Defining Magdalenian Cultural Groups in Franco-Cantabria by the Formal Analysis of Portable Artworks

Deﬁning Magdalenian cultural groups in Franco-Cantabria by the formal analysis of portable artworks ∗ Olivia Rivero & Georges Sauvet

The motifs, techniques and stylistic features of Upper Palaeolithic art offer enormous Paris potential for the investigation of social and cultural interactions in south-western France and northern Spain during the later stages of the last ice age. The key regions of Aquitaine, Cantabria and the Pyrenees clearly share an overall family resemblance, but detailed analysis of horse heads on portable objects of bone, antler and stone from Magdalenian contexts reveal that particular features can be attributed to different Madrid regions at different periods. Furthermore, the

0 km 500 patterns of interconnection are structured very N differently in the Upper Magdalenian than in the Middle Magdalenian, perhaps as rising temperatures in the latter period led to territorial expansion and social realignment. Keywords: France, Spain, Upper Palaeolithic, Magdalenian, mobiliary art, cave art, stylistic analysis, correspondence factor analysis, ascending hierarchical clustering

Supplementary material is published online at http://antiquity.ac.uk/projgall/rivero339

Introduction Style in archaeology may be used in different ways and serve various purposes (Conkey & Hastorf 1990). The pioneers in the study of Palaeolithic art used it as a chronological marker within an evolutionary perspective (Breuil 1952). Leroi-Gourhan (1965) subsequently proposed a chronocultural scheme of four stages, based on the evolution of animal representations using features such as the perspective of horns and hooves, the number of legs by pair and the presence of anatomical details and conventions. In this culture- historical approach, a unique evolution was assumed to take place everywhere and to follow

* CREAP “Cartailhac”, TRACES-UMR 5608, Universit´e de Toulouse-le-Mirail, All´ee A. Machado, 31058 Toulouse-Cedex 9, France (Email: [email protected]; [email protected])

C Antiquity Publications Ltd. ANTIQUITY 88 (2014): 64–80 http://antiquity.ac.uk/ant/088/ant0880064.htm 64 n rjcin ftepit ntepae endb h rnia xso nri lead by inertia achieved of is axes objects the principal of the Clustering by interpret. defined and planes visualise the mathematically to calculated in an easy or are in representations points absence points coordinates to of their the its cloud for of representing the 0/1 projections of 0/1 as inertia of and of (noted set axes values a The by of space. denoted dimensional range is determined individual a Each taking presence). attributes of set designed a well is CFA subjectivity. features, minimal the of with in sets typology appear with of objects objects questions of address these sets to of Associating features cluster. characteristic the the of vicinity and together clustered are profiles southern and Spain northern between relationships social have the (Fortea and Palaeolithic to and Upper of dynamics pursued the able vision during been their France already renewed is about have a art premise us to this Palaeolithic inform on led of and based Studies networks analysis time. exchange through morphological evolution of and knowledge formal processes our a improve the viewpoint, Interpreted in 65–70). between this 1987: role relationships (Ucko Palaeolithic social from active Upper and identity, the cultural an during the their hunters-gatherers identify playing to of archaeologists groups about by sign Therefore, also others used 1977). visual (Wobst be inform may interactions a it social to and as communication acts exchange, individuals information Style of of status. groups manufactured and to or given affiliation design individuals and form by particular the objects as considered millennia. be of should Style tens society. spanning period a over evolution stylistic of view linear this eas bet n rprisaesmerclytetd hi rxmt ntevarious the in proximity their (Benz analyses; treated, significant multivariate symmetrically is powerful are representations most graphical properties the of and one objects is because (CFA) analysis factor Correspondence methods and Materials raw of origin their industries, shows bone and time (lithic through data to networks etc.). archaeological due materials, the other interpenetration to of relation of variability in geographical level the dynamics the addition, their In quantitatively and exchanges. examining features cultural conduct stylistic of to morphological possibility us allowed of the occurrences distribution offering of number analysis The statistical artworks. in a richest Magdalenian the the on is and which art) Palaeolithic period, in animal represented representations frequently on most attention (the horses our of concentrated have We levels. archaeological well-defined in therefore offers approach. context, this stratigraphic for reliable support a variations best in the stylistic found when contemporaneous art, only Portable since meaningful. required, are still is framework chronological a & Rivero press; in the recent However, in Magdalenian. schemes (Ard the geometric Cave in clumsy Chauvet reached as first realism’ such the discoveries ‘photographic from the progress to technical Aurignacian of direction same the F suulyapidt ut-aaee aaesi hc ahojc sdsrbdby described is object each which in datasets multi-parameter to applied usually is CFA in style of role the aside sets it because misleading is approach external this Moreover, nti td,w xmn ag e fpral rwrso oe nlradsoefound stone and antler bone, on artworks portable of set large a examine we study, this In Alvarez-Fern ´ lvaRvr ere Sauvet Georges & Rivero Olivia ne 09.Hr hoooyi olne h ol but goal, the longer no is chronology Here 2009). andez ´ ce rne aesontedfclyo accepting of difficulty the shown have France) eche, ` 65 tal. et ci&Benz & ecri ´ 04 Fritz 2004; ci18) bet ihsimilar with Objects 1984). ecri ´ tal et 07 Sauvet 2007; . C niut ulctosLtd. Publications Antiquity tal et 2008, . n -

Research Defining Magdalenian cultural groups in Franco-Cantabria by the formal analysis of portable artworks ascending hierarchical clustering (AHC), using their factorial coordinates and the second- order central moment of a maximum partition as criteria of aggregation (Jambu & Lebeaux 1978). The dendrogram could be easily divided into three groupings that were distinguished in the early steps of the clustering process. Statistical tests were used to determine the features specifically attached to each group. Only associations between a feature and a group having a significance level higher than 95 per cent (with respect to a random distribution), were retained for discussion (Chenorkian 1996). CFA has been used in various archaeological domains and was found particularly useful in the case of artistic items when the objective was to correlate formal features with the human groups who produced them in order to study their differences, degree of independence and interrelationships (Villaverde Bonilla 1994: 203–209; Buisson et al. 1996; Tosello 2003: 507–15; Pigeaud 2005; Bourrillon 2009; Petrognani 2009). In the present study, a set of 273 horse representations from Magdalenian portable art was selected. Portable art (i.e. engravings on small pieces of bone, antler or stone) was chosen because it affords a much better account of chronology than parietal art, which is mainly dated by stylistic appreciation, leading often to circular reasoning. The Magdalenian period was chosen because it is the richest in artworks and because a large set of independent data is already available concerning the use of territories by hunters during this period (flint supply, Mediterranean and Atlantic shells used as personal ornaments, etc.). Very often, animals are represented in Palaeolithic art by their head or foreparts alone. In addition most pieces are broken, so we have focused our attention on the head and the neck, which are the parts of the body most frequently represented. This drawback is counterbalanced by the fact that the head is particularly rich in distinctive features. To be able to study the evolution of form through time, we have used the common distinction between Middle Magdalenian (MM) and Upper Magdalenian (UM) according to the archaeological horizons from which they come. Only 10 pieces could not be assigned precisely and were left as ‘Indefinite Magdalenian’ (IM). It is worth noting that the term ‘Middle Magdalenian’ was used without further subdivision owing to the rarity of pieces from the lower part of the MM in our corpus. For example, pieces from the so-called ‘Magdalenian III’ from La Marche were discarded due to the difficulty of describing them accurately from publications, and horses are almost absent in the ‘Cantabrian Lower Magdalenian’. In respect to geography, we have used only very large regional areas, taking the Cantabrian region as a whole and Aquitaine in a broad sense, including the Aveyron Valley and Quercy (Figure 1). Chronological and regional assessments are summarised in Tables 1 and 2. For the present study, the formal features used for the description of horse heads were not determined aprioribut were specially adapted to the corpus by a recursive process in order to obtain the best description. We retained 17 formal attributes, each with two or more values or modalities according to the variability and distinctiveness of the feature (Table 3). In other studies dedicated to the formal analysis of horse representations, the metrical proportions between the body and the head were found useful (Pales & Tassin de Saint-Pereuse 1981; Apellaniz & Calvo 1999; Pigeaud 2007), but measurements were found meaningless, in our study, for the analysis of heads alone. Our purpose being to highlight the criteria on which stylistic appreciation and differentiation could be made, we found that the most C Antiquity Publications Ltd.

66 atlo )L am;9 lVle 0 ueta 1 rig;1)Tre 3 but;1)Itrt;1)Aacu 16) Arancou; 15) horse Isturitz; with 14) Abauntz; sites 13) the Torre; and 12) Poeyma Urtiaga; study Vi 20) 11) this Espalungue; La Lumentxa; 19) in Saint-Michel; 3) 10) 18) discussed Valle; Caldas; Brassempouy; El areas 17) Las Dufaure; 9) the 2) Garma; showing Paloma; La France La 8) southern 1) Castillo; and CFA: Spain by analysed northern depictions of Map 1. Figure 1 aVce 2 ae;3)L ruae 4 otsrc 5 eCubt 6 aae 7 Fontal 37) Lafaye; 36) Courbet; Le 35) Montastruc; 34) Crouzade; La 33) Gazel; 32) Vache; La 31) 4 ora;2)Lsuu;2)Enl 26) Lespugue; 25) Gourdan; 24) 9 aid;4)L oc;4)Lmul 2 aMdlie 3 agreBse 4 ilpn 5 e yis 6 Peyrat; 46) Eyzies; Les 52) 45) Cepoy; 51) Villepin; Coteaux; 44) des Laugerie-Basse; Taillis 50) 43) Marche; Madeleine; La La 49) 42) Rochereil; Limeuil; 48) Raymonden; 41) 47) Soucy; Le 40) Lalinde; 39) ohrd aCil;5)L oteRfa;5)L Colombi La 56) Roffat; Goutte La 55) Caille; la de Rocher oa 19 2 7273 17 10 104 159 123 – 9 8 92 5 66 52 41 2 15 75 3 14 24 Total Magdalenian Indeﬁnite Magdalenian Upper Magdalenian Middle horses. Magdalenian of corpus the of distribution regional and Chronological 1. Table ` n;2)L u ’uobr;2)L a ’zl 9 eKrd ast 0 B 30) Massat; de Ker Le 29) d’Azil; Mas Le 28) d’Audoubert; Tuc Le 27) ene; atbinNrhr and Northern Cantabrian einPrne qian atr rneTotal France eastern Aquitaine Pyrenees region lvaRvr ere Sauvet Georges & Rivero Olivia ` r;5)Campalou. 57) ere; 67 a )Tt utlo )L ia )E ed;7 El 7) Pendo; El 6) Pila; La 5) Bustillo; Tito 4) na; ˜ ;2)LsEsp Les 21) u; ¨ tols 3 aBueNie 54) Noire; Baume La 53) Etiolles; ´ lge;2)Lbsie 3 Lortet; 23) Labastide; 22) elugues; ´ C niut ulctosLtd. Publications Antiquity ` s 8 Sainte-Eulalie; 38) es; ´ edeilhac;

Research Deﬁning Magdalenian cultural groups in Franco-Cantabria by the formal analysis of portable artworks

Table 2. Sites used for CFA and the number of horses analysed in each one. Cantabrian region Pyrenees Aquitaine Northern and eastern France

Las Caldas (20) Arancou (1) Courbet (5) La Baume Noire (1) La Vina˜ (1) Brassempouy (7) Les Eyzies (2) La Colombiere` (7) La Paloma (1) Bedeilhac´ (3) Fontales` (1) Cepoy (1) Tito Bustillo (3) La Crouzade (1) Lalinde (1) Campalou (1) La Pila (1) Dufaure (1) Laugerie-Basse (11) Etiolles´ (1) La Garma (2) Espalungue (7) Lafaye (1) La Goutte Roffat (1) El Pendo (5) Enlene` (2) Limeuil (24) Rocher de la Caille (5) El Castillo (1) Les Espelugues´ (1) La Madeleine (48) El Valle (2) Gourdan (7) Montastruc (22) Lumentxa (1) Gazel (1) La Marche (1) Urtiaga (1) Isturitz (12) Peyrat (1) Torre (1) Lortet (4) Rochereil (1) Abauntz (2) Labastide (8) Raymonden (1) Lespugue (2) Sainte-Eulalie (1) Le Mas d’Azil (22) Le Soucy (1) Le Ker de Massat (1) Taillis des Coteaux (1) Poeymau¨ (3) Villepin (1) Saint-Michel (2) Le Tuc d’Audoubert (1) La Vache (6)

useful formal characteristics were the manner of drawing the different parts of the outline (frontal line, mane, neck, mandible)—by a continuous line or a series of oblique hatchings, and sometimes by a mixture of both—and the attention paid to anatomical details such as the sense organs (eye, nostril, mouth, ears). For the eye, different modalities were observed (oval, point, stroke) and for the mouth a particular motif in the form of a hook was often found in the cut-out contours on hyoid bone (contours d´ecoup´es) (Buisson et al. 1996). Care was taken to indicate other anatomical details, such as the superior and inferior arches of the eye; the hairless zones around the snout or around the nostrils were also considered relevant. Other traits such as the indication of the lacrimal caruncle and/or the external commissure; the presence of a prominent crest and a forelock mane; a prolongation of the mandible at the rear of the head, going up to the ears in some cases; the presence of invasive hatching on the jaw and the cheek to indicate the facial coat; and the conventional delimitation of the mane by a double outline were also taken as potential markers of groups of artists.

Results For the CFA analysis, a table of 273 rows (the 273 representations of horses) and 53 columns (the values of the 17 attributes in Table 3) was used. Four columns (Foh, Moa, Fdm, Rmm) were classiﬁed as supplementary elements (SE) and did not participate in the construction of the axes of inertia, because they are very rare (less than 2 per cent of the objects have these properties). C Antiquity Publications Ltd.

68 rho h y ierElasn Ma Eys absent stroke Nop present Rmm mixed Eal Nostril Fdl Lcp Nkl linear present linear demarcation Fronto-nasal linear caruncle Lacrimal Mcp Ndl Mol eye the of Arch Code linear Value demarcation linear Sdl Nostril Neckline yes Msl linear outline linear Maxillary simple Attribute demarcation Snout crest Code in Mane Fol Value linear Mane outline Fronto-nasal Attribute nnse,adsm fte ih eatiue oieprecdatss(ieo2011). (Rivero artists inexperienced to attributed be might them of some and unfinished, ( organs sense of deprived elaboration, little showing representations niaino nu rdmraino otis( nostrils ( of eye demarcation or snout of indication h rnia enn etr fti ru,snei sue o nyfrteotie( outline the for only not used is Moh it since Nkh, group, this of feature defining principal the (attributes mouth hook-shaped or simple ie ( lines ( face whole the of (115 A Group C. attributes and 3, Table B by represented caruncle; groups (often lacrimal eye from the the of clearly with were indication oval A C) with an horses and group detailed B most separates (A, the 1 includes groups examples) Axis well-defined Three AHC. [1,2]. by plane found factorial main the in projection al .Atiue n ausue o h ecito fhrehasi adlna otbeart. portable Magdalenian in heads horse of description the for used values and Attributes 3. Table rusBadCetn ln xs2 ru 10eape)i hrceie by characterised is examples) (100 B Group 2. axis along extend C and B Groups F ple oti al ie h tagtowr eutsoni iue2b the by 2 Figure in shown result straightforward the gives table this to applied CFA Eaa o,Nkm). Mom, .Ms fteefiue r rbbysml kthswieohr a edeliberately be may others while sketches simple probably are figures these of Most ). ,btas o nenlaaoia eactos( demarcations anatomical internal for also but ), Fcp .GopAas nldsahbi ehiu sn ohhthn and hatching both using technique hybrid a includes also A Group ). betFaasn Fla Eya Eyo absent Flh absent oval hatching Eyp Mh Noa punctiform hook-shaped Ms Era Forelock Fda absent Fdm Rma Erp simple Lca absent Fdh Eaa present Eye absent absent absent mixed Eah Rmh hatching Mouth Nka Nkm Nda absent hatching hatching Ndm Nkh Ear Moa absent absent Sda Mom Ndh mixed Sdh mixed Rml hatching Mca Moh hatching absent hatching absent mixed linear hatching Mna mandible of Rear no Mdm Mdl indefinite Msm mixed double linear double Msh mixed simple Fne Foh hatching simple edge natural hatching lvaRvr ere Sauvet Georges & Rivero Olivia r,Np s Mh Ms, Nop, Erp, 69 d,Nda Sda, ailca rsn Fcp present coat Facial y,Lcp Eyo, n oace bv rblwthe below or above arches no and ) .Tetcnqeo acigis hatching of technique The ). d,Sh Eah Sdh, Ndh, ,teer h oti,ada and nostril, the ear, the ), C y,Na Ma Noa, Eya, niut ulctosLtd. Publications Antiquity betFca absent n h filling the and ) ,wt no with ), m,Flh, Rmh,

Research Deﬁning Magdalenian cultural groups in Franco-Cantabria by the formal analysis of portable artworks

Figure 2. CFA multivariate plot of the studied corpus of 273 horses from Magdalenian portable art in the main factorial plane [1,2]. The three coloured groups A, B and C are generated by ascending hierarchical clustering.

Group C (58 examples) comprises horses with a mane formed by a double line (Mdl), sometimes with hatching in between (Mdm), and horses with a prominent mane in the form of a crest (Mcp), these two features being frequently associated. In group C, hatching is no longer employed, and the outline is made by simple lines (Mol) as well as internal demarcations where they are present (Sdl). Thus, CFA applied to a corpus of horses from Magdalenian portable art led us to distinguish three main groups of figures characterised by sets of formal parameters that couldbeassimilatedtoformal concepts. The first of these, and the most elaborate, is constituted by figures making extensive use of hatching both for the outline and internal details (Figure 3). The second employs simple linear outlines without sense organs (Figure 4) and the third group contains horses with special features such as the so-called ‘double mane’ (i.e. represented by a double line), and/or a prominent crest (Figure 5). C Antiquity Publications Ltd.

70 ieo21) 0)Itrt atrRvr 00;G2 aGra(fe ieo21) a)Lbsie(fe Simonnet (after (after Caldas Labastide Las La1) C01) 2010); 1996); Rivero Mazo indicated. B (after Be1) & Garma 1914); Maury Utrilla La (after (after Laugerie-Basse Ga2) LB4–LB5) Abauntz 2010); A1) Rivero A: al. (after group et Isturitz to I03) belonging 2010); horses Rivero of Examples 3. Figure 99;MoMp eMsdAi atrClottes (after d’Azil Mas Le Mao-Map) 1989); lvaRvr ere Sauvet Georges & Rivero Olivia ´ dihc(fe az avt19) l clsi muls otherwise unless cm in scales All 1991). Sauvet & Jauze (after edeilhac tal. et 71 91;V–2 aVce(fe lte epre2003); Delporte & Clottes (after Vache La V1–V2) 1981); C niut ulctosLtd. Publications Antiquity

Research Deﬁning Magdalenian cultural groups in Franco-Cantabria by the formal analysis of portable artworks

Figure 4. Examples of horses belonging to group B: Be3) Bédeilhac (after Jauze & Sauvet 1991); I01) Isturitz (after Rivero 2010); Lf1) Lafaye (after Ladier & Welté 1991); Li2–Li3) Limeuil (after Tosello 2003); LP1) La Paloma (after Corchon´ 1986); U1) Urtiaga (after Barandiaran´ 1973); Ms1) Le Ker de Massat (after Barrière 1990); M36) La Madeleine (after Sieveking 1987); Mt7) Montastruc (after Sieveking 1987); Po1-Po2-Po3) Poeymau¨ (after Barandiaran´ & Laplace 2000). All scales in cm unless otherwise indicated.

C Antiquity Publications Ltd.

72 atrTslo20) 5 aVce(fe lte epre20) a–a)E al atrCorch (after Valle El Va1–Va2) 2003); Delporte indicated. & otherwise Clottes unless (after cm Vache La V5) 2003); Tosello (after iue5 xmlso ossblnigt ru :A)Aanz(fe til ao19) 2)L Madeleine La M24) Corch 1996); (after Mazo Lumentxa & Lu1) 2003); Utrilla Tosello (after (after Abauntz Soucy So1) A2) 1987); C: Sieveking group (after to Colombi Madeleine La belonging Cb3) horses 2003); Tosello of (after Examples 5. Figure ` r atrPilt&MnEte 00;L8 iei atrTslo20) 3)La M35) 2003); Tosello (after Limeuil Li8) 2010); Man-Estier & Paillet (after ere lvaRvr ere Sauvet Georges & Rivero Olivia 73 C niut ulctosLtd. Publications Antiquity n18) i)Limeuil Li5) 1986); on ´ n18) l clsin scales All 1986). on ´

Research Defining Magdalenian cultural groups in Franco-Cantabria by the formal analysis of portable artworks Discussion Having established the nature and content of these formal groups through CFA and AHC, our next objective was to analyse the chronological and regional composition of these groups, looking for differences that could be linked to different artistic traditions. To make the study as objective as possible, we have used a χ 2 test in order to determine whether the composition of the formal groups identified by AHC obeys a normal distribution or shows significant trends. The result of this test is unambiguous: the distribution of the various spatio-temporal subsets in the three groups is, at a very high probability level, clearly non-random (Tables 4 & 5). In other words, the groups distinguished by their component features are predominantly linked to a specific period and/or region. To summarise the main points of interest, group A mainly comprises figures belonging to the Middle Magdalenian (84 per cent), with the Pyrenees making the major contribution. Conversely, more than 90 per cent of the horses in group C come from the Upper Magdalenian, and in very large measure from Aquitaine. Two sites (Limeuil and La Madeleine) are particularly responsible for the dominance of Aquitaine in group C, since 22 of the 35 occurrences of the ‘double mane’ (Mdl) come from these two sites, as do 28 of the 57 occurrences of prominent crest (Mcp). Group B does not seem to be linked specifically to a region or period, but that is not surprising since linear outlines without any detail are not stylistically remarkable. The only striking point is the abundance of examples falling within group B, which represents one third of the total. The reason lies probably in the simplicity of these forms, avoiding technical difficulties, and could be executed by a relatively large number of individuals. The other, more complex, forms could be achieved only by skilled artists. The Cantabrian region shows a fairly balanced participation in the three groups with no clear trend towards any one of them in particular, but the small number of examples prevents reliable interpretation (Tables 4 & 5). The data provided by the statistical analysis must be interpreted as a contribution to the social anthropology of Upper Palaeolithic hunters-gatherers. In recent years, evidence from lithic and bone industries and from the circulation of flint and shells has highlighted the importance of mobility and emphasised the long-distance exchange networks between the groups of hunter-gatherers living in south-western Europe between 14 500 and 11 500 BP (17 700–13 400 cal BP). Most investigations, including those dedicated to symbolic representations, have concluded that the principal regions of southern France and northern Spain were closely connected during the Middle and Upper Magdalenian (Fritz et al. 2007; Sauvet et al. 2008; Rivero & Alvarez-Fern´ andez´ 2009). The intensity and variability of the interregional links, however, remain difficult to determine. It is not easy to understand why some types of object (technical tools or symbolic artefacts) and some formal features in portable or parietal art are widespread throughout the whole domain, while others remain confined to narrow territories. How do we reconcile the wide spatial diffusion observed on the one hand with the independence and individuality of small regional entities observed on the other? The results of our statistical analysis confirm this double and contradictory trend. For instance, we can assume that the formal model for depicting horses with extensive hatching (group A) was created and developed in the Pyrenees during the Middle Magdalenian since

C Antiquity Publications Ltd.

74 ofga,Tya,Lgae,i srltvl rqeti h atbinrgo Hro de (Hornos region Cantabrian the in frequent similar relatively a Combarelles, is presents Les it and (Font-de-Gaume, (Isturitz, art Lagrave), Quercy Teyjat, manes parietal and Rouffignac, in double Aquitaine also of in exists instances numerous mane few most double to distribution: a The Vache). seem only La however, with d’Azil, Pyrenees, influence Mas The Le of Valle). zone El this Lumentxa, outside Pila, remain La Pendo, El (Abauntz, region os ed eitdin depicted heads horse uigteMdl adlna.Ti scnre yalrerneo vdne including evidence, of range ( Corch large shells a and by flint of confirmed circulation is the This Magdalenian. Middle the during Corch ihycaatrsi fAutiedrn h pe adlna.Temdlsrasto Fontal spreads (Sainte-Eulalie, model Aveyron The of Magdalenian. valley Upper the the and during Middle Quercy Aquitaine the of to characteristic that limited highly demonstrates largely which are time. Magdalenian, A that Upper at group the occurred changes in of socio-cultural rare features important very the become that and noteworthy Magdalenian is It 1996). stegopAegaig,ie yeenoii ihawat fdtisadhatching, and details (Buisson of Aquitaine wealth and a region with Cantabrian origin the Pyrenean towards a expansion widespread i.e. a engravings, and A group the as is (Arias that noted already influence have Pyrenean authors a several seven confirm as but Garma sites, La two localised, these from more at two is clear model and Magdalenian particularly this Caldas Middle of Las the penetration from The during instances adopted. region and Cantabrian diffused the widely into was model in indicates Pyrenean represented which however, the well Courbet), also that Le is Montastruc, Aquitaine Madeleine, 3). La (Table (Laugerie-Basse, region A this group in dominant very numerically is it nteohrhn,hre rmgopC(ih‘obemn’adpoietcet are crest) prominent and mane’ ‘double (with C group from horses hand, other the On n20–06.Teecag ewrswr rbbya hi aiu extent on maximum their at probably were networks exchange The 2005–2006). on sfrfo admwt infiac rae hn9.% h otiuino each of contribution The 99.9%. than groups greater the three significance to the a subset to with subset random spatio-temporal from each far of is contribution the that shows freedom ecs)o (deficit); – – or (excess) symbolically: e e e e e neg neg neg neg Aq-UM neg Aq-MM Py-UM neg Py-MM Cneg CR-UM B CR-MM AnegnegGroup Aq-UM Aq-MM Py-UM 5. Table Py-MM C2723137 CR-UM B1332272322 CR-MM A94506287Group Middle AHC MM: three Magdalenian. Aquitaine, the Upper Aq: in UM: Pyrenees, subset Magdalenian, spatio-temporal Py: each region, to Cantabrian belonging CR: objects groups. of Number 4. Table ´ ´ tal. et χ 09.I sas hw ytegorpia itiuino etrssc as such features of distribution geographical the by shown also is It 2009). 2 etapidt h aai al .Tevalue The 4. Table in data the to applied test χ < 2 % elgbe(e) 2–10%: (neg); negligible 2%: au a vlae npretgso h lblvleadexpressed and value global the of percentages in evaluated was value otusd contours > 20%: lvaRvr ere Sauvet Georges & Rivero Olivia + Alvarez-Fern ´ ´ ecoup +++ ´ es (excess)or–––(deficit). oewihdslyeatytesm behaviour same the exactly display which mode ++ –ng– neg –– 75 ne 02 aob 05 iont2007; Simonnet 2005; Lacombe 2002; andez ´ + ecs)o dfii) 10–20%: (deficit); – or (excess) e e – – neg neg χ s n ece h Cantabrian the reaches and es) ` 2 = 0. o 0dgesof degrees 10 for 100.9 C niut ulctosLtd. Publications Antiquity ++ +++ tal. et 1999; tal et .

Research Defining Magdalenian cultural groups in Franco-Cantabria by the formal analysis of portable artworks la Pena,˜ Las Monedas, El Buxu, Candamo, Tito Bustillo) but very scarce in the Pyrenees (Niaux, Le Portel). The rarity of the cultural exchanges between Aquitaine and the Pyrenees during the Upper Magdalenian is in keeping with other observations. For instance, two typical Upper Magdalenian motifs such as horses with hypertrophied heads and schematic feminine figurines are abundant in Aquitaine and very rare in the Pyrenees (Cartailhac & Breuil 1907: 22; Bourrillon 2009). The links between the Pyrenees and Aquitaine are very slight at the end of the Magdalenian, whereas the links with the Cantabrian area seem relatively persistent, as shown by the motif of caprids in frontal view, which is mainly found in the Cantabrian region (57 per cent), is also abundant in the Pyrenees (33 per cent), but is rare in Aquitaine (10 per cent) (Rivero et al. in press). Together, these observations suggest that during the Upper Magdalenian the networks between the three main regions underwent a deep reorganisation. The relationships between the Pyrenees and Aquitaine appear much looser than in the previous period, because Aquitaine seems now to be engaged in new economic and cultural relationships with northern and eastern groups. The extension of schematic feminine figurines to Germany (Gonnersdorf)¨ is just one example of these new networks (Bosinski 2011). The major changes between the Middle and Upper Magdalenian that we perceive through our formal analysis of horse representations are also found in many other domains of portable and parietal art, and are confirmed by concomitant changes in lithic and bone industries. These relatively rapid changes seem to be correlated with an important redistribution of the exchange networks. The causes of this social and cultural reorganisation are not known. We can hypothesise, however, that the climatic improvement accompanying the Bolling¨ interstadial played a role in the circulation of big game and the expansion of new settlement sites, maybe in relation to a significant demographic increase. Eastward and northward movements of Aquitainian groups were probably linked to these phenomena, as well as an expansion of Cantabrian groups toward the interior of the Iberian peninsula via the Ebro Valley, where numerous Upper Magdalenian sites are known.

Conclusion Results presented in this work are based on the formal analysis of a representative corpus. Statistical tools such as correspondence factor analysis and ascending hierarchical clustering show that Magdalenian horse representations may be split into three formal groups according to specific features and that these groups present geographic and chronological distributions that are far from random. It is highly probable that the observed pattern reflects real tendencies and is worth discussion from an archaeological and anthropological point of view. The fact that a large number of horse figures using abundant hatching and anatomical details come from the Pyrenean Middle Magdalenian, and that a large number of much simpler figures, with linear tracing, ‘double mane’ and scarcity of detail, come from the Aquitainian Upper Magdalenian, may be seen as an indication of their territory of origin and development. The directions of diffusion and expansion are probably representative of the driving forces acting in society during the Magdalenian in south-western Europe. C Antiquity Publications Ltd.

76 04 Sauvet 2004; fpr yeensyei nobel ennfladw a ups httemxn of traditions. mixing different the with that groups suppose between relationships may the we symbolise and to meaningful aimed undoubtedly styles is ﬁgures style among panel Pyrenean study same pure the present these Together, in of the Pyrenees. presence as Their the style. that in in treatment Aquitainian exceptional archetypically a but are mane’, ﬁgures Aquitaine, ‘double in a frequent with particularly horse is a shows also is there bison, the to Aquitainian the intergroup (Ari Niaux in to of label perfectly cave a the conforming in as bison found acts a was it morphotype of that presence should is the style usefulness that instance, its accept For identity, we probable relationships. group If seems of transfer. marker the it during a and forgotten considered think, not be was we origin than without of widely point considered, their more be that much should instance, circulated region For probably another have rules. in features because the found context, within region fall given own are which a making their trends of those main in feature than typical the analysed instructive a outside more carefully be falling to be may Examples important and should exceptions probability. statistical was and of are it results interesting terms these but particularly in course, intuitively, treated Of less be features. or distinctive should stylistic main more Palaeolithic broad the The Upper specialists objectively data. Aquitainian by isolate raw and recognised of Magdalenian set already Middle large were Pyrenean a the from knowledge of extract characteristics to way objective most fe .Tsloi lte 1995. Clottes in Tosello G. after iue6 io n os rsnigtpclAquitainian (Ari Cave typical Niaux presenting in horse found and features Bison 6. Figure ftehreadbsno qiana tl ntePrna aeo iu icse above discussed Niaux of cave Pyrenean case the the in in style instance, Aquitainian For be of context. would bison particular paper, and the current horse account the attributions the into of chronological of taking scope and case, the each undated beyond in discussion, are necessary careful sites A most uncertain. sites quite because Colombi at are complicated found La also more Caldas, are (Las even examples few Magdalenian is a Middle but the period, to this within attributed falling occurrences the of cent oaas utvraeaayi sn ol uha F n H spoal the probably is AHC and CFA as such tools using analysis multivariate Nowadays, apriori tal. et supin,a let utrlecagsbtengop.Graphical groups. between exchanges cultural to clue a as assumptions, npes Fgr ) ti neetn ont ht nimdaeproximity immediate in that, note to interesting is It 6). (Figure press) in ` g,Prne) Drawing Pyrenees). ege, lvaRvr ere Sauvet Georges & Rivero Olivia 77 g)aditrrtdi hswy(Fortea way this in interpreted and ege) ` ae hw t aiu expansion per maximum 75 with Magdalenian Upper its the during shows ‘double it mane’ the instance, For only again trends. represent statistical they but that stressed be Magdalenian, should during Upper others the with and Magdalenian marked, the Middle during also dominant clearly features are some analysis formal recognise. to unable currently are we that distinctiveness regional markers carry in- stylistic probably global we as What generally conclusions. terpret similar to probably lead would withers, the on line double h idlg,the on legs, stripes as hind such the study, this in examined h hoooia rnssonb the by shown trends chronological The ayohrsyitcfaue not features stylistic other Many r) h aeo aeart cave of case The ere). ` C M niut ulctosLtd. Publications Antiquity sae el rthe or belly -shaped tal. et

Research Deﬁning Magdalenian cultural groups in Franco-Cantabria by the formal analysis of portable artworks

(Figure 6), the bison has a direct 14C date of 13 850−+150 BP (GifA-92501) that sets it unambiguously in the Middle Magdalenian. Thus, the ‘double mane’ of the adjacent horse should probably be assigned also to the Middle Magdalenian, despite the fact that double manes are much more frequent in the Upper Magdalenian. To go further into these questions of style, local origin and long-range diffusion of motifs and the related problems of acculturation, appropriation and reinterpretation, the next step of this research will require a careful examination of the technical characteristics of a vast series of objects from different areas. It is in the skill and originality of the artists in the manufacture of these objects that the answers to these difﬁcult anthropological questions are to be found (Rivero 2011). Acknowledgements This work was partly done in the framework of the research project ANR-10-CREA-001 (The arts of prehistory and the cultural dynamics of societies before writing).

References BUISSON,D.,C.FRITZ,D.KANDEL,G.PINÇON,G. SAUVET &G.TOSELLO. 1996. Les contours ´ ALVAREZ-FERNANDEZ´ , E. 2002. Perforated Homalopoma decoup´ es´ de tetesˆ de chevaux et leur contribution a` sanguineum from Tito Bustillo (Asturias): mobility la connaissance du Magdalenien´ moyen. Antiquités of Magdalenian groups in northern Spain. Antiquity Nationales 28: 99–128. 76: 641–46. CARTAILHAC,E.&H.BREUIL. 1907. Les œuvres d’art APELLANIZ,J.M.&F.CALVO. 1999. La forma del arte de la collection de Vibraye au Museum National. paleol´ıtico y la estad´ıstica. Analisis´ de la forma del arte L’Anthropologie 18: 1–36. paleol´ıtico y su tratamiento estad´ıstico (Cuadernos de Arqueolog´ıa 17). Bilbao: Universidad de Deusto. CHENORKIAN, R. 1996. Pratique archéologique statistique et graphique. Paris: Errance et Adam. ARIAS,P.,C.GONZALEZ´ SAINZ,A.MOURE &R. CLOTTES, J. 1995. Les cavernes de Niaux. Paris: Seuil. ONTAN˜ ON´ PEREDO. 1999. La Garma: un descenso al pasado. Catalogo´ de la exposicion´ . Santander: CLOTTES,J.&H.DELPORTE (ed.). 2003. La Grotte de Gobierno de Cantabria & Universidad de La Vache (Ariège). Fouilles Romain Robert.Paris: Cantabria. Reunion´ des Musees´ Nationaux. BARANDIARAN´ , I. 1973. El arte mueble del Paleol´ıtico CLOTTES, J., A. ALTEIRAC &C.SERVELLE. 1981. Cantabrico´ . Zaragoza: Monograf´ıas Arqueologicas.´ Oeuvres d’art mobilier magdaleniennes´ des anciennes collections du Mas-d’Azil. Préhistoire BARANDIARAN´ ,I.&G.LAPLACE. 2000. Temas, espacio Ariégeoise XXXVI: 37–70. decorativo y composicion:´ dos compresores magdalenienses de Poeymau(Pyr¨ en´ ees´ Atlantiques). CONKEY,M.&C.HASTORF (ed.). 1990. The uses of Saldvie I: 9–36. style in archaeology. Cambridge: Cambridge University Press. BARRIERE` , C. 1990. L’Ar t pariétal du Ker de Massat (Memoire´ de l’Institut de l’Art Prehistorique´ 5). CORCHON´ , M.S. 1986. El arte mueble paleol´ıtico Toulouse: UniversitedeToulouseleMirail.´ cantabrico.´ Contexto y analisis´ interno (Centro de Investigacion´ y Museo de Altamira 16). Madrid: BENZECRI´ ,J.-P.&F.BENZECRI´ . 1984. Pratique de l’analyse des données. Analyse des correspondances et Ministerio de Cultura. classification.Paris:Dunod. – 2005–2006. Los contornos recortados de la cueva de Las Caldas (Asturias, Espana),˜ en el contexto del BOSINSKI, G. 2011. Femmes sans tête. Une iconeˆ culturelle dans l’Europe de la fin de l’époque glaciaire.Paris: Magdaleniense medio cantabro-pirenaico.´ Munibe Errance. 57(3): 113–34. CORCHON´ , M.S., A. TARRINO˜ &J.MARTÍNEZ. 2009. BOURRILLON, R. 2009. Les representations´ feminines´ Mobilite, territoires et relations culturelles au debut dans l’Europe du Paleolithique´ superieur:´ une ´ ´ vision diachronique. Préhistoire, Art et Sociétés du Magdalenien´ moyen cantabrique: nouvelles LXIV: 105–15. perspectives, in F. Djindjian, J. Kozłowski & N. Bicho (ed.) Le concept de territoires dans le BREUIL, H. 1952. Quatre cents siècles d’art pariétal. Paléolithique supérieur européen (Actes du XV congrès Montignac: Max Fourny de l’UISPP, Lisbonne, 2006) (British Archaeological Reports international series 1938): 217–30. Oxford: Archaeopress.

C Antiquity Publications Ltd.

78 L F F J M P J P P L L AUZE AMBU ALES AILLET RITZ ORTEA ETROGNANI EROI ADIER ACOMBE AURY Pr ertie,d Territoires, observations. Pr 1927–1929). Jauze-Mandement (Fouilles pal tlsiu ucea td io aslatpari l’art dans bison du et cheval du stylistique pyr Remarques ’vyo,d Bruniquel de l’Aveyron, Li ’IP,Li l’UISPP, (ed.) Otte M. in ulCn dcoe el nvria de Cantabria. Universidad la de Ediciones PubliCan, Prehist Investigaciones de Internacional Instituto Tarascon-sur-Ari Pirineos los prehist en culturales fronteras y naturales Gonz ai:Mazenod. Paris: Soci olcinmjued ’r oiirpal mobilier l’art de majeure collection Pr rvrsd aMrh.III: Marche. La de gravures ai:Dunod. Paris: donn des l’analyse pour automatique magdal Pr mati Sp pal fronti territoires, ethniques, .S G. Nuil-u-i,An.Nouvelle Ain). (Neuville-sur-Ain, magdal l’abri de ai:Ophrys. Paris: ainldssoci des national olue 2001) Toulouse, ,L.&M.T ,C.,G.T ,B.&G.S -G g:Universit ege: ` ´ ´ ´ ´ hsor,Ate Soci et Art ehistoire, hsor Ari ehistoire hsor uSud-Ouest du ehistoire hsor.Tre thme usd(126 sud du hommes et Terres ehistoire. ´ el .&M-.L M.-O. & M. , .11.Fouilles 1914. J. , oihqeant eolithique eolithique ´ ´ .C W A.-C. & E , en ´ ,P.&E.M ,J.,C.F ´ ´ et ooi td Pr de et eologie AUVET rspremi eres ` OURHAN ´ ePr .20.Triorsdapoiineeten d’approvisionnement Territoires 2005. S. , lzUqio&X erds(ed.) Terradas X. & alez-Urquijo ´ r”dn ’r magdal l’art dans ere” es nJ abr .Braa(ed.) Barbaza M. & Jaubert J. in eens, ` ´ rcs(ca el reuni la de (Actas oricos ´ rcsd atbi ) 6–1 Santander: 165–81. 2): Cantabria de oricas ´ ne el rted B de Grotte la de enien ´ .2009. S. , ´ hsoiu Franc ehistorique ` g,2003) ege, &G.T OSELLO rpsd ulusensembles quelques de propos a ` ´ paeet,mobilit eplacements, RITZ .1965. A. , AUVET SI DE ASSIN al’ ultnd aSoci la de Bulletin ` ´ egeoise rsltiusa Tardiglaciaire. au lithiques eres ` AN ´ et 2–3 ai:CTHS. Paris: 329–53. : edeLi ´ ` g,2004) ege, ne eL Colombi La de enien ´ pev usyee ucarbone-14, du et style du epreuve ´ ,M.G e-magdal ´ aspiritualit La ´ shsoiuse scientifiques, et historiques es -E ELT ´ ehistoire EBEAUX rpsd traitement du propos A OSELLO 91 r mobilier Art 1991. . ` &G.S STIER LI 19–57. XLVI: ´ et Laugerie-Basse. a E ` ´ EAL16:163–75. 106): (ERAUL cags a“oinde “notion la echanges: ´ 91 avall La 1991. . ege. ` S ´ es ARCIA Pr AINT aFontal 8 35–104. 18: ` ¸aise XV 95–104. LXIV: 00 e ursd’art œuvres Les 2010. . ´ ne franco-ib enien hsor elatoccidental l’art de ehistoire ´ 1978. . 04 ’r pari L’art 2004. . XI 51–76. XXXI: (Monograf ´ equid AUVET ne,i .Czl,J. Cazals, N. in enien, ´ -P nclbaaen celebrada on ´ ,J.L.S ´ Atsd olqede colloque du (Actes e 11:50–56. 11(1): dihc(Ari edeilhac ´ EREUSE ´ et ´ e, ´ se bovid et es s nouvelles es: ` ´ em 07 Groupes 2007. . ´ cagsdrn la durant echanges Classification td d’une etude ´ ´ e.2 logiciels 2: ees. ANCHIDRI ´ rdoaede eridionale lvaRvr ere Sauvet Georges & Rivero Olivia ede ee ´ ere ` ´ a del ıas 1981. . Fronteras eolithique. ´ ultnd la de Bulletin eme ´ erique. ´ ´ es ege). etal ` ´ .Gap& congr Les AN ´ etal ´ . , ` es . 79 R P S S R noci La 2011. – R a art: cave Palaeolithic in style Determining 2007. – S IEVEKING AUVET AUVET IGEAUD IVERO IVERO IVERO ´ 02 Li 2012, Palaeolithic Eurasian the during displacements UISPP, Colloque cantabrique (Actes mobilier l’art de techniques analyses abz (ed.) Barbaza aeae:L Soci La Lavelanet: u (126 sud Pr la durant echanges tlsiuse omsduaedn latd la de ant “l’art silhouette” dans d’usage formes et stylistiques uatl Pr la durant P. & Fourment (ed.) N. Depaepe Jaubert, J. in l’Espagne, de Magdal (Thorign Mayenne-Sciences grotte la de chevaux rts Museum British Prehistoria contribuci c Medio Magdaleniense Salamanca. of University dissertation, PhD visi una Regi Pirineos: la de Magdaleniense el en pr soci S. G Bourillon, D. R. Petrognani, Fritz, C. el in en Medio, mobiliar Magdaleniense arte y adorno de Europa: elementos en intergrupales contactos de Evidencias 2010). pr Pal eain nr e soci les entre relations otrnse post-doctorants et doctorants rs.Futain des Fluctuations press. Cr 1 409–22. images. 81: horse from derived method new 2001) Toulouse, scientifiques, lper el paleol humanos 33–60. ´ hsoiu eFac Breu e Eyzies, Les – (Bordeaux France de ehistorique hsoiu Ari ehistorique ´ nc elsitrabo nr o grupos los entre intercambios los de onica ´ oihqesup eolithique ´ ,G.,C.F ,G.,J.F ´ et .21.L oiia elsguo humanos grupos los de movilidad La 2010. O. , ,O.,D.G ,O.&E. .20.Apoo e repr des propos A 2005. R. , ´ spr es ´ oo2 0—200a 000 000—12 20 ıodo .1987. A. , ai:Soci Paris: ne r enien ´ eme g,Belgium). ege, ` ´ -nCane aen) Singularit Mayenne). e-en-Charnie, ´ hsoius ecnrsinternationales Rencontres ehistoriques. ´ ndlan del on ´ 8 275–95. 68: nd pedzj ne rembla del mobiliar arte el en aprendizaje de on ´ hsor,Atsd XIeCongr XXVIIe du Actes ehistoire, ´ ORTEA congr RITZ A ertie,d Territoires, ARATE ´ ´ cn.Lapr pr L’apport ecent. e-magdal ´ ´ rniin,rpue tcontinuit et ruptures Transitions, LVAREZ odn rts Museum. British London: . ´ tcs acontribuci La ıticos. aaou fPlelti r nthe in art Palaeolithic of catalogue A natae e re Unpublished arte. del traves a on ege-Pyr ` rere rnee asl Nord le dans et France en erieur ´ ´ rt .Sue (ed.) Sauvet G. & arate ,J.F ´ et ` sntoa e soci des national es ´ et ´ ,C.F epr ´ lssmicrosc alisis ´ epr ´ C ´ hsor.Tre thme du hommes et Terres ehistoire. ..F M.A. & et ´ -F ORTEA niut ulctosLtd. Publications Antiquity cagssmoiusau symboliques echanges ´ hsoiu franc ehistorique ´ seurop es hsoiu Ari ehistorique ´ ne,i .Juet&M. & Jaubert J. in enien, ´ ´ nar-ieac:la antabro-pirenaico: ´ RITZ ERN oe fcnatand contact of Modes en ´ ´ paeet,mobilit eplacements, e XV:149–56. LXIV): ees ´ Blei el Soci la de (Bulletin 3–3 ai:CTHS. Paris: 235–63. : ANDEZ ´ o BP. nos &G.T &G.T ˜ ANO nCant on ´ enspnatle pendant eennes ´ lmniedes eliminaire ´ opico. ´ snain de esentations ´ npes Les press. In . 2009. . ´ et ndlat para arte del on ´ ´ shsoiuset historiques es OSELLO OSELLO Zephyrus ege-Pyr ` biaylos y abrica ´ ¸aise. rbjsde Trabajos Antiquity ’rtdes t L’Ar ` es .In 2008. . ,May es ´ ´ e, en ´ ´ e et ´ 61: e ´ ees. ´

Research Deﬁning Magdalenian cultural groups in Franco-Cantabria by the formal analysis of portable artworks

SIMONNET, R. 2007. Du silex des Pyren´ ees´ centrales aux TOSELLO, G. 2003. Pierres gravées du Périgord Magdaleniens´ a` Labastide, in N. Cazals, J. magdalénien: art, symboles, territoires (Gallia Gonzalez-Urquijo´ & X. Terradas (ed.) Frontières Préhistoire supplement 36). Paris: CNRS. naturelles et frontières culturelles dans les Pyrénées UCKO, P.J. 1987. Debuts´ illusoires dans l’etude´ de la préhistoriques (Actas de la reunion´ celebrada en tradition artistique. Préhistoire Ariégeoise 42: 15–81. Tarascon-sur-Ariège, 2004) (Monograf´ıas del UTRILLA,P.&C.MAZO. 1996. Arte mueble sobre Instituto Internacional de Investigaciones soporte l´ıtico de la cueva de Abauntz. Complutum Prehistoricas´ de Cantabria 2): 93–100. Santander: Extra 6(1): 41–62. PubliCan, Ediciones de la Universidad de Cantabria. VILLAVERDE BONILLA, V. 1994. Arte paleol´ıtico de la cova del Parpallo,´ vol. I. Valencia: Diputacio´ de Valencia. SIMONNET,G.,L.SIMONNET &R.SIMONNET. 1989. Art mobilier et art parietal´ a` Labastide, in J. Clottes WOBST, H.M. 1977. Stylistic behavior and information (ed.) L’Art des objets au Paléolithique. I: l’art mobilier exchange, in E.H. Cleland (ed.) For the director: et son context: 173–87. Paris: Ministere` de la research essays in honor of James B. Griffin Culture. (Anthropological Papers 61): 317–42. Ann Arbor: University of Michigan Museum of Anthropology.

Received: 17 September 2012; Accepted: 26 February 2013; Revised: 18 March 2013

C Antiquity Publications Ltd.