Case Study 7: A Mesolithic Landscape on the Bexhill to Hastings Link Road, East Sussex

Mike Donnelly (Oxford Archaeology South)


This case study describes the evaluation techniques, excavation methodologies and the post-excavation assessment approaches applied during work on a very large number of in situ flint scatters found during the Bexhill to Hastings link road project in East Sussex (Donnelly et al. 2019).

Around 260 scatters were identified totalling 465,000 struck flints. These scatters were difficult to uncover due to the three-dimensional nature of their preserved burial environment which necessitated a shift in standard operating procedures for their excavation. The potential of these scatters was immediately apparent, and every effort was made to fully record and recover all useful data from them.

The excavations revealed a nationally important early prehistoric landscape with activity dating from the late Upper Palaeolithic through to the Bronze Age, but primarily from the Late Mesolithic, thereafter buried under peat and estuarine silts.

Post-excavation assessment is ongoing but already the analyses have yielded over 100 radiocarbon dates and identified palaeoenvironmental sequences directly associated with the scatters. We also have very detailed spatial information from these scatters that strongly imply a relatively small typical group size, very probably a nucleated family group and this has wide-ranging implications for the interpretation of social structure, population densities and mobility patterns.

Background potential and evaluation

The area impacted by the proposed new road was known to have very good potential for well-preserved early prehistoric activity. However, in many ways the timing and methods used to evaluate this area were inimical to the recovery of flint scatters.

A fieldwalking exercise conducted in the autumn of 2012 recovered very little lithic material. This of course could be taken to indicate low potential. However, it is only by understanding the buried soil profile here that these results could be fully understood. In effect, this lack of material in the ploughzone indicated the potential for very well-preserved flint scatters to survive.

The evaluation occurred during late autumn 2012, a period in which substantial rainfall led to a rise in the local water-table making the identification of flint scatters extremely difficult. Moreover, machine stripping, confined to a single bucket width and tied closely to existing topography, resulted in truncation of numerous scatters (Fig 7.1). Even where the slope matched the underlying prehistoric topography, subtle changes in that land surface would have been removed, resulting in a significant reduction in flint numbers within the evaluation trench.

Only two main scatters were identified during evaluation. However, small numbers of flints were recovered from several locations that were identified as potential scatters, something that was confirmed by the later excavation in all but one instance. Assessment of the evaluation data resulted in a change in the methodology in favour of test pitting during the main phase of works at Bexhill.

Excavation methodologies

The excavation phase of works at Bexhill also included additional evaluation. Hand- or sometimes machine-dug test pits were used alongside augering to determine the potential at impact depth for several large areas. Of these, the hand-dug test pits were excellent for identifying potential scatters while the machine-dug pits had the same problems as the evaluation trenches. Machine-hand hybrid test pits were a suitable alternative, where the bulk of the often very deep overburden was removed by machine and a hand-dug test pit was placed in the base to remove the final layers.

These methods successfully identified an area of dense scatters on one site but very nearly missed a similar density of activity on another. Test-pit grids, like any form of evaluation, can miss key archaeological deposits.

One of the key decisions made at Bexhill was to strategically under-strip areas of the scheme where preserved old land surfaces had been identified. In effect this meant that any impacted areas below 5m OD were under-stripped, leaving a moderately thick layer of overburden (usually peat) over the flint scatters.

Where truncation had already occurred, the sites were stripped normally but the machines followed the predicted buried slope to minimise additional truncation. This strategy not only prevented the truncation of the scatters but also protected the undug portions from trampling by the large archaeological team and disturbance from the heavy pumps utilised for water management.

As with most in situ flint scatters a grid system was used to excavate these deposits. This evolved from individual grid squares covering areas of flint concentrations to finally covering entire sites in a series of 5m grids. We avoided the usual numbering system applied to infrastructure jobs where context numbers tended to be in the high thousands and instead allocated the numbers 1–99 for individual flint scatters. The scatter number was always used rather than the context number of the layer from which the flint was recovered, as in reality all the flint scatters belonged to the same archaeological horizon. The grids themselves used a simple lettering system.

All of this was done to minimise numbering errors and to massively reduce the amount of time needed to record and label these finds, allowing for the complete excavation of numerous scatters.

All significant flints (those greater than 10mm and any identified smaller tools) were 3D-recorded. This was carried out on site with either differential GPS, paired GPS with fixed base stations or robotic total station, all of which were relatively quick to implement on site, with between 600-1000 flints per day being recorded by each survey team. This allowed for remarkable resolution of activities on site and has allowed us to postulate quite complex group dynamics.

One example of this was at scatter 7 where it appears as if two knappers of varying skill attempted the same tasks while a very experienced knapper sat across a hearth from them and corrected any complex errors (Fig 7.3). Such a teaching scenario would have been impossible to infer through whole-earth sampling. Other more prosaic examples include identifying ‘butterfly’ gaps in the scatters where an individual had sat (Fig 7.4).

Half of all spoil from the grid squares was processed on site through a system of mechanical sieves originally obtained by OA for the CNDR project (see Case Study 6). The samples recovered through the sieving system were excluded from microwear analysis to avoid any complications.

Click to view the full image and then zoom in for more detail

Post-excavation assessment

All the lithic material recovered from excavation was assessed. This was carried out by a team of five under the supervision of the author. The assessment was quite detailed and included a full analysis and refitting exercise for six of the scatters. These works are ongoing but there are several key points that this assessment has currently raised.

As mentioned earlier, the resolution of the data has allowed some quite detailed narratives to be constructed and should allow artefact life histories to be examined (Conneller 2005). This should also allow us to link various scatters in the landscape and map these taskscapes in remarkable detail. These activity patterns strongly suggest that small groups of around two to four individuals were responsible for the bulk of the scatters. Some of the denser scatters contrast with this and often display quite low levels of knapping skill. These are currently interpreted as teaching areas and the implications of the two scatter types taken together is that a larger family unit was fragmented into smaller task-specific sub-units.

While this may imply simple environmental determinism, the nuances of the flint scatters, in terms of skills apparent, teaching dynamics etc, strongly suggest aspects of human behaviour that are often absent from early prehistoric narratives. They also indicate the importance of children in any social unit. What is currently absent from the Bexhill Late Mesolithic flint scatters is any form of recognisable base camp of the form that is frequently cited in the archaeological narratives of Late Mesolithic life.


The analysis also allowed us to tentatively identify five chronological phases in the Late Mesolithic assemblages largely based on microlithic forms. These have been tested by radiocarbon dating and have so far produced very encouraging results but much more work is needed here (Fig 7.5). The results of pollen analysis have identified concentrations of charcoal and episodes of woodland clearance that may indicate woodland management or the creation of water-edge clearings.

These results have the potential to greatly enhance our understanding of the Late Mesolithic in South-East England, but caution must be exercised as full analysis is still to be undertaken and it is not clear how representative the density of activity is for other areas.


Conneller, C, 2005 ‘Moving beyond the site: Mesolithic technology in the landscape’ in Milner, N and Woodman, P (eds) Mesolithic Studies at the Beginning of the 21st Century. Oxford, Oxbow, 42–55

Donnelly, M, Champness, C, Davies, S and Boothroyd, J, 2019 'Bexhill to Hastings Link Road: post-excavation assessment and updated project design'. Unpublished report, Oxford Archaeology