The Archaeological Potential of Secondary Contexts

Module 10 - Introduction

The module reviewed three interpretive models, from the perspective of the secondary context archaeological resource:

1. Gamble’s (1999) heuristic framework, linking locales and regions through the rhythms of social and technological behaviour.

2. White & Schreve’s (2000) framework of hominid colonisation, settlement and abandonment of Britain during the Pleistocene.

3. Ashton & Lewis’ (2002) model of late Middle Pleistocene population decline and possible strategies for re-colonisation.

The most marked differences between the models are in their relative degrees of reliance upon secondary context data. Gamble (1999) and White & Schreve (2000) employ secondary context data, but primarily as a supplementary support for the evidence from primary context sites. For example, White & Schreve’s (2000: 21) assessment of twisted ovate patterning in bifaces relies heavily upon data from primary context assemblages (e.g. Swanscombe (Upper Loam), Wansunt Pit, and Elveden), while mentioning secondary context sites primarily in passing. Gamble’s (1999) model, although multi-faceted, tends to rely upon primary context data in his discussion of social technology (e.g. Boxgrove knapping sequences and the Schöningen spears); paths and tracks (e.g. lithic transfer data from Caune de l’Arago); and generic/specific skills and their demographic implications (e.g. the comparison of Swanscombe and Olduvai Gorge), although secondary context data is highlighted with respect to paths and tracks (e.g. the use of quartzite in the Upper Thames region) and modelling population trends. The key point is that both Gamble (1999) and White & Schreve (2000) utilise restricted elements of the secondary context record. This is not a cause for criticism in itself, but it does mean that they are not required to deal with the wider problems associated with secondary context data — namely, unsystematic sampling at the regional scale, fragmented preservation, and derivation of the artefacts. By contrast however, the Ashton & Lewis (2002) model is explicit in its usage of regional, secondary context data. It therefore has to, and does, deal with the problems of secondary context data.

Overall, the Gamble (1999) model can be applied to regional, secondary context data sets, although in a piecemeal rather than wholesale manner. A number of small modifications have been proposed here, to account for the structure and character of the secondary context resource. As with the other models, this review of Gamble (1999) has illustrated that the most directly transferable models are those explicitly designed to work with secondary context data sets (e.g. Ashton & Lewis 2002). Models principally concerned with primary context data sets inevitably require greater degrees of adaptation. It is clear that: firstly, the secondary context archaeological resource has a wealth of potential value; and secondly, for this value to be exploited fully requires the development of specific and appropriate methodologies. Examples of these are therefore presented below.

Interpretive Frameworks for Secondary Context Archaeology
This module does not present a conceptual, interpretive framework in the style of Gamble (1999). Its intention is rather to highlight a series of key research questions in Lower and Middle Palaeolithic archaeology, and indicate whether those questions can or cannot be explored through regional, secondary context data sets. A series of specific interpretive frameworks are therefore discussed, alongside methodological examples of work undertaken as part of this project. Where relevant, links between the frameworks are identified, and preliminary connections are also drawn between the secondary context frameworks and the primary context archaeological resource. Finally, it is emphasised that the interpretive frameworks are intended to profitably exploit the structure of the data, and not to underpin a specific perspective of Palaeolithic societies.

The key research questions addressed here are adopted from the Research Frameworks for the Palaeolithic and Mesolithic of Britain and Ireland document (The Prehistoric Society 1999), and are divided between three themes (only the research questions relevant to the Palaeolithic were included here):

1. Colonisation and recolonisation:
1.1 Establishing the pattern of human interaction with, and possible impact on, faunas and vegetation of Britain and Ireland. This involves determining the earliest occupation in the Middle Pleistocene and relating this to well-dated and stratigraphically secure deposits in Britain, Ireland and Northern Europe.
1.2 Tracing through the archaeology, and where available the physical anthropology and bio-molecular evidence, the relations between Britain, Ireland and NW Europe. This should be done at various periods to determine the variety of source populations.
1.3 Establishing with greater precision the arrival of anatomically modern humans in Britain and tracing the patterns of Neanderthal extinctions in our data.
1.4 Establishing the pattern of recolonisation at various time periods but particularly following the Last Glacial Maximum at 18,000 uncal BP and throughout the Mesolithic as population expanded.

2. Settlement patterns and settlement histories:
2.1 How much of the Pleistocene actually saw human occupation in Britain and Ireland?
2.2 Were there, through time, appreciable shifts in the length of occupation as humans coped better with the recurrent challenges of climate change?
2.3 To what extent did successive populations in the Palaeolithic and Mesolithic intensify their subsistence behaviour and so develop new niches for adaptation?
2.4 Does the British and Irish evidence indicate a full settlement system within our current geographical limits? Alternatively how far are the settlement data illustrative of a logistic system of special task sites and where major base camps were located either on the continental shelf or on the continent?
2.5 What changes in landscape use and the organisation of technology are indicated by the provenancing and quality of lithic raw materials?
2.6 How soon after areas of Britain and Ireland became available for settlement (following glacial retreat and the establishment of useable food resources) were they in fact recolonised?
2.7 How do we compare activities which took place at cave and open sites and place them within a differentiated settlement system?

3. Social organisation and belief systems:
3.1 Apply the chaîne opératoire concept to the analysis of a ‘social technology’ rather than just to the mechanics of manufacture. The opportunity to include experimental studies, including use-wear analysis promises to make this a key area in the archaeological study of the production and consumption of material culture.
3.2 Animal bones need to be assessed as symbolic resources used in the reproduction of society as well as sources of calories.
3.3 Transference of social organisation into spatial patterning on camp sites needs to be systematically considered.
3.4 What was the regional scale of Palaeolithic and Mesolithic social systems as revealed in such concepts as Grahame Clark’s social territories (1975) and measured through artefact studies?

It is clear that many of these research questions cannot be answered from regional, secondary context data sets. This is due to a number of factors, including:

i. The lack of geochronological tools of sufficiently high resolution. This is particularly pertinent to the issues of the recolonisation delay after glacial retreat (2.6), and in combination with the problems of artefact re-working, the issues of recolonisation following the Last Glacial Maximum (1.4). In the case of issue 1.4, although high resolution geochronological tools are available for the last 40,000 years (e.g. Accelerator Mass Spectrometry dating), these refinements are of limited application due to the persistent problem of temporal artefact re-working. For earlier periods, the error ranges associated with coarser resolution geochronologies (e.g. OSL) and temporal artefact re-working were of comparable magnitudes – see Module 8. This is not the case for later periods, and the resultant disparity limits the applications of secondary context data to the understanding of high-resolution processes of re-colonisation, in which centennial timescales are of considerable significance (e.g. Housley et al. 1997).
ii. The loss through derivation of spatial information with respect to potential associations between different elements of the artefact record. This is particularly relevant to the issues of spatial patterning and social organisation (3.3), animal bones and symbolism (3.2), and the intensification of subsistence behaviour (2.3). In the case of issue 2.3, derived faunal elements frequently bear no evidence of either the agent of death (e.g. carnivore/hominid/natural causes) or whether hominid involvement was as a primary hunter or a secondary scavenger. Equally importantly, derived faunal remains cannot be used to construct age profiles which might indicate intensification of specialisation, since all the remains cannot be assumed to be contemporary. With respect to issue 3.2, if unmodified animal bones were being used as symbolic resources, there is no preserved contextual evidence within secondary contexts to demonstrate this.
iii. Inability to classify sites and settlements, due to re-working. This is relevant to the issues of contrasts between cave and open sites (2.7), and the nature of settlement systems (2.4), and is due to the simple fact that derived artefact assemblages are not sites. Moreover, the selective recovery of different artefact types from secondary contexts has severely restricted our ability to quantitatively measure differences between assemblages in space and time.

However, despite the inapplicability of regional, secondary context data sets to the research questions outlined above, there remain a wide range of issues that can be profitably addressed through the data sets. These questions highlight the key characteristic of the data, namely its extensive spatio-temporal coverage, despite the relatively coarse resolution associated with individual deposits and assemblages. This facilitates the investigation of:

i. The earliest occupation of Britain (1.1), as exemplified recently by the assignment of derived artefacts from the pre-Anglian sands and gravels at Warren Hill to MIS-13 (Wymer et al. 1991; Wymer 1999: 140), thus supporting other evidence for the earliest occupation of Britain, for example from Boxgrove (Roberts & Parfitt 1998).
ii. Links between Britain and northwest Europe, as traced through the archaeology (1.2), and recently explored by White & Schreve (2000).
iii. The modern human arrival in Britain (1.3). Although there is inevitably an issue of geochronological resolution (see the comments above), the difference here is that the focus is upon identifying the earliest dated example, rather than tracing a diachronic process through a number of dated occurrences. It is for this reason that the data cannot be used to address the second aspect of the research question (the pattern of Neanderthal extinctions in Britain).
iv. How much of the Pleistocene saw human occupation of Britain (2.1)? It is stressed that the data will provide a measure that operates at the regional scale, and at the MIS-level of temporal resolution. Yet while this is not generating a high-resolution model, it is the scope and abundance of the secondary context data that allow this work to undertaken at all. Estimates of occupation during the Pleistocene cannot be based on the small number of primary context sites currently available.
v. Changes over time in the length of occupation (2.2). These changes can only be mapped at the MIS-level, as this reflects the finest level of geochronological resolution that can currently be applied to regional data sets. Nonetheless, these data still indicate trends through time, as with the evidence for population decline over the Middle Pleistocene, modelled by Ashton & Lewis (2002) for the Middle Thames valley.

Despite the time- and space-averaging of the data, it is also possible to explore other questions that have traditionally been addressed through primary context data sets:

i. Raw material provenancing (2.5). Although the derived nature of the assemblages prevents the identification of specific raw material transfer distances, the structure of the data (e.g. the association between the assemblage and the fluvial system) supports a generic modelling of raw material procurement strategies based upon regional bedrock geology, the lithology of fluvial sediments, and the proportions of lithic types evident in the artefact assemblages. The tendency will be for the data to provide negative evidence with respect to long distance transfers and procurement, although the potential remains for the highlighting of geographically distant sources through the presence of exotics.
ii. The chaîne opératoire and social technology (3.2). While there is obviously no evidence for operational sequences within secondary contexts, there is an extensive body of material evidence that can provide indicators of the manufacturing process, and document distinctive aspects of artefacts (e.g. twisted profiles on ovate bifaces). The interpretation of these data will be dependent upon the specific interpretive and conceptual frameworks (e.g. compare McNabb & Ashton 1992; McNabb 1996; White 1998a, 1998b; Gamble 1999; Wymer 1999; Wenban-Smith et al. 2000), but all of these investigations do not need to be solely based upon the archaeology of primary contexts.
iii. The regional scales of social systems (3.4). This is assumed here to refer to the scale of the alliances and contacts that formed a part of hominid and social systems in the Palaeolithic. Since this is most commonly expressed through evidence for shared symbolic material culture (e.g. the Venus figurines of the Upper Palaeolithic), this has currently relatively little relevance to the Lower and Middle Palaeolithic. However, in those instances where the patterns in material culture are of a different (greater) scalar magnitude to that of the fluvial systems within which the derived archaeology occurs, recovered artefacts can still indicate generic scales of contact and interaction, although they cannot be tied to a specific time and place. Unfortunately, the probability of many of these artefacts surviving derivation and re-working is relatively limited.

Based upon these applications for regional, secondary context data sets, an interpretive framework is proposed. It is not intended to support a particular perspective with respect to the understanding and interpretation of Palaeolithic societies, but is rather designed to identify the structure of the resource, map relevant questions onto appropriate data, highlight linkages, and illustrate connections between primary and secondary context archaeology. The framework is summarised below.

The framework’s analytical units are defined in terms of the maximum spatial or temporal resolution that can be currently achieved. Spatial units are defined in terms of regions, which are defined here as the setting of the secondary context archaeological record – individual river systems. It is therefore recognised that regions will differ in size (e.g. compare the Thames and the Ouse), but stressed that as fluvial drainage systems they are structurally comparable. Temporal units are defined on the basis of the MIS record, either as single marine isotope stages or as sub-stages (e.g. Schreve 2001). The resolution of the analytical units varies in response to the behavioural domain that is being investigated. For example, in the case of the ‘Colonisation & Occupation’ domain, it is proposed that the population trends can only be modelled at the regional level (e.g. comparing population trends of the Thames region with the Solent River region) and on the basis of MIS-stages (e.g. comparing MIS-12 with MIS-10). This reflects the fact that regional populations should not be modelled on the basis of selected assemblages within that region, and that the sedimentary deposits representing portions of an individual MIS cannot be equated (and often not even identified) between different locations or regions on the basis of current geochronological tools.

There are not a limited set of specific methodologies associated with this framework, indeed it is hoped that this document will encourage the development of new approaches. However, and based on the earlier discussions of the three models, three sample methodologies are briefly outlined below, to illustrate some of the key considerations that need to be born in mind when dealing with these types of data (these considerations are not presented in any order of significance).

Population Modelling/Regional Distribution Patterns
There should be considerable overlap between population modelling and the analysis of regional distribution patterns, although there will inevitably be variations in terms of the focus of the interpretation of the results. Key considerations should include:

1. The availability of a robust geochronological framework.
2. Mapping of the secondary context deposits (e.g. fluvial sediments), aggregates extraction sites, urban development areas, and archaeological assemblages.
3. With respect to population modelling, explicit definition of the model’s assumptions regarding the links between artefact density values, rates of artefact discard, and population data, and how (or if) these change over time.
4. Standardisation of the data to take account of local and regional variations in deposit preservation, aggregates extraction activity, urban development, and antiquarian collector activity.
5. Investigation of the impact of temporal (vertical) re-working of archaeological materials, and the impact of the process upon the recorded results. Key factors in the assessment of the degree of re-working should include the relative preservation of the deposits associated with each terrace unit, and the physical condition of the artefacts.

Assemblage Characterisation
There is potential for considerable variation in approaches to assemblage characterisation, and the following points are therefore restricted to generic observations:

1. The availability of a robust geochronological framework.
2. Assessing the degree of spatial reworking of what is inevitably a palimpsest assemblage. Key factors would include the physical condition of the artefacts, and the sedimentary contexts.
3. Assessing the potential for vertical re-working (based on the location of the assemblage, the preservation of associated deposits, and the regional solid geology) and recognising the problems inherent in the analysis of a heavily time-averaged assemblage. This should be undertaken on case-by-case basis, and the option of rejecting an assemblage for analysis should always be available.
4. Documentary evidence (where available) relating to the collection of the assemblage and the potential presence of sample bias influencing the composition of the extant material.
5. Awareness of the range of data that can be recorded, which includes overall typology, evidence of the manufacturing process, physical condition (the état physique), raw material use (see also below), and non-standard typological variants (e.g. twisted profiles and pronounced asymmetry).

Raw Material Distributions
These approaches are dependent on an explicit understanding of the scalar elements of the analysis. Key points include:

1. The availability of a robust geochronological framework.
2. Modelling the theoretical spatial catchment of the assemblage or assemblages (e.g. Chambers in prep.).
3. If vertical derivation of the assemblage is demonstrated or suspected, then an evaluation of the potential for changing raw material availability over time must be undertaken (see White 1998b for a detailed example of this issue with respect to the Swanscombe bifaces).
4. Characterisation of the potential raw material sources, based on the regional solid geology and the lithology of the drift deposits.
5. Assessment of the proportions of different raw materials (where relevant) within the assemblage or assemblages (see also above).

It is emphasised that there are number of feedback loops between the different components of the interpretive framework. These loops illustrate the linkages between the themes, and demonstrate how results in one area will assist in the analysis and interpretation of other data. Generic examples might include how changes through time in population data (e.g. the decline in Middle Pleistocene populations documented by Ashton & Lewis (2002) in the Middle Thames) could be used to explore shifts in technological practice (e.g. the first appearance of Levallois technique in MIS-9/8). Alternatively, whether regional variations in the chronology of population peaks and troughs (e.g. the contrasts between the Solent River data (Hosfield in prep.) and the Thames data of Ashton & Lewis (2002)) could be considered in terms of variations in one or more elements, ranging from technology or raw material procurement to landscape location within the wider context of northwest Europe.

Following the tacking concept of Gamble (1996), linkages can also be drawn between secondary context data and their interpretive frameworks, and the on-site investigations that have tended to dominate Palaeolithic research. One of the best explicit examples of this has been Ashton & Lewis’ (2002) use of data from European Middle Palaeolithic sites (e.g. La Borde, La Cotte de St. Brelade and Wallertheim) to explore their long-term population patterns in the Thames valley. However, this does not have to be a one-way relationship, as Gamble reminds us:

“Population would ebb and flow into and out of the northern environments, as represented by Swanscombe, and this process [only observable through regional, secondary context data] will provide the strongest archaeological signature of variation in behaviour [traditionally the preserve of on-site studies] at the regional scale.”
(Gamble 1999: 140)

As a specific example, the fluctuating populations of Britain during the Middle Pleistocene (Ashton & Lewis 2002; Hosfield in prep; this Module) are an indicator that well preserved snapshots of behaviour may just be a day in the life, and does not necessarily prove the long-term presence of a well adapted hominid species.

In conclusion, this module began with a review of three recent models of Palaeolithic archaeology and hominid behaviour. The review indicated that regional, secondary context data sets are best investigated through the development of specifically-tailored frameworks and methodologies, although the three models did indicate profitable areas of research. Modified frameworks and methodologies have therefore been proposed, and it is hoped that these clearly demonstrate the considerable potential of archaeological secondary contexts.

Module 10 Outline

Module 10 Results

Module 10 Interm Report

References

Return to top