Knowledge of the distribution of Iron Age settlement in Scotland is intrinsically bound up with the Ordnance Survey and the completion of the 6-inch survey in the second half of the 19th century. These provided the foundation for David Christison‘s surveys of the last decades of the century (see Christison 1898); in studying earthworks, this was intrinsically focused upon the Iron Age settlement record. By then, however, it had already been severely damaged by the impact of agricultural improvements from the mid 18th century onwards. The pattern that can be elicited from OS maps is thus already a survival pattern. To take the Lothian Plain for example, little more than thirty earthworks survived, whereas aerial reconnaissance has now recorded over 200 settlement enclosures bounded by ditches or palisades.
To varying extents this differential survival of visible remains affects every dimension of the data. While individual valleys in the Border uplands or individual islands along the Atlantic façade may appear to present a relatively full settlement record, other areas may appear as completely deserted. Some may respond to traditional survey, but in the majority of these cases survey by simple observation of earthwork remains has been compromised by the history of land-use, both ancient and modern. For some areas, as noted for the Lothian Plain, aerial survey of cropmarkings may provide a tool to discover Iron Age settlements, but in its own way this is a technique that is as limited in its application as ground survey, constrained by such factors as the extent of freely draining soils, shifting weather patterns and the cultivation of cereals, to say nothing of any methodological issues (Cowley 2009). These factors play out at national, regional and local scales, affecting the survey record at every stage. Thus cropmarks form mainly in the east of the country, they are three or four times more frequent in Angus than in Aberdeenshire, and while they may be ubiquitous in gravel soils under cereals, less-freely-draining soils immediately adjoining may remain stubbornly blank. As a research tool, therefore, survey must be multi-facetted, fitting appropriate techniques to local circumstance. In this context Historic Land-use Assessment has some utility, for some aspects of the mosaic of land-use patterns that it records can be construed as a mosaic of potentials for different techniques, thus integrating the opportunities afforded by:
- ground observation, shadow photography and LiDAR in rough hill pastures or some woodlands,
- aerial photography on gravel soils under cereals, and
- arable field-walking for artefact scatters, coupled with geophysics.
- Emparkments under permanent pasture, areas of preserved rig-and-furrow cultivation, and many parts of improved ground in the uplands and the length of the west coast, may only respond to geophysical or multispectral scanning techniques.
it is clear that there is ‘no one size fits all‘ survey technique with a universal application in the Scottish landscape. Not only has the land-use mosaic created a series of different survey potentials, but the regional characters of the settlement record dictate a series of local and regional potentials for the application of the different techniques. At its crudest, the remains of a ploughed down broch mound have a different potential to a large fort with multiple ditches. A much slighter round-house might at best be fleetingly visible as a cropmark, but lost in the sampling resolution of a geophysical survey, and contain no artefacts to give its position away to surface collection.
These problems are no less a factor for wider artefact studies drawing upon both stray finds and stratified deposits. The stray finds are mainly the product of the very agricultural processes that are responsible for levelling so many Iron Age settlement earthworks. As a result distributions tend to be heavily skewed towards land that was improved during the 18th and 19th centuries. Metal-detecting, less skewed by agriculture, may serve over time to rebalance this inequality. Collection patterns also come into play, driven by the presence or absence of local collectors as the process of improvement progressed, and by the existence of a local museum playing a part in the preservation of discoveries and the record of their provenance. A further factor is that the soils across the greater part of Scotland are acidic, leading to the near absence of bone artefacts and faunal assemblages of any extent. Furthermore, in many areas, particularly in the arable lands of the south and east, there is a net loss of sediment year on year, whereas in machair deposits in the Northern Isles, the Western Isles and the Inner Hebrides the sediments may be accruing in calcareous sands. The creation of artificially deepened agricultural soils in the Northern Isles is also creating environments where there are not only remarkable levels of preservation of cultural material, but these are interleaved with stratified plant macro-fossils and faunal assemblages that have the potential to provide unparalleled insights into the exploitation of these sites or hinterlands and the agricultural methods deployed there.
Iron Age settlement patterns – understanding the evidence base
Maps of monument distributions are a common way in which spatial archaeological information is presented, and while these are rarely simple or necessarily well understood, they are often seen as true representations of past activity (see Halliday 2011 for discussion of these issues). Frequently, however, these distributions (or maps of recovery as they are now sometimes known) really represent biases inherent in archaeological information-gathering (see Cowley 2011, 45-7 for the example of burnt mounds). The results of aerial reconnaissance for archaeology are a good instance where the relationship between data and how it may reflect the past is not direct. Over the last 50 years this survey method has revolutionised understanding of much of lowland Scotland, recording many thousands of otherwise unknown levelled monuments revealed through differential cropmarking and vegetation. In some areas of southern and eastern Scotland the majority of known sites and monuments have been recorded in this way.
Generally, the effectiveness of aerial reconnaissance is conditioned by rainfall patterns, crop phenology, soil types and land use, producing best results over well-drained soils set to cereal crops in drier parts of the country. Beyond these physical factors, biases introduced by the interests and methodology of the airborne observer can pattern data strongly (papers in Brophy and Cowley 2005; Cowley 2002; Cowley and Dickson 2007). Detailed examination of the structure of the data highlights patterns that are difficult to explain. For example, comparison of cropmarked sites recorded in East Lothian during 1977 and 1992 shows markedly different returns between the two years, with only about a 30% overlap of sites recorded in both years. Thus, while the record compiled over many decades of intensive aerial reconnaissance is remarkable (Cowley 2008), it remains patchy- marked by dense clusters on well-drained soils set to cereals, thin scatters and complete blanks on heavier soils and pasture landcover. While the general processes of cropmark formation are understood (e.g. Wilson 2000), a close examination of the cropmarked information invariably raises questions about how representative it is.
The importance of these factors can be seen in the way in which the apparent clustering of rectilinear settlements in the vicinity of Traprain Law has become established in the literature (Armit 1997c; Armit and Ralston 1997, 218; Macinnes 1984b, 183-6). This ‘fact‘ originates in a distribution map published by Maxwell (1970) based on the limited suite of photographs available at the time. The rationalisation of the cropmarked site record at a regional level drawing together the results of subsequent decades of reconnaissance demonstrates that their distribution is much wider (Cowley 2009; Cowley et al. 2009). Moreover, the localized clusters, including the group near Traprain Law, may reflect little more than the general bunching of cropmarked sites on particular soils. Thus, the Traprain Law group is more likely to reflect the responsiveness of the soils to cropmark formation and to some degree at least the concentration of aerial survey in an area with guaranteed returns (see Cowley 2002 for a commentary on survey bias), than any real concentration of these farmsteads in the Iron Age.
This example illustrates the often selective use of the cropmark settlement record and the lack of contextualization. Later prehistory has suffered from a tendency to be a dustbin for all sorts of sites, generally enclosures, the contexts of which are not known on the basis of analogy with the few excavated sites. This was commented on over 30 years ago by Humphrey Welfare, writing after the first two seasons of aerial survey by RCAHMS (Welfare 1978), and his observations are still pertinent. On the one hand, Welfare pointed out the enormous potential of aerial survey, but identified that the use of frequently highly subjective typologies has only served to leave the interpreted picture fuzzy and confused. He also noted the requirement for research excavations to refine both chronology and cultural assignation in particular. The sheer mass of data that has been collected since Welfare‘s observations has further exacerbated the issues, as little has been marshalled in an interpretative framework, Indeed, in the past material has often been added to the RCAHMS database with imprecise ‘cover-all’ classifications, such as ‘enclosure‘ or ‘cropmark‘, in order to avoid imposing incorrect interpretations on sites. Moreover, even though knowledge has increased with new discoveries and excavations, it is only in the last few years that known material has been systematically revisited to review classifications. Thus, the sites recorded in the NMRS carried imprecise classifications that had limited utility for rationalising sites into morphological groups or robust regional settlement frameworks.
The foregoing discussion applies in equal measure to all archaeological data-sources, and illustrates the importance of understanding how data and knowledge are constructed. This understanding is dependent on thorough analysis of data structure, the rigorous critique of methodologies and insights that historiography alone can bring.
Robust understanding of the nature of survey datasets and their biases is critical
Survey methodologies need to integrate different techniques more effectively, tailored to a knowledge of local conditions.
Geophysical and multispectral scanning techniques offer ways into areas unresponsive to conventional aerial photography, as does fieldwalking and monitoring of metal-detecting finds.
Classification systems for sites (especially cropmarks) require regular critical review in the light of emerging knowledge.