Navigation – Plan du site

AccueilNuméros56Radiocarbon dates for the early s...

Radiocarbon dates for the early shouldered Tardigravettian from the rockshelter of Chinchon 1 at Saumane-de-Vaucluse and the chronology of the recent Provençal Upper Palaeolithic

Jacques Élie Brochier
p. 3-7
Cet article est une traduction de :
Les datations radiocarbone du Tardigravettien ancien à crans de l’abri de Chinchon 1 à Saumane‑de‑Vaucluse et la chronologie du Paléolithique supérieur récent provençal [fr]

Résumés

Le complexe industriel à pointes à cran et lamelles à dos tronquées de l’abri de Chinchon 1, dans le Vaucluse, a été diversement interprété au cours des dernières décennies : pour certains, il se rattache de toute évidence à une phase très finale, et classique, du Magdalénien supérieur ; pour d’autres, il se rattache, de façon tout aussi évidente, à une phase ancienne, italique, du Tardigravettien. Les résultats des études stratigraphiques, sédimentologiques et paléontologiques plaident pour accorder un âge ancien à ce complexe qu’il n’avait pas été, jusqu’à présent, possible de dater de façon absolue. Nous présentons ici des jalons chronologiques absolus nouveaux pour cet épisode. Ils sont analysés dans le cadre d’une modélisation chronologique bayésienne. Leurs rapports chronologiques avec le Tardisolutréen salpêtrien du Languedoc et avec la période de fréquentation la plus récente de la grotte Cosquer sont précisés. On montre que Tardisolutréen salpêtrien (ancien) et Tardigravettien ancien à pointes à cran sont partiellement synchrones et que, dans la région tardigravettienne où s’ouvre la grotte Cosquer, c’est pendant la phase ancienne du Tardigravettien à pointes à cran qu’ont été réalisées les peintures et gravures de la seconde période de fréquentation de la grotte.

Haut de page

Notes de la rédaction

Received July 16, 2012 - Accepted after review on February 23, 2014

Texte intégral

1Level C of the rockshelter of Chinchon 1 at Saumane-de-Vaucluse has been the subject of much controversy, and sixty years after the discovery of the site interpretations remain entrenched : for some authors the industry from the shelter belongs to the final Magdalenian whereas for others it is part of the early shouldered Tardigravettian. These two taxonomic alternatives clearly have archaeological and environmental consequences, as well as chronological implications.

  • 1 The term Tardigravettian (and not Epiperigordian or Epigravettian), coined in 1963, is not a rando (...)

2In 1961, Maurice Paccard presented the different archaeological units of the rockshelter. He identified the shouldered points from level C as characteristic points of the most recent Magdalenian levels of the classic region and emphasized the “unusual stratigraphic position” of such objects below levels B1 and B, which he attributed to the Magdalenian V and VI respectively. As early as 1962, Denise de Sonneville-Bordes contested this typological interpretation and identified these points as being typical Laugerie-Basse points. In spite of the position of level C below two Upper Magdalenian levels (levels B and B1) and an Azilian level (level A), the assemblage was attributed to a final evolutionary phase of the Upper Magdalenian (Figure 1). The presence of shouldered points and truncated backed bladelets (a characteristic association) led Georges Laplace, in 1964, to compare the industry from level C with that of the early Italic Tardigravettien1, with shouldered pieces, thereby resolving the problem raised by the strange stratigraphic position of these pieces below three clearly Magdalenian-Azilian levels. Maurice Paccard adopted this interpretation in his later publications (Paccard and Dumas, 1977).

Fig. 1. Shouldered objects from the early Tardigravettian level from Chinchon 1, Vaucluse (1 to 13) and “fléchettes”[“darts”] (14 to 19)

Fig. 1. Shouldered objects from the early Tardigravettian level from Chinchon 1, Vaucluse (1 to 13) and “fléchettes”[“darts”] (14 to 19)

3In 1966, Denise de Sonneville-Bordes reiterated her interpretation, followed by Max Escalon de Fonton (1966), who referred to “typically Magdalenian shouldered points”, rather than to Laugerie-Basse points. His assessment of the situation is the following : level C, Magdalenian VIa ; level B, Magdalenian VIb, level A, typical Azilian. He categorically refused any comparison between the shouldered industry from Chinchon 1 and that of the cave of La Salpêtrière, as suggested by Georges Laplace (Figure 2). In 1967, Jean Combier carried out a detailed analysis of Provençal and Languedoc shouldered industries and drew a parallel between the industries from Oullins, Chinchon 1 and La Salpêtrière. For him, level C from Chinchon 1 should not be considered as a terminal Magdalenian. However, in 2003, he briefly alludes to level C from Chinchon and, without any justification whatsoever, ascribes it to a “Magdalenian with no definite affinities”.

Fig. 2. Salpetrian. Shouldered objects from La Rouvière, Ardèche (1 to 5) and cave of La Salpêtrière (early Salpetrian, layer 6 and d)

Fig. 2. Salpetrian. Shouldered objects from La Rouvière, Ardèche (1 to 5) and cave of La Salpêtrière (early Salpetrian, layer 6 and d)

4The most recent researcher to have discussed the affinities of level C from Chinchon 1 is Guillaume Boccaccio (2005), who is in disagreement with Jean Combier’s latter allusion. For him, this level can clearly be attributed to an “early Epigravettian”, although he compares the shouldered points from level C to the Magdalenian pieces from Fontgrasse.

5The stratigraphic elements assembled by Eugène Bonifay (1964), and then by this author (Brochier, 1977 ; Brochier and Livache, 1978), validate field observations and underline the existence of facies and phenomena during the deposition of level C (high loessic component, deformations linked to frost) hitherto unknown in Tardiglacial deposits in the region. The revision of the large mammal fauna by Évelyne Crégut-Bonnoure (Crégut-Bonnoure and Paccard, 1997) led to the conclusion that this “cold” association was incompatible with those characterizing recent Tardiglacial episodes and thus pointed towards an earlier age for level C.

6Although the charcoal from the Paccard collections has disappeared, and even though it is currently impossible to obtain a reliable bone collagen date, new radiocarbon determinations obtained from fragments of bird eggshell are now available. The following section consists of a commentary of these dates in their regional chronological and archaeological context.

Material and chronological model

7The radiocarbon findings obtained for the early Tardigravettian horizon with shouldered points from the Chinchon 1 rockshelter have been integrated into a set of dates from Provence and Languedoc, which can be organized into a chronological model based on stratigraphic and archaeological data. The simultaneous calibration of 28 measurements, using Bayesian modelling, enables us to provide new elements in the interpretation of the evolution of the different techno-complexes (Buck et al., 1994, 1996 ; Litton and Buck, 1995, Zeidler et al., 1998) as well as compare calendar dates for the different events.

Early Tardigravettian and upper Magdalenian in Provence

8Only nine radiocarbon determinations are available for clarifying the temporal order of the evolution of the Tardigravettian and Magdalenian techno-complexes (Figure 3, Table I).

Fig. 3. Geographic location of the sites mentioned in the text

Fig. 3. Geographic location of the sites mentioned in the text

Tabl I. Radiocarbon ages used and chronological model subjected to Bayesian analysis

Tabl I. Radiocarbon ages used and chronological model subjected to Bayesian analysis

The events are not ordered within each phase. The seven measurements obtained from paintings in Cosquer Cave (recent period) make up a phase without any a priori information. The δ13C values followed by an asterisk are measured by AMS on the graphite produced from the sample. They are thus only approximate (but nonetheless sufficient for calculating age) as they are biased by the graphitization and the AMS measurement. They should not be compared to the measurements obtained in other disciplines (study of speleothems or palaeo-diet, for example) by mass spectrometry; ÂRC: Conventional Radiocarbon Age (CRA); v.n.c.: non communicated value. The outlier column gives (a) a priori and (b) a posteriori probabilities so that a measurement needs to be translated on the radiocarbon scale in order to be coherent with the other measurements. C/N: carbon/nitrogen ratio of bone. As we are not aware of the possible results of bone carbon and nitrogen measurements from the Chinchon 1 and Charasse 1 samples and some samples from La Salpêtrière, before measurement of the 14C activity, we cannot assess the validity of the results obtained (n.m.: not measured)

9A single measurement is available for the early Tardigravettian with unifacial points. It comes from charcoal in layer 5 of La Baume Rainaudes 1. The rarity of charcoal in the levels underlying and overlying layer 5 (Bazile-Robert, 1981) probably limited the incorporation of intrusive charcoal in the sample.

10The early shouldered Tardigravettian could only be dated at the rockshelter of Chinchon 1. The measurements were carried out on unidentified bird eggshell fragments (Paccard and Dumas, 1977). Methodological research by Long et al. (1983) and Freundlich et al. (1989), discussed by Higham (1994), showed that the proportion of 14C present in the calcite forming the eggshell accurately reflected the 14C activity of food and atmospheric carbon dioxide. Bird eggshell therefore appears to be a reliable dating material (Magee et al., 2009) and is sufficiently common in archaeological soils to be used when more classic materials are missing or too poorly preserved.

11Five measurements concern the oldest known levels posterior to the early shouldered Tardigravettian phase. All of these determinations were obtained using AMS on a single object and all of these are related to the classic Upper Magdalenian phase. At Chinchon I (level B1) and in the large Charasse rockshelter (Laplace, 1966b), these measurements were obtained on bone collagen. However, we do not have the carbon and nitrogen content of the bone, or the collagen content, and therefore we cannot accept these dates without reservations (Ambrose, 1990 ; Bocherens et al., 2005; Higham et al., 2006 ; Yizhaq et al., 2005). The last two measurements were obtained from two Pinus t. sylvestris charcoals from levels Cn and Bn from the large Venasque rockshelter in the Vaucluse (Brochier, 2008).

Solutrean and early Salpetrian in Languedoc

12Eight radiocarbon measurements are available for the chronological position of the early Salpetrian from La Salpêtrière (Table I). The radiocarbon dates for the early Salpetrian are deemed to be very coherent by Frédéric Bazile, who considers that the early Salpetrian develops between 19,000 and 18000 (Bazile, 1990) or between 19,000 and 17,000 radiocarbon years before the present (Bazile and Boccaccio, 2008).

13In 1980, Frédéric Bazile proposed combining the radiocarbon measurements MC 2168 and MC 2083 on the one hand, and MC 2084 and MC 2186 on the other hand, given that in each case the measurements had been made on a fraction of the same sample. Even though we cannot rigorously consider these sample pairs to be strictly identical, we will adopt the same approach here. We observe first of all that the measurements MC 2168 and MC 2083, on the one hand, and MC 2084 and MC 2186, on the other, are not significantly different (Ward and Wilson, 1978): T = 3.2; ν = 1; p = 0.072 for the first pair, T = 3.1; ν = 1; p = 0.079 for the second. These tests were made in the F14C space, and not in the conventional radiocarbon ages (CRA) space, due to the approximately Gaussian nature of errors on the radiocarbon scale. The combination can thus be applied. The results used in the chronological model are the following: MC 2168/2083: 19,230 ± 200 BP; MC 2084/2186: 18,700 ± 220 BP.

14Six radiocarbon measurements obtained on batches of charcoal or bone from Solutrean horizons were used in the model (Table I and Figure 4). We restricted our sample to horizons with a clear provenance and which were stratigraphically close to those bearing early Salpetrian artefacts. The aim is to use the stratigraphic information to constraint the calendar dates of the early Salpetrian samples.

Fig. 4. Schema of the chronological relations between the different phases of the model

Fig. 4. Schema of the chronological relations between the different phases of the model

They are based on knowledge of the stratigraphic sequences and of the evolutionary processes of the different complexes. The αj and βj represent the calendar dates from the beginning and the end of the phase respectively

15The fourteen radiocarbon determinations taken into account do not present all the required guarantees for conducting an analysis that leaves no doubts whatsoever. Firstly, they are not very accurate – which can hardly be considered unusual for dates obtained several decades ago – and secondly, in some cases, they are not very exact. However, the potential problems here stem mainly from a rather tenuous association between the dated material and the archaeological reality. For this reason, we fixed an a priori outlier probability of 0.40 for all of the dates from this site (outlier column (a) in Table I).

The paintings from Cosquer cave

16We retained seven of the numerous direct radiocarbon measurements taken from the paintings in Cosquer Cave (Valladas et al., 2001, 2005), which define the second pictorial phase of the cave (Tabe I). The GifA 96101 determination with a jellyfish-shaped sign has not been taken into consideration.

The structure of the model

17The chronological model subjected to Bayesian analysis is presented in Table I and Figure 4. It is not very constrained, in that the events are generally not ordered into phases. The αj and βj (Table I) represent calendar dates from the beginning and the end of the phases. It is not very likely that the calendar date of event i from phase j (θij) is exactly the calendar date of the first (or the last) event of the phase. The evolution of the Provençal Upper Palaeolithic is structured into three successive phases, within which the calendar dates of the different events are not ordered. The a priori information used is based on our knowledge of the evolution of the industrial complexes and on stratigraphic arguments. The last set of measurements, from paintings from the second phase of activity in Cosquer Cave is naturally not ordered either. No a priori chronological information links the three sequences to each other.

18We stressed our reservations above as to the exact nature of certain measurements. However, rather than eliminating these dates, we prefer to allot an a priori pi outlier probability to each of them and then to subject them to simultaneous Bayesian calibration. The procedure advocated by Christen (1994) gives us for each a priori pi probability, the a posteriori Pi probability, so that the yi measurement needs to be shifted on the radiocarbon scale in order to be coherent with the other measurements of the phase. We interpret the clause “needs to be shifted” by “is an outlier”. The detection of outliers and calibration are part of the same calculation process and there is no need to eliminate outliers as the resulting chronological information undergoes inverse weighting.

19Calculations were made with the Mexcal simulation software via the BCal web interface (Buck et al., 1999). Countless simulations, conducted with different initial parameters, showed that the parameters of interest converge on similar values. The different HPD (Highest Posterior Density regions, or intervals formed of years with the highest posterior densities of probability) and the a posteriori modes of distributions of probability densities are stable within margins of some decades. We also conducted a sensitivity analysis with different a priori values of probabilities so that certain radiocarbon measurements were outliers.

20It may seem natural to use the most recent calibration curve, IntCal 09, in spite of the potential problems which could subsist around the Heinrich 1 period (Reimer et al., 2009, p. 1116). However, since the publication of IntCal 09, combined 234U – 230Th and 14C coral (Durand et al., 2010) and speleothem dates (Southon et al., 2012) have revealed that the measurements – obtained on foraminifera from varved sediments in the Cariaco Basin and used in the construction of the 2009 curve – included a temporarily overly-high reservoir effect and should thus be excluded from the reference curve for the 12,500 – 14,500 14C BP period (Shakun et al., 2012 ; Paula Reimer pers. comm.). Pending the forthcoming publication of an improved IntCal version, all the estimates have been obtained using the IntCal 09 reference curve, apart from the four oldest Magdalenian dates which were calibrated with reference to IntCal 04 (tab. II).

Tabl. II. Radiocarbon ages from the Gravettian and the early Tardigravettian from Arene Candide

référence
échantillons
référence
laboratoire
matériau C/N delta 13C ÂRC
P 1-2
P 3
P 4
Tardigravettien ancien
à crans
R 745
R 2546
R 2550
charbons charbons charbons

v.n.c.
v.n.c.
v.n.c.
18560 ± 210
18950 ± 245
18820 ± 260
 
P 7.3
P 8
P 9
Tardigravettien ancien
à pointes à face plane
R 2533
Beta 48684
R 2541
charbons charbons charbons

v.n.c.
v.n.c.
v.n.c.
19400 ± 230
19630 ± 250
20470 ± 320
 
P 10
P 11-12
P 13
Gravettien OxA 10700
Beta 53983
Beta 53982
os charbons charbons 3,2

– 17,6
v.n.c.
v.n.c.
23440 ± 190
23450 ± 220
25620 ± 200

C/N: carbon/nitrogen collagenratio; ÂRC: Conventional Radiocarbon Age (CRA)

Chronological results, archaeological consequences

21The chronological estimates obtained for the early Tardigravettian with shouldered points from Chinchon 1 show unequivocally that this horizon is clearly earlier than the Magdalenian-Azilian B1, B and A sequence from this site (Table I). They invalidate comparisons made by several prehistorians between the shouldered points from level C and the Rochereil and Laugerie-Basse points, thereby excluding ipso facto the techno-complex from Chinchon C to a very recent Magdalenian episode.

22Conversely, they validate the observations and interpretations of Georges Laplace (1966a, p. 318), Maurice Paccard (1964, 1977) and the evolutionary process brought to light in 1976 (Livache, 1976 ; Brochier and Livache, 1978 ; Livache, 2000). The results of the factorial analysis of the contingency table using 12 typological descriptors and 13 (then 12) Vauclusian series are summarized in Figure 5. First of all (Figure 5a), the whole series is taken into account. Two groups of samples are opposed on the first factorial axis with, on the negative side, elements from the Tardiglacial Magdalenian-Azilian, and on the positive side, the very disparate Tardigravettian samples (early Tardigravettian with unifacial points (LFP), shouldered early Tardigravettian (CC), evolved Tardigravettian with truncated backed blades (S6)), which all have high frequencies of truncated backed blades (LDT). The segregation on the first factorial axis has a chronological value. The series from La Font-Pourquière (Figure 6), with exceptionally high frequencies of the foliate (F) and truncated backed blade (LDT) groups, has an overpowering contribution of 0.82 to the first factorial axis. When this is removed from the data table, it is possible to specify the evolutionary process characterizing the recent Vauclusian Upper Palaeolithic (Figure 5b).

Fig. 5. Graphic results (plans F1.F2) of the factorial correspondence analyses applied to contingency tables linking typological groups and archaeological levels

Fig. 5. Graphic results (plans F1.F2) of the factorial correspondence analyses applied to contingency tables linking typological groups and archaeological levels

a: La Font-Pourquière (LFP), Chinchon 1 (C), Soubeyras (S), Sabon-Ayme (SAy); b: without La Font-Pourquière. The modal classes of the BCE calendar dates are given in square brackets. Descriptors. G: end scraper, R: side scraper, PD: backed point, T: truncation, D: denticulate, Bc: bec, B: burin, LD: backed blade, P: point, Cr: shouldered blade and point, LDT: truncated backed blade, F: leaf-shaped piece

Fig. 6. Retouched tools from the open-air site of La Font-Pourquière (Vaucluse)

Fig. 6. Retouched tools from the open-air site of La Font-Pourquière (Vaucluse)

Early Tardigravettian with unifacial points. 1 to 6: points with flat and simple retouch, 9 to 12: backed points and bi-points, 13: bi-truncated backed bladelets. After Livache and Carry, 1975

23However, it is important to note that we are removing the oldest reference from the series, whether we follow the Italic stratigraphic sequences (Laplace, 1966a ; Livache and Carry, 1975 ; Palma di Cesnola, 1993), those from the Var region (Livache and Brochier, 2003), or the only Provençal radiocarbon date available for an industry of this type (Rainaudes 1, layer 5). This chronological estimate does not contradict the results obtained further east in the Arene Candide Cave, for the early Tardigravettian phase with unifacial points and for the end of the Gravettian phase (Table II).

24Before proceeding with the chronological analysis of the shouldered Tardigravettian, which constitues the main objective of this study, we will analyse the early Tardigravettian with unifacial points. Data related to this issue come from the analysis of a chronological model based on Arene Candide, the shouldered phase from Chinchon 1 and layer 5 from Rainaudes 1 (Tables I and II, Figure 7).

Fig. 7. Schema of the chronological relations between the different phases of the model

Fig. 7. Schema of the chronological relations between the different phases of the model

These phases are based on knowledge of the stratigraphic sequences and the evolutionary processes of the different complexes. The αj and βj represent the calendar dates of the beginning and the end of the phase respectively

25Figure 8, issued from this analysis, shows that the calendar date for Rainaudes 1– layer 5 [HPD 95 % : 24030 – 21770 cal. BCE] is very close to the date for the beginning of the Tardigravettian phase with unifacial points from Arene Candide (HPD 95 % from the beginning and from the end : [24560 – 21460 cal. BCE], [21730 – 20510 cal. BCE]). There is one chance in two (p = 0.45) that the date from layer 5 for Rainaudes 1 is included in this phase, one chance in two that it is slightly anterior and 9.5 chances out of 10 that it is posterior to the “Gravettian” phase of Arene Candide.

Fig. 8. Distribution of the a posteriori probabilities of the calendar dates from the beginning and the end of the early Tardigravettian with unifacial points from Arene Candide, as defined by Georges Laplace (1964), and the early Tardigravettian with unifacial points from layer 5 from the cave of Rainaude 1 in the Var

Fig. 8. Distribution of the a posteriori probabilities of the calendar dates from the beginning and the end of the early Tardigravettian with unifacial points from Arene Candide, as defined by Georges Laplace (1964), and the early Tardigravettian with unifacial points from layer 5 from the cave of Rainaude 1 in the Var

26After these clarifications concerning the chronology of the initial phase of the early Tardigravettian, we can at present return to the central question of this study.

27The results of the factorial analysis of the table with the 12 typological descriptors and the recent Vauclusian Upper Palaeolithic levels have been presented and commented upon on numerous occasions (Brochier and Livache, 1978 ; Livache, 2000, for example). Therefore, we will merely recall the most conspicuous elements here (Figure 5). In the first factorial plane, the levels are projected following a parabolic crescent characteristic of a priori ordered series. The first axis appears to be a chronological axis on which levels follow the stratigraphy. Whithin reference framework, everything points to a transition from Chinchon’s level C (with shouldered elements and truncated backed blades, i.e. a characteristic association of the early shouldered Tardigravettian) to Upper Magdalenian levels (with abundant backed bladelets and burins) evolving towards the Azilian (with endscrapers, sidescrapers and backed points) through the intermediary of the original level 6, at the base of the Soubeyras stratigraphy. In other words, nothing rules out the possibility that the very classic Vauclusian Magdalenian-Azilian sequence stems from the early Tardigravettian (Brochier and Livache, 1978 ; Livache and Brochier, 2003), as opposed to possible middle or early Magdalenian complexes, which are clearly absent from the region. It is clear that the radiocarbon analysis fully validates this chronological interpretation of the first factorial axis.

28The main results of the chronological analysis are grouped in Tables III and IV and illustrated by Figures 9 to 11. The shouldered Tardigravettian phase identified at Chinchon 1 seems to be rather long [HPD 95 % : 2100 – 6550 years] between [HPD 95 % : 22,870 – 21,230 cal. BCE] and [HPD 95 % : 19,460 – 16,300 cal. BCE], much longer that the early Salpetrian episode from La Salpêtrière [HPD 95 % : 50 –1950 years], which is clearly incorporated in this timescale (p = 0.98). The Salpetrian Tardisolutrean rapidly follows the Solutrean [HPD 95 % (β5 - α6) : 0 – 620 years], which is hardly surprising. The latter ends at La Salpêtrière, whereas the Chinchon 1 – C complex had been in position for several centuries (p = 0.99). There are only a few centuries [HPD 95 % : 50 – 2450 years] between the early shouldered Tardigravettian from Chinchon 1 and the early Tardigravettian with unifacial points from Rainaudes 1 (Figure 10), and probably also those from Font-Pourquière and Bouverie 1f, given the extreme originality of this facies (Livache and Carry, 1975). A long hiatus [HPD 95 % : 400 – 4800 years] separates the early Tardigravettian with shouldered points from Chinchon 1 from the first industrial complexes with a Magdalenian structure (Figure 11). This could be partially filled by the episode with truncated backed bladelets from Soubeyras 6 (S6), which the factorial analyses (Figure 5) consider as the end of the transition between the Tardigravettian with shouldered points and the Magdalenian-Azilian phase (Livache, 1976 ; Brochier and Livache, 1978). These new chronological data show that the repeatedly underlined shared dynamics – homodynamism (Laplace, 1966a ; Livache and Brochier, 2004) between the Atlantic Epigravettian series (Solutrean phase with flat retouch, Salpetrian shouldered phase) and the Italic Tardigravettian series (phase with flat retouch then shouldered early Tardigravettian phase) were not totally synchronous on both sides of the Rhône.

Fig. 9. a. Distribution of the a posteriori probabilities of the calendar dates from the beginning (α2) and the end (β2) of the early shouldered Tardigravettian from Chinchon 1 and from the beginning (α3) of the Magdalenian phase; b.

Fig. 9. a. Distribution of the a posteriori probabilities of the calendar dates from the beginning (α2) and the end (β2) of the early shouldered Tardigravettian from Chinchon 1 and from the beginning (α3) of the Magdalenian phase; b.

Distribution of the a posteriori probabilities of the calendar dates from the beginning (α6) and the end (β6) of the early Salpetrian from La Salpêtrière; c. Distribution of the a posteriori probabilities of the calendar dates from the beginning (α4) and the end (β4) of the recent activity phase in Cosquer Cave. The two grey vertical bands are centred on the distribution modes of α2 and β2

Fig. 10. Bayesian estimation of the time span separating the early Tardigravettian with unifacial points from Rainaudes 1 from the early shouldered Tardigravettian from Chinchon 1 (θ1 – α2)

Fig. 10. Bayesian estimation of the time span separating the early Tardigravettian with unifacial points from Rainaudes 1 from the early shouldered Tardigravettian from Chinchon 1 (θ1 – α2)

Fig. 11. Bayesian estimation of the time span separating the early shouldered Tardigravettian from the first industrial complexes with a Magdalenian structure (β2 – α3)

Fig. 11. Bayesian estimation of the time span separating the early shouldered Tardigravettian from the first industrial complexes with a Magdalenian structure (β2 – α3)

Tabl. III. Results of the Bayesian calibration of radiocarbon determinations from the recent Upper Palaeolithic in the Vaucluse and Var region, from the early Salpetrian and Solutrean from the cave of La Salpêtrière and from the recent phase of paintings from Cosquer Cave

Tabl. III. Results of the Bayesian calibration of radiocarbon determinations from the recent Upper Palaeolithic in the Vaucluse and Var region, from the early Salpetrian and Solutrean from the cave of La Salpêtrière and from the recent phase of paintings from Cosquer Cave

The αj and βj are calendar dates from the beginning and the end of each phase, the θi are calendar dates for each event. The values in brackets correspond to measurements which appear to be outliers after this calculation. The column HPD 95% cal. BCE gives BCE calendar intervals where αj, βj and θi have 95 chances out of 100 of falling within this range. The cal. BCE column mode(s) specifies the position of the modal value(s) of the a posteriori probability distribution of the calendar dates. Reference curves: IntCal04 and IntCal09 (Reimer et al., 2004, Reimer et al., 2009)

Tabl IV. Bayesian estimations of the beginning, the end and the duration of the archaeological phases

HPD 95 % mode
durée de la phase tardigravettienne à crans [alpha2 – bêta2 2 100 – 6 550 2 950
durée séparant le tardigravettien ancien à pointes à face plane de Rainaudes 1 de la phase tardigravettienne à crans [thêta1 – alpha2] 50 – 2 450 650
durée séparant la phase tardigravetienne à crans
phase magdalénienne [bêta2 – alpha3]
400 – 4 800 3 850
durée séparant le Solutréen du Salpêtrien ancien [bêta5 – alpha6] 0 – 620 50
durée de l'épisode Salpêtrien ancien [alpha6 – bêta6] 50 – 1950 50
durée de la fréquentation (récente) de la grotte Cosquer [alpha4 - bêta4] 1 300 – 4960 2650
HPD 95 % cal. BCE mode
premières structures typologiques magdaléniennes (alpha 3) 16 520 – 14 230 14 700
bornes de la phase récente de Cosquer bêta 4 20 250 – 17 450 19 250
alpha 4 23 450 – 20 850 21 700

29Very soon after the discovery of Cosquer Cave and the direct dating of many paintings (Clottes et al., 1992, 1997 ; Valladas et al., 2001, 2005), the latter were linked to Solutrean or Salpetrian “cultures” (Clottes and Courtin, 1994 ; Onoratini, 1992). Yet not even the slightest trace of these two industrial complexes had ever been observed east of the Rhône (Brochier et al., 1993 ; Livache and Brochier, 2003), which is the domain of the Tardigravettian with Italic affinities. It would have been more pertinent to note that the age of the paintings from the recent phase of Cosquer Cave (phase 2) corresponded to that of the Solutrean and early Salpêtrian in Languedoc. We observe that activity in Cosquer Cave extends over several centuries [HPD 95 % : 1300 – 4960 years] between 21,700 cal. BCE (modal value, [HPD 95 % : 23,450 – 20,850 cal. BCE]) and 19,250 cal. BCE (modal value, [HPD 95 % : 20,250 – 17,450 cal. BCE]). This lapse of time corresponds exactly to that of the early Tardigravettian with shouldered points from Chinchon 1 (Table III, Figure 9). There is a 0.95 to 0.99 probability that one of the phase 2 paintings from Cosquer Cave has a calendar date within the limits of the Tardigravettian phase. The only exception concerns Horse 7 (θ15), for which the probability is only 0.86.

30Therefore, based on the typological, geographical and chronological data, it appears to be more appropriate to link the painted and engraved phase 2 from Cosquer Cave to the early Italic Tardigravettian world rather than to the early Atlantic Epigravettian world. During this period, approximately 21,700 and 19,250 years before our era, evidence of a Tardigravettian presence remains rare, and is still poorly positioned from a chronological viewpoint, but is dispersed throughout the whole Provençal region, from the Vaucluse to the coast, the Rhône axis to Liguria.

Haut de page

Notes

1 The term Tardigravettian (and not Epiperigordian or Epigravettian), coined in 1963, is not a random or unreasonable choice (Laplace, 1997; Brochier and Livache, 2003, p. 49; Livache and Brochier, 2003, p. 43). At the onset of the Last Glacial Maximum, the Gravettian west of the Rhône documents the rapid development of flat retouch: which has become known as the Solutrean phase, the initial phase of the (Atlantic) Epigravettian. This original episode, which represents a dis- continuity, thus separates the evolved Gravettian sensu stricto from the Salpetrian (Tardisolutrean). Although no kinship can be identified today, the successive recent Upper Palaeolithic horizons follow the Epigravettian process. In the absence of a Solutrean interruption, east of the Rhône, as in the Italic Peninsula, the Gravettian continues to evolve synchronously into diverse complexes, termed Tardigravettian. The terms Tardigravettian and Epigravettian thus both have an evo- lutionary connotation where the technical and structural mutation of Solutrean and post-Solutrean complexes can be considered to be the diverging point in the evolution of the Gravettian. The term Tardigravettian thus indicates an evolutionary continuum from the Gravettian onwards, whereas, conversely, the term Epigravettian underlines the evolution that took place after the qualitative rupture with the Gravettian.

Haut de page

Table des illustrations

Titre Fig. 1. Shouldered objects from the early Tardigravettian level from Chinchon 1, Vaucluse (1 to 13) and “fléchettes”[“darts”] (14 to 19)
URL http://journals.openedition.org/galliap/docannexe/image/277/img-1.jpg
Fichier image/jpeg, 1,7M
Titre Fig. 2. Salpetrian. Shouldered objects from La Rouvière, Ardèche (1 to 5) and cave of La Salpêtrière (early Salpetrian, layer 6 and d)
URL http://journals.openedition.org/galliap/docannexe/image/277/img-2.jpg
Fichier image/jpeg, 1,4M
Titre Fig. 3. Geographic location of the sites mentioned in the text
URL http://journals.openedition.org/galliap/docannexe/image/277/img-3.jpg
Fichier image/jpeg, 1,4M
Titre Tabl I. Radiocarbon ages used and chronological model subjected to Bayesian analysis
Légende The events are not ordered within each phase. The seven measurements obtained from paintings in Cosquer Cave (recent period) make up a phase without any a priori information. The δ13C values followed by an asterisk are measured by AMS on the graphite produced from the sample. They are thus only approximate (but nonetheless sufficient for calculating age) as they are biased by the graphitization and the AMS measurement. They should not be compared to the measurements obtained in other disciplines (study of speleothems or palaeo-diet, for example) by mass spectrometry; ÂRC: Conventional Radiocarbon Age (CRA); v.n.c.: non communicated value. The outlier column gives (a) a priori and (b) a posteriori probabilities so that a measurement needs to be translated on the radiocarbon scale in order to be coherent with the other measurements. C/N: carbon/nitrogen ratio of bone. As we are not aware of the possible results of bone carbon and nitrogen measurements from the Chinchon 1 and Charasse 1 samples and some samples from La Salpêtrière, before measurement of the 14C activity, we cannot assess the validity of the results obtained (n.m.: not measured)
URL http://journals.openedition.org/galliap/docannexe/image/277/img-4.jpg
Fichier image/jpeg, 1,6M
Titre Fig. 4. Schema of the chronological relations between the different phases of the model
Légende They are based on knowledge of the stratigraphic sequences and of the evolutionary processes of the different complexes. The αj and βj represent the calendar dates from the beginning and the end of the phase respectively
URL http://journals.openedition.org/galliap/docannexe/image/277/img-5.jpg
Fichier image/jpeg, 372k
Titre Fig. 5. Graphic results (plans F1.F2) of the factorial correspondence analyses applied to contingency tables linking typological groups and archaeological levels
Légende a: La Font-Pourquière (LFP), Chinchon 1 (C), Soubeyras (S), Sabon-Ayme (SAy); b: without La Font-Pourquière. The modal classes of the BCE calendar dates are given in square brackets. Descriptors. G: end scraper, R: side scraper, PD: backed point, T: truncation, D: denticulate, Bc: bec, B: burin, LD: backed blade, P: point, Cr: shouldered blade and point, LDT: truncated backed blade, F: leaf-shaped piece
URL http://journals.openedition.org/galliap/docannexe/image/277/img-6.jpg
Fichier image/jpeg, 322k
Titre Fig. 6. Retouched tools from the open-air site of La Font-Pourquière (Vaucluse)
Légende Early Tardigravettian with unifacial points. 1 to 6: points with flat and simple retouch, 9 to 12: backed points and bi-points, 13: bi-truncated backed bladelets. After Livache and Carry, 1975
URL http://journals.openedition.org/galliap/docannexe/image/277/img-7.jpg
Fichier image/jpeg, 1,7M
Titre Fig. 7. Schema of the chronological relations between the different phases of the model
Légende These phases are based on knowledge of the stratigraphic sequences and the evolutionary processes of the different complexes. The αj and βj represent the calendar dates of the beginning and the end of the phase respectively
URL http://journals.openedition.org/galliap/docannexe/image/277/img-8.jpg
Fichier image/jpeg, 308k
Titre Fig. 8. Distribution of the a posteriori probabilities of the calendar dates from the beginning and the end of the early Tardigravettian with unifacial points from Arene Candide, as defined by Georges Laplace (1964), and the early Tardigravettian with unifacial points from layer 5 from the cave of Rainaude 1 in the Var
URL http://journals.openedition.org/galliap/docannexe/image/277/img-9.jpg
Fichier image/jpeg, 271k
Titre Fig. 9. a. Distribution of the a posteriori probabilities of the calendar dates from the beginning (α2) and the end (β2) of the early shouldered Tardigravettian from Chinchon 1 and from the beginning (α3) of the Magdalenian phase; b.
Légende Distribution of the a posteriori probabilities of the calendar dates from the beginning (α6) and the end (β6) of the early Salpetrian from La Salpêtrière; c. Distribution of the a posteriori probabilities of the calendar dates from the beginning (α4) and the end (β4) of the recent activity phase in Cosquer Cave. The two grey vertical bands are centred on the distribution modes of α2 and β2
URL http://journals.openedition.org/galliap/docannexe/image/277/img-10.jpg
Fichier image/jpeg, 759k
Titre Fig. 10. Bayesian estimation of the time span separating the early Tardigravettian with unifacial points from Rainaudes 1 from the early shouldered Tardigravettian from Chinchon 1 (θ1 – α2)
URL http://journals.openedition.org/galliap/docannexe/image/277/img-11.jpg
Fichier image/jpeg, 147k
Titre Fig. 11. Bayesian estimation of the time span separating the early shouldered Tardigravettian from the first industrial complexes with a Magdalenian structure (β2 – α3)
URL http://journals.openedition.org/galliap/docannexe/image/277/img-12.jpg
Fichier image/jpeg, 181k
Titre Tabl. III. Results of the Bayesian calibration of radiocarbon determinations from the recent Upper Palaeolithic in the Vaucluse and Var region, from the early Salpetrian and Solutrean from the cave of La Salpêtrière and from the recent phase of paintings from Cosquer Cave
Légende The αj and βj are calendar dates from the beginning and the end of each phase, the θi are calendar dates for each event. The values in brackets correspond to measurements which appear to be outliers after this calculation. The column HPD 95% cal. BCE gives BCE calendar intervals where αj, βj and θi have 95 chances out of 100 of falling within this range. The cal. BCE column mode(s) specifies the position of the modal value(s) of the a posteriori probability distribution of the calendar dates. Reference curves: IntCal04 and IntCal09 (Reimer et al., 2004, Reimer et al., 2009)
URL http://journals.openedition.org/galliap/docannexe/image/277/img-13.jpg
Fichier image/jpeg, 1,3M
Haut de page

Pour citer cet article

Référence papier

Jacques Élie Brochier, « Radiocarbon dates for the early shouldered Tardigravettian from the rockshelter of Chinchon 1 at Saumane-de-Vaucluse and the chronology of the recent Provençal Upper Palaeolithic »Gallia Préhistoire, 56 | 2016, 3-7.

Référence électronique

Jacques Élie Brochier, « Radiocarbon dates for the early shouldered Tardigravettian from the rockshelter of Chinchon 1 at Saumane-de-Vaucluse and the chronology of the recent Provençal Upper Palaeolithic »Gallia Préhistoire [En ligne], 56 | 2016, mis en ligne le 30 octobre 2017, consulté le 17 avril 2025. URL : http://journals.openedition.org/galliap/277 ; DOI : https://doi.org/10.4000/galliap.277

Haut de page

Auteur

Jacques Élie Brochier

UMR 7269 – LAMPEA, Aix-Marseille Université, CNRS, Ministère de la culture et de la communication, 5 rue du Château de l’Horloge, BP 647, F-13094 Aix-en-Provence, cedex 2

Haut de page

Droits d’auteur

Le texte et les autres éléments (illustrations, fichiers annexes importés), sont « Tous droits réservés », sauf mention contraire.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search