Archaeological Applications of Radiocarbon Chronologies and Statistical Models: Dating the Viking Age Settlement of Iceland (Landnám)
By Magdalena Maria Elisabeth Schmid
PhD Dissertation, University of Iceland, 2018
Abstract: This thesis aims to refine the accuracy and precision of radiocarbon (14C) datasets in order to better understand the timing of archaeological and palaeoenvironmental events relating to important issues like mobility, colonisation, human impacts and human responses to climate change. This is vital, because radiocarbon is one of the most important dating methods in prehistoric archaeology and Quaternary science and some 14C determinations can be anomalously older or younger than their stratigraphy suggests. Various ‘chronometric hygiene’ protocols have been developed aiming at enhancing the quality of 14C datasets by removing all potentially problematic 14C samples, but there is no generally accepted routine approach to chronology building utilising large 14C datasets. Existing practices that eliminate up to 95% of 14C dates can mean that so few dates remain in some locations that a robust chronology cannot be established. Despite their foundation in sound theory, without independent tests we cannot know if established protocols are apt, too strict or too lax. This research tackles this problem utilising Bayesian statistical modelling and tephrochronology.
The Viking age settlement of Iceland (Old Norse: Landnám) is an ideal laboratory to explore the potentials and the limits of chronometric hygiene and Bayesian statistical modelling because of a remarkable conjuncture of complementary dating methods of the archaeology and palaeoenvironment of first settlement (14C dates, ice core-dated tephrochronology, artefact typology, medieval literary texts and palaeoecology). The timing of Landnám is of major international significance because it represents a key stage in the greater Norse colonisation of the North Atlantic islands that led to the first European contact with North America. In recent years intensive archaeological research on the Viking age has produced significant new dating evidence and offers exciting opportunities to assess colonisation as a process. This thesis presents a first systematic and holistic cross-disciplinary regional study that critically synthesizes the spatio-temporal dynamics of settlement patterns in Viking age Iceland. The countrywide distribution of 19 tephra layers, 513 stratigraphically related 14C dates and diagnostic artefacts at 550 archaeological sites (300 settlements, 140 burial sites and 110 stray finds) are reassessed.
The first research objective of this thesis establishes a high-resolution chronology of the Viking age settlement of Iceland that is primarily based on tephra-dated settlements and burials (n = 261). While 85 sites can only be assigned general Viking age dates, less than 1% of the remaining 185 sites are from the pre-Landnám (pre AD 877) period, 48% from the Landnám (AD 877-938/939) and 51% from the post-Landnám (AD 938/939-1104) period. The combination of 335 reliable 14C samples yielded posterior probabilities of cal AD 863-881 (68%) and cal AD 751-893 (95%) for the overall onset of colonisation. Significantly, seasonal anthropogenic activities are dated before AD 877 in the southwest, while large-scale settlement (Landnám) from the coastal to inland zones happened after that date. The data support the hypothesis that Landnám was largely completed in twenty years after the deposition of the Landnám tephra layer of AD 877 ± 1.
The second objective establishes a new chronometric evaluation protocol that is based on Bayesian statistical modelling and uses robust constrains provided by medieval literary texts and ice core-dated tephra layers at Icelandic archaeological sites. This protocol promotes the most parsimonious exclusion of data – bulk sediments and samples where sufficient metadata is not published (e.g. material type, isotopes), while a variety of materials – short-lived taxa, wood charcoal with inbuilt ages as well as bone samples affected by marine reservoir offsets – allow robust chronologies to be established. In fact, the higher the density of 14C dates across a ‘Phase’, the higher the precision of posterior probabilities. Significantly, samples usually held to be at risk of inbuilt ages could still provide robust results in a large number of cases if appropriate prior assumptions are used and if the distribution of 14C dates through the ‘Phase’ is uniform. This has critical implications for 14C chronologies in world archaeology, as it means that current approaches may well be hampering investigations unnecessarily, by applying an overly prescriptive, and potentially biased, approach. In fact, this opposes current hypotheses about the robustness of uniformly distributed 14C datasets: Bayesian models are in fact sensitive to the distribution of dates and they will be biased if filtered datasets have certain dates removed, most significantly if the dates are from early contexts. They are also sensitive to any dominant inclusion of biased dates, such as high numbers of charcoal dates. Although short-lived taxa are always more accurate and precise than samples with inbuilt-age, these data can have limited use if we do not assess their stratigraphic relationships, as samples may not directly relate to the event in question.
The thesis also introduces a new software program ‘OxCal_parser’ for rigorous data entry when dealing with large datasets in OxCal. The new analytical and methodological approaches developed and tested in Iceland are then effectively extended to the example of East Polynesia, which successfully demonstrates the potential of the approaches developed in other geographical settings, and their relevance in these contrasting locations. The project has resulted in a series of scientific publications, and has created a new common resource of Viking age archaeology in Iceland. This resource will be of significant help in spotting dating errors, simplifying the process of dating, and allowing for a new type of metadata analysis of chronological data. This in turn will open up new avenues of research in interdisciplinary archaeological science and thus contribute to international debate on this core element of archaeological practice.