Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

download Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

of 15

Transcript of Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    1/15

    Journal of Archaeological Scienc e 1983,10,267-28 1

    The Interpretation of Stratified Sites:

    A View from Underground

    Paola Villa” and Jean Courtin*

    This paper reports on an experiment designed to study the role of trampling in the

    vertical dispersal of artifacts in the soil, and in the mixing of originally separate sets

    of materials. The experiment is part of a study of the archaeological stratigraphy and

    patterns of site use at a large stratified cave n southern France. The experiment was

    designed to replicate conditions prevailing at the cave. The results strongly suggest

    caution in the interpretation of living floors and stratified assemblages in sandy

    deposits.

    Keywords:

    EXPERIMENTAL ARCHAEOLOGY, CULTURAL STRATIFICA-

    TION, SITE FORMATION PROCESSES, TRAMPLING.

    Intruduction

    This article presents the results of a trampling experiment designed to study processes of

    formation of the archaeological record at stratified sites. The hypothesis to be tested was

    that trampling (a normal activity of prehistoric inhabitants of a site) can cause vertical

    dispersal of artifacts in the soil and can create false stratigraphic associations. The first

    part of this article provides the justification, background, and rationale of the experi-

    ment. In the second part we present the experimental procedures and analysis of results.

    The specific aim of the experiment was to provide information to aid in the interpre-

    tation of dispersal patterns and relationships between superimposed sets of materials at

    a particular site. We believe, however, that the results of this experiment are relevant to

    general studies of the formation of archaeological deposits. This information can contri-

    bute to a better assessment of the research potential of stratified sites and to the develop-

    ment of more rigorous methods for drawing inferences from archaeological remains.

    A case study

    Background

    The trampling experiment reported here is part of an ongoing study of the stratigraphy

    and patterns of site use in Fontbregoua Cave, about 100 km NNE of Marseille, southern

    France. Fontbregoua is a large cave (about 250 m2) under excavation since 1970 by

    Jean Courtin. It has yielded a long cultural sequence with Upper Paleolithic, Mesolithic,

    and Neolithic deposits totalling more than 9 m. The uppermost levels contain sparse

    Chalcolithic, Bronze Age and historic materials.

    aDepartment of Anthropology, University of Colorado, Boulder, Colorado

    80309.

    *Direction des Antiquit& Prehistoriquesde Provence,Alpes et CBte d’Azur,

    21-23 Boulevard du Roi Rend, 13617Aixen-Provence, France.

    267

    0305-4403/83/030267+

    15 03.00/O

    0 1983 AcademicPress nc. (London) Limited

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    2/15

    P. V ILLA AND J . COUR TIN

    The Neolithic layers (4 m) have yielded more than 24,000 objects including pottery,

    grinding stones, sickle blades, remains of domestic and wild faunas, and carbonized

    seeds of domestic wheat (mostly bread wheat), barley, legumes and acorns. The oldest

    Neolithic occupation has a radiocarbon date of 4750f 100 bc. Below, and separated by

    a layer of culturally sterile sand, are Mesolithic deposits with geometric microliths,

    abundant bird bones but sparse mammal bones, remains of fresh water turtles and fish,

    and wild legumes. The Upper Paleolithic occupation, with a radiocarbon date of 9250&

    100 bc, is known only from a very limited test trench. Bedrock has not been reached

    (Courtin 1975, 1976, 1978; Courtin & Erroux 1974; Cheylan & Courtin 1976).

    Significance of the site

    Being a deeply stratified site with a long history of occupation and repeated use, Font-

    bregoua provides data for the study of change and continuity in material culture. The

    site is especially important for a number of reasons. First, the cultural sequence spans

    the transition from hunting and gathering to farming, a turning point in human history.

    Second, the site has yielded a very large fauna1 sample and botanical remains for the study

    of hunting and farming practices. In particular, Fontbregoua is one of only four Ear ly

    Neolithic sites in southern France known to have yielded plant materials (carbonized

    cereal grains and legumes). Finally, an especially interesting feature of the site is con-

    tinuing use of the cave across the Neolithic transition. The continuing use of caves-

    together with open air sites-is not unusual in the Ear ly Neolithic of the Western Medi-

    terranean and provides an interesting contrast to the situation in the Near East, where

    caves, inhabited in Epipaleolithic times, were only sporadically used by Neolithic

    farmers. In southern France farming first appears on the coast at around 5500 bc. Some

    of the major domesticates-wheat, barley, perhaps sheep-are of exotic origin; their

    wild ancestors are only found in the Near East. Thus, the evidence suggests that farming

    was not a fully local, independent development. Its adoption must have been, at least to

    some extent, the result of a process of transfer of new technologies, materials and ex-

    pertise. The spread of the new economy eventually lead to the establishemnt of per-

    manent, year-round villages and to the abandonment of caves as habitation sites. How-

    ever, for the first two millennia after the appearance of farming, caves continued to be

    used side by side with open air sites. Storage or trash pits and fireplaces, with thick ash

    accumulations, are a conspicuous feature of the Neolithic deposits at Fontbregoua

    suggesting prolonged periods of habitation. By contrast, during the Mesolithic, the cave

    appears to have been used for short v isits only, The study of changing or continuing pat-

    terns of site use through time is a major goal of the Fontbregoua research project.

    Defining the Problem

    Assemblages and layers

    The first task facing the investgator of a deeply stratified site is to partition the deposits

    into layers or levels representing successive time units (Wheeler, 1954, p. 43). It is then

    possible to distinguish superimposed sets of material and group items into units of

    association (assemblages) to be analysed separately.

    Stratified sequences in prehistoric sites wil l often consist of a succession of contrasting

    types of deposits, bounded by discontinuities reflecting changes in type of sedimentation,

    temporary halts in deposition or erosional episodes. Within single geologic units, one

    or more units of cultural stratification may be distinguished. In the latter case, their

    boundaries are defined by the presence of steri le zones separating vert ical concentrations

    of materials, or by distinct interruptions in the accumulation of individual man-made

    deposits, such as ash or charcoal layers. Ideally each layer or level represents a well-

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    3/15

    STRATIFIED SITES

    269

    defined time unit bounded by clear discontinuity surfaces or by sterile zones. In fact,

    gradational boundaries are common. When the cultural material forms unpatterned

    vertical scatters within homogeneous deposits, arbitrary levels are used. Collections of

    objects from single stratigraphic units form separate analytical units and are called

    assemblages. In short, archaeologists subdivide and interpret a stratified sequence using

    criteria of diverse kind, based on a combination of natural and cultural processes of

    deposit formation(Wheeler, 1954, ch. 4; Michels, 1973, chs. 3 and 4; Harris, 1979, ch.

    5; Straus, 1979; Laville ef al., 1980, pp. 9, 13, 152). These same criteria have been used

    at Fontbregoua.

    Clearly the definition of stratigraphic units and their boundaries is important for the

    definition of assemblages. Boundaries between layers may be sharply delineated and

    continuous; they may also be diffuse, discontinuous and irregular. The distinctness,

    lateral continuity and temporal significance of these surfaces should be described and

    documented (Harris, 1979, pp. 49-80). Methods used for defining layer boundaries and

    for segregating superimposed sets of materials into assemblages should be crit ical ly

    evaluated.

    The assemblage concept plays an important role in Old World archaeology. In the

    “assemblage approach” of Palaeolithic archaeology, pioneered by F. Bordes (Binford,

    1981, p. 183; 1982) assemblage types are defined and compared on the basis of varying

    frequencies of artifact types. Each artifact contributes to the description and diagnosis

    of the assemblage. To be sure, assemblage typology and quantification are not emphasized

    in Neolithic or proto-historical archaeology. In these disciplines cultural-historical

    groupings and sequences are worked out on the basis of comparative studies of selected

    classes of artifacts. More attention is given to diagnostic artifacts than to frequencies of

    artifact types.

    However, in behavioral studies the assemblage again is the basic unit of analysis.

    Every item in the assemblage contributes to the diagnosis and description of the whole.

    Implicitly or explicitly the material is referred to a single episode of occupation or, at

    least, to a single mode of site use by a specific group of people. For instance, in studies of

    seasonality, diet, and subsistence patterns collections of bones from the same layer or

    level form the basis for estimating the minimum number of individuals represented in

    fauna1 remains and for analysing hunting or herding practices. In spatial analysis we

    try to distinguish areas in which various daily life activities were carried out. Again,

    sets of associated materials are the basic unit of analysis.

    Clearly, if all the elements of a set contribute to its diagnosis, then we should make

    sure that they do, in fact, belong to the ensemble. For instance, Grayson (1979, pp. 205,

    213) has suggested that the way a stratified sequence is subdivided into units of analysis

    seriously affects sample size and, consequently, measures of relative abundance of faunas.

    Sampling error may distort our understanding of past patterns of human behaviour; its

    effects should not be underestimated.

    In short, we need to ask:

    - Do strata boundaries, observed during the excavation, represent meaningful breaks in

    a sequence of occupations ? Are our samples discrete units or arbitrary slices of a tem-

    poral continuum?

    -

    Do assemblages from individual layers represent the discrete residues of distinct

    human groups or are they the aggregate of different episodes? Can we be sure that our

    stratigraphic units relate to a single mode of site use, if not to a single episode of

    occupation ?

    It is clear that the answers to these questions have an important bearing on studies of

    site use patterns. If we can define single phases of occupation, then we can study activi-

    ties and patterns of site use on a fine time scale. If our information is inadequate for

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    4/15

    270

    P. V ILLA AND J . COU RTIN

    such definition, then we should look only at gross patterns of change and continuity,

    focus our attention on the general picture, and develop detailed behavioural interpreta-

    tions only in the case of features and site structures-such as storage pits-that clearly

    associate contemporaneous materials.

    Assemblages and time.

    Confronted by the problem of defining the temporal duration of their samples,

    archaeologists distinguish between homogeneous and mixed assemblages (some New

    World archaeologists use the term “component”). Homogeneous assemblages are associ-

    ated sets of artifacts and bones, meaningful samples of a segment of a community. Mixed

    assemblages, on the other hand, are the telescoped residue of once separate and

    dissimilar events (Michels, 1973, pp. 22-25; Thomas, 1979, p. 231).

    This simple distinction hides a more complex situation. Homogeneous is an attribute

    that encompasses a graded series. The degree to which aggregates from individual layers

    can be treated as homogeneous assemblages depends upon the time scale we wish to

    adopt. An example from classical archaeology wil l serve to make this clear.

    Etruscan chamber tombs were often used for several generations. Pottery found in

    such a tomb may include, say, Middle Corinthian ware, bucchero ware and late black-

    figure vases. These grave goods span a period of over a century. On a coarse time scale

    the material is associated because it is found in the same container (a tomb) and is

    homogeneous in that it is a typical and meaningful sample of Etruscan culture and

    burial customs. On a fine time scale the late black-figure vases (first decade of the 5th

    century BC) and the Corinthian ware (first half of the 6th century BC.) are not con-

    temporaneous. They are in fact related to different events: the death and burial of a rich

    landowner and, much later, of his equally wealthy great grand-children. Seen from this

    point of view the tomb material forms a mixed assemblage.

    In prehistoric archaeology the strength of association of materials is traditionally

    evaluated in stratigraphic terms. Layers or levels are treated as containers of a sort

    (Spaulding, 1960, p. 211). But assemblages from individual layers represent temporal

    samples whose duration is very difficult to measure. Unlike historical materials, Stone

    Age materials are not clear time markers except on a very gross scale. Formal changes

    through time are too rare or unstable to allow time ranking with more than very few

    classes. Thus, neither stone artifacts nor Neolithic pottery provide direct criteria for

    separating the residues of different episodes of occupation that may have accumulated

    in a single layer. The time interval between the deposition of the first and last item in the

    set can rarely be specified. In most cases we do not know the scale on which an assem-

    blage is homogeneous. Fine dissecting of strata does not provide the degree of time

    control required by behavioural studies, because geological cycles often cover spans

    of time that are too long for the study of human activit ies on a fine scale (Bordes

    et al.,

    1972, pp. 17-19). The Neolithic deposits at Fontbregoua have an average sedimentation

    rate of 1 cm per 17 years. Thus, it is likely that materials now aggregated in a single

    layer were, in fact, discarded during separate phases of occupation and possibly different

    modes of site use. Estimates of sedimentation rates in other Stone Age caves indicate

    equally low values. One cm of deposit may represent from 5 to 167 years, with an average

    of about 14 years (Speth & Johnson, 1976, pp. 47-48).

    Furthermore, layers appear to be rather leaky containers. Recent evidence provided

    by conjoined pieces in Old World sites has shown that vertical migration and dispersal

    of artifacts across different cultural levels is a fairly common phenomenon (Cahen &

    Moeyersons, 1977; Van Noten

    et al.,

    1978, 1980; Bunn

    et al.,

    1980; Villa, 1982; Del-

    Porte, 1982, p. 161; see also Siriainen, 1977; Rowlett & Robbins, 1982). It is increasingly

    clear that in many stratified sites assemblages are not patterned sets of artifacts, tool-

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    5/15

    STRATIFIED SITES 271

    kits that can be referred to the activities of a specific social group (Binford & Binford,

    1969; Sackett, 1973; Whallon, 1973). Rather, they are aggregates of individual pieces

    with individual histories and origins (Binford, 1982, pp. 17-18).

    To understand patterns of site use, we must first investigate processes of artifact

    accumulation and dispersal in the soil. Stratigraphic studies have traditionally focused on

    the geologic context of cultural materials. Classic sediment analyses provide important

    data for reconstructing paleoclimatic histories and for temporal ordering of assemblages

    (e.g. Laville et al., 1980). However, the burial and stratification of cultural materials at

    repeatedly occupied sites, such as caves, are due to a complex interaction of human

    activit ies and geological processes. Comprehensive approaches to cave sediment forma-

    tion stress human and biological influences (Butzer, 1978, 1981, 1982, ch. 7; Goldberg,

    1979, 1980; Courty & Raynal, 1982).

    Many techniques and procedures used in the study of site formation and modification

    are borrowed from the earth sciences. Others are more strictly archaeological in origin,

    in that the information is provided by the artifacts themselves, their interrelationships

    and precise position in relation to strata boundaries. These latter methods of stratigraphic

    analysis are based on the use of vertical projections and plots of conjoinable pieces (e.g.

    Villa, 1982, pp. 280-281). Experimental and ethnographic studies of depositional en-

    vironments and disturbance processes have been designed to answer specific archaeolo-

    gical questions (Jewel1 & Dimbleby, 1966; Isaac, 1967; Stockton, 1973; Cahen & Moyer-

    sons, 1977; Gifford & Behrensmeyer, 1977; Reynolds 1974, 1979; Schick in Bunn

    ef al., 1980; see also Schiffer, 1975, p. 841; Lewarch & O’Brien, 1981, 307-311).

    These different approaches must be combined and integrated to reach a better un-

    derstanding of processes of cultural stratif ication and deposit disturbance. Our paper is

    not intended as a review of all these methods. The potential of geoarchaeology and the

    role of sediment analysis in such studies have been illustrated with concrete examples

    by Butzer (1982, pp. 77-122). We present here an example of the experimental approach

    that c losely reflects our professional interests. This approach should be seen as comple-

    mentary to sediment analysis. As indicated below, the need for experimentation arose

    as a consequence of studies of conjoinable pieces and of vert ical projections of the

    Fontbregoua material.

    Vertical displacement

    of

    archaeological remains at Fontbregoua

    The study of conjoinable pieces found at Fontbregoua (mostly pottery sherds and bones)

    is in progress. A preliminary analysis indicates that sherds belonging to the same pot

    have a vertical separation of up to 25-30 cm. Vertical dispersal often occurs across what

    during excavation had appeared to be distinct cultural horizons.

    A variety of factors may be suggested to explain the vertical displacements observed

    in the Fontbregoua deposits:

    (1) Soil fauna. Animal burrows have been observed especially in the upper (Chalco-

    lithic) layers; they are occasionally found in the lower units. However, these disturbed

    areas are easily distinguished from the surrounding intact matrix and pieces from these

    areas have been kept separate from the remaining materials. Earthworm castings are

    uncommon; however, the occurrence of diffuse layer boundaries suggests earthworm

    activit ies (Jacques E. Brochier, pers. comm.).

    (2) Tree roots. The cave ceiling collapsed in ancient times; thus the central cave area

    was exposed to sun and rain. When the excavation began, four live oaks and one cherry

    tree were growing in the cave. Their roots have been found 4 m below the surface. Since

    old roots rot away and may not leave visible traces, disturbance by tree roots may be

    expected even in layers where they have not been observed.

    (3) Alternate wetting and drying. According to Cahen dc Moeyersons (1977) alternate

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    6/15

    272

    P. V ILLA AND J . COUR TIN

    wetting and drying of sediments (by percolating rain water and sun) can lead to vert ical

    movement of artifacts, especially if sediments are unconsolidated due to biogenic activi-

    ties. It is interesting to note that some of the Fontbregoua layers and the sediments used

    in Cahen & Moeyersons’ experiment have a very similar grain size distribution (analysis

    of the Fontbregoua sediments is in progress). We have never observed clay cracks.

    (4) The digging and leveling activities of the prehistoric inhabitants. Pits, depressions,

    and large hearths in deep hollows are common in the Neolithic layers. Digging and

    leveling activities, necessary to the construction of such features, disturb older deposits

    and cause rearrangement and redistribution of archaeological material on younger sur-

    faces. It is possible to imagine situations in which pits or dug-out depressions will not

    leave visible traces in the soil. If the pit was in use for a short time, if it contained

    perishable material, if the fil l is similar in colour and texture to the layer in which the

    pit was dug, the feature will not be recognized and only the displacement of conjoinable

    pieces will betray the presence of disturbances.

    (5) Trampling. Recently several scholars have suggested that trampling is a kind of

    occupational disturbance that is not easily recognized and may cause considerable ver-

    tical displacement of artifacts (Stockton, 1973; Hughes & Lampert, 1977; Gifford &

    Behrensmeyer, 1977 ; Gifford, 1978; Butzer, 1981, p. 155). At the Jean Cros rock shelter

    in southeastern France sherds of the same pot have been found in three superimposed

    levels; their vertical dispersal was attributed by the archaeologist to trampling by the

    prehistoric inhabitants (Guilaine, 1979).

    A trampling experiment has been carried out by Stockton in Australia (1973). Small

    glass splinters (mean weight=4 g) were laid out on a level sandy surface, covered by

    5 cm of sand and intensively trampled for one day. When excavated by levels, the glass

    was found to be distributed over a depth of 16 cm. More than 50% of the sample had

    remained on or near the original surface; 22% had moved 2-5 cm upward; the rest of

    the sample had migrated downward to a maximum depth of about 10 cm below the

    original surface.

    Patterns of vertical dispersal of conjoined sherds in the central part of the Fontbregoua

    cave suggest that trampling may have been a contributing factor. The density of material

    indicates that the site had been intensively lived in; trampling on discarded material

    lying on the surface or slightly buried must have been a common occurrence. Some of

    the vertically displaced pieces are horizontally very close; thus it is unlikely that their

    vertical distribution is due to the presence of irregular or sloping l iving surfaces. Groups

    of conjoinable and vertically displaced sherds are scattered over large areas; this seems

    to suggest a non-localized agency of displacement such as trampling. To test this hypo-

    thesis, we decided to replicate Stockton’s experiment in such a way so as to approximate

    more closely conditions prevailing in the cave.

    Experiment Procedures

    On the rocky slopes, near the cave entrance, there are several artificial terraces built with

    dry stone walls to hold the screened backdirt of the excavation. During the summer,

    when the excavations take place, these terraces are used by the excavating team (an

    average of 12-15 people) for a variety of activities: screening, washing and sorting arti-

    facts, eating and resting.

    The material selected for the experiment was laid out on these terraces. We used

    material similar in size, shape and kind to that found in the cave, viz.:

    - flint flakes, blades and retouch flakes (down to l-2 cm in maximum dimensions);

    - mammal bones (shafts and articular ends, teeth and skull fragments);

    -

    marine shells (Cardium, Patella, Columbella, Cerithium);

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    7/15

    STRATIF IED SITES

    273

    - pottery sherds (modern, unglazed);

    -

    a few flat limestone pebbles, to simulate the polished celts found in the cave.

    Ninety-five y0 of all objects were 110 cm in maximum dimensions. Al l objects were

    measured, weighed, numbered and catalogued; the stone pieces were coated with a thin

    film of paint, in brilliant colours, to distinguish possible damage scars and to speed up

    the conjoining of broken fragments.

    To avoid downslope movement and image distortion in vertical projections, we flat-

    tened the sandy surfaces by repeated trampling before laying out the objects. Footprints

    no deeper than 1 cm were considered indication of an acceptable degree of compaction.

    The precise location of each object was recorded with three spatial coordinates, the

    relative elevation being measured with a transit. Some of the squares were covered with

    2-4 cm of sand or sand and rubble from the cave before trampling; in others the material

    was left uncovered. A sediment analysis of two squares done by Jacques E. Brochier

    (Laboratoire de Sedimentologie, FacultC Des Sciences, Marseille) showed that the matrix

    can be defined as a dry, loose, well-sorted sil ty sand (82-86% sand; 9-13% silt ) with very

    litt le clay (5-8%); the median size was 96-108 pm. Small quantities of limestone rubble

    (2-20 mm in diameter) were occasionally present in other squares; their proportions

    varied from 0 to 14.5%.

    In Stockton’s experiment intensive trampling was done for one day. We decided in

    favour of natural, casual and prolonged trampling done by the excavators while walking

    in and out of the cave to attend to their tasks. Al l the excavators wore only light sandals

    or went barefoot.

    The results we present now are based on the excavation and analysis of 292 pieces

    from 11 l-m squares, 4 of which had two superimposed levels. Eight squares were dug

    after 16 days; the others after 22, 32, and 36 days. All excavated artifacts were again

    recorded with their new spatial coordinates with the exception of a few objects (2.9%)

    which were recovered in the screens and could not be used in the analysis. The weather

    remained warm and sunny throughout, with only a light rain one night. Moisture from

    continuous water sieving operations considerably hardened the sand in one square.

    Resul ts

    Vertical displacement

    Significant vertical dispersal can be achieved even with a limited amount of trampling.

    The maximum range of vert ical separation we observed is 8 cm. After 16 days of casual

    trampling, 20% of the objects in square S (Figure 1) had migrated downward 5 to 7 cm

    below the surface.

    After 16 days of trampling, the material of two superimposed levels in square R,

    originally separated by 3 cm of sterile sand, was completely mixed and formed only one

    level (Figure 2). To prevent the effects of sand compaction in the mixing of objects,

    trampling was done in two stages. The lower level of artifacts was covered with 3 cm of

    sand and trampled for 20 days. Other objects were then laid out on the compacted sur-

    face and trampled for 16 more days.

    Table 1 gives the frequency distributions of artifacts by depth in 5 squares trampled

    for 16 days. The objects had not been covered by sand. Note that 21% of the objects

    did not move while 28% was displaced 3 to 7 cm below the surface. The few millimeters

    of upward displacement are the results of horizontal migration over a slightly irregular

    surface.

    Ten l-m squares covered by sand or sand and gravel were trampled for 16 to 36 days.

    The thickness of the protective layer varied from 2 to 4 cm. Some squares happened to

    be in areas outside the main traffic and were only minimally trampled. For the sake of

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    8/15

    274

    50 -

    cm

    80-

    +

    +

    P. V ILLA AND J . COUR TIN

    50

    lm

    Figure 1. Square S , upper le vel, uncov ered. Vertical projections of artifac ts.

    Top, before trampling, bottom, af ter 16 days of casual t rampling.

    c c ‘V c-

    I

    0

    m a 0

    0 0 0 c o o o o o

    0

    0 0

    c

    +

    lm

    Figure 2. Vert ical projections of art ifacts in square R, lower and upper levels.

    The lower level was covered with 3 cm o f sand and trampled for 20 days.

    Objects were then laid out on the compa cted surface and trampled for 16

    days. Top, the relative posit ion of objects in both levels before vert ical dis-

    placement occurred. The f igure is a compo site image as the two levels were

    deposited at dif ferent t imes. The elevat ion of each object was measured with

    a transit in relation to a f ixed datum. Bottom , af ter 36 days of t rampling.

    record a table of frequency distributions of artifacts in all covered squares is given below

    (Table 2). However, the reader is warned that displacement distributions in individual

    squares varied greatly.

    In general, downward displacement is smaller in covered than in uncovered squares

    even when trampling lasted longer. Upward displacement, of 1 cm or more above the

    original position, is limited; its frequency varies from 0 to 17% in individual squares

    with an average of 9-8o/o. The maximum observed value is +2-4 cm in a square covered

    with 4 cm of sand.

    We prefer to analyse distributions square by square and individually compare them

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    9/15

    STRATIFIED SITES

    275

    Table 1. Amount of disprclcement f rom the or@kal surface in uncovered squares

    Displacement n cm o/0of artifacts

    +0.9/-0.9

    21.0

    -l*O/-2.9 51.0

    -3.0/-4-g 23.0

    -5*O/-6.9 5.0

    N= 100

    Range: +0.3/--6.9

    Table 2. Amount of displacement f rom the or iginal surface in covered squares

    Displacement n cm 0/O f artifacts

    +1*0/+2.9

    +0.9/-0.9

    -1.0/-2-g

    -3*o/-4.9

    --5-O/-6*9

    N= 184

    Range: +2.4/-5.2

    to distributions in uncovered levels of the same squares. Table 3 shows vertical displace-

    ments in two squares, R and S, each with two superimposed levels. As mentioned earlier,

    in these squares sand was spread over laid out material and left to be trampled for 20

    days. A second layer of objects was deposited and walked over for another 16 days.

    The differences between the displacement distributions of the two lower levels appear

    to be due to the differences in the thickness of the covering layer (3 and 2 cm), but could

    also be partially attributed to more intensive trampling in square S which was close to

    the lunch table and to the drinking water. It is also clear that, as the vertical distance

    between the trampled surface and the artifacts increases, the effects of trampling de-

    crease proportionally until downward displacement effectively ceases. Vertical displace-

    ment is more limited in covered squares where objects were initially 2 to 4 cm below the

    trampled surface.

    Frequency distributions from the other covered squares are intermediate between

    those of the lower levels in squares R and S. In general, a comparison of frequency

    distributions of artifacts by depth in all squares suggests that the degree of vertical dis-

    placement depends on four major variables:

    (1) the intensity of trampling;

    (2) the degree of compaction of the sediments;

    (3) the thickness of the deposits covering the pieces;

    (4) the weight/size of the pieces (this point is discussed in the section “Sorting by

    weight”).

    During trampling, objects are constantly being displaced, sometimes lifted, sometimes

    pushed down. When the trampling process begins most pieces not covered by sand are

    pushed down and in a few minutes will disappear into the earth. But, after a while the

    same pieces may reappear on the surface, having been brought up by a human foot.

    Some pieces, however, appear to migrate gradually downward, below the zone of actual

    disturbance and beyond the reach of feet. Intensive trampling will increase the propor-

    tion of pieces pushed far below the surface.

    The degree of compaction and resistance to object penetration varied in different

    squares depending, among other factors, on the presence of rubble and moisture in the

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    10/15

    276 P. V ILLA AND J . COU RTIN

    matrix. Unfortunately, we have not been able to provide an objective measure of packing

    and we cannot quantify its effects on vert ical displacement.

    Table 3. Amount of displacement f rom the or iginal surface in covered and un-

    covered levels of the same square

    Square R

    Lower level

    (36 days, covered

    by 3 cm of sand)

    Displacement in cm

    %

    +1.0/+2.9 15.8

    +0*9/-0.9 84.2

    -l.O/-2.9

    -

    -3*o/-4.9

    -

    -5.O/-6.9

    -

    N=19

    Range +2*3/-0.8

    Upper level

    (16 days, not

    covered)

    29.4

    52.9

    17.6

    -

    N=17

    Range -O-l/-4.5

    Square S

    Lower level Upper level

    (36 days, covered (16 days, not

    covered

    Displacement in cm

    by 2 cm of sand)

    % %

    +1.0/+2.9

    +0.9/-0.9

    - l.O/--2.9

    -3-o/-4.9

    -5.O/-6.9

    -

    41.2 5.0

    29.4 30.0

    235 45.0

    ,I : : 20.0

    N=20

    Range +O.S/-5.2 Range +O*l/-6.9

    Sorting

    by

    According

    decreasing

    weight

    to Stockton, trampling results in moderate size sorting, with mean weights

    with depth. Our results are slightly different. Note that Stockton’s pieces were

    lighter than ours, having a mean weight of 4 g; our pieces had a mean weight of 12 g.

    Our correlation table (Table 4) suggests that pieces lighter than 50 g may move upward,

    Table 4. Correlation between weight and amount of displacement

    Amou nt of displacement f rom the or iginal surface (cm)

    We ight +2.9/ +0.9/ -l.O/ -3-O/ -S.O/

    @

    -l-O -0.9 -2.9

    -4.9

    -6.9

    1-9

    10-19

    20-29

    30-39

    40-49

    50-59

    60-69

    70-79

    80-89

    90-99

    100-109

    100-119

    120-129

    10 90 24 4

    4 27 4 I

    3 11 6 1 -

    4 1 1 -

    1 4 4 - 1

    3 1 - -

    1 2 - -

    - - -

    -

    1 1 - -

    -

    1 - - -

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    11/15

    S

    STRATIFIED SITES

    277

    downward or remain in place. Pieces heavier than 50 g tend to stay on or near the sur-

    face on which they were originally placed. However, the number of pieces heavier than

    50 g is very small and this proposition, though intuitively justifiable, remains unproven.

    We found no correlation between displacement and kind of material.

    Horizontal displacement

    The maximum observed horizontal displacement is 85 cm. In uncovered squares the

    the horizontal displacement is greater than in covered squares and most pieces are

    affected (Figure 3). The pieces had changed from a horizontal to an oblique or sub-

    vertical position in 4.3% of the cases; 21-O”h had been overturned.

    Figure 3. Horizontal displacements in squares R and S, near the cave en-

    trance. 1, Squares R and S, lower levels, af ter 36 days of t rampling. The

    objects w ere covered with 3 and 2 cm of sand respect ively. 2, Square R and

    S, upper levels, af ter 16 days of trampling. The objects had not been covered

    with sand. Arrows indicate direction and amount of hor izontal displacement.

    Aster isks indicate pieces that were not displaced.

    There is no obvious linear correlation between horizontal displacement and weight.

    Figure 4 shows that the most displaced pieces are light while the heavy pieces moved

    little. However, many light pieces were not displaced or were displaced only a short

    distance. Thus weight is not a good predictor of displacement. The missing third variable

    is the haphazard occurrence of effective scuffage.

    The observation that larger objects on paths get greater horizontal displacement (Wilk

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    12/15

    278 P. VILLA AND J. COU RTIN

    & Schiffer, 1979, p. 533) does not seem to apply to people with no or light footgear. It

    should be noted, however, that our trampled areas probably received less traffic than

    Wilk & Schiffer’s urban paths and that only 5% of our items were larger than 10 cm

    in any one dimension (see also Figure 4 and Table 4).

    Breakage

    Damage to pieces by breakage was limited since the substratum was nowhere very com-

    pact and artifact density was low (mean=20 artifacts per square metre). The fragmenta-

    50 -

    l

    (a 1

    40.

    5 30.

    i

    .o

    520-g

    l

    l

    L

     

    10-r’ l l

    .

    .

    l *

    y. l +

    .

    04

    =’ +a’. I.

    . l + .

    .

    .+ . l l

    .t

    l

    c

    120

    1

    I

    (b)

    50-

    .

    ,”

    40-

    i

    .

    .F N’

    s

    30-

    l . .

    20.

    l

    l .

    .

    IO- . . .

    .

    l

    : ’ l .

    . . . . .

    0

    IO 20 30 4b 50 60 70

    80

    Horizonta l displace ment (cm)

    Figure 4. Sca tter diagram of the relationship betwe en weight and horizontal

    displacement in (a) two uncovered squares trampled for 16 days and (b) two

    covered squares trampled for 36 days.

    ,.

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    13/15

    STRATIFIED SITES

    279

    tion index (no. of fragments x lOO/total no. of pieces) is higher in covered squares: 26

    versus 13 in uncovered squares.

    A few flints had very limited vis ible edge damage; predictably a few long blades were

    snapped in two. Bone and shells broke more easi ly than pottery or flint. However, the

    bone material was slightly weathered and the modern pottery we used is harder than the

    prehistoric pottery.

    Conclusions

    The experiment demonstrates that trampling can cause mixing of materials belonging to

    two separate levels. Thus, horizontally associated materials in a layer may derive from

    the mixing of distinct episodes of site use. Size sorting may appear to be a clue to the

    occurrence of vertical displacement since pieces lighter than 50 g are more mobile (both

    upward and downward). In sites with superimposed levels, however, it would be very

    difficult to know for sure whether small objects in a given level are displaced or in situ,

    unless information from conjoined pieces is used. It should be noted that, in Stone Age

    assemblages, many pieces are lighter than 50 g.

    The degree of vertical dispersal suggests that in caves whose rates of sedimentation

    are low, some archaeological items may be considerably younger than the matrix in

    which they are found.

    The horizontal displacement is not negligible. When doing a spatial analysis of sites

    where trampling and scuffing might have been a significant factor, it is probably wiser

    not to trust statistics based on precise measurements of horizontal distances (such as

    used in nearest neighbour analysis). Otherwise, one may be applying precise methods to

    ambiguous, imprecise data.

    The vertical displacement we observed (7-8 cm) is smaller than the displacement ob-

    served by Stockton (16 cm). Crucial differences were perhaps the use of small glass

    splinters in his experiment and the intensive trampling.

    Finally, both experiments suggest that in sandy deposits the effects of trampling are

    limited to a zone 10-16 cm deep. More intensive trampling or trampling over long spans

    of time may well cause larger-scale vertical displacements. However, the evidence, as it

    stands now, suggests that other agencies should be considered for large-scale vertical

    separations of archaeological items. Of course, these conclusions apply only to sandy

    deposits and to archaeological materials of the kind we used. Clearly the degree of

    consolidation of living surfaces at the time of occupation is a factor that should be con-

    sidered in stratigraphic studies.

    Note. While this paper was in review, we received a manuscript by Gifford et al. (in

    press) which dealt with two trampling experiments. These experiments support Stockton’s

    and our conclusions about the effects of trampling and the extent of vertical displace-

    ment sandy deposits.

    Acknowledgements

    The work of Paola Villa has been supported by grants from the American Philosophical

    Society and the Leakey Foundation. We also wish to thank Karl W. Butzer, John W.

    Fisher, Jr., Patricia Phillips and one unknown reviewer for their careful and constructive

    criticism of this paper.

    References

    Binford, L. R. (1981). Bones. Ancient Men and Modern

    Myths.

    New York: Academic Press.

    Binford, L. R. (I 982). The archaeology of place. Journal

    of

    Anthropological Archaeolog y 1,5-3

    1.

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    14/15

    280 P. VILLA AND J. COURTIN

    Binford, S. R. & Binford L. R. (1969). Stone tools and human behavior.

    Scient if ic American

    220, 70-84.

    Bordes, F., Rigaud, J. Ph. & de Sonneville Bordes, D. (1972). Des buts, problemes et limites

    de I’archeologie paltolithique.

    Quaternaria 16,

    15-34.

    Bunn, H., Harris, J. W. K., Isaac, Glynn, Kaufulu, Z., Kroll, E., Schick, K., Toth, N., &

    Behrensmeyer, A. K. (1980). FxJ iSO: an Ear ly Pleistocene site in northern Kenya.

    World

    Archaeology

    12, 109-136.

    Butzer, K. W. (1978). Sediment stratigraphy of the Middle Stone Age sequence at Klasies

    River Mouth, Tsitsikama Coast, South Africa.

    South African Archa eological Bulletin 33,

    141-151.

    Butzer, K. W. (1980). Cave sediments, Upper Pleistocene stratigraphy and Mousterian facies

    in Cantabrian Spain.

    Journal of Archaeological Scienc e 8,

    133-183.

    Butzer, K. W. (1982).

    Archaeology as Hum an Ecology.

    Cambridge: The University Press.

    Cahen, D. & Moeyersons, J . (1977). Subsurface movements of stone artifacts and their

    implications for the prehistory of Central Africa.

    Nature 266,

    812-815.

    Cheylan, M. & Courtin, J. (1976). La consommation de la tortue cistude Emys orbicular is au

    Postglaciaire dans la grotte de Fontbregoua (Salernes, Var).

    Bulletin du Mus ee d’Histoire

    Naturelle de Marseille 36,

    41-46.

    Courtin, J. (1975). Le Mesolithique de la Baume de Fontbregoua a Salernes.

    Cahiers Ligures

    de Prehistoire et d’ArchPologie 24,

    110-l 17.

    Courtin, J. (1976). La Baume Fontbregoua, Salernes, Var.

    Livret-Guide de 1’Excu rsion B2,

    IX Congress U. I .S.P.P. ,

    Nice, pp. 21-37.

    Courtin, J. (1978). Les animaux domestiques du Neolithique provencal: la faune sauvage et

    les debuts de l’elevage dans le Sud-est de la France.

    Bulletin du Mus ee d’Histoire Naturelle

    de Marseille 38,

    187- 194.

    Courtin, J. & Erroux, J. (1974). Apercu sur l’agriculture prehistorique dans le Sud-est de

    la France.

    Bulletin de la Soc iett Prehistorique Franc aise

    71, 321-334.

    Courty, M.-A. & Raynal, J. P. (1982). L’ttude geologique de sites archeologiques holodnes:

    I’approche de fait anthropique. In

    Le Neolithique de I’Est de la Fran ce.

    Societe

    archeologique de Sens, cahier no. 1, pp. 79-84.

    Delporte, H. (1982). A propos du Blot : methodologie et Bpistemologie de l’habitat au

    Paleolithique superieur. In

    Les hab itats du Paleolithique superieur.

    Colloque de Roanne-

    Villerest, sous la direction de J. Combier, 22-24 Juin 1982, pp. 152-162.

    Gifford, D. P. (1978). Ethnoarchaeological observations of natural processesaffecting

    cultural materials. In (R. A. Gould, Ed.)

    Explorat ions in Ethnoarchaeology.

    Albuquerque:

    Univers ity of New Mexico Press, pp. 77-101.

    Gifford, D. P. & Behrensmeyer, A. K. (1977). Observed formation and burial of a recent

    human occupation site in Kenya.

    Quaternary Research 8 , 245-266.

    Gifford, D. P., Damrosch, D. B., Damrosch, D. R., Pryor, J. & Thunen, R. L. (in press).

    The third dimension in site structure: an experiment in trampling and vertical dispersal.

    American Ant iquity.

    Goldberg, P. (1979). Micromorphology of sediments from Hayonim cave, Israel.

    Catena 6,

    167-181.

    Goldberg, P. (1980). Micromorphology in archaeology and prehistory. Paleorient 6, 159-164.

    Grayson, D. K. (1979). On the quantification of vertebrate archaeofaunas. In (M. B. Schiffer,

    Ed.)

    Advances in Archaeological Methods and Theory,

    Vol. 2. New York: Academic Press,

    pp. 199-237.

    Guilaine, J. (Ed.) (1979).

    L’abr i Jean Cros.

    Centre d’anthropologie des SocietCs urales,

    Toulouse.

    Hughes, P. J. & Lampert, R. J. (1977). Occupational disturbance and types of archaeological

    deposits.

    Journal of Archaeological Science 4,

    135-140.

    Harris, E.

    C .

    (1979).

    Principles of Archaeological Stratigraph y.

    London: Academic Press.

    Isaac, G. LI. (1967). Towards the interpretation of occupation debris: some experiments and

    observations.

    Kroeber Anthropologicul Soc iety Papers 37,

    31-57.

    Jewell, P. A. & Dimbleby, G. W. (1966). The experimental earthwork on Overton Down,

    Wiltshire, England : the first four years.

    Proceedings of the Prehistoric Society 32, 313-342.

  • 8/16/2019 Villa & Courtin 1983 the Interpretation of Stratified Sites - A View From Underground

    15/15

    STRATIFIED SITES 281

    Laville, H., Rigaud, J. Ph. & Sackett, J.

    (1980). Ro ck Shelters o f the Perigord. New

    York:

    Academic Press.

    Lewarch, D. E. & O’Brien , M. J. (1981). The expanding role of surface assemblages n

    archaeological research. In (M. B. Schiffer, Ed.)

    Advances in Archaeological Method and

    Theory,

    Vol. 4. New York: Academic Press, pp. 297-342.

    Michels, J. W. (1973).

    Dating Methods in Archaeology.

    New York: Seminar Press.

    Reynolds, P. J. (1974). Experimental Iron Age storage pits: an interim report.

    Proceedings of

    the Prehistoric Society 40,

    118-l 3 1.

    Reynolds, P. J. (1979). I ron A ge farm: the Butser Experiment. London: British Museum.

    Rowlett, R. M. & Robbins, M. C. (1982). Estimating orig inal assemblage content to adjust

    for post-depositional vertical artifact movement.

    World Archaeology 14, 73-83.

    Sackett, J. R. (1973). Style, function and artifact variability in palaeolithic assemblages. n

    (C. Renfrew, Ed.)

    The Explanat ion of Culture C hange.

    London: Duckworth, pp. 317-328.

    Schiffer, M. B. (1975). Archaeology as behavioral science.

    American Anthropologist 77,

    836-846.

    Siriainen, A. (1977). Pieces n vertical movement: a model for rockshelter archaeology.

    Proceedings of the Prehistor ic Society 43, 349-353.

    Spaulding, A. C. (1960). The dimensions of archaeology. Reprinted in (B. H. Fagan, Ed.)

    Introductory Readings in Archaeology,

    1970. Boston: Little & Brown, pp. 201-218.

    Speth, J. D. & Johnson, G. A. (1976). Problems in the use of correlation for the investigation

    of tool kits and activ ity areas. In (Ch. E. Cleland, Ed.)

    Cultural Change and Cont inuity:

    Essa ys in Honor of J. G. Grif f in.

    New York: Academic Press, pp. 35-76.

    Stockton, E . D. (1973). Shaw’s Creek shelter: human displacement of artifacts and its

    significance.

    Mankind 9, 112-l

    17.

    Straus, L. G. (1979). Caves: a paleoanthropological resource.

    World Archaeology 10,331-339.

    Thomas, D. H. (1979).

    Archaeology. New

    York: Holt, Rinehart & Winston.

    Van Noten, F., Cahen, D., Keeley, L. H. & Moeyersons, J. (1978).

    Les chas seurs de Meer.

    Dissertationes Archaeologicae Gandenses XVII, Brugge.

    Van Noten, F., Cahen, D. & Keeley, L. H. (1980). A Paleolithic campsite in Belgium.

    Scient if ic American 242, 48-55.

    Villa, P. (1982). Conjoinable pieces and site formation processes.

    American Ant iquity 47,

    276-290.

    Whallon, R., Jr. (1973). Spatial analysis of palaeolithic occupation areas. In (C. Renfrew,

    Ed.)

    The Explanat ion of Culture Change.

    London: Duckworth, pp. 115-130.

    Wilk, R. & Schiffer, M. B. (1979). The archaeology of vacant lots in Tucson, Arizona.

    American Ant iquity 44, 530-536.