Case Study 4: Investigating Prehistoric Landscapes with Lithic Scatters in the Lower Exe Valley

Olaf Bayer (Historic England)


This study used a number of multi-period surface lithic assemblages from a museum collection to investigate the changing character of prehistoric inhabitation in a landscape-scale study area. Key to this was engaging with both the contents (the stone tools of which they are comprised) and contexts (the topographic and cultural locations in which they are found) of lithic scatters.

In addition to lithic analysis research also included visits to scatter locations to understand their landscape setting, and the use of aerial photographs and geophysical survey to elucidate the nature of monuments in close proximity to scatters. The project was guided by four broad and overlapping research questions (see Table 4.1).

Table 4.1. Research questions

Research QuestionRelated questions
1. How can lithic scatters be used to understand the character and composition of inhabitation?
  • What types of activity created the scatter?
  • How intense is occupation?
  • Nucleated/dispersed?
  • Dense/low-level?
  • Focus of activity/background noise?
2. How can lithic scatters be used to understand the temporality of inhabitation?
  • When does occupation occur?
  • What is the duration of occupation?
  • Is it sustained or episodic?
  • A single event/ or an accretion of multiple events?
3. How can lithic scatters contribute to biographies of place?
  • What is there before, during and after a scatter?
  • What do scatters tell us about the prehistory, use and afterlife of monuments/locales?
  • How do scatters relate to monuments or topographic features?
4. What can lithic scatters tell us about scales of mobility/contact?
  • Where do raw materials come from?
  • What state do raw materials arrive/leave in?
  • Is the scatter balanced?
  • What is there and what is missing?
  • How do the scatters in the study area relate to each other/other places within the study area?
  • How do the scatters in the study area relate to places outside the study area?

The project study area was a 4km by 4km block of lowland mid-Devon surrounding the lower Exe basin (Fig 4.1).

The location was selected as it combined a large well-recorded surface lithic collection with an area of intensively studied archaeological cropmarks (Griffith 1990; 2001), and a recently examined late glacial to Bronze Age palaeoenvironmental sequence (Fyfe et al. 2003).

The extent was determined by the core of the Uglow lithic collection. Chiefly collected by amateur archaeologist John Uglow between the 1930s and late 1990s the collection contains assemblages spanning the Mesolithic to the Bronze Age.


A two-stage approach was taken to lithic analysis. An initial assessment was carried out on the entire collection (19,137 pieces) and its associated paper archive. The assessment quantified and roughly characterised each of the collection’s assemblages. This enabled informed decisions to be taken about the size (16,577 pieces) and geographic extent of the sub-sample of the collection to be examined in greater depth.

A relatively pragmatic approach was taken to selecting lithic artefact attributes examined as part of the fuller analysis. Priority was given to analyses (see Table 4.2) which it was hoped would provide answers to the research questions. All data were recorded and analysed in Excel spread sheets which were then exported to ArcGIS for plotting as maps. The basic unit of analysis was the individual lithic artefact, enabling maximum flexibility when interrogating the dataset.

Table 4.2 Lithic recording methodology

Attribute recordedRationale
Collection unit/grid referenceThe most detailed available grid reference for each artefact was recorded. This ranged between a centre point for an entire field to the centre of a 10x10m grid square. This primary location was then degraded to a secondary ‘field level’ location to enable all lithic assemblages of different sizes and from derived from different collection methodologies to be analysed on something approaching an equal basis.
Artefact typeTypological analysis was used to indicate a date and function for each artefact. A small range of chronologically distinctive artefacts were easily assigned to specific periods (e.g. late Mesolithic or Early Bronze Age). These artefacts accounted for approximately 3% of the total collection. Dating was much less specific for the majority of retouched artefacts and debitage.
Dorsal scar type (presence /absence of blade-based working)The proportion of artefacts displaying evidence of a blade-based reduction sequence was used as a crude technological index of assemblage date. Underlying this analysis is an assumption that blade-based working is indicative of Mesolithic, and to a lesser extent Early Neolithic, activity.
Raw materialRaw material type (e.g. nodular flint or greensand chert), colour (e.g. black or brown) and tone (e.g. dark, mid or light) was recorded. This identified the different types of raw material present; linked particular raw materials to certain sources, enabling a discussion of scales and directions of contact outside the study area; enabled a discussion of whether particular types/colours of raw materials were used for particular tasks, or in particular periods.
Reduction sequence (percentage dorsal cortex)The percentage of cortex retained on the dorsal face of each artefact was assigned to one of 6 classes. This enabled a discussion of which stages of the stone working process (from initial preparation of raw materials, to the use and maintenance of finished tools) are present at each scatter, and which are missing and by implication happened elsewhere in the landscape.
RetouchThe presence/absence of artefact retouch was used (in combination with reduction sequence) to identify the stages of the stone working and tool use process represented in a scatter.
Presence/absence of burningBurning was used as a crude indicator of domestic activity, the underlying assumption being that it reflects the accidental burning of artefacts in hearths.
Artefact weightArtefact weight was used alongside artefact count to quantify scatter size and relative proportions of raw material.

In addition to the complexities inherent to working with surface lithic assemblages, historic amateur collections can introduce a number of uncertainties and variables whose influence needs to be understood prior to analysis.

The Uglow collection is the result of over 60 years of fieldwork, conducted by a changing roll call of fieldworkers, driven by evolving research questions, and using different collection methodologies (from casual collection in the 1930s to systematic collection in 10m grids by the 1980s).

Table 4.3 summarises some of the issues that the collection presented and the actions taken to accommodate them.

Table 4.3. Working with lithic assemblages from museum collections

IssueAction taken
Were all lithic artefacts (not just diagnostic pieces) collected and retained?
  • Examine notes and correspondence in the collection archive, and published accounts
  • Interviews with surviving fieldworkers Wherever possible compare the composition of multiple episodes of collection from the same area
Is the location and extent of non-systematically collected scatters accurately recorded?
  • Carry out new systematic surface collection in areas of existing assemblages
To what extent is assemblage size determined by collection methodology?
  • Examine notes and maps in the collection archive to identify collection methodology used for each assemblage
  • Examine relationship between collection methodology and assemblage size
Is it possible to analyse assemblages derived from different collection methodologies on a like for like basis?
  • Over 1000 ‘sub-field’ collection units merged into 63 larger ‘field-level’ collection units to enable their analysis on something approaching a like for like basis
  • During analysis an emphasis placed on the composition of whole assemblages (i.e. proportion of artefact types/traits rather than absolute counts)
To what extent do blank areas reflect the absence of prehistoric activity or of fieldwork?
  • Interviews with surviving fieldworkers Rapid walkover survey in blank areas
  • Consideration of the impact of landscape scale taphanomic processes (colluviation/alluviation) and of twentieth century agricultural practice on scatter visibility/availability
  • Take an underlying assumption that blank areas generally reflect a lack of data
  • Existing scatters seen as positive data and a qualified approach taken when extrapolating from them into blank areas

The distribution of lithic artefacts within the study area is very variable. Particular concentrations occur in areas of the valley floor and on the hilltops on its immediate western edge.

The southern slopes of the Raddon Ridge produced a lesser quantity of lithic material, and even less is present on the Exe/Culm interfluve, the south-west and the western edges of the study area. The identification of chronologically distinctive lithic artefacts and a characterisation of the collection’s debitage indicates occupation in the study area spanning the Mesolithic (potentially the Early Mesolithic) through until at least the Early Bronze Age. This sequence begins in the Mesolithic with a small number of foci of activity mainly on the valley floor and its western edge.

During the Neolithic and Early Bronze Age the intensity of this occupation greatly increases, and extends beyond the valley floor into the wider hinterland.

Against this picture there is a recurring trend suggesting the long-term or repeated use of particular places. Put simply, places with evidence of Mesolithic inhabitation often have evidence for Neolithic and Early Bronze Age inhabitation. This occurs at small scales and low intensities throughout the study area. However, it is most apparent amongst the larger scatters on the valley floor and its western flank, and particularly at scatters N1 and N12 (see Fig 4.1).

Both scatters appear to have been a focus of inhabitation spanning millennia, from the Mesolithic and recurring throughout the Neolithic into the Early Bronze Age. It is suggested that the sustained inhabitation of locales such as these would have necessitated an engagement with both the physical traces, and cultural legacies of past inhabitation. It is amongst these ‘persistent places’ that the study area’s first monuments are built in the late 4th millennium BC (see Fig 4.2).

The study area’s lithic assemblages are chiefly composed of two raw materials, nodular flint and Greensand chert, with much smaller quantities of pebble flint, Haldon chert/flint and Portland chert. There is some evidence that trends in raw material use changed over time. Nodular flint seems to have been used throughout the entire lithic-using sequence whilst Greensand chert appears to have been particularly associated with Mesolithic activity. Portland chert is particularly associated with Neolithic artefacts.

None of the collection’s raw materials occur naturally in the study area; all have been imported. Notwithstanding the possibility that a minority component of the Greensand chert and pebble flint may have come from river terraces relatively close to the study area, most of the raw materials have been transported over distances of between 15 and 40km, and in the case of Portland chert up to 80km.

With the exception of Portland chert it is difficult to pin the collection’s raw materials to particular sources, but linking them to generalised source areas does give an idea of the direction and scale of routine movement or contact. A further observation is that although spatially distant, some of the potential raw material sources (Haldon Ridge and the western edge of the Blackdown Hills) are visible from parts of the study area.

Raw materials arrived in the study area in differing states of modification. Nodular flint and Greensand chert both arrived in a partially modified state, the flint more so than the chert, with the earliest stages in their reduction having occurred elsewhere in the landscape.

The situation with pebble flint and Haldon chert/flint is different, as it seems to have arrived in an almost unmodified state. Portland chert differs again in that it seems to have arrived in a very modified state, either as ready-made cores or finished artefacts, principally mid-Neolithic transverse arrowheads.

Analysis of raw material reduction sequences and the relative proportions of retouched or utilised material indicates that the manufacture and maintenance of stone tools and their subsequent use was widely distributed across the study area. Similarly, the incidence of artefact burning is relatively constant amongst the collection’s assemblages.

In the majority of instances there is little to suggest that specific tasks or activities were concentrated in specific places. However, some anomalous trends hint at a degree of separation and difference in Mesolithic taskscapes.

Both the low incidence of artefact utilisation/retouch, and imbalances in the Greensand chert reduction sequence indicate that whilst the initial stages of the chert working process took place on the valley floor (particularly at scatters N1 and N12), the products of this working were ‘consumed’ elsewhere in the study area and probably beyond.


The study has shown that it is possible to use what many would class as a ‘poor’ data set (historic collections derived from surface lithic assemblages) to further our understanding of prehistoric landscape inhabitation.

The analysis of surface lithic assemblages has successfully helped to address questions relating to the character and composition of inhabitation, biographies of place, and scales of mobility and contact. It has proved much harder to break apart large scatters into anything approaching the constituent events that created them, and to address questions about the temporality of occupation. This is in part due to the necessary aggregation of scatters with multiple collection units into single assemblages for the purposes of the current analysis.

It is considered that should the original detailed resolution be reintroduced to an analysis of these scatters a much more nuanced understanding of their spatial configuration would emerge. However, it is anticipated that a fine-grained understanding of the temporal aspects of these scatters (e.g. continuous vs episodic occupation) might remain elusive.


This project formed the basis of a PhD started at the University of Bristol (2006-8) and completed at the University of Central Lancashire (2008-11). Research was supported by a scholarship from the University of Central Lancashire, and by grants for the initial assessment of the lithic assemblages and for aspects of fieldwork from Devon County Council.

The Uglow collection is held by the Royal Albert Memorial Museum, Exeter.


Bayer, O J, 2011 Lithic Scatters and Landscape: the Mesolithic, Neolithic and Early Bronze Age inhabitation of the lower Exe valley, Devon. Unpublished PhD thesis, University of Central Lancashire

Fyfe, R M, Brown, A G and Coles, B J, 2003 ‘Mesolithic to Bronze Age vegetation and human activity in the Exe Valley, Devon’. Proceedings of the Prehistoric Society 69, 161–81

Griffith, F M, 1990 ‘Aerial reconnaissance in mainland Britain in the summer of 1989’. Antiquity 64, 14–33

Griffith, F M, 2001 ‘Recent work on Neolithic enclosures in Devon’, in Darvill, T and Thomas, J (eds) Neolithic Enclosures in Atlantic North-West Europe. Oxford, Oxbow Books, 66–77