nordland was the first to qualify and develop the idea, emphasising the possibilities for controlling the seaway’s traffic from Avaldsnes and for keeping watch over the sea route from the elevated settlement plateau. His conclusions are summed up in his characterisation of Avaldsnes as ‘the warrior farm’. Regarding Haraldr hárfagri and subsequent kings, Nordland’s assessment of the Old Norse literary evidence is still valid, although not undisputed – the same goes for Olsen’s discussion of Haraldskvæði (e.  g. Fidjestøl 1993:18). Regarding earlier periods, Nordland drew to the fore the concept ‘promontory kings’ and other indications that Iron Age kings purposely exploited the advantages inherent to island residence by extracting revenues from the seaway’s travellers.

2.3.2 Local interest in Avaldsnes 1935–89 On the periphery of the post-Weibullian scholarly endeavours, interest in Avaldsnes’ history was maintained by intellectuals living in the vicinity; three of them will be mentioned briefly. The journalist and author Heming R. Skre (1896–1943) connected Avaldsnes to what he considered to be a universal phenomenon of the past: the worship of the sun. He was a proponent of the opinion that ancient monuments and sites with cultic names in the region were organised along an elaborate system of lines that had their centre in the raised stone at Avaldsnes, Jomfru Marias Synål (‘St. Mary’s Sewing Needle’; Skre, Ch. 23). For the church’s 700-year anniversary in 1950, the Avaldsnes priest Lars Skadberg (1898–1966) published a book about the church and manor. The book is indebted to

30 

 A: Scholarly Background

the topographic traditions, in particular regarding its greatest strength: the author’s extensive knowledge of the sites, landscapes, and traditions. Since the 1980s, the psychologist and author Aadne Utvik (1936–) has worked at raising the local interest in Avaldsnes. Most significantly, in 1988, he published a series of articles in the local newspaper, which were later assembled in a book titled Vårt historiske Avaldsnes (‘Our historic Avaldsnes’). This series, together with the recent discovery of a subterranean passageway near the church (Bauer, Ch. 14:304–6) spurred considerable interest at the time. In 1989 the Archaeological Museum in Stavanger published a book aimed at the general public titled Avaldsnes, Norges eldste kongesete (‘Avaldsnes, Norway’s first royal seat’), which also was well received. The social and cultural role of all historical studies, including the strictly academic – namely, their relation to current concerns – is particularly visible in the non-specialist publications by H. Skre, Skadberg, and Utvik. Indeed, it was precisely local awareness and enthusiasm for Avaldsnes’ historic significance, evident in these writings, that initiated the boom in Avaldsnes research since 1993, ultimately leading to the research project that has resulted in this book (Skre, Preface).

2.3.3 A new beginning, 1993–2005: The Karmøy Seminars and Opedal 1998 In 1993, Karmøy Municipality established the Avaldsnes Project (Skre, Preface), initiating research on a variety of aspects connected to Avaldsnes. They developed a productive collaboration with Stavanger Maritime Museum, who conducted extensive underwater surveys in the Avaldsnes harbour, revealing mainly late-medieval and modern deposits and finds (Elvestad and Opedal 2001; Bauer and Østmo, Ch. 5:69). The archaeologist Per Hernæs (1997) contributed the volume on the prehistoric era to the three-volume Karmøys historie. In preparation for the St Óláfr’s Church’s 750th anniversary in 2000, the Avaldsnes Church Council called upon prominent scholars to write a book about the church, published in 1999 as Kongskyrkje ved Nordvegen (‘A royal church by the Norðvegr’, Langhelle and Lindanger 1999). The Bronze Age monuments in northern Kormt (Skre, Ch. 27:750–2) were explored by Lise N. Myhre in her 1999 book Historier fra en annen virkelighet (‘Histories from another reality’). Beginning in 1993 the Avaldsnes project organised a scholarly seminar series, Karmøyseminaret (‘The Karmøy Seminar’), on topics related to Avaldsnes’ history. The seminars, seven in the period 1993–2004, attracted leading Scandinavian and some international scholars, and were published by the municipality (Vea and Myhre 1993; Krøger and Naley 1996; Krøger 1997, 2000; Naley and Vea 2001; Jacobsen 2004; Kongshavn 2006). The seminars covered a wide array of topics, from Iron Age social and political structure to medieval trade, as well as the work of Tormod Torfæus. Several significant papers were published and are indeed referred to in relevant chapters in the present book.



2 Skre: Exploring Avaldsnes 1540–2005 

 31

Fig. 2.3: Professor of Archaeology, Bjørn Myhre (1938–2015). Photo: Terje Tveit, AM.

While Schreiner, Holmsen, and Krag looked to the north when discussing Avaldsnes’ significance in Viking Age politics, some contributions in the seminar publications look to the south and west, beyond what was to become the Norwegian Kingdom. Although Holmsen (1949:178) had noted that Haraldr – through his conquest of Avaldsnes and other manors – acquired treasure collected overseas by Vikings, the international perspectives introduced by Bjørn Myhre (1993) and developed by Egon Wamers (1997) and Arnfrid Opedal (1998, 2005) marked a significant shift in research perspective. Bjørn Myhre (1993; Fig. 2.3) sought the background for the 10th-century Norwegian Kingdom in the 7th–9th-century south and the west. Anglo-Saxon and Carolingian lords would have sought control over the trade in prestige commodities produced in the north, such as fur and walrus tusk. Since the 7th century the Karmsund Strait was the gateway to the resources in the north, and since the 8th a point of departure for traffic towards the British Isles and Ireland in the west. Thus, in Myhre’s thesis, the process that historians call ‘the unification of the realm’ should be regarded as the final phase in a development that began in the 7th century. During the three centuries preceding Haraldr hárfagri’s success there may have been several failed attempts to unify realms, and short-lived polities may have emerged only to dissolve with few or no traces (Myhre 1993:57–60). Primarily discussing the 9th–10th centuries, the archaeologist Egon Wamers (1997:18–21) considered the resistance to Danish dominance to have constituted a formative force in the creation of the kingdom of Haraldr hárfagri and his sons. The kingdom’s economic, military, and political base was built up in the lands he assumes they held overseas in the British Isles and Ireland. He suggests that the absence in

32 

 A: Scholarly Background

Rogaland of 10th-century imports may be connected to the expulsion of the Northmen from Dublin in 902; if so, Rygir (people from Rogaland) must have held prominent positions there. The close relations in the 9th century to Danish, Frankish, Anglo-Saxon, and Irish kingdoms, writes Wamers, inspired the creation of a Norwegian kingdom. Although these two and other significant papers that contextualised Avaldsnes appeared in the publications from the Karmøyseminaret, few if any of them contribute new information regarding the archaeology of the Avaldsnes area. That lacuna was soon to be filled, however, by two books by the archaeologist Arnfrid Opedal (b. 1965). After earning her master’s degree in 1994 with a thesis on the national undercurrent in Norwegian settlement research 1905–55, Opedal was affiliated with the Archaeological Museum in Stavanger. There she undertook research on the two ship graves near Avaldsnes: Storhaug, excavated by Andres Lorange in 1884, and Grønhaug, excavated by Haakon Shetelig in 1902. Opedal published two books in 1998, De glemte skipsgravene and Makt og myter på Avaldsnes (‘The forgotten ship graves’ and ‘Power and myths at Avaldsnes’), and in 2005 developed her research into a PhD thesis. The thesis was published in a revised form 2010 under the title Kongemakt og kongerike. Gravritualer og Avaldsnes-områdets politiske rolle 600–1000 (‘Royal Power and realm. Grave rituals and the Avaldsnes area’s political role 600–1000’). Opedal’s two books represented a leap in Avaldsnes research. On the basis of detailed analyses of the finds and the documentation from the Storhaug excavations, she argued that the grave rituals were intended to stabilise and mend the societal crisis following the king’s death by connecting to the world of the gods and bestowing the new king with power. While Opedal (1998:65) dated Storhaug to the late Merovingian Period (680–800), she concludes that the Grønhaug grave was built in the mid-10th century (1998:75). She places the Grønhaug funeral in the political calamities of that period, more specifically in an alliance between the Norwegian and the Anglo-Saxon kings to fight Danish expansion in Norway and England. She suggests (1998:200–4) that the grave is possibly that of Haraldr hárfagri. The connection drawn by Opedal (1998:109–40) between the two ship graves concerns kingship; she interprets the ship graves as manifestations of attempts to establish kingdoms in the region modelled on the Frankish kingdom. In this scheme, the polity was a Personenverbandsstaat centred on a mobile king and his retinue, the loyalty of which was maintained by gifts of luxuries and land. The Storhaug grave indicates the establishment around AD 700 of such a kingdom stretching from northern Rogaland to southern Hordaland. She points out as the kingdom’s nodes other coastal manors, and she stresses that sea travel connected the sites and facilitated kingship. This regional kingdom may have constituted the core from which Haraldr hárfagri expanded to establish a Norwegian kingdom in the decades around AD 900. Although the two ship graves in Kormt had been known by specialists, Opedal’s work brought them to the attention of the general community of scholars of the Viking Age, and she took the step of situating the ship graves within current strands of



2 Skre: Exploring Avaldsnes 1540–2005 

 33

research. This increased interest is mirrored in Frans-Arne Stylegar and Niels Bonde’s (2009) dendrochronological dating of Storhaug to 779 and Grønhaug to 790–5, and to Jan Bill’s (2015) discussion of the ship-grave custom, followed in 2016 by Stylegar and Bonde’s discussion of the same theme.

2.4 Why did kings settle in Kormt? The various strands of research into which Avaldsnes has been included seek answers to this chapter’s introductory question: why did the kings choose to settle in Kormt? Absalon Pederssøn Beyer (1567–70) was the first to address Avaldsnes’ position along the coastal sailing route, although not explicitly. His use of the term neße konger (‘promontory kings’), later used by Nordland (1950), was probably inspired by the military advantages he deduced from the maritime setting of Avaldsnes and other royal manors along the coast. Although Tormod Torfæus (1711) repeatedly mentioned sea travels in connection with Ǫgvaldr and other kings, juxtaposing them with land-based kings – Ǫgvaldr was initially one of the latter – he did not explicitly address this significance of this difference. That is rather in keeping with the scope of his scholarly inquiry; retelling and interpreting the original meaning of the old stories, rather than analysing them for general patterns. Gerhard Schøning (1771) was the first to identify Ǫgvaldr as a representative of one type of king, namely a pirate-king that roamed the seas. Ǫgvaldr’s predilection for piracy and his consequent need to protect his possessions are the two reasons he and other sea kings settled on Kormt, an easily defended island. Some years previously de Fine (1745) had identified a related rationale for Viking and Medieval Period kings settling at Avaldsnes, namely the site’s favourable position for defending the realm from external enemies. Christie (1842) was the first to describe how sailing conditions east and west of Kormt made Avaldsnes a favourable position for controlling traffic along the coast. Christie, along with other 19th-century writers such Hansen (1800) and Kraft (1829; 1842a), was interested primarily in interpreting the Avaldsnes monuments in light of the Old Norse literature. By the 20th century, however, scholars began to involve the site in their discussions of the creation of the Norwegian kingdom in the 10th–11th centuries. Based on the maritime position of Avaldsnes and the other early royal manors, Schreiner (1929) suggested that the need by chieftains further north for securing the coastal sailing route and defending against piracy was for Haraldr hárfagri the primary motive for unifying the realm. This idea was adopted more or less unaltered by Holmsen (1949) and Krag (2000). Although they do not refer to Olsen’s (1913) reading of the poem Haraldskvæði from c. AD 900, which identified Avaldsnes as Haraldr’s most prominent residence, their thinking appears to have been influenced by his analysis of it.

34 

 A: Scholarly Background

Nordland (1950) extended and deepened the line of thinking from Olsen and Schreiner. He built a detailed and complex, albeit conjectural argument, concluding that Snorri’s repeated references to Avaldsnes as the main manor of Haraldr and subsequent kings reflected a past reality. Nordland picked up Christie’s and Schreiner’s idea that Avaldsnes’ prominence was not due to its agricultural yield, but the threat it represented towards sea travellers. Poignantly, in reference to the ease with which passing ships could be cut off and the route could be observed from the elevated settlement plateau, Nordland called Avaldsnes ‘the warrior farm’. With a different perspective from that of Schreiner, Holmsen, and Krag, Myhre (1993), Wamers (1997), and Opedal (1998) saw the creation of a possible 8th–9th-century regional kingdom and a 10th-century Norwegian kingdom as the product of relations, peaceful and hostile, to kingdoms in the south and the west. Myhre’s long-term perspective as well as the international dimension introduced by Myhre, Wamers, and Opedal marked a significant shift in research related to Avaldsnes, pointing ahead to the ambitions of the ARM Project (Skre, Ch. 4). Acknowledgements: I am indebted to Marit Synnøve Vea, Karmøy Municipality and to Professors Emeriti Sverre Bagge and Else Mundal at the University of Bergen, and Professor Jon Gunnar Jørgensen at the University of Oslo for productive comments on an earlier version of this text. I am also grateful to Associate Professor Vibeke Roggen at the University of Oslo for kindly checking and correcting all Latin quotes and translations, and for supplying useful comments to the text. Transcription of Torfæus’ Latin text as well as the English translation have generously been placed at my disposal by Professor Torgrim Titlestad, Professor Emeritus at the University of Stavanger and the leader of Tormod Torfæus-Stiftelsen (‘Tormod Torfæus Foundation’), which organised the translation and publication of the Norwegian edition (2008–14).

Else Mundal

3 Avaldsnes and Kormt in Old Norse Written Sources This chapter provides an overview of references to Avaldsnes and to events occurring there in written sources from the Old Norse period. In written sources from the 13th century, Avaldsnes is named among the five royal estates belonging to Haraldr hárfagri in central Western Norway. In the narratives of an event that occurred during the reign of Óláfr Tryggvason, the god Óðinn appears and recounts the tale of a King Ǫgvaldr who lived on Avaldsnes in the distant past. This story, though of dubious factual reliability, nevertheless demonstrates that saga authors at the end of the 12th and throughout the 13th century were familiar with a tradition in which Avaldsnes was accorded a history as a royal estate beginning long before the time of Haraldr hárfagri. The Norwegian kings gradually took up residence in the towns, but sources for the end of the 13th and throughout the 14th century show that Avaldsnes continued to be a central, albeit occasional, place of residence for the king. From a reading of the late 12th- early 13th- century kings’ sagas, which recount events in Avaldsnes during the reigns of the two Christianising kings Óláfr Tryggvason and Óláfr inn helgi, it is evident that the original stories upon which the saga authors based their works had been modified over the centuries by oral tradition and removed from the historical basis that was perhaps there at the outset. In Óðinn’s tales about King Ǫgvaldr, the events are set in a time long before the reign of Óláfr Tryggvason. Normally, there would not be the slightest reason to have faith in stories about something that happened so long ago. But, in the case of Avaldsnes, where the archaeological excavations have shown that the place was a centre of power some hundreds of years before the unification of Norway, there are, after all, reasons to ask oneself whether the tradition about a King Ǫgvaldr has an historical core that has persisted through the centuries, likely supported by a place name and historical relics, such as burial mounds and memorial stones.

3.1 Kǫrmt Avaldsnes (in Old Norse, Ǫgvaldsnes) is mentioned quite frequently in medieval written sources, often in connection with the name Kǫrmt – the Old Norse for Karmøy, the island upon which Avaldsnes is situated. Avaldsnes was usually referred to as ‘Ǫgvaldsnes á Kǫrmt’, which suggests that the name Ǫgvaldsnes shared to a certain degree those associations evoked by the name Kǫrmt. By way of introduction to the discussion of Avaldsnes in Old Norse sources, it is therefore appropriate to begin with some comments on the name Kǫrmt. The name Kǫrmt occurs not only in narrative saga texts, but also in eddic poems and þulur (i.  e. enumerations of poetic words). For instance, it appears in the first half of strophe 29 of the eddic poem Grímnismál, amongst an enumeration of river names:

36 

 A: Scholarly Background

Kǫrmt ok Ǫrmt ok Kerlaugar tvær þær skal Þórr vaða hverjan dag,… (Dronke 2011)

Kǫrmt and Ǫrmt and the two Kerlaugar (‘tub baths’) shall Þórr wade each day,…1 In Grímnismál it is clear that Kǫrmt is the name of a river in the poem’s mythological landscape. The god Þórr must wade across this river in order to reach the meeting place of the gods by the ash tree Yggdrasil. One explanation for the name’s designation of a locality over which one can wade is that Kǫrmt has also been the name of the Karmsund (‘the Karm Strait’) (Storesund 2012). With its strong current, the long, narrow Karmsund features many riverine qualities, but the semantic content of the name Kǫrmt – derived from the same root as the word karmr (‘low protective wall’) – is better suited to an island, connoting a parapet that offers protection from the ocean beyond. Of the names enumerated in Grímnismál in full, the majority can be identified as names of actual rivers; the three names in the strophe cited above, however, stand out as names of mythical rivers. According to Magnus Olsen, they are the fabrication of the poet, who was familiar with the island name Kǫrmt, but not the actual locality (Olsen 1925). Kǫrmt, Ǫrmt, and Kerlaugar also occur as river names in þulur in all manuscripts of Snorra Edda.2 The heading of these þulur is “á heiti” / “áheiti”; one manuscript (AM 757 a 4to) has the variant “vatna heite”. In the þulur the names are grouped together, which suggests that they were gathered from the eddic poem, where they also appear together. Kǫrmt, however, is also found in the þulur as an “eyja heiti”.3 This could seem quite puzzling. The evident explanation is that Kǫrmt has been used as both an island name and a river name, and other examples are found of this. But it is equally probable that the name Kǫrmt had already acquired, for one reason or another, an almost mythical dimension, and was already being employed in mythological contexts as a name for a type of locality other than its original one. If we have here a name that at one point in time in effect has been elevated into the mythological sphere, this can thus also be of some interest for those associations connected to the name Ǫgvaldsnes, which is so closely linked to Kǫrmt.

1 The translations from Old Norse are my own throughout the article. 2 The þulur in the manuscripts of Snorra Edda are published in Finnur Jónsson 1912–15, AI (text according to the manuscripts with information about which manuscripts have the text) and AII (normalised text). “Á heiti” is found in AI:669–70, in BI:666–7. These þulur are forthcoming in volume 3 of the new edition of skaldic poetry, Skaldic Poetry in the Middle Ages. 3 Kǫrmt is also found as an “eyja heiti” in a þula in AM 748 II, 4to and AM 757 a, 4to, and additionally in one strophe in AM 748 II, 4to. Finnur Jónsson 1912–15, AI:689–90, BI:678–9; AI:652, BI:657.



3 Mundal: Old Norse Written Sources 

 37

3.2 The Royal estate Avaldsnes Snorri recounts in Heimskringla (Haralds saga ins hárfagra, ch. 37), that when King Haraldr hárfagri had reached old age, he often took up residence at his large estates in Western Norway, and then he tallies them in the following order: Alrekkstaðir (now Årstad), Sæheimr (now Seim), Fitjar, Útsteinn and Ǫgvaldsnes. Egils saga Skalla-Grímssonar (ch. 36) contains an enumeration of the same royal estates, here in sequence from south to north. According to 13th-century tradition, Avaldsnes had been a royal estate from the time of Haraldr hárfagri; however, in the saga of Óláfr Tryggvason by Oddr Snorrason munkr, which was originally written in Latin around 1190 but preserved only in Old Norse translation,4 Oddr relates (chs. 43–4) that Avaldsnes had been the seat of kings from time immemorial and that the place was named after King Ǫgvaldr,5 who had lived there long before the reign of Haraldr hárfagri. In Óláfs saga Tryggvasonar (ch. 64) in Heimskringla, Snorri recounts the same story, which he most certainly took from Oddr, but the two stories differ in details. Both authors recount an episode from a time when King Óláfr Tryggvason was in residence on Avaldnes: one evening, an elderly, one-eyed man with a low hat arrived and began to converse with the king. The king interrogated him on many topics, and the visitor could answer all his questions. The king asked whether he knew the identity of the person after whom the ness and the estate had been named, to which the guest answered Ǫgvaldr, a king and a great warrior. Oddr narrates that Ǫgvaldr became very fond of a cow that he always had with him in order to be able to drink her milk, while Snorri adds that Ǫgvaldr performed sacrifices to such a cow. Both recount that King Ǫgvaldr was slain in a battle against a king called Varinn and was buried in a mound in the vicinity of the royal estate, with the cow buried in a mound in close proximity. The morning after the guest had told this to the king, he disappeared, and the king understood that it was Óðinn who had visited them. Oddr relates further that on the fourth day after this visit, the king had the two mounds opened; human bones were found in one of the mounds, and cow bones in the other. Snorri’s account provides a detailed description of Avaldsnes. Both Oddr and Snorri mention the mounds near the estate; Snorri in addition mentions the memorial stones – “those that are still standing”6 – which Oddr does not mention. This could indicate that Snorri had been in person to see the estate on Avaldsnes. Bio-

4 The Old Norse translation of the work of Oddr Snorrason munkr is found in slightly different variations in AM 310, 4to (from the mid-1200s, Copenhagen) and in Sth. Perg. 4to, no. 18 (c. 1300, Stockholm). In addition there exists a fragment, De la Gardie 4–7 (c. 1270, Uppsala). 5 The name exists in slightly different variations. Parallel to Ǫgvaldr (with u-umlaut), we find Agvaldr (without u-umlaut), Ávaldr and the weak form Ávaldi could also be variants of the same name. 6 One of the memorial stones that had stood on Alvaldsnes was ruined in 1698 when the church caught fire, according to Torfæus (2011:93).

38 

 A: Scholarly Background

graphical circumstances also suggest a high likelihood of such a visit. On Snorri’s first voyage to Norway in the late summer of 1218, we do not know precisely where he landed, but it was some place in Western Norway. Over the winter, he resided with Skuli jarl, who was then in Eastern Norway. In the summer of 1219, Snorri travelled to Sweden, and the next winter resided once more with Skuli, who was then in Trondheim. Consequently, we can be quite certain that Snorri sailed past Avaldsnes at least twice.7

3.3 Events during the reigns of Hákon inn góði and Óláfr Tryggvason In the literature of the kings’ sagas, Avaldsnes is the setting for many notable events. In chapter 5 of the Norwegian kings’ saga Ágrip af Nóregskonunga sǫgum, c. 1190, Hákon inn góði is victorious over the sons of Eiríkr blóðox in the battle “í Kǫrmt við Ǫgvaldsnes”. The account of this battle in Ágrip is most probably the source for Snorri’s description of the same battle in Heimskringla (Hákonar saga góða, ch. 19). He specifies as well that the battle occurred “á Kǫrmt / … / á Ǫgvaldsnesi”.8 The battle most probably occurred in a place with conditions suitable for battle, or where the opposing armies happened to encounter one another. Regardless of whether or not the site of the battle was intentional, the relevant question is whether a victory – or a defeat – at a place such as Avaldsnes (according to tradition, an ancient seat of royal power) would garner special symbolic weight and therefore constitute a pivotal moment for King Hákon. In addition to the meeting with Óðinn, the sagas recount another dramatic event on Avaldsnes during the reign of Óláfr Tryggvason. In Oddr’s narration, the king summons to Nidarnes (in Trondheim) many seiðr-men and those skilled in magic. He has a ship made ready for them and orders their departure from the country. Among the exiles was a man called Eyvindr kelda, who Oddr says was the third or fourth man from Haraldr hárfagri. Before the time of their departure, Óláfr Tryggvason invites the men to a great feast and administers copious amounts of food and drink. When the seiðr-men become sufficiently drunk, the king has the building set aflame, and all are burned to death (‘burnt in’) with the exception of Eyvindr kelda, who escapes

7 Information about Snorri’s journeys in the country can be deduced by comparing the text in Sturlunga saga, chs. 185–8, and the text in Hákonar saga Hákonarsonar, chs. 58–62. 8 Snorri cites two strophes from the skaldic poem Hákonardrápa (strophes 6 and 7) by the Norwegian poet Guthormr sindri as the basis for his information about the battle on Avaldsnes, but the skaldic poem mentions neither Avaldsnes nor for that matter a particular battle, stating merely that the sons of Eiríkr often had occasion to witness Hákon’s power.



3 Mundal: Old Norse Written Sources 

 39

through the smoke vent over the fireplace. Oddr refers to Sæmundr fróði as the source for this story. Snorri tells essentially the same story, but he sets the burning-in of the seiðrmen in Eastern Norway. Eyvindr kelda is said to be the grandson of Rǫgnvaldr réttilbeini, one of the sons of King Haraldr hárfagri by the Sami girl Snæfríðr; Snorri by this detail creates a clear parallel between the burning-in of Eyvindr kelda and that of Rǫgnvaldr réttilbeini, who was burnt-in together with 80 seiðr-men by his halfbrother Eiríkr blóðøx under the orders of Haraldr hárfagri. In both variants of the story, Eyvindr kelda sends greetings to the king along with the message that he has escaped death. Both Oddr and Snorri recount that Eyvindr kelda later sought out the king on Avaldsnes. Oddr sets this event during Christmastide, while Snorri sets it during Eastertide. Both describe Eyvindr’s arrival on the island with a large company of seiðrmen and others skilled in magic (sailing in five longships in Oddr’s account, one in Snorri’s). Oddr has Eyvindr plotting an assault on the king and his people to kill them, but as they came up on the island and to the church that the king, bishop and all the people were inside, the sight of the holy church blinded them all. Rendered easy prey, the king’s men took them prisoner, and the next day the king had them taken to a skerry where they were struck down. Snorri’s account of Eyvindr kelda’s arrival at Avaldsnes unfolds slightly differently: Eyvindr covered his men with a cloak of invisibility, creating a great darkness to prevent the king and his men from seeing them. But when Eyvindr and his people arrived at the royal estate, it was as light as day, the darkness overcame them instead so that they were unable to see, and the king’s men easily took them prisoner. The greatest difference between Oddr’s and Snorri’s accounts is found in the story’s conclusion. While Oddr narrates that the seiðr-men were led to a skerry north of the ness where the Karmsund ends, where they were struck down,9 Snorri’s version has them led to an exposed skerry that went underwater at high tide, where they were bound and left to drown. As Finn Hødnebø has called attention to (Hødnebø 1992:129), the punishment meted out to the seiðr-men, according to Oddr, is in accordance with the punishment prescribed for practicing magic in the Older Gulating’s Law (NGL 4:18). There it is stated that those who practice magic shall be taken out onto the sea, struck in the back, and sunk. Snorri’s version of the execution of Eyvindr kelda and his men by means of confinement on a tidal skerry near Avaldsnes was likely his own fabrication; as Hødnebø points out, Snorri was perhaps influenced by the mythological narrative describing the slaying of the dwarves Fjalarr and Galarr at the hands of Suttungr the giant by similar means (Hødnebø 1992:130). Snorri has forgotten, however, that the tidal range occurring around Kormt was not

9 It is AM 310, 4to that gives this presentation. Sth 18 has a shorter presentation and states only that they were submerged in the place known as Skratteskjera (Sorcerers’ skerries).

40 

 A: Scholarly Background

Fig. 3.1: Seiðr-men on Skratteskjer. Illustration: Halfdan Egedius, 1899. Owner: The National Museum of Art, Architecture and Design © The image has been cropped.

the approximately two metres particular to his native region in Iceland, but a mere 30–40 cm. The narratives of events on Avaldsnes during the reign of Óláfr Tryggvason were subsequently incorporated into the manuscripts of Óláfs saga Tryggvasonar en mesta, which built upon Óláfs saga Tryggvasonar in Heimskringla, while also interpolating material from other sagas.

3.4 Events during the reign of Óláfr inn helgi A particularly consequential event, seen as a pivotal moment in the saga of Óláfr inn helgi, occurred on Avaldsnes during his reign. The narration of how Ásbjǫrn selsbani killed Þórir selr on Avaldsnes, something that led not only to his own death but also to the king’s death at Stiklestad, we know first and foremost from Snorri, but Snorri had both the so-called Oldest Saga and the Legendary Saga upon which to build. In Snorri’s version, Ásbjǫrn Sigurðarson, who was the son of the brother of Þórir hundr and the son of the sister of Erlingr Skjálgsson in Sola, travels south from Hålogaland seeking to buy grain after enduring many hard years of crop failure. Reaching the Karmsund, he moored up for the night by the Avaldsnes estate, where Þórir selr, the king’s steward, was in residence. In the morning he came to the ship, and Ásbjǫrn announced his errand to buy grain and malt to alleviate the crop failure in the north of the country. Þórir selr informed him that he should just turn back towards the north, as the king had decreed a ban against taking grain out of southern Norway. Ásbjǫrn



3 Mundal: Old Norse Written Sources 

Fig. 3.2: Ásbjǫrn hacks off the head of Þórir selr so that it lands on the table before King Óláfr. Illustration: Theodor Kittelsen, 1897.

 41

42 

 A: Scholarly Background

replied that if he would not be permitted to buy grain, he would travel on to stay with his maternal uncle, Erlingr Skjálgsson, in Sola, but he promised to pay another visit to Þórir selr at Avaldsnes on his way back northwards. Ásbjǫrn presented his wish to buy grain to his uncle, and Erlingr devised the solution that he could buy grain from his slaves, who stood outside the law and were not bound by the king’s prohibition against taking out grain, permitting Ásbjǫrn to travel northwards with a fully loaded ship. Returning to Avaldsnes, he fulfilled his promise to Þórir selr, but Þórir confiscated all of the grain from him, along with his beautiful sail and gave him an old one instead. When Ásbjǫrn reached home, he was little satisfied with his journey, and had to tolerate scornful words from his paternal uncle, Þórir hundr. The spring after this event, Ásbjǫrn travelled southwards on a longship with almost 90 men, and he arrived at Avaldsnes the Thursday of the Easter week. He moored up on the outer side of the island, and walked alone to the royal estate. It turned out that King Óláfr inn helgi was at a feast there. Ásbjǫrn entered the vestibule, and overheard Þórir selr telling the king about how he had taken the grain from Ásbjǫrn. As Þórir selr recounted that Ásbjǫrn had cried when he had the sail taken from him, Ásbjǫrn jumped into the room, drew the sword that he had hidden under his cloak, and hacked off Þórir selr’s head. The head fell on the table in front of the king, and the body at his feet. Skjágr Erlingsson, a relative of Ásbjǫrn, was in the king’s company and bade for mercy for Ásbjǫrn, but the king was furious and had Ásbjǫrn taken prisoner, announcing his intention to have him put to death. Skjálgr set off immediately with his company home to Sola in order to fetch help, but before he departed, he assigned the Icelander Þórarinn Nefjólfsson the task of keeping Ásbjǫrn alive until Easter Sunday. Þórarinn managed continually to find reasons to postpone the execution. When Skjálgr came to Sola and announced what had happened, Erlingr Skjálgsson immediately assembled a large army and sailed to Avaldsnes. With their superior strength, they forced the king to agree to a reconciliation with Ásbjǫrn. The agreement was that Ásbjǫrn, who had now received the byname selsbani (‘the killer of [Þórir] selr’), should take over the place of Þórir and become the king’s steward on Avaldsnes, after first going home to Hålogaland to sort out affairs there. His paternal uncle, Þórir hundr, persuaded Ásbjǫrn to remain seated on his farm; by doing so, he broke the agreement with the king. Ásbjǫrn selsbani was killed by one of the king’s men the summer after the event on Avaldsnes. Ásbjǫrn’s mother gave the bloody spear with which Ásbjǫrn was killed to Þórir hundr with the words that she desired that it should stand in the chest of Óláfr digri (Óláfr inn helgi). The event on Avaldsnes also led to renewed enmity between the king and Erlingr Skjálgsson. In Snorri’s account, there is a clear causal connection between the events on Avaldsnes and the king’s death in the battle at Stiklestad. This connection is not found in the versions older than Snorri’s. The narrative about Ásbjǫrn selsbani subsequently found its way into manuscripts of the Great Saga of St Óláfr, which builds upon Snorri’s separate Óláfs saga helga but interpolates material from other sagas.



3 Mundal: Old Norse Written Sources 

 43

Fig. 3.3: King Óláfr Haraldsson must force his way through the men of Erlingr Skjálgsson in order to return from the church where he had heard the Easter Mass to his chamber. Illustration: Erik Werenskiold, Unpublished sketch. Photo: Ellen C. Holte, MCH. Owner: Marit Werenskiold.

3.5 Avaldsnes at the end of the 1200s into the 1300s After the founding of the towns Bergen and Stavanger (Bergen, according to tradition, was founded in 1070 by King Óláfr kyrri; Stavanger likely somewhat later), one should believe that the rural royal estates in Western Norway declined in importance as centres of administration. The literature of the kings’ sagas, which narrate the kings’ journeys about the country, suggests that the kings increasingly took up residence in the towns. In Bergen, King Sverrir had his residence at Sverresborg; the royal installation on Holmen in any case dates back to the early 1200s. The sources speak little of events on Avaldsnes in the period between Óláfr inn helgi and the latter half of the 1200s, apart from the short version of Bǫglunga sǫgur (Bǫglunga sǫgur 1988:43), which mentions a skirmish between the baglers and the birkebeiners on Avaldsnes.10 Despite the scant information from the written sources, along with the gradual displacement of the old royal estates by the towns as political and administration centres, sources from the end of the 1200s and 1300s indicate that some royal estates, including Avaldsnes, retained their influence long after the foundation of the towns in the area.

10 The long version of the saga also mentions the event, but locates it with less precision to the Karmsund.

44 

 A: Scholarly Background

In the final chapter of Hákonar saga Hákonarsonar, the king builds a stone church on Avaldsnes – the fourth largest church in the rural districts of Norway. By the end of the 1200s and in the 1300s, royal charters were being issued principally in the towns, but some charters were issued on Avaldsnes: King Eiríkr Magnússon issued a charter there in 1297 (DI 2:67), and King Hákon V Magnússon issued many there in the early 1300s. He issued two charters there in 1308 (DN 3:71 and DN 2:90),11 one in 1309 (DN 3:81), and one in 1313 (DN 5:105). In 1314, Hafþórr Jónsson, the king’s son-in-law, was one of the granters of a charter on Avaldsnes (DN 4:107). From one charter in 1322 (DN 1:168), it can be surmised that a law-thing was convened on Avaldsnes. Some amendments were also issued on Avaldsnes in this period: one by Hákon V Magússon in 1313 (NGL 3:105–6) and another in 1314 (NGL 3:108–9), and one by King Magnús Eiríksson in 1355 (NGL 3:174). In the Icelandic Flatey-annals, under the year 1343, it is said that Ǫgvaldsnesbuza was wrecked in Grindavík (Storm 1888:402). This ship – or possibly an older ship with the same name – is also recorded many times in toll rolls from the English town King’s Lynn early in the 1300s. A buza is a large ship; in this period, the term designated a large merchant vessel. The ship was named after its place of origin, which signifies that Avaldsnes remained a significant centre for overseas commercial trade. In a letter from 1370,12 King Hákon VI Magnússon accused Hanseatic merchants of having burnt in 1368 his estate on Avaldsnes, other royal estates by the Karmsund, and a house for travelers (RN 7:46). In 1374, King Hákon VI Magnússon issued the last known royal charter to have been written in the Avaldsnes region (DN 15:29). The charter exists only in a late copy, and its location is given as “vdi Karirsund”, but ir in “Karirsund” is very likely an incorrect reading of m, and this is thus reason to consider the possibility that the location in the charter should be Karmsund. What the Old Norse text said that lies behind “vdi Karirsund” is not possible to establish with full certainty, but it was most likely “í Karmsundi”, which could indicate that the charter was issued aboard the king’s ship. It was in fact common practice for the king and his company to reside aboard the ship, possibly using a tent on land, when they were anchored in the harbour. Among other examples, King Magnús Erlingsson is said (Sverris saga, ch. 85) to have sailed northward along the coast of Western Norway, staying two or three nights in each harbour. He stayed in the Karmsund for two nights, and asked for news about King Sverrir from ships that came sailing through the strait from Bergen. On the other hand, it is hardly likely that King Hákon VI Magnússon would have stayed either on the ship or in a tent on land when he was in the vicinity of the royal estate on Avaldsnes. The fact that the charter is not localised to Avaldsnes, as

11 The location for this charter is not indicated, but the editors of RN (3:467) suggest that the charter was written on Avaldsnes, as another royal charter had been issued there two days earlier. 12 The king’s letter is a reply to the coastal towns’ response to the king’s earlier response to a letter of complaint from the coastal towns.



3 Mundal: Old Norse Written Sources 

 45

were the earlier royal charters that were issued by the king during a stay at the royal estate, most likely reflects that the royal estate, six years after the fire in 1368, had not yet been rebuilt. In the king’s response from 1370, the loss suffered by the king on Kormt after the devastation by the Hansa is valued at more than 2000 marks of burnt silver. This sum includes assets additional to the royal estate itself, such as the loss of other royal farms by the Karmsund and a forest, but it is reasonable to assume that the royal estate made up a large part of this appraisal. By comparing the valuation of the king’s loss at the Karmsund with valuations in the same response letter of other acts of destruction perpetrated by the Hansa, we see that the destruction of Marstrand with the castle, cloister, and church as well as a number of islands was appraised at more than 10,000 marks of burnt silver; a single farm by the Karmsund that belonged to one of the king’s men is valued at more than 40 marks of burnt silver, and the destruction of a whole area with many houses in the same place (perhaps a settlement in which farm houses were set close together; in Norwegian: ‘klyngetun’) is valued at more than 600 marks of burnt silver. It is difficult to establish how large a part of the king’s loss of 2000 marks of burnt silver was made up by the royal estate; in any case, the valuations suggest that the royal estate that was burnt in 1368 was an establishment with a considerable number of buildings. It is unknown whether Avaldsnes subsequently resumed its role as a royal estate; perhaps the ravages of the Hansa put an end to its long history as such. At any rate, with the onset of the period of unification there was no longer much need for maintaining royal estates throughout the country.

3.6 History and the creation of traditions When Avaldsnes is presented in Old Norse sources as a central place hosting many events of some consequence for Norwegian history, this is obviously connected with the fact that as a royal estate, the king would occasionally stay there. As the king stood at the centre of political activities in Old Norse society, and would be sought out by friends as well as enemies, it follows that the place of the king’s residence would consistently be the scene for dramatic events. Sites hosting such events lend themselves to the creation of traditions. It has been shown that people (such as kings and saga heroes) as well as places can attract traditions. The two kings responsible for Norway’s conversion to Christianity, Óláfr Tryggvason and Óláfr inn helgi, are without a doubt characters who attract traditional material and become attached to wandering stories and stereotypical motifs. Places can also attract tradition to themselves and Avaldsnes seems to have been such a place. Very likely this is connected with the toponym itself containing the man’s name Ǫgvaldr, the last element of which has the meaning ‘king/

46 

 A: Scholarly Background

ruler’.13 Of the royal estates belonging to Haraldr hárfagri in Western Norway, two are thought to contain a man’s name, the other being Alreksstaðir; this name, Alrekr, likewise has a final element that means ‘king’. It appears that in connection with royal estates that could be associated with a personal name, there, at one point or another, developed a tradition that connected the toponym of the royal estate to a king from long ago. Where such a toponym contained a name element that meant ‘king’, that would be a factor lending additional support for the formation of a tradition. Such traditions are not only connected to royal estates and to names that contain an element meaning ‘king’: in many Old Norse texts, often in fornaldarsǫgur, which narrate events taking place prior to the unification of the Norwegian state, there occur toponyms containing a man’s name – or a name element that has been interpreted as a man’s name. This element often forms the basis for stories that the place was named after a king who lived there long ago. One of the better-known examples of this is Balestrand in Sogn, which is thought to have been named after a King Beli in the fornaldarsaga Friðþjófs saga ins frœkna. Appearing in the same saga is a King Hringr in Ringerike (Old Norse Hringaríki), and in Haralds saga ins hárfagra in Heimskringla there is King Haki, to whom Hakadal owes its name. Alreksstaðir is likewise traditionally associated with a namesake king. The fornaldarsaga Hálfs saga ok Hálfsrekka, which likely appeared in written form in the 1300s, begins with the saga of a King Alrekr who lived in Alreksstaðir. The king had two queens; they did not get along well, so he had to send one of them away. In order to decide, he had the two queens compete to determine who could brew the best beer. One of the queens sought help from Freyja, the other from Óðinn; the latter queen won the competition and was allowed to retain her position as queen. As regards King Haraldr hárfagri’s royal estates, it seems that separate traditions have been conflated. Hálfs saga Hálfsrekka (ch. 2) also mentions King Ǫgvaldr.14 In this version, Ǫgvaldr does not live on Ǫgvaldsnes; rather, he had been killed in a fight against a Viking named Hæklingr and was buried in a mound at the location that would subsequently be named Ǫgvaldsnes after him. Once, when one of the Icelandic settlers lay ready to sail from Avaldsnes for Iceland, he asked when it was that King Ǫgvaldr had been killed; the dead king answered from the mound in the form of a stanza that it happened a long time ago. Regarding this type of tradition connected to Ǫgvaldsnes and Alrekkstaðir, it can seldom be determined with certainty whether it was a single (petty)king who pro-

13 Less certain is meaning of the first element in the name. One possible interpretation of Ag/Ǫg is ‘sword edge’ (Alhaug 2011, see under the lemma Agvald and Åvald; Kruken og Stemshaug 2013, see under the lemma Agvald). The word could also be a derivation of the root in ‘age’ (Old Norse agi), which means ‘respect, fear, admonition’. 14 In the manuscripts of Hálfs saga ok Hálfsrekka there are also found forms of the name that have inserted an n after the g (Ǫgnvaldr). Sophus Bugge comments on this “incorrect” form in his edition of the saga (Bugge (ed.) [1864–73], p. 4, note) and explains it as an analogy with the name Rǫgnvaldr.



3 Mundal: Old Norse Written Sources 

 47

vided the origin of the toponym, or whether it was the toponym that provided the name of the king. Without question, the place name contains a man’s name, but it is difficult to know whether this original man was the Ǫgvaldr referred to in the sources, or someone else. Assuming an historical core that there was once a King Ǫgvaldr on Ǫgvaldsnes, the fact that the particular name happens to contain an element meaning ‘king’ is suspiciously apt. Because the name is so suitable for he who carries it, it would be reasonable to expect that the name is to be understood as a by-name, as many names do indeed have origins in by-names. That the name Ǫgvaldr was rare,15 could also serve as an argument that the name was originally a by-name that had not quite established itself as an ordinary name. More plausible, however, is the evidence suggesting that Ǫgvaldr should be understood as an ordinary man’s name. The two elements composing the name – Ag/Ǫg and valdr – were widely distributed elements in Old Norse male names. The latter element, -valdr, is found in many common male names, like Þorvaldr, Gunnvaldr, Rǫgnvaldr, and Sigvaldr, among others. The former element, Ag/Ǫg, is less widespread, but is found in common names such as Ǫgmundr. The fact that both name elements are common, albeit not in combination with each other, supports the suggestion that Ǫgvaldr was a common Old Norse male name; if the name was rare within the collected Old Norse name material, it may yet have found more frequent use within a limited geographical area.16 It might be supposed that a name containing an element meaning ‘king/ruler’ would originally have been used by the highest social layer in a society, which would also explain why the name suits the bearer, as is the case with Ǫgvaldr on Ǫgvaldsnes. However, there is nothing in the Old Norse sources to indicate that the name in Old Norse times was reserved for members of ruling dynasties. Names with the element valdr had gained wide circulation in the Old Norse name material. This can indicate that the semantic content of ‘king/ruler’ had lost any former exclusivity; nevertheless, the interpretation of valdr, which is also a poetic word for ‘king’ in the language of the skalds, is sufficiently clear that it can be activated whenever desired. Whether Ǫgvaldsnes really was named after a man who was king in the distant past, or after a man who was not a king but was coincidentally called Ǫgvaldr, is impossible to establish. What is certain is that even with an historical core to the traditions regarding King Ǫgvaldr, the details of the tradition would remain unreliable. We have seen that variations in the traditions surrounding the same king are easily formed. This is also the case for Ǫgvaldr –one tradition is seen in the kings’ sagas and another in the for-

15 An overview of the existence of the name in Old Norse sources is found in Lind 1905–15 and 1931. 16 The overview of the existence of the name Ǫgvaldr (with variants) in Lind 1905–15 and 1931 shows that the name in the late Middle Ages has a concentration in Telemark, but the presence of the name in toponyms indicates that it had previously enjoyed a somewhat broader geographical distribution.

48 

 A: Scholarly Background

naldarsaga Hálfs saga ok Hálfsrekka. In the formation of a tradition can be discerned permanent and less permanent elements or motifs. Elements that are permanent, in the sense that they do not change significantly over time and do not split into multiple variants, can easily create the impression of greater historicity or “truth” than those parts of the tradition more prone to changing and splitting into variants. Consistency in a tradition is not necessarily a guarantee of a high level of truth. Permanence in the tradition would be an argument for a certain level of truth only if there is reason to believe that the tradition was formed not long after the events actually happened. If a relatively permanent element was fixed at a later point in time in the formation of that tradition, permanence in itself does little to verify any core of truth contained therein. However, in cases where the tradition or elements of it are seen to change easily, there is even greater reason to be skeptical regarding the level of historical truth it might preserve. Another hallmark of oral tradition is that it often contains stereotypical motifs recognisable in numerous other texts that build upon tradition – so-called wandering motifs. At the oral stage, such motifs could have moved at suitable times from a tradition linked to one hero to a tradition linked to another hero; at the written stage, an author could have gladly adapted such motifs from older written texts or from oral tradition. The author at the written stage, as well as the teller of the tale at the oral stage, could also compose freely around the subject at hand while altering the earlier tradition. At the oral stage, a new variant of the transmitted tale could enter the tradition if the new composition met with its audience’s approval. As concerns the tradition of King Ǫgvaldr and Avaldsnes, some elements of the tradition can be characterised as relatively permanent, while others have clearly been altered. There are elements of wandering motifs, as well as probable examples of new composition added at the written stage. Permanent elements in the tradition hold that there was once a real king who was called Ǫgvaldr, after whom Ǫgvaldsnes/Avaldsnes was named, and that the grave mound of this king is found on Avaldsnes. One reason that these pieces of information seem to have been a stable part of the tradition is easy to see. The oral tradition has here support in the form of the place name and in the form of the visible grave mound, which both virtually function as evidence that the narrative is true. But while the name of the king, Ǫgvaldr, has been preserved in the place name, the name of his opponent varies in the tradition. In the Old Norse kings’ sagas from the end of the 1100s and from the beginning of the 1200s (Oddr Snorrason munkr and Snorri), the king who fells Ǫgvaldr is called Varinn, in the fornaldarsaga he is called Hæklingr, and in Flateyjarbók from the 1370s (1:375), which builds upon the older kings’ sagas, he is called Dixin. The two kings responsible for the Christianisation of Norway were characterised as heroes in the stories told about them both contemporaneously and in later periods. These stories, however, to whatever degree they had been based on historical events, would subsequently attract wandering stories and narratives that perhaps origi-



3 Mundal: Old Norse Written Sources 

 49

nally had been connected to other figures. When the tradition began to form, both the selection of motifs and the shaping of certain motifs, including those rooted in historical events, would often be shaped in the direction of a stereotype. We can be certain that the two Christianising kings were both occasionally in residence at the Avaldsnes royal estate, and that the tales about Eyvindr kelda and Ásbjǫrn connected to the place likely have an historical core, though it may not be possible to find proof. It can likewise be supposed that Óláfr Tryggvason could have investigated the grave mounds at the place, which might have provided favourable conditions for the growth of stories that explain why he did so. It is impossible to determine how comprehensive the historical core of these stories is; what is certain is that a tradition rooted in historical events has been modified and expanded in the process of transmission. The visitation by Óðinn to one of the kings responsible for Christianisation, or to one of their men, is a motif so common in Old Norse texts that it can be considered a wandering motif.17 The time of year and location of the meeting can vary by the telling, but is usually set at one of the Christian festivals.18 The precise nature of the festival seems to be less important. According to Oddr Snorrason munkr, Óðinn turns up on Avaldsnes at Christmastime, while Snorri has this happening closer to Easter, probably because it is a better fit with the king’s travel schedule as he describes it. Another common element in the narrative has Óðinn demonstrating knowledge of the distant past with such a knack for storytelling that he keeps the king (in some versions one of his men) awake through the night in order that he will be sleepy in church the next day. The depiction of Óðinn as he appears to Óláfr Tryggvason likewise follows the trope that the god, though not identified by name, is recognisable to the tale’s audience in the image of a one-eyed old man wearing a low hat. As noted, the story of the burning-in of the seiðr-men and the killing of Eyvindr kelda potentially contains an historical core. At the same time, the representation of the magicians exhibits typical stereotypical traits connected with different genres of literature. As many have argued, the saga written by Oddr about King Óláfr Tryg-

17 Annette Lassen in her doctoral dissertation provides an overview of the representations of Óðinn, including encounters with Óðinn, in various Old Norse genres. Of special interest here is the presentation of Óðinn in kings’ sagas and þættir (Lassen 2011:135–51). Hallvard Lie provides a humorous discussion of the meeting between Óðinn and one of the members of King Óláfr Tryggvason’s court at the farm in Reina in Vika in Þáttr Þorsteins skelks (Lie 1992:211–216). 18 It is not only the meeting with Óðinn, but also meetings with dangerous powers and the unknown in general that are traditionally set at a festival. That the meeting with Óðinn was set during the Easter weekend is probably a later development from the Christian period. That the meeting with dangerous powers was set during the Christmas weekend, can also build upon pre-Christian notions. The Christian Christmas was set for the same time as the pre-Christian celebration of mid-Winter; the notion that meetings with dangerous powers are more likely in the darkest time of the year is hardly specific to the Christian tradition. In the kings’ saga Ágrip, for example, the narrative begins with King Hálfdan svarti losing his Christmas meal to invisible creatures on Christmas Eve (jólaptann). In the same saga, the meeting between King Haraldr hárfagri and the Sami girl Snæfríðr also occurs on Christmas Eve.

50 

 A: Scholarly Background

gvason was intended to portray the king as a saint. Whatever the case may be, the saga makes use of clearly recognisable hagiographical tropes connected to Christian literature; for example, the heathens/magicians are blinded by the sight of the holy church. On the other hand, Snorri’s representation of the magicians employs a trope common in secular Old Norse texts – the magicians have the power to conjure up thick fog or darkness. In the case of Eyvindr kelda and his followers, the spell of darkness that they had conjured up against the king and his followers backfires, and they are themselves overcome by darkness. Neither Oddr’s nor Snorri’s accounts of the killing of Eyvindr kelda and the seiðr-men can be called a typical traditional motif. Oddr’s presentation of the killing as being in accordance with the law can be interpreted as an authenticating detail, while Snorri’s representation should be understood rather as an example of his literary talent and his prioritisation of artistic concerns over fidelity to the sources. Several motifs in the representation of Ásbjǫrn selsbani too have likely had their formulation influenced by stereotypes in the oral tradition. When Ásbjǫrn hacked off the head of Þórir selr, the head fell on the table in front of the king. Such dramatic decapitations carried out as revenge are found in several Old Norse texts.19 The scene in which Ásbjǫrn’s mother goads his paternal uncle to take revenge and provides him with the murder weapon has close parallels in many texts, and could have been influenced by both oral motifs and older parallels in written texts. However, women goading their male relatives to vengeance was not an infrequent practice in Old Norse society, so the incident described here is not necessarily a result of the formation of tradition. Some of the narratives connected to Avaldsnes occur on two temporal planes. For example, in those tales that include the story about King Ǫgvaldr, the main plot takes place during the 900s or 1000s, alongside another narrative set in the distant past, at the time of King Ǫgvaldr himself.20 The story on the more distant plane of time tends to employ traditional tropes similar in degree of stereotyping to those used in stories connected to the Christianising kings. The king or the hero who, from within the mound, recites a strophe or responds when addressed, as King Ǫgvaldr does, is also a motif known from other sources. Needless to say, there is no historical core to be found here, though the motif could reflect actual beliefs held in pre-Christian times.21

19 A close parallel is found in Brennu-Njáls saga, ch. 155; a variant in which the head that has been struck off speaks while in mid-air is found in the same saga, ch. 158, and in Laxdæla saga, ch. 67. 20 The temporal plane of the author can also emerge in the text, as Snorri says about the boat-shaped ship setting: “those that are still standing”, or as Oddr Snorrason munkr says about the skerry where the seiðr-men were killed, that it has been “called Skratteskjer right up until our times”. 21 In the fornaldarsaga Hervarar saga ok Heiðreks konungs, Hervǫr communicates with her deceased father in his mound, Angantýr, by exchanging strophes with him.



3 Mundal: Old Norse Written Sources 

 51

Although the motif of the king who sacrifices to a cow or who has an extraordinary relationship with a cow that accompanies him everywhere and sustains him by her milk is not especially common, it is known from other sources and could serve as indication that Oddr is building upon oral tradition. For example, the motif is found in the fornaldarsaga Ragnars saga lóðbrókar, in which King Eysteinn in Uppsala has a cow, Síbilja, to whom he sacrifices. She leads his host into battle, and the king’s enemies are driven mad with fear simply from the sound of her bellowing. In the end, the cow perishes in battle against the sons of King Ragnarr. While the written fornaldarsaga is more recent than Snorri’s text, the motif of the king and the cow can possibly be an older motif in the oral narrative tradition. King Ǫgvaldr, sustained by the milk of the cow he always has with him, can thus also be associated with one of the primeval giants in Old Norse mythology – Ymir, sustained by the milk of the primeval cow Auðhumla – and the motif thereby plays a role in raising Avaldsnes to mythical stature.

3.7 Conclusion The written sources show that Avaldsnes as late as the 1300s was occasionally a site of residence for the king and his company and was an important place in central Western Norway. On this account, the sources are contemporary and reliable. The literature of the kings’ sagas recounts that Avaldsnes was one of the royal estates of King Haraldr hárfagri, and subsequently the scene of important events in the reigns of Hákon inn góði, Óláfr Tryggvason, and Óláfr inn helgi. The distance in time between the written sources and the events referred to is considerable here; around these stories a tradition has formed, through which the original material has been mediated and modified. Still, there is reason to search for a historical core in the narratives connected to Avaldsnes. These narratives describe battles in which kings or their sons are killed, attempts to attack the king at the royal estate followed by revenge on the attackers, and conflict between the king and one of the country’s most powerful dynasties in which the king is compelled into reconciliation; it is unlikely that narratives involving such dramatic events would have come out of nothing. In cases where the oral tradition is supported by skaldic strophes, the strophes could function as a stabilising factor. In cases where the oral tradition is not supported by skaldic strophes, there is reason to believe that the formation of the tradition could have produced broader modifications. The information connected to the most distant past, the time in which King Ǫgvaldr is said to have lived, is obviously of greatest interest for a book concerning the archaeological excavations at Avaldsnes. The temporal distance between the written sources and the events referred to is so great here that one cannot normally have faith that an historical core is to be found in the tradition. However, the crux of the matter is the question of whether the place name Ǫgvaldsnes has been a stabilising factor

52 

 A: Scholarly Background

in preserving through the centuries a tradition that a King Ǫgvaldr once existed and was buried in a grave mound near the estate which was named after him. No solid conclusions can be drawn in this regard. However, as the archaeological excavations indicate, Avaldsnes was in prehistoric times a seat of some kind of leader, which begs the question as to whether the tradition about King Ǫgvaldr might indeed contain an historical core. Whether or not the name Ǫgvaldsnes preserves information about an actual King Ǫgvaldr, the archaeological finds show that the tradition attached to the name is apt. The name Kǫrmt, which seems to have been elevated into the mythic sphere at a time before the poem Grímnismál came into being,22 could likewise be interpreted as an indication that this place in central Western Norway was known throughout the whole of the Old Norse culture.

22 The dating of eddic poems is often uncertain, and the poems are liable to varying degree to have changed over the course of their transmission through oral tradition until their commitment to writing, which likely first occurred in the 1200s. Most scholars agree that Grímnismál is from the pre-Christian period.

Dagfinn Skre

4 The Avaldsnes Royal Manor Project’s Research Plan and Excavation Objectives This chapter provides an outline of the scholarly problems that the Avaldsnes Royal Manor Project was designed to address, the central theme explored being the political institutions and processes in the first millennium AD. The research plan was developed during the 2007–9 pilot project phase, and was adjusted and supplemented during the 2011–12 excavations and the research and publication phase in the subsequent years. The first of the research plan’s two sections, the results of which are presented in the present volume, deals with Avaldsnes, Kormt, and the Karmsund Strait. The research plan included a series of selected themes, taking as a point of departure the rich and varied research strand on so-called central places. The central-place approach informed the choice of objectives for the 2011–12 excavations at Avaldsnes alongside a corresponding excavation and sampling strategy. Relevant specialists were invited to join the project. The second section of the research plan addresses the first-millennium history of political institutions and processes in the south-western coast of the Scandinavian Peninsula. The first results from this research are presented in this volume’s final chapter, which discusses Avaldsnes in a western Scandinavian context. The preliminary results presented in that chapter will be further developed in the next volume from this project.

Why did kings prefer to reside on the island of Kormt, a modestly fertile and windblown island, rather than in the more fertile and more densely populated regions further inland and along the fjords? This is the central question with which researchers of Avaldsnes have grappled for 450 years, and which the Avaldsnes Royal Manor (ARM) Project takes as its point of departure (Skre, Ch. 2). The short answer – proximity to the naval sailing route – can yet be developed to encompass discussions of most aspects of societal and political development in western Scandinavia through the first millennium AD. Two aspects of this vast field of research are addressed in this project: firstly, to discuss the shifting nature and context of a prominent western Scandinavian aristocratic site; and secondly, to reconsider the history of political institutions and processes in the first-millennium south-western coast of the Scandinavian Peninsula (Fig. 29.3) – both aspects with some potential bearing on the rest of Scandinavia and Germanic areas in general. The present volume will be devoted to the first part of the ARM Project research plan, the second volume to the latter. The chosen approach pursues a prominent strand of research in Scandinavian archaeology; the exploration of political sites, institutions, and processes in the first millennium AD. This strand can be traced back to the earliest writers of history (regarding Norway, see Skre, Ch.  2), but has been particularly vibrant during the past half-century. The application in the 1970s of anthropological research (notably, Service 1971; Sahlins 1972) introduced evolutionary models and social stratification as

54 

 A: Scholarly Background

frameworks for discussions of societal hierarchisation and political institutions and processes. The period’s large-scale surveys of ancient monuments in the three Scandinavian countries have provided an empirical base of unprecedented volume and quality for this research. In the following two decades, research on political institutions and processes was based primarily on cemetery studies and agrarian settlement history, as well as on special types of sites, such as hilltop fortifications, boathouses, rune stones, and courtyard sites (e.  g. Hyenstrand 1974; Magnus et al. 1976; Randsborg 1980; Hyenstrand 1982; Ambrosiani 1985; Myhre 1985; Solberg 1985; Myhre 1987; Ramqvist 1991; Hedeager 1992a). A shift occurred in the early 1990s towards qualitative and away from quantitative approaches and methods, leading to less model-based research. The Danish project ‘Fra stamme til stat’ (‘From tribe to state’, Mortensen and Rasmussen 1988, 1991) was instrumental in that reorientation. By bringing together archaeologists, historians, anthropologists, numismatists, and specialists in Old Norse literature and religion, the project set the path for an increasingly interdisciplinary research practice. Over the following decade a series of productive research themes were introduced or revitalised, most significantly the Iron Age hall (Herschend 1993), warfare (Olausson 1995; Nørgård Jørgensen and Clausen 1997), the history of landed property (Skre 1998; Zachrisson 1998; Iversen 1999), ethnic groups and territoriality (Callmer 1991; Näsman 1999), judicial organisation (Storli 2006; Iversen 2013a, 2015b), and central places (Brink 1996, 1997; Larsson and Hårdh 1998; Näsman 1998; Fabech 1999; Hedeager 2001; Jørgensen 2003; Söderberg 2005; Ljungkvist 2006; Skre 2007b). To varying degree, all these research strands involve the study of social stratification. Of the various strata, attention has been directed predominantly towards the social elite; this book applies the term ‘aristocracy’. The lack of formal nobility in first-millennium Scandinavia gives this term a less precise content than, for example, in high-medieval continental Europe. In the mainly agrarian subsistence economy of first-millennium Scandinavia, food production sufficient to support a group of people who were exempt from production activities is the first basic prerequisite for the existence of an aristocracy; the social group’s control of that production is the second. That position provides the opportunity to develop lifestyles and competences that more or less clearly separate that group from the rest of the population. The aristocratic lifestyle can take forms that are rarely or not at all reflected in the archaeological record – for instance, bodily gestures and oral language – archaeologists need to rely on those that are. Clearly, as indicated by rich depositions and huge mounds with lavish furnishing, these conditions were present in many Scandinavian regions in the Bronze Age, northern Kormt among them. From the early pre-Roman Iron Age, however, very few indications of social stratification are found in western Scandinavia. As at Avaldsnes (Bauer and Østmo, Ch.  8:154), agricultural yield increased during that period, and in the Roman Iron Age, a marked shift occurred. High-status grave furnishings and prominent grave monuments as well as luxury imports and huge buildings are found in most Scandinavian regions with a



4 Skre: Research Plan and Excavation Objectives 

 55

predominantly agrarian economy. Although the expressions of an aristocratic lifestyle found in the archaeological record vary through the first millennium, they are always present to varying degrees. They are probably only to some extent representative of the power and relative size of the aristocratic group. In the present context a variety of potentially aristocratic expressions will be considered (below; Skre, Ch. 27). Although some farms in western Scandinavia appear to have been large-scale in the late and most likely the early first millennium, the access to foodstuffs sufficient to feed the variety of specialists requisite to maintaining an aristocratic lifestyle – artisans, poets, warriors, servants, carpenters, and the like – will have necessitated access to surplus produce from other farms. The history of landed property has been explored intensively in the literature of the last twenty-five years (for references to Norwegian publications 1995–2008, see Skre 2011a:201–2). Although the number and extent of estates in the early first millennium is tentative, their existence is plausible. Thus, in this volume, the term ‘manor’ is applied to a farm that appears to be an estate-holder’s residence. In addition, political leaders may have had access to produce through the veizla institution. Prior to the introduction of royal taxes, fines, and the like in the 10th–12th centuries, the veizla appears to be the only redistributive mechanism for foodstuff other than land ownership (Skre, Ch. 29:798). Political institutions in first-millennium Scandinavia appear to have had both communal and aristocratic aspects; exploring institutions and processes therefore involves studying the communal. For instance, while local thing assemblies in the late Viking Age probably consisted of all land owners in the area, also such that possessed no farm but their own, the literary and judicial evidence clearly indicates that thing assemblies were dominated by the aristocracy. The relation between the communal and the aristocratic will be touched upon in this volume (Iversen, Ch. 26; Skre, Ch. 28) and explored further in the second volume.

4.1 Avaldsnes – a central place? The first part of the research plan, which this first Avaldsnes volume is intended to fulfil, is based primarily on the last 25 years of research into sites that have come to be identified as central places. These decades of exploration have allowed in-depth studies of the site or complex that hosted a variety of essential societal functions, for instance cultic rituals and feasts, thing meetings, and markets, and where aristocratic residences and prominent cemeteries were to be found. The ARM research plan aimed at applying the rich and varied research perspectives within this field, which primarily had been developed on sites in southern Scandinavia, Svealand, Vestfold, and Hedmarken (Fig. 1.2) onto a prominent west-Scandinavian site. A PhD thesis by Arnfrid Opedal (2005) and a Master’s thesis by Håkon Reiersen (2009) have demonstrated the potential of applying a central-place approach onto Avaldsnes and Kormt.

56 

 A: Scholarly Background

One of the reasons for choosing the central-place approach was to enable critical evaluation of the concept of the ‘central place’. The concept has primarily been applied to sites such as Helgö (Arrhenius and O’Meadhra 2011), Old Uppsala (Ljungkvist 2006, 2009), Uppåkra (Andrén 1998), Gudme (Hedeager 2001), and Tissø (Jørgensen 2010). North of Skagerrak in present-day Norway, Åker in Hedmarken (Hernæs 1989; Ingstad 1993; Pilø 1993) and Skiringssal in Vestfold (Skre 2007a, 2008, 2011b) display many of the same characteristics, and are here considered to be of the same type as the former (Fig. 1.2). The archaeological material from these sites consists of remains of numerous houses of which some are apparently aristocratic residences, thick and find-rich deposits often with substantial remains from craft production, and finds in deposits and graves of gold and other exotic materials and types. Together with written evidence and place names, the finds indicate juridical, social, and sacral activities. The search along the west-Scandinavian coast for central places of the type described above has been modestly successful. Surely, some sites there would have had central functions; however, aristocratic sites along the coast of western Scandinavia generally appear to lack several of the archaeological characteristics mentioned above. For example, find-rich deposits are rare or non-existent, and indications of extensive craft production and market sites are not commonly found in or near aristocratic sites. As opposed to southern and eastern Scandinavia, place names here that indicate sacral sites appear to be utterly few; securely identified theophoric names (i.  e. containing names of a god or goddess) may be counted on one hand, while those in the south and east approach 200 (Brink 2007b). However, numerous aristocratic sites from the first millennium AD did exist in western Scandinavia, from Rogaland in the south to Hålogaland in the north. Some are situated on the outer coast, others further inland along the fjords and in the two areas with continuous stretches of arable land, Jæren and Trøndelag (see maps in Figs. 29.5–5). Apparently, the centrality of these sites differed from that of aristocratic sites in southern and eastern Scandinavia. Parallels and differences between central places in southern and eastern Scandinavia and potential central places in western Scandinavia merit further exploration. If there are systematic differences, what caused them? Was centrality in the western regions of an entirely different nature from centrality in the south and east, or did the difference lie in the inclusion in central places of particular functions and exclusion of others? Were the differences in central-place features connected to differences in political institutions and processes? Or perhaps the concept of ‘central place’, or rather the content it has attained in Scandinavian Iron Age research, is in need of refinement (Skre 2010)? A research project centred on a site outside the central-place regions in southern and eastern Scandinavia, but which still appears to have had regional and possibly superregional political significance through much of the first millennium AD, would supply the opportunity to address such questions. Avaldsnes fulfils those criteria.



4 Skre: Research Plan and Excavation Objectives 

 57

Some of these questions will be discussed in Section E that concludes this volume; others will be pursued in the second ARM volume. The path towards understanding the nature and extent of Avaldsnes’ centrality pursued in the current volume is, firstly (Skre, Ch. 27), to identify the extent of aristocratic presence at Avaldsnes based on the research presented in Sections A–D. Secondly (Skre, Ch.  28), guided by the types of central functions identified in south- and east-Scandinavian central places, an attempt will be made to identify central functions that Avaldsnes may have had for people living in the vicinity as well as for those residing in the centre. Thirdly, the conclusions from these two chapters will be set in a west-Scandinavian context of aristocratic sites (Skre, Ch. 29). To provide an empirical basis for these analyses and discussions, the excavations at Avaldsnes as well as the exploration of the island of Kormt and the Karmsund Strait will be analysed to highlight a variety of aspects of centrality and aristocratic presence. These aspects and related research questions are outlined in the following.

4.2 Excavating Avaldsnes From the first-millennium monuments and finds in northern Kormt and along the Karmsund Strait, Avaldsnes appears to have been the most prominent manor in the area, at least in parts if not the entirety of that long period. The main aims of the ARM excavations were thus to identify indications of aristocratic presence and central functions at Avaldsnes, as well as the absence of such. The existing first-millennium archaeological evidence at Avaldsnes consists primarily of grave finds and monuments, some of which clearly indicate aristocratic presence. The vast majority of those that have been dated stem from the first half of the millennium (Østmo and Bauer, Ch. 12). Evidence of Viking Period aristocratic presence is predominantly literary and documentary. In the process of writing the ARM research plan it became clear that the existing evidence of both the early and the later periods was in dire need of reassessment, as well as substantiation and qualification through additional archaeological evidence (Mundal, Ch. 3; Stylegar and Reiersen, Ch. 22; Skre, Ch. 23). Although grave furnishings and monumentality supply significant information on aristocratic presence, graves represent points in time rather than trends that span decades and centuries. To identify more continuous trends and to date possible shifts in aristocratic presence and centrality, the Avaldsnes settlement site or sites from the first millennium would need to be identified and excavated. Furthermore, although it not an explicit component of the original excavation plan, the excavation proved that one additional type of evidence held great potential for exploring long-term trends in the site’s development: the history of agriculture at Avaldsnes. The excavation set out to address the following specific themes:

58 

 A: Scholarly Background

1. The types and numbers of buildings over time. Can periods without buildings be identified? 2. The character of the settlement. Do the functions and features of buildings, areas, deposits, artefacts, or biofacts indicate aristocratic presence or superregional networks? 3. The location of the farmyard. Was it stable, or was it moved at any time? 4. Graves and monumentality. Were there graves and monumental elements in addition to those already known, and can their chronology be outlined in greater detail? 5. Agricultural strategies and output. Can shifts in agricultural strategies be detected? Are there periods of increasingly intensive or extensive production? To produce an excavation strategy aimed at highlighting themes 1–3, the evidence from existing archaeological and geophysical surveys was analysed to identify indications of first-millennium buildings and occupation (Bauer and Østmo, Ch. 5; Stamnes and Bauer, Ch.  16). Indications such as postholes and cooking pits were found in widely dispersed areas, but the evidence was not sufficiently detailed or precisely dated to decide their extent or to reconstruct buildings or other types of constructions. On the basis of our analyses of this material and on a detailed LiDAR scan of the Avaldsnes headland, commissioned by the ARM Project, six areas (Areas 1–6, Fig 5.2) were identified as having potential for containing settlement features and deposits. In the excavation strategy designed to explore themes 1–3, the first step was to conduct initial survey trenching in those six areas followed by the opening up of larger excavation areas where the trenching had revealed settlement features and deposits. The minimising approach to further excavation was aimed at limiting intervention in these archaeological features and deposits to what was absolutely necessary to explore themes 1–3. After excavation, unexcavated features were carefully covered in preservation for future excavation. In addition to buildings in and around the farmyard, Avaldsnes would be expected to have had boathouses along the shore; the discovery, date, size, and construction of these would be relevant for themes 1–2. Several boathouse features had already been identified through visual surveying, and remains of one assumed boathouse were partially excavated in 2001. To identify and explore first-millennium boathouses, a methodology was designed that combined surface surveying, sea-level datings, and limited trenching (Bauer, Ch. 10). To highlight theme 4, limited trenching was conducted in the two monumental grave mounds Flaghaug and Kjellerhaug. The latter had not yet been dated, and the trenching was aimed at dating the mound and any subsequent phases of further build-up. This was also the aim of the trenching in the scant remains of Flaghaug (Østmo and Bauer, Ch. 12). In Areas 2–4 (Figs.  5.1–2), previous survey trenching had identified postholes and other indications of settlement. As the ARM trenching proceeded in 2011, it soon



4 Skre: Research Plan and Excavation Objectives 

 59

became clear that no first-millennium building remains could be securely identified in these areas. However, Area 2 presented excellent opportunities to explore the history of cultivation in the Avaldsnes headland. Because an understanding of this aspect of Avaldsnes’ history would contribute to identifying trends and shifts that might be linked to aristocratic presence and central functions, the excavation and sampling of these remains were included in the excavation plan as a fifth theme. While the project was focused on the first millennium AD, excavations revealed extensive remains from the subsequent millennium. In Area 1, quite unexpectedly, the ruins of a high-medieval masonry building were discovered. Less surprising was the identification in Area 1 of the remains of buildings and garden from the post-medieval rectory, although these were more substantial than anticipated. Following the excavation it was decided that these remains, in particular the masonry building, merited their respective chapters in this publication (Bauer, Chs. 14, 15). This inclusion did not substantially alter the chronological emphasis of the project, other than to extend the survey of the medieval literary and documentary evidence to include the evidence on the 13th–15th centuries (Mundal, Ch. 3). Because Areas 1, 5, and 6 lay in what was known to be the post-medieval rectory farmyard, excavations there were expected to reveal remains from that period. To improve the prospects of identifying such remains while avoiding unnecessary excavation, the ARM Project commissioned the historian Frode Fyllingsnes to survey public archives and produce a detailed overview of buildings and land use in the post-medieval era. His report (Fyllingsnes 2008) aided the identification of several features that occurred during excavation, in particular in Area 1, and during the writing of the history of the post-medieval rectory (Bauer, Ch. 15). The research plan included a strategy for scientific sampling of the site. Sampling methods were chosen that could potentially highlight all the themes 1–5 and contribute to dating features and deposits. All postholes and numerous other features were sampled to collect biological material: burnt animal bone fragments, charred plant macrofossils, and wood charcoal. In addition, bone fragments were collected manually during excavation. To solve particular research questions encountered during the post-excavation phase, the scope of biological analysis was extended to include phytolith analysis (silica microfossils of plants) and stable isotopes (Ballantyne et al., Ch. 19). Samples for radiocarbon datings were collected from all relevant contexts; macrofossils and charcoal were taken from biological samples, while charcoal was also collected manually during excavation. Additionally, extensive sampling from relevant contexts was conducted to map magnetic susceptibility, soil micromorphology, and soil chemistry, the latter samples also from grids. In accordance with research questions and sampling opportunities that occurred during excavation, a pollen profile and a single organic chemistry sample were collected and analysed (Macphail and Linderholm, Ch. 17). Extensive geochemical analyses using portable X-ray fluorescence (pXRF) were conducted on core samples that had been systematically collected prior to survey trenching and exca-

60 

 A: Scholarly Background

vations in all areas. Analyses of samples in Area 6 proved particularly relevant to the second of the themes that the excavation was set to address; these analyses are thus published here (Cannell et al., Ch. 18). Further details of excavation methodology and artefact recovery are described by Bauer and Østmo elsewhere in this volume (Ch. 5).

4.3 Exploring Kormt As already noted, surveys and analyses of prominent grave monuments (Opedal 1998, 2005) and certain central-place elements (Reiersen 2009) in Kormt had already been undertaken. Thus, fulfilling these aspects of the research plan did not demand new surveys, but could be based upon existing publications. However, exploration of central functions and aristocratic presence in the area required several additional studies of Kormt. Place-name studies are of demonstrated relevance to central-place studies (Brink, Ch. 24), as are analyses of ritual depositions (Zahrisson, Ch. 25). Recent studies have suggested that judicial organisation as it is known from the high medieval period has a great time depth. There is a need for such studies to be undertaken for larger regions than Kormt; to that end, Frode Iversen (Ch. 26) has studied the whole of Ryfylke. One category of artefacts from the ARM excavations had the potential for highlighting aspects of centrality and aristocratic presence: pottery. Analyses of this material and a reassessment of finds in Kormt and along the Karmsund Strait were thus included in the research plan (Kristoffersen and Hauken, Ch.  21). The results from these chapters are employed in Chapters 27–28, where the research problems in this part of the research plan are addressed.

4.4 Researching political dominance in southwestern Scandinavia in the first millennium AD The exploration of Avaldsnes and Kormt presented in this volume is meant to serve as a basis for implementing the second part of the ARM Project research plan, which aims at reconsidering the history of political dominance and institutions in first-millennium south-western Scandinavia – potentially with some bearing on the rest of Scandinavia and Germanic areas in general. The bulk of these studies will be presented in the second volume from the ARM Project, where the corresponding section of the research plan will also be presented. The final chapter in the present volume (Skre, Ch. 29) develops some of the themes that will be explored in these subsequent studies. The west-Scandinavian landscape sets rather rigid parameters for premodern communication. A communicative per-

4 Skre: Research Plan and Excavation Objectives 

 61

spective on aristocratic and communal sites highlights some structural features of settlement, economy, and society that deserve more attention in the historiography of kingship in western Scandinavia and beyond.

Section B Excavation Results 2011–12 B: Excavation Results 2011–12 5 Bauer and Østmo: Excavations and Surveys 1985–2012

Egil Lindhart Bauer and Mari Arentz Østmo

5 Excavations and Surveys 1985–2012 Surveys and excavations carried out at Avaldsnes 1985–2012 are described in this chapter, the main focus being on the Avaldsnes Royal Manor Project 2011–12 excavations. In sum, the campaigns conducted surface surveys, metal detecting, soil coring, test trenching, open-area excavation, as well as geophysical surveys and scientific sampling. Following a brief account of the extent and results of the 1985–2006 campaigns, the methodology and extent of the ARM excavations are described. Also addressed are the challenges related to investigating a site with such complex history including continuous activity in central areas as found at Avaldsnes.

This chapter has two main aims. The first is to clarify the state of knowledge prior to the Avaldsnes Royal Manor (ARM) Project excavations by providing an overview of previous surveys and excavations. These are presented below in chronological order, each survey to a certain extent representing different focus and objectives related to Avaldsnes as a historical and archaeological site. The second aim is to describe the methodology of the ARM excavations in terms of artefact recovery from topsoil, the combination of trenching and open-area excavation, the excavation of deposits and features, the digital documentation, and the sampling strategy. The scholarly objectives of the project are also briefly outlined (see Skre, Ch. 4 for details). In 1985 the first modern archaeological excavation was conducted at Avaldsnes. Prior to that year, except for a small excavation of a grave by Jan Petersen (1934), only amateur excavations were undertaken with accidental finds made during cultivation and groundworks. Additionally, visual surveys of monuments were undertaken (Skre, Chs. 2, 23; Stylegar and Reiersen, Ch. 22; Zachrisson, Ch. 25). Geophysical surveys are described and discussed elsewhere in this book (Stamnes and Bauer, Ch. 16) and thus mentioned here only in brief. Figure 5.1 provides an overview of all areas excavated or investigated by test trenching from 1985 to 2012. Detailed accounts on the individual field campaigns may be found in the respective reports (Hemdorff 1985; 1993; Rønne 1999a; Elvestad and Opedal 2001; Sjurseike 2001; Hafsaas 2005; 2006; Bauer and Østmo 2013). All radiocarbon dating results from these campaigns have been recalibrated and are supplied in Appendix II, together with all ARM calibrated radiocarbon dating results (for details regarding calibration and citation in text, see Skre, Ch. 1:7–9).

66 

 B: Excavation Results 2011–12

0

100 m

1985–1986

2005

Medieval harbour deposit

Grave mound

1992–1993

2006

Medieval ship wreck

Current sea level

1999–2000

2011–2012

2001

Road and car park Existing building

Fig. 5.1: Overview of all excavations at Avaldsnes, 1985–2012, including the results of the 1998– 2000 surveys indicating the presence of medieval harbour facilities. The numerous non-intrusive geophysical surveys are treated by Stamnes and Bauer (Ch. 16); a map of these surveys can be found there (Fig. 16.1). Minor test pits from surveys are not included. Illustration: I. T. Bøckman, MCH.

5.1 Surveys and excavations 1985–2006 5.1.1 Excavations 1985–6: Subterranean passageway The subterranean passageway was known from local tradition and was observed in 1923 during restoration work on St Óláfr’s Church, as documented in letters to the National Antiquarian Harry Fett from the parish priest Hove and the architect Moestue, who were in charge of the restoration (Hove 1923; Moestue 1923). In 1982, locals contacted the Archaeological Museum in Stavanger when the subterranean passageway was thought to have been rediscovered during the digging of a ditch for an electrical cable (Utvik 1982). The latter observation led to the museum undertaking an excavation in 1985–6. In Area 1 in the ARM Project’s excavation, the passageway ran about 30 m east to west, before turning north towards St Óláfr’s Church’s western tower (Figs.  5.1–2; Bauer,

 67

5 Bauer and Østmo: Excavations and Surveys 1985–2012 

10 9 8 7

1

6

3

5 2

4

Flaghaug

7

Cemetery

St Óláfr’s Church

Storehouse

8 10

1 5 sha

ug

Kjellerhaug

Kon g

2

6

Ba

Nordvegen

rn

9

Gloppe harbour

Gloppe

4 3

0

100 m

Excavated area

Road and car park

Area

Existing building

Grave mound

Fig. 5.2: The ARM Project excavation areas 2011–12 with names of topographical features and buildings mentioned in the text. Photo: KIB media. Illustration: I. T. Bøckman, MCH.

68 

 B: Excavation Results 2011–12

Figs. 14.9 and 14.11–12). The passageway was cut into weathered bedrock – saprolite – that could easily be removed for the purpose of constructing the passageway. The cut for the passageway was 0.5–0.6 m wide, about 1 m deep, and covered with large stone slabs, occasionally with slabs lining the walls. The 1985–6 investigation focused mainly on tracing the layout and direction of the passageway; it was not fully excavated. The depth and width of the cut were measured where the covering slabs were out of original position or provided small openings into the passageway (Haavaldsen 1987). The feature could not be securely dated, but was assumed to be from the Middle Ages (Hemdorff 1985; 1986). At the western termination of the investigated trench, where the passageway ran northwards, traces of two separate buildings, assumed to be more recent than the passageway, were found (Hernæs 1997:216). The passageway is discussed by Bauer elsewhere in this volume (Ch. 14:304–6).

5.1.2 Surveys 1990–3: Settlement remains In 1990, the Archaeological Museum in Stavanger conducted an extensive phosphate survey resulting in the identification of likely areas of prehistoric or medieval settlement (Forsberg and Haavaldsen 1990). In 1992–3, the museum carried out test trenching to search for settlement traces based on these phosphate indications (Hemdorff 1993). In Area 2, postholes, hearths, and cultural deposits were exposed. Ceramic sherds from the late Roman Iron Age and the Migration Period were recovered. Postholes and hearths were also found in Areas 3 and 4. The latter area contained ard marks, as well. In the area east of Area 1 and southwest of Area 8, Stone Age cultural deposits and flint artefacts were found (Hemdorff 1993:3). Most features located in 1992–3 were only recorded in plan and only a very small selection was cross-sectioned and dated. The main conclusion of the surveys was that all the elevated surfaces around the Kongshaug ridge bore traces of prehistoric settlement, possibly chronologically distributed with the older traces located furthest west and the younger (ranging from Roman Iron Age to the early Middle Ages) lying closer to the church (Hemdorff 1993:3). Results from these surveys are discussed in this volume (Østmo and Bauer, Chs. 6:89 and 7:103–4, 129, 131); some have been reinterpreted in light of the 2011–12 excavation results.

5.1.3 Excavations 1999–2000: Graves The excavations were conducted prior to construction of the Nordvegen History Centre, that is, in the northernmost part of Area 6, as well as east of that area. The main excavation was undertaken in the 20th-century rectory garden, east of Area 6. A restoration of the Kjellerhaug grave mound, including removal of certain recent constructions disturbing the mound, was also carried out (Rønne 1999a; Sjurseike 2001).



5 Bauer and Østmo: Excavations and Surveys 1985–2012 

 69

The excavations in the rectory garden revealed three circular stone packings with charcoal and cremated human and animal bones. Though human bones were not present in all three features, the secure presence of human remains in one and possible human remains in another led to the interpretation of the features as cremation graves. The three graves were radiocarbon dated to the Roman Iron Age and Migration Period (Østmo and Bauer, Ch. 12:245). Underneath the graves was a possible cultivation deposit dated to 1261–1125 BC (Beta-145267) (Bauer and Østmo, Ch.  8:141; Sjurseike 2001:6–7).

5.1.4 Surveys 1998, 2000: Harbour As part of Karmøy Municipality’s Avaldsnes Project, established in 1993, the search for a harbour and trading site from the Iron Age or Middle Ages was initiated. In cooperation with Stavanger Maritime Museum, surveys were undertaken in 1998 and 2000, both on land and underwater. Land surveys included surface surveys, phosphate prospecting, auguring, and excavation of test pits and small trenches close to the sea at Avaldsnes, as well as the neighbouring farms Bø and Utvik (Elvestad and Opedal 2001). Complementing previous surveys carried out by the Maritime Museum in the late 1970s and early 1980s, sub-sea surveys and minor trenching were targeting the inner and outer Gloppe Harbour, around the Gloppe Peninsula as well as the northern part of the Avaldsnes headland (Elvestad 2001:46–59). Finds both on land and under water indicated that a rather busy medieval harbour was located in the Gloppe area. Pottery dating from around AD 1250 to the middle of the 16th century supplied the time span for the activity. A substantial proportion of the pottery was dated to the 14th–15th centuries, suggesting an intense period of activity. Finds on land included building remains, boathouses, roads, and cultural deposits from the Middle Ages and the post-medieval period. Some of these possible boathouses were examined closely in 2012 for evidence regarding their construction and date (Bauer, Ch. 10:183–4). Underwater, foundations for piers, bridges, or sailing blockades were found, as well as cairns of ballast stones dumped from boats and thick waste deposits. Several loading sites were identified. A shipwreck from the High Middle Ages (AD 1224–63, T-14818) was discovered in Indre Gloppehavn – possibly sunk as foundation for a pier (Opedal et al., 2001:110). Most of the artefacts originated in what is now Germany or the Netherlands, testifying to the importance of the Hanseatic trade in western Norway. The finds lend credence to the previous assumption that the Hanseatic trading port called Notow/Nothau was located at Avaldsnes (Elvestad and Opedal 2001:6–7).

70 

 B: Excavation Results 2011–12

5.1.5 Surveys and excavations 2005–6: Graves and settlement remains A second round of survey excavations for the Avaldsnes Project in cooperation with the Museum of Archaeology in Stavanger took place in 2005–6; the main goal being to locate traces of the Viking Age royal manor (Hafsaas 2005; 2006). In 2005, the areas of investigation were Kongshaug and the harbour area east of the settlement plateau, at the Gloppe Peninsula. At Kongshaug, several graves, probably ranging in date from the 1st to the 10th century AD, were exposed (Østmo and Bauer, Ch. 12:243–5). Traces of settlement suggested that several buildings had stood at Kongshaug, probably in the pre-Roman Iron Age or earlier. Cultivation deposits were also exposed and excavated in the trenches dug at Kongshaug (Hafsaas 2005:14). In parallel with the 2005 survey, the Kongshaug ridge was surveyed using metal detectors, resulting in the recovering of multiple modern finds but also a 1.8-gram gold ingot (S12222a) from redeposited cultivation soil (Hafsaas 2005:14–15; Zachrisson, Ch. 25:701). The other part of the 2005 survey, at the Gloppe Peninsula, resulted in the discovery of three possible boathouses, four cairns, of which three were possible grave monuments, as well as other building remains, a road, and two wells or watering holes, probably from the post-medieval period (Hafsaas 2005:20–5; Bauer, Chs. 10 and 15; Østmo and Bauer, Ch. 12). A geophysical survey was carried out in 2006 in and west of the modern farmyard (Areas 1, 5, 6), considered to be the likely location for the Viking Age royal manor (see overview of all geophysical surveys in Stamnes and Bauer, Ch. 16:328–9 and Fig. 16.1). The surveys were followed by trenching providing the possibility to compare the results (Hafsaas 2006). The finds included postholes, wall ditches, cooking pits, cultural deposits, and a large stone packing, demonstrating that the prehistoric settlement remains extended over most of the early 20th-century farmyard. Artefacts and radiocarbon dating results placed the settlement traces in the period from the Roman Iron Age to the early Middle Ages (Østmo and Bauer, Ch. 7). In other trenches, possible remains from the high medieval farm were discovered (Bauer, Ch. 14). In addition to the investigations in the present-day farmyard, a limited excavation of a disturbed secondary inhumation grave within the stone packing at Kongshaug, mentioned above, was carried out (Østmo and Bauer, Ch. 12). The island Fårøy was surveyed with no findings, and test pitting within the possible boathouse remains located in 2005 did not provide definitive results (Hafsaas 2006).



5 Bauer and Østmo: Excavations and Surveys 1985–2012 

 71

5.2 The Avaldsnes Royal Manor Project 2011–12 The ARM excavations constitute the most extensive fieldwork conducted at Avaldsnes (Appendix I: ARM staff). The field work extended over two seasons and encompassed several excavation areas (Areas 1–10; Figs. 5.1–2). The aims and general strategies for the excavation are discussed in detail elsewhere in this volume (Skre, Ch. 4:57–60). In accordance with the excavation plan and the excavation permit from the Directorate for Cultural Heritage, excavated areas were minimised and deposits were left intact when possible. The main excavation areas (Areas 1–6) covered 5,228 m2, while minor trenches in other areas (Areas 7–10) amounted to 40 m2. In the excavation plan, the selection of excavation areas was based on results and interpretations from the 1985–2006 campaigns and on studies of the local topography, combined with knowledge gained from geophysical prospecting. A high-resolution LiDAR (Light Detection And Ranging) scanning (20 first-returns/m2), commissioned by the ARM in 2008 and conducted by Blom Geomatics AS on the Avaldsnes headland and the islands to the east to serve as a pre-excavation search for archaeological features, such as banks and depressions from prehistoric boathouses. The scanning results were of assistance in planning the field work in the harbour area; supplying high-quality documentation of features such as the Flaghaug grave mound remains and providing data for terrain models and illustrations of excavation results included in this volume. The choice of the main areas for excavation was based on the project’s scientific aims, namely the investigation of the settlement’s buildings, possible function-specific areas, changes in the farmyard, and monumentality towards the Karmsund strait in the east (Skre, Ch. 28). Thus, based on the available evidence, the main excavation effort was concentrated within the present-day farmyard (Areas 1, 5, and 6) and the adjacent field (Area 2), together assumed to comprise the main settlement area. During the 2011 season, Areas 1 (central part) and 2 were most intensely investigated. Towards the close of the excavation season, several new survey trenches were opened in Areas 5 and 6, as well as in other parts of Area 1, for the purpose of planning the 2012 season. Specifically, the goal of these surveys was to gain an overview of the potential for prehistoric settlement traces in areas not previously surveyed, as well as to assess the results of earlier surveys. As a result, early in the 2012 season, continuous excavation areas were opened in Areas 5 and 6. Furthermore, trenches were opened in the cemetery (the northernmost part of Area 1), in the former rectory garden (southeastern part of Area 1), in Areas 3 and 4, in the Flaghaug grave mound (Area 7), and in the harbour (Areas 8–10). At the conclusion of the ARM excavations, all exposed and remaining features were covered by fibre cloth before the excavated soil was redeposited in the trenches and the surface cover re-established. The depth of cultural deposits in the different excavation areas varied greatly, as did the stratigraphic complexity. In particular, Area 1, the north-eastern part of Area 5, and the southern part of Area 6 were heavily truncated by recent activities, leaving

72 

 B: Excavation Results 2011–12

the prehistoric remains disturbed and fragmented. As a consequence, a larger proportion of these areas had to be excavated to ascertain a cohesive image of them. There has been substantial activity at Avaldsnes since the late Stone Age. Throughout prehistory and in modern times buildings have been constructed, repaired, torn down, and rebuilt – in the same areas or in new locations. The central settlement plateau – Areas 1–2 and 5–7 – has seen the most intensive activity, resulting in large amounts of features and artefacts from vastly different periods, but consequently also truncation of older features. Truncations were especially visible in Area 1, where buildings from the post-medieval rectory and garden and the post–World War  II construction of a car park led to significant damage to building remains from the 4th–5th, 10th–11th, and 13th–14th centuries. Truncation of prehistoric features has caused mixing of material from different periods, resulting in diverging radiocarbon dating results between contextually related features or even within the same feature. In addition, bioturbation and vegetation movement, such as worm activity, erosion, and growing tree roots, has disturbed many features. What remained in Area 1 were shallow deposits between the bedrock and the makeup of the car park. Certain areas, like those between Areas 1, 2, 5, and 6, were unavailable for excavation due to standing buildings, vegetation, or infrastructure such as roads or cable ditches.

5.2.1 Artefact recovery from the topsoil Prior to the 2011 excavations, organised and supervised by ARM staff, a crew of metal detectorists from Rygene Metal Detector Club surveyed Areas 2 and 5 in a 20-by-20-metre grid. The Kjellerhaug and Flaghaug grave mounds were also surveyed. An equal amount of time was spent on each square in the grid in Areas 2 and 5, thereby ensuring a complete and equally intensive survey of all areas. Only artefacts located in the disturbed topsoil (the top 20 cm) were excavated during the metal detector survey. The survey yielded many artefacts; an assessment of the full assembly identified 29 artefacts as archaeologically significant (Fig.  5.3). The remainder were predominantly modern coins, nails, or other iron objects, quite similar to the 2005 metal detector survey at Kongshaug that resulted in only one prehistoric artefact. Apart from an 11th-century silver coin (Østmo, Ch. 20:518), no precious-metal objects were discovered. The meagre results from the 2011 metal detector survey were probably a reflection of previous illegal metal detecting; such activities are known to have taken place at Avaldsnes, on one occasion leading to a police investigation. Throughout the 2012 excavation, metal detector searches were sporadically carried out by the on-site staff. To retrieve artefacts from cultivated areas, a selection of 2-by-2-metre squares in Areas 2 and 3 were mechanically sieved. A custom-built, machine-driven sieve was utilised (Fig. 5.4). The modern topsoil and underlying cultivation deposits were sieved separately. During the two excavation seasons, 23 such squares were selected for



5 Bauer and Østmo: Excavations and Surveys 1985–2012 

 73

Fig. 5.3: Distribution of all metal-detector finds from 2011 and the location of the squares selected for sieving of plough soil. Illustration: I. T. Bøckman, MCH.

sieving, equalling approximately 3.2 % of the total opened ploughed area. The original goal was for 10–15 %, but in areas with thick cultivation deposits (soil depth varied greatly within each square), the sieving was very time consuming; it soon became apparent that it would be impossible to keep pace with the excavation targets. Following an evaluation of the recovered artefacts, it was decided to reduce the amount of soil sieved. Subsequent squares for sieving were only selected from areas likely to contain archaeological features. The deposits were dry-sieved with a mesh size of either one or two square centimetres, depending on the soil condition. Clayey and often wet soil decreased the sieving’s efficiency. The smaller mesh was impossible to use where the soil was wet. Conversely, the larger mesh allowed a large amount of the soil to pass straight through the sieve without revealing potential artefacts. The artefacts generated from sieving

74 

 B: Excavation Results 2011–12

Fig. 5.4: Machinedriven sieve in use. Photo: Cathrine Glette (upper), MCH (lower).



5 Bauer and Østmo: Excavations and Surveys 1985–2012 

 75

consisted mainly of modern ceramics, glass shards, clay pipe fragments, iron fragments, some slag and sintered clay, anthracite, a few lead fragments, flint, and other stone artefacts. Small amounts of burned bones were also recovered. Most of the material was discarded as it was of recent origin, corresponding with the results from the metal detector survey (Fig. 5.3).

5.2.2 Trenching and open-area excavation The excavation began with trenching or open-area excavation using a mechanised digger. Trenches between two to six metres wide were initially opened, taking into account previous survey results and topography. Most trenches were then expanded, based on the location and presumed continuation of exposed features (see Fig. 5.2 for the extent of excavation areas). In the harbour area, the investigations were sufficiently limited for the trenches to be dug by hand. The character of the overburden varied greatly from area to area. In the car park in Area 1, the overburden consisted of hard-packed gravel over thin cultural deposits and bedrock, while in the prehistoric field in Area 2 there were deep colluvial deposits covering stony subsoil, the upper strata of which had been truncated by modern ploughing. To preserve the integrity of particularly important monuments, some trenches in the interior of the Kjellerhaug grave mound and in the fortification remains were not dug all the way down to the subsoil or bedrock (Østmo and Bauer, Ch. 12; Østmo, Ch. 11). Earlier survey trenching (Hemdorff 1993) in Area 2 suggested that cultural deposits were quite thin. However, due to undulating bedrock, the ARM excavations revealed that between the survey trenches dug two decades earlier, the cultivation deposits were up to 1.3 metres thick, containing numerous archaeological features in different strata. Consequently, the soil in Area 2 had to be removed gradually in artificial horizons. While different stratigraphic levels were distinguishable in the trench profiles after excavation, this stratigraphy could not be discerned during excavation. Instead, once a feature appeared, the current horizon was maintained throughout the trench, thus exposing all features at that level. These artificial horizons simply represent the plough depth of younger cultivation and not the surface from which the archaeological features originally were dug; for this reason, features belonging to a wide time range occur at the same level. A similar approach, but on a smaller scale due to shallower deposits and a smaller excavation area, was applied to the colluvium in Area 6. Following excavation and documentation of the features in one horizon, mechanical soil stripping was resumed until a new horizon appeared. Generally, large constructions were only partially excavated. This applied to an almost 30-metre long stone construction uncovered along the eastern side of Area 6. After exposing its extent, crosscutting sections were established to reveal information about the feature’s con-

76 

 B: Excavation Results 2011–12

struction method and stratigraphy, as well as provide suitable places for sampling deposits in and beneath the construction. Apart from the sections, the remaining parts of the construction were left intact (details in Østmo, Ch. 11:210). The Kjellerhaug grave mound was treated similarly; after uncovering stratigraphic information and exposing sections for sampling, the main part of the mound was left intact (Østmo and Bauer, Ch. 12).

5.2.3 Single context excavation and sectioning The main excavation method employed on deposits and features was single-context excavation, although in most areas it was applied in a simplified version. Single-context excavation – as outlined by Edward C. Harris (1989 [1979]) – entails excavating individual contexts (deposits, cuts, and features) in reverse chronological sequence of their deposition, truncation, or construction. Deposition and construction may vary greatly in manner and duration, and can constitute anything from the momentary deposition of a part of a midden or the construction or repairing of a wall to the gradual accumulation over years and decades of a floor deposit inside a building or in a cultivated field. Separate events related to the same feature are treated as separate contexts. A posthole, for example, consists of multiple events and contexts, including but not limited to the hole dug for the post, remains from the post itself, stones and soil supporting the post, and deposits filling the posthole after the removal or decomposition of the post. Throughout most of the investigated areas at Avaldsnes, excavation consisted of topsoil stripping and exposure of features cut into cultural deposits, colluvia, or subsoil. In areas where individual features lay clearly delimited, the single-context method was simplified. Consequently, all events related to such individual features were recorded as part of the same context, called an archaeological object. Features containing separate contexts of importance for interpretation or sampling, for instance Kjellerhaug grave mound, were excavated and documented according to standard single-context methods. In areas with complicated stratigraphy, with multiple deposits and intersecting features, all contexts and their stratigraphic relationships were distinguished, excavated, and recorded. Delimited features such as postholes, cooking pits, hearths, ovens, and wall ditches were sectioned, allowing profile documentation. The fill from the excavated half was removed context by context whenever such sub-division of stratigraphic sequences within the feature had the potential to provide further information. The fill from selected features and deposits was wet-sieved with a mesh size of either 2 or 4 mm, depending on the feature and soil type. The fill of many features was sampled in its entirety for macrofossil recovery, thus eliminating the need for sieving. In addition to sectioning single features, several long profiles were left standing in areas with thick cultural deposits or complex stratigraphy. These profiles allowed rechecking of



5 Bauer and Østmo: Excavations and Surveys 1985–2012 

 77

the interpretations made during the course of the single-context excavation. Furthermore, the profiles were used for comprehensive sampling (see below).

5.2.4 Digital documentation Features, finds, samples, profiles, excavation areas, and topography were recorded using a Trimble TSC3 total station with millimetre accuracy. The data was imported into Intrasis (Intra-site Information System) version 2.2 b103. Intrasis is “a GIS designed for a combination of complex information data and geographical data” (http://www.intrasis.com/intrasis3_system.htm). Intrasis Analysis version 1.2.12 and ArcMap 10 were used for data analyses and map production, respectively. Field documentation employed contexts sheets on which features were drawn and described. Certain compulsory fields in the context sheets, such as colour, texture, and construction elements, provided consistency in the documentation. This allowed features to be sorted and organised, for example based on size, depth, or presence of certain material such as charcoal or fire-cracked stones. After recording in the field, the context sheets were transcribed to Intrasis, thus compiling a comprehensive GIS (geographic information system) of the excavation data. All features recorded in the field were automatically given the next available number in a consecutive sequence, serving as that feature’s unique identity. The feature denomination consists of an ‘A’ (archaeological object) followed by 4–5 digits, depending on how far into the sequence the features were recorded. After importing to Intrasis, errors in the recorded data were corrected, and features’ stratigraphic and contextual relationships to one another were entered. Meta-features, also called superstructures, such as buildings, which included several excavated features, were manually created in Intrasis (Fig. 5.5). The excavated features interpreted as components of the meta-feature were then related to it. The meta-features were assigned low numbers, for example building A10. Finds were coded similarly as features, with an ‘F’ (finds) followed by the next available number and a coded relation to the archaeological object from which they originated. Six-digit numbers were assigned to finds, samples, and features created post-excavation, in order to assign a unique identity to context-less artefacts or to artefacts found during sieving and therefore not recorded in situ. Six-digit numbers were also assigned to provide unique identities in cases where finds were divided, for example from burnt clay into sintered clay and burnt clay.

78 

 B: Excavation Results 2011–12

Fig. 5.5: Intrasis screenshot showing entering of data in Intrasis Explorer (upper) and map display and visual analysis in Intrasis Analysis (lower).



5 Bauer and Østmo: Excavations and Surveys 1985–2012 

 79

5.2.5 Soil sampling A comprehensive soil sampling strategy was carried out. Sampling included macrofossils, charcoal, micromorphology, soil chemistry, pollen, phytoliths, and soil cores (Table 5.1). In addition, environmental monitoring was carried out by the Norwegian Institute for Cultural Heritage Research (NIKU). For technical information about the samples and their analyses, see Ballantyne et al. (Ch. 19), Macphail and Linderholm (Ch. 17), Cannell et al. (Ch. 18), and Martens (et al. 2012, 2013, 2014, 2015). Tab. 5.1: Samples collected during the 2011–12 excavations. Sample type

Number

Charcoal Hummus Soil sample Macrofossil Magnetic susceptibility Micromorphology Environmental monitoring Pollen Soil chemistry Wood

40 4 8 519 33 63 24 9 1,044 5

Macrofossil samples from features and deposits were processed in a flotation tank, dried, and sorted on-site. A total volume of 2,800 litres of soil was processed. Due to observations in post-excavation analysis of macrofossils, a selection of unprocessed soil samples were selected for phytolith analysis and 100 cereals were selected for isotope analysis (Ballantyne et al., Ch. 19:460–1). Micromorphology and soil chemistry samples were collected as complementary sample packages from sections through individual features of specific interest and in sections through larger deposits, such as thick cultivated colluvia with intercutting features (Fig. 5.6; Macphail and Linderholm, Ch. 17:380). In addition, soil chemical samples were taken from horizontal grids covering activity areas of special interest, such as the dwelling areas and the production area in Area 6 (Macphail and Linderholm, Ch. 17:381–3; Cannell et al., Ch. 18:427–8). Extensive soil coring, using a 5 cm wide auger, was carried out prior to opening each excavation area (Fig. 5.7; Cannell et al., Ch. 18:427–8). In addition to providing valuable information about stratigraphy, the coring produced data about the deposits’ depths in the various excavation areas. Furthermore, the cores could be used for post-excavation subsampling of undisturbed matrices from deposits across the site. These samples were analysed using a portable x-ray fluorescence spectroscopy

80 

 B: Excavation Results 2011–12

Fig. 5.6: Profile 15653 after sampling for soil chemical analysis, micromorphology analysis, and pollen sequence. Photo: MCH.

(pXRF; Cannel et al., Ch. 18) to provide input on activities such as waste disposal and metalwork. Information of the depth of cores and their distribution across the site also offers insight into the human impact on the topography at Avaldsnes and the manner in which it has been transformed (Fig. 5.7). The combination of the lower-lying terrain and thick deposits in Area 2 shows both the filling in of undulating terrain and the transfer of soil from higher- to lower-lying plains, which is not to say that the most intense cultivation was located where the deposits were thickest, but rather that the topography has affected what areas have been subject to colluvial build-up. In the car park in Area 1, the shallow deposits represent the impact of modern activity in truncating prehistoric deposits. These quite different situations may have influenced the clustering of relatively well-preserved cooking pits and postholes in Area 2 and, conversely, the paucity of remains in Area 1. This is also a reflection of the fact that the terrain and landscape are not constants; rather, the more uneven early Bronze



5 Bauer and Østmo: Excavations and Surveys 1985–2012 

 81

Depth in cm 0–40 40,1–80 80,1–120 120,1–155

0

100 m

Fig. 5.7: Depth of deposits directly or indirectly affected by human activity recorded by coring. The depths represent the distance from the present-day surface to the bottom of deposits. Depths are affected by erosion and human activity. The deposits include, for instance, midden deposits, trampled or cultivated colluvium transformed into b-horizons, and burned surfaces representing land clearance. The depths do not correlate to the intensity of the use of different parts of the landscape but rather reflect how various activities have affected build-up and erosion of deposits in different landscape zones. Illustration: R. Cannell, I. T. Bøckman, MCH.

Age landscape of Avaldsnes became more regular following three millennia of agricultural and constructional activity.

5.3 Summary The 2011–12 excavations at Avaldsnes were planned and carried out to provide information about the Avaldsnes Royal Manor Project’s research problems (Skre, Ch. 1). Results from earlier surveys supplied background information for determining the excavation areas, but due to the limited scope of the previous surveys, several of the former interpretations have proven erroneous or could not be confirmed by the ARM

82 

 B: Excavation Results 2011–12

investigations. Thus, strategies and methods were revised during the course of the excavation period, in particular from 2011 to 2012. The excavations were further complicated by the massive amount of truncation and other disturbances of archaeological features, coupled with unavailability of certain areas for digging due to standing buildings, roads, the cemetery, and other infrastructure. The continuous activities across the settlement – especially at the main settlement plateau (Areas 1–2 and 5–7) – allow numerous conclusions about the site and its development over time. The development of agriculture, settlement, various activities, and the spatial organisation of the site are presented as seven chronological site periods (Østmo and Bauer, Ch. 6). The extensive sampling strategy of combining multiple sampling methods and analyses helped in filling out the picture of the changing character of the Avaldsnes site through these site periods.

Mari Arentz Østmo and Egil Lindhart Bauer

6 Site Periods and Key Contexts The traces of human presence at Avaldsnes date back to the Mesolithic; the character and extent of human impact on the site have varied through the many intervening millennia to present. The few scattered Stone Age artefacts that have been retrieved in Iron Age contexts supply scant information on the nature of the earliest human presence. Gradually, however, towards the Bronze Age, and increasingly through the Iron, Viking, and Middle Ages, the preserved remains consist of artefacts and as well as remains of buildings, graves, and deposits, revealing far more detailed information regarding the character of activities taking place at the Avaldsnes headland. To describe an overall phasing of the site, we have identified significant changes in these activities that define the transition from one phase to another. The result of the phasing is the identification of seven site periods (SP I–VII) that span the late Neolithic to c. AD 1900. This chapter serves as an introduction to the 2011–12 Avaldsnes Royal Manor Project excavations, presenting an overview of the excavation results and underlining the main contexts and findings that define the seven site periods and the transitions between them.

Significant changes in human activities at Avaldsnes allow the identification of seven chronological phases, which we call site periods (SP). The types of changes that define the transition between site periods are manifested within one or several of the categories that the Avaldsnes Royal Manor (ARM) Project excavations were intended to investigate (Skre, Ch. 4): the types and numbers of buildings, the character of the settlement, the layout and location of the farmyard, the functions and activities in buildings and farmyard, graves and monumentality, and agricultural strategies. The site periods are of varying duration, and though they are quite clearly defined, they are not absolute in the sense that changes occur simultaneously within all the categories. Some activities, individual features, or developments transcend site period limits, but based on the categories that the excavation aimed at identifying, we do not consider such overlap significant enough to merit division into additional site periods. Redeposited lithic artefacts were found in agricultural and cultural deposits, cooking pits, and various other types of Iron Age features scattered across most of the excavated areas. The artefacts include a microlith, flakes, microblades, and various types of cores, indicating that the artefact assemblage likely represents several separate events spanning most of the Mesolithic to the late Neolithic (Østmo, Ch. 20:526). A cultural deposit believed to be related to a Mesolithic or Neolithic settlement was recorded but not excavated in 1993. This is the only identified Stone Age cultural deposit, and since the remaining lithic material lacks contemporary contexts and offers inconclusive and sporadic evidence of early human presence at Avaldsnes, the lithic material alone cannot serve to define a separate site period. The lack of Stone Age contexts is affected by the ARM excavation’s focus on areas with most potential for settlement remains from the Iron Age in general and the Viking Age in particular. Pre-

84 

 B: Excavation Results 2011–12

viously recorded Mesolithic or early Neolithic settlement sites in the vicinity provide a more nuanced understanding of the use of the landscape in various parts of the Stone Age. Relevant sites include the pioneer settlement from the early Mesolithic at Uvik, 0.5 kilometres from Avaldsnes (Bang-Andersen 1988:41; Gjerland 1989; Hernæs 1997:3–4, 26; Fuglestvedt 2012:17), various Mesolithic and Neolithic settlement sites at Husøy, 1.5 kilometres from Avaldsnes (Hatleskog 1992), and two wetland offerings, one of them containing two flint daggers and more than 300 flint artefacts, at Uvik, 0.85 kilometres from Avaldsnes (Gjerland 1989:43; Hernæs 1997:39–42). The activities at the Avaldsnes headland become more tangible as the landscape was cleared around 2000 BC, permitting the definition of the site periods which are presented in the following. This chapter is based on more detailed discussions with relevant references elsewhere in this volume; see internal references for details.

6.1 Site Period I: c. 2000–c. 350 BC SP I spans the longest duration of the seven and is characterised by the introduction of agriculture followed by agricultural practices comprising husbandry and relocating fields. There are no excavated dwelling remains from SP I, but the remains of what is believed to be an economic building and the first phases of the Kjellerhaug grave mound, probably also the Flaghaug mound, can be dated to this period (Fig. 6.1). Around 2000 BC (Beta-333051), the earliest identified indication of land-clearing is represented by patches of a deposit containing charcoal likely originating from the burning of trees and brushwood; the dated sample was taken from an assumed burnt surface back-filling the void from a removed stone in Area 6, beneath the Kjellerhaug mound. A similar deposit containing charcoal believed to represent the same burnt surface formed the bottom of an animal-induced colluvial build-up captured in a section three metres further north-east, thus associating the earliest land-clearing to husbandry. Similar deposits with charcoal inclusions indicating land-clearing were found in other parts of the site, some providing slightly later dates, for instance in the central part of Area 2 and on the site of the Nordvegen History Centre. Tilled soils, combined with a dated deposit in the pollen sequence from Profile 13016 (Fig.  6.1; Macphail and Linderholm, Ch.  17), indicate well-established cereal cultivation of both wheat and barley and fertilisation in the 12th century BC at the latest (Beta347958). Stratigraphically older deposits in the same pollen sequence show that the cultivation of barley was established even earlier. Based on stratigraphic relations, the pollen sequence seems to reflect agricultural practices from some time before the 12th century BC until around the 2nd century BC, providing indications of a system by which fields and pastures were periodically relocated. Compared to SP II, fields appear to have been fairly small.



6 Østmo and Bauer: Site Periods and Key Contexts 

 85

The practice of relocating fields seems to have included periods of intensified cultivation, for example the fertilised deposit documented in layer 3 in Profile 13016. The sequence of deposits in Profile 13016 is similar to the sequence documented in Profile 15653 (Fig. 6.1; Bauer and Østmo, Fig. 8.2), which comprised deposits containing remains from burning, probably from land-clearing, pasture and possible cultivation around 1600 BC (Beta-304878). Still, the development in the areas around the two profiles differs with regard to an early deposit indicating intensified and fertilised cultivation as seen in layer 3 of Profile 13016. Such a deposit was not visible in Profile 15653. Such early traits were possibly not detected, truncated by later activity, or determined to be a later development in this part of the field. No SP I dwellings were identified, though they are assumed to have been present in close vicinity of the fields. Several postholes in Area 5 have radiocarbon dates from the late Neolithic, but these dating results are believed to stem from redeposited charcoal from older features. Only a single posthole is believed to have a Neolithic radiocarbon date not influenced by pollution from redeposited charcoal; this posthole is visible in the bottom of the stratigraphic sequence in Profile 15653 (A10500, TRa4226). A simple economic building or other type of construction (A11) from the middle part of SP I was excavated in Area 2, 3.5 metres east of posthole A10500, supporting the assumption of the presence of settlement at the Avaldsnes headland in this period. A11 was sealed between two dated cultivation deposits, which provide the rough dating of the construction to a 1000-year span (Beta-304878, TRa-4231). Other building remains from SP I may also be concealed outside of excavated and surveyed areas. Two grave monuments can be dated to SP I: the earliest phases in the Flaghaug and Kjellerhaug grave mounds (Østmo and Bauer, Ch. 12:231–42; regarding Flaghaug: Stylegar and Reisersen, Ch. 22:552–66). Only small sections of the Kjellerhaug mound were excavated; no burials were identified, nor were any artefacts or organic construction elements suitable for radiocarbon dating found. Nevertheless, the stratigraphic relations to surrounding dated features provide some information as to the possible date range for the mound’s construction. A charcoal-mixed deposit beneath the central part of the mound dated to the 18th–17th centuries BC (Beta-333050) provides a terminus post quem, while barley grain in a charcoal concentration within the mound make-up gives a terminus ante quem to the 8th–6th centuries BC for the two eldest construction layers with respective stone rows, though it remains unclear whether these represent contemporary construction elements or chronological sequences within this time frame. The mound was subject to subsequent use, probably in SP II and III (below).

86 

 B: Excavation Results 2011–12

BUILDINGS AND SETTLEMENT No dwellings located Single posthole TRa-4226 A10500

BC 1879–1767

GRAVES FLAGHAUG

ECONOMY BUILDING A11

Assumed first phase of construction

Relative dating based on the building’s stratigraphic location between two layers: Beta-304878 A3889

TRa-4231 A5882

Flag Flaghaug 3

BC 1608–1501 BC 795–595

AGRICULTURE

KJELLERHAUG First phase of construction. Stratigraphic dating: Terminus post quem Deposit beneath mound Beta-333050 A42891

BC 1733–1632

Deposits with indication of land clearing, husbandry, and cultivation

6

Terminus ante quem Charcoal concentration in mound Beta-333052 A39717

BC 794–568

01

1 15653

13

Initial land clearing followed by husbandry and periodically relocated fields Deposit indicating wheat and barley cultivation, fertilisation Beta-347958 A9601

5

Kjeller Kjellerhaug 1

BC 1257–1129

2

Earliest indication of land clearing Beta-333051 A44121

6 Trench 3, 2000 excavation

BC 2118–1961

4

1 0

50 m

3

Area number

15653 Profile number 10 Grave number

Estimated minimum extent of settlement Building Estimated minimum extent of cultivation Grave

SEA LEVELS Current BC 500 (+3,5 m) BC 2300 (+6 m)

Fig. 6.1: Main features dated to Site Period I. Key contexts and radiocarbon dates mentioned in the text are labelled in the upper map; an overall interpretation of the land use and spatial organisation within SP I is presented in the lower. Numbering of graves refer to map key in Østmo and Bauer, Fig. 12.1 and Tab. 12.1. Illustration: I. T. Bøckman, MCH.



6 Østmo and Bauer: Site Periods and Key Contexts 

 87

6.2 Site Period II: c. 350 BC–AD 200 In SP II, cultivated fields were expanded, buildings were possibly raised in Area 3, and the practice of preparing food in cooking pits was established. This period is also distinguished by a stone-packed cremation pit grave east of the Kjellerhaug grave mound, as well as possible secondary use of this mound (Fig. 6.2). Previously small, temporary cultivated patches were consolidated into a large, continuous field covering Area 2 and extending into Area 5, possibly also into Areas 3 and 4. This change can be seen from the increased colluvial build-up in Area 2, visible in Profile 15653 (Bauer and Østmo, Fig. 8.3). The colluvial deposits immediately overlie two SP I deposits that represent initial land-clearing, pasturage, and possibly cultivation in the early Bronze Age (Beta-304878) and late Bronze Age (TRa4231), indicating subsequent intensified and fertilised cultivation beginning at some point in the last two centuries BC (TRa-4230). Also, near the cluster of cooking pits immediately east of the Kongshaug ridge, a previous clearing cairn that likely marked the former southern border of a small cultivated patch was removed, leaving behind depressions in the ground that were back-filled with cultivation deposits from subsequent cultivation. The back-fill in one of these depressions was radiocarbon dated to the beginning of SP II (TRa-4213), providing a terminus post quem date for the consolidation of periodically relocated cultivated patches into a larger, permanently cultivated field. This does not provide a very precise dating of the cultivation and enlargement of the field, as it is unknown how long the charcoal had been circulating in the cultivation deposit before its secondary deposition. However, the stratigraphic relationship between removed cairns and the dated cooking pits in the area confirms the approximate dating to early SP II. Both the permanence and the enlargement must have resulted in a larger output from cultivation. This development forms a characteristic trait that contributes to defining the transition from SP I to SP II. The cultivated area was extended into the westernmost part of Area 5. Husbandry continued to be an important part of the farm’s subsistence; in early SP III turf used as building material in the boathouse A40 indicates grazing in the area prior to its construction, that is, in SP II. The introduction of cooking pits dug into cultivation deposits is another trait characteristic of SP II, and the practice continues through SP III. Seven cooking pits have been dated to SP II, the majority to the period’s first part, such as a cooking pit in Area 5 dated to the beginning of SP II, relatively contemporaneous to the removal of the above-mentioned clearing cairn (Beta-319015). Twelve cooking pits, four of which have been dated to SP II, lay in the western part of Area 2 (Bauer, Ch. 13), near the Kongshaug ridge, where settlement remains also appear to be concentrated. Three of the SP II cooking pits have radiocarbon dates stretching into SP III, during which period most of the cooking pits were dug. In addition to the cluster of cooking pits in the western part of Area 2, three single SP II cooking pits lay in different parts of the site: in Areas 2, 5, and 6.

88 

 B: Excavation Results 2011–12

BUILDINGS AND SETTLEMENT

COOKING PITS

Posthole, ditch, hearth, etc Trench 1992/D: Hearth TRa-10699 Structure 35

BC 184–AD 20

Cooking pit likely dating to SP II

No dwellings or buildings located Kongshaug: Settlement traces likely dating to SP II or III

Trench 1992/C: Hearth TRa-10700 Structure 31

GRAVES

Cooking pit radiocarbon dated to SP II

Possible burial indicated by the find of copper alloy masks

Beta-319015 A18656

Kongshaug stone packing Likely dating to 1st–5th c.

BC 356–201

Small stone packing Beta-145268 Grave 2

Area 3: Settlement traces likely dating to SP II–IV

BC 342–AD 2

KJELLERHAUG

AD 136–236

AGRICULTURE Cultivation deposit

Kong sh

aug

Joining of small, periodically cultivated patches to larger, permanently cultivated field

Trench 4, 2005 excavation

BC 361–208

Cultivation deposit indicating intensified and fertilised cultivation TRa-4230 A103

1 15653

Back-fill in removed clearing cairn, indicating expansion of field TRa-4213 A5300/A4079

2

16

Kjeller Kjellerhaug 1

6 10 Trench 3, 2000 excavation

BC 156–45

4

5

D 1992tion ava C exc

1 0

50 m

3

Area number

15653 Profile number 10 Grave number

Estimated minimum extent of settlement Estimated minimum extent of cultivation Cooking pit concentration Grave

SEA LEVELS Current AD 200 (+2,4 m) BC 150 (+3 m)

Fig. 6.2: Main features of Site Period II. Key contexts and radiocarbon dates mentioned in the text are labelled above; an overall interpretation of the land use and spatial organisation within SP II is presented in the lower. Due to intercutting/truncation in Area 5, some cooking pits may contain charcoal from neighbouring pits; while this may affect the dating of the individual pits, it still reflects the actual dating of this use of the area in SP II. Numbering of graves refer to map key in Østmo and Bauer, Fig. 12.1 and Tab. 12.1. Illustration: I. T. Bøckman, MCH.



6 Østmo and Bauer: Site Periods and Key Contexts 

 89

No remains of dwellings or other types of buildings from SP II were identified during the ARM surveys or excavations, though possible building remains were found during the 2005 survey at Kongshaug and in the 1992 survey around Area 3. Based on their stratigraphic relationship to a Roman Iron Age grave, these remains at ­Kongshaug likely date to SP II, although they could date as far back as to SP I. The postholes and dated hearths in Area 3 make this also a possible location of a settlement in SP II. Because the ARM research focus is on the first millennium AD, excavation of these remains was not prioritised. Although the identification as building remains is tentative, settlement on the headland is indirectly indicated by remains of agricultural practices as well as observations of latrine dumping in deposits predating the SP III settlement and graves in the area (Østmo and Bauer, Ch. 7; Macphail and Linderholm, Ch. 17). As for SP I, it may be assumed that the settlement in SP II was located outside of the areas excavated by ARM, and that the settlement was not organised in a well-defined farmyard similar to the one established in the following SP III. Two bronze masks, meant to decorate a cauldron or other object tentatively dating to the pre-Roman Iron Age or the late Iron Age, were found at Avaldsnes in the early 19th century. The artefacts’ relation to the Kjellerhaug grave mound is suggested by Stylegar (et al. 2011). Although no burial dating to that period was identified in the excavated parts of the mound, they may have existed elsewhere in the mound. The mound’s construction is complex with several phases, as is the Flaghaug mound roughly 100 metres further north (Stylegar and Reiersen, Ch. 22; Østmo and Bauer, Ch. 12). Three cremation burials, one of which was dated to the transition between SP II and III (Beta-145268), were found east of the Kjellerhaug grave mound (10, Fig. 6.2; Østmo and Bauer, Ch. 12:245). These were simple, shallow, circular stone-packed pits containing charcoal and burnt bones; all contained animal bones, and two also contained human or likely human bones (Sjurseike 2001:6, 8). The two remaining cremation burials were dated to SP III (Beta-145269, Beta-133887), indicating that the use of the area for burials was likely established in late SP II but maintained for centuries.

6.3 Site Period III: c. AD 200–600 In SP III, the site had a clear and structured spatial organisation and a monumental manifestation of power, represented by secondary phases and possible enhancements of the existing Kjellerhaug and Flaghaug grave mounds, new grave monuments such as the raised stones by Flaghaug, possibly also at Kongshaug, and hall building A10. All of these features were clearly visible from the strait to the east. Distinct zones may be identified, respectively, for cultivation, monumental graves, cooking pits, production and processing, dwelling and hall, and boathouses (Fig. 6.3). The activities in the three latter zones have not been securely identified before SP III, while those of the three former were initiated in SP I–II and culminate in SP III. The farmyard

90 

 B: Excavation Results 2011–12

COOKING PITS

BUILDINGS AND SETTLEMENT

Cooking pit radiocarbon dated to SP III

PRODUCTION

Posthole, ditch, hearth, etc

Cooking pit likely dating to SP III

Kongshaug: Settlement traces likely dating to SP II or III

Concentration A Youngest cooking pit: Beta-304877 A5031

Trench 2001 survey

AD 437–567

Establishment of farmyard in Areas 5 and 1.

Area 3: Settlement traces likely dating to SP II–IV

Flag Flaghaug 4-7

HALL A10 LONGHOUSE A13

21

AGRICULTURE

14

BOATHOUSE

Cultivation deposit

Wall bank

BOATHOUSE A40

1

15653

15

Kong sh

aug

Boundary between farmyard and field established in Area 5. Areas 2, 3, 4: permanent field in continuous use in SP III–VII. Individual layers have not been dated.

5

2 1

Area number

Posthole Beta-324651 A31003

AD 436–552

Kjeller Kjellerhaug 2

6

17 Trench 4, 2005 excavation

15653 Profile number 10 Grave number

Area 6: Features indicating activities like iron smithing, meat processing, etc.

Posthole Beta-324648 A31295 Nordvegen 11

AD 258–381

Layer Beta-324650 A30325

AD 426–533

9 Trench 3, 2000 excavation 34 35

4

GRAVES

Cairn Likely dating to SP I–V

FLAGHAUG

Mound Likely dating to SP II–V

At least four burials

KJELLERHAUG

3

23 0

50 m

Possible burial indicated by the addition of earth and stones

Mound, unknown size/location Likely dating to SP II–V Raised stones monument Likely dating to 0–500 AD

Small stone packing Beta-145269 Grave 4

AD 415–540

Small stone packing Beta-133887 Grave 1

AD 244–385

Cremation grave (secondary) Likely dating to SP II–III

Estimated minimum extent of settlement Building Estimated minimum extent of cultivation Cooking pit concentration Estimated extent of production area Boathouse Grave

SEA LEVELS Current AD 550 (+1,95 m)

Fig. 6.3: Main features of Site Period III. Key contexts and radiocarbon dates mentioned in the text are labelled above; an overall interpretation of the land use and spatial organisation within SP III is presented in the lower. Numbering of graves refer to map key in Østmo and Bauer, Fig. 12.1 and Tab. 12.1. Illustration: I. T. Bøckman, MCH.



6 Østmo and Bauer: Site Periods and Key Contexts 

 91

(longhouse, hall, and production area) established early in SP III continued to be in use through SP IV–VI. The cultivation continued within a slightly increased area that covered the western half of Area 5. In the central part of Area 5 a boundary was established between the field to the west and a designated farmyard to the east. The border is seemingly not physically marked at this stage, but seems to be maintained throughout the entirety of SP III. In Area 2, the build-up of colluvial deposits around Profile 15653 continued. The field seems to have been permanent and in continuous use, as the micromorphology samples from the cultivation deposits show little evidence of periods of disuse or reestablishment of grass turfs. Numerous cooking pits were dug into cultivation deposits in Areas 2 and 5. The cooking pits in Area 5 formed a concentration of such features (concentration A in Bauer, Fig. 13.1) close to the settlement and farmyard. The radiocarbon dates show that cooking pits were used throughout SP III, the youngest slightly predating AD 600 (Beta-304877). The establishment of a defined farmyard in SP III is attainable first and foremost through the remains of two buildings: A10 and A13. Few features from A10 were preserved. This building was located to the north-east in Area 1, at the top of the slope towards the sea, and thus was clearly visible from the strait. The building remains comprised a hearth and a wall ditch; radiocarbon dates from hearth and ditch indicate use from the mid or late 3rd century until the early 5th century AD, that is, the first half of SP III. This period for A10 partially overlaps with the use of a three-aisled longhouse (A13) in Area 5. The latter building contained features radiocarbon dated to the entire SP III. However, the earliest 14C dates from the hearth are probably due to the significant input of large oak logs in the fuel selection (the “old wood” problem is discussed in detail in relation to longhouse A13 in Østmo and Bauer, Ch.  7:128). Other features in longhouse A13 date to the 4th century onwards, and the longhouse probably belongs to the latter part of SP III. The area between the buildings was too truncated to permit any conclusions regarding the activities taking place there in SP III. However, Area 6 is another defining aspect of the SP III farmyard, situated immediately south-east of A13 and dedicated to production activities such as iron smithing resulting in slags from primary and secondary processes. Other types of activities include what was likely smoking or curing of meat in a low-temperature oven. Furthermore, there were microstratigraphic indications of crafts using ashy fluxes. Parallel to some of these processes, waste was dumped there. Several graves date to SP III, of which those in the Flaghaug grave mound are the most prominent (Østmo and Bauer, Ch. 12:231–5; Stylegar and Reiersen, Ch. 22:564–6). In what originally appears to have been a Bronze Age grave mound, at least four burials were interred in SP III (4–7, Fig. 6.3). The well-furnished of these dates to the 3rd century, the first part of SP III. The raised stone monument just south-west of Flaghaug most likely was built in the first part of SP III (14, Fig. 6.3; Skre, Ch. 23:663);

92 

 B: Excavation Results 2011–12

similarly, an addition of earth and stones in the south-western part of the Kjellerhaug grave mound that lay atop a piece of plano-convex slag likely originated from the SP III iron smithing immediately south of the mound. In addition to these three monuments there are the two above-mentioned cremations east of Kjellerhaug grave mound (9, 11, Figs.  6.3 and 12.1), 14 recorded grave mounds or cairns across the Avaldsnes headland (15, 21, 23–35, Fig. 6.3), as well as the already mentioned stone packing dated to SP II or III with a secondary cremation burial at its southern end (16–17, Fig. 6.3) at Kongshaug. The mounds are believed to be of an SP III–IV origin, while it is possible that some of the cairns date to the Bronze Age. In Area 8, boathouse A40 was built in the first half of SP III, possibly as early as the late 3rd century (Beta-324648). The boathouse was rebuilt or adjusted in late SP III, in the 6th century (Beta-324650, Beta-324651). The boathouse originally had room for a 20-metre long rowing ship. The hearths in A10 and A13 as well as the more prominent cooking pits located closest to the farmyard were fuelled with oak logs. Compared to the mixed fuel assemblages in other contexts, the structured selection of oak fuel is possibly an example of conspicuous consumption, due to this valuable construction material likely being scarce in Kormt in this period (for other possible explanations regarding the fuel selection addressed in this volume, see Østmo and Bauer, Ch. 7:125, 132–3; Ballantyne et al., Ch. 19:480–1). As such, the fuel selection supplements the image of a magnate farm comprising a central settlement area with two buildings clearly visible from the strait surrounded by a boathouse for a vessel with military potential, at least three monumental tombs, and a production area, as well as a large cultivated field.

6.4 Site Period IV: c. AD 600–900 Some aspects of the spatial organisation of the site that became fixed in SP III continue into SP IV. However, adjustments to this organisation were made, such as a possible relocation of dwellings, the designation of Area 6 mainly for food processing and storage, and the construction of a palisade on the eastern edge of the settlement plateau (Fig. 6.4). The archaeological record indicates substantial changes from around AD 600, including the lack of substantial dwelling remains within the excavated areas. Although building remains from the 7th–8th centuries are found in western Norway, the period is characterised by a smaller number of identified buildings compared to the preceding Roman Iron Age and Migration Period. However, it appears that the Avaldsnes farmyard continued to be used, and the border towards the infield was maintained. Indirect indications of settlement make it likely that dwellings were located close by, but outside of the excavated areas or in areas that were heavily truncated in later periods, thus removing any trace of dwellings or other buildings. Boat-

6 Østmo and Bauer: Site Periods and Key Contexts 

Possible relocation of buildings; no coherent building remains found

BUILDINGS AND SETTLEMENT

Posthole, ditch, hearth, etc

Fertilisation pratices near Area 5 suggests settlement in the vincity.

 93

FORTIFICATION

PRODUCTION

STONE CONSTRUCTION A20

Area 6: Features indicating processing and storage of foodstuffs. Possibly some metalwork.

Single posthole Ua-45349 A46825

AD 548–605

Pit Ua-45358 A44483

Area 3: Settlement traces likely dating to SP II–IV

AD 555–615

20

AGRICULTURE

22

Cultivation deposit

aug

15653

Areas 2, 3, 4: permanent field in continuous use in SP III–VII. Individual layers have not been dated.

BOATHOUSE Wall bank

20646

Kong sh

High concentration of grains in Area 6 may indicate instensified cultivation, expansion of fields, and/or relocation of storage area.

1

BOATHOUSE A40

5

2

BOATHOUSE A41

6

35285

19 Trench 4, 2005 excavation

GRAVES

4

Inhumation grave Likely dating to 1st milennium AD

1 0

50 m

3

Area number

15653 Profile number 10 Grave number

Boat grave Likely dating to 7th–10th century Inhumation grave (secondary) Likely dating to SP III–V

Estimated minimum extent of farmyard Possible location of main settlement Estimated minimum extent of cultivation Fortification Estimated extent of production area Boathouse Grave

SEA LEVELS Current AD 800 (+1,6 m)

Fig. 6.4: Main features of Site Period IV. Key contexts and radiocarbon dates mentioned in the text are labelled above; an overall interpretation of the land use and spatial organisation within SP IV is presented in the lower. Numbering of graves refer to map key in Østmo and Bauer, Fig. 12.1 and Tab. 12.1. Illustration: I. T. Bøckman, MCH.

94 

 B: Excavation Results 2011–12

house A40 fell out of use by the first decades of the 7th century, simultaneous to the construction of a new, smaller boathouse (A41). SP IV also sees production processes in Area 6, but the activities change from a varied production to more dedicated processing and storage of foodstuffs. Geochemical indications of low-scale brass work could belong to SP IV or V, though the activity cannot be stratigraphically connected to any specific phase (Cannell et al., Ch.  18:451; Østmo, Ch.  9:176–7). The character of Area 6 also changed to that of an enclosed and protected area as a result of the building of a large stone construction (A20) constituting the base of a palisade on the eastern edge of the plateau. The continued cultivation from SP III, through SP IV, and up to modern times can be observed in Profile 15653 in Area 2, Profile 35285 in Area 3, and Profile 20646 in Area 5 (Bauer and Østmo, Figs. 8.3, 8.5–6). The cultivation deposits do not indicate any changes during SP IV regarding the cultivated area’s extent, intensity, or methods of cultivation. The high concentrations of grain in Area 6 could however indicate intensified cultivation or extension of the cultivated area, or alternatively a relocation of grain processing compared to earlier Site Periods; the two are not mutually exclusive. Probably, the changes in activities in Area 6 coincided with the construction of palisade A20 during the first half of the 7th century. Features lying under A20 have radiocarbon dates ending c. AD 600. Waste deposits containing latrine dumping seem to be part of an activity parallel to and partially post-dating the SP III production, likely ending around AD 600 or shortly thereafter. The earliest grain-rich features date to the mid-7th century and are located west of A20. In addition to grain processing, there are indications in Area 6 of food preservation by means of seaweed ash. This activity may have begun a little earlier as some fragments are found in SP III features. Food processing and storage appears to be one of the activities and resources that the palisade was intended to protect. The lack of other SP IV remains above the palisade remains may indicate that it was in use throughout SP IV. The placement of such a palisade on a geological scarp with a vista over the Karmsund strait would also have added to the monumental façade of the settlement towards seafaring travellers. Other manifestations of ritual and power dated to this period include a boat grave, an assumed inhumation grave without visual marking that lacks secure dating, and a secondary inhumation in the stone packing at the Kongshaug ridge (20, 22, 19, Fig.  6.4; Stylegar and Reiersen, Ch.  22; Østmo and Bauer, Ch.  12). Except the older cairn and the stone packing, the graves at the Kongshaug ridge do not have visual markers above ground today, but whether this was the case originally is unknown. As mentioned under SP III, several of the graves at Kongshaug and other possible grave mounds across the Avaldsnes headland have not been dated by radiocarbon or artefacts and are of an assumed Iron Age origin. Several building remains in Area 5 do not form any coherent buildings. A waste pit and a posthole are radiocarbon dated, with the dating spectrum stretching into the earliest part of SP IV (Ua-45358, Ua-45349). Although clear SP IV building remains have not been identified in Area 5, fertilisation practices in the western part of the



6 Østmo and Bauer: Site Periods and Key Contexts 

 95

area suggest that settlement continued there or in the close vicinity. Whereas the deposits in Area 2 were fertilised primarily using dung, the deposits in the western part of Area 5 included a higher proportion of settlement waste. The area occupied by the present cemetery north of Area 5 and west of the raised stones is a possible location for the settlement of SP IV. Other alternatives are the area south-south-west of Area 6 or along the eastern edge of Area 1, which was later occupied by a high-medieval masonry building (Bauer, Ch. 14) and the post-medieval rectory (Bauer, Ch. 15).

6.5 Site Period V: c. AD 900–1250 The activities at Avaldsnes in SP V were very much a continuation from SP IV, with one substantial difference: on the same spot where the building A10 stood in early SP III, a row of postholes that constitute a possible building (A14) was raised in the 10th or early 11th centuries. This possible re-establishment of a building with a vista of the Karmsund strait, highly visible to travellers, characterises SP V (Fig. 6.5). A14 is poorly preserved and initiates a period of rather vague or indirect settlement indications. Other parts of the site bear witness to stability and continuation with regard to cultivation (Areas 2 and 5) and the processing of grain (Area 6). A14 consisted of a row of postholes cut into the wall ditch of SP III building A10. The row is probably the remains of a wall, though an alternative interpretation as a fence cannot be dismissed. However, three depressions in the bedrock forming a parallel row to the west suggest a possible western wall (Østmo and Bauer, Fig. 7.9). As the material available for radiocarbon dates is limited to fragments of charred birch in the back-fill of postholes without preserved post prints, it is impossible to consider whether the dates represent the time of construction or deconstruction of the building. Calculated probability for the different sub-sequences within the 1 and 2 Sigma range of each sample combined with the overlap of radiocarbon dating results from the postholes may indicate a slightly higher chance for an overall date to late 10th century and early 11th century. However, it remains possible that the building was raised earlier in the 10th century. A radiocarbon date from the hearth of SP III building A10 spanning from the mid-11th to mid-12th century (TRa-4240) is probably the result of an intrusion. Although the connection between the SP V post- and stakeholes in Area 1 is difficult to confirm, their number, alignment, and narrow dating frame indicate that at least one building was raised there in the 10th–11th century. Two postholes in Area 5 date to the early 11th to mid-12th century (Ua-45371 and Ua-45380). A large concentration of heated and cracked stones covered the postholes. The stone layer contained loom-weights and schist griddle stones likely dating to the 9th–13th century while the near absence of ceramic sherds may push the dating towards the late Viking Age, that is, the early 11th century. Such deposits containing

96 

 B: Excavation Results 2011–12

BUILDINGS AND SETTLEMENT Ditch A18206 Possible border between farmyard and field Posthole in ditch A18206 Ua-45348 A48560

AD 990–1029

Single posthole Ua-45380 A52453

AD 1015–1148

Single posthole Ua-45371 A46764

AD 1044–1157

GRAVES

Posthole

FLAGHAUG Secondary burial Flag Flaghaug 8

BUILDING A14 Hearth TRa-4240 A8957

AD 1041–1154

A14

AGRICULTURE Cultivation deposit

1

Areas 2, 3, 4: permanent field in continuous use in SP III–VII. Individual layers have not been dated.

5 2

6

FOOD PROCESSING AND STORAGE A18505 Concentration of heated and cracked stones, used for heating water. The layer also contained loom-weights and schist griddle stones Continued grain processing indicated by corn-drying kiln Ua-45360 A44031

AD 1033–1152

4

3 0

50 m

1 10

Area number Grave number

Estimated minimum extent of farmyard Possible location of main settlement Estimated minimum extent of cultivation Estimated extent of production area Grave Hypothetical, small wooden church preceeding the one built c. 1250

SEA LEVELS Current AD 1000 (+1,3 m)

Fig. 6.5: Main features of Site Period V. Key contexts and radiocarbon dates mentioned in the text are labelled above; an overall interpretation of the land use and spatial organisation within SP V is presented in the lower. Numbering of graves refer to map key in Østmo and Bauer, Fig. 12.1 and Tab. 12.1. Illustration: I. T. Bøckman, MCH.



6 Østmo and Bauer: Site Periods and Key Contexts 

 97

heaps of stones – discarded after having been used for heating water – are a normal occurrence in the outskirts of farmyards in this period (Østmo and Bauer, Ch. 7). Along with the objects used for weaving and bread baking, the concentration of stones indicates that the SP V farmyard was lying close by, possibly in Area 5 or in the heavily truncated western part of Area 1. This fits rather well with a ditch and a row of small posts in the central part of Area 5 that seem to reaffirm the border between the field to the west and farmyard to the east. One of the postholes dates to the turn of the first millennium AD (Ua-45348). In Area 6, the only grain-rich feature postdating the last decades of the 10th century was a corn-drying kiln dated to AD 1033–1152 (Ua-45360). Though the grain processing in this area does seem to continue well into SP V, this use of the area may have seen its highest intensity in late SP IV and the first century of SP V.

6.6 Site Period VI: c. AD 1250–1368 The most prominent feature of SP VI is the standing St Óláfr’s Church, the construction of which began around 1250. A masonry cellar from a building (A12) on the eastern edge of the settlement plateau to the south of the church, several stone pavings, a subterranean passageway, and an octagonal masonry building are other features of the SP VI complex (Fig. 6.6). Characteristic for SP VI is the masonry architecture, representing a substantial change in the monumentality, functionality, and construction technology of the known features of the royal manor at Avaldsnes. The high-medieval complex was erected within an existing royal manor estate, although there are few archaeological sources that can illuminate the form and layout of the manor in SP VI. A preceding church is mentioned in written sources, likely in the shape of a wooden church in the same area as the SP VI stone church. A Romanesque masonry church predating the present Gothic church is also possible. In AD 1308 the church became a royal collegiate chapel, and documentary evidence indicates that the king stayed at Avaldsnes on several occasions in the early 14th century (Bauer, Ch. 14; Mundal, Ch. 3:44). Based on the excavated features, a royal manor complex of buildings and features is suggested, situated in the same general area where the farmyard was established in SP III. Known elements in the complex comprise the church, several stone pavings, including a stone-paved walkway at least 25 metres long (A32545) in Area 5, building A12, a subterranean passageway in Area 1, and a demolished octagonal building that according to documentary evidence was standing in the cemetery south of the church (Bauer, Ch.  14). The exposed part of the subterranean passageway was more than 30 metres long; it is assumed to date to the Middle Ages based on parallels from medieval Iceland as well as the stratigraphic relationship to a stone foundation for a building, believed to date to the Middle Ages (Haavaldsen 1989a:76–7). The presence of the subterranean passageway

98 

 B: Excavation Results 2011–12

Ditch A18206; division between field and farmyard in SPV Dated backfill indicating relocation of the farmyard further west: Ua-45344

AD 1299–1394

ROYAL MANOR COMPLEX

St Óláfr’s Church

C. 1250

Subterranean passageway

Storage/dwelling building A12 Likely dating to late 13th c.

Stone-paved walkway

Octagonal building

1

AGRICULTURE Cultivation deposit

2

PRODUCTION Little activity in Area 6

5

Areas 2, 3, 4: permanent field in continuous use in SP III–VII. Individual layers have not been dated.

Waste layer/ deposit Ua-45368 A12780

AD 1221–1261 6

4

3 0

50 m

1

Area number

Royal manor complex Buildings Estimated minimum extent of cultivation

Grave

SEA LEVELS Current AD 1250 (+1 m)

Fig. 6.6: Main features of Site Period VI. Key contexts and radiocarbon dates mentioned in the text are labelled above; an overall interpretation of the land use and spatial organisation within SP V is presented in the lower. Illustration: I. T. Bøckman, MCH.



6 Østmo and Bauer: Site Periods and Key Contexts 

 99

signals the need for escape routes in turbulent times; for example, the Hanseatic raid of Avaldsnes in 1368, which included the burning down of royal properties (Mundal, Ch 3:44). This attack appears to have precipitated the downfall of Avaldsnes as a royal manor and marks the end of SP VI. The excavated remains of building A12 were a masonry cellar with a doorway towards the strait, likely used as a storage room for goods combined with a dwelling for the early 14th-century royal collegiate. The cellar was part of a larger building, as it likely had one or two additional floors above and continued northwards outside the excavated area. Previous surveys have documented activity in the harbour area at Gloppe in SP VI. Large underwater waste deposits related to trade activities indicate intensive use; finds include medieval pottery and a shipwreck likely dated to the 13th century (Opedal et al. 2001:108–13). Although the upper floors of A12 may have served as residence for clergy and possibly the visiting king, direct evidence of other dwellings was not identified. The stone-paved walkway in the west indicates that the division between field and farmyard was moved slightly further west compared to its location in SP V. Back-fill in the ditch that served as the division in SP V was dated to the 14th century (Ua-45344); thus, material from farmyard activities in SP VI seems to have been redeposited in the earlier ditch. The paved walkway is difficult to date precisely but its orientation and similarity to pavings south-west of A12 suggests its contemporaneity with building A12 and St Óláfr’s Church. In Area 6, deposit A12780, cutting into several older features and containing burnt clay, charcoal, cereal, and burnt animal bones, was dated to the mid-13th century (Ua-45368), that is, much later than other features in this area. The lack of coherent buildings or other constructions related to the dated deposit could suggest that there was little activity here in SP VI, or alternatively that such remains were truncated by later activities in the rectory farmyard.

6.7 Site Period VII: 1368–1900 After the 1368 fire, no royal documents issued at Avaldsnes have been preserved (Mundal, Ch. 3:44), suggesting that the administrative functions of the royal manor ceased. This development was probably connected to the increased role of towns in royal affairs, and because, due to the joint kingship 1380–1814 between Denmark and Norway, the king was residing in Copenhagen. However, activity in the Avaldsnes farmyard area did not cease after the 1368 fire. While there are no clear building traces from the following century, there are deposits overlying A10/A14 that are radiocarbon dated from the mid-15th to the mid-17th century (TRa-4233, TRa-4239, and TRa-4244). Written sources, paintings, archaeological artefacts, and additional radiocarbon dating results contribute to the knowledge of the farm at Avaldsnes in recent centu-

100 

 B: Excavation Results 2011–12

ries, but such features or constructions beyond the time frame investigated by ARM were not excavated and documented unless they covered features targeted due to the project specific aims (Fig. 6.7). The post-Medieval rectory was situated immediately to the south of St Óláfr’s Church, in the area where the farmyard appears to have been situated since SP III. While the evidence is scant for the early period, it is richer for the period following the fire that destroyed the rectory in 1698, of which several presumed traces were found. The rectory was rebuilt on the same site after the fire. Excavated building remains of the post-Medieval rectory post- and predating the 1698 fire consisted primarily of collections of stones forming rows or other more or less distinguishable albeit fragmented patterns. Most of the stones were probably part of foundations; the rectory farmyard consisted of several buildings, including dwellings and economic buildings that were built, rebuilt, or torn down through the centuries, some of which are described in documentary sources (Fyllingsnes 2008; Bauer, Ch. 15). A few remains of the old rectory are still preserved in Area 1, south of the church: a small building that previously was the storehouse from the old farmyard and a stone-built well (A11062). The well was probably in use for centuries, but dated deposits from the bottom of the construction were from AD 1697–1917 (Beta-319021). Paintings and surveys at the Gloppe Peninsula reveal several buildings and building remains, of which some likely relate to the Hanseatic presence in the area in SP VI–early VII. Trade is documented by written sources as well as voluminous waste deposits in the strait around Gloppe (Opedal et al. 2001:113–20). From the 16th century onwards the priests at Avaldsnes are supposed to have managed an inn at Gloppe, although it seems this activity did not produce significant income (Lindanger 1999:174–80). By the 18th and 19th centuries a new trade centre developed at Gloppe, which had also become the location for the local thing assembly (Lillehammer 1989). Several excavated features testify to the rectory garden, which is also visible in paintings of Avaldsnes (Bauer, Figs. 15.1–3). By the end of SP VI, a new dwelling was built for the rector at the site of the present Nordvegen History Centre, marking the end of perhaps 1700 years of a the main dwelling being situated in the farmyard on the large plateau overlooking the strait. In the mid-16th century the church fell into disrepair, and for a long time only the chancel was in use. The rest of the church fell into ruin until it was restored in the 19th century. In the 1920s, St Óláfr’s Church achieved its present form, including a large western tower, presumably similar to the medieval version of the church. By an impressive act of camouflage, the newly restored church avoided demolition by the German occupiers during the Second World War – they feared it would aid the navigation of English air pilots – thus it remains to provide the signature silhouette of Avaldsnes as a place of history and power.

6 Østmo and Bauer: Site Periods and Key Contexts 

RECTORY Few traces of buildings older than the 1698 fire. Several later building foundations from the period after the fire, as well as waste and settlement deposits were uncovered. The rectory was moved to the present Nordvegen site at the end of SP VII.

SETTLEMENT DEPOSITS TRa-4233 A6488

AD 1448–1613

TRa-4239 A19880

AD 1450–1615

TRa-4244 A20326

AD 1491–1632

 101

CHURCH

Stone-built well likely in use for centuries; bottom deposits dated Beta-319021 A11062

St Óláfr’s church Estimated location of cemetery walls: Before the expansion in the 1840s

AD 1690–1926

Storage house (still standing)

1840s–1900

Garden deposit Collapsed garden wall

AGRICULTURE 1

Cultivation deposit Areas 2, 3, 4: permanent field in continuous use in SP III–VII. Individual layers have not been dated.

5 2

6

Nordvegen

4

0

50 m

3

1

Area number

14th–19th century rectory area Buildings: Arne Berg’s suggested plan of the rectory in the mid19th century Storage house (still standing) 19th–20th century rectory area Maximum extent of churchyard before 1900 Estimated maximum extent of cultivation Grave Flaghaug excavated and removed in1834-5 SEA LEVELS Current

Fig. 6.7: Main features of Site Period VII. Key contexts and radiocarbon dates mentioned in the text are labelled above; an overall interpretation of the land use and spatial organisation within SP VI is presented in the lower. Illustration: I. T. Bøckman, MCH.

102 

 B: Excavation Results 2011–12

Fig. 6.8: The St Óláfr’s Church camouflaged by the German authorities during the Second World War. Owner: The Directorate of Cultural Heritage’s archives.

Mari Arentz Østmo and Egil Lindhart Bauer

7 The Prehistoric Settlement and Buildings This chapter describes and discusses three identified and one possible prehistoric building at Avaldsnes, alongside an overview of the settlement’s organisational layout during various periods of prehistory. The buildings vary in age, size, and function; all these aspects are discussed. Each building is treated in its own section, with an initial description that clarifies the building’s essential construction elements, followed by discussion of its possible functions. The poor preservation of the buildings, particularly in Area 1, complicates the interpretation, as aspects of their construction and functions remain unknown. The archaeological features are the main material for the discussion, supplemented by macrofossil and micromorphology analyses, osteology, soil chemistry, and artefacts. The excavations, coupled with previous surveys at Avaldsnes, indicate a settlement that existed over a long period of time, during which the site’s organisation became increasingly fixed and the site’s various functions increasingly localised in specific areas.

The excavation plan for the Avaldsnes Royal Manor Project was conducted with the stated aim of examining those elements that might elucidate the settlement’s social status, as well as identifying possible reorganisations of the settlement that could imply changes in social status. The fragmentary building remains uncovered were difficult to interpret, but a tentative assessment from the stated perspective is still plausible. Prior to the excavations of the current project, no complete prehistoric buildings were known at the Avaldsnes headland, although postholes and other construction remains had been identified in Areas 1–6 and on the Kongshaug ridge in the course of surveys carried out by the Museum of Archaeology in Stavanger in the period 1992– 2006 (Bauer and Østmo, Ch. 5:68–70). As previous investigations were limited to surveying, most of construction remains were documented only superficially, and only a small selection were radiocarbon dated or dated by stratigraphic relations. As such, the 2011–12 excavations provided a wealth of new information in terms of the chronology, spatial distribution, and function of the features excavated. Some of the areas surveyed earlier were not included in the excavation areas of the Avaldsnes Royal Manor Project, leaving unknown the character of these settlement remains, particularly the features at Kongshaug. This elevated ridge running north– south for approximately 200 meters near the western border of Area 2 was dominated by grave monuments of various forms and dates (Østmo and Bauer, Ch.  12:243–5), most prominently a large stone packing in the southern end containing at least two graves. This grave monument was built over a group of postholes and cooking pits. The grave monument is believed to date from late Site Period (SP) II or early SP III, while the settlement traces must therefore predate this (Hafsaas 2005:12–17; 2006:22– 3). It is unknown whether these postholes were components of dwellings or merely fences or other minor constructions. An undated, rectangular posthole was docu-

104 

 B: Excavation Results 2011–12

mented underneath the present-day barn by the southern end of Area 6 (Hemdorff 1994). The posthole’s size (approximately 0.5 × 0.4 m) could imply a building, but the degree of modern disturbances precluded further examination in the adjacent area. Features discovered during earlier surveys point to the likelihood that while the settlement at Avaldsnes included other buildings in addition to those described below, these constructions are either lost due to disturbance from later activities – especially in the central parts of the modern farmyard – or unknown as they are located outside the limits of the excavation areas. The three buildings that could be distinguished with certainty consist of a Bronze Age building (A11) from SP I located in Area 2 and two buildings from early and mid-SP III interpreted as a possible hall building (A10) and a longhouse (A13), located in Areas 1 and 5, respectively (Fig. 7.1). A possible fourth building (A14), located in Area 1, dates to early SP V; its function has not been defined due to poor preservation. All building remains were exposed by stripping away the overburden with a mechanised digger. The excavation was complicated by modern disturbances, thick colluvium, modern roads, and obstructing buildings; hence the great variation among the buildings in their degree of preservation and availability for exposure. Numerous other construction remains such as postholes occur across the site, with comparatively higher density in Areas 5 and 6, but none could be identified as consistent with buildings. The excavation results provide a fragmented representation of prehistoric settlement; the possibility of Iron Age buildings that have been disturbed beyond recognition or are located in unexcavated areas cannot be dismissed. Still, two primary periods of settlement are identified, the first in SP I and the second from SP III to SP V. However, some aspects of the spatial organisation as it existed from SP III gradually emerged during SP II. In SP II, the cultivated areas were expanded and permanently established, while the first buildings in what was to become the farmyard appeared in SP III.

7.1 SP I – Building A11: Building of uncertain function Building A11 was built sometime between c. 1600–600 BC in the central northern part of Area 2, covered by roughly a meter of colluvial cultivation deposits. The building consisted of a U-shaped ditch (A13306), probably indicating the line of the walls. Two postholes lay inside the building, but could not be securely related to A11. The wall ditch’s north-western side measured 6.7 meters, the south-western 6.3 meters, and the north-eastern 4.6 meters. The north-eastern wall was possibly shorter due to truncation by later cultivation. Ard marks are cut into deposits both stratigraphically above and below the wall ditch, demonstrating that the area was cultivated both before and after the building stood there (Bauer and Østmo, Ch. 8:144–5). The ditch was up to

7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 105

A14 features A10

A13

A11

0

20 m

Prehistoric building Excavated area Road and car park Existing building Cemetery wall

Fig. 7.1: The location of buildings A10, A11, A13, and A14 (?). Illustration: I. T. Bøckman, MCH.

40 cm wide and cut up to 18 cm into the subsoil. The area within the wall ditch measured about 38 m2, but it is uncertain whether this represents the entire building. The slight curvature in the southern part of the ditch’s south-western side suggests the building’s termination. A 0.6 m wide opening in the ditch’s north-eastern part may represent an entrance. The ditch was diffuse in this area, so the presumed opening might alternatively be a result of truncation. A collection of stones lay scattered slightly off-centre within the wall ditch area. Between and around the stones were several imprints of removed stones. An interpretation as a hearth was considered, but the lack of fire-cracked stones and charcoal makes this unlikely. Soil chemistry samples from the area where the wall ditch was located showed decreased values

106 

 B: Excavation Results 2011–12

of magnetic susceptibility immediately around these stones and features compared to those in the area to the south and west (Macphail and Linderholm, Ch. 17:388–92, Fig. 17.6.b) – another indication it was not a hearth. No artefacts indicative of the building’s functions were recovered from the wall ditch or any other part of the area. The building’s stratigraphic position between dated deposits placed it in the Bronze Age. In the north-eastern part, the wall ditch cut into a deposit (A4216) dated to 1608–1501 BC (Beta-304878). The agricultural deposit (A5882) covering the wall ditch was dated to 795–595 BC (TRa-4231). Consequently, the building was constructed and in use between these periods. It is unlikely that the wall ditch was part of a larger building stretching further south-east. Firstly, this is due to the aforementioned inward curvature of the south-western side of the ditch, which suggests an almost quadratic shape for the building. Secondly, if the building had extended further south-east, building features such as postholes or continuing wall ditches would be expected, particularly as the surface sloped down towards the south-east, lower than the exposed building remains, and therefore probably would have been less disturbed by cultivation. The uneven shape of the cut for the trench seen in Profile D and partly in Profile B could represent poorly preserved postholes within the ditch (detail in Fig.  7.2); however, such features are not sufficient evidence for drawing definite conclusions about the wall construction or any roof construction. The ditch’s curved corners make sill-beams less likely than possible postholes placed directly into the ditch. A wattle-and-daub wall is possible, but such a construction would probably have left traces of stakeholes within the ditch; none were found in the four excavated segments. It is possible that the ditch, rather than forming part of the wall, represented a drainage ditch below the building’s eaves. In that case, the roof-bearing construction must have been further into the building. Only two postholes were found inside the building, and their stratigraphic relationship to the building was uncertain. Stakeholes littered the area; their stratigraphy demonstrated that, rather than belonging to the building, they are related to a later phase, possibly as parts of hay-drying racks or fences concentrated along the northern edge of the cultivated field or pasture. There are many uncertainties regarding the construction and thus building type. Based on the observations accounted for above, there is little evidence that building A11 was a dwelling. There is no evidence of a roof or a hearth, so the construction was probably simple and might have functioned as some sort of outbuilding, perhaps a byre or a storage building. The soil chemistry mapping indicated that the surfaces outside the building were manured and cultivated (Macphail and Linderholm, Ch. 17:388), while diverging readings suggest a possible division between the surrounding surfaces where the agricultural activities took place and the area inside the building. Details regarding the building’s construction remain uncertain. Similar building remains exist in Rogaland and in other parts of the country as well as from other time periods. Børsheim et al. (2002:167–71) thoroughly account for a building with a U-shaped ditch at Gausel, along with similar building occurrences

7 Østmo and Bauer: Prehistoric Settlement and Buildings 

0

1

 107

3m

B A

D

C

A

NE

SW

B

E

W

Wall ditch

C

NNW

SSE

D

NW

SE

Posthole Stakehole Ard mark Imprint of removed stone

0

50 cm

Stone Excavated area

Fig. 7.2: Wall ditch with profiles, and other features in and around building A11. For the soil chemistry mapping, see Macphail and Linderholm, Ch. 17:Fig. 17.6. Illustration: I. T. Bøckman, MCH.

108 

 B: Excavation Results 2011–12

in Rogaland (Bårdsgård 1981; Haavaldsen 1984; Løken 1987; Steen 1995; Skare 1998). On the basis of similarities with the other buildings, the Gausel building is dated to the late Iron Age; however, some material from the Gausel building was radiocarbon dated to the late Stone Age. The dating of the other buildings spanned from the early Bronze Age through the pre-Roman Iron Age to the Merovingian Period. The U-shaped ditch at Stavnheim in Hå was of similar size to the ditch at Avaldsnes, although somewhat wider and more irregular (Børsheim et al. 2002:fig.136). This construction was dated to the pre-Roman Iron Age and had no postholes from roof-bearing posts. Similarly to the Avaldsnes building, it had an opening in one corner, although it is unknown whether this is a coincidence or whether the void represents an entrance. In fact, several of the U-shaped wall ditches in Rogaland had a small opening (Børsheim et al. 2002:fig.136), indicating the presence of doorways. Most of the buildings with U-shaped or round wall ditches had clear traces of postholes in a square pattern. The lack of such postholes in the Avaldsnes building suggests a different construction method or that such construction traces were shallow and removed by later cultivation. Near Værnes church in Nord-Trøndelag, a similar feature was exposed in 1999– 2000 (Gundersen 2001:22). This building consisted of a U-shaped wall ditch measuring 6.7 by 6 meters, thus of similar size to the building at Avaldsnes. No traces of internal constructions were found either within the ditch or in the area inside the building. The macrofossil evidence provided no clue as to the function of the building, and no interpretation of the building was suggested. The building was dated to AD 245–415 (Gundersen 2001:22). If the dating of the buildings in Nord-Trøndelag and in Rogaland is correct, it indicates that this building type was in use throughout much of prehistory, at least from the Bronze Age and well into the Iron Age. Similar features found in other parts of the country demonstrate further that this building type was not exclusive to western Norway (e.  g., Diinhoff 2005a:78; Nilsen 2005:4). As of yet, no one has been able to present a convincing interpretation of this type of remains. However, that lack of hearth in most of the buildings makes it unlikely that the building type represents a dwelling; thus, it probably is related to farm-economic activities.

7.2 SP III – Building A10: Possible hall building Building A10 lay at the eastern edge of the settlement plateau, in a part of Area 1 containing multiple building remains. The distribution of building remains, coupled with the radiocarbon dating results, suggests that two separate buildings lay in the same location: one from early SP III (A10) and one from early SP V (A14, see section below); see Figs. 7.1 and 7.4. The remains from the earlier of the two buildings were more accessible and could contribute to a suggested building plan of two hearths, a



7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 109

Fig. 7.3: Overview of the area where remains of buildings A10 and A14 were found. Facing north. Stripes from the teeth of mechanical diggers are clearly visible in the area south and west of building A10’s hearth. Photo: MCH.

wall ditch, and cluster of five postholes potentially bearing some relationship to the building. The remains from the more recent building are dealt with in a later section; first, an account of some general observations of the area. The features from the two buildings consisted of postholes, a wall ditch, and two hearths, the larger of which yielded three radiocarbon dates to early SP III and one to SP V. Recent activities had truncated most of the features, making them fragmented and shallow. The truncations were caused by a range of activities, including those related to the high-medieval royal manor, the St Óláfr’s Church, and the post-Medieval rectory. Modern disturbances include the levelling of the terrain to create a car park and a cable ditch (see impact of disturbance in Fig. 7.3). The later use of the area is attested by patches of cultural deposits and features with finds dated to medieval or more recent times. Macrofossils from all features in this area were relatively scarce, restricting the datable material. In addition, micromorphological analyses showed that the features in the area had been affected by bioturbation (Macphail 2012a:8–12, 38–48). Vertical burrowing by earthworms may have contributed to homogenisation

110 

 B: Excavation Results 2011–12

between overlapping deposits or features. In fact, the younger postholes (overview in Fig. 7.4) cut into the wall ditch could only be observed where the cut was deeper or wider than the ditch, as there was hardly any colour contrast between different fills (cf. Canti 2003a:139). Another indication of worm activity was a “pea grit horizon” – a layer of tiny stones at the bottom of certain features, where the bedrock prevented further burrowing (Canti 2003a:143). The pea grit included tiny fragments of modern brick from the overlying layers, indicating vertical transportation via worm burrows. In addition to these disturbances, the generally scarce volume of macrofossils indicates that the 14C dating results should be treated with caution. Dating results are more reliable when they form a consistent pattern or are in line with interpretations based on the spatial relationship between features. Two such cases are the set of SP III dates from hearths and wall ditch fill from building A10 and the set of SP V dates from postholes within this ditch; it is thus reasonable to infer that the latter are the remains of a later construction (A14). The features belonging to the building from SP III have been identified based on their spatial relationship, supported by coinciding radiocarbon dates. The features comprise two hearths (A5793 and A8957) and a wall ditch (A9231), and possibly a disturbed cluster of postholes north of the hearths (Fig.  7.4). An estimated ground plan, based on the curvature of the wall ditch and the internal distance between wall ditch and hearth, indicates that the building was 18–20 meters long by about 6 meters wide, thus covering an area of 108–120 m2. The wall ditch was preserved for a length of 16.7 m, but truncated at both ends. In the north it was obscured by the cemetery wall, while in the south it was cut by a modern ditch. Nonetheless, given its curved ends, it is unlikely that the wall ditch extended significantly into the present-day cemetery or beyond the modern ditch. The wall ditch, measuring about 0.3 m wide and up to 0.2 m deep with straight sides and a flat bottom, was cut into the green schist bedrock. While no trace of a corresponding western ditch was found, the bedrock formation does slope towards the west, leaving open the possibility that a ditch could have been dug only in the covering soil; any remains of such a feature would have been removed at the time the car park was levelled, if not earlier due to farmyard activities. The ditch infill has a terminus post quem date to AD 236–333 (TRa-4246), which corresponds with the fuel in hearth A8957 that was dated to AD 243–423 (Beta-222063, TRa-4236, Beta-304879), and with that of hearth A5793 dated to AD 255–345 (TRa-4235) (Fig. 7.4). It should be noted that there is also a date (TRa-4242) from the ditch infill that corresponds with an overlying late- to post-medieval dirt floor (Macphail and Linderholm, Ch. 17:383). Vertical movement due to earthworm activity provides a probable explanation for this phenomenon as well as for the homogenisation that renders the ditch infill indistinguishable from that of the postholes. Only the shape of the cut into bedrock – slightly wider or deeper than the ditch – together with the smaller size of the clustered stones compared to those in the ditch infill allowed the postholes to be identified. Three of the identifiable postholes in the ditch were not radiocarbon dated and could in theory belong to building A10, but as will be discussed later, given their

7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 111

A18745 TRa-4237

AD 258–381

A5793 TRa-4235

A8957 TRa-4236

AD 258–381

AD 255–345

A8957 Beta-222063

A9231 TRa-4242

AD 243–345

AD 1522–1646

A8957 Beta-304879

AD 347–423

A9231 TRa-4246

AD 236–333

Estimated extent of building Features in A10 Posthole Wall ditch Hearth Estimated extent of hearth Flue Other feature Cemetery wall Bedrock Modern ditch Cable ditch Excavated area 0

1

3m

Fig. 7.4: Suggested plan of building A10 with radiocarbon dates. Illustration: I. T. Bøckman, MCH.

112 

 B: Excavation Results 2011–12

shape and size this row of posts are more plausibly related contextually and dated to SP V. One larger posthole lay stratigraphically below the ditch and was dated to AD 135–219 (TRa-4245). As no other traces of posts were found in the ditch, it is likely that the building had a palisade wall construction with standing planks or split logs set directly in the ditch or upon a sill in the ditch. A similar wall construction has been suggested for a hall building at Eide in Gloppen dated to late Roman Iron Age and early Migration Period (Diinhoff 2009:26–7). This building had wall ditches with traces of sills from either a post-and-plank or palisade wall construction. The walls were curved towards the gable end, the sill plate having been cut in short segments to allow for curvature. The lack of physical traces in the ditch in A10 precludes definitive conclusions regarding further details of the wall construction, but standing planks in a sill beam would be in line with the contemporary buildings from other sites in Rogaland (Løken 1991:66–7). There were no preserved traces of roof-bearing construction, either because the postholes had been shallow and removed by later activity or because roof-bearing posts had been placed on stones. Located along the central axis of the building approximately one meter north of the circular hearth was a group of five postholes, one of which was radiocarbon dated to AD 258–381 (TRa-4237), and another tentatively dated by pottery to AD 350–450 (S12768/12, Kristoffersen and Hauken, Fig. 21.5); that is, seemingly to the period when the building was standing. The radiocarbon-dated posthole was 40 cm wide and only 7 cm deep and the best preserved of the group. It is uncertain whether the remaining features are contemporary. The individual postholes were observed at different levels due to modern disturbance as the border of the levelled car park ran through them. As the group of postholes was located along the central axis of the building it is unlikely that they were part of a roof-bearing construction. Possibly they were part of an inner wall or other internal installation, but the poor state of preservation prevents firm conclusions. As their eventual constructional function within A10 is unclear, it remains likewise uncertain whether the pottery and overlapping radiocarbon date indicate some possible relation to the building or whether they simply contain secondary deposited material originating from activities within A10. The large, semi-rectangular hearth was heavily disturbed by recent activities, and its original extent is preserved only in the northern and most of the eastern side of the hearth (estimation of original extent in Fig. 7.4). A furrow in the bedrock ran from the north-eastern corner of the hearth and ended with a slight depression filled with humus and silt. It was initially suggested that this may have functioned as a flue, but as the fill of the furrow was mainly mineral and contained humus and charcoal only where it ran into the hearth, this may rather be a natural bedrock formation. The hearth A8957 contained charcoal mainly dated to the early SP III (Beta-222063, Beta-304879, TRa-4236), but there was also a date to AD 1041–1154 (TRa-4240). The north-eastern part of the hearth, which contained the later date, had two distinct charcoal lenses, indicating reuse (cross-sections in Fig.  7.5). Such reuse was con-

7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 113

A5793

A5793

A8957

B

A (mirrored)

A C

A8957

B C

0

Hearth fill

Stone

Dislodged stone

Charcoal layer

Fire-cracked stone

Bedrock

1m

Micromorphology sample

Fig. 7.5: Cross-sections of hearths A8957 and A5793. Illustration: I. T. Bøckman, MCH.

firmed by the micromorphology analyses (Macphail and Linderholm, Ch.  17:395), likely related to an ongoing process of cleaning out the hearths when the building was in use. An intrusion of younger material, rather than existence of a later SP V hearth overlapping the same location, is the most likely reason for the chronological gaps between dates. A smaller, circular hearth of similar date lay one meter further north, slightly displaced from the building’s central axis and not in physical contact with the possible flue. The differences in size, shape, and fuel types may suggest that the two hearths differed in function, but precisely how they differed is unknown. Some information related to their construction may be lost as both were preserved as shallow cuts into the subsoil, stopping at the bedrock. Indications of maintenance, such as cleaning out ashes, were observed in micromorphology samples from the semi-rectangular hearth. Different layers of charcoal were visible in the sections, suggesting repeated episodes of use and ash removal. The circular hearth contained a scant number of fire-cracked stones as well as a charcoal-rich fill with more concentrated charcoal towards the bottom; however, there was no definitive indication that its function was as a cooking pit. In some of the better-preserved parts of A8957, the sides and parts of the bottom of the hearth were packed with small, angular stones, though this was not consistent throughout the feature and seemed more likely to have been the border

114 

 B: Excavation Results 2011–12

component of a hearth rather than the more random fill characteristic of a cooking pit. While the primary fuel in the large hearth was oak, almost all of the charcoal in the smaller hearth was birch (Ballantyne et al., Ch. 19:480, Tab. 19:4). Many of the oak fragments in the larger hearth displayed weakly curved growth rings, indicating use of large logs for firewood. Ballantyne et al. (Ch.19:481, 507–8) suggest that because wood of this type, highly valued as timber, was probably scarce in the immediate landscape and would have incurred a long regeneration period after cutting, its presence in the large hearth might represent the use of discarded construction parts for fuel – an interpretation supported by the discovery of iron nails in the hearth. The distribution of oak across the site shows particular concentrations in hearths inside buildings and adjacent cooking pits, indicating a structured selection of fuel. Possible interpretations of this pattern are discussed in detail by Ballantyne et al. (Ch.  19:507–8; see also general interpretations later in this chapter); in their view, any connection between the selection of oak and prominent hearths or cooking pits becomes more likely if related to the entertaining of guests. The poor state of preservation clouds details of A10’s construction, making the interpretation of the building’s function challenging. Its most striking feature is its location on the plateau between the two monumental grave mounds, Kjellerhaug and Flaghaug. Both Flaghaug and Kjellerhaug seem to be multi-phased monuments first constructed in SP I, though certain younger phases, such as the 3rd century AD central grave from Flaghaug, are most prominent (Østmo and Bauer, Ch. 12:231–55; Stylegar and Reiersen, Ch. 22:574–614). Given the building’s location at the midpoint between the funerary monuments and its visibility from the strait, it can be reasonably inferred that the building was part of a grouping at Avaldsnes that presented a monumental façade towards the sea route through the Karmsund strait. The building’s spatial and temporal relationship to the wealthy Flaghaug grave raises the question of whether it functioned as a hall. Hall buildings are defined by Frands Herschend (1993:182–3; 1998:16) as buildings belonging to big farms that consist of a large, open room, occupy a distinct position on the farm, feature hearths that were used for neither cooking nor facilitating a handicraft, and contain artefacts distinct from those from the dwelling part of the main house on the farm. Based on these criteria, although possibly with special emphasis given to the size of the building, numerous halls have been identified in Norway during investigations in recent decades (Diinhoff 2010:84). Building A10 is unfortunately too poorly preserved to allow consideration of several of the criteria concerning construction details or artefact collection; the relevant aspects will be discussed in the following. Although length by itself in many instances has evidently been taken as the main criterion when identifying hall buildings, the layout of the hall as a building type did not remain constant throughout the Iron Age. The hall originated as a separate building extraneous to the traditional longhouse (Løken 2001:76–81; Herschend 1998:17). The free-standing halls of the 3rd and 4th centuries were small or medium-sized buildings within the settlement, developing into a larger dwelling house with several



7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 115

rooms in the early Migration Period (Herschend 2009:252–60). The hall was traditional in its structural similarity to ordinary dwelling houses, and at the same time novel in that its architecture reflected the needs of the owner, rather than the community. Familiar features took on new functions, for instance hearths used for heat and light rather than cooking (Herschend 1998:37). Amongst the halls, great structural variation is seen with respect to building size; whether a building features specific entrance rooms or entrance directly into the main room; whether it features one or several entrances; whether it features a high-seat placed in the side aisle, the corner, or the short-end; and the quantity of artefacts indicative of gender-specific zones (Herschend 1998). Herschend (1993:184–5; 1998:17) interprets the hall as a room suited not only for representation of economic and military leadership, but also for positioning the family as separate from the subsistence-level activities centered on the main longhouse. To the extent that the preserved remains of A10 permit conclusions on its size and layout, it does not diverge significantly from the contemporary embryonic halls at other Scandinavian sites, for example at Vallhagar in Gotland and Uppåkra, but this cannot be confirmed on the basis of the surviving archaeological features (Herschend 1993:fig. 183; Larsson and Lenntorp 2004). Carstens (2015) underlines the importance of place for the location of hall sites – they were typically built at an elevated position, where they could be spatially associated with large grave mounds or places of cultic worship, and usually show traces of cooking, presumably connected to feasting. Due to the lack of construction details in building A10, these latter aspects become most relevant for considering the building’s social status and function. A10 was clearly part of a larger settlement with a position allowing for control of the Karmsund strait. The settlement was characterised by monumental grave mounds, and the building’s location may have allowed it to contribute to Avaldsnes’ visual impact towards the strait, providing a public façade (Herschend 1998:39–42). Furthermore, numerous cooking pits spanning the same period of prehistory are indicative of social gatherings. The collection of finds from the building is scarce (S12768/1–28), consisting of half of a pair of tongs or pliers, four nails, a few small shards of pottery, and pieces of glass. Based on the quality of the glass, these fragments likely came from overlying remains dating to the period AD 1400–1600. The nails found in the hearth are difficult to assess in terms of date but could be an indication of building material being reused as fuel. Besides the pliers, there are no indications of crafting in the hearth, as metalwork would likely have left traces observable either during flotation or in the analyses of micromorphology or soil chemical samples. Thus, the pliers cannot be definitively associated with crafting in the building. Regarding Herschend’s (1993) criterion that the hearth in halls are not used for cooking or handicraft, but rather functioning solely as sources of light and heat, Carstens (2015:17) contends that such functional diversity is difficult to assess as hearths would have been subjected to at least occasional cleaning, but accepts that a find of multiple hearths within a single building could be taken as an indication that at least some of them were used solely for lighting

116 

 B: Excavation Results 2011–12

and heating. The macrofossil material from the hearths in A10 is too scarce to provide information on cooking activities (Ballantyne et al., Ch. 19:481). Amongst the few preserved burnt bone fragments is one from the lower front leg of a red deer stag (Ballantyne et al., 19:481). Bones from wild animals are found in both settlements and graves throughout the Iron Age, though domesticated animals dominate the bone assemblages (e.  g., Jennbert 2002:110–11; Mansrud 2006:142; Nilsson 2006:63). Some authors have argued that finds of bones from wild animals such as deer are indicative of sites related to cult activities, whereas others hold that the distribution of wild animal bones is not particularly higher at such specific or high-status sites (Ballantyne et al., Ch. 19:497–8 with references; Jennbert 2002:11). Though the small fragment of bone from Avaldsnes does not provide enough information to assess whether consumption of venison and the hunting of large animals is linked to high status in this region in the Iron Age, this association has been generally accepted due to a close connection between such activities and landownership (Andrén 1997:470; Oehrl 2013:508). Trapping undertaken by farmers can be differentiated from high-status hunting with horses, hounds, or birds of prey, which moreover served as a demonstration of power and wealth due to the parallels between this mode of hunting and Iron Age warfare (Dahlgren 2001; Ahrland 2013:442; Lie 2004:48). Whether such cultural notions were current in the Avaldsnes area during the Roman Iron Age is unknown. In Roman England, however, bones of deer are mainly found at high-status sites such as Roman villas and near the lodgings of high-ranking officers at military garrisons, indicating that red deer hunting was subject to regulation (Allen 2015:178). Though Roman influence can be seen in, for example, the high-status artefacts of the late 3rd-century Flaghaug grave, it is by no means certain that the concept of a noble hunt had reached Scandinavia at this point. While antler finds can be merely those shed naturally, bones in the proper context are an absolute indication of hunting. The fragment found in building A10 is likely waste from food or slaughter, possibly having seen secondary use as fuel (Ballantyne et al., Ch. 19:506). A single bone fragment does not constitute substantial grounds for deductions regarding the general diet and subsistence economy at Avaldsnes and its hinterland in the Roman Iron Age. However, the fact that bones from wild animals are rare at Iron Age settlements makes its presence here an interesting detail amongst other aspects of the building’s social milieu. In summary, the preservation of construction elements is poor and cannot by itself provide a clear interpretation of the building’s function. A central hearth may indicate a central or main room; the possible preference of oak logs for fuel in the central hearth sets it apart from ordinary cooking pits, while the fragment of red deer bone may possibly be related to high-status hunting. All other aspects, such as the absence of a byre or household activities, are known through merely negative evidence and cannot in themselves explain the function of the building. The strongest argument for interpreting A10 as part of a high-status settlement is the context of the monumental grave mounds (Østmo and Bauer, Ch. 12), the contemporary boat-



7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 117

houses (Bauer, Ch. 10:183), and the nearby cooking pits (Bauer, Ch. 13). The building’s location becomes meaningful when considering its visibility from the Karmsund strait. The hall would have risen up from the settlement ridge, monumentally positioned between two grave mounds – clearly a statement of the farm-holder’s status.

7.3 SP III – Building A13: Longhouse The longhouse (A13) in Area 5 was a three-aisled, trestle-framed construction with several repair events or adjustments, possibly representing different building phases. The building was probably in continuous use throughout most of SP III. The building remains were covered by a modern cultivation deposit (A107) and a deposit containing heated and cracked stones (A18505) dated to the 11th century on the basis of artefacts. Concentrations of such stones indicate the presence of a dwelling nearby, as they were probably used to heat water in affinity to the farmyard (Pilø 2005:136–7). The preserved remains from building A13 below these two deposits consisted of 37 postholes, two hearths, and three segments from two different wall ditches (Fig. 7.6). The building was at least 28 meters long, 8 meters wide, and oriented north to south. The distance between the northernmost and southernmost trestle postholes was just over 24 meters. The building’s original length cannot be determined due to substantial recent disturbances in the north and a gravel road covering a possible continuation to the south. A rocky outcrop about 12 meters to the south of the excavation area marks a likely maximum southern extent of the building. The northernmost building remains were faint and shallow due to construction works at the site in the 18th and 19th centuries. These activities appear to have completely removed any possible continuation northwards of the identified building remains. The even and slightly sloping ground extending over 100 meters north of the excavation area is well suited for building and does not delimit the possible maximum. By adding a reasonable distance to the gables from the preserved postholes, the building’s minimum length is estimated at 28 meters. In Iron Age longhouses, the distance between the trestles commonly varies within the building to accommodate function-specific needs. In order to create a capacious hearth room in the dwelling part of the building, the distance between the trestles is usually greater than it is in the byre or storage rooms (Løken 1991; 1998). The distance between the posts within the trestles in the hearth room in the middle of the building can also be greater, with the posts placed closer to the wall to open the room even further (Løken 1991:66–7; 1999). No such variations could be measured in A13, making it difficult to discern functional areas within the building or to calculate its original size based on the layout of construction remains. However, an indication of a differently constructed hearth room is given by the diameter of the post imprints in some of

118 

 B: Excavation Results 2011–12

the postholes closest to A13’s northern hearth, which were generally larger than those found elsewhere in the building (3, 6, 7 in Fig. 7.6; below). The position of the postholes and the wall ditches indicates that the building was approximately 8 meters wide. Based on the postholes’ distribution, the building was interpreted as a three-isled, trestle-framed construction, with a series of post pairs and tie beams supporting the roof – the common Iron Age farmhouse type (Nærøy and Børsheim 2005:186–8). At least 24 postholes for roof-bearing posts were distributed among a minimum of ten trestles, of which nine had both postholes preserved (all except from trestle 1 in Fig. 7.6). The features’ layout and stratigraphy suggested a history of several repair events or adjustments made to the building. Some of the building features could not have existed contemporaneously. Such mutually exclusive structural elements in practical terms were particularly visible around the northern hearth and by the entrances. Around hearth A46300, the distance between the trestles was too short for them all to be contemporary. Rather, in this area there seem to be four trestles from two different building phases (3 and 7, 6 and 8 in Fig.  7.6). Double postholes and postholes located directly adjacent to one another (1, 3, 10, 13, 15 in Fig. 7.6) were other signs of the building’s development or multiphase history. In theory, the composition of the postholes’ fill – the colour, texture, or contents such as charcoal or burnt clay – could be expected to help distinguish the construction phase, repair/modification phases, or deconstruction/tearing-down phases, but in this case no pattern could be discerned. Posts were probably replaced while the building was standing, possibly due to decay or in an effort to adapt different parts of the building to changing functions. The hearths in the longhouse were radiocarbon dated using charcoal, while the postholes were dated using material from either the infill of the features or the likely post imprints, in the cases where such existed. The dating results varied from 1952 BC to AD 543–600. The wide age spread of the dated features as well as inherent uncertainties related to dating secondary deposits prevent definite delineation of building phases. The wide range of dates is discussed below. The building’s height is difficult to estimate. However, post imprints for thirteen of the postholes were visible in section, either as darker soils or as void spaces between supporting stones (Fig. 7.7). The imprints were not detected in plan prior to excavation; therefore, no information could be gathered about the posts’ shape. The imprints’ diameters varied from 30 to 74 cm, and seven were greater than 57 cm. Compared to contemporary longhouses in western Norway, these dimensions are truly exceptional. For instance, House 8 at Gausel had roof-bearing posts with a diameter of 16–20  cm, while House 4/10 at the same site had posts 15–17  cm in diameter (Børsheim et al. 2002:113–15, 41). The Roman Iron Age/Migration Period hall at Eide had post imprints 26 and 28  cm in diameter Diinhoff 2009:13–15), and a Migration Period/Merovingian Period longhouse excavated at Sandane airport had posts probably 15–20  cm in diameter (Olsen et al. 2010:135). Four of the largest postholes (in trestle 3, 6, 7, 8 in Fig. 7.6) in the longhouse at Avaldsnes lay in the area around the northern hearth, indicating that the posts were more massive here than in other parts

 119

7 Østmo and Bauer: Prehistoric Settlement and Buildings 

A48787 Beta-333047

A48801

A46300

BC 1952–1880

Beta-332885

BC 1878–1770 Beta-453514

AD 128–215 A47199 Ua-45352

A46673 Beta-333048

BC 1871–1692

AD 410–532

A46796 Beta-333049

A48716

AD 349–429

1 A18246

A46858 A46996

2

A46617

A48688 Ua-45362

3

AD 444–595

4

5 6 7

A45557 Ua-54363

8 11

AD 434–544

9 10

12

13 14 15 A48640 Ua-45347

AD 543–600

A52562 A52540

Estimated minumum extent of building

Features in A13 Posthole Hearth Wall ditch Other feature Bedrock Modern ditch Excavated area

0

3m

Fig. 7.6: Building A13 with all building elements and radiocarbon dates. The numbering of trestles etc. (right) refers to the discussion in the text. Illustration: I. T. Bøckman, MCH.

120 

 B: Excavation Results 2011–12

30–39 cm 40–49 cm 50–59 cm 60+ cm Post imprint defined by: soil colour void space between supporting stones shape of posthole cut

Fig. 7.7: Diameter of post imprints in excavated postholes of building A13. Illustration: I. T. Bøckman, MCH.

of the building, suggesting that this part of the building was taller, possibly because it constituted the mid-point of the construction. Wall ditches would have been used for draining water dripping from the eaves or possibly would have contained the actual wall posts and sills. In the case of the longhouse at Avaldsnes, the latter is more likely; if the former were case, the distance between walls and roof-bearing posts would have been significantly smaller. Situated between the postholes for the trestles and the wall ditches were other postholes, interpreted as related to entrances (below) – due partly to their proximity to the post-



7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 121

holes from the trestle construction, but primarily to the two western postholes corresponding to an opening in the wall ditch (4 in Fig. 7.6). The building’s multiphase history is confirmed by postholes interrupting the entrances, of which there probably were two: one in the centre of the remains of the western wall trench (4 in Fig. 7.6) and one in the centre of the eastern wall’s southern half (12 in Fig.  7.6). Identification of the second was less convincing than the first, but the dating result (Ua-54363) of the posthole infill for one of the entrance-related features indicates its contextual relationship to the longhouse. On the other hand, the eastern wall ditch segment (11 in Fig.  7.6) interrupted the eastern entrance – a clear sign that the two features were disassociated. Eight small postholes in two rows and a wall ditch segment (A18246) extended from the western entrance at an angle. The respective orientations of the postholes and the wall ditch diverged slightly from one another, indicating the two features were not part of the same construction. The postholes suggest a passageway leading up to the entrance, perhaps as a fenced aisle or a planked walkway. The aisle was about four meters long and two meters wide; its termination against the building corresponded with an opening in the north–south-oriented wall ditch, suggesting contemporaneity. The east–west-oriented wall ditch might belong to another entrance phase or to a feature altogether unrelated to the longhouse. Similar angled entrances to Iron Age longhouses are not frequent. However, at Hove-Sørbo, Merovingian House 33, a possible passageway for leading cattle into the building is suggested, lending support to an interpretation of the building as a barn or byre (Bjørdal, pers. comm. 2014). A similar situation may have prevailed at Avaldsnes though there are no clear indications of a byre in this part of the building. At Gausel, there were angled entrances to House 8 and House 4/10, but these consisted of a stone pavement between the two buildings (Børsheim et al. 2002:104–7, 119). There, the angled entrance led into a large central room that may have functioned as either a byre or a hallroom, based on macrofossils or architectural details and large hearths respectively; the contradicting indications possibly relating to different building phases (Børsheim et al. 2002:239). A connection between the paved passageway and the hallrom is suggested, leading into into a particularly important social arena. Similary, reconstructions and change of functions in parts of A13 make it difficult to establish what kind of room the angled passageway led into, see below. However, the spatial relation to contemporary cooking pits (Bauer, Ch. 13) which likely relate to the entertaining of guests west of A13, make it more likely that the entrance was for people rather than cattle. The longhouse area contained two hearths (5 and 14 in Fig. 7.6), of which the northern was the larger. It was centrally positioned between the two rows of trestle postholes, roughly midway along the stipulated building’s long axis. A modern drainage ditch cut the hearth’s eastern part, truncating its original, probably roughly quadratic shape. A flue led out from the hearth’s south-western part. In section, two different extents of the hearth were visible, perhaps related to different building phases. Radiocarbon dating of charred alder from the hearth resulted in the 14C-date 1878–1770

122 

 B: Excavation Results 2011–12

BC (Beta-332885), whereas charred oak was dated to AD 128–215 (Beta-435314). The feature’s location firmly connects it to the SP III longhouse, and the early date of the alder is closely synchronous with radiocarbon dates from two neighbouring postholes in the building (A48787 to 1952–1880 BC, Beta-333047, and A47199 to 1871–1692 BC, Ua-45352; see Fig. 7.6). Diverging dates within a building present a recurring challenge and may have different explanations. Examples of the presence of old turf can be found regionally, for instance in Building 3 at Kjernevikveien in Stavanger, and super-regionally, for instance at Ringdal, Vestfold where turf with Mesolithic dates retrieved in Iron Age houses has been interpreted as either fuel or building material (Sandvik et al. 2012:4; Gjerpe and Østmo 2008:132–3). The presence of such turf is reflective of activity in the building during the Iron Age; however, this does not seem to be the case with the charcoal of Neolithic origin in building A13. Microscopic assessment of the charcoal has led Ballantyne et al. (Ch. 19:482) to conclude that it is most likely represents the product of redepositing of waste from earlier activity, rather than the residue of waterlogged fuel. A case with similarly diverging radiocarbon dates is seen in houses II and IV at Veien, Buskerud (Gustafson 2016:110–11). Both houses could be dated stratigraphically to the late Roman Iron Age and Migration Period, whereas the 14C-results provided dates from the Neolithic, Bronze, and pre-Roman Iron Age. Consequently, all macrofossils in the area apparently originated from activity pre-dating the two houses (Gustafson 2005g:50–2). Likewise, at Avaldsnes, the Neolithic dates seem to originate from charcoal related to earlier activities included and redeposited in younger features. In hearth A46300, the charred alder stands out both due to its 14C age and because the context is otherwise dominated by charred oak providing a date to the transition between SP II and III (above and Fig. 7.6). This is still c. 200–300 years older than other dated postholes, but explicable by an expected substantial age of the oak log fuel. Finds from the building (below) concur with the remaining radiocarbon dates, which seem to represent the building’s actual period of use in late SP III. The smaller hearth lay in the building area’s southern part, truncated by the same modern drainage ditch as the northern hearth. The southern hearth was oriented north to south and lay parallel with, but displaced one meter east from the building’s long axis. Considering its proximity to the eastern row of posts, it is doubtful that the hearth was contemporary with the second trestle from the south (13 in Fig. 7.6). The hearth was dated using roundwood from willow/poplar to AD 543–600 (Ua-45347). The radiocarbon dating results suggest that the building was in use for a long time; rather than distinguishing specific building phases, the broad span of the dating results indicates the entire period of the building’s use. The spatial layout of building remains is the strongest evidence for identifying different phases: the internal spatial relation among certain structural elements contradicts contemporaneity (4 and 6, 10 and 12 in Fig. 7.6). In the present case, the radiocarbon dating results could be expected to suggest that some parts of the building belong to an early phase and some to a later phase. Such an attempt at analysis should begin with those features where the dated



7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 123

material originated as closely as possible to the building’s use phase – that is, with post imprints and charcoal from the hearths. There are two postholes which are dated using material from the post imprint: A46796 and A48688 (Figs. 7.5–6). Together, the dating results from these postholes suggest at least two broad building phases in each end of this spectrum: the first from the first half of SP III and the second from the latter half of SP III. The posthole with the youngest date (A48668) overlaps in time with a dated posthole situated immediately to the east: A45557. The latter posthole is interpreted as part of an entrance, perhaps from a later building phase. The posthole dated to the early SP III (A46796) lies immediately east of the building, practically blocking the western entrance (12 in Fig. 7.6). Their likely non-contemporaneity suggests that this entrance is also from the later building phase. As mentioned above, the limited trestle distance in the area around the hearth indicates different phases. Possibly, postholes A46796 and A48716 could belong to a trestle (6 in Fig. 7.6) in an earlier phase, while postholes A46858 and A46673 could belong to a later trestle (7 in Fig. 7.6). The dating results from A46673 (albeit from the posthole’s backfill) support this hypothesis. Pottery found in the imprint fill in posthole A52562 indicates that it was not an imprint of the decomposed wooden post, but rather the backfill resulting from the post’s removal. It is possible, however, that this backfill is mixed with waste from activities within the house. Nevertheless, the artefact resulting from the imprint demonstrates the difficulty of dating postholes, whether by artefacts or radiocarbon dating (or both in combination), even when a post imprint is preserved. The southern hearth’s dating result partly overlaps with that of posthole A48688, suggesting the hearth’s relationship to the later building phase. The layout of postholes, wall ditches, hearths, and entrances did not allow a division into functionally different building parts, such as dwelling, byre, barn, and storage (e.  g., Herschend 2009:14). A ditch (A46617) running east to west across most of the building area was investigated as a possible division for a tri-partite longhouse, but given the stratigraphy of the area a connection to the building is unlikely (Macphail and Linderholm, Ch. 17:400). The aisle leading up to the western entrance suggests that this ditch could be part of a fenced passage for cattle, leading from the outfield through the infield into a byre part of the longhouse, but this does not accord with the entrance leading into a hearth room. If the aisled entrance and the hearth were contemporary, it could be that the entrance led into a dwelling room, but whether this room was located in the building’s middle section or closer to one of the ends is unknown, as the building’s length has not been established. The two postholes belonging to the western entrance which lay between the opening in the wall ditch and the trestle posts suggest the presence of a dormer projecting beyond the roof plane, with the door situated between the longhouse’s outer wall and the inner roof-bearing construction. A similar construction is possible for the eastern entrance. All excavated postholes and hearths were sampled for macrofossil analysis (Ballantyne et al., Ch. 19:482). During floatation, several artefacts were furthermore recovered (Fig. 7.8 and below). Soil chemistry samples were collected in a grid (Cannell,

124 

 B: Excavation Results 2011–12

Burnt clay

Burnt bone

Sintered clay

Tooth

Glass Pottery

Flint

Mica

Iron fragment

Slag

Spindle whorl

Fig. 7.8: Overview of artefacts and macrofossils from the building features in A13. Illustration: I. T. Bøckman, MCH.



7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 125

Ch.  18; Macphail and Linderholm, Ch.  17). Neither sampling scheme was able to establish functional differences in the longhouse. Specifically, no presence of a byre or stable in any part of the building could be found (Ballantyne et al., Ch.  19:482; Macphail and Linderholm, Ch.  17:400). There were elevated MS values in Area 5’s south-eastern part, but this might rather be caused by proximity to the later production activities in Area 6 (Østmo, Ch. 9), approximately 30 meters further south-east. Only a single posthole (A46764) contained a large amount of grains – insufficient evidence for suggesting that the posthole was inside a byre or storage part of the building. As was the case in the possible hall building (A10) from early SP III, oak was the primary fuel used in the two longhouse hearths. The dated material from the larger hearth, which produced the older radiocarbon date (Beta-332885), was alder (Alnus), and not representative for the fuel found in the hearth. As mentioned in relation to A10, oak regenerates slowly and is a valuable building material. The symbolically wasteful use of a valuable building material as fuel could be interpreted as conspicuous consumption (Veblen 1899). Alternatively, cut-offs from oak logs intended for building construction or building elements were reused as fuel after repairs, as is suggested for A10 (Ballantyne et al., Ch. 19:480–1, 507–8). However, no iron nails or rivets were found in the hearth in A13 that would have substantiated such an interpretation. Ballantyne et al. (Fig. 19.6) show that the distribution of oak logs indicates a structured selection of fuel, where oak logs were preferred in the large hearths inside the building as well as larger cooking pits located on the more visually prominent spots in the farmyard. A somewhat similar situation was observed at Gausel where oak dominated the hearths inside dwellings. Functional qualities such as oak being a heat-efficient fuel that produces less sparks, well suitable for certain craft processes as well as reducing the risk of house fires were suggested as possible explanations for this preference (Børsheim et al. 2002:235) The artefacts recovered from the excavated longhouse features (Fig. 7.8) provide no further clue as to the building’s function or to the dating of the building phases. The artefacts consisted of mainly burned and sintered clay, burned bones, slag, flint, and ceramics (S12780/1–70). There was also a heavily corroded iron nail (S12780/1), as well as two iron fragments (S12780/2), which possibly had been part of the same needle-like object. Some pieces of mica, apparently intentionally cut along the edges, came from the northern hearth. The mica was distributed throughout the hearth and might have been used as temper in pottery production. The various building features contained 19 sherds (S12780/5–15) from 16 ceramic vessels. With one Migration Period exception (S12780/11), none of the sherds could be dated with greater accuracy than to the early Iron Age. The Migration Period sherd was found in posthole A52540’s backfill in the south-western part of the building remains; the sherd was undecorated, soapstone-tempered, and possibly from an undecorated bucket-shaped pot (Kristoffersen and Hauken, Ch. 21:529). Only one of the sherds (S12780/14) bore traces of decoration, in the shape of two parallel angled lines. This sherd was recovered from the

126 

 B: Excavation Results 2011–12

post imprint of roof-bearing post A52562 in the southern half of the building remains. Another posthole five meters to the north contained a small soapstone spindle whorl (S12780/4), typologically datable only to the Iron Age. In a sample from posthole A48801 – the northernmost of the building remains – a piece of curved, white, semiopaque glass (S12780/3) was recovered. The glass could be from a prehistoric vessel, but the possibility cannot be dismissed that it is a more recent intrusion into this heavily disturbed part of Area 5. In a sample from another posthole, A46996, in the middle part of the building remains (8 in Fig. 7.6), five pieces of iron slag (S12780/17) were recovered from a sample. One posthole contained a polecat/ferret’s tooth and two other teeth, possibly from the same animal. Furs from such animals were used in high-status fur clothes (Ballantyne et al., Ch. 19:499), although it remains uncertain whether the presence of the teeth in one of the longhouse’s features implies manufacturing or other use of such a material. Apart from suggesting a combination of household and production activities, the combined artefacts contribute little to establishing the age and function of the longhouse.

7.4 SP V – Building elements A14: Possible building remains In the same location as the early SP III building (A10), there were postholes that may represent a SP V building (in the following referred to as A14). In an even more fragmented state than building A10, the remains of building A14 consisted only of nine postholes, none of these forming complete trestles in a roof-bearing construction. Six of the postholes were quite similar to each other, forming a row overlapping and cutting into the SP III wall ditch. Of the six cutting into the ditch, three were dated to SP V, while three remain undated. Another three features were dated to SP V: two stakeholes and a posthole lying directly west of the line of postholes within the ditch. The postholes within the ditch formed a line that could correspond to a wall construction, but lacking a relation to roof-carrying constructions it cannot be ruled out that these postholes are remains of a simpler construction, such as a fence. The radiocarbon dates of the postholes in the ditch correspond well with those of the postholes west of the ditch. The close dating results support the interpretation of a contextual relationship for the area between the postholes, though a suggested ground plan produced on such a basis remains highly tentative (Fig. 7.9). An additional possible posthole was observed at the surface of building A10’s wall ditch but could not be distinguished from the ditch’s fill during cross-sectioning. A sample taken from the section provided a date overlapping with the SP V postholes (TRa-4234) and may indicate the presence there of a posthole, despite homogenisation of the deposits filling the features rendering such a posthole indistinguishable. In addition, one of the samples from the large early SP III hearth (A8957) produced a later date of AD 1041–1154 (TRa-4240), though

7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 127

A19788 TRa-4234

AD 985–1020

A18736 Beta-319016

AD 1035–1152

A19799 Beta-319017

AD 988–1023 A18687 Beta-222064

A8957 TRa-4240

AD 901–1021

AD 1041–1154

A18677 TRa-4243

AD 904–1015

A19829 TRa-4241

AD 907–1017 A19839 TRa-4238

AD 903–1016

Roman Iron Age fireplace Roman Iron Age wall ditch Suggested plan of fragmentary building Features in A14 Posthole Posthole (uncertain) Other feature Cemetery wall Bedrock Modern ditch Levelled area Excavated area

0

1

3m

Fig. 7.9: Suggested ground plan for the possible building A14 with all related construction elements and radiocarbon dates. Illustration: I. T. Bøckman, MCH.

128 

 B: Excavation Results 2011–12

this dating shows that the sample likely represents a disturbance caused by activities related to SP V, rather than a relationship to the actual hearth. Three vague and shallow depressions in the bedrock were discovered while removing a deposit containing modern rubbish in search for a western wall of building A10. The depressions were approximately 5  cm deep, and their fill was mixed with modern rubbish. No undisturbed prehistoric fill was recognised, due either to its absence or to modern intrusions obscuring a transition between deposits. The depressions were recognised as possibly man-made only after they had been emptied. Their size, shape, and interspatial distances correspond well with the bottoms of the postholes cut into the ditch about 6 meters further east; this correspondence furnishes the only evidence to argue for including them in the interpretation of A14. If they functioned as postholes, they could represent a western wall; otherwise, all that can be said for A14 is that it was a construction with a row of posts, forming either a wall in a building or a fence. A posthole with a contemporary radiocarbon date (Beta-222064) could be contextually related to the row of post, but remains highly tentative as traces of roof-bearing trestles are otherwise lacking. The poor state of preservation of the building remains makes it difficult to ascertain whether they are part of a building or another type of construction, such as a fence. The combined SP V dates in postholes, stakeholes, and likely intrusion in the hearth (A8957) would indicate a rather concentrated period of activity in early SP V. Though the radiocarbon dates from A14 span the period AD 901–1021, the calibration curve indicates a slightly higher probability for a date to the second half of the 10th century (Fig. 7.9, Appendix II). As the dated fragments of birch were taken from the backfill in the postholes and no post prints were observed, it is unknown whether the dated samples reflect the time of construction or backfill from when the building was torn down. No further details of the construction can be deduced from the preserved remains. If the vague depressions in the west are included in the interpretation, the building would be of a width similar to A10; however, the length or any other details would still be unattainable. High phosphate readings across the area with building remains, especially in its western part, have been interpreted as a reflection of intensive occupation, but such data cannot be connected to a specific period of use (Macphail and Linderholm, Ch. 17:392–6). Although the placing of posts into the older wall ditch must have facilitated the erection of the construction from SP IV, it is highly questionable whether this co-location was intentional – primarily because it is unlikely that such a ditch would have been visible a few hundred years after the removal of A10. Still, given the location’s prominence overlooking the Karmsund strait, a large construction there would be visible to travellers by sea: an ideal site for demonstrating status.



7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 129

7.5 Organisation of the prehistoric settlement The excavations at Avaldsnes exposed features related to dwellings and other types of buildings, food production, processing of grain and other foodstuffs, funerary customs, and military aspects such as seafaring and fortification (overview in Østmo and Bauer, Ch. 6). A layout of the prehistoric organisation of the settlement can be suggested based on the spatial relationship of the remains. The settlement spanned a long time, but two main phases are distinguishable: one in SP I and one SP III-IV.

7.5.1 Settlement in SP I–II In SP I–II, up to the 2nd century AD, the main identified features and activities at the site were the earliest phases in the grave mounds Kjellerhaug and Flaghaug (Østmo and Bauer, Ch. 12:231–42; Stylegar and Reiersen, Ch. 22:563), building A11 in Area 2 east of Kongshaugen, and the initiation of agriculture with signs of land clearing by fire and a patchwork of smaller fields that were relocated after shorter periods of use and which in SP II gradually were developed into a larger continuous cultivated area. Sporadic postholes indicate possible buildings in other locations across the site, mainly underlying the cultivated soil in Areas 2, 3, and 5 (Fig. 7.10); however, these were too fragmentary and dispersed to form distinct buildings or other types of constructions. In Area 5, radiocarbon dates from the postholes span most of SP I, probably due partly to sporadic activity during this SP and partly to the inclusion of redeposited Neolithic charcoal waste in younger back-fill. The high-status Bronze Age grave monuments at Reheia indicate that the high-status centre at this time was not located at Avaldsnes, but a few kilometers further north. The Kjellerhaug grave mound and the earliest phase in the Flaghaug grave mound show that Avaldsnes was nevertheless a site suited for symbolic communication based on its visibility from the strait; as no artefacts have been recovered, the question of social status at this time is left unanswered. Building A11 from SP I appears to have fallen into disuse fairly quickly. In the time period between this building and building A10 from SP III, there are no known buildings, only traces of other activities such as cultivation and the use of cooking pits. Based on conditions elsewhere, hiatuses as well as reorganisations of the prehistoric settlement are to be expected (cf. Gjerpe 2010:15). However, the large size of the areas left unexcavated, together with the possibility that other building remains have been disturbed beyond recognition in the most central parts of Avaldsnes, prohibit a definite conclusion in this matter. As the agriculture from SP I seems to have been present in the form of patchwork fields relocated in the landscape, a similar movement could have characterised the dwellings and other buildings. Speculatively, settlement traces on Kongshaug could relate to this form of settlement. Latrine waste and other microstratigraphic indications of settlement in several areas, for example underlying the Flaghaug grave mound, suggests a more widespread character of set-

130 

 B: Excavation Results 2011–12

SP I–II

SEA LEVELS Current AD 200 (+2,4 m) BC 500 (+3,5 m) BC 2300 (+6 m)

0

50 m

SP III–V SEA LEVELS Current AD 1000 (+1,3 m) AD 550 (+1,95 m)

Building

Fortification

Boathouse

Grave

Estimated minimum extent of settlement

Estimated minimum extent of cultivation

Estimated extent of production area

Raised stones monument

Fig. 7.10: The organisation of the prehistoric settlement. The two maps depict accumulated evidence from SP I–II (upper) and SP III–V (lower). A patchwork of relocated fields and buildings is characteristic of SP I–II. SP III–V saw a farm structure with a large continuously cultivated area, a farmyard with dwellings, a possible hall, and structures related to crafts as well as processing of agricultural produce. Near the farmyard is a harbour with boathouses, as well as grave monuments in prominent locations. Though not all these elements are present throughout SP III-V, the general spatial structure of the farm seems to have been stable. Illustration: I. T. Bøckman, MCH.



7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 131

tlement in SP I and II. Though Stylegar and Reisersen (Ch. 22) argue convincingly for a early Bronze Age phase in Flaghaug, the extent of this phase compared to expansions related to the SP III burials remains uncertain. The time-frame for the underlying settlement-related deposits remains wide at either SP I or II. The remains of agricultural activities imply a more or less continuous use of the entire headland, though of varying local intensity. An intensive cultivation with manured, permanent field(s) gradually taking form during SP II does, however, support a shift in the land use at this time and corresponds with a development resulting in settlement of a different character in SP III onwards.

7.5.2 Settlement in SP III–V Within the context of concurrent activities, the buildings dated to SP III and V at Avaldsnes contribute to an understanding of the settlement structure (Fig. 7.10). As emphasised elsewhere in this volume, it is possible that additional buildings may have existed beyond the extent of the previous surveys and the current project’s excavation. Intensive cultivation east of the Kongshaug ridge from SP II onwards evolved into a continuous and permanent field on the flatter areas surrounding Kongshaug. The presence of long-lasting buildings dating to SP II–V in the field in Area 2 in the western part of the settlement plateau seems unlikely: firstly because the limited number of postholes discovered were insufficient evidence for identifying buildings, and secondly because the cultivation, at least in Area 2’s central part, seems to have been continuous based on observations in the micromorphology analyses from Profile 15653 (Bauer and Østmo, Ch.  8:145; Macphail and Linderholm, Ch.  17:401). Temporary constructions might have been placed there when fields were relocated; some of the features were substantial enough to have belonged to proper buildings, but most probably represented remains of temporary shelters or other small constructions, perhaps erected for seasonal work or temporary gatherings. Some of the undated features beneath the thick colluvium might also be part of constructions predating the permanent field and the settlement from SP III further east. This is likely also the case for the postholes at Kongshaug, though their terminus ante quem dates cannot distinguish between SP I or SP II. The field in the site’s western part remained in use over a long period, with possible interruptions during short periods of disuse, as observed in the microstratigraphy in the earliest part of the colluvial sequence in Profile 15653 (for a discussion of agricultural activity at Avaldsnes, see Bauer and Østmo, Ch. 8; Macphail and Linderholm, Ch. 17:400–3). A ditch (A18206) containing postholes from a wooden fence running roughly north to south through the middle of Area 5 marked the boundary between the settlement area and the field. The ditch’s back-fill was dated to the medieval period, but the spatial distribution of cultivation deposits and other feature types shows a transition in this area (as discussed below), indicating that this boundary

132 

 B: Excavation Results 2011–12

probably was older, possibly dating back to SP II. The field in Area 2 extended into the western part of Area 5, although micromorphological analyses showed marked differences in the fertilisation in these two parts of the field. While animal dung was the primary fertiliser in the middle of the excavated field in Area 2, the cultivation deposits immediately west of building A13 contained charcoal, latrine waste, and other kind of refuse, such as iron slag and glass fragments, indicating an infield close to the settlement (Macphail and Linderholm, Ch. 17:402). The deposit is interpreted as the eastern delineation of the cultivated colluvial formation otherwise observed and documented in Area 2. The deposit’s stratigraphic relationship to many of the cooking pits and other pit features found in a concentration immediately west of building A13 spanning 356 BC–AD 615 (Bauer, Ch. 13; Beta-319015, Ua-45358) supports the conclusion that this extent of the cultivation deposit was established during SP II and continued through SP III. The building remains were clearly separated from this food preparation area, indicating long-lasting functional differences between the two areas. The continuation of the aisled entrance to the longhouse passed through or just south of this cooking pit concentration. Such a layout featuring cooking pits near the entrance of Iron Age longhouses is known from other sites, for instance at Eide, Forsandmoen, and in Fedjedalen (Diinhoff 2005a:82–3; Dahl 2009:101; Bjørdal 2011:5). Unlike the cooking pits in other parts of the farm further away from the farmyard, several of the features in the area close to the longhouse had been fuelled with oak logs (Ballantyne et al., Ch.19:474). A mix of wood species is not uncommon in cooking pits, often thought to represent collection of available fuel from the surrounding woodland (Gustafson 2005g). A distribution pattern has been observed by Lars Erik Gjerpe (2009:142) in his analyses of charcoal from cooking pits and hearths in Vestfold in the period 200 BC to AD 600. He notes that a higher proportion of mixed assemblages are found in specialised cooking pit fields compared with hearths and cooking pits at Roman Iron Age and Migration Period settlements in the same region. In pits and hearths at settlement sites such as Gulli, Ringdal, Elgesem, and Rødbøl, there was a tendency toward a more frequent presence of pits containing fuel of a single taxon; oak and pine were over-represented (Gjerpe 2009:143). The make-up of the surrounding forest, collection strategy, and preferences are suggested to have contributed to the distribution pattern in Vestfold (Gjerpe 2009:143). In another study from Vestfold, the charcoal contained very few fragments of old trunks, indicating the presence of a managed forest that was cleared, pollarded, and subjected to several strategies to extract the woodland resources (Mikkelsen and Bartholin 2013:96–8). The Vestfold flora is obviously quite different from that of Norway’s western coast, but it is to be expected that the woodland resources were equally managed at Avaldsnes with respect to firewood and fodder collection, gathering and management of building material, or other farm-economic productions. Considering the coastal heathland formation (Prøsch-Danielsen and Simonsen 2000a:199–201) and the decreasing presence of oak in analysed pollen samples, Macphail and Linderholm (Ch. 17:418, Fig. 17.5) argue that oak used as a construction material (e.  g. for shipbuilding) was



7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 133

probably not local. Ballantyne et al. (Ch. 19:480), however, refer to the presence of small ancient woodlands with oak and pine at Kormt today and argue that while all wood types may possibly have been gathered at Avaldsnes, gathering large oak and pine logs would likely require greater time and resources compared to other fuels. Ballantyne et al. (Ch. 19:480–1) argue that the distribution of oak logs reflects a structured selection of fuel, possibly for practical reasons connected to the fuel’s heat-producing quality or for the specific smell or taste it contributes to the food prepared in hearths or pits. It should be noted that two of the ovens in Area 6 include a large proportion of oak, adding to the argument that this structured selection was at least partly functional. At the same time, the inclusion could also have held symbolic meaning – massive oak logs may constitute a form of conspicuous consumption when used as fuel in the hearths or cooking pits most closely related to the dwelling and entertaining of guests. Such an interpretation would bind together contextually the large area of cooking pits, the longhouse, and the possible hall, alluding to special occasions of feasting. Of relevance here is Lydia Carstens’s (2015:17) suggestion that if the hall was intended for a specific selection of guests, the rest of their retinue might have been entertained elsewhere, but likely in proximity to the hall. Whatever the reason for this selection of fuel, it seems planned and structured as the other cooking pits, most ovens, charcoal lenses and the like show a different fuel assemblage in greater correspondence with the vegetation in the close surroundings (Ballantyne et al., Ch. 19:480). The phosphate distribution across the site confirms the interpretation of Areas 1 and 5 as the main occupation locations, with values five times higher compared with the field in Area 2 (Macphail and Linderholm, Ch. 17). The high phosphate values correspond to the higher density of building remains in Areas 1 and 5 compared with the remaining excavated areas. This indicates that the activity was either of high intensity and/or of long duration. Topography was probably a deciding factor for the long-term use of this settlement area. The possible hall building from SP III in Area 1, on the eastern ridge of the settlement plateau, was visible from the strait, and could have been important for communicating power and possession of the area. The building lay between monumental grave mounds located at each end of the eastern ridge of the settlement plateau (Skre, SP III in Fig. 28.2), underlining the farm-holder’s high status. Contemporary with the possible hall and the longhouse, the southernmost tip of the settlement was a production area with ovens used for high-temperature processes (Østmo, Ch. 9:171–7; Macphail and Linderholm, Ch. 17:385, Fig. 17.4). There were also settlement traces in Areas 1 and 5 that were not part of the hall or longhouse, but rather belonged to pre- or post-dated constructions. Some of the postholes in Area 5 had SP I dates, but these are likely caused by inclusion of older material in younger features. Close to the settlement, the boathouses from SP III and the transition to SP IV were investigated (Fig. 7.10). Additional boathouse remains are possible in other areas

134 

 B: Excavation Results 2011–12

around Avaldsnes (Bauer, Ch. 10), but the proximity of the boathouses in question would have facilitated quick use of sea-faring vessels as needed. The two-phased SP III boathouse in Area 8 demonstrates functional continuity in this area, explicable by topography. The sea level in SP III prohibited practical use of Gloppe – then an island, now a peninsula – as a harbour; with the gradual land rise, this area became more available for such functions. While the postholes constituting A14 indicate continuity of the SP III-farmyard area into SP V, the general lack of buildings from SP IV has raised the question of a possible hiatus in the settlement at Avaldsnes. With the existing data not accounting for large unexcavated areas particularly around the St Óláfr’s Church and the present-day barn (Bauer and Østmo, Fig. 5.2), this question cannot be properly addressed. However, certain findings from around the site indicate settlement in SP IV as well, although the buildings themselves are not identified. Care must be exercised, as two dated features in building A13 extend into SP IV. A small boathouse east of the settlement plateau was dated to AD 582–637 (Ua-45327). Other indications include a few postholes, ditches, and pits in Area 5 (Ua-45358, Ua-45349, Ua-45374), and a rather large group of small postholes in Area 6 dated to SP IV and the early SP V. These latter postholes were likely part of small buildings or constructions related to storage or food processing, although no certain buildings were identified (Østmo, Ch. 9:169–71). Immediately east of the postholes in Area 6, a deposit containing latrine waste sealed ovens dated to SP III (Ua-45334, Ua-45336). As the deposit itself contained charcoal dated to AD 414–532 (Ua-45332), it likely formed through episodic dumping, probably during the transition between SP III and IV (Macphail and Linderholm, Ch. 17:406; Østmo, Ch. 9:181). Latrine waste indicates a nearby dwelling, likely located in Areas 1 or 5 or in unexcavated areas nearby. A surveyed posthole under the present-day barn is a possible indication of such a building (Hemdorff 1994). Immediately overlying the deposit containing latrine waste was a large stone construction (A20) extending along the eastern edge of the farmyard plateau, probably built as a fortification during early SP IV (Østmo, Ch. 11:215–18). The sporadic features and indirect indications of settlement from early SP IV are too numerous to dismiss this period as a hiatus in the settlement. Moreover, while the boundary in Area 5 towards the field in the west seems to have continued throughout SP IV, the central farmyard may have extended further than indicated by the buildings A10, A13, and A14. The St Óláfr’s Church and cemetery immediately to the north and the present-day barn immediately south of the identified buildings constitute locations that could potentially have complemented the suggested activity areas shown in Figure 7.10. In Area 5 there are also indications of settlement in SP V in the form of a deposit containing mostly fire-cracked stones. Such stone heaps occur frequently on the outskirts of settlements of the late Iron Age and early Middle Ages (Pilø 2005:136–7). The deposit could be related to the remains from the royal manor that lay in the same area in SP V (Bauer, Ch. 14).



7 Østmo and Bauer: Prehistoric Settlement and Buildings 

 135

7.6 Summary Avaldsnes is a site with a long and extensive history. Settlement traces are scattered over the entire site. The traces of the settlement in SP I and II are not conclusive; they are documented indirectly in the form of latrine waste under Flaghaug and probably directly in the form of postholes in survey trenches on Kongshaug and possibly in Area 3, none of which were excavated due to their expected pre-AD date. It is not clear how these traces relate chronologically to building A11 from SP I below thick cultivation deposits in Area 2. A11 appears to have fallen into disuse fairly quickly, and the next clearly identified building is A10 from early SP III. The concentration of SP III and V features on the large settlement plateau in Areas 1 and 5 suggests that the settlement organisation was fairly fixed from SP III onwards, with a large settlement plateau east of a large field for cultivation and pasture. The two identified buildings from SP III and the possible building remains from SP V comprised a possible hall, a longhouse, and a possible building of unknown function. There were no identified SP IV buildings, though indirect settlement indications and the production in Area 6 suggest continuation of the overall farmyard structure. Stratigraphy and radiocarbon dates show activities at Avaldsnes throughout SP I–VII, but the state of preservation varied greatly. Though the buildings are too poorly preserved to provide details by themselves of the social status of the farm to which they belonged, their relation to features such as grave monuments, boathouses, a possible palisade, cooking pits, and areas for food processing and storage indicate an overarching context of a manor encompassing numerous activities, several of which testify to the holders’ social, economic, and military power.

Egil Lindhart Bauer and Mari Arentz Østmo

8 Prehistoric Agriculture

This chapter presents and discusses aspects of husbandry and cultivation as well as the organisation of settlement and land use at Avaldsnes from the transition between the late Neolithic and the early Bronze Age and throughout the Iron Age (SP I–III). The material comprises various agricultural deposits analysed through excavation, archaeological profiles, macrofossils, pollen, soil chemistry, micromorphology, and radiocarbon dating. Geophysical surveys were also utilised. The analyses demonstrated that husbandry and/or cultivation took place over most of the investigated part of Avaldsnes; certain areas were exposed to more intensive use than others. Husbandry began after land clearance in SP I, in the transition between the Stone Age and the Bronze Age. Cultivation, probably in numerous small field patches, may also have begun then or a little later. Remains from a late Bronze Age construction have been identified in the area of early agriculture. Fields and buildings appear to have been relocated periodically. Traces of buildings and other features demonstrate that the fields were used for other activities when not being cultivated. Colluvial deposits indicate that cultivation was intensified in SP II in the western part of the excavated area. A similar development is seen further east, though development of this field appears to have occurred somewhat later. During SP II and SP III, the settlement’s organisation seems to have become fixed, with cultivation of a large and continuous main field, a main area of settlement, and an area for processing and production on the southeastern tip of the settlement plateau, which – extending into SP IV – contained evidence of activities such as threshing, corn-drying, salt production, and limited smithing as well as episodic grazing and cultivation. The main field lay adjacent to and west of the farmyard, and was at times clearly separated from the latter by a ditch, probably containing a wooden fence. This organisation of the settlement was probably maintained until the latter part of SP V. Although the agriculture at Avaldsnes might have been enough to sustain the household, depending on its size, it is possible that grain was brought to Avaldsnes from other farms as tribute or taxes, thus creating a surplus at the farmholder’s disposal. Phytolith and isotope analyses were conducted with the aim of addressing such a hypothesis; however, the results were inconclusive.

The two main aims of this chapter are to identify areas of cultivation and husbandry and to explore how the organisation of agriculture changed over time. Prior to the excavations, the project aim was to explore ecofacts or traces of activities that could illuminate the social status of the farm, such as surplus production or collection of commodities in terms of tax to accommodate a large, high-status household. The excavated material in relation to agricultural production proved to be of limited value in addressing these questions, although it is likely that the processing and storage of grain was reorganised and concentrated on the southern tip of the settlement plateau in SP IV (Østmo, Ch. 9:169, 178 and discussion in Ballantyne et al., Ch. 19:485, 497). Cultivation and husbandry at Avaldsnes occurred in parallel since SP I; however, the time of introduction of each is difficult to determine with precision. Most surveyed or excavated areas have been grazed and cultivated in the past, for a longer or shorter

138 

 B: Excavation Results 2011–12

duration and with varying intensity. The most prolonged cultivated area encompassed Areas 2, 3, 4, and the western part of Area 5 (Fig. 8.1), which together at times made up a continuous field of at least 2.9 hectares. The elevated ridge called Kongshaug (Fig. 8.1) interrupted the field, but this ridge seems likewise to have been cultivated at times, certainly in the historic era and possibly also in prehistory, although the extent of such cultivation is unknown (Hafsaas 2005:10–18). No secure date exists for the beginning of cultivation west of this ridge, but the cultivation deposits differed from the colluvial deposits east of the ridge mainly in terms of thickness; they are likely of similar origin and date. To the north lies the present-day cemetery, precluding further investigation of the extension of the prehistoric field there. The field’s southern and western extent at various times is also unknown, but traces of cultivation deposits combined with the even topography make it likely that the prehistoric field corresponded fairly well with the present-day cultivated area of approximately 6.7 hectares used mainly for pasture and fodder production (Fig. 8.1). The first estimate (2.9 hectares) is based on actual exposed cultivation deposits within the excavated area and offers a minimum extent for the prehistoric field, whereas the latter estimate (6.7 hectares) is the maximum extent based on topography. The cultivated area’s size is not constant, but rather variable over time – at least one expansion appears to have taken place during the first half of SP II. Several attempts have been made to estimate the size of the farmland of other Iron Age farms. The results vary largely from 5–15 hectares in southwestern Norway, 0.4–0.75 hectares in Öland, and 1–3.5 hectares in Gotland (Gjerpe 2010:14, with references). The remains of agricultural activities at Hørdalsåsen in Vestfold suggest a combination of a large, 1.5–2 hectare intensively cultivated field and an area of several smaller plots of 255–360 m2. Both these areas of cultivation were established during the pre-Roman Iron Age and went out of use in the early Roman Iron Age and the Migration Period respectively (Mjærum 2012b:117; 2012a:187). By comparison, the main field at Avaldsnes is rather large and was subject to long-term use. The example from Hørdalsåsen does imply that a larger main field could have been combined with other, smaller plots. Lars Erik Gjerpe (2010:14) suggests that an intensively cultivated field of 0.3 hectares per resident would produce a surplus of grain. Accordingly, the main field at Avaldsnes could be expected to have provided grain for 10–22 people. In addition, other farming activities such as husbandry, as well as fisheries, would have contributed to the subsistence. Modern agricultural deposits covered the field, thus demonstrating a persistent tradition of using the area for cultivation and pasturage. The cultivation deposits in Area 2 were 130 cm thick at the deepest. The excavations exposed several horizons, demonstrating different phases of cultivation activity. Radiocarbon dating of the bottommost deposits, which contained charcoal from land-clearing by fire, showed that land use, in the form of either cultivation or pasture, was established during SP I (BC 1608–1501, Beta-304878). The investigations suggest that in the early phases, cultivation occurred in small patches that were relocated in unknown intervals. Regardless

8 Bauer and Østmo: Prehistoric Agriculture 

 139

Exposed prehistoric agricultural deposit

Trench from 2001 survey

Clearing cairn

Excavated area

20646

6

01

haug

15653

13

5

6

35285

Kong s

2

1

4

0

100 m

3

Present-day field Assumed extent of prehistoric field

Fig. 8.1: Upper: Overview of trenches with exposed prehistoric cultivation deposits and features. Sections referred to in the text are indicated. Lower: Present-day cultivated area and assumed minimum extent of SP II–V cultivation. Illustration: I. T. Bøckman, MCH.

140 

 B: Excavation Results 2011–12

of how often the cultivated patches were moved or of the duration of periods in which the area was not cultivated, roughly the same area would have been used for field and/or pasture on and off for over 3.5 millennia. The general understanding of agricultural practices in Norwegian prehistory indicates that the organisation of the landscape with areas used extensively or intensively for producing crops, fodder, and pasturage can be traced to as far back as the late Bronze Age. A general intensification of the cultivation of grain, partly due to a transition to larger, continuous fields, and of manuring took place in the Roman Iron Age and Migration Period (Myhre 2002:137–41). Relative to the general picture, the continuous field at Avaldsnes was probably established as a ‘main’ field somewhat earlier, already in the first half of SP II, although it may also have been extended during the second half of SP II. Cultivation deposits outside the main field were shallower, indicating shorter duration, lower intensity, and lower soil enrichment. Depending on the terrain, these three factors will have reduced colluvial formation across the farm.

8.1 Husbandry Traces of husbandry occurred in several areas across the farm, probably tracing back to the earliest agricultural activity and contributing to the development of an open landscape. In several locations in southern Norway, husbandry might slightly pre-date cultivation; after cultivation began, the activities were carried out in parallel, with manure from livestock used for fertilisation (Myhre 2002:50–7; Olsen et al. 2010:21–3; Mjærum 2012b; T.B. Olsen 2013:160–1). Such may be the case at Avaldsnes. Pollen from certain cereals and weeds indicates cultivation, while pollen from other plants indicates animal management. Pollen from Calluna and Plantago lanceolata is found throughout the pollen sequence in Profile 13016 (Figs. 8.1–2; Fig. 17.5), suggesting grazing and manuring in the area (Hjelle et al. 2006; Macphail and Linderholm, Ch. 17:387). The increased occurrence of these pollen types apparently corresponds with the deforestation in the sequence’s lower part, suggesting land-clearing to facilitate pasture and subsequent cultivation. Grazing has been suggested in the area of the SP III boathouse (A40) and the sites of the Kjellerhaug and Flaghaug grave mounds due to traces of animal trampling and dung found in the course of micromorphological analyses (Macphail and Linderholm, Ch. 17:404). Such analyses have suggested that the construction of boathouse A40 (Bauer, Ch. 10) employed grazed turf and was situated on a surface that had been grazed. Similarly, turf gathered from locally grazed pastures was employed in the construction of both the Flaghaug and the Kjellerhaug grave mounds (Macphail and Linderholm, Ch. 17:415–16). The primary burial in Flaghaug is problaby of a Bronze Age date, while the secondary dates from early SP III (Stylegar and Reiersen, Ch. 22:574– 614), while the first phase of Kjellerhaug must have been constructed between 1733–



8 Bauer and Østmo: Prehistoric Agriculture 

 141

1632 BC (Beta-333050) and 794–568 BC (Beta-333052); that is, during SP I. The area must have been grazed prior to this. Indications of more intensive animal grazing were found in the preserved deposits beneath the Kjellerhaug; animal trampling as well as dung inputs were identified. The area had likely been cleared of stones before a trampled deposit filled voids left by the removed stones. The deposit was dated to 2118–1961 BC (Beta-333051), raising the possibility that the clearing of this land for pasture dates back to the transition between the late Neolithic and the early Bronze Age. During the excavation prior to the construction of Nordvegen history centre to the east of the Kjellerhaug grave mound, a deposit containing charcoal was dated to SP I: 1261–1125 BC (Beta-145267). This could represent land-clearing for pasture, but a possible clearing cairn could also imply that the surface had been cultivated (Sjurseike 2001:8–10). The presence of animals led to erosion in the activity area south of the Kjellerhaug grave mound. Periodic use of the area for grazing allowed grass to regrow and form topsoil, which in turn was trampled again. Intensified activity, with increased accumulation of trampled soil but less evidence of topsoil formation, indicated a more long-lasting stockyard in places (Macphail and Linderholm, Ch. 17:405). Arable colluvia immediately overlay this trampled old surface, and a grain of barley from the upper part of the deposits is dated to the first half of SP II (Ua-45354; Østmo, Ch. 9:180). The formation of these deposits and their likely reworking through trampling or possibly episodic cultivation in both prehistoric and modern times resulted in a complicated stratigraphy with features from both SP III and IV found at different levels within the deposits. The disturbed stratigraphy reflects the variable impact of later activities and long-term formation processes. However, the dated features – as well as the mentioned grain of barley – provide a terminus post quem date to early SP II as well as a series of fixed temporal points at which the formation of the deposits began. The deposits’ stratigraphic relation to stone construction A20, as well as other features, indicate that the deposits had been formed by early SP III (Østmo, Chs. 9, 11). Still, the colluvia were likely continuously formed and reworked over periods of grazing, stocking, possibly cultivation, and general human trampling related to farmyard or production processes throughout SP III. To the west of stone construction A20, such reworking may have continued into SP V as some of the late SP IV intercutting features were also partly covered by colluvium or trampled (Østmo, Ch. 9) while the overlying stone construction A20 from SP IV concealed and prevented such reworking and disturbing of features in the eastern part of the plateau (Østmo, Ch. 11). Neither excavations nor analyses of soil chemistry or macrofossil samples could conclusively locate traces of stables at Avaldsnes. While dung fragments identified in the wall ditch on the eastern side of building A10 together with soil chemical traces found in the southern part of longhouse A13 are suggestive of animal activities, these finds merely indicate the presence at some point of animals or their manure, rather than permanent stabling in connection with the farmyard.

142 

 B: Excavation Results 2011–12

8.2 Cultivation Prehistoric cultivation deposits covered large parts of the excavated areas, with the thick deposits in Areas 2, 3, 4, and 5 constituting a main field (Fig. 8.1). The deposits accumulated primarily through colluvial activity from higher-lying areas – a process enhanced by soil-tilling. Cultivation continued as the colluvial sediments accumulated in the lower-lying areas. Fertilisation with animal dung and midden material from the settlement increased soil accumulation. Types of activities identified through micromorphological analyses of thin sections include arding or ploughing, colluvial build-up, and fertilisation. These analyses furthermore showed periodic stabilisation of the deposits, with subsequent revegetation, again covered by new colluvium. Pollen evidence reinforced the hypothesis of such ongoing revegetation of the cultivated fields. Additionally, the absence of certain plant pollen types indicated that the fields were not permanent (Macphail and Linderholm, Ch. 17:387–8). Rather, the fields were probably relocated, allowing grass to regrow and hence become usable for pasture. A pollen sequence was extracted from Profile 13016 (Figs. 8.2, 17.5) in the northwestern part of Area 2 (Fig. 8.1). A chronology for the pollen sequence was suggested by the radiocarbon dating of extracted humic material from a deposit corresponding to layer 3 (A9601), giving a date to SP I (1257–1129 BC, Beta-347958). The result must be treated with caution, however, as the dated material provides an average age for the organic matter collected. A nearby SP II cooking pit radiocarbon dated to 356–201 BC (TRa-4227) is cut into the overlying layer 2 (A11837). The other cooking pits in this area were cut into colluvial deposit A102 overlying all of the layers documented in Profile 13016. A selection of these was radiocarbon dated, all to the first half of SP II. The analysed pollen sequence thus seems to reflect the earlier phases of agriculture in SP I, extending into SP II, c. 1500–200 BC. The pollen diagram shows a transition in layers 5–3 from forested to cleared ground with a marked increase of Poaceae (the grass family), which encompasses both edible cereals and plants suitable for animal-grazing, as well as the presence of Plantago lanceolata (ribwort plantain), indicating intensive grazing, and Hordeum (barley), demonstrating that the area was used for both cultivation and pasturage. Certain weeds in layers 3–2 indicate that the fields at this time were fertilised with dung, and that the fields were not permanent. Layer 3 is probably a response to colluviation caused by a period of intense local cultivation in the latter part of SP I – that is, the late Bronze Age (Macphail and Linderholm, Ch. 17). The possibility of periodic change in the use of the land, with a decrease in agricultural activities, is indicated by a slight increase in tree pollen contemporary with a decrease in grass pollen in layer 2 of the pollen sequence as seen in Figure 17.5 in Macphail and Linderholm (Ch. 17). As this decrease is stratigraphically younger than the dated layer 3 (Beta-347958) and older than several cooking pits dated to the first half of SP II (TRa-4224, TRa-4225, Beta-319014, and TRa-4227), the change in land use can be assumed to have taken

 143

8 Bauer and Østmo: Prehistoric Agriculture 

Cultivation deposits A101–103 removed before establishment of section

1 2 3 4 5 LAYER 3 Beta-347958

BC 1257–1129

1

Cultivation deposit

2

Cultivation deposit

3

Cultivation deposit

4

Peat layer

5

Peat layer

0

1m

Subsoil Stone Pollen sample Fig. 8.2: The truncated Profile 13016, seen towards the north-west. Illustration: I. T. Bøckman, MCH.

place towards the end of SP I. Barley is recorded from layer 5 and upwards, whereas wheat is present from layer 3; both grains were therefore cultivated by the latter part of SP I. There are no indications of the duration between layers 5 and 3, so the initiation of cultivation of barley can only be given a terminus ante quem date of 1257–1129 BC (Beta-347958). Ard marks located in the area, cutting into layer 2, contained soil equal to layer 1 in Profile 13016. Consequently, the soil-tilling represented by the ard marks occurred later than the SP I cultivation represented by layer 3 and prior to the formation of the above-lying colluvial deposits (layers 1–3 in Fig. 8.3) and the above-mentioned cooking pits dated to the first half of SP II. The pollen sequence seems to reflect roughly the same time period as is represented by layers 4 and 5 in Profile 15653. Profile 15653 (Figs. 8.1 and 8.3) provides an overview of the sequence of deposits in one of the thickest parts of the main field.

144 

 B: Excavation Results 2011–12

1 2

6

4

3

5

POSTHOLE A10500

0

1m

Modern topsoil Prehistoric cultivation

1: TRa-4230

BC 156–45

1

Initial cultivation

2: TRa-4231

2

BC 795–595

Posthole A10500 Topsoil/modern agricultural layer

Stone

1

Cultivation deposit

Dislodged stone

2

Cultivation deposit

SAMPLES

3

Cultivation deposit

4

Cultivation deposit

5

Initial land clearing and cultivation deposit

Soil chemistry

6

Charcoal lens

Macrofossil

Subsoil

Pollen

Micromorphology

Fig. 8.3: Profile 15653, cutting north to south through the prehistoric field in Area 2, seen towards the west. Illustration: I. T. Bøckman, MCH.

The earliest phase visible in the profile consisted of two deposits containing charcoal (layers 4 and 5 in Fig. 8.3). The bottommost deposit (A4216) probably represents the initial agricultural activities, encompassing pastures, fodder gathering, and possibly crop cultivation. The deposit was mixed with subsoil – probably from soil tilling – and charcoal from land-clearing by fire. The latter traces are common in western Norway, visible as charcoal-containing horizons in the bottom of prehistoric fields (e.  g., A.B. Olsen 2013a:136, 40; Slinning 2013:194). The bottommost deposit in Profile 15653 contained the most charcoal of the two charcoal-containing deposits, and was dated to SP I, specifically 1608–1501 BC (Beta-304878), slightly earlier than layer 3 in Profile 13016. It is reasonable to assume that the initial agriculture at Avaldsnes was established by this time but it is not clear whether it consisted primarily of husbandry or represented a combination of husbandry and crop cultivation. The immediately overlying deposit (A5882) constituted a gradual transition between A4216 below and the seemingly homogeneous colluvium above. This transitional agricultural deposit also belongs to SP I and was dated to 795–595 BC (TRa-4231). Building A11, a possible outbuilding (Østmo and Bauer, Ch. 7:106), was situated stratigraphically between deposit A4216 and A5882, thus also belonging to SP I. While the building remains were too scarce to firmly identify the building’s function, the lack of dwelling-type



8 Bauer and Østmo: Prehistoric Agriculture 

 145

features allows the assumption that the building was related to agricultural activities, for instance as a byre or a storage building. As a mix of deposits, it is possible that the sampled material from deposit A5882 contains charcoal from the older deposit below. Such mixing of material is known to have been the cause for conspicuously old dating results of cultivation deposits in other cases (e.  g., A.B. Olsen 2013a:160). In Area 2, the earliest dating result from a cultivation deposit less likely to be contaminated by older material is from the first half of SP II (156–45 BC, TRa-4230). This deposit (A103) is the bottommost of the colluvial deposits in the central part of Area 2 (Fig. 8.3). However, a similar reservation concerning the deposit’s dating applies here. Cultivated soils can contain charcoal and other material from different burning, cultivation, or fallow periods, mixed over the course of centuries by soil-tilling and animal trampling. One single dating result from a charcoal fragment merely indicates when some of the soil was cultivated – it does not establish when cultivation began (Slinning 2013:201). The youngest feature underlying the colluvial deposits in Profile 15653 was a late SP I posthole dated to 511–404 BC (TRa-4220), giving a terminus post quem date that corresponds well with the dating of A103 to the first half of SP II. The bottom of the colluvium must therefore have been formed shortly before AD 1. It is not unlikely that this development is related to the removed clearing cairn indicative of an expansion of the field in the first half of SP II, observed further to the west (Fig. 8.1 and details below), resulting in a continuous field from mid-SP II. Colluvial cultivation deposits with an almost complete lack of stones made up the middle and thickest phase of Profile 15653. Although thick cultivation deposits indicate a long duration of soil-tilling, the rate of accumulation can be greatly affected by such factors as slopes causing erosion or by soil-addition, for example from frequent fertilisation with manure and household refuse. Consequently, it is problematic to extrapolate duration from thickness. Three horizons of colluvial deposits were identified, but with barely any difference in colour or texture to distinguish them. On a micromorphological level, however, there were differences between the deposits, most notably the already-mentioned stabilisation episodes with grass growth. At the time of these growth episodes, the field could have been used periodically for grazing, which would provide a natural fertilisation of the soil. Although most of the field in Area 2 had very thick colluvial deposits, certain parts in Areas 3, 4, and 5 were much shallower. The subsoil’s depth varied between 20 and 130 cm below the present-day surface. A GPR survey (VIAS 2009; Stamnes and Bauer, Ch. 16:328–9) detected two depressions in the underlying bedrock in the field’s northern part (Fig. 8.4). This is where the deepest deposits were uncovered (Fig. 8.3). These turfy deposits accumulated prior to land-clearing, at a time when the two depressions were wetter, almost bog-like. Erosion that followed from cultivation caused soil to be transported from the higher-lying land surrounding the depressions, thereby filling the latter. Other parts of the field had shallower deposits, particularly in the east and south where the subsoil

146 

 B: Excavation Results 2011–12

0

GPR 50–55 cm

Excavated area 1992–1993

Relative reflection strength

Excavated area 2011–2012

High Low

Fig. 8.4: GPR time slice (depth: 50–55 cm) showing two large depressions in the bedrock in Area 2 (lower centre and left). The homogeneous cultivation deposits that fill the depressions show up as light areas, while more shallow bedrock and certain stone-rich features appear as dark anomalies. The western edge of the eastern 1992 survey trench (Hemdorff 1993) is visible. Illustration: I. T. Bøckman, MCH.

20 m

 147

8 Bauer and Østmo: Prehistoric Agriculture 

1

2

7

6

5

3 4

8 0

Turf/modern agricultural deposit

Modern drainage ditch

1

Cultivation deposit

Stone

2

Cultivation deposit

3m

SAMPLES

3

Cultivation deposit

4

Cultivation deposit

5

Cultivation deposit

6

Cultivation deposit

7

Peat layer

Macrofossil

8

Leach layer

Pollen

Micromorphology Soil chemistry

Fig. 8.5: Profile 35285, cutting north to south through the northern part of Area 3, seen towards the west. Illustration: I. T. Bøckman, MCH.

was exposed just below the topsoil. It is likely that the earliest prehistoric cultivation deposits in these shallower areas were ploughed away during later soil-tilling – a process intensified by erosion. Thus, the overview of exposed prehistoric cultivation deposits in Figure 8.1 shows the locations where prehistoric cultivation deposits were preserved, but the actual history of the cultivated area remains conjectural. Profile 35285 (Fig.  8.5), to the south, showed a matrix similar to that of Profile 15653, with the same kind of turfy deposits below homogeneous colluvia. A larger number of deposits were distinguishable in this profile than in Profile 15653 to the north. The reason for this might be that soil-tilling had been less intense in this area, resulting in fewer disturbances to the deposits. In the southern part of Area 3, the cultivation deposits were much shallower (left part of Profile 35285, Fig. 8.5), probably a consequence of the subsoil’s levelness, with no significant depressions similar to those in the northern part of Area 3 and in Area 2. A sequence of deposits with traces of initial land-clearing were also uncovered in the southern part of Area 3, but these traces were preserved only in minor, sporadically occurring depressions. The cultivation deposits exposed in Area 4, south-west of Kongshaug, were similar to the colluvial deposits in Area 2. However, no separate phases were visible and the deposits were not as thick as those in Area 2. As in Area 3, the possible reason for the thin deposits was the lack of depressions in which the colluvium could accumulate, perhaps in combination with increased soil erosion due to the slope to the south-west. There is no reason why the area on the western side of Kongshaug could not be part of the same permanent field from Area 2. However, the magnetic susceptibility survey showed marked differences between Areas 2, 3, and 4 (Fig. 16.8), suggesting that the three areas were separate fields. Nevertheless, apart from the mentioned varying depths, no visible changes appeared in deposits or features during excavation, which could explain the differing magnetic susceptibility values. Since the differing values

148 

 B: Excavation Results 2011–12

1

2

3 5

4 6

7

0

Turf/modern agricultural deposit

Stone

1

Ploughed-out stone-paved walkway

Fire-cracked stone

2

Post-medieval cultivation deposit

3

Cultural deposit

4

Cultivation deposit

5

Pit

6

Cultivation deposit

7

Cooking pit Subsoil

1m

SAMPLES Micromorphology Soil chemistry Macrofossil

Fig. 8.6: Profile 20646, cutting east to west through the western part of Area 5, seen towards the south. Illustration: I. T. Bøckman, MCH.

corresponded to the areas separated by the present-day road, the presumed cause for the variation was differing activities after construction of this road, consequently related to recent agriculture. Profile 20646 (Fig. 8.6) in Area 5’s western part was located in the eastern part of the main field. Neither turfy deposits nor any clear traces of land-clearing were found in the bottom of the matrix, but thick, homogeneous cultivation deposits were visible, as in Area 2 and the northern part of Area 3. The whole of Area 5 was cultivated in modern times, as documented in photographs from the early 20th century (e.  g., Alsvik and Alsvik 2001:159). This recent cultivation extended further east than that of SP II–IV. Micromorphological analyses from Profile 20646 showed that cultivation in this area had caused less truncation of the subsoil than that documented in Profile 15653 (Fig. 8.3). This suggests shovel-tilling rather than arding in the earliest phase of cultivation in the eastern part of the field. The same analyses showed that household refuse was used for fertilising from layers 4 and above in the prehistoric cultivation matrix. This refuse consisted of charcoal, latrine waste, and other kinds of refuse such as iron slag and glass fragments, and was different from the mainly animal-dung fertilising visible in Profile 15653 (Macphail and Linderholm, Ch. 17:383, 402). The proximity to the farmyard, along with the probable disposal of household refuse in the field, can explain the difference of fertilising practices in Area 5 compared to the field further west. The cultivation deposit containing household refuse was not dated, but lay stratigraphically above the cooking pits in the area, the youngest of which was



8 Bauer and Østmo: Prehistoric Agriculture 

 149

dated to the transition between SP III and SP IV (Ua-45358; Bauer, Ch. 13:260), indicating that the household refuse originated after this time. The oldest cultivation deposit (layer 6 in Fig. 8.7) was of limited eastern extent. This eastern delimitation of the cultivation deposit suggests that a fairly established boundary separated the farmyard from the cultivated area through one and a half millennia of cultivation, from the first half of SP II until the high medieval complex was established during SP V (Bauer, Ch. 14). The cultivation deposit pre-dated most of the cooking pits in Area 5 (Figs. 8.6–7). The earliest of the pits, dated to 356–201 BC (Beta-319015), had no stratigraphic relationship to the deposit; the oldest pit with direct relation to the cultivation deposit assigns the deposit a terminus ante quem date to late SP III (AD 433–534, Ua-45359). It is impossible to discern a precise date for the establishment of either cultivation or the boundary. Clues to the established boundary, in addition to the cultivated area’s delimitation, were a ditch (A18206) running north to south just west of longhouse A13 (Østmo and Bauer, Ch. 7:131) and a number of postholes and stakeholes within the ditch and running parallel to it. Part of the ditch cut into bedrock, probably to allow for posts or stakes in a wooden fence to be driven sufficiently far down into the ground. Seven metres south of the location where the ditch cuts into bedrock, a short section of what is presumably the same ditch has been preserved. The boundary’s continuation further to the south and to the north was uncertain, however. Charcoal from the ditch’s fill was dated to SP V/SP VI (AD 1299–1394, Ua-45344), and one of the postholes close to the ditch was dated to SP IV/SP V (AD 990–1029, Ua-45348). Despite scarce traces of the physical boundary, the functional boundary was probably stable for a long time. The excavated features’ contextual relationship with the prehistoric cultivation – demarcating its eastern boundary – leaves open the possibility that the boundary existed already in SP III.

8.3 Other cultivation features A number of prehistoric features were exposed in the spaces between the main field’s cultivation deposits. Such features included cooking pits, settlement traces, and traces directly or indirectly related to cultivation (Fig. 8.8). Cooking pits and settlement traces are discussed in detail in this volume (Bauer, Ch. 13; Østmo and Bauer, Ch.  7, respectively). The cultivation features consisted of ard marks, imprints of removed stones, and a partly removed clearing cairn. There were no visible lynchets, neither positive nor negative, despite the thick cultivation deposits and colluvial activity. Ard marks were exposed in two concentrations that were visible against the burnt deposits in the matrix’s bottom parts. The homogeneity of the overlying deposits rendered such narrow and shallow features invisible during excavation, but, as men-

150 

 B: Excavation Results 2011–12

0

10 m

Cooking pit Cultivation deposit Ditch A18206 Posthole Excavated area Fig. 8.7: Ditch A18206 and postholes in Area 5, creating a boundary between the farmyard and the cultivated area with cooking pits. Illustration: I. T. Bøckman, MCH.

8 Bauer and Østmo: Prehistoric Agriculture 

0

10

 151

20 m

Cooking pit Agricultural feature Settlement feature Excavated area

Fig. 8.8: Overview of archaeological features exposed between and below cultivation deposits in Area 2. Note that the features lay at different stratigraphic levels and are not all contemporary. Their dating results span SP I–III: 1879 BC–AD 567. Illustration: I. T. Bøckman, MCH.

tioned, the micromorphology showed arding and/or ploughing in these later deposits as well. Traces of a removed clearing cairn (A5300) demonstrate an expansion of the field slightly east of the Kongshaug ridge. The backfill from a removed stone in the ground adjacent to the removed cairn was dated to 361–208 BC (TRa-4213), providing a terminus post quem dating for the subsequent cultivation. Apart from a few remaining stones, the cairn had been removed, leaving an irregular imprint that indicates the earlier field boundary. A similar feature was exposed during the excavation at Gausel in 1997–2000, tentatively dated to the early Iron Age (Børsheim et al. 2002:233–4). The feature at Avaldsnes was about the same size and of similarly irregular shape as the

152 

 B: Excavation Results 2011–12

one exposed at Gausel. The terminus post quem dating of the backfill in the imprint of the removed stone, along with the stratigraphic relationship to the overlying cultivation deposits A103/A102, indicate that the cairn’s removal likely occurred during the first part of SP II. Imprints of individual removed stones were also discovered in other parts of the field at Avaldsnes. Most of these features had the tell-tale oval shape created when the stones were pulled up sideways out of the ground. The features between the cultivation deposits not directly related to cultivation demonstrate the occurrence of other, temporary activities in the area. These features support the hypothesis that the fields were relocated or allowed to lie fallow for periods during which the area was utilised for other purposes. It is not possible to establish a detailed chronological sequence for the features as only a fraction of them have been dated. Furthermore, it is impossible to estimate how many shallow features have been removed due to prehistoric soil-tilling. Those preserved were the only features cut deep enough to remain undisturbed by later cultivation or other activity. Consequently, while the age gap between the features and the colluvia covering them can be great, the dated features still provide terminus post quem dates for the over­ lying horizons and terminus ante quem dates for the deposits into which they are cut, forming reference points in the chronology. One such fix point, a cooking pit dated to AD 437–567 (Beta-304877) shows that deposit A103 and other low parts of Profile 15653 are older than this, whereas A102 and other higher-lying deposits are younger. The profile thus appears to cover a time period spanning from the late Bronze Age (SP I) through the Iron Age (SP II–IV), and likely on through medieval and modern times (SP V–VI). Traces of at least two constructions pre-date the intensified cultivation represented by the homogeneous colluvial deposits. SP I building A11 (Østmo and Bauer, Ch. 7), mentioned above, truncated the bottommost surface with traces of land-clearing by fire and was thus probably from the early period of agriculture. The single trace from the other construction was a posthole (A10500) visible in the bottom of Profile 15653 (Fig. 8.3). The posthole was filled with soil from the burnt deposit and was likely dug shortly after the area was burned, with the dug-up soil filled in around the erected post. The posthole was dated to 1879–1767 BC (TRa-4226), making it the oldest dated feature from the earliest site period that testifies to human activity in this area. A building in this location does not preclude agriculture. It is possible that parts of the area were used for agriculture at this time, although concentrated in small patches of pasture or cultivation rather than occupying a continuous field. Late-Neolithic fields in western Norway appear to have been irregularly shaped, 500–1000 m2 in area, and adapted to the terrain (A.B. Olsen 2013a:139). With the exposed posthole as the only preserved construction element, there is no evidence that a building actually stood in this location. The posthole could be related to another type of construction altogether. However, the posthole’s size is indicative of a roof-bearing post or another type of large post. If a building indeed stood there, it could have been the type of two-aisled longhouse common to the early agricultural period in western Norway (A.B. Olsen



8 Bauer and Østmo: Prehistoric Agriculture 

 153

2013a:130). The other scattered postholes elsewhere in the field, for instance those in the north-eastern part of Area 2 or the southern part of Area 3, could be remains from buildings related to cultivation or animal grazing, for instance byres or stockades, or from entirely different, possibly temporary constructions. A large concentration of stakeholes was found around the early building remains in the northern part of the field (Fig. 8.8). These features were probably part of simple constructions, such as fences, livestock enclosures, or hay-drying racks. The large number of stakeholes, in conjunction with their proximity, suggests that the stakeholes are associated with more than a single or a few features, probably spanning a long time period. Fields and pastures were normally established near the settlement for easier management. If occasionally relocated, such a system could potentially remain sustainable for several generations within a limited area. Fields or pastures that occupy too large an area or support an excessive number of grazing animals would in time be expected to exhaust the areas, necessitating the opening of new fields or pastures and leading to new buildings and changes in the organisation of the settlement (A.B. Olsen 2013a:140–1). The features in the field at Avaldsnes could represent traces of one or more stages of such processes of change. Further, it is possible that the entire farm area was abandoned for periods of time, even for decades or more. The small increase in betula, visible in layer 3 in the pollen diagram (Macphail and Linderholm, Fig. 17.5), could perhaps indicate such a period of disuse.

8.4 Discussion: change in the agricultural organisation At the present stage of research, the introduction of agriculture in western Norway is estimated to have occurred over a 1000-year period between 2700 and 1700 BC (A.B. Olsen 2013a:129). Relying on the dating of the pollen sequence from Profile 13016 and the assumption that the dated burnt horizons in various parts of the investigated area represent land-clearing, agricultural activities at Avaldsnes seem to have been established during the latter part of this period – that is, in the transition from late Stone Age to early Bronze Age. It is not entirely clear whether early agriculture comprised only husbandry or when cultivation was introduced, as only a terminus ante quem dating to the late Bronze Age is provided for the cultivation of barley. At this early stage of SP I agriculture at Avaldsnes, the fields were regularly relocated, probably making rather extensive use of the landscape. Traces of agricultural activities were found across the modern farm in the areas of the Kjellerhaug grave mound, at the Nordvegen history centre, and in the bottom layers of the main field. The settlement traces exposed in the north-eastern part of Area 2 and the southernmost part of Area 3 substantiate the hypothesis of temporary buildings and a farmyard that shifted position around the areas together with the agricultural patches. Some of these tempo-

154 

 B: Excavation Results 2011–12

rary fields, however, could have been used intensively for shorter periods, as is seen in the fertilised cultivated colluvium in Area 2, dated to 1257–1129 BC (Beta-347958). Consequently, the various settlement traces might be related to the earliest patchwork cultivation associated with Bronze Age – or even Neolithic – agriculture. During the first half of SP II an intensification of cultivation seems to occur. This interpretation is based on the expansion of the cultivation in Area 2 and the accelerating accumulation of thick cultivation deposits indicating the area’s transition into a permanent field. This development towards a landscape of intensively cultivated manured fields near the farmyard, surrounded by areas or fields used extensively for pasture and fodder production, is well known from other sites (Myhre 2002:127–32). Relevant parallels are found at Forsand in Rogaland (Prøsch-Danielsen 2001), Hjelmeset in Sogn og Fjordane (T.B. Olsen 2013), and Hørdalsåsen in Vestfold (Mjærum 2012a; 2012b). It seems likely that the accumulation of thick colluvial deposits marks the transition from small cultivated patches to large continuous fields. In a large field, soil would be constantly redeposited throughout a continuous area and accumulate differently from the way it would in small patches. If this presupposition is correct, the continuous field was established in the first half of SP II, based on the dating of the bottommost of the colluvial cultivation deposits in Area 2, as well as cooking pits also dated to the first half of SP II that cut into the cultivation deposits. The early SP II dating of the back-fill of a removed stone indicates that the removal of a clearing cairn just east of Kongshaug and other stones in the field correspond with this establishment. For comparison, at Hørdalsåsen a large main field was established and intensively manured in the pre-Roman Iron Age (Mjærum 2012b:117), whereas the permanent field at Hjelmeset was established during the transition to the Roman Iron Age (T.B. Olsen 2013:160). By the latter half of SP II at Avaldsnes, the boundary between field and buildings in the farmyard to the east seems to have been made permanent. Closer to the harbour, cultivation deposits covered the early SP III boathouse, having developed after the boathouse fell out of use. Prior to the construction of the boathouse, the area was used for pasture. The youngest boathouse-related deposit was dated to late SP III (AD 539–600, Beta-324647). While the possibility of cultivation around the boathouse whilst in use cannot be dismissed, cultivation post-dating the boathouse is certain. The deposits’ substantial thickness demonstrates colluvial activity, confirmed by micromorphological analyses, which would have been accentuated by the slope (Bauer, Ch.  10:188). Similarly, cultivation deposits found in Area 6 consist of accumulated soil produced by fertilising as well as erosion from the Kjellerhaug grave mound. This led to the gradual covering and sealing of deposits and features on the southern tip of the settlement plateau immediately south of the mound. These arable culluvia were dated to the first half of SP II (182–64 BC, Ua-45354) and the first half of SP III (AD 243–335, Ua-45357). As mentioned, datings of deposits have relied on the basis of single seeds and thus should be treated with caution. The deposits south of the grave mound were thinner than those in the field further west, and a high frequency of traces of processing, such as threshing,



8 Bauer and Østmo: Prehistoric Agriculture 

 155

corn-drying, salt production, and metalworking, demonstrate that the area was used for other purposes both before and after the cultivation, and possibly during intervals between periods of cultivation. Consequently, the cultivation in this area should not be considered continuous, as in the field west of the farmyard. Rather, this area might have been employed as a limited patch of cultivation. This activity could have occurred earlier than the establishment of the permanent field in Area 2, but it can also be contemporaneous with the main field, though subjected to less intensive and discontinuous use. The area might also have been cultivated in later periods, as suggested by features showing effects of colluviation and soil-tilling (Macphail and Linderholm, Ch. 17:403). The agricultural deposits demonstrate that cultivation occurred in other parts of the farm, both before and after the establishment of the main field west and southwest of the farmyard. These deposits therefore represent not only the chronological development of the agricultural organisation at Avaldsnes, but also an agricultural organisation with different areas designated for different use.

Mari Arentz Østmo

9 The Production Area The plateau south of the Kjellerhaug grave mound has seen varied use from the very beginning of SP I all the way into modern times. With the notable addition of the construction of the Kjellerhaug grave mound in the middle of SP I, Area 6 in SP I and II was dominated by agricultural activities. A wider range of production was initiated in SP III and IV. The data includes tangible evidence in the form of refuse from iron smithing, whereas copper alloy work, glass bead production and possible pottery production are merely suggested by geochemical measurements or clustering of finds. The processes that took place in SP III are characterised by waste from ferrous metalworking, as well as ovens used likely for curing meat and possibly for glass-bead production. A clustering of pottery sherds in Area 6 may relate to local pottery production in SP III, although this is more likely explained as patterns of waste disposal. In SP IV–V production activities are mainly associated with food processing and storage. Large concentrations of grain were found in drying kilns as well as postholes and ditches that may have belonged to storage constructions. Macrofossil and geochemical measurements also indicate the use of seaweed ash to preserve animal commodities, dating mainly to SP IV–V. Geochemical analyses indicate copper-alloy work of rather limited volume; although these chemical signatures cannot be dated by stratigraphy, the likely use of brass may indicate a late SP IV or early SP V date. Iron smithing and other SP III production activities seem to have a wider spatial distribution within Area 6, whereas the grain-related activities in SP IV are concentrated in its central and western parts. This variance in distribution is probably due to the construction shortly after AD 600 of a fortification on the plateau’s eastern rim, thereby sequestering that part of the plateau. This created a vertical and horizontal stratigraphy for the various remains: processes from SP III are found across the plateau whereas the processes from SP IV–V were focused west of the fortification (Østmo, Ch. 11). This chronology of activities and use of space is supported by radiocarbon dates from a number of features.

The land underneath the Kjellerhaug grave mound and the area to the south of it were cleared for pasture in the late Neolithic. Analysis of layers sealed under the mound indicates that the area had been used for livestock early in SP I. South of the mound an arable colluvium was detected. The soil derives from cultivation deposits, presumably firstly redeposited as mound fill in Kjellerhaug. It remains uncertain whether the colluvial formations were tilled again after the build-up or only prior to it. Features such as postholes and hearths were discovered at different levels within the colluvium, but as the stratigraphy and radiocarbon dates not always coincide, this indicates both a gradual build-up of the layers and the varying impact of later trampling or other farmyard activities obscuring the upper part of features. A few cooking pits, one of which is dated to late SP I, predate a major change taking place at the transition to SP III, when the area became the site for a range of production processes (Fig. 9.1). Due to the coarse chronology provided by radiocarbon dates, it remains unknown whether all processes were contemporaneous.

158 

 B: Excavation Results 2011–12

A32030 Ua-45332

A44031 Ua-45360

AD 1033–1152

A12178 Ua-45376

A37744 Ua-45334

AD 535–601

A20032

AD 882–971 A20476 Beta-319019

AD 414–532

A52220

AD 781–894

A12142

A52326 A10438 Ua-45377

A37770 Ua-45336

AD 882–971

AD 237–333 A40222 Ua-45356

BC 358–204

A52790 Ua-45351

AD 238–333

A20

A12780 Ua-45368

AD 1221–1261 A37846 Ua-45335 A401438

AD 345–415

Ua-45339

AD 236–333 Ua-45337

AD 265–406 0

Cooking pit

Waste layer

Oven

Other structure

Pit

Grave mound

Ditch

Trench

Remaining deposit

Excavated area

1

5m

Fig. 9.1: Overview of Area 6: the production area with dated features other than postholes. For the latter, see figure 9.8. The stone-built fortification A20 and underlying deposits were not removed, except in the two trenches across A20.These trenches were excavated to expose underlying layers and features, uncover stratigraphic relations, and establish good sampling contexts. Except for the trenches and some areas where stones were exposed to reveal extent and construction details, the fortification A20 and Kjellerhaug grave mound were preserved for future excavation; thus, additional production traces likely remain in situ. See details in Østmo, Ch. 11 and Østmo and Bauer Ch. 12. Illustration: I. T. Bøckman, MCH.

The activities of SP III occurred across the entire plateau south of Kjellerhaug. Some of these have left definite and tangible remains, while others can be inferred indirectly. Among the processes with tangible remains is ferrous metalwork, manifest in slag fragments and spheroidal slag dispersed in various features across Area 6. Some hearths are also believed to be associated with this production. Based on microstratigraphic observation, one oven located south in Area 6 has been interpreted as a low-temperature installation used for food preparation, possibly for the curing or boiling of meat.



9 Østmo: The Production Area 

 159

Analyses of burnt clay showed no chemical signatures reflecting ferrous or non-ferrous metalwork and may relate to other types of ovens and productions processes. Among the processes suggested as taking place in this period are pottery production and glass bead manufacture. As many as 434 sherds or 66 % of the total prehistoric pottery material from the 2011–12 excavations comes from Area 6. This seemingly structured spatial pattern has two possible explanations: pottery production or waste disposal. Supporting evidence for pottery production is meager, whereas pXRF-and other soil chemical readings indicate waste disposal as the more likely explanation; both suggestions, however, will be addressed in this chapter. Glassy slag with high contents of phosphate and calcium observed in micromorphology analyses (hearth 37770, Fig. 9.1) implies a form of artisan work that may have included the use of ashy fluxes. The kind of artisan activity this represents remains speculative. Glass bead manufacture is a possibility based on the chemical signatures, though this lacks supporting evidence in terms of waste remains. The transition from SP III to IV is characterised by a restructuring of space and the introduction of new types of production. The A20 stone construction is believed to form a base of a defensive rampart or wall, functioning as a fortification. This construction sealed off the eastern edge of the activity surface from SP III, reducing the size of the flat surface area available for production activities, while simultaneously affording a measure of protection to the remaining production area. Activities here focused on food processing, preservation, and storage. High concentrations of grain were found in a corn-drying kiln, postholes, and ditches, likely reflecting both the drying and the storage of grain (Fig. 9.8). Fragments of charred seaweed, combined with chemical signatures indicative of a wider and denser distribution of charred seaweed or seaweed ash, are interpreted as indications of the preservation of commodities such as meat or cheese through the use of seaweed ash or black salt (Ballantyne et al., Ch. 19:490, 499–501; Cannell et al., Ch. 18:436–9). The charred seaweed could be related to the actual preservation process as well as storage of preserved products. While pXRF-analyses of soil sampled in a horizontal grid cannot be related to certain features or phases, they do indicate a form of low-volume copper-alloy work, likely with brass. Cannell et al. (Ch. 18:439–43, 451–2) argue that this copper-alloy work has Viking Age parallels across Scandinavia, raising the possibility that these geochemical signatures may date to SP IV or early SP V. The spatial restructuring and the construction of A20 implies that activities from SP IV onwards have been protected behind a fortification. The end date of the fortification remains uncertain; however, the area of the A20 saw no later activity until the modern period (Østmo, Ch. 11). The features west of the fortification were a mix of remains from SP III to early SP IV with no clear internal stratigraphy, due to the prehistoric colluvial formation, possible cultivation, stocking of animals, and muddy trampling by humans in this area, as well as disturbances from activities related to the modern farmyard (Bauer and Østmo, Ch. 8:154–5). Consequently, only radiocarbon-dated features provide chronological information on the activi-

160 

 B: Excavation Results 2011–12

ties there. Waste from metalwork and charred grain were occasionally found mixed together in the same contexts, probably as a result of truncations and re-deposition of older waste in younger features. In some cases there is also a lack of diagnostic waste. Thus, it is not always evident what processes had taken place in individual ovens or hearths. In addition, the waste from the various processes was not only found in the ovens or hearths, but also in postholes, deposits, and cooking pits. Because the occurrence of the production waste there is likely due to re-deposition, such contexts form terminus ante quem dates of the production that produced that specific debris; nevertheless, similar types of production may have taken place in Area 6 at a later date. In this chapter, the archaeological features will be presented first, followed by a discussion of the indications of metalwork, treatment of cereal, and other processes, drawing on artefacts, ecofacts, and analyses of micromorphology and soil chemistry.

9.1 Ovens, hearths, and kilns The features associated with production and processing included six ovens, hearths or kilns. The features portray an internal variation and could reflect a range of processes, from metalworking to grain drying and processing of other foodstuffs. Stratigraphic information and radiocarbon dates from the features show that the variation is also chronological: beginning in early SP III and continuing through SP IV and into SP V. The features will be presented here in chronological order. Oven A401438 (Figs.  9.2.a–b) was located in the south-eastern part of Area 6, underneath A20. The oven was only partially exposed within a trench and consisted of a bowl, a fire pit, and a flue (A45470, A39340, and A37190, respectively). The feature was cut partly into subsoil and partly into greenstone schist bedrock. The flue was lined and covered by flat stones, which, assuming they originally followed the cut into the bedrock, may have been slightly displaced towards the south-east during the later construction of A20. The bowl was the deepest part of the oven, lined with very compact, greyish sandy clay and filled with a humic soil with inclusions of dung, presumably originating from a later period when the area was used for animal stocking (Macphail and Linderholm, Ch. 17:404–5; Bauer and Østmo, Ch. 8:141). The construction bears superficial similarities with L-shaped corn-drying kilns with regard to the placement of the flue in respect to the oven (Monk and Kelleher 2005:302; Monk and Power 2012). However, in A401438 it seems the flue was designed to lead air into the fire pit rather than to lead heated air from the fire pit into the chamber. The elevated fire pit may have presented a challenge with circulating hot air and smoke into the chamber, but depending on the chamber’s architecture this may have been compensated by a chimney effect that would have secured the circulation of hot air, as seen

9 Østmo: The Production Area 

 161

Section B

Section A

Section C

Ua-45337

AD 265–406

Section A

Section C

Section B

0

Flue

Waste layer

Fire-pit

Cut

Bowl

Bedrock

Clay lining

Stone

Charcoal layer

Macrofossil sample

Cultivation deposit, fill layer

Soil chemistry sample series Micromorphology sample

Fig. 9.2.a: The oven A401438. Illustration: I. T. Bøckman, MCH.

1m

Area 6

162 

 B: Excavation Results 2011–12

Fig. 9.2.b: Photos from different stages of excavation of oven A401438 with main features delineated: (i) flue before excavation, facing northwest; (ii) flue and fire pit after removal of upper part of stone lining, facing northwest; (iii) emptied flue and fire pit, cut in the bedrock visible, facing northwest; (iv) emptied flue, fire pit, and bowl, cuts in the bedrock visible, facing southeast. Photo: MCH. ­Illustration: I. T. Bøckman, MCH.

9 Østmo: The Production Area 

 163

STONE CONSTRUCTION A20

HEARTH A37744

HEARTH A37770

Ua-45334

Ua-45336

AD 535–601

AD 237–333 Ua-45357

AD 243–335 0

Waste layer

SAMPLES

Hearth A37770

Macrofossil sample

Fill layer, cultivation deposit, subsoil

Soil chemistry sample series

Old ground surface

1m

Area 6

Micromorphology sample

Removed stone Stone Fire-cracked stone

A20

Fig. 9.3: Profile 41719: section through the possible hearths A37744 and A37770, facing north. Illustration: I. T. Bøckman, MCH.

for example in the reconstruction of the Hyrdehøi pottery oven from the late Bronze Age (Ingvardson 2005:50; Monk and Kelleher 2005:95). Micromorphology and soil chemistry analyses from the bowl indicate a low-temperature oven most likely used for preparing meat or drying cereals (Macphail and Linderholm, Ch. 17:410–11). The signatures of decaying meat found in the fill of the bowl (Macphail and Linderholm, Ch. 17:410) may indicate that it was used for smoking and drying meat. A charcoal layer dated to AD 265–406 (Ua-45337) covered the bottom of the fire pit and dates the oven to early SP III. Further north, also covered by the fortification A20, two possible hearths, A37770 and A37744, were partially exposed in a trench dug through this fortification. Both continued into the northern trench wall. Judging from its exposed part, A37770 was a simple round or oblong pit placed at the edge of the plateau where the terrain falls towards the sea in the east (Figs.  9.3–4). It measured approximately 1.5 metres in cross-section and was 30  cm deep, cutting partly into bedrock, partly into subsoil, with a few stones placed towards the bottom of the cut. The fill comprised three layers with varying content of charcoal fragments and red burnt clay; the subsoil bore signs of in situ burning. An unidentifiable piece of corroded iron, sintered clay, and a helical blue glass bead found in the backfill may be related to the production in the pit; alternatively, they were coincidentally redeposited there.

164 

 B: Excavation Results 2011–12

Fig. 9.4: Hearth A37770 (centre) prior to excavation, facing north. The bedrock forming the edge of the plateau is visible to the right. Photo: MCH.

Somewhat similar pits, but with larger inclusions of slags and metallurgic ceramics, have been examined at other Norwegian sites – for example, a group of hearths at Rødbøl in Vestfold (Rønne 2008) and Hurdal in Akershus (Bergstøl 2005). A37770 did not seem interconnected with other features, as seen at the late Roman Iron Age smithy at Fossdalen, where a separate pit for the bellows lay in direct continuation of the hearth (Kristoffersen 1988). However, it is possible that such associated elements were obscured by the overlying fortification in the unexcavated part of the feature. The placement on the edge of the plateau, precisely at the point where the terrain begins to slope downward, could also permit placement of a bellows. Soil micromorphology has shown the presence of spheroidal slag and indications of temperatures of around 1000  ̊C, fitting well with an oven or hearth used for metalworking (Macphail and Linderholm, Ch. 17:407). The hearth was radiocarbon dated to early SP III, that is, AD 237–333 (Ua-45336). Hearth, A37744 (Figs. 9.3 and 9.5), only 1.15 metres further west, was of a different character. It measured 53 cm in diameter, 15 cm deep, and was lined with flat stones. A thin deposit containing charcoal and some red burnt patches of subsoil was discovered underneath and partly between the stones and included a few fragments of



9 Østmo: The Production Area 

Fig. 9.5: Plan (above) and section (below) of the possible hearth A37744. Photo: MCH.

 165

166 

 B: Excavation Results 2011–12

sintered clay and slag. No finds or waste inside the pit could otherwise illuminate its use; relying on its similarities with younger hearths from Sigtuna, it is assumed to be related to metalworking (Söderberg 2002:258, fig. 3). It was radiocarbon dated to AD 535–601 (Ua-45334). Two ovens, A10438/52326 and A12142/52220, both consisting of two parts – a hearth interlocked with a shallow pit partly filled with stones forming a figure-ofeight-shaped type of oven – lay further north-west on the plateau (Fig. 9.6). Both were poorly preserved, consisting of deposits no more than 8–14 cm thick, presumably representing the bottommost section of the oven constructions. A10438/52326 was recognised primarily on the basis of its resemblance in terms of layout and composition to the slightly better preserved A12142/52220. A similar, better-preserved oven dated to the late Iron Age excavated at Toten, Oppland, has been interpreted as part of a smithy (Loktu and Hovd 2014). The lack of slag and inclusion of no more than four fragments of sintered clay in the Avaldsnes ovens means that an association with iron smithing is unlikely; however, it is possible that they were used as grain-drying kilns, as more than 30 charred grains were found in each of the thin deposits. Grain-drying kilns of a similar construction are known on the British Isles (Monk and Kelleher 2005; Ballantyne et al., Fig. 19.8). The Avaldsnes specimens are closer to the oven from Toten in size terms and are smaller than the parallels from the British Isles, where the bowl alone measures about 1 m in diameter. The small size could be due in part to their fragmented state of preservation. Only one of the ovens was dated by radiocarbon, giving a date range of AD 882–971 (Ua-45377). Oven A44031, a keyhole-shaped kiln, was discovered in the northern part of Area 6 (Fig.  9.7). The chamber was shallow and contained charcoal lenses and concentrations of red burnt clay presumably from the collapsed oven lining. The clay found in the chamber was not preserved in solid form – possibly, indication that it had not been exposed to very high temperatures, as clay burnt at temperatures below 450°C may be less resilient to water and more likely to dissolve (Brorsson 2005:80). Still, it should be noted that it remains unknown whether the clay is representative for the oven as a whole, as other parts of the oven lining may have been removed or truncated. The bulk samples from the bowl fill contained large amounts of grain and some spheroidal slag. The relatively low MS readings around the oven (Fig. 9.9; Macphail and Linderholm, Fig. 17.17.b) in tandem with the indication of lower temperatures provided by the dissolved clay means that this was more likely a grain-drying kiln than an oven associated with metalwork. If this is the case, the spheroidal slag likely is the result of re-deposition. A long flue containing charcoal led into the chamber; it is expected that a fire pit lay at the mouth of the flue, which would have been intended to keep sparks from entering the drying chamber and setting fire to the grain. The large amount of charcoal found in the flue is puzzling; one possibility is that it was intended to create an oxygen-reduced atmosphere inside the oven. A radiocarbon date from one of the oat grains shows that the oven was in use in the period AD 1033– 1152 (Ua-45360).



9 Østmo: The Production Area 

 167

Fig. 9.6: Figure-of-eight-shaped oven A12142/52220 prior to excavation (above) and cross-section (below). Photo: MCH.

168 

 B: Excavation Results 2011–12

N

A B

W E

S A

B

N

S (reflected)

W 0

E

1m

(reflected)

Flue Bowl Clay lining Silt containing large amounts of charcoal Stone

Fig. 9.7: Oven 44031 with flue. Photo prior to excavation. Photo: MCH. Illustration: I. T. Bøckman, MCH.

Area 6



9 Østmo: The Production Area 

 169

In addition to the described features there were a few cooking pits or pits that in light of artefacts, ecofacts, and soil chemical analyses could either be associated with ironworking or contain redeposited materials.

9.2 Buildings, fences, and other constructions Seventy-five postholes and seven stakeholes were excavated in Area 6. The postholes were generally round and small with cross-sections measuring below 40  cm and with no preserved post imprints. It was not possible to identify buildings on the basis of these postholes, though some formed rows that may have been sections of walls or fences. These findings make it unlikely that larger buildings such as longhouses would have been standing here. Rather, this has been an area of varied use with fences relating to husbandry, sheds, lean-tos, or buildings of relatively simple construction related to production processes or storage. Most of the postholes contained cereal, and radiocarbon dates on carbonised grain from a selection of 11 postholes show that ten of these were from SP IV–V and only one from SP III: A50424 (Ua-45375) (Figs.  9.8–9). Charred grain is normally preferred over charcoal as dating material; firstly because grain is an annual plant whereas the life period of the trees from which the charcoal derives may produce dates significantly older, and secondly, at multi-phase sites, the circulation and redepositing of charcoal may result in radiocarbon dates that reflect earlier activities (Løken et al. 1996:57; Diinhoff 2005b:110; Gustafson 2005g:50–5; 2016:111). While the same may be the case for grain, the latter is generally believed to be less susceptible to such reworking within deposits. As a rule, larger volumes of either charcoal or grain are more likely to bear relation to their context as compared with single fragments or grains. Most postholes here were grain rich; of those few postholes that did not contain cereals, two were selected for carbon dating on the basis of charcoal to clarify whether they represented activity or a time period different from the grain-containing postholes. Consequently, one posthole was dated to early SP III and the other to early SP IV (Fig. 9.8). Given that 11 out of 13 dated postholes were dated to SP IV–V, it seems likely that the dated and undated postholes mainly represent activity in this period, and that most of them have been part of lighter constructions related to the processing and storage of grain. In Ireland, postholes have been found close to grain-drying kilns; these finds have been interpreted as roofs, kilns attached to storage sheds, and screens or windbreaks (Monk and Kelleher 2005:84). In fact, experiments with corn-drying kilns show that the results may be greatly affected by wind and weather; thus, roofs and screens serve as important control mechanisms in the drying process (Monk and Kelleher 2005:102–4). One of the two postholes dated to SP III (A50424) contained a larger concentration of spheroidal slag and vitrified or sintered clay and is likely related to metalwork on the plateau.

170 

 B: Excavation Results 2011–12

±

A10161 Ua-45369

A10197 Beta-319018

AD 689–775

AD 781–894

A53576 Ua-45381

AD 893–971 A49884 Ua-45370

A32087 Ua-45333

AD 605–647

AD 1668–1950 A49724 Ua-45373

AD 872–971 A12036 Ua-45364

A49699 Ua-45343

A12178

AD 898–986

AD 389–530

A50677 Ua-45372

A51007 Ua-45367

A50424 Ua-45375

AD 214–326

AD 605–647

AD 890–970 A12060 Ua-45366

AD 890–971

A50691 Ua-45345

AD 656–764

A50604 Ua-45365

AD 894 - 972

0

5m

Dated posthole

Dated material: grain

Posthole

Dated material: charcoal

Other structure Grave mound Excavated area

Fig. 9.8: Overview of postholes in Area 6 with datings. Illustration: I. T. Bøckman, MCH.

9 Østmo: The Production Area 

A20032 A10373 A10385 A52702 A45593

 171

A37744 A37770

A32050

A50369 A50414 A50424 A50837

5m

Cooking pit or possible hearth

A25

0

A20

526

A52790

MS

Hearth

112,1–500

Posthole

500,1–750

Charcoal concentration

750,1–1000

Leached waste deposit

1000,1–1250

Layer

1250,1–1500

Stone construction A20

1500,1–2000

Other structure

2000,1–6434,3

Grave mound Trench Excavated area Fig. 9.9: Overview of MS readings and relevant features for metalwork; see Macphail and Linderholm (Fig. 17.17.b) for further detail. Darker colours signify higher MS readings. Please note that the cooking pits included in this map have only a low volume of slag; however, their locations make them relevant. Illustration: I. T. Bøckman, MCH.

9.3 Metalwork Remains of hearths or ovens, one piece of plano-convex slag, and spheroidal slags, as well as soil chemical signatures, bear witness to the metalwork that has taken place in Area 6. The presence of iron slag and the lack of moulds, crucibles and pieces of copper alloys, silver, or gold indicated that metalwork primarily consisted of iron-smithing. In Fig. 9.9 the most relevant features in respect to metalwork is combined with Macphail and Linderholm (Fig. 17.17.b) analyses of MS readings. The apparently equal distributions of burnt and sintered clay and of slags (Figs. 9.10 and 9.12) led to an initial assumption that the clay represented the remains

172 

 B: Excavation Results 2011–12

BURNT CLAY IN LAYERS AND STRUCTURES

Grams per feature 0,01–1 g 1,01–5 g 5,01–15 g 15,01–30 g 30,01–62,6 g

Trench

Excavated area

0

10 m

SINTERED CLAY IN LAYERS AND STRUCTURES

Grams per feature 0,01–5 g 5,01–50 g 50,01–150 g 150,01–250 g 250,01–458 g

Fig. 9.10: Overview of the spatial distribution of burnt and sintered clay. With a few possible exceptions, the analysed clay does not indicate metalwork, but rather other activities. However, the distribution shows some correspondence with the distribution of slags; see Figure 9.12. The distributions in Figs. 9.10 and 9.12 show amount per context, not density of finds in each context. Since much of the waste may be the result of re-deposition, the associated processes did not necessarily take place where the waste was found, but probably in the near vicinity. Illustration: I. T. Bøckman, MCH.



9 Østmo: The Production Area 

 173

Fig. 9.11: Plano-convex slag (above right), spheroidal slag from different contexts (left), and a mix of slag from bloom refining and forging found in the same context (centre) (Scale 1:2). Photo: Terje Tveit, AM.

of ovens or metallurgic ceramics, as found at several sites with traces of Iron Age metalworking (e.  g., Bergstøl 2005; Rønne 2008; Reitan 2010). A selection of clay and slags was subjected to thorough analysis by Bernt Rundberget and Unn Pedersen (2015) with the intention of identifying non-ferrous metalwork or specialised processes such as brazing or carburisation (Gustafsson and Söderberg 2005; Söderberg 2014). The analysis consisted of visual characterisation and measurements by a pXRF (model Niton XL3t GOLDD+), with the result that only four sintered fragments and none of the burnt clay had high Fe readings indicative of ironwork. No indications of non-ferrous metalwork were found (Rundberget and Pedersen 2015). Thus, with a few possible exceptions, the clay does generally not reflect metalwork and should rather be interpreted as waste from other types of ovens or processes. From Area 6, 564.7g slag was collected, of which 3.26  g were spheroids and 308.4 g was a single lump of plano-convex slag (Fig. 9.11). The slag underwent similar analyses as the clay in order to identify visual and chemical signatures representing various processes (Rundberget and Pedersen 2015). A small fraction of the fragments bear the characteristics of smelting slag, but considering their low volume, they likely originate from a primary process of bloom refining (Rundberget and Pedersen 2015). Most of the slag is waste from iron forging and as such a byproduct of secondary processes. One single piece of plano-convex slag formed in the bottom of a hearth was found in the make-up of the Kjellerhaug grave mound in the northern end of Area 6 (Rundberget and Pedersen 2015). Although found out of context, it is reasonable to assume it originates from a hearth in Area 6. Fifty-two pieces of spheroidal slag were found in postholes, ovens, ditches, and layers. Spheroids form during processes of

174 

 B: Excavation Results 2011–12

bloom refining and forge welding; while their scarcity indicates that they originate from secondary processes, the analyses could not provide firm conclusions regarding their origin (Jouttijärvi and Andersen 2005:328–9; Jouttijärvi, et al. 2005:301–2; Jouttijärvi 2009:975; Dorling 2011:35; Rundberget and Pedersen 2015). The distribution of different types of slags does not correspond to any spatial pattern reflecting different areas for primary or secondary processes (Rundberget and Pedersen 2015). In fact, slags of both kinds could be found within the same contexts, for example in postholes directly next to each other, such as A50414 and A50424. Similarly, the fragments seem to be mixed due to cultivation, trampling, or truncations, such as in the redeposited fill A25526 in fortification A20 (Figs. 9.9 and 9.12). Figure 9.12 rather indicates that the distribution reflects separate processes occurring at separate times within the same concentrated areas as well as post-depositional activities. Consequently, the ironworking in Area 6 encompasses at least two and likely several different processes and techniques. The spatial distribution of slag coincides quite closely with that of spheroidal slag, although spheroids appear in fewer features (Fig. 9.12). The significance of this combined spatial pattern is confirmed by the soil chemical analyses. Macphail and Linderholm (Ch. 17:406) interpret the generally high MS readings as the results of prehistoric metalworking. The readings also show two areas of especially high MS values measuring approximately 4x4 m2, which Macphail and Linderholm (Ch.  17:406, Fig. 17.17.a) identify as two smithies or forges. The south-eastern area of high MS levels overlaps with a concentration of slag, spheroidal slag, and sintered clay. Whereas the analysed clay does not show direct contact with iron or other metals, it does indicate high temperatures. Furthermore, although no hearth or forge was identified there, one 16 cm deep feature, categorised as a cooking pit (A45593), with a few fire-cracked stones, 17 fragments of sintered clay, and one spheroidal slag could be related to metalworking activity. At the Roman Iron Age ironworking site at Rødbøl, Vestfold, small charcoal-rich hollows were interpreted as hearths. While bearing a noted morphological similarity to cooking pits, their function as ironworking hearths is clearly indicated by the smithing debris they were found to contain (Rønne 2008:93–109). A few well-preserved workshops, such as one of Migration Period date in Skeke in Sweden and one from the Viking Age at Viborg-Søndersø in Denmark, may serve as models for lesser-preserved sites. These workshops featured installations including a large post to hold the anvil, support for bellows, charcoal storage, and a casting box in the form of a flat stone covered in sand (Thomsen 2005:280–2; Hjärtner-Holdar 2012). Similar functionalities could be expected in other sites; these examples provide an interesting perspective on the high amount of spheroids found in posthole A50424. As at Rødbøl, where the distribution of spheroids and hammerscales led to identification of three or four forges (Jouttijärvi 2009:977–8), the concentration of spheroids in A50424 (22 pieces, almost half of the total amount from Area 6) may be indicative of a proximity to a forge and anvil. The radiocarbon date from the posthole A50424 is AD 214–326 (Ua-45375).

 175

9 Østmo: The Production Area 

SLAG IN LAYERS AND STRUCTURES

Grams per feature 0,01–5 g 5,01–20 g 20,01–50 g 50,01–100 g 100,01–458,7 g

Trench

Excavated area

SPHERIODAL SLAG IN LAYERS AND STRUCTURES

Pieces per feature 1 piece 2 3 4 22

Fig. 9.12: The spatial distribution of slag and spheroidal slag in Area 6. Illustration: I. T. Bøckman, MCH.

0

10 m

176 

0

 B: Excavation Results 2011–12

1m

Cut Stone Area 6

Fig. 9.13: A20032: possible working surface within a smithy. Left: plan and section drawing of the feature. Right: photos during excavation. Photo: MCH. Illustration: I. T. Bøckman, MCH.

The north-western peak of MS values (Fig. 9.9) does not correspond directly with concentrations of slag, spheroids, and sintered clay, although some of the adjacent features may bear some association with these elevated measurements. One of these features is A20032, which contained two spheroids, a tiny piece of slag, and burnt clay, as well as a glass bead. A20032 is of unclear function and consisted of central, flat, semi-rectangular stones as well as smaller stones placed along the sides of the cut (Fig. 9.13). It is possible that the flat stones were so placed for the purpose of forming a well-kept, flat and clean working surface for the various processes taking place within the smithy, though an alternative interpretation is suggested in the following. A compact hearth or charcoal lens containing burnt clay, which lay directly northeast, may be associated with assumed metalworking in this area. In the same area Cannell et al. (Ch. 18:439–43, Fig. 18.3.a) have found indications of copper-alloy work, likely in brass based on the combined elevated levels of copper and zinc. Though



9 Østmo: The Production Area 

 177

there is spatial overlap, it is not likely that this assumed low-scale copper work is the origin of the elevated MS readings; such a low volume of copper-alloy work, according to Cannel et al. (Ch. 18:453), would not necessarily imply formal structures and a significant amount of waste. This also explains the absence of tangible remains, leaving the chemical signatures as the sole indication of such activity. The indications of metalworking in Area 6 are both direct in terms of slags and features interpreted as hearths or ovens and indirect in terms of the MS values and pXRF measurements. The spatial distribution of smithing debris partially corresponds to the higher MS values measured in the western part of the plateau, indicating the possible location of two smithies. Some of the contexts interpreted as related to metalworking were concealed by the fortification on the geological edge of the scarp. The dated features related to metalworking seem to be a part of an activity phase initiated at the very beginning of SP III and ending around AD 600; as such, these features contribute to defining SP III. Metalworking waste found in younger dated features may represent re-deposition, although later instances of metalworking cannot be excluded.

9.4 Processing of agricultural products Farming consisting of cultivation and grazing seems to have been established at Avaldsnes early in SP I, although grazing may have slightly predated cultivation (Macphail and Linderholm, Ch. 17:405; Bauer and Østmo, Ch. 8:140). Overall, archaeological features contained only a few macrofossils, highlighting the high concentrations of grain in various contexts in Area 6. The charred grain from Area 6 makes up 72 % of all grains from the excavation. The grain-rich features were dated to SP IV and V with smaller clusters of consistent radiocarbon dates providing support for contemporaneity (Ballantyne et al., Ch. 19, particularly Tabs. 19.7–8, Fig. 19.8). Some kilns or ovens contained concentrations of both grain and spheroidal slag, suggesting the possibility of multifunctional ovens. The keyhole-shaped kiln A44031 is an example of this, containing more than a hundred carbonised grains as well as 3 spheroids and an iron nail (Ballantyne et al., Tab. 19.8). However, it is probable that metalworking waste from earlier phases was redeposited in younger contexts, as supported by the fact that all ovens containing fewer than 3 grains are of an SP III date, with the youngest dated to AD 535–601 (Ua-45334) (Fig. 9.1; Ballantyne et al., Fig. 19.8). By contrast, the earliest of the grain-rich features is dated to AD 689–775 (Ua-45369). The two figure-of-eight-shaped ovens A12142/52220 and A10438/52326 (Figs.  9.1 and 9.6) both contained burnt bone and sintered clay, with no direct indication of metalworking. Both were found to contain more than 30 grains of various types and may have served as grain-drying kilns, although some degree of multi-functionality is pos-

178 

 B: Excavation Results 2011–12

sible. A barley grain from A10438 was dated to AD 882–971 (Ua-45377), corresponding well to the adjacent grain-rich features, especially ditch A12178, but also postholes A53576, A49724, A12060, and A12036 (Figs. 9.1, 9.8). A construction (A20032; Fig. 9.13) with a previously suggested relation to a smithy may offer an alternative interpretation. Measuring approximately 100 by 70 cm and consisting of two flat stone slabs resting on or lined by smaller slabs, A20032 resembles constructions found in buildings from the 6–7th centuries in Rogaland. Such constructions have been interpreted as tusser, an installation for drying grain on a slab heated from below (Petersen 1933:7–8, 24–6, 87; Talve 1960:437–9, fig. 136). It has been suggested that similar constructions from Swedish sites were used for the baking of bread (Bergström 2007:160–1). The interpretation of A20032 remains inconclusive; the fill beneath the stone contained only 16 grains and a small amount charcoal, as well as two pieces of spheroidal slag. These features, in combination with undated postholes of similar shape, size, and archaeobotanical content are likely remains of contemporary granaries, small storage buildings, lean-tos, or screens designed for fencing off and sheltering workspaces. Apparent lack of charred chaff and straw from the excavation in general and from grain-rich Area 6 in particular led to a working hypothesis that grain was either threshed and cleaned outside the excavated areas or brought to Avaldsnes already cleaned as part of a taxation system. Analyses of phytoliths as well as stable isotopes were conducted with the aim of shedding light on these matters (Ballantyne et al., Ch.  19:490–7). However, the analyses showed that chaff and straw were indeed present, but likely designated to a use other than kindling, such as thatching or fodder, as they were not subjected to charring. The stable isotopes showed continuity in growing conditions, and a shift towards large amounts of imported grains from SP IV onwards seems unlikely. Rather, the origins of grain from SP I–V seem to be the same, although import from areas with similar δ13C plant values cannot be completely ruled out. Nevertheless, the high concentrations of grain as well as the presence of ovens or kilns do show a shift in the organisation of the treatment and storage of agricultural produce. A large volume of grain-drying kilns from Ireland have been dated to the 4th–13th centuries with a peak in the 6th century. Monk and Power (2012) have suggested that this increase may have been a response either to a wetter and colder climate or to a higher demand for grain to feed a segment of the population that was not self-sufficient, such as craftspeople, warriors, or an élite. The high concentrations of cereals and kilns may represent an intensification of the crop-growing area or simply a new spatial organisation of the area. Unfortunately, the analysed pollen sequence from Area 2 (Macphail and Linderholm, Ch. 17:385–8) does not cover SP IV–V and cannot answer these questions. From SP IV, Area 6 seems to have been designated for treatment and storage of various foodstuffs. Based on dated grain-rich postholes and ditches as well as dated grain-drying kilns/ovens, this designation of the area was likely begun in the 7th century and continued through the Viking Age with the youngest grain-rich feature,



9 Østmo: The Production Area 

 179

the kiln A44031 dated to SP V. The social and mythological significance of bread and its association with high-status fortified settlements may shed light on this concentration of grain and area for food storage (regarding the palisade A20: Østmo, Ch. 11).

9.5 Other production processes Charred seaweed was found in 16 of the excavated contexts at Avaldsnes. Six of these lay in Area 6, amongst which oven A37770 was dated to SP III and a posthole was dated to AD 781–894 (Beta-319018). The implications of these findings are treated in further detail by Ballantyne et al. (Ch. 19:499–501), who argue that the most plausible explanation for the routine charring of seaweed is the production of seaweed ash known as ‘black salt’, which can be used for preservation of foodstuffs. By the Hallstatt period, the extraction of salt from mines in north-eastern France, central Germany, and the Austrian Alps was a well-established practice, while production of sea salt dominated in England and the Mediterranean area (Gräslund 1973:284–5). Geological sources of salt are absent on the Scandinavian Peninsula, whereas the processing of sea salt by natural evaporation or boiling of saltwater is known from historical periods; the apparently well-established production and taxation of salt in Scandinavia during the Middle Ages suggests that the technology likely had long been available in the region, although archaeological traces for the prehistoric and early medieval periods are scarce (Gräslund 1973:287–8; Øye 2002:362–4; Larsen 2013:44–9). Archaeological excavations on Læsø, Denmark, have however shown that the practice of boiling saline groundwater have roots at least back to the Middle Ages (Vellev 1996:40–81). The alternative use of black salt as a means of preserving meat, fish, seal, whale, and cheese has been documented by several ethnographic sources from the North Atlantic region and has support in archaeological remains in the Norse settlement on Greenland. Charred seaweed has been documented at a wide range of North Atlantic sites dating from the Neolithic onwards, mainly in the latter part of the first millennium AD (Ballantyne, Ch. 19:500–1). Geochemical analyses of some of these sites has shown that the use or production of seaweed ash for food preservative leaves geochemical traces, though the findings in analyses conducted on material from Area 6 is not sufficient for drawing firm conclusions (Cannell et al., Ch. 18:439, 450). However, Cannell et al. argue that the combined indications of archaeobotanical and soil-chemical analyses do suggest that seaweed was used either as fuel or as food preservative in the form of ash. While seemingly a component of a regional traditional technique for conserving foods with black salt during the medieval and historic periods (Clément 1914; Vellev 1996:8–15), charred seaweed has scarcely been documented at Norwegian prehistoric sites. Charred seaweed in contexts dating back to SP III in Area 6 provide indications both of the knowledge and use of black salt during the Iron Age in general,

180 

 B: Excavation Results 2011–12

and that Area 6 was used specifically for the processing and possibly also storage of commodities other than grain. The food-oriented use of the plateau could therefore have predated the grain-rich contexts, as indicated by the traces of decaying meat in the low-temperature oven A401438, which was possibly used for smoking and curing. Char that might originate from cooking by roasting was also documented in the micromorphology analyses through the colluvial formation immediately south of the Kjellerhaug grave mound, providing further support to the food-oriented use of the area (Macphail and Linderholm, Ch.  17:408; Bauer and Østmo, Ch.  8:154–5). A radiocarbon-dated charcoal inclusion (Ua-45354) as well as stratigraphic relations indicate that this was a colluvial build-up that formed gradually, probably during SP II–V. In addition to metalworking and food processing, calcium- and phosphate-rich silicate slags observed in micromorphology samples indicate craftworking involving use of ashey fluxes, such as glass bead manufacturing (Macphail and Linderholm, Ch.  17:407–8). There is however no further evidence of such a production process, though five glass beads were found across the plateau, one of these in the same context as the studied micromorphology sample. The collection of 434 pottery sherds found at Area 6 comprises both tableware and cruder pots. This relative intra-site clustering of pottery may be explained as either traces of a local pottery production or as a pattern of waste disposal. Kristoffersen and Hauken (Ch. 21) have analysed and compared the sherds of bucket-shaped pots to regional corpus. They argue that the material can be seen as part of an innovative, although not particularly high-quality local craft tradition that might be rooted in a late Roman Period elite milieu situated along the Karmsund Strait. With the concentration of pottery sherds in Area 6 it is worthwhile considering Avaldsnes or its immediate surroundings as a possible place for such production. Based on an assumed connection between goldsmith workshops and production of bucket-shaped pots, the clustering of gold artefacts at Avaldsnes could provide an indication of a production site, especially as suitable clay for pottery production is to be found at Bøvågen (Kristoffersen and Hauken, Ch. 21:556). These are, however, only circumstantial indications. At Augland, Vest Agder, the remains from a massive pottery production was excavated in the 1970s. In addition to approximately 55,000 sherds and wasters, the site contained pits for processing of clay, several ovens, and a possible storage area for drying pots (Rolfsen 1980). The indications of pottery production at Avaldsnes seem meagre and tentative by comparison; furthermore, Augland may be unique as sites with remains of a production of this magnitude are otherwise not known. In addition to the clustering of pottery and the circumstantial indications, there is little evidence to support the presence of pottery production. However, most of the fragments of clay spread across Area 6 (Fig.  9.10) are not related to metalworking and could potentially be associated with other types of ovens or waste from activities such as pottery production. The lack of wasters, ovens associated with pottery production, and clay reservoirs speaks against pottery production and necessitates considera-

9 Østmo: The Production Area 

 181

Shards per feature 1–2 shards 3–5 6–10 11–20 21–23 Trench

Excavated area

0

10 m

Fig. 9.14: Distribution of bucket-shaped pottery in Area 6. Illustration: I. T. Bøckman, MCH.

tion of other explanations for the relative clustering of pottery in Area 6. Much of the material is disturbed and redeposited in trampled or possibly cultivated layers or in the waste layers preserved under A20 (Fig. 9.14). The waste layers also contain inclusions that have been interpreted as episodic dumping of latrine waste in addition to waste from metalworking and burnt and sintered clay. It is therefore likely that the eastern part of Area 6 was used for waste disposal, leading to farmyard middening in some places (Macphail and Linderholm, Ch.  17:406). According to Kristoffersen and Hauken (Ch. 21:528–40), the layers contained both early and late bucket-shaped pottery, which also could indicate waste disposal as the cause for this clustering. Both the youngest pottery and dated charcoal from these waste deposits (for example Ua-45332 from a lens in A32050; Fig. 9.3) slightly predate or just barely overlap with the very youngest suggested date range for underlying hearth A37744. This could imply that the area was subjected to the dumping of waste simultaneously with the working of hearths, or that at least some of the deposits containing waste had been at least partially relocated or levelled out prior to the building of A20. It should be noted that the pXFR measurements and other soil-chemical measurements of soil sampled from a horizontal grid in the central part of Area 6 do show high levels of phosphate, sulphur, and loss-on-ignition, likely reflecting patterns of the disposal of organic and general waste (Cannell et al., Ch. 18; Macphail and Linderholm, Ch. 17).

182 

 B: Excavation Results 2011–12

9.6 Summary The production processes occurring in Area 6 during SP III–V exhibit several main tendencies. The production during SP III occurred across the plateau and includes activities ranging from food preservation to various forms of iron-smithing. There are chemical indications of other artisan work that would be compatible with glass bead manufacture, but this remains speculative in the absence of documented physical waste. The clusters of pottery sherds in Area 6 could be explained by pottery production or patterns of waste disposal. While the latter might have more direct support in the soil-chemical analysis, firm conclusions are not possible. During SP IV a fortification was built on the eastern edge of the plateau, delimiting the area available for production processes. Besides possible copper-alloy work of limited volume, the area seems specialised and designated for food processing and storage. As this area lay protected behind a fortification, it may have been deemed important to secure and defend these subsistence goods in the southern farmyard area. Acknowledgements: Unn Pedersen and Bernt Rundberget carried out a significant analysis of slags and burnt and sintered clay from the production area, producing a visual characterisation and measurements by pXRF. I have been very fortunate to be able to use their results and am most thankful for their valuable input regarding the interpretation of the metalworking occurring in Area 6.

Egil Lindhart Bauer

10 Two Iron Age Boathouses Remains of two Iron Age boathouses are presented and discussed, followed by an interpretation of the relationship of one of the buildings to the military aspect of Avaldsnes’s power. The archaeological features and artefacts exposed during excavation, and comparisons with other sites, form the basis for the discussion. Micromorphological and macrofossil analyses, GPR surveys, and observations from coring complement the excavation results. The investigations have demonstrated that the two boathouses belonged to different periods of the Iron Age; the larger boathouse was two-phased, first constructed in the Roman Iron Age and rebuilt in the Migration Period, while the smaller was from the Merovingian Period. Both boathouses testify to the importance of maintaining readily available sea-going vessels at Avaldsnes, probably for economic as well as military purposes.

Boathouses are circumlittoral constructions used for storage and maintenance of sea-going vessels, as well as sheltering them from the environment when not in use. While most farms along the coasts would have needed boats for performing numerous everyday activities, ownership of larger ships was more exclusive; these were primarily military vessels intended for swift transport of troops. The following discussion aims to evaluate whether the two prehistoric boathouses investigated at Avaldsnes in 2012 were constructed for sheltering specifically military vessels. One of the two Iron Age boathouses (A40) was unknown prior to the 2012 investigation. This boathouse was at least 15–19 metres long, 4.9–5.6 metres wide at the entrance, and angled outward to approximately 6.9–7.9 metres further into the construction. Its two phases are dated to the Roman Iron Age and Migration Period, respectively. The boathouse was slightly broader in the earlier than it was in the later phase. The other boathouse (A41) was described during a 2001 survey (Krøger and Opedal 2001:40–1), but was not excavated prior to the 2012 investigation. Its remains were found to be approximately 11 metres long and up to 3.8 metres wide internally; the entrance breadth was not established. The building dates to the Merovingian Period. A third construction, located at Gloppe, the peninsula east of Avaldsnes, has been interpreted as a boathouse (A46) from the Roman Iron Age (Krøger and Opedal 2001:39). However, its height above sea level, the nature of the artefacts, and a radiocarbon dating (AD 1692–1920, Ua-45328) of birch charcoal established by the 2012 excavation have demonstrated that the boathouse was in use during more recent centuries, consequently excluding it from analysis in the present chapter. Possibly, the remains can be connected to one of the buildings in a 1779 painting of Avaldsnes (Bauer, Fig. 15.3). Three other potential structures previously interpreted as boathouses (Hafsaas 2005:21–3) at the Gloppe peninsula were subsequently identified as natural formations. All three features would have been submerged in the early Iron Age. The 2012 investigations furthermore disregarded a row of stones on the eastern side of the

184 

 B: Excavation Results 2011–12

Gloppe peninsula. The stones proved to be part of a fence, rather than a boathouse wall. Four other possible boathouse features along the western coast of the Gloppe peninsula were left uninvestigated – three of them due to their limited size. The fourth was roughly the same size as A41, but due to uncertainty over whether it was a natural formation, its investigation was not prioritised. These four possible boathouses would have been suitable for small boats. While important to the household economy for making use of the various coastal resources (Rolfsen 1974a:133), these features are not relevant when considering the military aspect of Avaldsnes’s naval capabilities. Norwegian boathouses were described and to some extent investigated already in the 18th and 19th centuries, but it was not until Erik Hinsch’s (1961) investigation of the boathouse at Bjelland on Stord, Hordaland, that relatively modern excavation techniques were employed (Rolfsen 1974a:14–18; Wickler and Nilsen 2012:106). Perry Rolfsen’s (1974a) pioneering work on the boathouses in Jæren on the southwestern coast of Norway is still important in current boathouse research. Rolfsen surveyed 49 boathouses, of which four were excavated. Rolfsen considered artefacts found in the excavated boathouses, datings of the constructions, functional interpretation, and their relative chronology.

10.1 Methods and problems The 2012 investigation was carried out on the basis of narrow trenches dug roughly perpendicular to the long axis of the two boathouses (A40 and A41). Five trenches were dug in the largest boathouse (Fig. 10.3), four from within the walls and extending towards the central area and one across the central part of the boathouse. To avoid two modern drainage ditches cutting through the boathouse’s long axis in the southwestern part of the boathouse, three aligned trenches were dug, rather than a single, continuous trench. In the smaller boathouse, a single trench was dug from within the northern wall southward across the construction’s mid-section, almost to the base of the southern wall. One or more sections in each trench were documented by profile drawing, and a number of samples were taken to investigate micromorphology, macrofossils, and soil chemistry. The sections exposed the constructions’ stratigraphy – most importantly, the wall and floor sequences (Nilsen and Wickler 2011:122; Wickler and Nilsen 2012:109). The micromorphology analyses (Macphail and Linderholm, Ch. 17) have substantiated and added to the stratigraphic interpretations carried out in the field. To reduce intervention in the boathouses, the trenches were dug only 0.5  m wide, exposing only small surfaces. Stratigraphic excavation proceeded within the trenches until reaching the subsoil. Coring around the larger boathouse contributed to delimiting the inner gable end towards the northwest, where the walls were diffuse in the exposed sections.

10 Bauer: Iron Age Boathouses 

0

Boathouse

 185

50 m

SEA LEVELS Current

AD 550 (+1,95 m)

AD 200 (+2,4 m)

AD 800 (+1,6 m)

Fig. 10.1: The location of the two Iron Age boathouses and grave monuments on the settlement plateau’s eastern edge. Sea levels around AD 200, 550, and 800 are estimated, and original topography is reconstructed. Illustration: I. T. Bøckman, MCH.

In 2013, Area 8 was surveyed using ground-penetrating radar (GPR). The survey was intended to establish a more secure delimitation of the larger boathouse than the limited excavations and coring were capable of providing, as well as to locate other possible boathouses in the vicinity. The ground was waterlogged during the survey, leading to poor results (Stamnes and Bauer, Ch. 16:371–3). The survey did not locate other boathouses in Area 8.

186 

 B: Excavation Results 2011–12

A methodological problem is presented by the limited scope of the investigations of boathouse features, usually consisting of narrow trenches or small test units. Such limited investigations have led to uncertainties regarding the boathouses’ interpretation (Wickler and Nilsen 2012:109), as will been seen with regard to both boathouses at Avaldsnes. The boathouses at Stend, Bjelland, and Nordbø (below) exemplify the significantly greater amount of information that can be gathered from such buildings when completely exposed. Another methodological problem concerning premodern boathouse sites is the fact that many are destroyed or remain undetected due to limited traces above ground. While some boathouse types had massive outer earthen wall banks or stone walls, some were constructed with modest or no earthen walls (Grimm 2002:105, 14). The larger boathouse at Avaldsnes remained unknown until 2012 due to such simplicity in its construction. On the other hand, features historically recorded as boathouses may rather reflect a past misinterpretation of natural depressions as boathouses. Bjørn Myhre (1985:50) argues that the distribution of boathouses in Hordaland, Rogaland, and Vest-Agder appears to reflect historical political organisation; however, any reliable map of boathouse distribution will show only a minimum of the possible boathouse sites, due to the reservations mentioned above (Grimm 2001:59).

10.2 Boathouses – the building and its use Boathouses are probably as old as the boat itself. Norway’s long and rough coastline necessitated protection of valuable vessels, in summer and winter, from sun, cold, wind, and water (Rolfsen 1974a:12; Stylegar and Grimm 2005:256). Despite their widespread use throughout the Iron Age, relatively few prehistoric boathouses have been excavated, especially when compared to excavated dwellings from the same period. More than 50 boathouses have been archaeologically investigated (Stylegar and Grimm 2005:254; Wickler and Nilsen 2012:107). However, as mentioned above, most of these investigations are based on narrow trenches. Prehistoric boathouses varied greatly in their construction (e.  g., Schjelderup 1995), but as a rule the construction was oblong with a gable end facing the water. A boathouse usually consisted of three components: the building proper, the landing place extending into the water, and the slipway between the building and the landing place. The slipway and the interior of the building could have timber or stone flooring, if not simply earthen. Prehistoric boathouses tend to have curved walls, while a majority of medieval boathouses are rectangular. Usually a boathouse sheltered a single vessel, but the size of a boathouse could vary. Most were 10–12 metres long, probably intended for boats for fishing and local transport, while some were up to 40 metres long and intended for other types of vessels (Rolfsen 1974a:11, 20–4; Bill and Grimm 2002:197; Stylegar and Grimm 2005:256–8).



10 Bauer: Iron Age Boathouses 

 187

In Old Norse texts, two words describe boathouse constructions: naust and hróf. The first construction type was a massive house, while the second was a light shelter. Furthermore, each type occurred in sub-types (Schnall 1978:286–8; Stylegar and Grimm 2005:256). Nausts occurred mainly in Norway, whereas hrófs were more common in the rest of Scandinavia and the North Atlantic islands. Boathouses were primarily timber constructions, but the walls could be of stone or a mixture of stone and material such as gravel, sand, or turf. The chosen construction type seems to have been determined by local building material and geographical conditions. Postholes as footing for roof and wall constructions were common in boathouses, at least in the naust type. Hróf-type boathouses – more simply consisting of insignificant earthen walls with a light roof – are difficult to discern in the landscape today. Hróf constructions without posts as part of the roof construction, or even without a roof, are also known (cf. Bill and Grimm 2002:199; discussion of A40 below). Some boathouses feature a keel trench or conduit along the central axis intended for guiding the vessel as it was hauled into the building. Some constructions contain one or more hearths as well as cultural deposits of varying thickness. Common artefacts in boathouses include pottery, nails, and tools related to fishing and woodworking. Such features, deposits, and finds demonstrate that boathouses could house a variety of activities; for instance, cleaning, maintenance, and repair of boats and equipment related to sea-travel. Boathouses could also be used as worksites or provide room for temporary occupation, or even be used as banquet halls (Rolfsen 1974a:94–5; Myhre 1985:38; Grimm 2001:64–5; 2002:108–9; Stylegar and Grimm 2005:256, 8; Wickler and Nilsen 2012:108). Prehistoric boathouses varied greatly in size, exemplified for instance by the 49 boathouses surveyed by Rolfsen (1974a:42, fig. 7) in Jæren. Among these, only five exceeded 20 metres in length. Inger Storli (2006:82) classifies boathouses as large when their length exceeds 20 metres. Oliver Grimm (2001:56, 62) refers to boathouses 18–20 metres long as huge. Grimm (2001:57) believes boathouses up to 10–15 metres were constructed for vessels used in local activities such as fishing. Rolfsen (1974a:133) reasons similarly: large boathouses contained sea-fearing vessels, while small boathouses contained vessels used along the coast. Small vessels needed only a small crew; sea-fearing vessels, on the other hand, required a larger crew for rowing. Rolfsen reasons that large vessels therefore belonged not to a single farm, but rather were part of a larger organisation, controlled by chieftains or wealthy farmers. The sheltered vessel’s size can be inferred from that of its respective boathouse. Accordingly, military qualities can be ascribed to large-sized Iron Age boathouses, as these could shelter vessels capable of transporting a large crew which could also be warriors (e.  g., Grimm 2001). At the same time, Grimm (1999:55; 2001:64–5) stresses the multifunctional nature of boathouses – ships with military potential, such as the Nydam ship, could also have been used for transportation of crew and goods in peacetime. The boathouses themselves could likewise have been used for various activities, as mentioned above.

188 

 B: Excavation Results 2011–12

10.3 Boathouse A40 Of the two Iron Age boathouses discovered at Avaldsnes, boathouse A40, in Area 8, was the larger. Based on the assumption that boathouse size can reflect vessel type, it will be argued that this boathouse was suited for sheltering a vessel of military significance. As such, boathouse A40 will be the focus of the following discussion. The construction was unknown prior to the Avaldsnes Royal Manor Project’s 2012 investigation. It was barely visible aboveground as a faint northeast-to-southwest-oriented depression in the slope east of the Flaghaug grave mound and the settlement plateau. The depression lay perpendicular to the water, situated above a good natural harbour, sheltered by the Bjørnholmen islet to the north. For more information on Kormt’s topography, see Skre, Ch. 28. The investigation was carried out on the basis of five 0.5  m wide hand-dug trenches, four from within the wall banks into the boathouse interior and one in the central part of the boathouse. Two modern drainage ditches cut northeast to southwest through the boathouse interior, truncating features and breaking the stratigraphic sequence between the boathouse’s central part and the walls on both sides (Profile 28857 in Fig. 10.5 and Profile 33430 in Fig. 10.3). Observations in the trenches demonstrated that the boathouse had two phases of wall banks, cuts for roof-bearing posts, and floor deposits related to the two respective wall bank phases. In addition, there was a central keel trench or conduit from the earlier phase. The wall banks seem to have angled inwards, making the boathouse narrower closer to the entrance. The entire construction was covered by a colluvium that was probably formed by eroded deposits caused by agricultural activity further up the slope towards the farmyard to the southwest. The wall banks, combined with observations from coring, enabled the delimiting of the boathouse. The stratigraphy observed in the profiles (Figs. 10.3 and 10.5) suggests the walls’ outlines in the two phases, but the exact length and breadth of the boathouse phases were hard to ascertain. A gravel road covers the northeasternmost remains of the boathouse, rendering them inaccessible for excavation. Soil tilling had disturbed the western wall banks, obscuring their original extent, while the eastern wall banks were well preserved. There were no remains of inner plank walls (like in the Stend boathouse, cf. Myhre 1977) or wall ditches, but the wall banks and the floor deposits helped with delimiting the interior. Truncation of the old ground surface, visible in the sections on both sides of the boathouse, showed the initial cut for the boathouse interior, but the finished construction was somewhat narrower (Profile 30376 in Fig. 10.3, Profile 28857 in Fig. 10.5). The distance between the wall banks’ inner boundary and the conduit – presumed to correspond to the construction’s central long axis – established the boathouse’s inner breadth. Of the two boathouse phases, the earliest was the broadest, with the interior measuring 7.9 metres in the boathouse’s broadest part. The later building phase’s breadth was a little less than 7 metres. The later phase seemed to be a reestablishment of the wall banks by means of



10 Bauer: Iron Age Boathouses 

 189

Fig. 10.2: The location of boathouse A40 at the bottom of the slope east of the settlement plateau. The Bjørnholmen islet shelters the location. The stone row extending from the shore at a different angle is modern. Facing north-east. Photo: MCH.

adding soil onto the already existing banks, rather than a completely new construction (Figs. 10.3 and 10.5). A speculative estimate of the entrance breadth for the earliest phase is 5.6 metres, while the later phase is 4.9 metres; estimates are based on the wall bank segments’ positions, as seen in all sections (Figs.  10.3 and 10.5). However, considering the limited areas opened for the investigation, the estimates should be taken with a grain of salt. The boathouse’s walls apparently were not curved, as for example in the boathouses at Stend in Fana (Myhre 1977:34–5) and Bjelland on Stord (Hinsch 1961:12–13). Rather, they seem to be angled outwards from the entrance, widening further into the boathouse. As the inner gable wall in the boathouse at Avaldsnes is not exposed and has not been accounted for, it cannot be ruled out that the construction continued further to the southwest, in which case the boathouse was longer than estimated. Coring in the area southwest of the trenches located soil indicative of wall banks; however, it is possible that this soil is collapsed or ploughed-out wall deposits. These deposits might have eroded downslope, making the boathouse seem shorter. The GPR-data from the post-excavation survey could neither confirm nor disprove the outlined boathouse phases, as the walls banks only showed as stronger reflections

190 

 B: Excavation Results 2011–12

Profile 33430

Profile 32432

1 10

3

1

2

6

4

5

33 43 0 2 43 32

303

76 310

35

Profile 30376

5

2

Profile 31035

1

4

10

8

5

9

7

0

WALL BANK 1 Earlier phase 2

Later phase

3

Uncertain phase

7

Conduit

8

Possible use phase

9

Construction layer, levelling

Later phase

6

Uncertain phase

Stone

SAMPLES

FLOOR LAYER 4 Earlier phase 5

Modern drainage ditch

10

Turf, colluvium, modern agricultural deposit, modern redeposited soil, abandonment phase, subsoil

Micromorphology

Old ground surface

Macrofossil

Fig. 10.3: Profiles in boathouse A40. Illustration: I. T. Bøckman, MCH.

Soil chemistry

1m



10 Bauer: Iron Age Boathouses 

 191

in a few areas (Stamnes and Bauer, Ch. 16:373). The above-mentioned boathouses at Stend and Bjelland were both ship-shaped – that is, widest in the middle. The impression that the walls in the Avaldsnes boathouse were not curved, but rather gradually widened further into the boathouse could indicate that the inner gable end in fact has not been found, entailing a construction several metres longer. A ship-like shape for the Avaldsnes boathouse would make it approximately 30 metres long – significantly longer than what could be observed on the surface. Such a length would better fit the length–breadth ratio of other contemporary boathouses, as further discussed below. Two postholes for roof-bearing posts were identified in the southeastern trench (Fig. 10.5), one belonging to the earlier construction phase and one to the later. Both were located partially within the wall banks, thereby freeing maximal floor space for the vessel in the boathouse’s dug-down interior. Stratigraphy and radiocarbon dating confirmed the postholes’ relationship to separate construction phases (Profile 31310 in Fig. 10.5). The smaller posthole (A31295) belonged to the earlier boathouse phase from the Roman Iron Age (AD 258–381, dated using alder charcoal from the posthole fill, Beta-324648). The larger posthole (A31003) was from the Migration Period (AD 436–552, dated using birch charcoal from the post imprint, Beta-324651). The postholes’ proximity suggests a similar layout for the two boathouse phases, with posts placed in roughly the same place. This substantiates the above-mentioned hypothesis of the later boathouse phase as a reestablishment of the earlier phase, rather than a new construction. A possible third posthole was exposed in the northeastern trench. This was smaller than the two postholes mentioned above, and it lay further into the boathouse. It is uncertain how – or even if – this feature formed part of the boathouse construction. It could alternatively relate to activity within the boathouse. Despite the limited amount of exposed constructional features, a discussion of the boathouse’s roof construction will be attempted, beginning with the postholes. The younger of the two large postholes contained a post imprint 37 cm in diameter. Coupled with the dimensions of the stones used for bolstering it, this suggests that the post belonged to a substantial roof construction. Rather than employing the trestle-framed construction common in early Iron Age longhouses, it is possible that the boathouse lacked tie beams connecting the roof-bearing posts on each side of the building. Instead it could be a simple rafter construction (Johansen 2002:18) of the kind suggested by Hinsch for the Bjelland boathouse. A construction with tie beams would be impossible if the boathouse had been intended for vessels with high stems (Hinsch 1961:17). The Nydam ship had a stem height of 4 metres, the Sutton Hoo ship 3.8 metres, and the later Kvalsund ship 4.1 metres (Rolfsen 1974a:104). Tie beams would consequently restrict the boathouse to vessels with low stems. Collar ties close to the roof’s ridge (Fig. 10.5) and lintel beams connecting the posts along each wall could possibly compensate for the lack of tie beams; the inner gable end also might have been reinforced by a single tie beam (Rolfsen 1974a:100–1). Another possibility is that the roof-bearing posts were indeed connected by tie beams, but with posts high enough to create the required internal height for the vessel’s stems. While the post

192 

 B: Excavation Results 2011–12

imprint in the posthole from the later construction phase demonstrates a considerable thickness for the roof-bearing post, this does not necessarily entail great height. The posthole was 0.5 metres deep, and to accommodate a ship with 4-metre high stems, at least 4.5-metre high posts would be required if the posts were connected by tie beams. Such great height would significantly affect the building’s stability, which leads us to another part of the boathouse construction: the stabilising wall banks. One of the main reasons for employing tie beams in a roof construction is to stiffen it, leading pressure from the roof straight down onto the walls. Without tie beams, the pressure from a simple rafter roof would instead be forced down and outwards, necessitating wall banks to withstand it (Hinsch 1961:13–14). Several excavated boathouses, for instance at Bjelland (Hinsch 1961), at Stend (Myhre 1977), and at Jæren (Rolfsen 1974a:101) feature such exterior supportive wall banks constructed of stone, turf, sand, and gravel (Myhre 1980:166). While the Avaldsnes boathouse had wall banks, these could not have performed much of a supportive function, as they did not adjoin any internal plank walls. According to Rolfsen (1992:45), wall banks necessitated inner wooden walls to prevent collapse. However, to prevent decomposition of the wooden planks, the only reasonable construction method would be to place a layer of stones or coarse gravel nearest to the inner wooden walls. That the wall banks in the Bjelland boathouse lay directly against the inner plank wall was evidenced by stone and gravel having collapsed into the room as the wall decomposed (Hinsch 1961:7). In the Avaldsnes boathouse, there was no trace of wall ditches or wall planks. The floor deposits had accumulated over the inner demarcation of the wall banks, stratigraphically above the respective wall phases to which they belonged. Consequently, a boundary consisting of a plank wall between the wall banks and floor deposits would be impossible. Such a boundary would have led to the opposite stratigraphic sequence, with the wall banks having collapsed over the floor deposits upon removal or decomposition of the wall planks, as in the Bjelland boathouse. A few stones were exposed in the trenches through the southeastern wall bank in the Avaldsnes boathouse, but these were not nearly as numerous, as large, or as regularly placed as in the wall banks in the Stend or Bjelland boathouses (Hinsch 1961:12; Myhre 1977:32). The only sizable stones exposed in the wall banks in the Avaldsnes boathouse were those related to the posthole from the later construction phase. There was furthermore no trace of substantial stone walls outside the wall banks exposed in the trenches, as these would have shown in the GPR data from the 2013 survey (Stamnes and Bauer, Ch. 16:373). The lack of massive supporting wall banks and wall ditches, combined with the postholes’ location partially within the wall banks, suggests that the Avaldsnes boathouse was a different type of construction from the other boathouses mentioned above. Arguably, the boathouse shared some traits with the hróf construction type. While the wall banks could not support the outward pressure from the roof construction, they would help lay the foundation for the posts, supplementing the stabilising effect of the bolstering stones. The lack of significant walls to support the pressure



10 Bauer: Iron Age Boathouses 

 193

from the roof suggests that the roof construction was light. Such a light construction is suggested for one of the two boathouses at Harre Vig in Denmark. The walls, or even the ship itself, might have supported the light roof construction. Another possibility is that the boathouse at Harre Vig was not roofed (Bill and Grimm 2002:208). A caveat for this interpretation, which is not discussed, is that the construction indeed had postholes for a roof construction, though none were exposed by the narrow trenches (Bill and Grimm 2002:203) dug through the construction. The Avaldsnes boathouse had substantial posts, so an alternative to a light roof construction is oblique posts outside the wall banks. Such posts would replace the supporting function of tie beams or massive wall banks by transferring the pressure from the roof construction into the ground. Neither the excavations nor the GPR survey could confirm or disprove this, but it remains a definite possibility. Oblique external posts supported the walls in the boathouse at Norbø in Rennesøy (Auestad 1992). This boathouse had earthen wall banks, with two rows of circular postholes in the boathouse interior and oval postholes 2–3 metres beyond these rows. The shape of the cuts indicated that the oblique posts were inclined at 25 degrees. Consequently, they could not have been part of the roof, as the angle would be too low. Instead, the posts were probably supports for the roof, transferring pressure from the roof construction into the ground. The boathouse was 29.5  m long, 5–6  m wide, and dated to AD 1030–1220 – much younger than the Avaldsnes boathouse (Auestad 1992:5–7). As with the Avaldsnes boathouse, the Norbø boathouse had a ditch dug in the centre. Auestad (1992:7) interprets this as a drainage ditch. Equally likely, however, it could have been a conduit for the vessel’s keel. The ditch might of course have had a double function. Regardless of its intended use, flowing water would gather in such a ditch. Flat stones forming part of a pavement lay close to this ditch (Auestad 1992:7). In the Avaldsnes boathouse, a flat stone on top of a trampled activity surface (Macphail and Linderholm, Ch. 17:412) in the boathouse’s central part (Profile 31035 in Fig. 10.3) might have been part of a similar pavement. The stone must belong to the later use phase, as it lay stratigraphically over the central conduit. No pattern of stones forming a pavement showed up in the GPR data. Such a pavement might have been removed before cultivation began in the area (Bauer and Østmo, Ch. 8:154). Viewed in cross-section, the conduit in the Avaldsnes boathouse’s centre had angular edges and a flat base. It formed a 30 cm wide and 10 cm deep cut into the subsoil. No stones were observed in the excavated conduit profile, but wooden planks possibly lined the conduit when the boathouse was in use. Such planks would further ease transportation of vessels into and out of the building. Radiocarbon dating of birch charcoal (AD 426–533, Beta-324650) from the fill in the conduit assigns it to the later boathouse phase. This suggests in-filling of the floor surface as part of the establishment of the later boathouse phase. Two ceramic sherds (S12776/2–3), neither of them from bucket-shaped pots (Kristoffersen and Hauken, Ch. 21:529), lay in the bottom of the conduit. Consequently, the sherds were probably deposited during activity related to the earlier boathouse phase.

194 

 B: Excavation Results 2011–12

It follows from the foregoing discussion that the Avaldsnes boathouse could not consist of a confined room with continuous plank walls (see Hinsch 1961:8–9 for a suggestion of how the walls in the Bjelland boathouse were constructed). The postholes demonstrate that the construction was roofed, unlike the simplest hróf-type of boathouse. The boathouse was probably more of a framework construction with a simple rafter roof resting on postholes, over a depression with wall banks on each side. The degree of bioturbation visible in the micromorphology of the boathouse’s floor deposits supports this hypothesis. The analyses suggest that – although it was roofed – the boathouse’s sides were open to the environment (Macphail and Linderholm, Ch. 17:413–14). Consequently, in addition to lacking the supporting function, the Avaldsnes boathouse lacked the insulation also provided by such walls (Myhre 1977:49). After the last glacial period – as the continental ice sheets melted – sea levels rose rapidly. However, the melting ice sheets also caused the land mass to rise due to diminishing weight. Consequently, both these motions must be accounted for when assessing sea-level changes at the coastline. During the period from 12,000 to 6,000 years ago, the Norwegian coastline saw great variation in sea level due to both land upheaval and melting ice elsewhere in the world causing sea level to rise. Over the past 2000 years, the global mean sea level has been relatively constant and close to present-day levels. The relaxation of the land mass in response to deglaciation still causes vertical land motion in Norway, however (Masson-Delmotte et al. 2013; Simpson et al. 2015:7, 18, 20). Vertical land motion, or land upheaval, is essential to consider when discussing when prehistoric boathouses were in use – or rather could be in use. Boathouse A46, mentioned in the introduction to this chapter could not have been in use in the Roman Iron Age, since the site was submerged. The Norwegian Mapping Authority (Kartverket) provides current numbers of the land upheaval in Norway. For Avaldsnes, the annual land upheaval is 1.4 mm (Kartverket 2016a). This number is absolute and does not take into account sea-level rise. Relative sea level takes into account both land rise and sea-level change; in other words: relative sea level is the level of the sea surface relative to the shore (Simpson et al. 2015:18). As mentioned, the absolute sea level has been fairly constant during the last two millennia. Consequently, the change in relative sea level equals the land upheaval; thus, the annual drop in relative sea level at Avaldsnes is in the following considered to be 1.4 mm. Since the modern gravel road running through Area 8 (Figs. 10.2 and 10.4) covered the boathouse’s presumed lower end and entrance, determining the sea level in SP III is essential for estimating the length of the boathouse. The Caledonian bedrock at Kormt is complex, with differences between the northeastern and southwestern parts of the island along the Kormt fault (Prøsch-Danielsen 2006:11). These differences have caused varying degrees of land upheaval in various parts of the island (Rolfsen 1974a:24). According to Rolfsen (1974a:23) and Myhre (1977:58), the sea level by the

10 Bauer: Iron Age Boathouses 

 195

3m

S G S

4m

G S G

5m

S G S

6m

0

10 m

Earlier phase

Micromorphological thin section M34217 SEA LEVELS

Later phase

Current

Conduit

AD 550 (+1,95 m)

Lighter areas estimate the maximum length

AD 200 (+2,4 m)

Posthole Modern gravel road

Fig. 10.4: Boathouse A40’s two construction phases, with estimated sea level around AD 200 and 550. Being based on limited evidence from five narrow excavation trenches, the boathouse’s outline is tentative. The three south-western trenches were the first to be dug, with placement based on the surface appearance of the boathouse feature. After investigation, it became clear that the boathouse’s orientation diverged somewhat from expectations. Two modern drainage ditches cut north-east to south-west through the boathouse interior, truncating features and breaking the stratigraphic sequence between the boathouse’s central part and the walls on both sides. The disturbance of the boathouse features by cultivation in the construction’s western part is another reservation regarding estimates of the building’s outline. The micromorphological thin section (M34217; see Macphail and Linderholm, Ch. 17:413) from the north-eastern trench displays alternating sandy (S) floor and gravel (G) deposits, numbering five and four, respectively. The gravel was possibly introduced by storm surges. Photo: R.I. Macphail. Illustration: I. T. Bøckman, MCH.

196 

 B: Excavation Results 2011–12

two early Iron Age (AD 150–600) boathouses at Nes on the southwestern side of Kormt was 1.5 metres higher when the boathouses were in use. According to Prøsch-Daniel­ sen (2006:25), the difference was 2 metres. According to the Norwegian Mapping Authority (Kartverket 2016b), the land upheaval at Nes has been 1.3  mm per year. Applying this number, the relative sea level at Nes around AD 400 would have been c. 2.1 m higher than today. Allowing for a slight rise in sea level since the Roman Iron Age/Migration Period, Prøsch-Danielsen’s estimate seems sound. A relative sea level roughly two meters higher than the current sea level would not pose a problem to the two boathouses at Nes, however, with their present height above sea level constituting 5.75 m and 6 m, respectively (Rolfsen 1974a:22). Change in relative sea level is not the only factor to consider when discussing boathouses’ location and age. On top of this come tidal variations. Relative sea level and tidal variations might contribute to a more precise determination of the Avaldsnes boathouse’s seaward extent. The annual land upheaval at Avaldsnes, 1.4 mm infers that the relative sea level at Avaldsnes c. AD 500 would have been 2.1 meters higher than at present. Again, allowing for a slight rise in sea level, a relative sea level two meter higher seems probable. This corresponds to the estimates of other researchers (Lindanger 1999:162). With the sea level two metres higher, the area just northeast of the gravel road (Figs. 10.2 and 10.4) would have been below sea level, limiting the boathouse’s extent in this direction. The Avaldsnes boathouse was in use for an extended time period, however. Applying the annual rate of 1.4 mm, the relative sea level at the time of the earliest spectrum of the radiocarbon dating result (AD 258–381) from the Roman Iron Age boathouse phase would be c. 2.5 metres higher that today. Adding 46 cm for the greatest astronomical tidal water at Avaldsnes suggests that the earliest boathouse phase’s entrance could not lie lower than c. 3 m above the current sea level. This indicates that the entrance to the Roman Iron Age boathouse phase lay somewhere below the gravel road. In the trench dug through the boathouse just southwest of the gravel road, alternating fine soil and coarse gravel characterised one of the floor deposits (A33570, Profile 32432 in Fig. 10.3). Five assumed occupation surfaces and four gravelly deposits alternated within a 75 mm high micromorphological thin section (M34217, detail in Fig. 10.4). This suggests that the boathouse’s entrance lay close to the tidal range consequently affected by sedimentations from intermittent storm surges. The alternating deposits are too few to be caused by regular high and low tide, as the mean high tide at Avaldsnes is only 16  cm above the mean sea level (Kartverket 2016a). Meteorological factors such as wind and air pressure can have a significant effect on astronomical tidal water. Particularly large deviations can appear when storm flooding occurs simultaneously with maximum high tide (Sælen and Weber 2015; Weber 2016). The Norwegian Mapping Authority’s recurrence intervals describe a sea level superseded after an average span of years, adjusting for weather factors. For instance, once every 20 years the sea level at Avaldsnes reaches 101 cm above the mean sea level (i.  e. roughly half a metre higher than the maximum high tide). The correspond-



10 Bauer: Iron Age Boathouses 

 197

ing number for a five-year interval is 91 cm. Probably, the alternating deposits in the Avaldsnes boathouse represent such flooding, possibly recurring once every 5–20 years. This accords well with the location of the trench where the alternating deposits were found, situated approximately 3.5 meters above the current sea level. The boathouses in Brunnavik at Sola in Jæren to the south of Avaldsnes demonstrate how such constructions were gradually moved lower in accordance with land upheaval. This practice substantiates the benefit of keeping the vessels as close as possible to the waterline (Rolfsen 1974a:122). While it is not possible to pinpoint the exact location of the Avaldsnes boathouse’s entrance, it seems reasonable to assume that it was somewhere underneath the gravel road running perpendicular to the water’s edge. As mentioned above, this would place the entrance at approximately 3 metres above the current mean sea level, thus corresponding to the situation at Stend, where the boathouse also was in use in the Roman Iron Age and Migration Periods (3rd–late-6th century). The boathouse at Stend lies approximately 2.5 m above the current sea level at high tide and approximately 3.5 m at low tide (Myhre 1977:31; Grimm 1999:50–1; 2001:tab. 1). Land upheaval in Norway is greater further in from the coast (Simpson et al. 2015:figs. 4.1., 4.2). However, numbers from the Norwegian Mapping Authority (Kartverket 2016c) show that the difference between current annual land upheaval rate at Avaldsnes and Stend is small, only 0.2 mm. Given the 1.6 mm annual land upheaval at Stend, the relative sea level there would have been higher by 2.9 metres c. AD 200 and by 2.3 metres c. AD 600. A slight rise in sea level should be allowed for. While the mean high tide at Stend is +40 cm, the variation in astronomical tidal water at Stend is +78/-79 cm, and the recurrence interval for a high tide of +1 meter is only one year (Kartverket 2016c). These numbers suggest that the Stend boathouse would be prone to flooding with even greater frequency than the Avaldsnes boathouse. Consequently, a relative sea level in the 3rd to 6th centuries higher than 2.3–2.9 meters seems unlikely. Alternatively, the boathouse’s earliest phase should be considered to be pushed forward in time, perhaps to the late 4th century (cf. Grimm 1999:50). Boathouses varied in length and breadth, but up to and including the Viking Age they remained quite narrow relative to their length; boathouses became considerably broader in form during the Middle Ages (Myhre 1985:40; Stylegar and Grimm 2005:258). An established method for typologically dating boathouses is carried out by calculating the aspect ratio between the inner length and the entrance’s inner breadth, the latter determining the size of the vessel that could enter the boathouse (Rolfsen 1974a:115; Myhre 1985:39). Based on studies of the boathouses at Jæren, viewed in relation to boathouses from Hordaland, Myhre (1985:44) suggests that an aspect ratio greater than 4.5 combined with a height greater than 1.5–2 m above current sea level indicate that a boathouse should be dated to the Roman Iron Age or the Migration Period. The land upheaval along the Norwegian coast has varied, as discussed above. The difference between past and current sea level is greater in the north than in the south, and likewise greater in the east than in the west. Conse-

198 

 B: Excavation Results 2011–12

quently, the length–breadth ratio appears more reliable to use for dating purposes than the height above sea level, according to Myhre (1985:42–3). The radiocarbon dating results and finds from the Avaldsnes boathouse correspond with its location at approximately 3  m above the current mean sea level, according to Myhre’s dating estimation. It is impossible to establish the exact length– breadth ratio for the boathouse’s two phases. The entrance has not been found, and the inner gable end’s location could only be estimated by coring and the wall banks’ curvature as observed in the profiles. Consequently, the boathouse’s length can vary by several metres; 19 metres is a safe estimate. If we accept the estimated entrance breadths of the two phases, the ratio for the earlier phase is 3.4; for the later phase, the ratio is 3.87. These ratios contradict Myhre’s dating method, which results in a reduced ratio for younger boathouses. The ratios are highly tentative, however; firstly, because the boathouse entrance was not exposed, and secondly, because the remains’ extent towards the southwest is likewise tentative – a greater length would skew the ratio figures significantly. Crumlin-Pedersen (1997:189) notes that the early Viking Age ships from Tune and Gokstad (and also Oseberg) are broader than the late Viking Age warships from Ladby and Skuldelev. Therefore, while the length–breadth ratio for boathouse A40 seems to contradict Myhre’s dating method, the ratio of various known vessels rather indicates that the index is difficult to use without taking other factors into consideration. For instance, as Crumlin-Pedersen remarks, differences in proportions between the early Viking Age vessels from Norway and the late Viking Age vessels found in Denmark might be caused by varying navigational conditions rather than different age. However, the relatively low aspect ratio of the Avaldsnes boathouse in fact suggests a continuation of the construction further up the slope, consequently accommodating a longer vessel, which lends credence to Myhre’s dating method. Applying Myhre’s suggestion of a 4.5 aspect ratio for the Roman Iron Age or the Migration Period would give the Avaldsnes boathouse a length of 25 m and 22 m in its respective phases. A longer construction would increase the possibility of curved walls, similar to those of several fully excavated contemporary boathouses. The 2013 GPR survey unfortunately did not allow a delimitation of the boathouse. In several of the trenches there were two distinguishable floor deposits. The deposits varied in thickness, from about 5 to about 20 cm, and were clearly affected by cultural activity. Micromorphological analyses showed trampled soil in both phases of floor deposits in the southeastern trench, close to the wall bank. With the vessel taking up the central part of the boathouse, the area along the walls would logically have been subjected to more trampling than the central floor area. The thin sections furthermore contained charcoal fragments and thin, semi-horizontally oriented wooden fibres – presumably woodworking remains from boat maintenance or construction. Soil slaking associated with iron staining was also visible in the thin sections from the later use phase in the boathouse’s central part. This soil slaking was possibly caused by saltwater drip from hauled-up boats with rusty iron rivets. In

10 Bauer: Iron Age Boathouses 

 199

Collar tie

Ra fte r Lintel beam

Roof-bearing post

Profile 28857 1

2

5

3 4

Profile 31310 1

28

5 6

85

7

313

7

8

10

Beta-324648

AD 258–381 0

WALL BANK 1 Earlier phase 2

Later phase

FLOOR LAYER 3 Earlier phase 4

Later phase

5 6,7

8

Beta-324651

AD 536–552 1m

Construction layer

Turf, colluvium, subsoil

Stone

Posthole

Modern drainage ditch

Macrofossil sample

Old ground surface

Fig. 10.5: Suggested reconstruction of boathouse A40, seen in cross-section from the landward side. Profiles 28857 and 31310 – dug through the south-eastern wall bank – display some of the features on which the interpretation is based. Illustration: I. T. Bøckman, MCH.

200 

 B: Excavation Results 2011–12

addition, it is possible that strongly iron-stained charcoal in floor deposit A33570 (4 in Profile 32432 in Fig. 10.3) could be associated with such iron-rich drip; alternatively, it represents remains from occupation deposits older than the boathouse (Macphail and Linderholm, Ch. 17:412). The deposit (A28805, Profile 31035 in Fig. 10.3) related to the boathouse’s later use phase also contained charcoal. Fireplaces are frequently found in boathouses (Nilsen 1998:94; Grimm 2002:107). Although the trenches in the Avaldsnes boathouse exposed no such features, a disturbed or cleaned-out fireplace in another part of the boathouse is a possible source of the charcoal. Boathouses should be considered multifunctional buildings (Grimm 2002:109). Relatively thick cultural deposits in many boathouses (e.  g., Rolfsen 1974a:57; Nilsen 1998:90; Wickler and Nilsen 2012:115), including in A40, suggest a use of the buildings broader than merely as storage for the vessels. Boat-related activities, such as cleaning, maintenance, and repair, as well as storing of boat-related equipment, may have been common. Standing empty for long periods, boathouses might even have been employed as temporary dwellings, and meals might have been consumed there. Fireplaces might have been used to heat food or tar for maintenance (Rolfsen 1974a:94–5). Ceramic sherds are commonly found in boathouses (Rolfsen 1974a:95; Myhre 1985:37). The large amount of sherds from high-quality ceramic pots in boathouses at Jæren (the amount was greater than in the nearby longhouses), has led Rolfsen (1974a:96–7) to suggest that the boathouses were used for storing pottery intended for distribution by boat, most likely as designated trade goods. The ceramics from the Avaldsnes boathouse was limited to five sherds, thus insignificant compared to the excavated boathouses at Jæren. Analyses of organic remains from one of the sherds in the Avaldsnes boathouse, however, testify to maintenance activities: Birch tar covered the sherd’s surface (Isaksson 2014a:4–6). This suggests that the ceramic pot contained tar for treating the wooden vessel stored in the boathouse. Despite the narrow trenches, a range of artefacts, apart from the ceramics, was recovered from the boathouse. Of these, twenty iron fragments (S12776/1) might be parts of nails or rivets, and two pieces of petrified wood (S12776/11) might be parts of a wooden nail. There were also burnt and sintered clay fragments (S12776/6) distributed in the centre of the boathouse and along the interior of its southeastern wall. Several flint artefacts (S12776/7–10), including fragments of blade-like artefacts, probably came from Stone Age settlement deposits identified further up the slope to the southwest (Hemdorff 1993:3–4; cf. Bill and Grimm 2002:208), subsequently transported down the slope by colluvial activity. Burnt bone fragments (S12776/12) were also found, but it was not possible to determine their species. Agricultural activity – both cultivation and animal grazing – probably contributed to artefact relocation in the area. The macrofossil samples from several of the boathouse deposits contained charred seaweed, as well as traces of burnt soil or turf. It is possible that turf covered the boathouse roof and collapsed into the boathouse during a fire or as the building fell into disuse.



10 Bauer: Iron Age Boathouses 

 201

10.4 Boathouse A41 The smaller of the two Iron Age boathouses (A41) lay sheltered on the western side of the narrow bay between Avaldsnes and the Gloppe peninsula – an island in the early Iron Age – in Area 9. The boathouse was identified in 2001, but not excavated (Krøger and Opedal 2001:40). The 2012 excavation identified the boathouse’s construction, which consisted of wall banks and roof-bearing posts. The construction method suggested it was from the Iron Age, but its height of two metres above sea level suggested that it was in use somewhat later than boathouse A40 further north. On the surface, the feature stands out as a depression between two parallel banks from which a few boulders protrude. In the northwest, the banks are curved towards each another, but do not meet to form a continuous gable wall. This is possibly due to disturbances from later construction and activity in the area. Based on surface observations, the boathouse was about 11 metres long and 3.8 meters wide inside. The entrance’s breadth was not established, complicating attempts at dating the boathouse using Myhre’s aspect ratio between the inner length and the entrance’s inner breadth. In continuation of the southern wall, boulders lay in an uneven line towards the water. Further south, two other stone rows extend into the water, but these belong to other, probably later, harbour constructions. There were also remains of other stone features in the area, including a building (below) that was probably younger than the boathouse (Krøger and Opedal 2001:40). A single 0.5 m wide trench was hand-dug from within the northern wall southward across, and perpendicular to, the long axis of the boathouse. The wall construction consisted of turf and large boulders. A boulder in the excavation trench lay well into the surface upon which the turf wall was built, indicating that some of the stones in the construction had already been laying there. Just inside the wall, there was a stone-shod posthole with a post imprint. Radiocarbon dating of willow/poplar charcoal from the imprint deposit placed the feature in the Merovingian Period (AD 582–637, Ua-45327). The entire posthole was exposed in the excavation trench. The posthole’s location is marked in Figure 10.6, as it was not visible in the trench section. No corresponding posthole was found along the opposite wall, either because the trench was not dug far enough towards the southern wall or due to the narrowness of the excavation trench. The post imprint was approximately 20 cm in diameter, indicating the diameter of the actual post. The post was set as closely as possible to the inner side of the wall, freeing the maximum amount of space for storage inside the boathouse. No trace of posts was found further into the boathouse, indicating that the roof construction was supported by wooden posts in line with the exposed posthole, possibly in combination with the stone and turf walls. The wall banks were much larger in this boathouse than those of boathouse A40 further to the north, indicating that the banks held a more important function in boathouse A41. The posthole’s position just inside the wall suggests that the banks contributed to resisting the outward pressure from an assumed angled roof. Soil from the wall bank covered the posthole,

202 

 B: Excavation Results 2011–12

POSTHOLE A25343 Ua-45327

AD 582–637 0

1m

0

10 m

Posthole A25343

Turf, infilling, leach layer, subsoil

SEA LEVELS

Old ground surface or cultural layer

Current

Wall bank

AD 1000 (+1,3 m)

Old ground surface

AD 800 (+1,6 m)

Estimated outline of wall banks

AD 550 (+1,95 m)

Stone Dislodged stone Macrofossil sample Fig. 10.6: Profile 25205 through the northern part of boathouse A41, displaying the wall construction and the location of the excavated posthole. Illustration: I. T. Bøckman, MCH.



10 Bauer: Iron Age Boathouses 

 203

suggesting a collapse into the boathouse after the building fell out of use. There was no trace of a wall ditch or wall planks in the trench. The excavated trench provided only basic information about the construction and no information about alternative uses or different construction phases. In the boathouse’s interior, there was a thick, bog-like turf deposit. The bottom part of this deposit (“Turf, etc.” in Fig.  10.6) might represent in-filling of the boathouse after it fell out of use; however, no clear stratigraphic transition was visible to distinguish such in-filling. Close to the surface, the deposit contained large amounts of modern refuse, demonstrating that the boathouse’s interior depression was used for garbage dumping in recent times. No clear floor deposits were identified; however, no micromorphology samples were taken from this part of the profile. No charcoal was preserved, although there was a dark, humic deposit below the refuse deposit. While it is possible that this deposit could represent a waterlogged cultural deposit related to activity in the boathouse, its character was rather that of an old ground surface. It could alternatively be related to the later building immediately to the west. The later building is unexcavated and little is known about it from previous surveys. A relationship to the boathouse is possible. However, the large amount of stone features from what appear to be several phases render the area difficult to interpret (Krøger and Opedal 2001:40–1). As in the larger boathouse in Area 8, macrofossil samples from this boathouse contained charred seaweed and traces of burnt soil or turf, suggesting similar use and/or roof cover of the two boathouses (Ballantyne et al., Ch. 19:499).

10.5 The boathouses at Avaldsnes in a military context A boathouse’s size reflects the size of the sheltered vessel. The practical breadth of a post-built construction, employed by both boathouses at Avaldsnes, is the distance between posts in the same pair. The entrance’s breadth limits the size of the ship capable of entering the boathouse, while possible tie beams between the posts restrict the height of the vessel’s stems (Hinsch 1961:17), as long as the stems were not removable. The girth of the posts in the larger boathouse (A40) suggests a great internal height. Based on the size estimates, the boathouse could have housed a vessel at least 19 metres long and around 5 metres wide. However, as discussed above, the construction could have extended further to the southwest, thus providing room for a longer vessel. The free space – around a metre – on both sides further into the boathouse would have facilitated movement and activities around the vessel, such as maintenance work. There was probably also some room for storage in boathouse – if not necessarily for trading goods, then at least for maritime equipment (Rolfsen 1974a:97; Grimm 2002:109).

204 

 B: Excavation Results 2011–12

Iron Age ships varied significantly in size and proportion. Crumlin-Pedersen (1997:185, 92) consequently discusses the problem of representativity as regards ships found and recorded. Do the relatively few ship finds reflect standards for the various periods of prehistory, if there were any standards to speak of? Nonetheless, any further discussion of the type of vessel sheltered in the Avaldsnes boathouse calls for their comparison with certain known vessels. The boathouse’s size suggests that it could house a vessel of a size comparable to that of the Kvalsund ship, which was 18 metres long and 3.2 metres wide. The Kvalsund ship, however, dates from the Merovingian Period (Rolfsen 1974a:104). If the boathouse’s length was somewhat greater than the estimated 19 metres, it could have sheltered a vessel similar to the Roman Iron Age Nydam ship, a rowed vessel, 23.4–23.7 metres long and 3.3–3.7 metres wide, found in the Nydam bog near Jylland’s southeastern coast (Rieck 1994:9, 28; 2002:80). Vessels of this kind were suited for transporting a crew which could also be warriors. Such vessels thus had a definite military potential (Grimm 2002:109, 11). The Nydam ship was built from oak trees felled between AD 310 and 320, and the vessel’s estimated use period was 30–40 years (Rieck 2002:79). The dating of the Nydam ship corresponds with that of the earlier phase of the larger Avaldsnes boathouse. According to Grimm (2002:114), a vessel of the Nydam type needed a crew of 30 men, while Rieck (1994:69) estimates that figure at 45 men. Myhre (1985:53–4; 1997:180–1) assumes that local chieftains mobilised armed forces from their residence and from a certain area around the boathouses’ location. The person holding Avaldsnes must have demanded the fealty of a number of people capable of equipping the vessel stored in his boathouse with sufficient crew for battle. Sea transport was highly preferable to land transport (Albrethsen 1997:211, fig. 3). If military potential is reflected in the number of warriors at one’s disposal, then the ability to deploy those warriors effectively, which a ship of the Kvalsund and Nydam sizes would facilitate, is a reflection of military power. A simple boathouse construction method does not necessarily reflect a lack of military function; it is rather the vessel’s size that is significant in discussions of military potential and power. Furthermore, a simple boathouse construction in itself does not indicate a small vessel. The aforementioned boathouse at Harre Vig had an internal length of 24 metres and was 5.6–6.2 metres at its widest. The boathouse’s opening was estimated at approximately 3.5 metres (Bill and Grimm 2002:203, 207). It is important to remember, however, that a vessel’s size and potential number of warriors are not the only reflections of military power – of equal importance are the warriors’ discipline and training, weaponry, tactics, and strategic information (Albrethsen 1997:210). Therefore, the definition of boathouses constructed for “military” vessels must take into account other factors in addition to their size. The other boathouse’s size (A41) indicates that it was built for a smaller vessel, probably about 10 metres long. It was probably used for local, farm-related activities such as fishing and transport (Grimm 2001:57). However, considering the above,

10 Bauer: Iron Age Boathouses 

Early Iron Age Possible early Iron Age

Middle Ages Possible Middle Ages Skipreide boundary

0

 205

50 km

Fig. 10.7: Distribution of known boathouses in the early Iron Age and Middle Ages in Rogaland and Hordaland (Based on Grimm et al. 2006). One caveat in this distribution is the large group comprised of boathouses dating either to the Iron Age or the Middle Ages. Illustration: I. T. Bøckman, MCH.

a 10-meter long vessel could possibly perform auxiliary functions during military operations, such as transportation of equipment, rations, or a small number of men. Such aspects of military logistics in Iron Age warfare are often overlooked (Albrethsen 1997:211–12). Nevertheless, military use is only one of the possible functions of vessels housed in boathouse A41. According to Myhre (1985:50–1), large boathouses from the Norway’s Middle Ages (i.  e. AD 1050–1537) are spread evenly along the southwestern coast in accordance with the skipreide organisation of that time. Boathouses from the Roman Iron Age and Migration Period, however, are often found in groups and tend to be concentrated near economic and political centres (Fig. 10.7). Certain places, such as North Jæren, feature extraordinary concentrations of large boathouses. The lack of a definite dating of many boathouses gives rise to some uncertainties in the distribution. The possibility of numerous undiscovered or destroyed boathouses furthermore complicates the matter. Nevertheless, the uneven distribution suggests a naval organisation

206 

 B: Excavation Results 2011–12

in the early Iron Age different from that in the Middle Ages, when most of the country was ruled by a unitary royal power. In earlier periods, huge boathouses located at chieftains’ farms testify to their military power, reflected in the number of warriors they could mobilise (Myhre 1985:50–1). These power centres often lay in close proximity to gathering places, such as court sites, where troops could be mobilised before initiating large sea-based military operations or smaller raids (Grimm 2001:61; 2002:111, 6; Iversen, Ch. 26). Harald Egenæs Lund (1965:292–3) was one proponent of a military function for court sites. The argument is disputed, however. According to Storli (2006:82–3), there is no clear connection between large boathouses and farms connected to court sites in northern Norway. Furthermore, Storli (2006:142) remains unconvinced of the notion that the court sites were controlled by particular chieftains or chieftains’ farms. Rather than representing a strong and stable sociopolitical system under a single leader, the court sites in his view could represent an institution akin to the Icelandic thing sites (Iversen, Ch. 26). Grimm (2010:44) presents three functional interpretations of court sites, each with varying degrees of speculation. He prescribes the highest degree of speculation to an interpretation with political and military functions, which would require a highly stratified society controlled by “petty kings”. In choosing the location for a boathouse, local topography with natural harbour qualities is undoubtedly a decisive factor. Additionally, a location’s ease of mobilisation must also have been important consideration; to that end, it would have been beneficial to cluster boathouses (Rolfsen 1974a:131). The good natural harbour and position along important traffic arteries are probably two reasons why boathouses dated to the early Iron Age are found at several later skipreide centres (Myhre 1985:51). Such a continuity of naval organisation may be present at Avaldsnes, as this farm gave its name to the later skipreide. Grimm (2001:59; 2002:114) points out that, by this reasoning, many more late Iron Age boathouses than have been identified should have been expected, especially at central places. He argues moreover that the person buried in the Flaghaug grave mound at Avaldsnes (Østmo and Bauer, Ch. 12; Stylegar and Reiersen, Ch. 22) must have controlled retinues and vessels with military potential. Geographically, Avaldsnes commanded a strategic position for controlling the strait; effective control would have required several vessels, meaning that a higher number of boathouses should be expected here. The topography around boathouse A40 allows enough room for at least two additional boathouses to the northwest, and perhaps even one to the southeast. The GPR survey carried out in November 2013 showed vague traces of another construction to the northwest of the boathouse in Area 8. However, these anomalies could be geological rather than archaeological features. Additional boathouses may rather have been placed in entirely different areas, for instance closer to Bøvågen to the west. Several areas in southern and western Norway feature noticeable boathouse clusters. The largest concentration is found around Hafrsfjord, near Stavanger (Myhre



10 Bauer: Iron Age Boathouses 

 207

1997:178). Spangereid, at the southern tip of Norway, has a concentration of seven large boathouses within an area of approximately 500 by 250 metres. The topography suggests that the majority of the structures date to the late Roman Iron Age and Migration Period. Some of these lie close to a hill fort and court site. Even if only a few of these boathouses were in use at the same time, a considerable number of men still would have been needed to crew the vessels stored there. Several of the boathouses lay side-by-side, indicating contemporaneity. Spangereid is important because of its strategic location close to the southernmost cape of Lindesnes – a dangerous point on the passage along the coast where ships would frequently need to wait for favourable wind. Control of Spangereid meant control of seafaring around the southern tip of what later came to be Norway. Control could be exercised by restricting access to a canal leading to a route through the inner fjord system, circumventing the dangerous sea passage (Stylegar and Grimm 2005:261–2). A similar situation is present at Avaldsnes: the Karmsund strait was preferable to the rough seas on the seaward side of Kormt, and the tidal current Salhusstraumen of up to 5 knots approximately 1.5 kilometres north of Avaldsnes would often require vessels to wait before passing.

10.6 Conclusions Several features previously interpreted as boathouse remains have been discarded as boathouses, while a boathouse previously assigned an early Iron Age date has proved to be modern. A number of minor features, possibly boathouses, remain uninvestigated, as they more probably represent the remains of shelters for small fishing boats or similar vessels. Although limited in extent, the excavations of the two Iron Age boathouses have provided essential data allowing conclusions on construction method as well as minimum size estimates. Uncertainties are related to specific construction details of the large boathouse at Avaldsnes. As the feature’s exact size is also uncertain, estimates of the length–breadth ratio remain tentative. These uncertainties accentuate the limitations of investigating such features using only narrow trenches. A complete excavation of the boathouse would undoubtedly yield a great deal of useful information about this apparently uncommon type of boathouse construction, as the feature seems to be well preserved. Most importantly, a full excavation would reveal the construction’s length, which is important for discussing the boathouse’s military context. The post-excavation GPR survey was unfortunately unable to establish the Roman Iron Age/Migration Period boathouse’s delimitation or to locate other boathouses in the area. The Roman Iron Age/Migration Period boathouse at Avaldsnes is more apt for consideration in a military context than the Merovingian Period boathouse, as the smaller size of the latter restricts the military capacity of the vessel that could have

208 

 B: Excavation Results 2011–12

been sheltered there. Even if the larger boathouse housed a vessel similar to the Nydam ship, the construction itself was not as massive as contemporary boathouses in other locations. There were neither outer supporting walls nor inner plank walls. Instead, the construction seems to have been partly open to the environment on both sides. Nevertheless, the vessel’s size is of significance when discussing military potential and power. Boathouse A40 at Avaldsnes allowed for a ship at least 19 metres long – perhaps longer if the construction continued up the slope to the southwest. Many uncertainties remain regarding the maritime military organisation of the early Iron Age, particularly because the number of boathouses either undiscovered or destroyed since their construction and use remains unknown. There are furthermore many unanswered questions regarding the cultural, political, and military aspects of the Iron Age maritime world. Still, it can be argued that a transition occurred in maritime military organisation from the early to the late Iron Age. Initially clustered at political and economic centres, boathouses later were spread more evenly along the coast to accommodate the skipreide, and later leidang, system, both established by the early kings of Norway, probably in the 10th–11th centuries. The boathouses at Avaldsnes represent the earlier phase, a time when chieftains equipped their wartime vessels with retinues from their own farm or from surrounding farms. Avaldsnes became the eponymous farm for the later skipreide. The conspicuous lack of boathouses from the Middle Ages at Avaldsnes can be explained by the removal of their traces by later activity, or as the result of the use of construction methods that left only scarce traces above ground, as in the case of the large boathouse found in 2012.

Mari Arentz Østmo

11 A Late Iron Age Palisade Facing the Karmsund Strait A stone construction, 2.2–3.7 metres wide, about 30 metres long, 0.3–0.6 metres high, and probably built in the 7th century, was discovered on the edge of the Avaldsnes settlement plateau. It probably represents the remains of the foundations for a wooden palisade. Its original extent is unknown, and some stones were likely removed over the centuries for reuse in fences and buildings. The construction’s parallel rows of stones and the scatter of stones between them, as well its position atop a steep slope on the edge of the plateau, are comparable with the low stone walls or ramparts of 3rd–6th-century fortifications elsewhere on the Scandinavian Peninsula. Around 400 such fortifications are known in Norway, in Sweden more than 1000. While there are possible exceptions such as Halsstein in Trøndelag, only rarely do Norwegian fortifications appear to have been permanently settled. The Avaldsnes fortification represents another possible exception, indicating that the phenomenon of hilltop settlement and enclosed farms is not restricted to continental Europe and Sweden. Material for comparison is provided by approximately 40 Iron Age settled fortifications that have been found in Sweden, especially in the Mälar region, most of which were abandoned by the mid-6th century, and therefore pre-date the Avaldsnes palisade. Nonetheless, a few sites in the Östergötland show continued use into the late Iron Age and the construction of the 8th-century Birka fortifications, though approximately a century younger than A20, provides a late Iron Age parallel. As part of a palisade that contributed to defence of the settlement, the stone construction served a functional purpose; it also would have contributed to Avaldsnes’ monumental façade towards the Karmsund Strait, and may have met other demands such as enclosure, demarcation, and the like. The main problem in identifying the precise nature of the Avaldsnes construction is the uncertainty regarding its original extent. Its defensive capacity would be limited if it provided only an incomplete enclosure of the area, leaving other parts of the headland easily accessible to attackers. This chapter describes the physical remains of the stone construction and considers the validity of the suggested interpretations as a palisade foundation with function similar to those at hilltop settlements otherwise found in Sweden and in Germanic areas close to the Roman Limes.

In the eastern- and southernmost parts of Area 6, above the steep slope facing the Karmsund Strait, the stone construction A20 was exposed (Fig.  11.1). Area 6 comprises a flat surface south of the monumental Kjellerhaug grave mound, about 1 metre higher than the terrain to the north and west. The area is delimited to the east and south by a geologically shaped scarp with a steep drop of about 3–5 metres, followed by uneven and rocky terrain sloping towards the sea. While its interpretation as a palisade foundation was suggested quite early during the Avaldsnes Royal Manor (ARM) excavation, other possible interpretations were considered and discarded. The lack of building remains ruled out interpretation as a house terrace similar to those found at, for example, Lejre in Sjælland (Christensen 2015:71–4) or Runsa, Gamla Uppsala, Helgö, and Fornsigtuna in the Mälar region (Ljungkvist 2006:50–75; 2013; Olaus-

210 

 B: Excavation Results 2011–12

son 2009:47, 2011b). Protruding bedrock made it unsuited for use as a built-up road, such as the contemporary road at Norra Gärdet at Uppsala (Duczko 1993:35; 1996:41, 115–22). It appears most likely that A20 is the remains of a palisade foundation of unknown original height and length. The structure was subjected to a limited excavation focusing on exposing the borders and examining the stratigraphy and construction. The uppermost stones in A20 and soil in parts of the construction were removed; only two cross-cutting trenches in the northern and southern segments, respectively, were fully excavated through underlying deposits and down to the bedrock. These sections also provided suitable contexts for sampling. The construction was dated to Site Period IV (AD 600–900; below, and Østmo and Bauer, Ch. 6).

11.1 The physical remains A20 consisted of two parallel rows of large stones 2.2–3.7 metres apart, with the gap between them filled by a scatter of stones of various sizes mixed with soil. The two rows followed the shape of the plateau edge, beginning at the south-eastern base of the Kjellerhaug grave mound and stretching south-south-west for almost 30 metres, interrupted only by a protruding formation of bedrock in the middle of the stretch. The size and shape of the stones in the construction varied within different parts, and it was unclear whether the preserved construction had been part of a larger structure. In his description of archaeological monuments at Avaldsnes, Johan Koren Christie (1842:334) mentions a flat area south of the mid-19th-century rectory garden that likely corresponds to the plateau in Area 6. According to Christie, this flat area was sharply cut in a straight line towards east. It seems likely that the stone construction contributed to this sharply defined eastern edge, and that A20 at this time might have been better preserved – though likely not of greater height – than it was prior to excavations in 2012. In that case, quite a few stones would have been removed after Christie’s visit, resulting in a reduced volume and a less defined eastern edge. According to local tradition at least going back to the first half of the 20th century, the area carried the name Borgjå, meaning ‘The Fort’ (Utvik 2008, and pers. comm. 2016). The age of this tradition is uncertain, but the plateau above the sloping terrain or even visible stones in the stone construction could have contributed to such an association. A foundation for a small shed was placed directly on top of A20, containing broken, modern window glass. The digging of the foundation trench would have exposed stones; thus, their presence will have been known. Protruding bedrock divided the construction into a northern and a southern segment; these segments were slightly angled and did not form a straight line. The northern segment also decreased in width towards the north where it curved moderately westwards and onto the base of the Kjellerhaug grave mound. The eastern row

11 Østmo: A Late Iron Age Palisade 

Location of the fortification in the prehistoric landscape.

0

Fortification

Bedrock

Stone

Grave mound

5m

Excavated area

Fig. 11.1: Overview of the stone construction A20 situated along the scarp edge. Illustration: I. T. Bøckman, MCH

 211

SEA LEVELS Current AD 800 (+1,6 m) AD 550 (+1,95 m)

212 

 B: Excavation Results 2011–12

of stones was more irregular and less continuous than the western; probably because stones have slid down the slope. Thus, the width of the construction may have been reduced. Where the foot of the grave mound met the slope, a dip in the terrain was filled by two levels of stones. The border stones were not uniform in terms of size, shape, or placement. Border stones in the construction were placed in one or two layers, probably depending on the shape and size of the stones. The slightly curved northernmost part consisted of two layers of rounded stones with cross-sections measuring maximally between 0.3 and 0.6 metres. In the straight part of the northern segment, the western border stones were fairly rectangular and relatively thin slabs, compared to the otherwise rounded stones or boulders within the construction (Figs.  11.1–2). The slabs measured 1 by 0.5 metres ± 20 cm and were only approximately 15–20 cm thick. They were neatly placed along the construction’s border, partly overlapping to form two layers. In the southern part there was only one layer of stones in the western border row (Fig. 11.3). These stones were rather large boulders, mainly in the shape of rounded cubes with cross-sections measuring maximally 0.7–0.9 metres, with an internal spacing of approximately 0.3–0.6 metres. The south-eastern border row was only partly uncovered during excavation and seemed to be constructed of rounded stones with cross-section measurements between and 0.35 and 0.45 metres, more similar to the border stones in the slightly curving eastern side of the northernmost segment than to the opposing western row. The gap between the border rows was filled by stones of varying sizes and shapes ranging mainly within cross-sections of 0.2 to 0.6 metres mixed with humic soil. The placement of stones in the outer rows seem to have occurred before the addition of soil, but most of the stones between the rows, regardless of size, seemed to have been added simultaneously with the soil: the soil lay both between and underneath stones. In the photos and measurements, the southern part appears to include fewer and slightly smaller stones, but this is partly a skewed image as less soil was removed during excavation of this segment, leaving a number of stones completely or partly covered. In the northern segment, however, a greater number of stones were exposed during excavation and some of the uppermost stones and soil fill was removed in order to reveal construction details. This exposition showed that the area immediately north of the protruding bedrock contained fewer stones than the area further north, possible due to modern truncations in the former. In the southernmost end of the construction, the stones were more thoroughly exposed during excavation with the aim of identifying an eventual end or continuation. Instead, the construction petered out, leaving unanswered the question of whether this was an original end or one created by subsequent truncation; the latter appears the more likely alternative. Immediately south of A20, an approximately 30  cm deep natural step in the bedrock running east–west marks the south-eastern end of the plateau (Fig.  11.3). South and east of this, the terrain becomes increasingly uneven and slopes more



11 Østmo: A Late Iron Age Palisade 

 213

Fig. 11.2: The northern segment of the stone construction. The northernmost part curving up onto the Kjellerhaug, two layers of stones visible (above, facing south-southeast). The northern part of the main segment of A20 (below, facing north). Photo: MCH.

214 

 B: Excavation Results 2011–12

Fig. 11.3: The southern segment of stone construction A20: The northern part of that segment (above, facing east); the southernmost part slightly curving towards west (below left, facing south); the natural step in the bedrock where the terrain starts to slope towards the south and east (below right, facing north). Photo: MCH.



11 Østmo: A Late Iron Age Palisade 

 215

sharply towards the strait. The plateau continues towards the west and south-west, but modern buildings and truncations made excavations futile. Similarly, to the north, the stone construction could be traced as far as the foot of the partly reconstructed Kjellerhaug grave mound, where it was found to be slightly overlying the mound. It remains unknown whether this represents the actual end of the construction or if it originally had run across and beyond the grave mound. The stone construction theoretically could have circumscribed only the plateau immediately south of the Kjellerhaug. Alternatively, the stone construction could have continued to the north of the grave mound along a slightly elevated ridge, running all the way to the Flaghaug grave mound approximately 125 metres further north. Immediately north of the Kjellerhaug grave mound, modern elements such as a parking lot and pedestrian walkway have transformed the landscape. The excavation trenches on the parking lot showed an area greatly affected by recent activities; preserved deposits and features from prehistoric times were limited. The medieval stone-built cellar prevented any further search for a possible continuation in that direction (Bauer, Ch. 14).

11.2 Dating Samples for micromorphology analyses were retrieved from the section walls of the two narrow trenches dug across the construction and into the underlying deposits. Analyses of these samples show that the A20 stones were placed mainly on layers made of general farmyard waste with inclusions of episodic latrine dumping and debris from ovens (layers 1 and 4 in Fig.  11.4), but also occasionally on features (hearth A37770, cooking pit 9 in Fig. 11.4) or partially on an underlying surface formed by colluvial processes (Fig. 11.4), possibly induced by cultivation or animal stocking further west (Østmo, Ch. 9; Macphail and Linderholm, Ch. 17:415). Based on stratigraphy, datable finds in the waste deposits, and the placement of some the A20 border stones within the layer, it seems likely that the waste deposits were subjected to some degree of levelling or reworking prior to or in relation to the construction of A20, causing some of the stones to be set deeper into the waste deposits. Redeposited pottery in the waste deposits (layers 1 and 4 and similar patches outside the trenches) included bucket-shaped pottery of early and late types, partly mixed, though there was a slight clustering, possibly from the same or similar pots with early pottery (AD 350–450/500) in and near layer 4 and late pottery (AD 450–550) in and near layer 1 (Kristoffersen and Hauken, Figs. 21.4 vs 21.14 and 21.20 respectively). One of the waste layers (layer 1 in Fig. 11.4) sealed off two small ovens or furnaces, which were dated AD 237–333 (Ua-45336) and AD 535–601 (Ua-45334) (Fig. 11.4). In addition, the dated charcoal from the overlying waste layers [layers 1 and 4 in Fig.  11.4, Ua-45332 (not sampled from section) and Ua-45339] provided a terminus post quem date for A20 to AD 414–532 and AD 236–333 respectively. As these radiocarbon dates are overlapping

216 

 B: Excavation Results 2011–12

A20

1

3 2

HEARTH A37744 Ua-45334

HEARTH A37770 Ua-45336

AD 535–601

AD 237–333

LAYER A32030 Ua-45332

LAYER A34995 Ua-45357

AD 414–532

AD 243–335

(sample not taken from section)

A20

A20

3

4 6

5

7 8

LAYER A25526 Ua-45338

AD 677–770

9

LAYER A35150 Ua-45339

COOKING PIT A37846 Ua-45335

AD 236–333

AD 345–415

(sample not taken from section)

0 1

Waste layer

8

Cooking pit

2

Hearth A37770

9

Cooking pit

3

Fill layer A25526

Colluvium

4

Waste layer

Old ground surface

5

Levelling deposit

6

Flue

Turf, fill layer, cultivation deposit, subsoil

7

Posthole

Tree root

1m

Removed stone Stone Fire-cracked stone SAMPLES Macrofossil sample Soil chemistry sample series Micromorphology sample

Fig. 11.4: Two trenches through A20 showing the underlying waste layers, ovens, and cooking pits as well as the placement of the border stones in A20. Some stones in the northern trench have been removed to permit the establishment of the section. Illustration: I. T. Bøckman, MCH



11 Østmo: A Late Iron Age Palisade 

 217

or slightly earlier than the respective underlying hearths, they likely represent redeposited waste reflecting a general period of artisanal production occurring during SP III, ending around AD 600 (Østmo, Ch. 9). The many fragments of Roman Iron Age and Migration Period pottery found in the waste layer support this conclusion. The A20 stone construction is difficult to date directly as it does not include contexts that can provide primary radiocarbon dates. It was covered only by the grassturf and lacked overlying dateable contexts or finds providing a terminus ante quem, except for the foundation for a modern gazebo or garden outbuilding mentioned above. Two charcoal samples from the infill between the stones in A20 were radiocarbon dated Cal AD 347–421 and Cal AD 677–770 (Ua-45331 and Ua-45338 respectively). It is uncertain what these bits of charcoal date, however, as the soil is redeposited from older contexts and thus contains finds and inclusions pre-dating the construction and use period of A20 (below). In better-preserved palisade foundations or ramparts, charcoal concentrations have been taken to represent wooden elements within the stone wall intended to stabilise the wall and possibly support a wooden superstructure (Olausson 1995:149). At many of the Swedish fortifications and hilltop settlements the palisade has been successively burnt, destroyed, and rebuilt several times (Olausson 2009:37). In A20, the preserved stratigraphy and amount of charcoal were too sparse to serve as reliable evidence of burnt construction elements. The construction layer between the stones (layers 1 and 5 in Fig. 11.4) contained a part of an arrow with the blade either rusted away or broken off. The arrow (S12772/2) has a socket, but its type cannot be precisely identified due to the lack of a blade; however, it is likely either R213 or R535. Peter Lindbom (2006; 2009) argues that tang arrows are the indigenous type in Scandinavia whereas socket arrows stem from an elite foreign influence. He classifies R213 and R535 as arrows of a western Norwegian tradition in the period AD 300–500 with precedents in the Roman auxiliary arrows probably brought to Scandinavia by returning warriors (Lindbom 2006:4). R535 resurfaces with the establishment of the Merovingian kingdom in the 6th century and is found in Vendel, Valsgärde, as well as the mountainous and valley areas of eastern Norway up to c. AD 600 (Gudesen 1977:108; Lindbom 2006:195–204). Because burials from the Merovingian Period in Rogaland contain few arrows altogether and none with sockets (Rønne 1999b, catalogue), a reasonable estimate for the age of the S12772/2 arrow in A20 is the period AD 300–500. As such it is seemingly one of several older inclusions in redeposited soil within A20, along with sherds of pottery likely of a Migration Period date (e.  g. S12772/85; Kristoffersen and Hauken, Fig. 21.8). In light of the recurring evidence of rebuilding at Swedish sites, it would be problematic to put too much weight on single fragments of charcoal inclusion in the infill between the stones as an indication of its period of use. Rather, the terminus post quem dates from underlying features and redeposited artefact inclusions in combination with horizontal stratigraphy are the best available indications as to when the construction was built. Contextualising the construction in terms of the various phases of

218 

 B: Excavation Results 2011–12

production in Area 6 clearly suggests that in SP III the entire area south of the Kjellerhaug grave mound, as well as beneath the stone construction, was used for metalwork, possibly pottery production and other artisanal work (Østmo, Ch.  9:158–9). From AD 600, a new type of production was established, consisting of the processing and likely storage of grain and other foodstuffs. All features with high concentrations of grain and 14C dates belonging to SP IV lay clearly west of the stone construction, seemingly respecting a construction standing at that time. The A20 infill’s lack of modern or medieval pottery, clay pipes, or well-preserved animal bone – the type of finds omnipresent in medieval and post-medieval deposits examined during the ARM excavation – supports a pre-medieval date. Taken together, the vertical stratigraphic relations and the terminus post quem dates based on finds and dated charcoal indicate that A20 was constructed no earlier than c. AD 600; the horizontal stratigraphic relations to food-related processing and the lack of medieval finds makes it likely, though not certain that the construction was erected already early in SP IV and that it remained functional or at least preserved as a substantial ruin throughout SP IV and early SP V. Only two features in Area 6 date to the 11th-13th centuries AD, impeding conclusions on the activities taking place here as well as a date for the downfall of A20. It remains uncertain whether A20 was reduced to its current state of preservation immediately after it fell out of use or whether the process occurred in several stages, permitting later phases of superimposed activities. In the latter case, any traces of such activities would have been eradicated by later removal and reuse of stones from the construction.

11.3 Fortifications and hilltop settlements The approximately 1,400 fortified sites on the Scandinavian Peninsula share common traits such as stone foundations or ramparts of stone and earth that supported wooden palisades. These sites are found mainly at elevated spots partly protected by natural obstacles in the landscape; some of these appear to be focal points in a social-political landscape, occupying a strategic position permitting control of the vicinity. Most sites dating to the 4th–6th centuries are interpreted as defensive fortifications, whereas non-military enclosures normally pre-date the 4th century (Olausson 1995:156; Skre 1998:266–7; Ystgaard 2003:22, 25–7). On the Swedish mainland, most sites in the Mälar region were abandoned by AD 550, while in Östergötland a few sites continued to be used into the late 7th century (Olausson 2009:38). Nevertheless, as small number of fortified sites were established in the late Iron Age, such as at Birka and Stora Karlsö (Zachrisson 2009). The fortifications in Norway and Sweden from the period AD 300–550 generally consist of irregular semi-circular ramparts, benefitting from natural obstacles such as cliffs or steep slopes (Näsman 1996:147; Olausson 2010:26; Hedenstierna-Jonson et al.



11 Østmo: A Late Iron Age Palisade 

 219

2013:286–7). Several researchers have argued against identifying these fortifications by the traditional term bygdeborg, generally translated as hill fort, which carries the association of providing refuge for the district – Norwegian bygd (Skre 1998:266; Ystgaard 2003; 2014:164). Following these researchers, I will use instead the more neutral term fortification. Late Roman and Migration Period fortifications without traces of settlement could have functioned as temporary refuges for groups of warriors in times of trouble, used actively as a base for either defence or attack, as satellite fortification within a hierarchy. The lack of building remains and space for storage of provisions, livestock, and the like suggests that such fortifications were unlikely to have served as refuge for a district population (Hedenstierna-Jonson et al. 2013:287). In order to distinguish between fortifications with and without settlement and to underline the former’s similarities with the höhensiedlungen in Germanic areas immediately north of the Roman Limes, I will employ the term “hilltop settlement” (Olausson 2009, 2010; Hedenstierna-Jonson et al. 2013:287–8). Hilltop settlements are generally complex, reflecting an imported, elite, Roman-influenced lifestyle that combined fortification with elite dwellings, craft production – especially non-ferrous metalwork and other prestigious objects – the milling of grain and the baking of bread. The evidence shows that hilltop settlements were characterised by elite artefacts as well as a prestigious diet, which included meat and bread, according to osteological analyses and finds of rotary querns (Olausson 2009:52; Zachrisson 2009:43–5; Ystgaard 2014:150). The most relevant comparisons for the Avaldsnes stone construction are to be drawn from the approximately 40 Swedish hilltop settlements from c. AD 300–550. Although most of these were abandoned by the time the Avaldsnes construction was built, the few Swedish sites that are contemporary with A20 appear to be rooted in the Migration Period fortification tradition, and are therefore the most suited amongst available parallels. In Ireland, enclosed settlements, or cashels, are also known; although their layout and construction differ, they provide a contemporary parallel elsewhere in north-western Europe. Regarding the approximately 40 Swedish hilltop settlements, Olausson (2009; 2010) draws parallels to the Alemannic hilltop settlements along the Roman Limes. These settlements, inhabited by Germanic leaders and their vassals, were characterised by fortifications, military equipment, non-ferrous metalwork, and Roman imports, and were generally abandoned by AD 500 (Steuer 1994; Bücker and Hoeper 2000:219–23). Based on finds of reticella glass and other imports, some of these hilltop settlements were resettled by elites in the 7th century (Bücker and Hoeper 2000:228). Across Europe, a trend has been observed: a clear decrease in hilltop and fortified settlements in the centuries following the decline of the Roman Empire and a subsequent re-emergence in the 8th–10th centuries of a variety of fortifications (Boschetti 2016:129–31; Christie 2016:52–3; Tys et al. 2016:175–6, 183). As exceptions to this general trend, some Roman defences persisted in western Europe after the decline of the Roman empire, though no longer part of a military system (Tys et al. 2016:176). ln addition, fortified enclosures were emerging

220 

 B: Excavation Results 2011–12

in northern Britain and Ireland as early as the 5th and 6th centuries, respectively. Whereas fortified settlements in Ireland increase in number towards the 7th century, Pictish northern Britain experienced the reverse trend, somewhat obscuring the overall pattern (Comber 2016; Noble 2016). The Irish ringforts vary in shape and material, with earthen or stone-built enclosing walls. While the ringforts of the 9th and 10th centuries were increasingly defensive, the walls of settlements from the 6th and 7th centuries had “modest defensive capabilities, though possibly more for display purposes than practical application” (Comber 2016:12). As such, the early ringforts, contemporary with the Avaldsnes construction, were defendable rather than actively defended – in a pinch, they would protect people and livestock from wild animals and raids, but would be less effective against attack by a larger force of warriors. Thus, the abandonment of most Scandinavian fortifications and hilltop settlements c. AD 550 conforms with the general trend across Europe. Precursors to the general European renewed investment in fortification from the 8th century are seen in some areas, such as the above-mentioned insular examples. A similar deviation is seen in renewed or continued activity at singular fortifications on the Scandinavian Peninsula, at least into the 7th century (Ystgaard 2013, 2014:212–13; Olausson 2009:59–60). A small number of defensive structures were also built in Scandinavia during the Viking Age, but these are massive ringforts, linear embankments, or underwater pile barriers (Hedenstierna-Jonson 2006:49; Olausson 2009; Pedersen 2016). In Denmark, such linear defences were initiated already in the Roman Iron Age (Jørgensen 2003). According to Ulf Näsman (1996:148), most strategic defence systems in Denmark in the Early Iron Age were, as with most of the approximately 1,400 on the Scandinavian Peninsula, designed as defence against attacking forces, and were not permanently manned. The approximately 40 Swedish hilltop settlements from c. AD 300–550 are often situated along communication routes (Näsman 1996:148; Olausson 1996:157). Similarly, Avaldsnes occupies an elevated plateau overlooking the strait, which was the key to control over seaways communication along the coast. Compared to heavily defended hilltop settlements such as Runsa (Fig.  11.5), the defensive elements at Avaldsnes are somewhat different; the stone construction is less voluminous and is only known from a small segment. Other sites, such as the enclosed magnate farm at Lindö Utmark, apparently have more in common with Avaldsnes; there the enclosure permitted defence of the farm and the enclosing wall may have had additional functions such as demarcation of various jurisdictional and societal borders. In SP IV Avaldsnes lacks building remains from dwellings, though several indirect indications of settlement were observed. These include dumping of latrine waste, lack of observed pauses in the agricultural production, and continued activity in the southern end of the farmyard. Settlement indications are thus much stronger than on the numerous AD 300–550 fortifications elsewhere in Norway. As the full extent and character of the Avaldsnes SP IV settlement remain unknown, we cannot address in detail the similarities or differences compared to the large dwellings and halls found



11 Østmo: A Late Iron Age Palisade 

Fig. 11.5: Examples of defended and defendable Swedish parallels: Hilltop site, Runsa borg (upper, Olausson 2011a:fig. 1); Enclosed magnate farm, Lindö Utmark, Kärrbo parish (lower, Olausson 2009:fig. 1).

 221

222 

 B: Excavation Results 2011–12

at Runsa or similar large hilltop settlements. Nonetheless, its proximity to an area with contemporaneous processing and storage of agricultural produce as well as indication of copper alloy work permits a discussion of similarities of other aspects of hilltop sites.

11.4 Construction and functionality of the Avaldsnes palisade At Swedish fortifications and hilltop settlements, repeated sequences of destruction by fire and resurrection are not uncommon, indicating that the palisades were rebuilt after destruction (Olausson 1995:149; 2009:37). By destroying fortifications and halls in an ostentatious display of victory, a successful attacker could garner prestige (Holmquist 2010). Although initially built in the mid-8th century, the fort rampart of Birka followed the traditional construction of Migration Period fortifications (suggested reconstruction in Fig. 11.6; Olausson 1997:402; Hedenstierna-Jonson 2006:49; Lundström et al. 2009:112; Hedenstierna-Jonson et al. 2013:294). The Birka battlement seems to have rested on frames of horizontal beams on top of the stone base, while other sites show the use of beams within the rampart combined with a palisade or a superstructure (for a presentation of various types of timber and stone fortifications, see Engström 1984:41–3). It is difficult to assess the possible construction techniques of the palisade in A20 – probably, only the very bottom has been preserved. Nonetheless, there are indications of at least two layers of stones on the northern segment; the border rows there are evenly placed, creating an even, vertical front. The construction as a whole is less clear in the southern part, where there is a high degree of internal variation. The row of massive stones placed slightly apart constitute the western border of the construction, whereas the eastern border is marked with smaller stones more closely packed. These differences in shape, size, and placement of stones demonstrate that uniformity may not have been a requirement for the construction to function as intended; alternatively, it was constructed in separate segments or phases, or in a rush, making use of all available stones. To function properly as a fortification, a palisade would need to close off a defended area. Much of the Avaldsnes headland remains unexcavated, making it impossible to assess whether A20 was one component in a broader defensive construction or merely an entirely local palisade enclosing a vulnerable or significant part of the settlement. Michael Olausson and Guy Halsall distinguish between defended and defendable sites, of which the latter are sites where a rampart’s military aspect seems to be of secondary importance while the primary function is to seal off, proclaim superiority, or add to a monumental character (Halsall 2003:215; Olausson 2009:48). Olausson (2009:49) exemplifies this with the town wall of Birka and



11 Østmo: A Late Iron Age Palisade 

Fig. 11.6: Suggested reconstruction of the fort rampart with battlement at Birka (upper, Lundström et al. 2009:fig. 53). Overview of the military installations at Birka (lower, Hedenstierna-Jonson 2006:fig. 4).

 223

224 

 B: Excavation Results 2011–12

the enclosed magnate farm at Lindö Utmark in Västmanland, Sweden, arguing that they were “fortifiable only to a certain extent” by comparison with the heavily fortified sites such as Runsa Borg (Fig. 11.5; Olausson 2009:49). The defendable sites have insufficient ramparts or a placement not truly elevated from the surroundings, but the ramparts would still provide a limited degree of protection. Based on the excavated remains, A20 seems to have obstructed access to a part of the settlement from the east while adding to a monumental façade towards the strait; as such the Avaldsnes fortifications may be understood as defendable rather than defended. Though walls such as the town rampart at Birka, bore a military appearance, they also would have been understood as providing societal and jurisdictional demarcation, displays of social status and the power of the ruler or leader in addition to defence (Halsall 2003:215–16; Olausson 1993:83; Olausson 2009:48). If not capable of withstanding the attack of an invading army, they could nevertheless offer defence against more minor attacks or play a strategic role in stalling an attack, providing refuge while buying time for negotiation (Halsall 2003:22; Olausson 2009:40, 59). Defensive aspects of such fortifications have generally received more attention than their offensive capabilities. Regarding Birka, Hedenstierna-Jonson (2006:66) points out that researchers have focused on the difficulties posed to defenders by the length and many openings in the town rampart, rather than its possible function as a “tactical base[s] for offensive warfare”. The presence of ramparts from the very start at Birka suggests that a military presence was a prerequisite for the security of the activities happening there. In Olausson’s (2009:44–5) general interpretation, the underlying causes for founding hilltop settlements were never solely military; moreover, their military aspect over time became secondary to the elite settlement’s prerogative of co-locating crafts, trade, and political, economic, military, and symbolic power. In support of this thesis, Olausson (2009:38) cites the short, intensive period of fortification-building followed by periods no less affected by warfare but without fortifications. The military strategy of fortifications at hilltop settlements, argues Olausson (2009:36–7, 58–9), would have allowed the defenders to wait for the optimal time for battle outside the walls, hopefully deterring attack in the first place. The attackers’ strategy would be to draw the defenders out by plundering the vicinity. The changes in defensive constructions reflected a general shift from a hierarchy of fortifications and hilltop settlements to combinations of linear defences at a larger scale to control waterways, secure harbours, and control the movement through bottlenecks in the landscape (Hedenstierna-Jonson et al. 2013:289, 299). Such changes paralleled changes in power structures, military strategy, and weaponry to suit new ways of battle (Ystgaard 2014:140–4). In terms of its strategic location with control over the strait and the construction of a stone-built base assumed to have carried a wooden superstructure, Avaldsnes shares some characteristics with the Swedish hilltop settlements. In its other aspects, Avaldsnes appears to have been less fortified than the larger hilltop settlements in the Mälar district, closer in character to defendable, enclosed magnate farms such



11 Østmo: A Late Iron Age Palisade 

 225

as Lindö Utmark. Shifting historical contexts may also have contributed to these differences. As suggested for the late Iron Age defensive structures at Birka, a military presence may have been important, but a palisade or rampart may have been used in offensive warfare as much as in defence; other aspects of the enclosure may have been equally important. Although it is uncertain whether the palisade terminated at the base of the Kjellerhaug grave mound or continued across the mound, the mound would have contributed to the physical stature of the palisade. The inclusion of graves is notable also in the town wall at Birka. There, some are believed to be younger graves dug into the wall while others are older graves included in or overlaid by the wall (Olausson 1993:69–77). Though undoubtedly a practical arrangement, such a foundation in an older monument could also have been intended to demonstrate a connection to the past or invoke a relation to existing power structures. This manner of relating to the past has been suggested by Lund and Arwill (2016:417–18, 422), exemplified by the late Iron Age ship settings at Glavendrup, where the stem and stern stones are set into Bronze Age grave mounds (Østmo and Bauer, Ch. 12). Common traits of the hilltop settlements in Sweden include the presence of craft remains. The plateau west of A20, Area 6, bore traces of ovens, some of which were dated to SP III, some to SP IV and SP V, as well as clusters of small postholes dating mostly to SP IV and early SP V. However, in contrast to the Swedish hilltop settlements with abundant traces of non-ferrous metalwork, geochemical analysis at Avaldsnes were indicative only of low-scale copper alloy production (Cannell et al., Ch.  18; Østmo, Ch. 9:176–7; Olausson 2009:50–2; 2010:7–9). Remains from iron smithing date mainly to SP III, whereas brasswork may tentatively be ascribed to SP IV or early SP V and is therefore likely contemporaneous with A20 (Cannell et al., Ch. 18:451). In addition to craft, rotary querns and indications of a voluminous storage capacity for cereals and finds of bread are common traits at Swedish hilltop settlements. This indicates the possibility of cereals being collected at such places, and underlines the importance of everyday commodities in maintaining a hilltop settlement of this type. Nor should we overlook the symbolic value of bread as a Roman-inspired luxury food, or the mythic-poetic motif of bread served to the elite in Rigsþula (Ljungkvist 2006:101–2; Olausson 2009:52–5; 2010:9–10). The large volume of cereals found in Area 6 dated to SP IV – A20’s assumed period of use – is interesting in this perspective. The designation of this area for the processing and possibly storage of grain suggests that at least one aspect of the strategy behind the A20 construction was to secure the provisions for those dependent on them. The intention and importance of the palisade thus seem rooted in both the actual protection provided and the authority and power expressed in the landscape.

Mari Arentz Østmo and Egil Lindhart Bauer

12 Grave Monuments at Avaldsnes At least 36 graves are known from Avaldsnes headland and the immediately adjacent islets and islands. These include two monumental mounds, a raised stones monument, four stone packings, and 15 mounds or cairns that have been identified based on morphological traits across the headland; written sources account for one additional eradicated mound with a raised stone. Several of the monuments contained multiple burials. In addition, four stray finds likely representing graves have not been securely connected to known monuments. Most of these graves have not been the objective of professional excavations, although some documentation is preserved from the 19th-century excavation in Flaghaug. Nine of the graves of various forms were excavated or superficially examined professionally in the 20th century, in addition to the limited excavations of the remains of Flaghaug and Kjellerhaug grave mounds by Avaldsnes Royal Manor Project in 2011–12. The aim of the Flaghaug investigation was to obtain information on its construction and to assess the potential for recovery of artefacts from the remains. The details of the individual secondary burials in Flaghaug are published by Frans-Arne Stylegar and Håkon Reiersen in this volume. The aim of the investigation of Kjellerhaug was to reveal the construction details and stratigraphic relations of the mound as well as to recover datable material in order to obtain a more precise date for the monument. Although the excavations were limited in scope, they succeeded in providing datings of the monuments, along with insights into the landscape and the monuments’ history. This chapter also provides an overview of all recorded graves and grave finds at Avaldsnes, which forms the basis for discussing the development of the ritual landscape at Avaldsnes and its relation to the contemporary settlement. The large number of grave monuments at Avaldsnes demonstrates the site’s importance in terms of ritual activity and demarcation of social status in the Bronze and Iron Ages. The large grave mounds and the raised-stone monument held a special position, in part due to their clear visibility to people sailing through the strait.

Some of the grave monuments at Avaldsnes are still visible today, while others are known from excavations or from written evidence or stray finds. Most of the monuments are prominently situated at elevated locations across the headland (Fig. 12.1). Two grave mounds were partially investigated during the in 2011–12 ARM excavations: Flaghaug and Kjellerhaug. Both mounds, situated along the eastern edge of the elevated settlement plateau, were integral to the visual impact of the Avaldsnes farm. As the primary components of a monumental façade towards the strait, the mounds’ dating and construction details provide valuable insights to the development of the Avaldsnes farm as a site of political power. Presented in this chapter are the results from the ARM investigation together with an account for the remaining grave monuments known from previous surveys and excavations. The degree of detail in available information spans from observed monuments of an assumed date to well-documented graves excavated within recent decades (Fig. 12.1; Tab. 12.1).

228 

 B: Excavation Results 2011–12

25-28

29 30

32

31

3-8 33

21 20 22 24

14

15

1-2

Kongshaug

16

9-11

19 17

34 35

23

0

Grave mound

36

18

Grave cairn Stone packing Small stone packing Boat grave Unmarked grave

200 m

2

B5776

Copper alloy ring

Scale 1:2

13

Raised stones monument Grave mound, estimated location and extent

12

Cremation grave (secondary) SEA LEVELS Current AD 600 (+2m)

S6810 Gold pendant Scale 1:2

B2774 Gold ring Scale 1:2

Lost bronze masks Scale approx. 1:5

B916 Lance head Scale 1:4

BC 1800 (+5m)

Fig. 12.1: Overview of graves at Avaldsnes, numbers referring to Tab. 12.1. Monuments of certain and approximated location are included in the map, while finds from unidentified monuments are represented by artefact drawings. The drawing of copper alloy spiral ring (36) is based on a similar ring (B2752, Voss, Hordaland County). Drawings of S6810 and B22774: H. Reiersen. Bronze masks: L. Sagen (Fig. 25.8) B916: L. Tangedal. Illustration: I. T. Bøckman, MCH.



12 Østmo and Bauer: Grave Monuments 

 229

Tab. 12.1: Grave monuments and grave finds at Avaldsnes listed in chronological order. ‘Map key’ relates to Figure 12.1. Each colour signifies one monument. The dark shade indicates the primary grave, the lighter shades the secondary. The sorting of the secondary graves in Flaghaug is based on Stylegar and Reiersen, Chapter 22. The ID numbers refer to the national cultural heritage database (Askeladden). Map key

Grave

Dating

1

Kjellerhaug: multiphased grave mound

SP I

2

Kjellerhaug: at least one secondary phase but no identified burials

SP I–IV

3

Flaghaug: multiphased grave mound

SP I

4

Flaghaug: Grave 2

SP III, 3rd c.

5

Flaghaug: Grave 3

SP III, 4th c.

6

Flaghaug: Grave 4

SP III, 3rd c.

7

Flaghaug: Grave 5

SP II–III, 1st–mid 6th c. AD

9

Small stone packing Small stone packing Small stone packing Unknown monument, stray find

SP III, 3rd–4th c. SP II–III, 2nd– 3rd c. SP III, 4th–5th c. SP III, 3rd c.

Unknown monument, stray find

SP III, 3rd c.

10 11 12

13

Artefacts

Copper alloy face masks, deposition of bone and plant remains,

Year of inves- References tigation/artefact recovery 1999, 2011–12 (ARM)

Rønne 1999a, ID34379

Early 19th c., 2011–12 (ARM)

Stylegar et al. 2011, ID34379

1720’s, 1834–41, 2011–12 (ARM)

Skadberg 1950, Stylegar and Reiersen, Ch. 22. ID34379

Sword, neck ring and finger rings of gold, Hemmoor bucket, etc. C718, C24819, B606– 12, B614–15, B617 Westland cauldron, B605 Hemmoor bucket, B604, C24820 Organic urn, not preserved

Stylegar and Reiersen, Ch. 22

Stylegar and Reiersen, Ch. 22 Stylegar and Reiersen, Ch. 22 Stylegar and Reiersen, Ch. 22

None

2000

Sjurseike 2001

None

2000

Sjurseike 2001

None

2000

Sjurseike 2001

Gold pendant, thimble-/basket-shaped, S6810 Gold fingering, serpent head type, B2774

1940

Reiersen 2009:37 with refs., Zachrisson, Ch. 25

1852

Reiersen 2009:36–7, 64–9; 2011:162

230 

 B: Excavation Results 2011–12

Tab. 12.1: (continued) Map key

Grave

Dating

Artefacts

14

Raised stones monument, including Jomfru Marias synål Removed mound with raised stone, Kongshaug

SP II–III, AD 1st–5th c. AD, possibly 3rd c. SP II–V

None

16

Large stone packing, Kongshaug

17

15

Year of inves- References tigation/artefact recovery Skadberg 1950, Hernæs 1997, Skre, Ch. 23, ID34379 Nicolaysen 1862–6:346, 806

None

Mid-19th c.

SP II–III, first five centuries AD

none

1934, 2005–6

Christie 1842, Petersen 1934, Hafsaas 2005; 2006

Cremation grave, secondary, Kongshaug

SP II–III

Burnt bone (sherds of pottery?)

2005

Hafsaas 2005

18

Unknown monument, stray find

Late SP III

Lance head and spear head

19

Inhumation grave, secondary, Kongshaug

SP III–V, late first millennium AD

Ten glass beads, weaver’s baton, pecked roundbutted adze, S12222

Probably plundered c. 1840, excavated 2005–6

Hafsaas 2005; 2006

20

Boat grave

Iron nails, S12220

2005

Hafsaas 2005

36

Unidentified grave mound

SP IV–V, 7th–10th c. Late SP IV– early SP V

Two bronze spiral finger rings, B5776

1902

Reiersen 2009:52

8

Flaghaug: Grave 6

SP IV–V, 9th–10th c.

Balance scale, B615

21

Grave cairn, Kongshaug

None

2005

Hafsaas 2005

22

Grave without visual marker, Kongshaug Eight grave mounds from across the settlement.

SP II–V, 1st millennium AD SP II–V, 1st millennium AD SP II–V, 1st millennium AD

None

2005

Hafsaas 2005

SP II–V, possibly also SP I.

None

23–30

31–35

Five grave cairns from across the settlement.

None

Reiersen 2009:65–69

Stylegar and Reiersen, Ch. 22

Skadberg 1950, Haavaldsen 1989b, ID14712, ID4904, ID65614, ID65615 Haavaldsen 1989b, Hafsaas 2005, ID115870, ID4905, ID4892



12 Østmo and Bauer: Grave Monuments 

 231

12.1 The Flaghaug grave mound The Flaghaug grave mound (3, Fig. 12.1) is situated at the very prominent north-eastern tip of the Avaldsnes settlement plateau. From this position, it would have formed an imposing silhouette against the sky for all who sailed through the strait to the east, as well as from Bøvågen in the north and north-west (Figs. 22.2, 28.2). In 1834, Johan Lyder Brun, rector at Avaldsnes in 1832–48, initiated a full excavation. In their analyses of the artefacts and written evidence from the 19th-century excavation, Stylegar and Reiersen (Ch. 22) propose that the mound contained a primary burial from the Bronze Age, four secondary burials from the early first millennium AD, and one secondary burial from the late first millennium AD (3–8, Fig. 12.1). This section deals exclusively with the mound’s construction based on the information retrieved during the limited 2012 investigation of the scant remains of the mound. These observations and analyses provide information on the landscape in which the mound was built, including some details on the mound’s construction that will be correlated to Stylegar and Reiersen’s suggestions. This is not an uncomplicated task: the grave mound remains were investigated by means of a single trench cutting perpendicularly through the north-eastern part of the bank – the only visible remains of the mound (Fig. 12.2). The trench’s two profiles revealed traces of Brun’s excavation, the old surface on which the mound was built, and three different deposits of mound fill, including a row of stones related to the mound’s construction. As a result of expansions, additions of chambers, and repeated opening of the mound for secondary burials, it is assumed that the stratigraphic sequence would not be uniform throughout the bank. The following observations offer only a small glimpse into the construction of the Flaghaug mound. The profiles revealed preserved mound fill (Fig.  12.3) within the bank, which could represent different mound phases. A build-up consisting of different deposits and construction elements would be in line with observations for other Bronze Age mounds in Rogaland (e.  g. B. Myhre 1981:70; L. Myhre 1998:86). The presence of multiple construction phases is a recurring feature in Danish Bronze Age barrows, though the varying quality of the documentation of deposits and construction elements excavated in the 19th century complicate interpretation of the development of individual mounds (Jensen 2002:156). The observed stratigraphy has also been affected by later additions and truncations caused by secondary burials in the Iron Age, as well as by the aforementioned excavations. The deposits sloped downward towards the mound’s presumed outer edge (right in Fig. 12.3) and towards the mound’s interior (left in Fig.  12.3), suggesting that the entire mound area inside the bank had been emptied of soil. Micromorphology analyses from the thicker part of the bank showed that the bottommost mound fill (layer 3 in Fig.  12.3) contained latrine waste and other occupation waste; the deposit though not necessarily a farmyard midden sensu strictu, shares some characteristics with farmyard middens. The extent of the layer, coinciding with the remaining bank and placement of smaller stones at the edge of

232 

 B: Excavation Results 2011–12

Fig. 12.2: The remains of Flaghaug, visible as a semi-circular bank outside the cemetery wall. The cemetery wall cuts across the centre of the mound. The remaining bank indicates a diameter of approximately 34 m. Facing S-SW. Photo: MCH.

the deposit, indicates rather that it is a construction layer, and that the dumping of occupation waste to form the base of the mound may have been a convenient way of disposing of such deposits. Similarly, a likely ad hoc disposal of human waste in a Viking Age grave mound ditch at Heimdaljordet in Vestfold indicates a combination of planned and casual management of human waste at this site (Macphail et al. 2016:14). The occupation waste deposit in Flaghaug suggests occupation in the area from where the soil was gathered – possibly in the vicinity of the constructed mound (Macphail and Linderholm, Ch. 17:416). The mentioned row or cluster of small stones (Figs.  12.3–4) demarcating the outer border of the waste deposits may have been laid down during the initial stage of the mound construction in order to establish its intended shape and size; alternatively, it was placed there as a final demarcation of that construction stage. Too little of the stone row was exposed to confirm either hypothesis; more defined rows of stones observed in Kjellerhaug are discussed below. As the Flaghaug mound is believed to have been established in the early Bronze Age, it is likely that the occupation waste represents settlement in Site Period I (SP I) (approximately 2000–350 BC; Østmo and Bauer, Ch. 6). However, the exact extent of the mound in the Bronze Age is not known, raising at least the theoretical possibility

12 Østmo and Bauer: Grave Monuments 

 233

Profile 44023 1

2 3

5

4

Profile 44025 (mirrored)

1 2

3

4 Truncation

0

1m

Redeposited soil

02

Old ground surface

5

5

Mound fill: redeposited latrine and occupation waste

4

44

Mound fill: construction layer, possibly disturbed

3

02

Mound fill: grazed turf

2

44

1

3

Turf, infilling/redeposited soil, subsoil

Stone SAMPLES Micromorphology Macrofossil Soil chemistry Fig. 12.3: Sections through the north-eastern part of the Flaghaug bank. In Profile 44023, layer 1 seems to have eroded down the bank and built up over layer 5 after the 19th-century excavation, while the transition between layers 4 and 2 seems to be disturbed by truncation in Profile 44025. Illustration: I. T. Bøckman, MCH.

that this northern part of the mound was constructed during the SP III phase of secondary burials, providing a terminus ante quem date for the occupation. Another mound deposit (layer 1 in Fig. 12.3) was situated stratigraphically above the stone row and could suggest a later expansion of the mound. However, the two deposits are quite different, with the variation of building material more likely a reflection of the suitability for the mound construction or the availability of soil for building. The micromorphology analysis suggested that this mound material, layer 1, was composed of turf gathered from locally grazed pastures. Grazed turf seems to have been a commonly used construction material at Avaldsnes across site periods, judging from such use in the Kjellerhaug grave mound, as well as in the wall banks of

234 

 B: Excavation Results 2011–12

Fig. 12.4: Stone row delimiting the bottom-most mound fill of the grave monument. Facing S-SW. Photo: MCH.

boathouse A40 (Macphail and Linderholm, Ch. 17:379, 415–16; Bauer, Ch. 10:200–1). This is a recurring trait in South Scandinavian barrows. The use of such turf likely rendered large areas temporarily unsuitable for grazing or cultivation, thus representing a relatively costly investment in the monument (Jensen 2002:156). The turf deposit in Flaughaug was possibly compacted in a moist state (Macphail and Linderholm, Ch.  17:416). The use of moist sods or turfs is another trait known from Danish Bronze Age barrows. The use of moist turf may have been an aspect of the practice of constructing the mounds, possibly including deliberate moistening of the turf, which incidentally or intentionally may have produced anaerobic conditions that contributed to the preservation of organic materials in the monument (Holst et al. 2001:133–5; Jensen 2002:160–7). At Avaldsnes, this does not seem to have been consistent practice, and the moistness may alternatively stem from the wet local climate (Macphail and Linderholm, Ch.  17:416). The primary burial in Flaghaug was constructed with a cairn covering the central wooden chamber (for a reconstruction of the burials’ location within the mound, Stylegar and Reiersen, Fig. 22.3). Another deposit of varying thickness was observed between the layers discussed above. In the south-eastward-facing section, the layer was only 2–3 cm thick, whereas in the north-westward-facing section it was up to 25 cm (Fig. 12.3). In the east, Profile



12 Østmo and Bauer: Grave Monuments 

 235

44025, the transition to underlying layer 3 was sharply defined. Layer 2 may represent a younger disturbance in the north-eastern part of the mound that later was covered by excavated mound fill. It is impossible to conclude whether this disturbance relates to secondary burials in the north-eastern part of the mound, to plundering of the mound, or even to the 19th-century excavation trench (Stylegar and Reiersen, Ch. 22:556–66). The surface of the ground prior to the mound’s construction has been identified (layer 4 in Fig. 12.3). In the south-eastward-facing profile, this layer covered the mound fill as a redeposited version of the same layer, possibly as the result of the previous excavations. There were additional redeposited soils and some truncations visible in the profiles, demonstrating several digging events in the mound, probably in the 19th century. The original mound fill preserved in the excavated part of the bank make it probable that some original mound fill is preserved all the way around the semi-circular bank. Thus, it appears that the rector supervising the 1834–5 excavation choose to leave some of the original mound in place to preserve the monument’s original extent. If the bank preserved today is indeed the mound’s original extent, this means that it was approximately 34 metres in diameter (Stylegar and Reiersen, Ch. 22:552–4).

12.2 The Kjellerhaug grave mound The name Kjellerhaug derives from two earthen cellars (‘kjeller’) constructed within the mound in the 19th century (Rygh 1863:10). The mound has been reduced from its former stature by the cellars, and also by an electricity cable cutting through the mound, a stone-built fence, and plantings of several trees (1, Figs.  12.1 and 12.5). Paintings of Avaldsnes in the 18th century do however portray the Kjellerhaug mound as comparable in volume and prominence to the Flaghaug mound (Bauer, Fig. 15.3). No grave has been identified in the mound, although the crofter who constructed the cellar for the rectory was said to have become noticeably wealthy afterwards, hinting at an undisclosed find of some kind (Christie 1842:332). Stylegar et al. (2011) have suggested Kjellerhaug as the original context for two unique copper-alloy human masks fitted to an unknown object (2, Fig.  12.1). The masks are lost, but were documented in drawings and written notes by Lyder Sagen, first in 1803, and again with some adjustments in 1812 and 1830. Here, it is stated that the masks were found deep in the soil at Avaldsnes around 1800 and allegedly sent to Copenhagen shortly after (Stylegar et al. 2011:11–13). Similar stylistic traits may be seen in Celtic or Celtic-inspired metalwork from the pre-Roman Iron Age, insular masks such as those appearing on hanging bowls from the 8th century AD, and Scandinavian Viking Age masks such as those fitted on the sleighs in the Oseberg burial (Stylegar et al. 2011:16–19, 21–2).

236 

 B: Excavation Results 2011–12

0

Kjellerhaug

High-voltage cable

Stone fence

Buildings (modern)

Row of trees

Trench 1999

10 m

Trench 1999 approximate location

Fig. 12.5: Kjellerhaug with its known disturbances and construction elements prior to the ARM excavation. If the masks (Fig. 12.1) were indeed found in the Kjellerhaug mound, they were likely found close to the cellar construction observed in the north-eastern trench from 1999. Illustration: I. T. Bøckman, MCH.

Stylegar et al. (2011:13–14) argue that the generally shallow soils at Avaldsnes restrict soil depths of at least 1.2 metres (more than two ells) to wetland or a mound. The construction of the cellar in Kjellerhaug in roughly the same time period has led them to conclude that this was the origin of the masks. The line of arguments is convincing, despite the ARM excavations revealing that relatively deep deposits are to be found in several places across the farm, for example in the cultivation deposits in certain parts of Areas 1, 2, 3, and 10 (Bauer and Østmo, Fig. 5.7). While there are other possible locations for the find, the contemporaneity of the find with cellar construction together with the abrupt wealth that allegedly followed support such a connection. Prior to the ARM excavation, details on the mound construction were limited and the monument had not been dated with certainty. The 1999 survey identified remains of the stone-built cellar and the disturbed mound fill in the north-east part of the mound, while the remains of a seemingly undisturbed central cairn were revealed under the soil mantel in the south-western part of the mound (Fig. 12.5; Rønne 1999a). The modern installations were removed and measures were taken to restore some of the mound’s height (Rønne 1999a; Sjurseike 2001). Further investigation was still



12 Østmo and Bauer: Grave Monuments 

 237

needed to gather information of the construction and the monument’s date. This was the main aim of the ARM investigation, as the mound remained an important aspect of the visual and ritual monumentality of Avaldsnes. The ARM investigation targeted the south-western and south-eastern areas of the mound as they were most likely to contain undisturbed stratigraphy. Three trenches were opened, one in the south-western and one in the eastern parts of the mound; additionally, the survey trench from 2006 in the south-east was reopened. The first trench was only partly excavated; archaeological documentation was recorded, and probes were installed for a long-term monitoring of preservation of cultural deposits (Martens and Bergersen 2015). The remaining trenches were fully excavated and samples were taken from the sections through the mound to conduct analyses of palynology, macrofossils, micromorphology, and soil chemistry. In addition, the transition zone between the base of the mound and the adjacent plateau south of the mound was stripped of topsoil in order to examine stratigraphic relations between the mound and other archaeological features. Prior to the ARM excavations the Kjellerhaug mound was generally believed to be of an Iron Age date. However, Stylegar (et al. 2011:19) suggest that the masks may represent a secondary grave in a Bronze Age mound. The stratigraphic dating of the construction elements and radiocarbon dates from selected deposits support this suggestion. The examined construction deposits are not consistent throughout the mound and suggest that different parts may have been built using different techniques and deposits. Alternatively, these differences may represent different chronological stages in the monument related to secondary burials. As the radiocarbon dates generally have produced terminus post quem dates, it is not always possible to discern between these alternative explanations. The earliest phases of the Kjellerhaug mound are captured in the eastern and south-eastern sections of the mound (layer 1, Fig. 12.6). The eastern trench does not cut deeply towards the centre of the mound, but rather intersects the outer brim (Fig. 12.6, Profiles 45490 and 45494). Here, closest to the centre of the mound, turf formed part of a compact mixed fill including burned byre waste, anthropogenic soil, and sub-soil (Fig.  12.6, layer 1). The underlying ground surface also indicates trampling by animals, with inclusions of byre waste and dung reflecting stocking of animals prior to the building of the mound. While turf is present in this mound, it does not share the characteristics of the compacted moist, grazed turf observed in Flaghaug. Furthermore, compared to Flaghaug and younger mound construction deposit in Kjellerhaug, the included turf in layer 1, as well as in layer 2, reflects local occupation rather than grasslands (Macphail and Linderholm, Ch. 17:415–16; Macphail pers. comm.).It should be noted that the samples for micromorphology analysis were taken from the outer edge of construction layer 1 and that internal variations within the mound construction may be reflected in particular samples (Fig.  12.6, Profiles 45490 and 45494). A charcoal lens within construction layer 1 gave a radiocarbon date to 1733–1632 BC (Beta-333050) providing a terminus post quem date for the mound. A

238 

 B: Excavation Results 2011–12

Profile 45494 (mirrored) 1 4

4 45 549 49 0 4 45

3

46

2 45

45

4

46

45 43 7

46

6

2

Profile 45490

1

2

3

4

Charcoal lens Beta-333050

BC 1733–1632

Profile 45466

Profile 45464 5

13

9

2

Profile 45462

4 6

7

12 Beta-333052

8

9

10

BC 794–568

12

Profile 45437

11

1

Mound fill: mixed fill containing turf and burned byre waste

2

Mound fill: humic soil with turf inclusions

3 4

Mound fill: as 5, but mixed with fragmented grassland turf Mound fill

5

Mound fill: redeposited subsoil

6

Turf formation

7

Charcoal deposit

8

Charcoal deposit

0

9

Mound fill, inbetween stones

Stone

10

Mound fill

Tree root

11

Mound fill, inbetween stones

12

Trampled surface

13

Colluvial deposit Old ground surface, animal trampled coluuvium Bedrock Turf, cultivation deposit, disturbance infilling, subsoil

SAMPLES Micromorphology Macrofossil Pollen Soil chemistry

Fig. 12.6: Kjellerhaug located on the edge of the settlement plateau, facing the strait to the east (Fig. 12.1). Stratigraphic build-up of deposits and placement of relevant samples are documented in several sections. Location of excavated trenches and documented sections are delineated. Illustration: I. T. Bøckman, MCH.

1m



12 Østmo and Bauer: Grave Monuments 

 239

row of stones of various sizes that seem to follow the edge of layer 1 (Profile 45490, Fig.  12.6) may have had a function similar to that of the row of stones observed in Flaghaug, marking what may have been the outer border of the mound at what seems to be the primary stage. A second construction layer is visible in four sections in the eastern and south-eastern areas of the mound (layer 2, Fig.  12.6). It cannot be determined with certainty whether this later addition is part of constructional elements within a mound built in a continuous sequence or whether it represents a later construction stage in the monument’s history. However, it is interesting to note that secondary burials in Danish barrows often are placed south or east of the primary burial (Holst 2013a:64–6). The second construction segment, including fill and stone row, was uncovered south-east of the boundary of the primary monument, extending the monument approximately three metres towards the south and east. This extension may well represent a secondary monument stage, though no burial related to such a secondary stage was uncovered during the excavation. Similarly to the mound fill in the earliest stage, the outer border of layer 2 was marked by a belt of smaller stones (layer 9, Figs. 12.6–7). The underlying trampled surface (layer 12, Fig. 12.6) contains inclusions of remains from early land clearance around 2118–1961 BC (Beta-333051; Bauer and Østmo, Ch. 8:239) and does not contribute to a narrower timeframe than that provided through stratigraphic relation to the earliest construction stage (layer 1). A charcoal-rich inclusion overlying layer 2 provides a terminus ante quem date to 794–568 BC (Beta-333052), consequently dating both construction phases to the period 1733–568 BC. A few sherds (S12772/65) found on the trampled surface beneath the mound have a general date to the Bronze Age and likewise do not narrow down the wide timeframe (Kristoffersen and Hauken, Ch. 21:529). While the burnt bones, cereals, and other food plant inclusions in layer 7 could imply offerings in the mound, the micromorphology provides a divergent interpretation based on inclusions of latrine waste (Macphail and Linderholm, Ch. 17:415; Ballantyne et al., Ch. 19:470, 498). Such material would imply settlement-related midden deposits rather than offerings. The outer belt of stones and the charcoal-rich deposits appears to have been truncated at some point and subsequently back-filled with a sandy construction layer (layers 9, 8, 7, and 5, respectively, Fig. 12.6). Within the yellow sandy silt, quite resembling the natural C-horizon in the area, a thin turf formation (layers 5 and 6, respectively, Fig. 12.6) indicates a period, possibly as long as a decade, where grass-turf was allowed to reform before additional sandy silt, quite similar in character, was deposited over the turf horizon (Macphail and Linderholm, Ch. 17:415). These sandy deposits represent additional construction stages, and unlike the earlier phases with no observation of the time-indicating pedogenic or biological processes accounted for by Holst (2013b:234–40), the turf formation shows that the construction, at least at this stage, was discontinuous. These sandy layers have not been dated directly, but the stratigraphic relation between the stone-built palisade base A20 and the mound provides a terminus ante quem date to around AD 600 (Østmo, Ch. 11:217–

240 

 B: Excavation Results 2011–12

Fig. 12.7: The row of stones marking the outer border of the second construction stage in Kjellerhaug. Top: After removal of top soil, (layers 9, 5, and 6 visible). Middle: After removal of younger deposits and upper stones (bottom stones and layer 2 visible). Bottom: Belt of stones documented in section. See Fig. 12.6 for numbering of deposits. All photos facing N-NE. Photo: MCH.



12 Østmo and Bauer: Grave Monuments 

 241

18). This relation is not captured in the sections, but some of the northernmost stones of A20 abutted the mound’s south-eastern edge (Østmo, Fig. 11.2). Micromorphology analyses of layer 4 indicate farmyard midden deposits with latrine waste inclusions similar to those observed in the waste layers immediately underlying A20 further south (Østmo, Fig. 11.4; Macphail and Linderholm, Ch. 17:406). In addition, the trampled or cultivated colluvial deposit A25600 with a barley grain dated to the last two centuries BC (Ua-45354) had been building up over the stone border of the mound (layers 13 and 9 respectively, Fig. 12.6). As A25600 seems to have been affected and reformed parallel to the production activities of SP III, this does not permit any narrowing down of the timeframe for these mound construction elements. The cairn observed under the soil mantle in the 1999 survey was not present in the areas excavated in eastern and south-eastern parts of the mound, but is likely identical to the stone packing uncovered in the western trench and in the transition between the southern part of the mound and the adjacent plateau (Figs.  12.6 and 12.8). A piece of plano-convex slag was uncovered underneath the stone packing (Østmo, Fig. 9.11). The assumed connection to the SP III metal work on the adjacent plateau provides a terminus post quem date for the construction of the stone packing to AD 200. It is unclear how this construction phase relates to the phases observed in south-eastern and eastern parts of the mound. These limited excavations have revealed that the Kjellerhaug mound is a complex monument. The temporal relation between some of the construction stages is not clear, and some elements may be close to contemporary and rather relate to how the mound was constructed while others clearly represent separate chronological stages in the mound’s history. The lack of diagnostic finds and the scarcity of deposits containing representative datable material has produced a rather coarse chronology, based mainly on combinations of terminus post quem and terminus ante quem dates. The Kjellerhaug and Flaghaug mounds share some characteristics with other Bronze Age mounds in the region as well as with contemporary barrows in southern Scandinavia. These traits include construction with grazed turf and a row or belt of smaller stones marking the outer border of mound construction deposits. Furthermore, the presence of secondary burials seems to be quite typical in the large mounds from the Bronze Age, as is addressed in the discussion below. It is clear that the numerous construction elements seen in Kjellerhaug may represent several secondary phases during which the monument was enlarged or adjusted, likely contemporary with a secondary burial. The investigated sections and surfaces cannot be assumed to provide a full history of the mound. Rather, the many variations within the documented sections indicate that it is highly likely that other secondary phases would be uncovered in other parts of the mound, providing ample possible contexts for the copper masks.

242 

 B: Excavation Results 2011–12

Fig. 12.8: The stone packing uncovered in the southern (top) and the south-western (bottom) edge of the Kjellerhaug mound. Photo: MCH.



12 Østmo and Bauer: Grave Monuments 

 243

12.3 ‘Jomfru Marias synål’ and other raised stones Immediately north of St Óláfr’s Church’s northern nave wall is a 7.2  m high raised stone, known as the Virgin Mary’s Sewing Needle (‘Jomfru Marias synål’) (14, Fig. 12.1). It leans towards the church wall, and the upper end has marks from trimming, probably due to local tradition prophesying the advent of Doomsday should the stone touch the wall (Hernæs 1997:206). During the restoration of St Óláfr’s Church in the late 1830s, a rather large piece was cut off the stone on the side closest to the church (Skadberg 1950:20). Skre (Ch. 23:654–9) has examined the evidence for additional stones and their possible location, arguing that the ‘Needle’ is a component of a triangular raised-stone burial monument, albeit with a different layout from that previously suggested by Hernæs (1997:212, 19, 22). The raised stone monument likely dates to the first five centuries AD, which could possibly be narrowed down to the 3rd century AD. The second of the presumed three stones has also been located. Because the medieval church was constructed over the area in which a grave centrally placed within the monument could be expected, the details of a possible burial are lost. See Skre (Ch. 23) for a comprehensive discussion of the monument.

12.4 The Kongshaug graves At the elevated ridge called Kongshaug, situated west of the settlement plateau (Fig.  12.1), earlier surveys have revealed several grave monuments (Christie 1842; Petersen 1934; Hafsaas 2005; 2006). The mid-19th century written sources (Nicolaysen 1862–6:346, 806; Rygh 1863:10) describe an inconspicuous mound with a central raised stone on the ridge. The raised stone is said to have been torn down in the 1860s (15, Fig. 12.1). Other recorded graves found in northern and southern clusters on Kongshaug have been superficially investigated, and only one was fully excavated in 2005 and 2006. The following description will present the graves in each cluster in chronological order as far as their date ranges permit. The circular stone packing (16, Fig. 12.1) on the southern end of the ridge described in the mid-19th century seems to be identical with a stone packing observed by Jan Petersen in 1934, which was covered by soil to facilitate cultivation on the ridge (Christie 1842:332; Rygh 1863:10–11; Petersen 1934). This stone packing was also observed in the 2004 geophysical survey, in the 2005 archaeological survey, and in a small excavation in 2006 (Sandnes and Eide 2004:fig. 10; Hafsaas 2005:12–13; 2006). The grave monument is not visible today as it has been covered by soil. A simple iron sword is said to have been removed from this monument in 1841 (Christie 1842:332; Nicolaysen 1862–6:346). The monument was semi-circular, measuring 20 by 17 metres. The primary burial in the stone packing, presumably centrally located, has not been identified.

244 

 B: Excavation Results 2011–12

Hafsaas (2005:14; 2006) dates the stone packing to the first four centuries AD based on dating of similar monuments, in particular a quite similar stone packing from Time in Rogaland that contained several burials. Multiple burial events are also found in the Kongshaug stone packing: a cremation grave (17, Fig. 12.1) was identified along the southern edge of the stone packing, underneath the two outermost stones that were removed under excavation. The bones had been cremated elsewhere and deposited without waste from the cremation pyre in a 2 × 6 m area. As the bones have not been subject to osteological analysis, it has not been confirmed whether they are human; the 2006 excavation could not ascertain whether the bone deposit continued further underneath the monument or was located mainly outside of the monument. In the latter case, it is possible that the bones represent offerings at the burial site, rather than a secondary burial. Further examination is necessary to arrive at a conclusion on the matter. The bones have not been radiocarbon dated. While cremation was more common in the Roman Iron Age, the practice was maintained throughout the first millennium AD (Solberg 2000:76, 135, 186, 223). In the western part of the stone packing, a plundered inhumation grave was identified (19, Fig. 12.1). The artefacts from the burial included ten glass beads, three ceramic sherds, and a Mesolithic pecked round-butted greenstone axe. Hafsaas (2005:16–17) has also suggested that the simple iron sword found in 1841 originated from this secondary burial and that it may actually have been a weaver’s batten. Based on the beads and batten, the grave may date to the period AD 200–1050. A 1.81 g gold ingot (S12222a) was retrieved near the grave, albeit within the redeposited cultivation soil added when facilitating for cultivation at Kongshaug in the 19th century (Reiersen 2009:42, 71; Zachrisson, Ch. 25:701–3). The northern cluster of Kongshaug graves comprises a small cairn ten metres in diameter and two graves without visual aboveground markers. One of the latter was a boat grave, identified by the presence of nails, some of which formed lines indicating the shape of the boat (Hafsaas 2005:10–11). These three graves (20, 21, and 22, Fig.  12.1) were only partially exposed and only documented as they appeared after removal of topsoil and not examined further. Hafsaas (2005:10–11) suggests that the two latter graves are secondary burials in the cairn, though there is little support for this in the documentation of the graves’ internal stratigraphic or spatial relations. The cairn cannot be precisely dated based on morphology, but is likely from the first millennium AD. Boat burials were a practice initiated in the Merovingian Period and peaking in the Viking Age, indicating that this burial is from the 7th–10th centuries (Solberg 2000:186–7, 222–3). The second grave was partly uncovered immediately south of the boat grave. Besides its proximity to and the presence of fill resembling that of the boat grave, there is little information available as no artefacts were retrieved; while apparently of rectangular or semi-rectangular shape, neither end of the grave could be identified within the trench. However, there are no indications that this was a cremation grave. As inhumation graves are increasingly common from the 3rd century onwards, par-



12 Østmo and Bauer: Grave Monuments 

 245

ticularly in the Merovingian Period, the possible date range is wide (Solberg 2003:77, 135, 186). One curious find from Kongshaug is a gold pendant shaped as a thimble or basket (S6810; 12, Fig. 12.1). The pendant likely originates from a high-status female burial from the 3rd century, but was found out of context, and the exact location of the grave is uncertain (Reiersen 2009:37, 64–6; Zachrisson, Ch. 25:702).

12.5 Other graves from the Avaldsnes headland East of Area 6 lay three circular stone packings (9, 10, and 11, Fig. 12.1) with diameters between 0.9 and 1.8 metres covering shallow pits (1000°C) silicate slags imply ironworking (as corroborated by archaeological finds), while glassy slags suggest other artisanal work employing ‘ashey’ fluxes. Mapping possibly identified two 4 × 4 m and 4 × 3 m work spaces/buildings within the complex. One specific feature in Area 6 (Profile 45418) in an area of hearths attracted special scrutiny: the dark, very fine charcoal-rich ‘wet’ fill showed high MS levels and very high phosphate concentrations, with organic inclusions containing up to 19.2 % P (SEM/EDS). EDS found the presence of sulphur; a single organic chemistry sample found lipids, indicating the presence of fats of terrestrial (but not marine) animals. In addition, other extracts showed that these fats had not been heated, and that herbivore faecal waste had been introduced. These equivocal findings, along with high magnetic susceptibility readings throughout and very high phosphate concentrations, may suggest a fire installation, albeit one more likely associated with the later decomposition of meat and perhaps a later use of the space for stocking or stabling of herbivores. Roman Iron Age/Migration Period boathouse A40 was constructed from local pasture turf. Buried occupation layers indicated that the boathouse floor was near sea level, and that episodic high tides buried trampled soil floors with sands and gravel. These floors included wood fragments, possibly indicative of woodworking/ boat maintenance. Speculatively, anomalous iron staining in the boathouse deposits may stem from the effects of saltwater drip on rusty iron fittings. By contrast, the post-medieval boathouse A46 (Area 10) was characterised by a midden spread, rich in latrine and non-calcareous ash waste. Enigmatic inclusions did not have the microchemistry of ‘tar’, and were tentatively interpreted as possible ‘paint’ or pigment. Interestingly, the soils buried by the Kjellerhaug grave mound record grazing and stock management from the early Bronze Age, which is consistent with interpretations based on pollen analysis in Area 2. Although pasture turf was a major component in the mound, some layers are dominated by farm-settlement disposal, rich in charcoal and latrine waste. The Roman Iron Age Flaghaug grave mound also includes a make-up that involved the disposal of midden waste, some of which is very high in calcium-iron-phosphate compounds that also include sulphur (S) and fluorine (F). Lastly, a high-medieval house-floor deposit was preserved in Area 1. This essentially domestic, beaten floor included human coprolitic remains as well as evidence of hightemperature activities: strongly heated rock fragments and charcoal slag (fuel ash). Some overall findings can be suggested for the Avaldsnes settlement. The phosphate box plots and distribution maps, as well as punctuated soil micromorphology

420 

 C: Scientific Analyses 2011–12

samples, identify that the overall chemical impact of cultural activities reflects a settlement of significance; at least in the respect of an overall input of nutrients. This seems to be a response to intensive use and a significant population, which in turn is a consequence of local political decisions and actions at the Avaldsnes site during various phases of use. The combined studies certainly indicate a focus on grazing and animal management associated with cultivation employing manuring beginning perhaps in the early Bronze Age. Occupation seems to have intensified in the Iron Age and Migration Periods, with chemical signals resulting from evident latrine-waste disposal, while magnetic susceptibility maps clearly show both hearths and industrial furnaces in action. Patchy Viking Age and medieval results suggest the continuation of these intensive activities. Despite the mixed chronologies of many of the studied areas, it is clear that the combined soil and pollen investigations of mapped areas and soil profiles have yielded much spatial and period-specific characterisations of the settlement’s morphology over time. In addition, the importance of this site is evident from the findings of phosphate accumulation as well as other manifestations of long-term occupation – that is, from the results of intensive use over a coherent period of more than two millennia. Phosphate amounts and concentrations alone also suggest an intensity of activity that probably reflects a site of great importance compared to rural settlements studied elsewhere in Norway, for example along the E18 Gulli–Langåker corridor. Acknowledgements: The authors gratefully acknowledge the input of the ARM excavation and project team, especially in terms of supplying samples, site information, and discussion; particular thanks are due to Egil Lindhart Bauer, Mari Arentz Østmo, and Dagfinn Skre of MCH, University of Oslo and Rebecca Cannell of the University of Bournemouth. Laboratory assistance and data were kindly supplied by Jan-Erik Wallin (pollen analysis), Fredrik Olsson and Philip Jerand (GIS work; at the Environmental Archaeology Laboratory (Umeå University), Sven Isaksson (organic chemistry; Stockholm University) and Kevin Reeves (University College London). Additionally, Roger Engelmark is acknowledged for valuable discussions.

Rebecca J. S. Cannell, Paul N. Cheetham and Kate Welham

18 Geochemical analysis using portable X-ray fluorescence This chapter presents the results from geochemical analyses of soil samples from Area 6, where remains from production and a palisade were identified (Østmo, Ch. 9). The samples were taken during the 2011–12 excavation campaign, and later analysed using portable X-ray fluorescence (pXRF) to obtain the elemental composition. The geochemical data from the samples was then statistically treated using principal component analysis to simplify the data volume. This reduced the elemental quantities into activity types, which could be further refined to areas and, from selected cores taken, by stratigraphic layer. The analytical results suggest that several different activities occurred in Area 6 in the late Iron Age and early Middle Ages. In addition to samples taken on a regular grid, an oven feature was sampled for comparison, and to aid interpretation of its function. The sampled oven feature, dated to early Site Period V, appears to have been used for organic processes such as corn-drying rather than metalworking. In the same stratigraphic phase, food-processing, potentially involving seaweed ash or a by-product of this process, was identified. To the west of this feature, copper alloy–working on a small scale possibly occurred during the Viking Age. Prior to this, perhaps as early as the Roman Iron Age, the site had a different function, one that might have involved the addition of organic waste such as dung collected as a by-product of animal stocking, or more likely the middening of organic waste. The waste management and use of space appears delineated; despite the small area, clear clustering patterns appear. As portable XRF technology and its application – particularly to soils in archaeology – are relatively new, its suitability warrants debate. This chapter therefore also contains a short review of the use of pXRF in archaeology.

Studying use of space in the past offers great insights into past cultural and social norms, resource use, and economy, as well as potentially revealing ritual or culturally motivated practices. Space is active; it is much more than the sum of objects that compose it. Social space constantly imposes and engages with those that inhabit, construct, and mentally define that space within their own cultural and individual ontologies. Space is both passive and active in everyday life. Interiors, exteriors, open spaces and landscapes are invariably intertwined with individual perceptions of the social and physical spaces they occupy (Lefebvre 1991:410–12; Hem Eriksen 2015:38– 43). It is therefore important that archaeology not only document the archaeological features such as walls or postholes, but also investigate what happened between, within, and without the structured space. How people act within a space, and how space is defined and reproduced, can reveal social structures relating to power and hierarchy as well as the economy of a settlement or area. Thus, the internal spaces within a house or external spaces of a settlement, perhaps defined either by proximity to archaeological features or by landscape constraints and perceptions, are tools we can use to understand the inhabitants of that particular space. In practice, though, social spaces are often represented by nothing more than an apparently featureless soil and a set of artefacts deposited within that soil. The

422 

 C: Scientific Analyses 2011–12

most common means for the archaeologist to understand space is to look at the distribution of artefacts, and relate these to other sources of information such as previous excavation data, documentary sources, or archaeological features. This can be highly productive; however, what remains is often fragmented and disturbed, and as some materials are more durable than others, the surviving objects are unlikely to be representative of the past human activities we seek. Archaeological geochemistry is a developing technique that aims to access past human activities more directly by looking at the surface on which the activities took place. This is achieved by measuring the major and trace chemical remains within soils and sediments. As people live, they produce waste on all scales; with repeated activity, waste becomes an increasingly significant component of the soils and sediments where the activity took place. Over time, these inputs are subject to a plethora of forces that break down and disturb the traces. Yet one hundred years of single- and multi-element research suggest that the chemically significant remains of past activities can be measurable in the present day. These chemical remains can be measured and employed to better understand the range of activities that occurred on a site, and where they most frequently took place. This article presents the results from sampling a well-preserved, archaeologically complex area of the site for multi-elemental geochemical analysis. The results, in combination with the other scientific results and the archaeological evidence, further enhance our understanding of the use of space and the activities that have occurred on the site over time. The nature and structure of past activities can then be related to the overall ARM Project objectives of understanding the social, political, and economic structure of royal manor sites in the period of kingdom formation in Norway (Skre, Ch. 4).

18.1 Portable XRF in archaeology The chosen method for analysis was portable X-ray fluorescence (pXRF). X-ray fluorescence is a method for identifying the elemental composition of an object or substance. It works by using high energy X-ray photons to excite an electron in the K or L shell of an atom. Since the excitation source emits slightly more energy than the binding energy of the inner K (or L) shell electron, this electron leaves the atom. Electrons shift down shells within the atom, so that an L (or M) shell electron replaces the lost K (or L) shell electron. Outer shell electrons have higher energy states than inner shell electrons. Consequently, as they replace the lost electron they release secondary energy: X-ray fluorescence. The energy released is element specific and known. Non handheld (i.  e. laboratory based) X-ray fluorescence instruments have been applied to archaeological research for over fifty years; although not the most common instrument used for soil analysis, XRF has been repeatedly and successfully applied (e.  g., Shackley 2011a; Charlton 2013; Frahm and Doonan 2013).



18 Cannell et al.: Geochemical analysis using pXRF 

 423

The first truly portable, handheld instrument on the commercial market was produced in 1994 by Thermo Scientific/Niton1. The downsizing of instruments from bench-top to handheld was the result of the development of smaller components (Pantazis et al. 2010). Alongside this advancement, the improvement in detectors (e.  g., the silicon drift detector, SDD) and fundamental parameters (FP) software have transformed the size, ease of use, and accuracy of XRF technology (Liritzis and Zacharias 2011). While ease of use and improved software calibration do not eliminate the need for empirical calibration and the use of recognised international standards (Shackley 2011b), they remove the exclusivity of the technology from highly specialised laboratories (Frahm and Doonan 2013). Improvements in pXRF have only recently come to the point where they are being widely applied to archaeological materials as a quicker and cheaper means of multi-elemental analysis. The majority of published studies in archaeology relate to obsidian sourcing (published examples: Phillips and Speakman 2009; Craig et al. 2010; Shackley 2010; 2011b; Speakman et al. 2011; Frahm 2013; Frahm et al. 2013; Frahm and Doonan 2013; Frahm and Feinberg 2013; Speakman and Shackley 2013). Aside from obsidian, pXRF is also readily applied to ceramics, glass, paints, and metal artefacts, as well as a range of lithic types. Provenance studies form the vast majority of published work (examples including stone provenance, paints, pigments, and glazes in ceramic and glass can be found in Pappalardo et al. 2004; Oikonomou et al. 2008; Bonizzoni et al. 2011; Forster et al. 2011; Liu et al. 2011; Frankel and Webb 2012; Liu et al. 2012; Tantrakarn et al. 2012; Barbera et al. 2013). As the metal composition of historic and prehistoric artefacts features considerable variation due to differences in raw materials and production (sometimes rendering unnecessary the highest possible precision in analysis), pXRF is highly suited to provenance and composition analysis of metals (Karydas et al. 2004; Gliozzo et al. 2011; Heginbotham et al. 2011; Martinón-Torres et al. 2012). Aside from the field of provenance analysis, few studies have been published in the field of archaeology using truly portable XRF (i.  e. handheld). There is, however, considerable potential in using pXRF on archaeological and historical sites to improve contextual interpretation and sampling, as demonstrated by the in-situ analysis undertaken by Donais and George (2012). Potential applications of pXRF include in-situ conservation as well as researching past activities that do not leave clear artefactual or other physical evidence and further refining the understanding of those that do. Portable XRF was chosen for this study as part of a larger research project into the use of pXRF on archaeological soils and as a method that allowed a more flexible sampling and analytical approach. Using pXRF, cores could be analysed directly and non-destructively, while samples taken from a defined surface could be processed rapidly when compared to the sample preparation requirements for more traditional

1 See http://www.niton.com/en/portable-xrf-technology.

424 

 C: Scientific Analyses 2011–12

instrumental choices such as ICP-MS or ICP-AES (inductively coupled plasma mass spectrometry or inductively coupled plasma atomic emission spectrometry). The speed of analysis means pXRF is more affordable, allowing for a greater sample density. Furthermore, as it is non-destructive, samples can be repeatedly analysed or stored for future research.

18.2 Archaeological geochemistry As mentioned in the introduction, the aim of this research is to measure anthropogenic enhancement in soils and sediments, which can be used to define past settlement and distinguish functional areas within structures and sites. Many factors inherent to the soil affect the retention of major and trace elements from past activities. For instance, the pathways for elemental retention vary according to the soil’s pH, particle size and type (e.  g., type of clay, proportion of sesquioxides), and organic content. Other factors include the recycling of nutrients by plants, later land use, sample depth, and soil processes over time (Rimmington 1998). As the retention mechanisms are element-specific and dependent on their physicochemical form, even in near-identical conditions, relative and absolute elemental retention will vary. To further complicate the matter, the mechanisms are not completely understood for every element (Tack 2010:10–11). Not all of the factors will be covered here in depth, but they are discussed to some degree in recent review papers (Oonk et al. 2009a; 2009b; Wilson et al. 2009; Walkington 2010) and other published case studies (e.  g., Jones et al. 2010; Luzzadder-Beach et al. 2011; Vyncke et al. 2011). Due to the potential uncertainties in the interpretation of geochemical data, archaeological geochemistry is best undertaken where additional data is available, whether from geophysical prospection or excavation, and is employed most effectively when addressing suitable and specific research questions.

18.2.1 Accuracy using pXRF on soils The compromise in using pXRF is its reduced accuracy; whilst the instrument can detect to parts per million for all but the lightest elements, modern ICP instruments can achieve parts per billion for the majority of elements. However, recent research comparing techniques on soil samples has shown a strong correlation and thus agreement between the results from pXRF and ICP-AES (Schneider et al. 2015). For this research, it was deemed that accuracy on the order of parts per billion would not improve or enhance interpretations in archaeological soils, as elements present in such ultra-low quantities cannot be used to clearly define past activities in this context. Therefore, portable XRF was chosen as an economic and flexible technology that could produce results relevant to the ARM Project’s research aims.



18 Cannell et al.: Geochemical analysis using pXRF 

 425

The general consensus holds that pXRF instruments are internally stable and precise to the degree sufficient for answering archaeological questions pertaining to provenance studies (Nazaroff et al. 2010; Shackley 2010; Goodale et al. 2012). Instrument precision and accuracy is easily measured and reconciled with the correct use of certified reference standards. Accuracy in situ with unprocessed samples can be problematic due to factors such as moisture content, porosity, and uneven surface geometry. These factors, especially moisture content, disproportionately affect results for lighter elements (Schneider et al. 2015). Similar problems can also occur with ex-situ analysis, which can be considerably alleviated and improved by simple sample processing. In archaeological geochemistry, it is the intra-site variability that is interpreted, rather than the absolute values; each site has a unique combination of environmental and anthropogenic influences (Bethell and Smith 1989). Therefore, as the instrument is internally stable and precise, intra-site analysis of data from sediment analysis in conjunction with the correct sample treatment could provide a means of interpreting past activity types and distribution.

18.3 Avaldsnes The archaeological and historical significance of the Avaldsnes site is extensively discussed in this volume by engaging previous research, historical texts, and not least the campaign of excavation and research undertaken by the ARM Project. The project’s overarching aim was to deepen archaeological understanding of early royal power at Avaldsnes in late Iron Age Norway. This includes its economic and military roles and its position as a centre for the manufacture and trade of high-status items as well as other primary economic resources, such as food (Skre, Ch. 4). The potential for geochemical analysis to provide evidence to forward these research aims was identified prior to excavation as a method capable of measuring and locating production and storage activities, particularly in situations where the physical remains available to archaeologists are scant or ephemeral. Thus, as a technique complementary to the other scientific and traditional methods employed during the excavation and the research project as a whole, geochemical analysis can provide supporting evidence for the interpretation of excavated archaeological features. The geochemical sampling for multi-elemental analysis at the site was focused on Area 6, which was excavated in 2012. This area was selected for practical and archaeological reasons. The silty loam cambric subsoil appeared consistent on the selected archaeologically defined surface. The revealed archaeology was complex and multiphase; however there was no immediately visible structure or pattern, such as a clearly defined house as discovered in other areas, complicating efforts at interpretation. Modern disturbance presented a constant challenge during the Avaldsnes excavations. In Area 6, although modern intrusions such as cable trenches are present,

426 

 C: Scientific Analyses 2011–12

these were nevertheless isolated and appeared defined, hence sampling could potentially avoid these areas. Area 6 can be physically defined. A bedrock scarp, formed by changing geology from green schist to metaphoric sandstone, limits the area to the east and south; the scarp allows a vista over the land and coast to the east, toward Bukkøya, Karmsund, and the mainland. Modern disturbance, buildings, and raised bedrock limited the available surface to the immediate south and west, whilst to the north the substantial Kjellerhaug burial mound formed another boundary, delimiting an area for study. Of course, it cannot be assumed these defined boundaries had the same, or indeed any, relevance in the past. During the excavation of Area 6, eighty-five postholes were recorded, along with several oven features, cooking pits, stakeholes, and stone settings. There is clearly a multi-period aspect to all areas of the site: according to 14C results, Area 6 was occupied from 300 BC to AD 1160, excluding the modern disturbances. Hence, in Area 6 there were not one, but several spaces representing different phases of use. The changing use of the geographically constrained space, as well as the activities represented within it, could be highly enlightening, in order to understand the role of the site throughout the period as a centre for royal power. The majority of the dates relating to the area sampled for this analysis fall within the Viking Age. Although a wall ditch, ovens, numerous postholes and a substantial stone-built feature were excavated, no clear pattern related to buildings could be identified from the archaeological features. Geochemical results could potentially define features and specific activity areas, as well as detecting the function of oven features. For example, the purpose and use of an oven, such as for metal working, domestic activities, or large-scale food processing, can have significant implications for the economic and social structure of the site. This was demonstrated for the Silchester site by Cook (et al. 2010), who analysed samples from hearths and furnaces using laboratory-based XRF in order to specify hearth and oven function to the metals worked or domestic activity. Within buildings, ‘zones’ or ‘functional areas’ can be chemically defined in environments as varied as Neolithic Orkney to Hellenistic Turkey, and could therefore help define the use of space in Area 6 (Jones et al. 2010; Vyncke et al. 2011).

18.4 Methods 18.4.1 Sampling Two approaches to sampling were employed at Avaldsnes: coring to investigate vertical stratigraphy, and gridded sampling to take small soil and sediment samples from horizontal, archaeologically defined layers.



18 Cannell et al.: Geochemical analysis using pXRF 

 427

Coring (as opposed to augering) as a sample method has not received extensive discussion in archaeology, and is very rarely undertaken on the scale employed at Avaldsnes. In archaeology, coring is predominantly used for shallow submerged sites (e.  g., Horlings 2013), pollen sampling (Ghilardi and O’Connell 2013), lake sediments for environmental reconstruction (Støren et al. 2008), and, above all, in alluvial landscapes (Passmore et al. 2002; Brown 2009). The method has a longer history of application in archaeology in the U.S. than in Europe, but has become more common globally in recent years (Canti and Meddens 1998). There are many examples of augers used for soil sampling without mention of the specifics (e.  g., Salisbury 2012), but as noted by Gauss (et al. 2013), augers frequently lose soil as they are drawn up, sample sizes can be very small, and fine stratigraphy can be blurred or lost. A great deal of information regarding soil- and site-formation processes is held in the layer interfaces, which are obviously lost in the strip-and-map method of excavation. Maintained sections can be far apart and therefore unrepresentative of the site as a whole. Taking undisturbed cores prior to excavation has the dual benefit of providing prospection information as well as full soil profiles and sample material, which, if the core locations are surveyed precisely, can be related to subsequent excavation results. The disadvantages inherent in this method are primarily the time it requires to achieve significant depths and the disturbance it potentially causes – for instance, creating an oxidising area in anaerobic soils, which can be detrimental to preservation in the area around the core. This is a concern if the area is left unexcavated or there is a significant time lapse between coring and excavation. In this study, this occurred in only a small section of Area 2 and relating to four cores; the majority of the cored areas were either well-drained, and therefore oxidised, or rapidly excavated. Another disadvantage to coring is the potential for hitting and damaging an archaeologically significant object, deposit, or feature. Similar damage occurs, however, when excavation is conducted with mechanical excavators, shovels, and spades; damage is a risk regularly taken when balancing the time constraints against the information potential. As the cores taken at Avaldsnes were 5  cm in diameter and therefore liable to disturb only a very small area, the risk was deemed offset by the potential for information; upon examining the cores, the only archaeological finds were two very small slag fragments, from a total of 364 cores. In applicable areas, cores were taken prior to excavation and in reference to ground penetrating radar (GPR) data provided by a survey conducted by the Vienna Institute for Archaeological Science (VIAS; Stamnes and Bauer, Ch. 16:328–9). Core locations were surveyed using a total station theodolite2. Cores confirmed the depth of the archaeological deposits and provided valuable site formation, environmental details, and material for later subsampling. The corer used was a Van Walt liner

2 Trimble S3.

428 

 C: Scientific Analyses 2011–12

sampler set (manufactured by Eijkelkamp) for hard soils, which takes undisturbed samples measuring 300  × 49  mm in transparent, single-use plastic liners. The distortion of the stratigraphy and compaction to the soil is minimal in most conditions, although topsoil can be slightly compressed due to the higher organic content and porous structure. The authors have also observed compression in waterlogged clays and histic (peaty) horizons; however, these factors were minimal at Avaldsnes. The maximum depth achieved was 2.35 m, although the majority of cores were less than 1 m in depth. Over the 2011 and 2012 seasons, 364 cores were taken (200 in 2011, 164 in 2012), many of which were later recorded prior to subsampling and analysis by Dr. J. Linderholm at MAL (Macphail and Linderholm, Ch. 17). Additionally, two deep cores provided background reference samples for geochemical analysis presented in this chapter, and four cores from Area 6 were subsampled to test whether doing so would add temporarily to the geochemical analysis. For in-situ and ex-situ observations and analytical results from the cores, data analysis and visualisation software3 was used to create interpolated surfaces and sections from the cores to check for anomalies and stratigraphic integrity. This created a flexible, integral database for later reference. The majority of the data presented in this chapter was taken on a horizontal grid set out over an excavated horizon. Specifically for Area 6, geochemical samples were taken from an archaeologically defined layer. Samples were taken using a predefined 1  × 1  m grid system at the end of the excavation season to minimise disruption to the excavation of archaeological features. Where present, samples were taken from the base of archaeologically defined layer 25600, others from the immediate subsoil. These layers are pedologically very similar. For each sample point, observations of the soil and archaeology were used to determine which samples would be taken, and ultimately analysed. Archaeologically defined layers do not always relate to soil horizon processes; samples can be from different soil horizons, and thus potentially have different processes and subsequent compositions. These factors can influence the chemical and physical composition of samples, hence the retention capacity of the sample for anthropogenically sourced inputs. The samples taken from the grid represent a combination of potentially selective leaching and accumulation in the soil without temporarily. Naturally this has inherent problems for interpretation, which is discussed below. The inclusion of cores in geochemical anlaysis was intended, as a pilot study, to see if temporarily could be added to the technique. Samples from the cambric, silty loam upper B horizon in Area 6 were prioritised and selected. In addition, samples located in archaeological features or beside known modern intrusions were discarded. The only exception was the deliberate sampling of an oven to determine its function during use. All samples were taken by means of disposable plastic

3 Rockworks 15 by Rockware.



18 Cannell et al.: Geochemical analysis using pXRF 

 429

spoons for ease and speed, and to reduce cross-contamination. Sample points were then surveyed using a total station theodolite4.

18.4.2 Instrumental settings A Niton (Thermo-Scientific) XL3t GOLDD, with a 45 kV, 0–200 µA x-ray tube was used with manufacturer’s filter settings in mining mode, using fundamental parameters calibration. In total, 493 readings were taken from 189 samples, including samples used to test the effects of moisture content (to be published elsewhere). The sample preparation involved drying the samples for 24 hours at 105°C; the samples were then sieved to less than 1 mm before being crushed with a clean ceramic pestle and mortar. Two cores from Area 4 were subsampled in all horizons to serve as background reference samples (Bauer and Østmo, Ch.  5:67). For analysis, samples were placed in plastic cups with a 4 µm polypropylene membrane as the sample window. For the pXRF analysis, the filter times were set by incrementally extending the filter time from 20 up to 120 seconds. Initially, the main filter time was set, as all other filter times and results are dependent upon the main filter. Once this was set, the light, high, and low filters were incrementally extended in a similar manner. The light and low filters were of primary interest as these measure lighter elements (Z=