Solifluction-induced modifications of archaeological levels: simulation based on experimental data from a modern periglacial slope and application to French Palaeolithic sites

Journal of Archaeological Science 35 (2008) 99e110 http://www.elsevier.com/locate/jas Solifluction-induced modifications of archaeological levels: simulation based on experimental data from a modern periglacial slope and application to French Palaeolithic sites Arnaud Lenoble a,b,*, Pascal Bertran a,b, François Lacrampe c a PACEA/Institut de Préhistoire et de Géologie du Quaternaire, bâtiment de géologie, avenue des facultés, F-33405 Talence, France b INRAP, Centre d’activité les Echoppes, 156 avenue Jean Jaurès, F-33600 Pessac, France c Archéosphère, Domaine du Haut-Carré, Bât. C5, 351 cours de la Libération, F-33 405 Talence Cedex, France Received 29 November 2006; received in revised form 19 February 2007; accepted 25 February 2007 Abstract The taphonomic study of Petit-Bost, Croix-de-Canard and Cantalouette II, three Palaeolithic sites that were recently discovered near Périgueux and Bergerac (Dordogne, France) in a colluvial context, has enlightened the difficulty of adequately appreciating the relative role of cultural and natural processes in site spatial patterning. Periglacial solifluction was thought to have played a significant role in site formation. Because the nature of the modifications induced by solifluction was still poorly understood, a simulation was made using data of soil movement recorded at La Mortice (French Southern Alps, 3100 m in elevation) in a modern periglacial environment. The results show that, for a knapping location, the first steps of deformation are typified both by a downslope translation of the location center and by an anisotropic diffusion of the artifacts. The knapping spot becomes elongated along the slope, with a dense relic concentration of artifacts in the upslope portion. This type of pattern has been obtained after 100e200 years of simulated displacement according to the climatic and soil conditions that characterise the La Mortice site. The ultimate stages of deformation show that the artifact distribution tends to homogenise on larger surfaces and resemble a random distribution. The ability of the simulated patterns to closely fit those observed in archaeological contexts is evaluated at three sites from Southwestern France. At Petit-Bost, the hypothesis of limited solifluction explains accurately the association of both cultural (artifact concentrations) and natural (artifact preferred orientation) features. At Croix-de-Canard, long-term solifluction can be proposed. By contrast, the simulated patterns do not describe the structures observed on the steeper slopes at Cantalouette II, where the knapping spot transforms into distinct solifluction lobes. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Site formation processes; Periglacial solifluction; Simulation; Middle Palaeolithic; Southwestern France 1. Introduction Since Schiffer’s (1972) work, the idea that all Palaeolithic sites have undergone modification to a variable extent after their abandonment by the inhabitants has become firmly established (Schiffer, 1983, 1987; Binford, 1981; Butzer, 1982; Bertran, 1994; Waters, 1992). Different types of taphonomic * Corresponding author. Present address: INRAP, Centre d’activités les Echoppes, 156 avenue Jean Jaurès F-33600 Pessac, France. Tel.: þ33 (0)5 57 01 00 10. E-mail address: arnaud.lenoble@inrap.fr (A. Lenoble). 0305-4403/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.jas.2007.02.011 analyses have been proposed to estimate these modifications. The alteration and the degree of fragmentation of the artifacts, grain-size distribution, the fabric (orientation and dip), the distribution of the refittings, and the spatial patterning of artifacts are among the most commonly used criteria. The results are interpreted by comparison with experimental data, or data derived from geomorphic observations. However, the former are rare and dedicated to only a few biological and sedimentary processes such as fluvial transport (Schick, 1986), overland flow (Kirkby and Kirkby, 1974; Poesen, 1987; Lenoble, 2005), ploughing (Steinberg, 1996), needle-ice creep (Bowers et al., 1983), and the activity of the soil fauna (Cahen and A. Lenoble et al. / Journal of Archaeological Science 35 (2008) 99e110 Moeyersons, 1977; Johnson, 2002; Araujo and Marcelino, 2003). The duration of the experiments is often too short and, consequently, the modifications observed are not strong enough to be compared to archaeological cases. The suitability of the experimental models varies largely according to the studied sedimentary processes. As shown by Schick (1986), the modifications undergone by lithic assemblages in a fluvial context are mainly due to few efficient floods, which can be accurately documented by short-duration experiments. By contrast, other processes like soil creep are characterised by the compilation of minimal displacements over long lapses of time, typically a few millimetre or centimetre a year, that affect collectively the whole artifact assemblage. Experiments alone cannot provide adequate descriptions of the modifications that result from these slower processes. As a consequence, the impact of creep and other related slow sedimentary processes is generally intuitively inferred in archaeological sites. Periglacial solifluction offers an example of such a slow process. As defined by Harris et al. (1997), the term is applied to a slow creep of soil down a slope typical of periglacial environments. The displacement involves distinct mechanisms: (1) slow downslope movement of soil because of frost heaving and thawing (frost-creep), (2) heaving and creeping of stones due to needle ice formation at the soil surface (pipkrakes), and (3) localized mass displacement of water-saturated sediment (gelifluction). Soil displacement frequently generates turf-banked lobes in alpine environment and stone-banked lobes in deserts, but sheet without lobes are possible as well. This process can occur on grades as shallow as 2 degrees, and, consequently, is widespread in periglacial landscape (Ballantyne and Harris, 1994). This process is well-known by arctic archeologists because sites are frequently encountered in solifluction deposits (e.g. Esdale et al., 2001; Hopkins and Giddings, 1953; Johnson, 1946; Mackay et al., 1961; Rahmani et al., 2005). Solifluction also occurred in mid-latitude European regions that experienced periglacial climates during the Pleistocene glacial periods (Gullentops and Deblaere, 1992; Vallin et al., 2001; Van Vliet-Lanoë, 1988). This is the case in Southwestern France where permafrost developed at the height of cold episodes (Texier and Bertran, 1993; Van Vliet-Lanoë, 1996). In this region, solifluction is widely reported from Palaeolithic sites (Bertran, 1994; Texier, 2001; Couchoud, 2003; Lenoble, 2004). However, the influence of this natural process on site patterning remains conjectural because of no suitable model exists. These difficulties arose during the study of two Middle Palaeolithic sites and an Upper Palaeolithic site recently discovered along a future highway near Périgueux and Bergerac (Dordogne, France). For each of these sites, Petit-Bost, Croixde-Canard, and Cantalouette II, geoarchaeological investigations showed that solifluction might have played a significant role in sediment deposition. In order to provide descriptions of archaeological modifications induced by solifluction before final site burial, a computer simulation based on the experimental data collected during the TRANSIT program, which took place in a modern periglacial environment in the French Alps (Texier et al., 1998; Todisco et al., 2000), was carried out. The results of the simulation are presented here and compared to the archaeological patterns recorded at Petit-Bost, Croix-de-Canard and Cantalouette II. 2. Archaeological data: Croix-de-Canard, Petit-Bost and Cantalouette II sites The site of Croix-de-Canard, excavated by L. Detrain, is located on an alluvial terrace in the Isle valley about 30 km west of Périgueux (Fig. 1) and contains distinct Middle Palaeolithic levels. The most recent, referred to as ‘‘locus 2’’, is the only one investigated here. The ‘‘locus 2’’ level is buried in clayey sand colluvium that covers alluvial gravels at a depth of 0.6 m. The slope of the artifact level is 3 degrees. Because the site has not been deeply buried by colluviums and should have underwent strong modification due to sedimentary and soil-forming processes, a detailed taphonomic study has been 4° 0 4° 8° N Paris 48° 48° FRANCE Bordeaux 44° 44° 0 250 km 4° Croix-de-Canard PERIGUEUX Neuvic Petit-Bost Vern I sl e MontponMénestérol Mussidan Cantalouette II Le Bugue re 100 Vé BERGERAC Sainte-Foyla-Grande Dordogne 0 Fig. 1. Site location. zè Lalinde 10 km 200 m NGF 100 m 101 A. Lenoble et al. / Journal of Archaeological Science 35 (2008) 99e110 carried out. The main conclusions are the following (Detrain et al., 2005; Lenoble and Bertran, 2004): - the artifacts have a strong preferred orientation parallel to the slope (Fig. 2). Such an orientation has never been observed in undisturbed sites. By contrast, the orientation intensity falls within the area known for solifluction (Fig. 3). Therefore, the hypothesis of significant modification of the archaeological level due to periglacial slope dynamics during the Pleistocene cold phases seems highly plausible. - the lack of clear concentration and the low artifact density again suggest that initially more clustered units were scattered by natural processes. However, a high refitting rate has been found (37%), similar to that found in wellpreserved Palaeolithic sites. isotropic fabric 0.2 0.8 argiliturbation 0.4 Elongation Index Isotropy Index 0.6 E3/E1 1-(E2/E1) 0.6 0.4 runoff + bioturbation runoff 4 - Like at Croix-de-Canard, the artifacts have a significant preferred orientation, which testifies to downslope movement due to solifluction (Fig. 3). Abundant evidence of former ice bodies in the soil (ice-wedge casts, platy structure due to lenses of segregation ice) strongly supports this hypothesis. - By contrast to Croix-de-Canard, the archaeological data suggest negligible site modification. The main arguments are: (i) a clear artifact clustering (Fig. 4); (ii) a high refitting rate (up to 20%), each refitted block involving a large number of pieces (118 for the most important). The 0.8 3 1 absence of perturbation The taphonomic study enables us to conclude that strong alteration of the site patterning by solifluction has occurred, although the integrity of the lithic assemblage seems to have been largely preserved. The Petit-Bost site, excavated by L. Bourguignon, is located in a similar geomorphologic context, 3 km west of Croixde-Canard. The main archaeological level (‘level 1’) has a mean slope gradient of 4 degrees. It is buried under 1.3 m of sandy clay colluviums. The taphonomic analysis shows (Bourguignon et al., 2005): solifluction 0.2 planar fabric 2 linear fabric Fig. 3. Fabric of Croix-de-Canard, sector 2 (1 and 2), Petit-Bost (3) and Cantalouette II, Solutrean level (4). The areas typifying the processes of disturbance are indicated according to Lenoble and Bertran (2004). refittings are grouped within a unique artifact concentration. This is interpreted as a knapping area mixing knapping activities and throwing out of discarded flakes, rather than a typical knapping spot because of the elongated shape and the large dimension of the area. At this stage of the study, the taphonomic analysis at PetitBost cannot be used to explain why undeniably ‘cultural’ characteristics, particularly with respect to the spatial distribution of the artifacts, and evidence of modifications due to sedimentary processes are present in the same archaeological level. The site of Cantalouette II, excavation conducted by L. Bourguignon, is situated on a plateau at the northern side of the Dordogne valley, about 10 km east of Bergerac (Fig. 1). slope N sector 2 Middle Palaeolithic N = 49 Fig. 2. Fabric of elongated flakes, Croix-de-Canard, sector 2. Schmidt (equalarea) stereogram, lower hemisphere. 0 1m Fig. 4. Artifact distribution map of level 1, Petit-Bost. The dotted lines indicate the location of the survey trenches. 102 A. Lenoble et al. / Journal of Archaeological Science 35 (2008) 99e110 Three main archaeological levels have been preserved in a cluster of sinkholes (Bourguignon et al., 2004). The uppermost level has yielded a large Solutrean assemblage. This level is buried in massive sandy clayey silts, resting at variable exposures and at several depths, according to its location in the sinkholes. Several highly-concentrated clusters of flaked stones have been retrieved (Fig. 5). These clusters have sharp boundaries and are surrounded by empty or low-density areas. The following characteristics were noted during excavations. - The concentrations are made up exclusively by flakes related to the production of large laurel leaf bifaces. This homogeneity supports the idea that these areas represent knapping locations. Moreover, the debris originating from the same nodule are found in a single cluster. This can be assumed from distinctive physical features of flint debris in terms of color and veining. - Contrastingly, different lines of evidences point to the action of periglacial processes: (i) preferred orientation of isolated pieces located on the slopes (Fig. 3), (ii) polygonal distribution pattern of pieces resting at the bottom of the sinkhole, and (iii) abundant marks of former segregation ice in the fossils soil (platy structure and ice-wedges casts). Thus, like at Petit-Bost, the Solutrean level of Cantalouette II presents associated features indicative of modification by natural processes and evidence of cultural organization of artifacts, as shown by the homogeneity of flaked waste and the clear clustering of artifacts. 3. The TRANSIT experiment The TRANSIT program has been designed to document the modifications of lithic and bone assemblages by periglacial processes (Texier et al., 1998). A set of experiments has been carried out on a gentle slope at La Mortice, the southern French Alps, 3075 m in elevation. The climate is characterised by a mean annual air temperature close to 3  C, which implies discontinuous permafrost, and mean annual precipitation between 1500 and 2000 mm. Slope dynamics involve mainly solifluction, which creates stone-banked lobes about half a metre thick (Fig. 6). Downslope displacements are a few centimetres a year. At some periods, other sedimentary processes occur, such as aeolian transport of sand and thin slats of schist, and overland flow during summer rainstorms. These are poorly recorded in the deposits, because the associated sediments are rapidly reworked by solifluction and integrated into the lobes. A 0 10 m N B N S F area 79,2 TRENCH 79 Fig. 5. (A) Artifact distribution map of the Solutrean level, Cantalouette II. The dotted lines indicate the limits of the excavated area; and (B) close-up horizontal distribution of the area F with mapping of the sector used to construct Whallon’s diagram. 103 A. Lenoble et al. / Journal of Archaeological Science 35 (2008) 99e110 Fig. 6. Stone-banked solifluction lobe. This lobe is mainly composed of brown schist fragments and moves slowly over grey limestone debris. Fig. 8. View of a cell after 3 years of experiment. Some artifacts that were initially located inside the white square are now outside. In cross section, the deposits are stratified as a consequence of the piling of successive lobes (Bertran et al., 1993, 1995). One of the experiments was dedicated to recording artifact displacements on the ground surface. The gradient is between 8 and 14 degrees. Marked pieces were distributed among five test areas located at the surface of lobes and movements were measured over a 5 years period (1991e1996) (Figs. 7 and 8). Repeated heaving and settling of the soil due to ice lenses formation (frost-creep), needle ice activity at the surface (pipkrakes), and saturated soil deformation upon thawing (gelifluction) were the main processes responsible for artifact displacement (Coutard and Ozouf, 1996). Statistical analyses by Todisco (1999) and Todisco et al. (2000) indicate that mean displacements range between 1.8 and 3.6 cm/yr (Fig. 9). Few displacements were greater than 15 cm. The interannual variability is high and mainly controlled by the number of days with rain during the summer. At this altitude, most of the freezing and thawing cycles occur in July, August and September, since the ground remains covered by snow during the rest of the year. Their efficiency (i.e. their ability to trigger ice growth in the soil) depends largely on the soil water content and therefore, the amount of precipitation during the previous days. Some displacements are one or two orders of magnitude greater than the mean values, cellule 1 : displacements 1991 - 1995 e fr on t A ob slope l 0 30 cm B frequency 0.4 geometric mean = 1.4 cm + / - 0.6 / 3.2 cm N = 81 0.2 0.1 0 0 5 10 distance (cm) Fig. 7. Close-up of an experimental cell during emplacement in 1991. Fig. 9. (A) Artifact displacements recorded from 1991 to 1995 at cell 1, TRANSIT experiment, La Mortice. The length of the arrows is proportional to the movement. (B) Distribution of the displacements. 104 A. Lenoble et al. / Journal of Archaeological Science 35 (2008) 99e110 A frequency 0.3 geometric mean = 29.0 cm + / - 16.4 / 51.3 cm N = 50 0.2 0.1 0.05 0 0 50 100 15 0 cm frequency 0.3 B geometric mean = 31.2 cm + / - 10.7 / 90.4 cm N = 21 0.2 0.1 0.05 0 0 50 100 150 200 230 300 cm Fig. 10. ‘Unusual’ displacement distribution, TRANSIT experiment, established for lithics (A) and bones (B) all cells mixed. and reach over 1 m in a single year (Fig. 10). They affect mainly charcoal and bones and are attributed to overland flow and wind action. Significant differences in the displacements related to the original location of the pieces also were observed. Typically, flakes or bones resting on the stony fronts and lobe borders move at lower velocities than those located on the finer-grained central part of the solifluction lobes. 4. Principles of the simulation For the simulation, artifact displacements induced by solifluction are computed. The displacements are described by two parameters: length and orientation. At each cycle, a set of data is generated for each variable so that they present similar statistical characteristic (mean and standard deviation) to the measured data in the TRANSIT experiment. This allowed the significant variability of the measurements to be included. This is an important point because this variability plays an important role in artifacts dispersion. The length of displacement follows a log-normal distribution. This can be described by the geometric mean and the mean square of the values after transformation into log 10 (Caine, 1968). Long, ‘accidental’ displacements (i.e. displacements above four times the average) are considered separately. They amount to 4% of the total recorded values. A Student ttest indicates that this proportion does not differ significantly between the lithic and bone sample. The maximum covered distance reaches 3 m for bones (Fig. 10B), whereas it is only 1 m for lithics (Fig. 10A). Displacements are all oriented downslope. Their orientations fit a normal circular (Von Mises) distribution (Baschelet, 1981) and the angular standard deviation has been computed. Each increment in the calculation is equivalent to one year in the TRANSIT experiment. Thus, the artifact coordinates become (the Y axis being parallel to the slope): Xn þ 1 ¼ Xn þ dcos a Yn þ 1 ¼ Yn þ dsin a with d: displacement (cm), a: angle between the displacement direction and the slope (rd). Computations were made with Datadesk (version 6.1) software, which has a graphic interface enabling one to visualize artifact distributions at each step. Some assumptions are introduced into the simulation. The principal assumption concerns the initial (i.e. before deformation) artifact distribution, which is assumed to represent a knapping location for the following reasons. 1. Archaeological data at Petit-Bost, Croix-de-Canard and Cantalouette II show that flint knapping was an important activity at these sites. The presence of knapping spots, which have become significantly modified by sedimentary processes, is a plausible hypothesis. 2. Knapping spots are well-documented and have been reproduced experimentally (Newcomer and Sieveking, 1980; Hansen and Madsen, 1983; Boëda and Pellegrin, 1985; Bertran et al., 2006). 3. This kind of pattern has been used for other taphonomic experiments (Schick, 1986; Barton and Bergman, 1982), allowing comparisons with our data sets. 5. Dimensional analysis of variance To quantify the simulated modifications of an artifact distribution, the Dimensional analysis of variance, first proposed by Whallon (1973), is used. This analysis allows one to detect artifact concentrations and estimate their size from the map of an archaeological level. A square or rectangular grid is placed over the map, the grid units being 1/64 m2, i.e. 12.5 cm  12.5 cm. 105 A. Lenoble et al. / Journal of Archaeological Science 35 (2008) 99e110 be seen, corresponding to grid units of 1/16 m2 (i.e. 0.25 m  0.25 m) and 1 m2, respectively. The former reflects the two distinct heaps separated by the leg of the knapper, while the latter is the spot itself. The Variance/Mean ratio is independent from the total area of the studied surface. However, it is correlated to some degree to the relative position of the concentration in the grid (Djindjian, 1991). This explains the low ratio for a grid unit of 1/4 m2, the concentration being equally shared among four adjacent squares. 5 confidence interval variance / mean 4 concentrations 3 2 random distribution 1 6. Results 0 1/64 1/16 1/4 1 16 4 64 The simulated modifications of a 120-piece knapping spot consist of first in a downslope translation of the entire cluster, which reaches ca. 4 m in a century (Fig. 12). The ‘unusual’ movements, although being few (4%), have a significant contribution (25%) to the whole displacement (Fig. 13). Second, there is a dispersal of artifacts, resulting in an increase of the cluster’s size. The artifact scatter, due to greater downslope than perpendicular movement, becomes a slopeward elongated concentration with an ellipsoid or fan-like shape. Artifact density remains higher upslope than downslope. The sharpness of the cluster boundaries progressively decreases as a consequence of diffusion, as clearly shown on the Whallon’s diagram (Fig. 14). Fifty to a hundred simulated years are enough for the artifact distribution to become random, while the concentrations corresponding to the two internal heaps disappear after 20 years. This process of diffusion and homogenisation can also be illustrated by the evolution of the artifact number by unit surface (Fig. 15). Comparison between the simulated patterns and the archaeological case of Petit-Bost show a good agreement for displacements corresponding to a few hundred years. The similarities are: Grid (m2) uniform distribution Fig. 11. Whallon’s (1973) diagram. The bold curve corresponds to the knapping spot used in the simulation. On this graph, the area of distribution statistically similar to random distribution is shown by the shaded area. The areas where the Variance/Mean ratio is indicative of concentrations or uniform distribution are indicated as well. The mean number of items by surface unit and the associated variance are calculated. The grid units are then grouped into successively larger blocks, each twice the area of the next smaller block, and calculations are repeated (see Whallon, 1973, for detailed method). The ratio Variance/Mean is high when artifact concentrations occur and approach 0 for a uniform distribution (i.e. if the content of all unit surfaces is equal) (Fig. 11). The ratio approaches 1 when the distribution is random, i.e. when the distribution of the contents follows a Poisson distribution (Djindjian, 1991). A diagram Variance/Mean as a function of the size of the grid unit allows identification of the artifact concentrations, which correspond to peaks of variance. Significant departure from randomness of distribution is tested with X2. On the diagram, the area of the values, which cannot be statistically differentiated from randomness, can be drawn. Fig. 11 shows the diagram for the knapping spot used as the initial spatial pattern in the simulation. This initial scatter is divided into two halves as obtained by Newcomer and Sieveking (1980) or Boëda and Pellegrin (1985) when the knapper is seated. In Fig. 11, two peaks of the Variance/Mean ratio can 0 0 0 A B C 1 2 2 2 3 4 5 0 1 2 X (m) 3 Y (m) 1 Y (m) 1 slope Y (m) - the concentration is characterised by an elongated shape parallel to the slope; - the upslope part is denser than the downslope part of the concentration. The latter has indistinct limits; the number of pieces per surface unit decreases progressively downslope. 3 3 4 4 5 5 0 1 2 X (m) 3 0 1 2 3 X (m) Fig. 12. Evolution of a knapping spot (120 artifacts) after 20 (A) and 50 (B) years of simulated solifluction. 106 25 solifluction plus unusual displacements 20 15 10 solifluction 5 100 200 300 400 500 years Fig. 13. Comparison between the simulated displacement due to solifluction only and due to solifluction plus ‘unusual’ movements. The hypothesis of a former knapping spot that has been deformed by solifluction describes adequately the overall characteristics of the lithic concentration at Petit-Bost, with respect both to artifact distribution and preferred slope-parallel artifact orientation. However, precise estimation of the deformation rate remains difficult. The general shape of the concentration is close to that obtained by simulation for 50e100 years, while the maximum artifact scatter on the slope, which reaches approximately 10 m, suggests more significant displacement, which fits with ca. 500 years of solifluction. The Whallon’s diagram does not show a marked concentration (Fig. 16), but the 5 5 1/4 1 4 16 4 1 -4 64 2 slope Y 2 4 1 -2 0 2 6 8 -4 4 -2 0 2 4 2 4 2 4 X X Grid 5 Among the numerous clusters of Cantalouette II, the ‘‘F area’’ has been closely examined (Fig. 5). This concentration -2 0 1/16 - the lack of sorting of the artifacts during downslope movement; - the dimension of the excavated surface (250 m2). A large part of the original lithic assemblage has been recovered despite dispersion. 100 yrs initial state 1/64 Variance/Mean ratio increases slightly for grid units of 1 and 2 m2. Therefore, the spatial distribution seems to be best explained by 200 years of solifluction. Dimensional analysis of variance applied to ‘‘locus 2’’ at Croix-de-Canard indicates that no artifact concentration can be detected (Fig. 17). The distribution tends to be uniform for unit surfaces greater than 1/4 m2, and the Whallon’s diagram is similar to that obtained by simulation for a long period of deformation (more than 500 years). In contrast to what is observed at Petit-Bost, the refittings are scattered over the entire excavated surface, and do not cluster in a particular place, although the refitting rate is high (37%). The refitted sets of pieces are numerous, but each involves only a few artifacts (maximum 17, mainly 2e4). Taken altogether, the site characteristics suggest a full homogenisation of the occupation level, which can be explained in light of the simulation by a longlasting solifluction activity. Despite strong modification of the lithic assemblage, preservation of its integrity may result from: Y length of concentration center displacement (m) A. Lenoble et al. / Journal of Archaeological Science 35 (2008) 99e110 20 yrs 5 200 yrs 6 0 8 Y Y 2 10 4 1 12 1 6 -4 -2 0 2 4 -4 -2 X 500 yrs 50 yrs 0 1 5 18 2 20 4 22 Y Y 5 0 X 6 1 -4 -2 0 X 2 4 24 -4 -2 0 X Fig. 14. Evolution of the knapping location and the Whallon’s diagram during the simulation. The scale of the artifacts maps is in meters. 107 300 variance / mean ratio concentration (artifacts / m2) A. Lenoble et al. / Journal of Archaeological Science 35 (2008) 99e110 0.25 m2 200 1 m2 100 5 1 20 4 m2 0 100 200 300 400 500 years Fig. 15. Evolution of the maximum artifact concentration (N/m2) during the simulation for 0.25, 1 and 4 m2 surfaces centred on the knapping spot. Each curve is the mean of 30 runs. The mean square is in grey. involves 2500 items larger than 2 cm and lies on a slope of 10e15 degree. Like at Petit-Bost, the concentration has an elongated shape oriented downslope (Fig. 5B). However, neither ellipsoid or fan-like shape, nor decreasing of the artifacts density in the slope direction can be observed. By contrast, this high-density area shows the following characteristics unlike the simulated distributions (Fig. 18): 1/64 1/16 1/4 1 4 16 64 grid (m2) Fig. 17. Whallon’s diagram, sector 2, Croix-de-Canard. features are known to characterize sorted solifluction lobes (Washburn, 1979; Bertran et al., 1995). This interpretation is strengthened by the fabric of the artifacts, which shows a unimodal orientation transverse to the slope in the front (Fig. 20), indicating compression. The objects displacement can be used to assume the time lapse during which solifluction occurred. Supposing that the original knapping spot was situated at the upslope part of the present feature, the stretching of the spot is about 2 m. - The upslope part yields scattered pieces. - Artifact density increases downslope. - The lateral and downslope limits are sharp. The latter takes the shape of a steep lobate front, ten to fifteen centimeters thick and transverse to the slope. - The size of the flakes increases downslope. This lateral size sorting is accompanied by a vertical size sorting, the larger objects covering the smaller ones. - Two significant peaks of concentration, underlines by the Whallon’s diagram for the grids 0.25 and 1 m2 (Fig. 19). A rough examination of the artifact distribution (Fig. 5B) shows that the best expressed peak (0.25 m2) is related to the flakes accumulation in the front of the concentration. variance / mean ratio This pattern is dissimilar to experimental knapping spots and to the simulated distributions as well. However, similar 5 1 1/64 1/16 1/4 1 grid 4 16 (m2) Fig. 16. Whallon’s diagram for level 1, Petit-Bost. 64 Fig. 18. Close-up view of the area F concentration, Cantalouette II. Vertical size sorting of flakes explains why only large pieces are visible at the surface of the concentration. 108 A. Lenoble et al. / Journal of Archaeological Science 35 (2008) 99e110 7. Discussion variance / mean ratio 10 5 1 1/64 1/16 1/4 grid 1 4 16 (m2) Fig. 19. Whallon’s diagram, area F, Cantalouette II. Measured rates of present-day solifluction lobe movement in mid-latitude environments range between 0.5 and 1 cm per year for moderate slopes (French, 1976; Bertran and Coutard, 2004). This indicates that between 100 and 200 years was necessary to produce such a deformation. N Area F, front N = 33 Fig. 20. Fabric of the artifacts in the front of the lobe, area F, Cantalouette II. The fabric associates a dominant orientation of a-axes perpendicular to the slope and a minor downslope orientation with an upslope dip. The key feature of the simulated modifications of a knapping spot by solifluction is a downslope translation of the centre of gravity of the spot together with an anisotropic diffusion. This results in a decrease in artifact density and gives the concentration an elongated shape. However, a dense relict area may remain upslope during the first stages of deformation, while the downslope limits become diffuse. With respect to the TRANSIT data, this kind of pattern typifies 100e200 years of solifluction activity. The corresponding downslope translation is respectively 4 and 8 m. Further stages of deformation are marked by a progressive homogenisation of the artifact distribution. A random distribution is reached after 400e500 years. Comparisons between archaeological and simulated data show that artifact redistribution by solifluction may explain the main features observed both at Petit-Bost and Croix-deCanard. It particularly enables the identification of associated ‘cultural’ features (a clear artifact concentration with a high refitting rate, originally identified as a knapping area), and obviously natural ones (the preferred slope-parallel orientation of the pieces) that seemed incompatible at first. According to this interpretation, the knapping area should correspond to a former, more classical knapping spot, which have been subsequently deformed in a periglacial environment. A similar interpretation also can be proposed for Croix-de-Canard, which should reflect an ‘ultimate’ stage of degradation. Permanence of ‘residual’ refittings, each involving a more limited number of pieces than at Petit-Bost, is then an important characteristic of the Croix-de-Canard assemblage. Finally, the characteristics of each sites and differences between sites can be interpreted both as reflecting solifluction-induced modifications after abandonment of the site but prior to its burial. The simulation allows the successive deformational patterns to be quantified. However, precise estimations of the time lapse during which the sites have undergone solifluction remains difficult, due to the assumptions associated to the simulation. These are: - the measured displacements used here have been made in a specific context that of La Mortice. The relative role of the processes involved in soil movement (frost-creep, pipkrakes, gelifluction) depends to a large extent on the physical and climatic parameters of the site. Therefore, the measurements are not necessarily representative of all sites where solifluction occurs. As an example, the TRANSIT data have been collected on slope gradients ranging from 8 to 14 degree. More gentle slopes may have some influence on the results. However, the data known for modern periglacial slopes do not enable one to draw clear conclusion about this problem. Drainage amelioration in connection to slope increase tends to reduce the possibilities of ice growth in the soil, and as a consequence, the gradient does not act always as a determinant factor on displacement. In some cases, slope and velocity can be uncorrelated or inversely correlated (Smith, 1992). The role of slope cannot be isolated from the other A. Lenoble et al. / Journal of Archaeological Science 35 (2008) 99e110 parameters (grain-size, precipitation, depth of freezing, topographic position.) that interact in soil displacement (French, 1976). In many periglacial milieus, solifluction is often associated with other processes such as overland flow and snow avalanches, which influence to a variable extent the displacement of the stones at the ground surface. Low frequency events with a high geomorphologic impact (rainstorms, exceptional snow falls) may play a crucial role on slope dynamics. The short field experiments (5 years) do not allow such phenomena to be taken into account. - Simplification of the processes that operate in natural milieus has been made in the simulation. The lateral variability of movements in solifluction lobes, lower on coarse grained than on matrix-rich frost susceptible areas (Coutard and Ozouf, 1996), has been ignored. In particular, artifacts were placed upslope of lobe fronts, and thus experimental data do not allow for taking into account the role of lobes. This would have lead to more complex deformation, particularly when considering large soil surfaces. Burial of the artifacts is also a complex mechanism, which is poorly described by the available experimental data. In other respects, the TRANSIT experiment does not enable one to simulate the influence of the initial artifact distribution on the final pattern. Nevertheless, the relationship between velocity and soil grain-size is well-known. Thick artifact concentrations that impede frost penetration in the underlying soil during daily freeze-thaw cycles may play an important role in deformation. The discrepancy between simulated pattern and observed features at Cantalouette II, located on a steeper slope than at Petit-Bost, reflects such an interaction between artifacts distribution and solifluction. This has resulted in the formation of a lobate, size-sorted front. For this reason, the simulation probably best applies to shallow, sheet-like solifluction that typifies low slope gradient (Akerman, 1993) and shallow freeze-thaw cycle environment where pipkrakes play an important role in soil movement. This is particularly the case in semi-arid environments in mid and low latitudes (Schumm, 1967; Francou and Bertran, 1997), but also on south-facing slopes in the northern hemisphere (Bertran and Coutard, 2004). 8. Conclusion This study provides new data concerning the modification of archaeological assemblages by solifluction. The simulated patterns that may typify artifact distribution as a function of time of exposition to solifluction are precisely documented. Despite some limitations, comparison to the Petit-Bost site shows that it can correctly describe archaeological cases and enables an understanding a set of observations that could not be explained intuitively. It also suggests that criteria like the presence of clear lithic concentrations and high refitting rates cannot be used as unambiguous evidence of a lack of perturbation. According to the proposed scenarios, Petit-Bost and 109 Croix-de-Canard should be considered as different stages in site degradation due to periglacial processes. The case of Cantalouette II indicates that other deformation patterns linked to solifluction are also possible, due to the interaction between lobe fronts and artifact concentration. The simulation enlightens some issues that require further investigation in order to better appreciate the processes operating upon burial, such as the influence of the original artifact distribution or the role of the stone-banked lobe fronts. Acknowledgements Special Thanks to Dr William Banks for improving the translation of this draft. References Akerman, H.J., 1993. Solifluction and creep rates 1972-1991, Kapp Linné, West Spitzbergen. In: Frenzel, B. (Ed.), Solilfuction and Climatic Variation in the Holocene. G. Fisher, Stuttgart, pp. 225e250 (European Science Foundation). Araujo, A.G.M., Marcelino, J.C., 2003. The role of armadillos in the movement of archaeological materials: an experimental approach. Geoarchaeology 18, 433e460. Barton, R.N.E., Bergman, C.A., 1982. Hunters at Hengistbury: some evidence from experimental archaeology. World Archaeology 1, 237e248. Baschelet, E., 1981. Circular Statistics in Biology. Academic Press, London. Ballantyne, C.K., Harris, C., 1994. The Periglaciation of Great Britain. Cambridge University Press, Cambridge. Bertran, P., 1994. Dégradation des niveaux d’occupation paléolithiques en contexte périglaciaire et implications archéologiques. Paléo 6, 285e302. Bertran, P., Coutard, J.-P., 2004. Solifluxion. In: Bertran, P. (Ed.), Dépôts de pente continentaux: dynamique et faciès, pp. 84e109 (Quaternaire Special Issue, Paris). Bertran, P., Bordes, J.-G., Barre, A., Lenoble, A., Mourre, V., 2006. Fabrique d’amas de débitage: données expérimentales. Bulletin de la Société Préhistorique Française 103, 33e47. Bertran, P., Francou, B., Pech, P., 1993. Stratogenèse assistée à la dynamique des coulées à front pierreux en milieu alpin, La Mortice, Alpes méridoniales, France. Géographie Physique et Quaternaire 47, 93e100. Bertran, P., Francou, B., Texier, J.-P., 1995. Stratified slope deposits: the stonebanked sheets and Lobes model. In: Slaymaker, O. (Ed.), Steepland Geomorphology. Wiley & Sons, London, pp. 147e169. Binford, L., 1981. Behavorial archaeology and the ‘‘Pompei premise. Journal of Anthropological Research 37, 195e208. Boëda, E., Pellegrin, J., 1985. Approche experimental des amas de Marsangy. Archéologie expérimentale. In: Les amas lithiques de la zone N19 du gisement de Marsangy: approche méthodologique par l’expérimentation. Association pour la Promotion de l’Archéologie de Bourgogne, Beaune, pp. 19e36. Bourguignon, L., Ortega, I., Sellami, F., Brenet, M., Grigoletto, F., Vigier, S., Daussy, A., Deschamps, J.-F., Casagrande, F., 2004. Les occupations paléolithiques découvertes sur la section Nord de la deviation de Bergerac: résultats préliminaires obtenus à l’issue des diagnostics. Préhistoire du Sud-Ouest 11, 155e172. Bourguignon L., Bertran P., Djema H., Henry-Duplessy S., Duplessy M., Teresa-Matamores J., Lahaye C., Bechtel F. 2005., Petit-Bost (Neuvicsur-l’Isle, Dordogne). Excavation Final Report, Bordeaux, Archaeological Regional Service, unpublished report. Bowers, P.M., Bonnichsen, R., Hoch, D.M., 1983. Flake dispersal experiments: non-cultural transformation of the archaeological record. American Antiquity 48, 553e572. Butzer, K., 1982. Archaeology as Human Ecology. Cambridge University Press, Cambridge. 110 A. Lenoble et al. / Journal of Archaeological Science 35 (2008) 99e110 Cahen, D., Moeyersons, J., 1977. Subsurface movements of stone artifacts and their implications for the prehistory of Central Africa. Nature 266, 812e815. Caine, N., 1968. The log-norma distribution and rates of soil movement: an example. Revue de Géomorphologie Dynamique 18, 4e46. Couchoud, I., 2003. Processus géologiques de formation du site moustérien du Roc de Marsal (Dordogne,France). Paléo 14, 51e68. Coutard, J.-P., Ozouf, J.-C., 1996. Modalités de la cryoreptation dans les Massifs du Chambeyron et de la Mortice, Haute-Ubaye, Alpes françaises du sud. Permafrost and Periglacial Processes 7, 21e51. Detrain L., Bats J.-C., Bertran P., Colonge D., Detrain L., Fourloubey, C., Grigoletto F., Lenoble A., L’Homme X., 2005. ‘‘La Croix de Canard’’ (Neuvic-sur-l’Isle, Dordogne). Excavation Final Report, Archaeological Regional Service, Bordeaux, unpublished report. Djindjian, F., 1991. Méthodes pour l’archéologie. A. Colin, Paris. Esdale, J.A., Le Blanc, R., Cinq-Mars, J., 2001. Periglacial geoarchaeology at the Dog Creek site, northern Yukon. Geoarchaeology 16, 151e176. Francou, B., Bertran, P., 1997. A multivariate analysis of clast displacement rates on stone-banked sheets, Cordillera Real, Bolivia. Permafrost and Periglacial Processes 8, 371e382. French, H.M., 1976. The Periglacial Environment. Longman, New York. Gullentops, F., Deblaere, C., 1992. Erosion et remplissage de la grotte Scladina. In: Otte, M. (Ed.), Recherches aux grottes de Sclayn. Le contexte, vol. 1. Etudes et Recherches Archéologiques de l’Université de Liège, Liège, pp. 9e31 (No. 27). Hansen, P.V., Madsen, B., 1983. Flint axe manufacture in the Neolithic. An experimental investigation of a flint axe manufacture site at Hasprup Vaenget, East Zealand. Journal of Danish Archaeology 2, 43e59. Harris, C., Davies, M.C.R., Coutard, J.-P., 1997. Rates and processes of periglacial solifluction: an experimental approaches. Earth Surface Processes and Landforms 22, 849e868. Hopkins, D.M., Giddings Jr., J.L., 1953. Geological background of Iyatayet archeological site, Cape Denbigh, Alaska. Smithsonian Miscellaneous Collections 121 (11). Washington. Johnson, F., 1946. An archaeological survey along the Alaska highway. American Antiquity 11, 183e186. Johnson, D.L., 2002. Darwin would be proud: bioturbation, dynamic denudation, and the power of theory in science. Geoarchaeology 17, 7e40. Kirkby, A., Kirkby, M.J., 1974. Surface wash at the semi-arid break in slope. Zeitschrift für Geomorphologie N.F. 21 (Suppl.), 151e176. Lenoble, A., 2004. L’abri Caminade. In: Texier, J.-P., Kervazo, B., Lenoble, A., Nespoulet, R. (Eds.), Sédimentogenèse des sites préhistoriques du Périgord. Association des sédimentologistes français, Paris, pp. 43e51. Lenoble, A., 2005. Ruissellement et formation des sites préhistoriques: référentiel actualiste et exemples d’application au fossile. In: British Archaeological Report International Series, No. 1363. Oxford. Lenoble, A., Bertran, P., 2004. Fabric of Palaeolithic levels: methods and implications for site formation processes. Journal of Archaeological Science 31, 457e469. Mackay, J.R., Matthews, W.H., MacNeish, R.S., 1961. Geology of the Engigstciak archaeological site, Yukon Territory. Arctic 14, 25e52. Newcomer, M.H., Sieveking, G., de, G., 1980. Experimental flake scatter-patterns: a new interpretative technique. Journal of Field Archaeology 7, 345e352. Poesen, J., 1987. Transport of rock fragment by rill flows e a field study. Catena Supplement 8, 35e54. Rahmani, N., Todisco, D., Desrosiers, M.P., Gendron, D., Bhiry, N., 2005. The geoarchaeology of the Tayara Site (Hudson Strait, Nunavik), Canada. In: Poster Presented at the First Developing International Geoarchaeology Conference, Saint-John, Canada, October 21e23, 2005. Schick, K., 1986. Stone age in the making: experiments in the formation and transformation of archaeological occurrences. In: British Archaeological Report International Series, No. 319. Oxford. Schiffer, M.B., 1972. Archaeological context and systemic context. American Antiquity 37, 156e165. Schiffer, M.B., 1983. Toward the identification of site formation processes. American Antiquity 48, 675e706. Schiffer, M.B., 1987. Formation Processes of the Archaeological record. University of New Mexico Press, Albuberque. Schumm, S.A., 1967. Rates of surficial rock creep on hillslopes in Western Colorado. Science 55, 560e561. Smith, D.J., 1992. Long-term rates of contemporary solifluction in the Canadian Rocky Mountains. In: Dixon, J.C., Abrahams, A.D. (Eds.), Periglacial Geomorphology, Proceedings of the 22nd Annual Symposium in Geomorphology. John Wiley and Sons, Chichester, pp. 203e221. Steinberg, J.M., 1996. Ploughzone sampling in Denmark: isolating and interpreting site signatures from disturbed context. Antiquity 70, 368e392. Texier, J.-P., 2001. Sédimentogenèse des sites préhistoriques et représentativité des datations numériques. In: Barrandon, J.-N., Guibert, P., Miche, V. (Eds.), Datation, XXIe rencontres internationales d’archéologie et d’histoire d’Antibes. APDCA, Antibes, pp. 159e175. Texier, J.-P., Bertran, P., 1993. Données nouvelles sur la présence d’un Pergélisol en Aquitaine au cours des dernières glaciations. Permafrost and Periglacial Processes 4, 183e198. Texier, J.-P., Bertran, P., Coutard, J.-C., Francou, B., Gabert, P., Guadelli, J.-L., Ozouf, J.-C., Plisson, H., Raynal, J.-P., Vivent, D., 1998. TRANSIT, an experimental archaeological program in periglacial environment: problematic, methodology, first results. Geoarchaeology 13, 433e473. Todisco, D., 1999. Vitesses et contrôle climatique des coulées à front pierreux de la Mortice (Haute-Ubaye, Alpes françaises du sud). Unpublished Master dissertation, University Paris 1-Sorbonne, Paris. Todisco, D., Bertran, P., Pech, P., 2000. Déplacements superficiels et contrôle climatique des coulées à front pierreux de la Mortice, Haute-Ubaye, Alpes françaises du Sud. Permafrost and Periglacial Processes 11, 97e108. Vallin, L., Masson, B., Caspar, J.-P., 2001. Taphonomy at Hermies, France: a Mousterian knapping site in a loessic context. Journal of Field Archaeology 28, 419e436. Van Vliet-Lanoë, B., 1988. Le rôle de la glace de segregation dans les formations superficielles de l’Europe de l’Ouest. Processus et Héritage. State thesis, University of Paris 1, Sorbonne, p. 854 Van Vliet-Lanoë, B., 1996. Relation entre la contraction thermique des sols en Europe du Nord-Ouest et la dynamique de l’Indlandsis weichsélien. Comptes Rendus de l’Académie des Sciences, Serie II A: Sciences de la Terre et des Planetes 322, 461e468. Waters, M.R., 1992. Principles of Geoarchaeology. The University of Arizona Press, Tucson. Whallon, R., 1973. Spatial analysis of occupation floors. I: application of dimensional analysis of variance. American Antiquity 38, 266e277. Washburn, A.L., 1979. Geocryology: A Survey of Periglacial Processes and Environments. Edward Arnold, London.