ENVIRONMENTAL ARCHAEOLOGY LAB.
REPORT no. 2017-012
Environmental archaeological analysis from the archaeological excavations at Ørland kampflybase, Vik 70/1, Ørland,
Sør-Trøndelag, Norway. 2015-2016
Philip Buckland, Johan Linderholm, Sofi Östman, Samuel Eriksson, Jan-Erik Wallin & Roger Engelmark
DEPARTMENT OF HISTORICAL, PHILOSOPHICAL & RELIGIOUS STUDIES
i
Contents
List of tables ... ii
List of figures ... iii
Introduction ... 1
Site background ... 1
Research questions ... 2
Methods and materials ... 3
Sampling ... 3
Plant macrofossil analyses ... 3
Soil chemistry and physical property analyses ... 8
Pollen analyses ... 8
Plant macrofossil results overview ... 9
Soil chemical and physical property results and initial interpretation ... 11
Analysis and interpretation ... 17
Introductory remarks ... 17
General patterns in the plant macrofossil sample data ... 18
Principal Component Analysis (PCA) of taxa and features ... 18
Plant macrofossil preservation ... 21
Plant groups in relation to feature types ... 21
Other general archaeobotanical points... 22
General patterns in the geoarchaeological data ... 25
Felt A - Activity areas, refuse layers and cooking pit fields ... 28
Felt A: Hus 01 ... 28
Felt A/E: Hus 05 and Hus 14 area ... 31
Hus 05 ... 31
Hus 14 and area between this and Hus 05 ... 34
Felt A: Hus 09 ... 35
Felt A: Area of samples 150727 and 153414 ... 39
Felt A: Hus 31 and area to its west ... 40
Felt B – Buildings and cooking pits from the pre-Roman Iron Age ... 42
Felt B: Pits and postholes near sample 336941 ... 42
Felt B: Hus 03, 06 and 07 ... 43
Overview of area ... 43
Hus 03 ... 43
Hus 06 ... 43
Cooking pit area around sample 336885 ... 48
Hus 07 ... 49
Felt B: Hus 08, Hus 11, Hus 13 ... 52
Wells and cultivation layers ... 52
Hus 08 ... 53
Hus 11 ... 53
Hus 13 ... 53
Felt B: Hus 10 ... 54
Felt C - Farmsteads from the late Roman Period/migration period ... 55
Felt C: Cultivation layer area, Intrasisid 500344, sample 338830 etc... 55
Felt C: Hus 15 area ... 57
Western area ... 57
ii
Central area (around cooking pit feature 517508, sample 338647) ... 57
Nothern area ... 58
Hus 15 ... 58
Felt C: Hus 02, 04, 12, 16, 17, 18, 19 and Felt D: Hus complex 21-24, 26, 28-30 ... 59
Overview of area ... 59
Felt C: Hus 02, 04, 12, 16, 17, 18, 19 ... 63
Hus 02 ... 63
Hus 04 ... 68
Hus 12 ... 68
Hus 19 ... 68
Hus 16 ... 68
Hus 17 ... 69
Hus 18 ... 70
Area between Hus 02 and Hus 04 ... 71
Felt D: Hus 21-30 ... 72
Felt E ... 75
Felt E, Hus 24,26,19,14... 75
Felt E: Grophus 15 ... 83
Conclusions and points for further discussion ... 85
References ... 87
Appendices ... 89
Appendix 1. Glossary of plant names ... 89
Appendix 2. Pollen analysis report - Jan-Erik Wallin ... 91
Appendix 3. Plant macrofossil sample results and data ... 95
Additional finds and information from plant macrofossil samples ... 109
Sampled submitted for radiocarbon dating (14C) ... 116
Appendix 4. See separate digital file for soil chemistry and physical properties data ... 119
List of tables Table 1. Geoarchaeological methods and abbreviations as used in this report. ... 8
Table 2. Descriptive statistics from the soil chemistry analysis of the various contexts related figure in the analysis. ... 12
Table 3. Felt A plant macrofossil samples and results, including plant type summaries (in bold). ... 96
Table 4. Felt B plant macrofossil samples and results, including plant type summaries (in bold). ... 100
Table 5. Felt C plant macrofossil samples and results, including plant type summaries (in bold). ... 104
Table 6. Felt A, extra finds and information from plant macrofossil samples ... 109
Table 7. Felt B, extra finds and information from plant macrofossil samples. ... 111
Table 8. Felt C, extra finds and information from plant macrofossil samples ... 113
Table 9. 14C sample submissions ... 116
iii
List of figures
Figure 1. The full Ørland site area with topographic situation and location of defined buildings (Hus). ... 5
Figure 2. Site topographic situation and sampling points for surface samples (green dots) (n=1632). ... 6
Figure 3. Site topographic situation and sampling points for feature samples (green squares) (n=589). ... 7
Figure 4. Graphical summary of plant taxa found. Total number of seeds per taxa (and straw/spikes) found in all samples (thin blue bars), ordered from highest to lowest, and number of samples containing each taxa (thick black bars). Note the log scale and that empty bars represent one seed per taxa. (See Appendix 1. Glossary of plant names for English and Norwegian names). ... 10
Figure 5. Frequency distribution of Cit-P data (left, surface samples n=1632), (right, feature samples n=589). ... 11
Figure 6. Frequency distribution of MS data (left, surface samples n=1632), (right, feature samples n=589). ... 11
Figure 7. Frequency distribution of LOI data of feature samples (n=589). ... 12
Figure 8. Boxplots describing Cit-P variation within each feature category. ... 13
Figure 9. Boxplots describing MS variation within each feature category. ... 14
Figure 10. Boxplots describing organic matter (LOI) variation within each feature category. ... 15
Figure 11. Boxplots describing variation in P quota within each feature category. ... 15
Figure 12. Principal Component Analysis loading of the 1st and 2nd components using presence only data (i.e. no abundance or volume measurements). The proposed separation of crop processing, grain storage and burning/food production is indicated. ... 18
Figure 13. PCA loading for the first two components, central cluster of points showing tendency of most taxa/finds towards storage and to a lesser extent heating. ... 19
Figure 14. PCA model for the plant macrofossil and sample finds data, classified by feature type (provided by the archaeologists) and using presence only data. The first two components are shown. ... 20
Figure 15. Relative abundance of plant groups per excavation feature class (i.e. based on the number of seeds per class). ... 23
Figure 16. Relative occurrence of plant groups per excavation feature class (i.e. based on the number of samples in each class where each plant type occurs). ... 24
Figure 17. Spatial distribution of soil phosphate content (Cit-P) over the investigated area at Ørland (n=1632). ... 26
Figure 18. Spatial variation of soil MS data over the investigated area at Ørland (n=1632). ... 27
Figure 19. Hus 01 area: Sample points. ... 28
Figure 20. Hus 01 area: Plant macrofossil analysis results summarised. ... 29
Figure 21. Felt A, Hus 01, Cit-P survey. ... 29
Figure 22. Felt A, Hus 01, Cit-P survey and features. ... 30
Figure 23. Felt A, Hus 01, MS features. ... 30
Figure 24. Hus 05 and Hus 14 area: Macrofossil sample points. ... 31
Figure 25. Felt A/E, Hus 05 Cit-P. ... 32
Figure 26. Felt A/E, Hus 05 MS. ... 33
Figure 27. Hus 05 and 14 area: Plant macrofossil analysis results summarised. ... 34
Figure 28. Hus 09 area: Sample points. ... 36
Figure 29. Hus 09 area: Plant macrofossil analysis results summarised. ... 36
Figure 30. Felt A, Hus 09, Cit-P survey and Cit-P of features. ... 37
Figure 31. Felt A, Hus 09, Cit-P survey and P quota of features. ... 37
Figure 32. Felt A, Hus 09, MS survey and features. ... 38
Figure 33. Area of samples 150727 and 153414: Plant macrofossil analysis results summarised (note: pie charts not over sample points). ... 39
Figure 34. Hus 31 area: Sample points. ... 40
Figure 35. Hus 31 area: Plant macrofossil analysis results summarised. ... 41
Figure 36. Macrofossil samples from pits and postholes near sample 336941. ... 42
Figure 37. Hus 03, 06 and 07: Sample points. ... 44
Figure 38. Felt B: Hus 03, 07, 07, 08 and 13, Cit-P. ... 45
Figure 39. Felt B: Hus 03, 07, 07, 08 and 13, P quota. ... 46
iv
Figure 40. Felt B: Hus 03, 07, 07, 08 and 13, MS. ... 47
Figure 41. Pits and postholes around sample 336885, showing seeds, straw and charcoal proportions. ... 48
Figure 42. Hus 03, 06 and 07: Plant macrofossil analysis results summarised. ... 50
Figure 43. Hus 03, 06 and 07: Straw fragments. ... 51
Figure 44. Hus 08, 11 and 13 and nearby wells. Sample points ... 52
Figure 45. Hus 08, 11 and 13 and nearby wells. Plant macrofossil analysis results summarised. ... 53
Figure 46. Hus 10. Sample points and plant macrofossil analysis results summarised. ... 54
Figure 47. Felt C: Area of sample 338830, intrasisid 500344 ("cultivation layer"), sample points and straw fragments 55 Figure 48. Felt C: Area of sample 338830, intrasisid 500344 ("cultivation layer"), plant macrofossil analysis results summarised. ... 56
Figure 49. Hus 15 area. Sample points. The western area is adjacent to the large feature towards the left-hand frame. . 57
Figure 50. Hus 15 area. Plant macrofossil analysis results summarised. ... 58
Figure 51. Spatial distribution of soil phosphate content (Cit-P) over the Felt C and D. Northern house complex 2, 4, 12, 16-19; Southern house complex 21-24, 26, 28-30. ... 60
Figure 52. Spatial distribution of soil phosphate content (Cit-P) over the Felt C and D, with relative presence of carbonate/shell in samples noted. ... 61
Figure 53. Spatial variation of soil Magnetic Susceptibility data over the Felt C and D. ... 62
Figure 54. Hus 02, 04, 12, 16, 17, 18 and 19. Sample points ... 64
Figure 55. Spatial distribution of soil phosphate content (Cit-P) over the Felt C; detail over Hus 02 and 04. Parallel surface sample lines inside houses. ... 65
Figure 56. Spatial distribution of soil phosphate content (Cit-P) over the Felt C; detail over Hus 02 and 04, with feature responses added. ... 66
Figure 57. Spatial variation of soil Magnetic Susceptibility data over Felt C; detail over Hus 02 and 04. ... 67
Figure 58. Hus 02, 04, 12, 16, 17, 18 and 19. Plant macrofossil analysis results summarised. ... 69
Figure 59. Hus 02, 04, 12, 16, 17, 18 and 19. Approximate charcoal proportions in macrofossil samples. ... 70
Figure 60. Felt D Hus 21-30 Cit-P and features. ... 73
Figure 61. Felt D Hus 21-30 MS. ... 74
Figure 62. Felt E, Cit-P survey samples. ... 76
Figure 63. Felt E, Cit-P feature samples. ... 77
Figure 64. Felt E, MS survey samples. ... 78
Figure 65. Felt E, MS, feature samples. ... 79
Figure 66. Felt E, Hus 24,26,19,14 Cit-P survey samples. ... 80
Figure 67. Felt E, Hus 24,26,19,14 Cit-P feature samples. ... 81
Figure 68. Felt E, Hus 24,26,19,14 MS feature samples. ... 82
Figure 69. Felt E, Grophus 15 (Intrasisid 2044477) Cit-P. ... 83
Environmental archaeological analysis from the
archaeological excavations at Ørland kampflybase, Vik 70/1, Ørland, Sør-Trøndelag, Norway. 2015-2016.
Philip Buckland1, Johan Linderholm1, Sofi Östman1, Samuel Eriksson1, Jan-Erik Wallin2 & Roger Engelmark1
1The Environmental Archaeology Lab, Umeå University, Sweden
2 Pollenlaboratoriet i Umeå AB
Introduction
A total of 322 bulk samples, 267 bulk subsamples and 1632 survey samples from the excavation of Iron Age settlements at Ørland, Vik, Sør-Trondelag, were analysed at the Environmental Archaeology Laboratory (MAL) at Umeå University. The overall aim of these analyses was to look for evidence which could help identify possible prehistoric activity areas, understand building functions and divisions, and shed light on land management around the farmsteads.
Three sets of methods were applied to the samples: 1) plant macrofossil/archaeobotanical analysis, 2) soil chemical and magnetic analysis (geoarchaeological) and, to a small extent 3) pollen analysis.
The three sets of samples were analysed as follows:
322 bulk samples analysed for carbonised plant macrofossils, soil chemistry and properties (a small subset of 9 samples submitted for pollen analysis)
267 subsamples from features (where macrofossils have been analysed by another lab) analysed for soil chemistry and properties
1632 surface survey samples analysed for soil chemistry and properties
Samples collected during the field seasons of 2015 and 2016 were submitted to MAL by NTNU- Vitenskapsmuseet in Trondheim with project leader Ingrid Ystgaard as primary contact person, and further information provided by Magnar Mojaren Gran. This report is a compilation and interpretation of the analytical results from these samples. Parts of the survey sample analyses have been presented earlier in a preliminary report (Ericsson 2016). Johan Linderholm has been in charge of the project at MAL and Philip Buckland has coordinated and compiled the final report.
Due to the complex and large scale nature of the results, conclusions on individual structures and areas are included in the appropriate sections of the analysis and interpretation, rather than in a final conclusion. We hope that the integrated approach of this work illustrates the importance of multiproxy analyses, where each set of results complements the others to allow for a more holistic interpretation of the site. It should be noted that this does not mean that interpretation will be easier or simpler with multiple lines of evidence - reality is complex, and multiple methods have the ability to reveal this complexity where individual methods may show a misleadingly simple result.
Site background
The site of Ørland kampflybase extends over an area of over 90 acres, has been excavated since 2014, and is one of the largest projects dealt with by NTNU-Vitenskapsmuseet. The excavation has been undertaken due to the expansion of Ørland hovedflystasjon (airbase) in connection with the government decision on the purchase of new fighter jets.
The local bedrock is composed of granites, overlain by marine sediments of shifting thickness.
Sediments in the investigated area consist mostly of beach sands and gravels with patches of shell bank and shell rich areas. The dominant soil type in the investigated area is ploughed agricultural
soils. In the western parts, wetter soils dominate but on the somewhat higher grounds soils are significantly drier.
The earliest traces of human activities in this area are from the Bronze Age, when relative sea level change first exposed the land. Settlement and agricultural activities expanded throughout the Iron Age and into the medieval period. The settlement area of Vik is located on a gravel rich ridge and has most likely been an attractive settlement site for a long time. Its proximity to the sea, well- drained soils and soils well suited for agriculture provide a combination of favourable living conditions.
This large archaeological excavation differs somewhat in scale from others in Norway, as commercial projects are most often limited by the area affected by the infrastructural project.
Excavation was initiated with a full scale removal of the topsoil to allow for the rapid exposure of archaeological features in what is an area considered as having a high cultural historical value. A large number of structures and features were thus made visible, not only features connected to buildings and graves, but also those ones easily missed in smaller surveys such as wells, fences, agricultural remains and areas of other types activities connected to the everyday life. However, there is a risk with such an approach in that potentially valuable signals recorded in the plough soil may be lost. This may have affected the possibilities for interpreting some the soil chemistry and properties data, as described below.
The site appears to have experienced multiple phases of occupation, and thus many features from different ages are likely to have become superimposed. Disturbance related to a later historical dwelling/agricultural settlement is evident at the site and in its surroundings.
Research questions
Such a large scale excavation, with a very large number of environmental archaeology samples, potentially allows for broad spatio-temporal questions to be asked of the material when integrated with the archaeological interpretation. These may include the relationship between different kinds of settlement structures, the use of land over time and the relationship between burial/settlement.
Extensive radiocarbon dating, which has been provided for in subsamples from this material, will help understand the temporal patterns and phasing of different activities as well as help relate this to other sites.
The two main research questions, forming a common thread through the project are expressed as (Ystgaard 2017):
1. “To obtain an understanding of the relationship between the development of landscape and settlement history from the Bronze Age to the middle ages”
2. “To investigate structural changes in the spatial and social organization of the settlement and agriculture from the Bronze Age to the middle ages”
Further questions of interest to the project are related to the development of agriculture, economy, and settlement activities in the area; the use of marine and terrestrial resources; whether different features are connected to traces of a fishing or agricultural economy; where livestock may have grazed and what kind of crops grown. It is also hoped that the combination of plant macrofossil and soil chemical analyses will assist in the understanding of spatial variation within structures and activity areas. We anticipate that the integrated humanities and natural sciences based approach in this report will help provide some material for working towards answering at least some of these questions, or if not, at least provide the basis for more questions to be formulated.
Methods and materials
Sampling
Sampling was undertaken by NTNU archaeological staff in connection with the excavation process and after discussion with Johan Linderholm from MAL. Bulk samples were collected from archaeological features and stored in six litre plastic buckets or three litre plastic bags. Sample size varied between 0.5 – 5.5 litres and samples were ascribed sample numbers (Prov nr). Soil survey sampling was conducted after removal of the Ap-horizon by excavator. Sampling grids were used with distances ranging from 1, 2, 5 10 and up to 20 m depending on the archaeological contexts and need for precision. In two of the structures (Hus 02 and 04), parallel lines of samples were collected inside the buildings along axial lines inside the longhouses.
Figure 1 provides an overview of the research area, with identified structures marked as “Hus”, and Figure 2 shows the surface sample points. A total of 1632 surface samples were submitted for analysis covering an area of 4.5 ha. Figure 3 shows the location feature subsamples for which geoarchaeological analyses were undertaken, including samples not submitted to MAL for archaeobotanical analyses. The total number of geoarchaeological samples analysed in this category was 589, of which 322 were subsamples from plant macrofossil samples submitted to MAL.
The samples arrived to Umeå packed on pallets and were then organized and marked with a local sample ID (MAL nr). From the bulk samples, subsamples were extracted in the lab for soil chemical analysis and pollen analysis (where requested). To ensure a statistically representative subsample of the bulk samples, the material was poured out on a tray and ca 10 ml of soil representing the whole sample collected and processed separately according to analysis method (see method descriptions below). Due to EU restrictions on the handling of soils from outside of the EU, the macrofossil samples were subsequently heated to 170°C, a temperature sufficient to kill organisms, before processing. The smaller soil survey samples were assigned a local ID and stored in drying room at 30°C before processing.
Sample organisation and processing was undertaken by Tone Hellsten and Guido Mariano at MAL.
Plant macrofossil analyses
Prior to analysis, samples were stored in a drying room (+30°C) to eliminate moisture and reduce the risk of mould which could prevent accurate 14C dating. (This has the unfortunate effect of damaging non-carbonised, subfossil, material and alternative strategies must be considered if waterlogged samples are to be submitted with such large sample batches in the future). Sample volume was estimated before floatation and washing with water through 2 mm and 0.5 mm sieves.
The resulting material (flotant) was sorted and identified under a stereo microscope (8x) with the help of MAL’s plant macrofossil reference collection and reference literature (Cappers et. al. 2006).
Only charred/carbonised material was extracted from samples and the amount of woody charcoal estimated. (Note that non-charred material found in carbonised contexts should always be treated with suspicion as there is a high probability of it being contaminant). Material for 14C was extracted during identification and weighed. Only cereals and straw fragments have been submitted for dating. Charcoal proportions, when given, were assessed in addition to any seeds and straw fragments found in the samples. Charcoal was returned to NTNU for submission to another laboratory for charcoal analysis and additional 14C dating.
Plant macrofossil identifications at all levels of detail are referred to as “taxa” (“taxon” in the singluar). When preservation is at its best, cereal identification can be performed at the subspecies level, such as Hordeum vulgare var. vulgare (hulled barley/agnekledd bygg) or Hordeum vulgare var. nudum (naked barley/naken bygg). With suboptimal preservation, cereals can be identified at best to the species level, e.g. Hordeum vulgare (barley/ bygg), or at worst simply as cerealia (indet).
Half cereal grains or small pieces and fragments are referred to as cerealia fragmenta. When a
macrofossil looks like a particular species but lacks the species specific characters necessary for a 100% reliable identification, it is referred to as “cf. taxa” (cf. Triticum, means “looks like” wheat).
This system is applied to all of the botanical material. Plant names are given in the text as
"Scientific/Linnean name (English/Norwegian)" and Appendix 1. Glossary of plant names is provided for further reference (including Swedish names).
Other material potentially of archaeological significance encountered during the macrofossil processing was also recorded and their volume or quantity estimated. This includes bones, ceramics and other small pieces of archaeological remains.
A complete list of macrofossil and other remains is provided in Appendix 3. Plant macrofossil sample results and data.
Macrofossil analyses were undertaken by Sofi Östman, Jenny Ahlqvist and Roger Engelmark, and the interpretation assisted by Philip Buckland. Radoslaw Grabowski kindly provided additional advice on the interpretation of some of the results.
Figure 1. The full Ørland site area with topographic situation and location of defined buildings (Hus).
Figure 2. Site topographic situation and sampling points for surface samples (green dots) (n=1632).
Figure 3. Site topographic situation and sampling points for feature samples (green squares) (n=589).
Soil chemistry and physical property analyses
Prior to all analyses all samples were dried at 30°C. Samples were then passed through a 1.25 mm sieve and any presence of material of cultural significance noted (such as bone, charred material, ceramics etc.). The chemical methods employed here are the same as those used in Swedish soil chemical studies following the methodological approach of Engelmark and Linderholm (1996 and 2008). The parameters analysed and abbreviations used are explained in Table 1. The term
“geoarchaeological” is sometimes used synonymously with all of the analyses.
Table 1. Geoarchaeological methods and abbreviations as used in this report.
Abbreviation Method Description
MS Magnetic Susceptibility Magnetic susceptibility measured on 10g of soil, with a Bartington MS3 system with an MS2B probe (Dearing 1994).
Data are reported as SI-units per ten grams of soil, (corresponding to Xlf, 10-8 m3 kg-1) (Thompson & Oldfield 1986).
MS550 Magnetic Susceptibility after burning at 550oC
Magnetic susceptibility after 550° C ignition (units as above)
LOI (%) Loss On Ignition Soil organic matter, determined by loss on ignition at 550° C, in percent (Carter, 1993).
Cit-P Inorganic phosphate content (mg P2O5/100g dry matter)
Extraction with 2% citric acid (corresponding to the Arrhenius method (Arrhenius 1934 and 1955))
Cit-POI Total phosphate (mg P2O5/100g dry matter) (inorganic & organic)
Extraction with 2% citric acid on ignited soil
P quota Cit-POI /Cit-P Ratio of inorganic & organic to inorganic phosphate
These methods have been developed and adapted for soil prospection and the bulk analysis of occupation soils and features. Analysed parameters comprise organic matter (loss on ignition [LOI], Carter 1993), two fractions of phosphate (inorganic [Cit-P], and sum of organic and inorganic [Cit- POI]) (Engelmark & Linderholm 1996, Linderholm 2007) and magnetic susceptibility (MS-χlf) and MS550-χlf (Clark 2000, Linderholm 2007, Engelmark & Linderholm 2008). These analyses provide information on various aspects concerning: phosphate, iron and other magnetic components and total organic matter in soils and sediments, and their relationship to phosphate. (Further details can be found in Viklund et al., 2013).
In this study, surface samples were analysed for Cit-P and MS. All feature/macrofossil subsamples were analysed with all five parameters (and P quota is calculated from two of these). The interaction of all of these parameters allows for a more detailed understanding of the implications of each of them in terms of archaeological research questions.
Soil chemical analyses were undertaken by Samuel Eriksson, Johan Linderholm and Guido Mariani and interpretation by Johan Linderholm and Samuel Eriksson.
Pollen analyses
Samples were treated according the standard methodology for pollen preparation as described by Moore et al. 1991. Concentrated pollen was placed on a slide and coloured with saffron-dyed glycerine. Pollen taxa were identified under microscope using the keys of Beug (1961) and Moore et al. (1991), counted, and summarised for this report. All pollen samples derive from subsamples of nine bulk sampled archaeological features. Pollen analysis was undertaken by Jan-Erik Wallin, Pollenlaboratoriet AB. For the full pollen report, see Appendix 2. Pollen analysis report - Jan-Erik Wallin.
Plant macrofossil results overview
A complete list of plant macrofossils found, as well as other finds encountered during processing and a list of radiocarbon dating submissions can be found in Appendix 3. Plant macrofossil sample results and data.
A total of 322 plant macrofossil samples from archaeological features were analysed, of which 166 samples produced 2285 identifiable carbonised seeds. The number of seeds found per sample varied considerably, from a single fragment to over 700 fragments (sample 332970, Hus 07). Cereal grain fragments were by far the most abundant macrofossil find across the site, with a total of 843 fragments being found (Figure 4). However, fragmentation leads to overrepresentation in that more than one fragment may come from the same grain, and it is essentially impossible to reconstruct the number of whole grains without underestimating their abundance instead. At the genus level, Hordeum (barley/bygg) species dominate, with 438 grains more or less equally split between seeds only identifiable to the species level, Hordeum vulgare var. vulgare (hulled barley/agnekledd bygg), and Hordeum vulgare var. nudum (naked barley/naken bygg). Stellaria media (common chickweed/vassarve) seeds were found in large numbers, mainly concentrated in a single sample (332357, also in Hus 07). The distribution of cereal straw fragments, including occasional spikes which held the actual cereal grains, is of interest and is described where relevant.
Plant taxa were classified into four broad cultural-ecological classifications: Cultivated, Weeds/Ruderals, Meadow/Pasture/Wetlands and Other. Ruderals are plants which commonly grow on waste or disturbed, nutrient rich ground. A list of classifications for each taxon is provided in Appendix 1. Glossary of plant names. These classifications are used consistently in the summary diagrams and maps in the analysis and interpretation below. Classification was performed with respect to the taxa's dominant characteristic, but should not be considered as absolute fact. Stellaria media (common chickweed/vassarve), for example, may have been used as a food stuff for people and animals, but is mostly indicative of disturbed, nutrient rich soils.
Overall, the archaeobotanical material of the site is dominated in numbers by cultivated grain crops, followed by weeds and ruderals. In terms of species diversity, cereals make up at least half of the taxa found. The frequency distribution of plant taxa at this site is essentially in line with what can be expected from archaeological investigations. Further summary statistics and a discussion of these points are presented in the analysis and interpretation below.
Figure 4. Graphical summary of plant taxa found. Total number of seeds per taxa (and straw/spikes) found in all samples (thin blue bars), ordered from highest to lowest, and number of samples containing each taxa (thick black bars). Note the log scale and that empty bars represent one seed per taxa. (See Appendix 1. Glossary of plant names for English and Norwegian names).
Soil chemical and physical property results and initial interpretation
Soil chemical and physical property results are provided in a separate digital appendix due to the large number of samples.
Figure 5 shows the frequency distribution of Cit-P for both categories of analysed samples (survey and feature). The Cit-P dataset from the surface samples shows a bimodal distribution, which is related to the presence of shell banks that lowers the response from the citric soluble phosphate, resulting in the first lower peak. This was already observed in the 2015 samples and this pattern persists when the 2016 samples are added. From the second peak and upwards there are several samples with a significant phosphate accumulation which can be interpreted in terms of human input and habitation. The feature samples have significantly less shell bank influence and thus no bimodality. Instead, a strong log-normal distribution is evident with a significant phosphate input.
Figure 5. Frequency distribution of Cit-P data (left, surface samples n=1632), (right, feature samples n=589).
Figure 6. Frequency distribution of MS data (left, surface samples n=1632), (right, feature samples n=589).
The MS data is log-normally distributed (Figure 6) showing a median of 24 (Table 2). This is to a certain extent dependent on the geological background and general marine origins of the sediments.
A few outliers are found in the material and the feature samples show a bimodal distribution caused by a mix of heat-related and non-heat related features in the dataset. The loss on ignition (LOI;
organic matter, charcoal and some biogenic carbonate) frequency distribution for the feature samples is provided in Figure 7, and individual results are described where relevant below.
Table 2. Descriptive statistics from the soil chemistry analysis of the various contexts related figure in the analysis.
CitP MS
Figures Location number median mean StDv median mean StDv
Figure 17 & 18 All data 1632 86 85.4 31.8 24 30.6 21.7
Figure 51 - 53 Felt D 479 78.4 73.5 38.3 38 39.5 17.4
Figure 55 - 57 Hus 02 & 04 128 45 71.8 52.2 41.5 43.2 18.8
Figure 62 - 64 Felt E 652 80 85 23.7 18 18.9 9.2
Figure 25 & 26 Hus 05 114 92 90.8 14.1 11 12.5 3.8
Figure 30 - 32 Hus 09 80 104 93.8 31.7 21 22 6.6
Figure 21 - 23 Hus 01 215 102 92.1 35.9 21 22.8 7.6
Figure 7. Frequency distribution of LOI data of feature samples (n=589).
Box plots of the feature samples divided into different feature categories (Figure 8 - Figure 11) provides an even clearer picture of the feature distributions. The categories used in these box plots are mostly those defined during the excavations: postholes, posthole impressions (note that Hus 05 in Felt E contributes half of these), waste pits, cooking pits and hearths. The two last categories, layers and pits, combine all defined layers into one category and in the pit category, undefined pits and “nedgravning” are combined.
The phosphate distributions (Figure 8) show that the spread varies in each group. Postholes are quite clustered around the median (shown by the box) but with several outliers (points), whereas cooking pits (hearths, layers and pits) show a wider spread around the median (larger box). The hearth category shows quite a number of higher phosphate rich objects. In general, “pits” in settlement contexts are filled up with culturally impacted sediments, often related to specific uses but with potentially complex re-use histories (see analysis and interpretation below). Postholes have a different catchment and should more represent the surrounding, point specific areas.
Figure 8. Boxplots describing Cit-P variation within each feature category.
The MS box plot (Figure 9) shows similar trends for the posthole samples as for phosphate concentrations. Heat related features, hearths and cooking pits, have the highest medians and top readings, which is to be expected. However, in the case of cooking pits, cases with low readings can be explained in terms of either sampling, location, classification or use. The sampling issue is that the sampled layer in a pit may not reflect the actual situation of use (a problem equally relevant for the plant macrofossil interpretation, see below). Location is another aspect in this case as the ground soil varies from wet to dry, with consequent differences in taphonomy and soil formation. Location also relates to function, and the relationships between structures and activity areas. A broad field classification may also lead to this category being broader, as several different functions are possible. In addition, the use of a cooking pit may not have been that intensive. The same line of reasoning is applicable to hearths but this category is theoretically “easier” to understand.
The box plot distribution of loss on ignition data (LOI; organic matter, charcoal and some biogenic carbonate) in the different feature categories is shown in Figure 10. As can be expected, the cooking pits and hearths are found in higher levels, as charcoal (which is abundant in these features)
contributes to the overall loss on ignition. Postholes show a similar pattern as in the previous two parameters.
Finally, Figure 11 shows the P quota for the feature categories. In general, all “pits” and layers show the highest readings and postholes show the most outliers. There are only 12 instances where high P quotas may be related to presence of shell material, so general use and catchment are more plausible explanations here and the results probably reliable for archaeological interpretation.
Figure 9. Boxplots describing MS variation within each feature category.
Figure 10. Boxplots describing organic matter (LOI) variation within each feature category.
Figure 11.Boxplots describing variation in P quota within each feature category.
Analysis and interpretation
Introductory remarks
In parallel with the complexity of the archaeology of multiphase, long-term sites, interpreting environmental archaeology data in terms of building and area usage from these sites is complicated.
There is a tendency for material to become mixed, especially in sunken features such as pits or postholes, and especially when buildings are demolished to make way for new ones on the same site. Postholes may be dug into earlier cultural layers and features, and the origin of fill materials may not always be connected to the construction, usage or demolition phases of the building under examination. For structures which have been used for a long time there is the problem of composite evidence - i.e. that our evidence may not be snapshots single phases of activity, but take the form of a mix of material from multiple usage phases (see Grabowski 2014 for more discussion on these lines).
There are also a number of difficulties to be considered in relation to the archaeological and environmental interpretation of pits or depressions of any kind. Purposefully created pits may be emptied after use, reused and repurposed, and present a difficult stratigraphy even if layers are well preserved, documented and sampled stratigraphically. The central or largest single context may represent the phase of last use, and the pit may subsequently have been covered as land was levelled for new activities or constructions. However, rather than being abandoned whilst full, it may have been abandoned after emptying, and filled with rubbish after its last use, perhaps with mix of waste or surface material from the area around pit or a waste disposal area (as is common in medieval urban deposits). The latter factors make pits especially difficult to date reliably as the age and source of charcoal in the features is difficult, and perhaps sometimes impossible, to assess. Pits also have multiple potential uses, including as food cooking pits, but also for fat reduction (usually leading to high phosphates), roasting (often producing low phosphates), heating stones, mixing tar for waterproofing, storing/maturing dung, or as waste pits, latrines, and wells or animal water holes.
Some features may have been used for several of these purposes over their lifetime, each of which may or may not have left its traces in the environmental record. Lipid analysis could be useful in fine tuning the investigation of some of these features, but the method is still not immune to the above stratigraphic issues. Good stratigraphic sampling and context documentation is naturally key to understanding the results of any analysis.
Below follows a presentation of the analysis and interpretation of the plant macrofossil and geoarchaeological evidence. We have divided samples into what appear to be cohesive or logical areas of analysis from the environmental point of view. These may not always be the same for the macrofossils and geoarchaeology, so some separation of the analysis areas is inevitable, as is some repetition of descriptions to help contextualise the interpretations. We have used the terms building and structure interchangeably and avoided the use of house unless there are indications of domestic activities or established house divisions. We have used the term stable universally to describe the stalling of any type of animal, rather than the more ambiguous terms byre or barn, which may include the storage of fodder.
In light of the above, and until more dating evidence becomes available, some of the interpretations below should be considered preliminary and open to revision in the light of new phasing evidence.
It is important to remember then, however, that radiocarbon dates must never be discarded - always explained, even if they do not fit preconceived ideas, and that if the dating does not fit with the archaeological interpretation then either of them should be open to discussion.
General patterns in the plant macrofossil sample data
Detailed statistical analysis of carbonised archaeobotanical remains, especially where preservation is variable or poor, is unadvisable due to the potential for patterns emerging as a result of taphonomic, rather than cultural or biological factors. With enough samples, however, it may be possible to use presence/absence (occurrence) data to identify some broad trends in the type of material found in features with different field classifications. This is something of a "big data"
approach to environmental archaeology that is only made possible through either large scale excavations or the combination of data from multiple sites.
Principal Component Analysis (PCA) of taxa and features
178 of the 322 samples produced non-seed charcoal and 56 samples produced straw/spikes from cereal plants. The loading from simple principal component analysis (PCA) on the occurrence of all sample finds, including macrofossils, indicates that features containing charcoal tend not to have preserved seeds (Figure 12).
Figure 12. Principal Component Analysis loading of the 1st and 2nd components using presence only data (i.e. no abundance or volume measurements). The proposed separation of crop processing, grain storage and burning/food production is indicated.
Similarly, samples containing many cereal grains/fragments or charcoal are a different set from those containing straw/spike fragments. This clearly indicates a separation of samples representing broadly defined activities: (crop) processing, (grain) storage and burning/heating/food production.
The PCA model was calculated from a dataset where the abundance of each macrofossil or find in a sample was replaced by the number 1. Whilst this may introduce artefacts in the model (see below), it prevents superabundant taxa from having extreme effects on the results (i.e. skewing the model in a specific dimension) and allows quantitative (e.g. number of cereal grains) and qualitative data (e.g. relative amount of charcoal) to be combined in the same analysis.
Whilst these three signals are strong, the majority of macrofossils and sample finds are clustered in the centre of the PCA loading results for the first two components (within the dotted box in Figure 12 and as a close-up in Figure 13).
Figure 13. PCA loading for the first two components, central cluster of points showing tendency of most taxa/finds towards storage and to a lesser extent heating.
Zooming in on this central cluster, however, shows a general trend towards the storage component in the bottom left of the diagram, and to a lesser extent, heating/cooking (down and to the right). A tentative conclusion could be that the vast majority of macrofossil samples come from areas which represent the farmstead activities of storing and using cereal products as well as heating (for undefined purposes), with a few outliers representing other activities such as animal stabling, crop
production and other non-cereal related processing activities. Evidence for crop processing appears to be limited to a restricted set of samples. This is something that should be investigated further, paying attention to the dating of features and their spatial distribution over the site.
Whilst the loading diagrams for the plant taxa and sample finds present some clear patterns, the PCA model itself (Figure 14) is more difficult to interpret, giving weight to the complexity of a multi-phase, multi-purpose site. The apparent diagonal striping may be an artefact of the use of presence only data, but there appears to be a tendency for postholes to cluster to the top left and bottom right. Cooking pits seem also to be predominantly in a bottom right cluster, and it would be worth exploring these data in more detail, looking at the geochemistry and plant macrofossils in the same model.
Figure 14. PCA model for the plant macrofossil and sample finds data, classified by feature type (provided by the archaeologists) and using presence only data. The first two components are shown.
Plant macrofossil preservation
There are a number of simple but important factors to consider when interpreting quantitative plant macrofossil data: different species of plant produce different numbers of seeds; different seeds or other remains preserve differently in different conditions; human activities may intentionally or unintentionally concentrate or reduce the abundance of a particular type of group of plant. There are indications that much of the material in this investigation has been well burned, rather than carbonised, and that many fragile seeds will most likely have been destroyed. Preservation of seeds in the samples is generally poor, and there is therefore a bias which has led to the preservation of only thicker walled seeds. The coastal location of the site and the porous sediments with little organic matter may also have negatively affected preservation; a water table which fluctuates through the archaeological features helps to both mechanically break down even carbonised remains, which are less susceptible to breakdown through biological and chemical processes. Later drainage and ploughing will most likely also have negatively affected preservation. Collectively, this has unfortunately produced a material where little quantitative comparison of the relative abundance of plant species (c.f. Viklund 1998b) can be undertaken for individual structures.
In the light of these taphonomic issues, the following figures (Figure 15 and Figure 16) should thus be regarded only as a guide for identifying general trends, and that an integrated analysis of specific contexts, plants, geoarchaeology and the archaeology should be used to provide more information on some of the specifics of possible activities on the site.
Plant groups in relation to feature types
As Figure 15 illustrates, the macrofossil samples are clearly dominated by finds of cereal grains, with unidentified cerealia and barley species making up the majority of these finds (see Figure 4 and comments related to this). However, the extremely high number of fragmented grains (843 - more than twice as abundant as the next highest taxa) suggests poor preservation conditions which have most likely differentially affected the preservation of other seeds. Cereal grains are large and robust when carbonised, whereas the seeds of many other plants are smaller, fragile and light. For the crops this is of course the result of millennia of selection through cultivation to produce plants with larger and more abundant grains. For the weeds and other plants different natural selection processes have applied (even if the cultural selection of crop plants may have influenced the genetic pathways of some other species, especially weeds), including seed and pollen dispersal strategies, pest resistance and environmental adaptation. Furthermore, sampling strategies will heavily influence the type of plant macrofossils found, and if more (potentially) domestic features are sampled then there will inevitably be a bias towards culturally important plants. To balance this factor it is important that, if possible, stratigraphic samples (e.g. peat bog and lake cores) away from the influence of the site are examined with other methods (such as pollen and fossil insect analyses) in order to build a picture of the vegetation history of the area. This may then be contrasted with on- site finds, and any potential human impact in the landscape signals correlated with site data.
Even if weeds and ruderals are underrepresented, they do make up the next most abundant type of plant found in the samples. If we reduce the data to a simple occurrence matrix, where only the presence of a taxa in a sample is considered, rather than the number of seeds found (essentially changing every cell in the macrofossil results tables to '1', as was done for the PCA analysis above), the biodiversity of taxa becomes more balanced (Figure 16). Cereal taxa are less dominant (and would be even less so if the fragments and grains of cerealia indet. were to be combined), and a more diverse weed/ruderal flora is at least partially expressed. This diverse weed flora is commonly considered a feature of pre-industrial agriculture. However, the much smaller number of finds of other seeds still gives an impression of a site dominated by cereals, as the later maps also illustrate.
Other general archaeobotanical points
A number of features produced finds of straw and spikes from cereal plants (see especially
Felt B: Hus 03, 06 and 07 as well as Felt C: Cultivation layer area, Intrasisid 500344, sample 338830 etc.). Whilst the preservation of these macrofossils appears to be variable across the site, and we cannot assume that they were not present in areas where preservation is poor, they do allow us to hypothesize on potential crop processing areas within buildings, and in the latter case in open spaces. The site also displays a high diversity of cereal types, which is typical for what is considered in Sweden as a transitional phase in humanity's agricultural history in the Bronze Age and Iron Age.
The presence of naked Hordeum vulgare Var. nudum (barley/bygg) together with Secale cereale (rye/rug) is particularly indicative of this (Viklund 2008). At this site, the latter is only found in one sample (522516, between Hus 02 and Hus 04, see also Figure 4), which would suggest an earlier, rather than later date for the entire site. There is a conspicuously low amount of finds of Avena sp.
(oats/havre) at the site considering its probable age, which could be explained by either a dominance of older sampled contexts or local agricultural specialisation almost exclusively towards barley.
Figure 15. Relative abundance of plant groups per excavation feature class (i.e. based on the number of seeds per class).
Figure 16. Relative occurrence of plant groups per excavation feature class (i.e. based on the number of samples in each class where each plant type occurs).
General patterns in the geoarchaeological data
This section includes both data from the 1632 soil survey and the 589 archaeological feature samples. Note that archaeobotanical data for 276 of the feature samples is lacking (Felt D and E) as these analyses were undertaken by another lab. It is important that these results, when they become available, are interpreted at the raw data level in combination with the geochemical data presented here.
The Cit-P distribution map (Figure 17) shows an overview of the full extent of the investigation. In this map, the concentration levels (blue colour bands) have been chosen to show the starting point of phosphate accumulation in the different areas. In the southern area (Felt C and D) there is a clear indication of where areas with shell deposits are found (low Cit-P). Both aggregations of buildings in Felt E are located on shell rich areas. However, in Felt B this is not the case. There are areas that show evident phosphate accumulations in association with the different groups of buildings, but not directly on top of them.
In Felt A and E, the connection between phosphate accumulation and buildings seem to be closer than for C and D. In the northern part of Felt E, there seem to be an activity area, highlighted by a phosphate accumulation, which relates to the presence of cooking pits.
The overall variation in MS readings (Figure 18) shows a partially similar pattern to the phosphate distribution. The southern area (Felt C and D) and Felt B have generally higher readings than the northern Felt A and E. Shell rich sediments will show lower readings as the iron content is presumably lower here and any heat applied to this will have less effect on MS measurements than in to the surroundings. This is a clear illustration of one reason why the interpretation of geoarchaeological data should never be based on absolute values alone, and that the context of soils and sediments is extremely important. In the western part of Felt E there is a clear gradient that shows areas that have held a higher groundwater table in the past. This is also the case in the southern part of Felt E. Most probably, the settlement was located on the edge of significantly wetter land. In the cooking pit rich area in Felt A, the MS readings are distinctly higher than other parts in the northern area and this is emphasised by the phosphate accumulation in the same area.
Figure 17. Spatial distribution of soil phosphate content (Cit-P) over the investigated area at Ørland (n=1632).
Figure 18. Spatial variation of soil MS data over the investigated area at Ørland (n=1632).
Felt A - Activity areas, refuse layers and cooking pit fields Felt A: Hus 01
Samples from Hus 01 (Figure 19) and the potential cultivated layer to its southeast (Intrasisid 101087, sample 152471) produced only a few individual cereal grains and weed/ruderal seeds (Figure 20). This is insufficient material to interpret with respect to the building's usage. Very small amounts of charcoal (<7 ml in most samples) were found in all samples, suggesting that either the longhouse was not burned down or the burnt layers were lost in the machine excavation process.
Two samples on and close to the floor layer (Intrasisid 148321) contained a higher proportion of charcoal than the other samples.
Figure 19. Hus 01 area: Sample points.
Hus 01 was intensively mapped for geoarchaeological analyses in a 7 by 21 m 1 m dense grid (Figure 21). Areas in the eastern and south-western part show the presence of shell bank material, where Cit-P correspondingly drops. Substantial Cit-P concentrations can be found in the north- western part of the structure. Postholes (Figure 22) show general and distinct phosphate accumulations, and are coherent with surroundings, apart from the shell bank area to the east of the house where the concentrations are still high. This building clearly represents an intense phase of habitation.
Judging from MS data, a hearth may have been located in the eastern part of the house (Figure 23).
In addition, the features in the mid-section are somewhat higher and distinctly higher than the surroundings. MS data does not indicate a burned house.
Figure 20. Hus 01 area: Plant macrofossil analysis results summarised.
Figure 21. Felt A, Hus 01, Cit-P survey.
Figure 22. Felt A, Hus 01, Cit-P survey and features.
Figure 23. Felt A, Hus 01, MS features.
Felt A/E: Hus 05 and Hus 14 area
Figure 24 shows the area of Hus 05 and Hus 14. The geochemical results from the broader area of the adjacent Felt E are shown in Figure 66 to Figure 68.
Hus 05
On the basis of the posthole distribution, Hus 05 is probably an elevated house standing on several posts (Figure 25 and Figure 26). Ninety samples were collected in a one by one m grid, so the sample density is quite high. Ten posthole samples were analysed in this context for both soil chemistry and plant macrofossils. In Figure 25 the phosphate data is presented in a map. Both the immediate area of the posts as well as the post fills show accumulation of phosphate concentrations.
The drying or smoking of marine products could account for this accumulation. This is however, an extremely tentative explanation if only based on phosphate accumulation and further archaeological evidence should be sought before drawing any conclusions.
Figure 24. Hus 05 and Hus 14 area: Macrofossil sample points.
MS readings, both for grid- and feature samples are generally quite low (Figure 26), a clear indication of wet soil conditions in this case. Wet conditions in the soil would certainly promote the need for an elevated structure irrespective of the function of the construction. There is an increase in MS to the north of what has been interpreted as a ramp/bridge to the platform. A tentative interpretation could be that the low MS values show the extent of wetter grounds at the time of use, and thus an explanation for the need of a bridge to the structure from the area demarcated by higher MS. The two samples from Hus 05 (Figure 24) which produced plant macrofossils contained only individual weed seeds, and provide no useful interpretative information on the function of the
structure. The building was not burned, and carbonised material produced in the structure. If the structure was indeed a raised pole building, this result is not surprising as overspill or subfossil remains from a raised floor are unlikely to have survived in the supporting postholes on destruction.
Figure 25. Felt A/E, Hus 05 Cit-P.
Although there is the possibility that these features represent postholes for drying racks, the geochemistry strongly indicates the location as either a wetland or waterlogged area, which would perhaps be unsuitable for drying produce or nets.
Figure 26. Felt A/E, Hus 05 MS.
