Journal of Archaeological Science 38 (2011) 1080e1089 Contents lists available at ScienceDirect Journal of Archaeological Science journal homepage: http://www.elsevier.com/locate/jas A proxy of potters’ throwing skill: ceramic vessels considered in terms of mechanical stress Enora Gandon a, *, Rémy Casanova a, Patrick Sainton a, Thelma Coyle a, Valentine Roux b, Blandine Bril c, Reinoud J. Bootsma a a b c Institut des Sciences du Mouvement E.J. Marey (UMR 6233), Université de la Méditerranée, 163 avenue de LuminyeCP 910, 13009 Marseille, France CNRS, Maison de l’Archéologie et de l’Ethnologie, Préhistoire et Technologie (UMR 7055), 21 Allée de l’Université, 92023 Nanterre cedex, France Ecole des Hautes Etudes en Sciences Sociales, 54 bd Raspail, 75006 Paris, France a r t i c l e i n f o a b s t r a c t Article history: Received 6 April 2010 Received in revised form 7 December 2010 Accepted 10 December 2010 This study is aimed at evaluating potters’ skills according to the mechanical characteristics of the vessels they produced. It focuses on wheel thrown vessels. In a first stage, considering that the difficulty of throwing ceramic vessels is to a significant extent determined by the risk of collapse of the thrown structure, we applied the method of finite element modelling to derive an index of mechanical stresses operating within a vessel. Validated via compression tests, the so-called Von Mises stress index was employed as a global index of mechanical difficulty. Because this index allows comparisons between vessels of different form, mass, and clay properties, it provides a more powerful tool than existing techno-morphological taxonomies. In a second stage, in order to relate the Von Mises stress index to throwing difficulty, we analysed the geometrical and mechanical characteristics of vessels thrown by eleven expert potters invited to reproduce four different model forms with two different masses of clay. The results demonstrated that reproductions revealed subtle but systematic deviations from the model forms that allowed a decrease in the mechanical difficulty. More difficult forms showed larger degrees of mechanical optimisation. These results, in combination with a new analysis of data from Roux’s (1990) study with potters of different skill levels, indicate that skill resides, at least to a certain extent, in the capacity to marshal the operative mechanical constraints. In other words, the latter, measured by the Von Mises index, provides a useful signature of a potter’s skills. ! 2010 Elsevier Ltd. All rights reserved. Keywords: Pottery Ceramic Skill Constraint Mechanical stress Expertise Craft Technique 1. Introduction Classifying ceramics by form is a classic exercise in archaeology, mainly used to characterise chrono-cultural periods or to identify functions (e.g., Rice, 1987; Orton et al., 1993). Morphological typologies can, however, also be used for other purposes, such as evaluating the skills that potters acquired (e.g., Balfet, 1984; Berg, 2007). In the present framework, a skill is defined as “a form of behaviour acquired through learning” (Bril, 2002: 115), and its development regarded as determined by the social context in which it develops. Evaluating the skills of the potters who produced archaeological assemblages paves the way to rich fields of * Corresponding author. Tel.: þ33 0491 17 22 55; fax: þ33 0491 17 22 52. E-mail addresses: [email protected] (E. Gandon), remy.casanova@ univmed.fr (R. Casanova), [email protected] (P. Sainton), thelma.coyle@ univmed.fr (T. Coyle), [email protected] (V. Roux), blandine.bril@ ehess.fr (B. Bril), [email protected] (R.J. Bootsma). 0305-4403/$ e see front matter ! 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.jas.2010.12.003 interpretation, including the ways production and transmission were organised (e.g., Roux and Corbetta, 1990; Vitelli, 1993; Crown, 2001, 2007; Budden, 2008) and the mechanisms underlying changes in ceramics (e.g., Gelbert, 1997, 2003; Gosselain, 2000). Analysing ceramic forms from a skills point of view entails describing the vessels not as abstract and purely geometric shapes, but, in line with Gibson’s (1979) arguments, concretely as objects. In other words, the description must be linked to the shaping process and to the physical properties of the object the potter produced. According to Ingold (2001: 22), “it is the activity itself e of regular, controlled movement e that generates the form, not the design that precedes it. Making, in short, arises within the process of use, rather than use disclosing what is, ideally if not materially, readymade”. Evaluating potters’ skills from ceramic vessels presupposes e mutatis mutandis e recognising that a vessel is subject to physical constraints and, in line with Bernstein’s (1967) suggestion, considering that the skill level is reflected in the extent to which these physical constraints have been marshalled. This hypothesis E. Gandon et al. / Journal of Archaeological Science 38 (2011) 1080e1089 was the starting point for a preliminary study by Roux (1990), who proposed a classification of wheel thrown ceramic forms according to how difficult they are to produce. Within the framework of this techno-morphological taxonomy, forms are described in terms of absolute and relative dimensions, and classed successively according to the size, the morphological category (open or closed), and the relevant proportions (Fig. 1). Field studies with Indian and French potters, examining their appreciation of the difficulty encountered in producing specific vessels, have validated the taxonomy in two different cultures (Roux, 1990). Describing the vessel’s form by means of its absolute and relative dimensions, the taxonomy pragmatically incorporates the mechanical constraints underlying the risk of collapse. However, it does not provide a unified global measure of the latter, not only because geometry is reduced to the major proportions only, but also because mass is considered separately from geometry. In this paper we explore the part played by mechanical constraints in the difficulties encountered by expert potters in throwing different types of vessels on a wheel. The first part of this paper presents a mechanical analysis of ceramic vessels, using the well-established method of finite element modelling. The aim of this analysis is to offer an index that expresses the level of difficulty of throwing a vessel. This index should summarise the combined influences of relevant characteristics of the vessel, such as form, mass, and clay properties. The analysis culminates in a global index of mechanical stresses known as the Von Mises stress index. We then present an experiment aimed at testing the relevance of this index for assessing the difficulties involved in the throwing of vessels of different shapes and sizes. Rather than asking potters to classify vessels according to their difficulty, in the present framework we evaluated the vessels they actually threw in terms of 1081 reproducibility and deviation from prescribed model forms. For simple forms we expected potters to reliably and consistently reproduce the models proposed. For more difficult forms we expected potters to adapt the vessels’ form so as to alleviate the operative mechanical stress, perhaps accompanied by a larger variability over repeated productions. Results are discussed in terms of the operative relations between geometry and mechanical stress. We conclude that the Von Mises stress index provides a useful global index for the classification of ceramic vessels in terms of difficulty, paving the road towards an assessment of potters’ skills. 2. Materials and methods 2.1. Mechanical modelling The mathematical method of finite element modelling is commonly used in mechanics to predict collapsing risk in material structures. The method integrates three elements: the geometry of the object, the material characteristics of the object’s matter, and the (mechanical) behaviour law relating stresses operating in and on the object to the object’s deformations. It is widely used in civil and aeronautical engineering, for instance to evaluate whether material structures, such as dams, bridges and boats can resist irreversible deformations under varying loading conditions, and to identify the weak points where bending or twisting may occur. In the present framework, we examined vessels in their immediate post-throwing state, when the clay was still wet. In the course of throwing, mechanical stresses increase as the vessel evolves towards its final form, with the walls becoming thinner and the clay wetter. Thus, for the purposes of interpreting skill from existing Fig. 1. The techno-morphological taxonomy proposed by Roux (1990) is based on the principle that vessels with more acute angles a, b, and g (see top panels) are more difficult to produce. These angles are globally captured by proportions (see bottom panels) relating base diameter (B), maximal diameter (MD), aperture diameter (A), and height (H). For an open vessel (left panel), the classification ratios used are A/H and A/B. For a closed vessel (right panel), the difficulty of execution is captured by MD/H, MD/B, and MD/A. The larger these ratios, the greater the difficulty, due to the danger of the vessel collapsing. In addition to these ratios, the height of the maximum diameter (HMD) is considered as a complementary principle of classification for closed vessels: the lower the maximum diameter height, the greater the difficulty. For each vessel, wall thickness was determined at 4 levels: middle of the base (T.1), bottom (T.2), middle (T.3) and top (T.4) of wall (see bottom panels). 1082 E. Gandon et al. / Journal of Archaeological Science 38 (2011) 1080e1089 vessels, this immediate post-throwing state is considered the most pertinent. As at this stage neither the potter nor the (no longer rotating) wheel exert time-varying forces on the structure, a static model may be adopted. Because thrown vessels are theoretically symmetrical, we used a symmetrical model allowing mechanical stresses in the entire structure to be computed from its profile. In a first step, we examined the vessel under the load of its own weight. Because under the load of its own weight the noncollapsing vessel does not deform plastically (i.e., irreversibly), the wet clay’s behaviour was modelled as an elastic material, with classic parameters obtained from the literature for Poisson’s ratio e the ratio of decrease in the thickness (lateral contraction) to its increase in length (longitudinal extension) when stretched e and Young’s modulus e the modulus of elasticity characterising stiffness. Poisson’s ratio (n) was set to a value of 0.3 (Yin, 2006) and Young’s modulus (E) to a value of 3.0 MPa (Gagou et al., 2008). The density (r) of the sandstone clay used was experimentally determined by weighing controlled volumes as 1.95 g/cm3. The digitised face of the vessel was imported into the modelling software (CASTEM 2000"), together with the thickness of the wall, measured at four different levels (see Fig. 1). From these latter data, the vessel’s geometry was reconstructed as a surface in 3D and summarised by a distribution of quadrilateral elements. On each element the software computes local stresses with three components: the radial traction-compression stress srr, the axial traction-compression stress szz, and the shear stress srz. To summarise the three components, we used a global norm traditionally used in clay studies (Lemaitre and Chaboche, 2004), the Von Mises norm, computed as: sVonMises sffiffiffiffiffi" ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi#ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 ¼ $ ðsrr $ szz Þ2 þðsrr Þ2 þðszz Þ2 þ 3$srz2 2 With this method, we obtained a Von Mises stress distribution map of the vessel under the load of its own weight (Fig. 2). Because the location and magnitude of the maximum Von Mises stress value determines where and when the structure risks collapse, this value provides a pertinent index of mechanical stresses in the vessel. The error associated with the derivation of the Von Mises values was less than 5%. In future work, this error could be further decreased by replacing the estimation of wall thickness along the surface through interpolation of discrete measurements with complete data from digitised walls. The model was validated by comparing observed deformation under the influence of an external force to vessel deformation predicted by the behaviour law 3 ¼ Ss, where 3 is the local deformation, S the operator of flexibility which integrates the material’s parameters E and n (Lemaitre and Chaboche, 2004), and s the internal stresses. An expert potter threw six vessels of each of three different forms (cylinder, bowl and open-necked vase1). Using a classic mechanical testing protocol, three specimens of each form were compressed with a controlled force applied by means of a penetrometer. To this end, a perforated plastic plate was centred on top of the vessel, covering its aperture. The penetrometer was positioned in the middle of the plate so as to apply an evenlydistributed pressure. The tests were filmed to capture each vessel’s profile before and after deformation and to detect the exact moment of onset of irreversible deformation. Pressures exerted at this moment were 159.2 & 15.8 kPa for the cylinder, 117.5 & 8.7 kPa for the bowl, and 75.2 & 4.5 kPa for the open-necked vase. The 1 The three forms used for the mechanical tests during the validation stage were drawn from the personal repertoire of the expert. They do not correspond to the forms tested in the experiment, which were chosen to be impersonal. Fig. 2. Von Mises stress distribution map for an open-necked vase under the load of its own weight. The legend presents a correspondence between colour coding and Von Mises stress expressed in Pa. The vessel is modelled as a finite element structure, with individual elements forming a chain of volumes. Constituent stresses are computed in each element taking into account its place in the structure. Under the load of its own weight, the collapse zone (zone with the highest Von Mises stress values) of this vessel is situated in its base. remaining three specimens of each form were vertically cut and photographed. From the photographs, the mean (between-specimen) wall thickness at four different locations (see Fig. 1) was calculated for each form and integrated into the model. For each form, the threshold pressures were then incorporated into the software as load acting on the top of the vessels and we calculated the deformation resulting from its application (see Fig. 3, second panel, for an example). For all three forms, the model reliably reproduced the experimentally observed zones of deformation (see Fig. 3, third panel, for an example). Finally, the modelling results revealed that the threshold stress level, defined as the local maximal Von Mises stress value in the vessel under the pressure exerted at the moment of deformation, was equivalent for the three different forms (18 kPa & 15%). Given that both the location of deformation zones and the threshold local stress levels rendered consistent results, the model was deemed suitable for analysing experimental vessels. Note that while compression via external forces was used to validate the model, in the remainder of this paper we focused on the mechanical stress (represented by the maximal Von Mises stress index) operating in the vessel under the load of its own weight (as in Fig. 2). 1083 E. Gandon et al. / Journal of Archaeological Science 38 (2011) 1080e1089 Fig. 3. Deformation of a vessel under the influence of an external pressure exerted on the top (compression test). The photographs show an experimental vessel before and after deformation occurring at the level of the neck. As can be seen (right panel) from the Von Mises stress distribution map of the (not yet deformed) vessel under the influence of the external pressure, the maximal stress values (indicative of the future collapse zone) are located at the level of the neck. The panel between the two photographs shows the model before (red) and after (blue) deformation, under the same conditions. As for the experimental vessel, the model deforms at the level of the neck. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 2.2. Experiment Selected on the basis of responses to a questionnaire evaluating their experience in wheel-throwing and use of sandstone clay, eleven expert right-handed French potters, nine men and two women, agreed to participate in the experiment. With a wheelthrowing experience of 15e40 years (mean & sd: 29 & 7.7 years), the participants’ age varied from 36 to 59 years (51 & 6.5 years). The experiment took place in a large room of a dedicated pottery centre in the Bourgogne area. All throwing was performed on the same electrical wheel (Brent) using sandstone clay. Participants were asked to reproduce four different model forms using two different quantities of clay, for a total of eight experimental conditions (Table 1). The forms (referred to as cylinder, bowl, sphere and conical vase, respectively) were presented on (2D) drawings without any indication of dimension. The quantity of clay provided for each trial corresponded to a mass of 0.75 kg or 2.25 kg. The choice of the model forms was based on Roux’s (1990) taxonomy, which suggests that for open forms the bowl is more difficult to produce than the cylinder, while for closed forms the conical vase is more difficult to produce than the sphere. In contrast to the other three forms, the conical vase proposed was not typical of classic ceramic forms. In the framework of our objectives, it was included as a particularly difficult form, due to the very small height of maximal diameter. As detailed in Table 1, the combination of model forms and masses gave rise to a range of (maximal) Von Mises stress indices varying between 2.6 and 18.8 kPa. Table 1 The eight experimental conditions were defined by combining four different forms (cylinder, bowl, sphere and vase) with two clay masses (0.75 and 2.25 kg), giving rise to maximal Von Mises stress indices varying from 2.6 to 18.8 kPa. Form Cylinder Bowl Sphere Vase 2D drawing For the purposes of the present experiment, the participants were instructed to accurately reproduce the proportions of the models, to throw vessels with the thinnest walls possible, and to refrain from embellishment operations once the vessel was thrown. Participants practiced the task the day before the experiment, producing at least one vessel under each of the eight experimental conditions. During the experiment proper, the order of the different conditions was randomised within each block of eight trials so as to avoid systematic learning effects. With an experimental session comprising five blocks e potters thus producing five specimens in each experimental condition e each participant produced a total of 40 vessels. 2.3. Methods 2.3.1. Data recording Experimental sessions were videotaped using a Panasonic NV-GS320 camcorder and the image of each vessel immediately after throwing was extracted from the films for further analysis. In order to determine the wall thickness, throwing vessels were transversely cut the following day using a fine cord and photographed using a Canon A720 camera. From the post-throwing images extracted from the video recordings we recovered for all vessels the absolute dimensions (in mm) of base (B), height (H) and aperture (A), complemented with maximal diameter (MD) and height of maximal diameter (HMD) for the closed forms. As for the mechanical tests, wall thickness was measured from the photographs of cut vessels at four levels (see Fig. 1). Values of thicknesses are presented in Table 2. Table 2 Between-participants average wall thicknesses of thrown vessels measured at 4 levels: middle of the base (T.1), bottom (T.2), middle (T.3) and top (T.4) of wall. Form Mass (kg) T.1 (mm) T.2 (mm) T.3 (mm) T.4 (mm) Cylinder 0.75 2.25 0.75 2.25 0.75 2.25 0.75 2.25 7.3 9.7 9.7 12.3 6.4 9.1 6.6 8.9 10.5 14.7 18.2 27.3 15.4 21.8 14.2 20.5 6.5 8.9 8.5 11.7 6.5 8.0 9.0 11.1 5.2 6.4 6.4 7.8 6.8 8.9 6.9 8.8 Bowl Sphere Mass of clay (kg) 0.75 2.25 0.75 2.25 0.75 2.25 0.75 2.25 Von Mises (kPa) 2.6 4.1 6.9 12.4 6.5 9.9 12.5 18.8 Vase 1084 E. Gandon et al. / Journal of Archaeological Science 38 (2011) 1080e1089 2.3.2. Data analysis The analysis was focused on three sets of dependent variables: (i) absolute dimensions (H, B, and A for open forms, complemented with MD, and HMD for closed forms; see Fig. 1), (ii) proportions (A/B and A/H for open forms; MD/B, MD/H, and MD/A for closed forms; see Fig. 1) and (iii) Von Mises indices obtained via modelling. Our goal was to compare model and thrown vessels in terms of each of these sets of dependent variables. In order to allow such a comparison on the absolute dimensions, the model vessels e defined as geometric forms (proportions) without intrinsic scale e were homeomorphously transformed to match the mean between-participant height of the vessels thrown in each condition (see Table 3). Combining these absolute model dimensions with local extrapolations of the measured wall thicknesses of the thrown vessels in each condition (see Table 2) allowed the derivation of the models’ maximal Von Mises stress indices reported in Table 1. Accuracy of reproduction was assessed via comparisons between model and thrown vessels according to the means of the absolute dimensions, except, of course, for the vessel height H, which served to scale the models’ absolute dimensions in the different experimental conditions. Using the between-trial variabilities over the five specimens thrown by each participant in each condition, reproducibility was assessed via the coefficients of variation (CV ¼ 100% * standard deviation/mean). From the measured absolute dimensions we then calculated the thrown vessels’ proportions, identified in Roux’s taxonomy. Deviations from the model forms were calculated as the percentage difference between the proportions of the thrown vessels and model forms. Differences in mechanical stress between the model and thrown vessels were characterised using the maximal Von Mises stress indices. The effects of Form (cylinder, bowl, sphere, and conical vase) and Mass (0.75 and 2.25 kg of clay) were evaluated using analyses of variance (ANOVA) with repeated measures on both factors. Significant (a ¼ .05) main effects and interactions were further explored using NewmaneKeuls post-hoc test procedures. vessels and average thrown vessels. As can be seen from Fig. 4, deviations from the model forms were largest for aperture in the open forms and for maximal diameter in the closed forms. The five reproductions of each experimental condition allowed quantification of the (intra-individual) between-trial variability in absolute dimensions, captured by the coefficient of variation (CV). The mean CVs of the absolute dimensions are reported in Table 4. Inspection of the CVs revealed that the geometrical deviations from the model forms, depicted in Fig. 4, were produced in a systematic fashion. Indeed, CVs of the absolute dimensions were overall small, between 2 and 6% for the open forms (cylinder and bowl) and between 2 and 11% for the closed forms (sphere and vase). Repeated-measures ANOVAs on the CVs averaged over different absolute dimensions demonstrated that the factor Mass did not significantly influence the intra-individual variability (F(1, 10) ¼ 1.23, ns), while the factor Form did (F(3, 10) ¼ 21.83, p < .001). Post-hoc analysis (p < .05) of the latter effect revealed that the CV of the vase (7.4%) was larger than the CV of the sphere (5.6%), which in turn was larger than the CVs of the bowl (4.6%) and the cylinder (3.8%). The absolute dimension with the least intra-individual variability was the base for the cylinder (3.3%), the aperture for the bowl (3.4%), and the maximal diameter for the sphere and vase (3.2% for each). Interestingly, these lowest-variability absolute dimensions also showed the largest deviation from the model forms (Fig. 4). Overall, the analysis of the absolute dimensions revealed that potters did not faithfully reproduce the model forms, deviating from them in a (form-specific) systematic manner. Although the more difficult forms e according to Roux’s (1990) taxonomy e revealed a larger variability in absolute dimensions, in general between-specimen variability was low. The level of standardisation of the experimental vessels as measured by the CVs (5.5% on average) was close to the 6% reported by Roux (2003). 3.2. Proportions 3. Results 3.1. Absolute dimensions Table 3 summarises the absolute dimensions of the model and thrown vessels. For the thrown vessels, both form and mass were found to influence the absolute dimensions of H, B, A, MD, and HMD (all main effects and interactions p < .001, except for the interaction Form ' Mass for MD, ns). As was to be expected on the basis of allometric scaling laws, the increase in size was not proportional to the threefold increase in the amount of material available (as can be observed in the percentage increase reported in Table 3). In order to visualise the results summarised in Table 3, Fig. 4 graphically presents, for each form and mass, the scaled model In order to better understand the observed systematic deviations from the model forms, we next analysed the proportions (i.e., relative dimensions) of the thrown vessels. To this end, we calculated the ratios A/H and A/B for the open forms (cylinder and bowl) and the ratios MD/H, MD/B, and MD/A for the closed forms (sphere and vase). Proportions of the model vessels and thrown vessels are presented in Table 5. According to the taxonomy proposed by Roux (1990), the larger these ratios, the more difficult the task. Fig. 5 presents the ratios of the thrown vessels as a function of the model ratios. We observe that, except for the cylinder, all the other forms produced revealed systematic negative deviations from the model ratios. Participants consistently produced bowls, spheres and vases of smaller Table 3 Absolute dimensions of model and thrown vessels. Only defined by their proportions, the model vessels were scaled to the average height of thrown vessels for comparison purposes. H: Height, B: Base, A: Aperture, MD: Maximal Diameter, HMD: Height of Maximal Diameter. The increase in absolute dimensions between the 0.75 kg vessels and the 2.25 kg thrown vessels are expressed as percentage increase (% inc). Model Vessels Thrown Vessels Form Mass (kg) H (mm) B (mm) A (mm) Cylinder 0.75 2.25 0.75 2.25 0.75 2.25 0.75 2.25 177.2 286.5 92.3 138.1 120.7 186.9 99.9 150.8 101.9 164.6 73.7 110.5 84.4 130.4 91.8 138.7 98.5 159.2 272.5 408.8 83.9 129.7 64.5 97.4 Bowl Sphere Vase MD (mm) 171.7 265.4 214.2 323.4 HMD (mm) H (mm) 57.0 88.1 30.3 45.8 177.2 286.5 92.3 138.1 120.7 186.9 99.9 150.8 % inc 61.7 49.6 54.9 51.0 B (mm) 101.3 142.5 86.5 127.8 97.0 142.4 113.3 169.5 % inc 40.7 47.8 46.8 49.7 A (mm) 104.8 143.9 214.5 330.3 86.3 116.7 68.8 89.5 % inc MD (mm) % inc HMD (mm) % inc 37.2 54.0 35.3 30.2 158.8 234.0 164.9 242.6 47.4 47.1 59.5 91.0 37.4 52.4 52.8 40.4 1085 E. Gandon et al. / Journal of Archaeological Science 38 (2011) 1080e1089 0.3 0.25 0.2 0.15 0.15 0.1 0.1 0.05 0.05 0 -0.1 -0.05 0 0.05 0.1 0 -0.25 -0.2 -0.15 0.2 0.2 0.15 0.15 0.1 0.1 0.05 0.05 0 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0 -0.2 -0.1 -0.05 0 0.05 0.1 -0.15 -0.1 -0.05 0 0.05 0.15 0.1 0.2 0.15 0.25 0.2 Fig. 4. Graphical representation, with scale in m, of model (grey) and average thrown vessels (black) for each of the four forms and two clay masses. proportions than the models presented, in line with the predictions drawn from Roux’s taxonomy. Average deviations of thrown vessels from model proportions are presented in Table 6. A repeated-measures ANOVA on the percentage deviation averaged over the different proportions revealed that Mass influenced the deviations from the model proportions to a certain extent (F(1, 10) ¼ 7.38, p < .05), with an average deviation of $18.5% for the 0.75 kg masses and $16.8% for the 2.25 kg masses. Form had a clear-cut effect (F(3, 30) ¼ 61.44, p < .001), while the interaction Mass ' Form was not significant (F (3, 30) ¼ 0.8, ns). Post-hoc analysis (p < .05) of the main effect of Form indicated that deviations from the model ratios were larger for the vase ($26.9%) and bowl ($25.2%) forms as compared to the sphere ($10.7%). The latter in turn demonstrated larger deviations from the model ratios than the cylinder (þ2.4%). As can be seen from Fig. 5 and Table 6, for the bowl, the A/B ratio decreased on average by 30.7% and the A/H one by 19.6%. These effects were largely due to the decrease in A and, to a lesser degree, the concomitant increase in B (see Fig. 4). Both closed forms also revealed a decrease in ratios (Fig. 5). Mainly due to the decrease in MD, MD/H and MD/A decreased respectively by 8.8% and 4.7% for the sphere and 22.4% and 21.2% for the vase. The concomitant Table 4 Mean coefficients of variation (CV) of the absolute dimensions of thrown vessels. H: Height, B: Base, A: Aperture, MD: Maximal Diameter, HMD: Height of Maximal Diameter. Form Mass (kg) H (%) B (%) A (%) MD (%) HMD (%) Cylinder 0.75 2.25 0.75 2.25 0.75 2.25 0.75 2.25 4.4 5.2 5.1 4.6 5.2 5.1 7.4 6.6 3.4 3.3 5.8 5.1 5.7 4.9 7.7 6.0 3.1 4.1 3.9 2.9 8.7 7.5 10.3 10.1 3.4 3.0 3.6 2.9 6.3 5.9 9.5 9.9 Bowl Sphere Vase increase in B gave rise to a decrease in MD/B of 18.5% and 37.0% for the sphere and vase, respectively. 3.3. Mechanical stresses As can be seen from Table 1, the mechanical stress e indicated by the maximal Von Mises index e of the model forms to be reproduced varied as a function of both form and mass. Fig. 6 presents the maximal Von Mises indices for the models and the thrown vessels. As can be seen, compared to the models presented, levels of mechanical stress were lower in all the thrown vessels, except for the 0.75 kg cylinder. For the three standard forms (cylinder, bowl, and sphere), the mechanical stress in the thrown vessels (with both 0.75 and 2.25 kg of clay) covaried linearly (R2 ¼ 0.986) with the mechanical difficulty indicated by the maximal Von Mises indices of the models to be produced, according to VMthrown ¼ 0.53 VMmodel þ 1.74. This result is all the more interesting in light of the fact that the maximal Von Mises index allows the inter-classification of different forms and masses. These analyses in fact reveal two concurrent phenomena. First e as indicated by the positive slope in the VMthrown vs. VMmodel relation e more mechanically taxing models systematically gave rise to thrown vessels being characterised by higher maximal Von Mises indices. Second, except for the 0.75 kg cylinder, all thrown vessels revealed a decrease in mechanical stress relative to the models presented, as revealed by the positive intercept and less-than-unity slope of the VMthrown vs. VMmodel relation. Expert potters were thus capable of reproducing vessels of varying mechanical difficulty, while adapting the vessel’s form in order to diminish the operative mechanical stress. For the (unfamiliar) conical vase shape, the vessels produced revealed maximal Von Mises indices that were considerably lower than expected on the basis of the linear relation described above. While larger for the 2.25 kg than for the 0.75 kg clay masses, the maximal Von Mises indices of the vases were, respectively, 68.2% and 67.1%, smaller than prescribed by the models. 1086 E. Gandon et al. / Journal of Archaeological Science 38 (2011) 1080e1089 Table 5 Proportions of model vessels and average proportions of thrown vessels. For the open vessels: aperture/height (A/H) and aperture/base (A/B). For the closed vessels: maximal diameter/height (MD/H), maximal diameter/base (MD/B) and maximal diameter/aperture (MD/A). Between-participants averages of the within-participant standard deviation capturing the variability over repeated reproductions are presented in parentheses. Model Vessels Form Mass (kg) A/H A/B Cylinder 0.75 0.56 0.97 2.25 0.56 0.97 0.75 2.96 3.70 2.25 2.96 3.70 Bowl Sphere Vase Thrown Vessels MD/H MD/B MD/A 0.75 1.42 2.04 2.05 2.25 1.42 2.04 2.05 0.75 2.14 2.33 3.32 2.25 2.14 2.33 3.32 4. Discussion Designed as a first step towards an assessment of potters’ skill as derived from ceramic assemblages, in this study we examined the geometrical and mechanical characteristics of vessels of different difficulty thrown by expert potters. Difficulty was experimentally varied by asking participants to reproduce (four) different model forms using (two) different quantities of clay. Absolute size was not prescribed: we solicited the largest possible form by asking participants to throw vessels of the form prescribed with the thinnest walls possible. Analysis of the absolute dimensions of the vessels thrown revealed that for each form participants indeed threw larger vessels when using the larger quantity of clay. They did not however faithfully reproduce the model forms, as revealed by differences in proportions (i.e., relative dimensions) between thrown vessels and model forms. Except for the cylinder e the simplest form e the deviations from the model forms invariably showed a lowering of throwing difficulty as assessed by Roux’s (1990) techno-morphological taxonomy. For the open forms, A/H A/B 0.60 (0.04) 0.51 (0.04) 2.34 (0.17) 2.42 (0.14) 1.03 (0.04) 1.01 (0.03) 2.50 (0.15) 2.62 (0.14) MD/H MD/B MD/A 1.33 (0.09) 1.26 (0.06) 1.68 (0.15) 1.65 (0.13) 1.66 (0.10) 1.66 (0.09) 1.48 (0.10) 1.46 (0.08) 1.87 (0.17) 2.03 (0.17) 2.46 (0.23) 2.77 (0.27) participants produced bowls that had smaller ratios of aperture over base and height. For the closed forms, participants produced spheres and vases that had smaller ratios of maximal diameter over aperture, base and height. Corresponding to another way of lowering throwing difficulty that can be derived from the taxonomy, the height of the maximum diameter of the closed forms also increased relative to the model forms, an effect especially visible for the (conical) vase (see Fig. 4). These deviations from the model forms were produced in a reproducible, standardised manner, as revealed by the low level of intra-individual between-specimen variation (5.5% on the average). Hence, interpreted within the framework of Roux’s technomorphological taxonomy, the deviations observed suggest that potters tend to reduce throwing difficulty for the more difficult forms. However, as Roux’s (1990) taxonomy of throwing difficulty is based on proportions of archetypal forms, it cannot be used to derive an absolute measure of throwing difficulty, nor to provide a criterion for classifying forms of different size. To obtain a global classification criterion, applicable to all vessel shapes and masses, we hypothesised that throwing difficulty is, at least partially, determined by mechanical constraints, notably those determining the risk of collapse of a thrown vessel under the influence of its own weight. Mechanical stress tests conducted in the first part of the study on wheel thrown vessels allowed the Von Mises stress index, derived via finite element modelling, to be validated in the present context as a useful global descriptor of the operative mechanical constraints. Table 6 Average deviations (in percentages) of thrown vessels from model vessels’ proportions. For the open vessels: aperture/height (A/H) and aperture/base (A/B). For the closed vessels: maximal diameter/height (MD/H), maximal diameter/base (MD/B) and maximal diameter/aperture (MD/A). Form Mass (kg) A/H (%) A/B (%) Cylinder 0.75 2.25 0.75 2.25 0.75 2.25 0.75 2.25 6.8 $8.6 $20.9 $18.4 7.1 4.4 $32.4 $29.1 Bowl Sphere Fig. 5. Average proportions (ratios) of thrown vessels as a function of the model ratios for each condition for the two clay masses (0.75 kg in white and 2.25 kg in black) and the four forms (C ¼ cylinder, B ¼ bowl, S ¼ sphere and V ¼ vase). Vase MD/H (%) MD/B (%) MD/A (%) $6.6 $11.0 $21.7 $23.1 $18.6 $18.4 $36.7 $37.5 $8.6 $0.8 $25.8 $16.5 E. Gandon et al. / Journal of Archaeological Science 38 (2011) 1080e1089 1087 Fig. 6. Maximal Von Mises stress values for the models and thrown vessels. The eight experimental conditions are ordered (left-to-right) following increasing maximal Von Mises values for the model vessels. The two masses are represented in grey (0.75 kg) and black (2.25 kg). The results obtained with respect to the Von Mises stress criterion confirmed and substantially extended the analyses based on the techno-morphological taxonomy, limited to comparing forms of the same size. Integrating both form and mass, the Von Mises index allowed inter-classification of vessels thrown under the eight different experimental conditions. The mechanical analysis of the models indicated that the larger cylinder was more difficult than the smaller cylinder but less difficult than the small sphere. The small sphere was approximately of the same difficulty as the small bowl. The large sphere was more difficult than the small bowl but less difficult than the large bowl, with the latter’s difficulty being approximately equal to that of the small vase. The large vase was the most difficult. For the cylinders, bowls, and spheres, potters threw vessels that followed the same ordering. Thus, for these forms, the difficulty of execution as captured by the level of mechanical stress can be accurately predicted from the difficulty of the models that potters were attempting to reproduce. An interesting result from a complementary video-based observational analysis revealed that the time spent in observing the model during throwing varied as a linear function of the mechanical difficulty of the model (Observation time ¼ 0.28 VMmodel e 0.67; R2 ¼ 0.878). While producing vessels of higher mechanical difficulty for more taxing models, potters did not faithfully reproduce the model forms. The mechanical analysis unambiguously revealed that the observed systematic deviations from the model forms are to be attributed to mechanical optimisation. Hence, notwithstanding the organising character of the intended form, in the throwing process potters adjusted the vessel’s form so as to alleviate the mechanical stress within the vessel to a certain extent. This observation resonates well with Ingold’s (1993: 461) assertion that “it is precisely because the practitioner’s engagement with the material is an attentive engagement rather than a mere mechanical coupling e because he watches, listens and feels as he works e that skilled activity carries its own intrinsic intentionality, quite apart from any designs or plans that it may be supposed to implement”. The starting point of our research programme into the evaluation of potters’ skills according to the characteristics of ceramic vessels is the hypothesis that the difficulty of throwing a vessel is to a significant extent determined by the operative mechanical constraints. This hypothesis was tested in the present contribution by comparing the degree of mechanical stress e synthesised by means of the Von Mises stress index e with the degree of task difficulty derived from Roux’s techno-morphological taxonomy of throwing difficulty. This taxonomy allows for several comparisons. First, it states that, for a given form, larger-sized vessels are more difficult to throw. For each tested form, clay masses of 2.25 kg systematically gave rise to larger-sized vessels than clay masses of 0.75 kg. Because the larger-sized vessels were always characterised by higher Von Mises stress indices, this first comparison corroborates our hypothesis. Second, the taxonomy states that open form vessels with larger ratios of aperture over height and base are more difficult to throw. Comparing the open forms experimentally tested (i.e., cylinder and bowl), our results demonstrated that, relative to the cylinder, the bowl revealed both larger A/H and A/B ratios and higher Von Mises stress indices, again corroborating the hypothesis. Third, the taxonomy states that closed form vessels with larger ratios of maximal diameter over height, base and aperture are more difficult to throw. Comparing the closed forms experimentally tested (i.e., sphere and vase), our results demonstrated that, relative to the sphere, the conical vase revealed both larger MD/H, MD/B and MD/A ratios and higher Von Mises stress indices, once again corroborating the hypothesis. Integrating both mass and form into a global measure, and thus going beyond the mere categorical classifications allowed by Roux’s (1990) techno-morphological taxonomy, the Von Mises stress indices allowed the different-sized cylinders, bowls, spheres and vases to be arranged along the same mechanical stress dimension (Fig. 6). Based on the foregoing comparisons and the informal results on the duration of model observation during the throwing process, we suggest that this arrangement also corresponds to experienced task difficulty. Participants in the present study were all expert potters. The results discussed so far indicate that, even for these highly skilled experts, more mechanically taxing forms were more difficult to throw, as revealed on the one hand by the correspondence between mechanical stress in the models proposed and in the vessels thrown, and on the other hand by the systematic mechanical optimisation produced, visible in the subtle yet methodical geometrical and mechanical differences between models proposed and vessels thrown. It is also noteworthy that even these expert potters were found to encounter severe difficulties when asked to reproduce the unfamiliar and mechanical strenuous conical vase form. To further validate the merits of the Von Mises index for the evaluation of potters’ skill, we analysed the productions of experts and intermediate potters as they appear in a dataset reported by Roux (1990). This dataset, including the relevant dimensions of vessels thrown by a group of seven expert potters and by a group of seven intermediate potters asked to reproduce four different forms (with two specimens per condition), allowed calculation of the degree of mechanical stress (i.e. Von Mises index) in the models and vessels thrown by the two groups. As can be seen from Fig. 7, while Roux’s expert potters revealed the same type of mechanical optimisation reported in the present 1088 E. Gandon et al. / Journal of Archaeological Science 38 (2011) 1080e1089 Fig. 7. Maximal Von Mises stress values for models and vessels thrown by two groups of potters of different expertise levels (experts in grey and intermediates in black), calculated on the basis of the data from Roux (1990). Intermediate and experts potters are adults, their skill levels differing with regard to the type of usual productions. experiment, the vessels they produced nevertheless demonstrated higher Von Mises indices than the vessels produced by the intermediate group. Thus, expertise is indeed visible in the degree to which the mechanical constraints of the form to be produced are marshalled. 5. Conclusion In the present contribution, we examined the possibility to characterise and classify ceramics in terms of throwing difficulty, defined via a measure of the mechanical stresses operating within the vessel under the influence of its own weight. In a first step, we demonstrated the validity of the Von Mises stress index, which incorporates the combined influences of form, mass, and clay properties, for assessing the sturdiness of a given vessel. By having potters attempt to reproduce different model forms with different quantities of clay, in a second step we compared throwing difficulty as measured by the Von Mises stress index to that derived from Roux’s (1990) techno-morphological taxonomy, grounded in geometrical proportions. These comparisons confirmed the potential of the Von Mises stress index to provide a valuable scale for evaluating throwing difficulty. Moreover, comparisons between model and thrown vessels revealed that expert potters were implicitly attuned to mechanical difficulty, with thrown vessels systematically deviating from model forms so as to reduce the risk of collapse. In a final step, we demonstrated that expert potters were better able to marshal the mechanical constraints than less experienced potters, demonstrating that this implicit attunement develops as a result of learning the skill of throwing. To conclude, we would like to insist on the importance of an analysis of ceramics taking into account the skill point of view. As a general rule, adoption of the wheel and its mastery has been progressive (e.g., Knappett, 2005; Roux, 2010). By assessing the skills developed in relation to the range of wheel-made ceramics, it is now possible to interpret the evolution of this range in terms of an evolution of the potters’ cultural perception of the properties of the wheel. Cultural perception of technical tasks is transmitted in the course of apprenticeship and defines ways of ‘doing’ of cultural groups (e.g., Dobres, 2000). Change in perception will express innovations, challenging the question of the dynamics underlying evolution in both producers’ behaviour and consumers’ demand. On this ground, variability within morphological types could be questioned in terms of a potter’s ability to reproduce types depending on its forming difficulties, which leads us to transmission studies and the genesis of new shapes in the course of apprenticeship. In the end, characterising distribution of skills over social groups will enable us to approach issues such as craft specialisation and socio-economic structures of ancient groups, on the one hand, and statuses of objects which may be unravelled through the understanding of the skills required for their making, on the other hand. Studying ceramic assemblages in terms of skills corresponds first of all to an anthropological approach to artefacts. In this regard, the analysis of the numerous technically heterogeneous ceramic assemblages, including wheel-made ceramics from the 5th to the 1st millennium BC, should greatly benefit from it. Because it allows interpretation of the time-frozen traces of ancient dexterity, the Von Mises stress criterion would seem to provide a useful window into the assessment of potters’ skills from ceramic archaeological assemblages. Considering the possible inferences on the socio-historic dimensions, this approach should be complementary to the others. The present contribution suggests that the assertion by Orton et al. (1993: 23) that “. every pot was (i) made or used at a certain time; (ii) made at a certain place; (iii) used for a certain purpose or purposes” may be usefully complemented with “(iv) made with a certain level of skill”. Acknowledgements We express our gratitude to the CNIFOP (Centre National de Formation aux Métiers de la Céramique) for providing us with access to facilities and material during the experiment. References Balfet, H., 1984. Methods of formation and the shape of pottery. In: Van Der Leeuw, S.E., Pritchard, A.C. (Eds.), The Many Dimensions of Pottery. Universiteit van Amsterdam, pp. 173e197. Berg, I., 2007. Meaning in the making: the potter’s wheel at Phylakopi, Melos (Greece). Journal of Anthropological Archaeology 26, 234e252. Bernstein, N.A., 1967. The Coordination and Regulation of Movements. Pergamon Press, London. Bril, B., 2002. L’apprentissage de gestes techniques: ordre de contraintes et variations culturelles. In: Bril, B., Roux, V. (Eds.), Le Geste Technique. Réflexions Méthodologiques et Anthropologiques. Erès, Ramonville Saint-Agne, pp. 113e150. Budden, S., 2008. Skill amongst the sherds: understanding the role of skill in the Early to late Middle Bronze Age in Hungary. In: Berg, I. (Ed.), Breaking the Mould: Challenging the Past through Pottery. Archaeopress, Oxford, pp. 1e18. E. Gandon et al. / Journal of Archaeological Science 38 (2011) 1080e1089 Crown, P.L., 2001. Learning to make pottery in the prehispanic American Southwest. Journal of Anthropological Research 57, 451e469. Crown, P.L., 2007. Life histories of pots and potters: Situating the individual in archaeology. American Antiquity 72, 677e690. Dobres, M.A., 2000. Technology and Social Agency. Outlining a Practice Framework for Archaeology. Blackwell Publishers, Oxford. Gagou, Y., Padayodi, E., Atcholi, K.E., Saint-Grégoire, P., 2008. Study of the mechanical behaviour of clay, a natural material for house construction. International Scientific Journal for Alternative Energy and Ecology 62, 187e193. Gelbert, A., 1997. De l’élaboration au tour au tournage sur motte: difficultés motrices et conceptuelles. Techniques et Culture 30, 1e23. Gelbert, A., 2003. Traditions céramiques et emprunts techniques dans la vallée du fleuve Sénégal. Editions Epistèmes, Paris (Bilingual CD-ROM). Gibson, J.J., 1979. The Ecological Approach to Visual Perception. Houghton-Mifflin, Boston. Gosselain, O.P., 2000. Materializing identities: an African perspective. Journal of Archaeological Method and Theory 7, 187e218. Ingold, T., 1993. Technology, language, intelligence, a reconsideration of basic concepts. In: Gibson, K.R., Ingold, T. (Eds.), Tools, Language and Cognition in Human Evolution. Cambridge University Press, Cambridge, pp. 449e472. Ingold, T., 2001. Beyond art and technology: the anthropology of skill. In: Schiffer, M.B. (Ed.), Anthropological Perspective on Technology. University of New Mexico Press, Albuquerque, pp. 17e31. 1089 Knappett, C., 2005. Thinking through Material Culture. An Interdisciplinary Perspective. University of Pennsylvania Press, Philadelphia. Lemaitre, J., Chaboche, J.L., 2004. Mécanique des Matériaux Solides. Dunod, Paris. Orton, C., Tyers, P., Vince, A., 1993. Pottery in Archaeology. Cambridge University Press, Cambridge. Rice, P.M., 1987. Pottery Analysis: a Sourcebook. University of Chicago press, Chicago. Roux, V., 1990. Elaboration d’une taxinomie pour mesurer les difficultés de tournage des céramiques préhistoriques et protohistoriques. In: Roux, V., Corbetta, D. (Eds.), Le Tour du Potier: Spécialisation Artisanale et Compétences Techniques. Editions du CNRS (Monographie du CRA), Paris, pp. 103e151. Roux, V., 2003. Ceramic standardization and intensity of production: Quantifying degrees of specialization. American Antiquity 68, 768e782. Roux, V., 2010. Technological innovations and developmental trajectories: social factors as evolutionary forces. In: O’Brien, M.J., Shennan, S.J. (Eds.), Innovation in Cultural Systems. Contributions from Evolutionary Anthropology. MIT Press, Cambridge, Massachusetts and London, England, pp. 217e234. Roux, V., Corbetta, D., 1990. Le Tour du Potier: Spécialisation Artisanale et Compétences Techniques. Editions du CNRS, Paris. Vitelli, K.D., 1993. Franchti Neolithic Pottery, Vol. 1. Indiana University Press, Bloomington. Yin, Z., 2006. Modélisation du comportement visqueux de l’argile naturelle. In: Proceedings of the XXIVème Rencontres Universitaires de Génie Civil, June 1-2, 2006, (La Grande Motte).
© Copyright 2025 Paperzz