Sei sulla pagina 1di 177



Direttori Daniele Malfitana · Jeroen Poblome · John Lund

Comitato scientifico S. E. Alcock (Brown University, R.I.) · P. M. Allison (Australian National University) · M. Bonifay (Centre Camille Jullian - UMR 6573, CNRS) · R. Brulet (Université Catholique de Louvain) · L. Chrzanovski (International Lychnological Association) · F. D’Andria (Università di Lecce) · M. de Vos (Università di Trento) · K. Dunbabin (McMaster University, Ontario) · M. Feugère (Equipe TPC - UMR 5140, CNRS) · I. Freestone (Cardiff University) · M. Fulford (University of Reading) · C. Gasparri (Università di Napoli “Federico II”) · E. Giannichedda · F. Giudice (Università di Catania) · A. Hochuli-Gysel (Fondation Pro Aventico, Avenches) · S. Ladstätter (Österreichische Akademie der Wissenschaften) · M. Lawall (University of Manitoba) · M. Mackensen (Ludwig- Maximilians-Universität, München) · D. Manacorda (Università di Roma Tre) · D. Mat- tingly (University of Leicester) · M. Mazza (Università di Roma “La Sapienza”) · D. Michaelides (University of Cyprus) · M. D. Nenna (Maison de l’Orient et de la Méditer- ranée, Lyon) · M. O’Hea (University of Adelaide) · E. Papi (Università di Siena) · D. P. S. Peacock (University of Southampton) · N. Rauh (Purdue University) · P. Reynolds (University of Barcelona) · G. Sanders (The American School of Classical Studies at Athens) · F. Slavazzi (Università di Milano) · K. W. Slane (University of Missouri-Colum- bia) · N. Terrenato (University of North Carolina, Chapel Hill) · M. Torelli (Università di Perugia) · H. von Hessberg (Universität zu Köln) · A. Wilson (University of Oxford) · D. Yntema (Vrije Universiteit Amsterdam)

Consulente di redazione per la grafica e la fotografia Giovanni Fragalà




· 2007





Amministrazione e abbonamenti Accademia editoriale ® Casella postale n. 1, succursale n. 8, i 56123 Pisa Tel. +39 050542332 · Fax +39 050574888

Abbonamenti (2007):

Italia: Euro 40,00 (privati) · Euro 60,00 (enti, con edizione Online) Abroad: Euro 60,00 (Individuals) · Euro 80,00 (Institutions, with Online Edition) Prezzo del fascicolo singolo: Euro 85,00

I pagamenti possono essere effettuati tramite versamento su c.c.p. n. 17154550 o tramite carta di credito (American Express, Visa, Eurocard, Mastercard)

Uffici di Pisa: Via Santa Bibbiana 28 · i 56127 Pisa Tel. +39 050542332 · Fax +39 050574888 E-mail: accademiaeditoriale @

Uffici di Roma: Via Ruggiero Bonghi 11/b · i 00184 Roma Tel. + 39 06 70493456 · Fax + 39 06 70476605 E-mail: accademiaeditoriale.roma @

La Casa editrice garantisce la massima riservatezza dei dati forniti dagli abbonati e la possibilità di richiederne la rettifica o la cancellazione previa comunicazione alla medesima. Le informazioni custodite dalla Casa editrice verranno utilizzate al solo scopo di inviare agli abbonati nuove proposte (Dlgs. 196/2003).

Autorizzazione del Tribunale di Pisa n. 22 del 15-ix-2004 Direttore responsabile: Fabrizio Serra

Sono rigorosamente vietati la riproduzione, la traduzione, l’adattamento, anche parziale o per estratti, per qualsiasi uso e con qualsiasi mezzo effettuati, compresi la copia fotostatica, il microfilm, la memorizzazione elettronica, ecc., senza la preventiva autorizzazione scritta della Fabrizio Serra · Editore ® , Pisa · Roma, un marchio della Accademia editoriale ® , Pisa · Roma. Ogni abuso sarà perseguito a norma di legge.

Proprietà riservata · All rights reserved © Copyright 2008 by Fabrizio Serra · Editore ® , Pisa · Roma, un marchio della Accademia editoriale ® , Pisa · Roma

Stampato in Italia · Printed in Italy

La Accademia editoriale ® , Pisa · Roma, pubblica con il marchio Fabrizio Serra · Editore ® , Pisa · Roma, sia le proprie riviste precedentemente edite con il marchio Istituti editoriali e poligrafici internazionali ® , Pisa · Roma, che i volumi delle proprie collane precedentemente edite con i marchi Edizioni dell’Ateneo ® , Roma, Giardini editori e stampatori in Pisa ® , Gruppo editoriale internazionale ® , Pisa · Roma, e Istituti editoriali e poligrafici internazionali ® , Pisa · Roma.

issn 1971-9051


«Facta. A Journal of Roman material culture studies»


Jeroen Poblome, Daniele Malfitana, John Lund, Tempus fugit, «Facta» manent. Editorial statement


Michel Feugère, Techniques, productions, consommations: le sens des objects


Johan Modée, Outline of a New Theory of Artifacts


Enrico Giannichedda, Lo scavo, i residui e l’affidabilità stratigrafica


Gary Reger, Regions Revisited. Identifying Regions in a Greco-Roman Mediter- ranean Context


Lucio Fiorini, Mario Torelli, La fusione, Afrodite e l’emporion


Marcello Mogetta, Nicola Terrenato, Architecture and Economy in an early imperial Settlement in Northern Etruria


Marie-Dominique Nenna, Production et commerce du verre à l’époque impé- riale: nouvelles découvertes et problématiques


Fabrizio Slavazzi, Un piatto in porfido da Cremona. Note su una classe di vasel- lame di lusso


Helga Di Giuseppe, Proprietari e produttori nell’alta valle del Bradano


Indirizzi degli autori



Notes of the editors

acta» is an international, peer-reviewed journal aimed at creating a forum for Roman ma- terial culture studies. The Journal is published annually. «Facta» gives central place to

wider interpretative studies on how artefacts were produced and used throughout the Roman Empire, from Republican/Hellenistic times into Late Antiquity. The journal is interdiscipli- nary in nature, promoting the role of ancient history, technology, design, archaeometry, socio- political studies, socio-cultural studies, economy, demography, ethnology and other scientific disciplines, in archaeology. The journal promotes reflexive approaches to Classical Archaeolo- gy. «Facta» considers it highly desirable for Roman material culture studies to find a new élan and balance, in order to answer the many challenges posed by the evolution of the wider in- terdisciplinary field and Roman archaeology in general. The latter, performed in interdiscipli- nary ways, seems especially suited to driving our field of research in innovative directions and to new heights. We trust that the creation of this new Journal has a strong potential for pro- viding more visibility for Roman material culture studies in many ways, which are not neces- sarily covered by existing international journals or conference proceedings. The first issue of the Journal will entertain papers on a wide range of topics, whereas in subsequent numbers such will be combined with papers devoted to specific themes. For the next issue, «Facta» is in- terested in papers on the theme of “Rhosica vasa” and “the complex relationship between sherds and coins”, along with non-thematic papers. Besides it will propose reviews of books sent to the editors and considered of importance for methodological approach and results.

« F

It is intended to publish a series of «Facta. Supplementa», focussing on particular themes covered by the Journal. The first supplement is entitled «From Amphorae to modelling the Late Roman economy» and comprises the Acts of a roct-Conference at the University of Ghent in Belgium (5-6 December 2005).


All papers result from careful reflection and the argumentation is clearly presented, with atten- tion to the broader significance of the results. Papers make an original and significant contribu- tion to the field of Roman material culture studies. Papers with the potential to stimulate fur- ther discussion in the Journal will be preferentially accepted. Accessibility to the non-specialist reader is to be kept in mind, and citations in ancient languages should always be translated.

Editorial procedure

The Journal is run by the Editors, with the appreciated help of the Editorial Advisory Board. Submitted work is reviewed by each of the Editors, as well as at least two members of the Ed- itorial Advisory Board. «Facta» is in favour of revealing authorship to reviewers, as well as re- vealing the identities of reviewers to authors. Authors of papers which have been accepted need to take the detailed comments of the reviewers into account before final submission.


«Facta» is scheduled to appear by the end of each year. The optimum time for submission is between January and August of every given year.


notes of the editors

Style guidelines

The manuscript is supplied to the Editors in a current word for Windows version, submitted by email, and by first class airmail including a cd/dvd, accompanied by a print-out and hard copy of all illustrations, drawings, tables and graphs. Use top, bottom, right and left margins of 2 cm on an A4 format. Use single line spacing and 6 pt spacing-after for paragraphs. Do not use double returns, or any other additional formatting of the text. Choose Times New Ro- man, 14 pt-bold-centred for the title and author(s), 12 pt-justified for the abstract and body text and 10 pt-left for the endnotes. The preferred language for publications submitted to «Facta» is English, but the journal will also consider contributions in Italian, French or German. «Facta» acknowledges that the use of one or other language can be preferred when a paper is intended to contribute to a specific debate in the world of Roman archaeology. Non-native speakers in any of these languages are advised to have their contributions corrected before submitting. In case these languages are considered problematic, please contact the Editors. Each manuscript is headed by a fully descriptive title and comprises an abstract in English. Along with full address information for every author, also information on the funding source for each of the parties involved in the article needs to be submitted. All manuscripts which do not follow the formal guidelines shall be returned to the authors, who must make the appropriate corrections before re-submitting.


Each contribution is preceded by an English abstract of up to 300 words. The body text is not formatted in columns. A list of references follows the paper, using the Harvard system. Ref- erences should be complete. Endnotes never contain complete references, only abbreviated references with surname of author(s) (if more than two, use et al.), year of publication, refer- ence to specific pages or illustrative material (e.g. Hayes 1972, pp. 33-35). In case more than one publication of the same year of the same author is cited, the year is followed by a, b and so on (e.g. Hayes 1993a, pp. 16-19). All drawings, illustrations, graphs and tables follow at the end of the paper. A manuscript for an article may comprise up 15 such pages including notes and bib- liography. Please contact the Editors in advance if there is need for more space.


All illustrative material should be free of copyright. All computer-generated illustrative ma- terial should be submitted as a high quality print out and as a digital file (preferably .tiff or .jpg, with a minimum resolution of 300 dpi). Scales should be included in the illustration or be clear- ly indicated. Original photographs and drawings will be returned to the corresponding author after printing, upon simple request. Tables should be formatted as simple as possible, using simple lines between rows and columns. Graphs are only generated in a current Excel for Win- dows version, and supplied both as print-out and as file, including the raw data. A separate list of captions should mention creditors and source, whenever necessary. Reference to illustra- tive material in the body text is formatted as follows: (Fig. 10) (Fig. 10-11) [used for both draw- ings and other types of illustrations], (Table 1) (Table 1-3) and (Graph 1) (Graph 5-6). A pa- per includes maximum 10 figures (including tables and graphs).


References to modern, scholarly literature should follow the Harvard system, according to which the full bibliographic reference to a work is given only in the list of references. For for- matting this list, please consult

notes of the editors


The following standard abbreviations are used:

110 ad – 232 bc – Cat. – cf. – ed. – esp. – i.e. Measurements are indicated as follows: H = height; W = width; L = length; Th = thickness; Diam = diameter; cm = centimetre; m = meter; km = kilometre.


Only one proof will be send to the corresponding author.


The corresponding author will receive a pdf-file of the final version of the article, as printed.

Books for review

Books for review should be sent to one of the Editors, who, where appropriate, can contact specialist reviewers.

Contacting the editors

If issues arise which are not covered by these guidelines – or in the case of doubt about their application – please contact the Editors for clarification.

Daniele Malfitana d.malfitana @

Jeroen Poblome Jeroen.poblome @

John Lund John.Lund @


Jeroen Poblome · Daniele Malfitana · John Lund

I n late 2006, as the three of us finalized editing «Old Pottery in a New Century: Inno- vating Perspectives on Roman Pottery Studies. Atti del Convegno Internazionale di Studi

Catania, 22-24 aprile 2004»,1 we asked ourselves what might be a logical next step to fur- ther and hopefully enrich the field of research, in which we have passionately engaged for more than a decade. The answer – and our inspiration – came from Susan E. Alcock’s concluding re- marks to the said volume,2 which stressed the need for ceramologists to broaden their view and address pottery in the wider context of Roman artefact studies, by tying ce- ramic evidence to other relevant artisanal phenomena. In short: to study the ancient material culture in its totality, with a view to clarifying the complex wider implica- tions of such evidence for understanding a host of issues concerning for instance the economy, daily life, politics, religion, and ultimately the history of the ancient Roman world. A fruitful encounter with Fabrizio Serra of the ‘Fabrizio Serra · Editore’ convinced us that a useful move towards realizing this goal would be to launch a new journal as a forum for our endeavour. Hence «Facta» was conceived. We chose the name, because the Latin word covers all human creations, material as well as ideological. «Facta» stands for all kinds of man-made works from the humble handicraft to the capolavori of precious artistic expressions.3 The journal is intended as an honest and open-minded intellectual and scholarly endeavour. Our background is transparent: a fascination with broken Roman pottery, lots of it for that matter, from a disparate range of sites and regions in the Central and Eastern Mediterranean. What originally brought us together were our trials-and- errors in attempting to make ancient sites, regions, and their material culture speak, and we remain convinced that such personal, deep-rooted interests will make «Facta» work better than any given institutional or project-based link.4 We acknowledge the many strengths of the discipline of Roman ar chae ology,5 which has by now established a firm typo-chronological framework for most artefact classes from most regions of the empire.6 But to decode the messages inherent in these «reluctant witnesses to the past»,7 scholarly traditions must be in constant flux in order to remain relevant to the wider academic environment, and this is why the journal wants to bring scholars and disciplines together, in more ways than one.

1 Malfitana-Poblome-Lund 2006.

2 Alcock 2006.

3 Cato, Agr. 21, 5: «De rebus corporalibus: de metallis, quae in instrumentis, vasis, signis sunt »; Porph., «quae- cumque perfectissima esse volumus significare, ad unguem facta dicimus».

4 We have set forth our conceptual background in Lund-Malfitana-Poblome 2006 and Poblome-

Malfitana-Lund 2006.

5 See Woolf 2004.


jeroen poblome · daniele malfitana · john lund

We live in an age of specialization, which is not in itself a bad thing. Yet, in the ar- chaeological community, specialization runs the risk of scholars working and com- municating in self-perpetuating and more or less self-contained circles focusing on a single category of finds, such as objects of glass, metal ware, terracottas, pottery (sub- divided into table wares, transport amphorae, cooking wares, lamps), gems, coins etc. It often seems as if the involved scholars work along parallel lines: posing compara- ble questions, formulating similar theories, and even suggesting the same answers, but without being aware of what is going on in other scholarly circles dealing with other artefact classes. It is our ambition and hope that «Facta» may unite all students of the worldwide scientific community engaged in all aspects of Roman material cul- ture studies.1 To be sure, the study of artefacts requires master ing a great deal of tech- nical and observation skills, providing more than sufficient raison d’être for the exist- ing scholarly circles focusing on the study of one or other type of artefact.2 Such societies are hotbeds for the exchange of basic data, developing best praxis and guid- ing new talent, and they certainly have every reason to continue to exist, provided that their debate avoids becoming internalized and that they remain in touch with wider methodological developments in other domains of science. «Facta» wants to bridge theory and praxis by providing a platform for innovative methodological and theoretical approaches towards Roman material culture by cross-fer ti lization between the traditional fields of study.3 We are confident that the coupling of skilfully collected catalogues of one or the other type of artefactual evi- dence with an innovative methodological and theoretical matrix will spark new im- pulses for the study of each of the categories involved. This practice will not only im- prove the current strategies of communication inherent in such Roman material culture but should also provide corrections for stale scholarly traditions. The ordinary Roman households and communities were characterized by a hybrid, contingent, context-specific material culture, but the current traditions of research can only take us so far in understanding how daily life in antiquity was organized. An-

1 «Facta» is part of a wider international scholarly context, and has benefited from the experience of

other initiatives, which have gone before us. On the Italian scene, reference may be made to «Dialoghi di

archeologia» founded by R. Bianchi Bandinelli in 1967 and followed by the research encapsulated in the vol- umes «Società romana e impero tardo-antico», and then the journal «Ostraka» (1992) directed by Mario Torel- li; we may also mention «Archeologia e cultura materiale» (1975) as well as the «Workshop di archeologia. Paesaggi, costruzioni, reperti» (2004) directed by Andrea Carandini and Emanuele Greco. More recently, other, more specialized, reviews have emerged, such as «Eidola. International Journal of Classical Art His- tory», directed by Irene Favaretto and Francesca Ghedini and dedicated to research in iconography, iconol- ogy and ancient art history in general; mention should also be made of «Musiva & Sectilia». An Interna- tional Journal for the Study of Ancient Pavements and Wall Revetments in their Decorative and Architectural Context (2004), directed by Federico Guidobaldi, and finally of «Marmora». An International Journal for Archaeology, History and Archaeometry of Marbles and Stones (2005) directed by Lorenzo Laz- zarini. In Belgium, «roct», representing the international network focusing on «Roman Crafts and Trade», directed by Marc Waelkens, has paved the way for «Facta» in many of its workshops.

2 For instance, for the Rei Cretariae Romanae Fautores,

the Late Roman Coarse Wares, Cooking Wares and Amphorae in the Mediterranean network, l’Association Inter-

nationale pour l’Histoire du Verre at, the Crafts network on l’archéologie de l’artisanat et économie de l’Empire romain, Instrumentum at and the International Lychno- logical Association at

tempus fugit, «facta» manent. editorial statement 15 cient artefacts should not primarily be studied as archaeological tools (for dating etc.), but as evidence of the many functions fulfilled by artisanal products, composed, used and ultimately discarded by the inhabitants of the Roman commonwealth and its shadow zones.1 We are convinced that any collection of material culture is non-coincidental, and «Facta» will give pride of place to contributions illustrating such collections retrieved from excavation and/or survey work. We wish to understand better how and why Ro- man material culture was produced and circulated, how and why the inhabitants of the Empire tapped into it, possibly changing that very culture by the simple act of consumption. Ultimately, we want to investigate how, why and in which contexts Ro- man material culture was used. How fast it changed (or needed to change), became redundant and was discarded or recycled for another material life.2 The non-coincidental character of Roman material culture implies that its study must be concerned with all aspects of ancient daily life. People, communities and their artefactual assemblages together form society, thus providing meaning to Ro- man material culture in the way that it sustained habits, routines and traditions of life, but also contained gender patterns, systems of values, beliefs and memory.3 Human agency is crucial in this respect,4 and «Facta» will strive to document such in its pages. By incorporating implicit or explicit messages, Roman artefacts are closely engaged with contemporaneous culture, which is likewise very visual in order to reach out to most layers of society.5 The notion of context (or indeed contexts)6 is of great importance,7 ranging from the variable quality of archaeological deposits or surveyed collections, to the actual context of usage. On a small scale, the latter comprise finds associated with architec- tural units and their development, households with all their members in constant evo- lution, communities and their urban or other places of shared living, or regional catchment areas with high quality mineral or other resources. On a large scale, «Facta» considers the context of artefactual usage in the Roman Empire as a histori- cally specific demonstration of the many possible effects of the Empire, when and where it existed. This does not at all imply that the outcome of research in material culture is known, but we suspect that it will enable us to recognize some processes of more universal value, such as cultural bricolage, social and technological emulation between crafts, the existence of hybrid identities, the working of human agency and the concept of cultural relativism.8 «Facta» also wishes to be a forum for wider debates in classical studies. The role, position and development of technology is one potential area for such a debate, and the possible contribution of a study of crafts to ancient socio-economic patters

1 Cf. Allison 1999, and for a very British example, Hingley-Willis 2007; for a very Pompeii example,

Allison 2006.

2 Cf. the illuminating approach taken by Peña 2007.

3 Appadurai 1986; Cumberpatch-Blinkhorn 1997; Alcock 2002; Sørensen 2004; Meskell-Preucel

2004; Johnson 2004.


Boivin 2004. 5 Huskinson 2000; Clarke 2003.


It should not be forgotten that each object forms part of innumerable contexts through (and even af-

ter) its use-life, cf. Nørskov 2002. The archaeological context, defined by Martin Carver (Carver 2005, p.107) as the «prime stratigrafic unit» is only one of these, albeit naturally an important one.


jeroen poblome · daniele malfitana · john lund

another.1 Even if ancient technology is in evidence everywhere, and comprises much more than artefacts for that matter, its study by ancient historians or archae- ologists is rarely considered as central to the functioning of Roman society.2 Unfor- tunately, much technological, archaeometrical work can often seem overly science- specific and out of touch with the main issues in the archaeological debate. «Facta» wishes to support technological studies of whichever nature, that use their scien- tific rationale in order to highlight the variegated role material culture played in Ro- man life and thereby demonstrate that no technology is stagnant, but socially, cul- turally and possibly politically, embedded and crucial to our understanding of how ancient individuals and communities functioned by finding creative answers to their challenges in life.3 The most socially embedded types of technology, that of cook- ing and eating,4 as well as the technology of dealing with the death,5 are prime ex- amples of the multi-layered use of versatile and evolving material culture. It is paradoxical that while most Roman material culture specialists consider their ef forts to be highly relevant for the study of ancient mechanisms of production and exchange, their work finds little resonance in the general debate on the ancient econ- omy, or perhaps rather socio-economy. «Facta» aims at bringing these parties closer together by intentionally leaving the archetypical debate between so-called primiti- vists and modernists, or any other – ists, behind, and let the material speak for itself. In most cases, the object of Roman material culture studies is broken and humble, but then again there is lots of it, forcing us to contemplate on the role of the producer, consumer and every party in between these two. As far as «Facta» is concerned, there was money to be earned in the production and distribution of Roman material cul- ture, most obviously at the level of production, but the higher the output, the more levels of society got involved, in non-innocent ways.6 Apart from theoretical/disciplinary inspiration, social matrix, cultural message, as- pects of contexts, technological input and socio-economic impact, regionalism forms another interesting arch of tension between cohesion and diversity in Roman mate- rial culture, which was composite and complex yet at the same time also displaying lines of similarity and visual recognition. For example, at no point in history, before or after, have roofs looked so recognizably similar over an incredibly vast geographi- cal area, because they employed the classic tegula and imbrex. And at the same time, most ceramic assemblages contain a local or regional component, next to generic and inter-regionally distributed wares such as sigillata or transport amphorae. “Connec- tivity of micro-regions” was the ultimate simplified explanation P. Horden and N. Purcell brought to bear on the notion of cohesion in the history of the Mediterranean Sea and its coastlands over more than three millennia, at the same time warning us, that such processes of interdependence were not inevitable, but contingent, context- specific, unplanned, reversible and possibly in conflict.7

1 «Facta» will also be a forum for broadly based studies, which transgress the «Roman» period, as

normally defined, if they present new results concerning ancient craft production of technological and methodological importance. An example of this is the contribution of L. Fiorini and M. Torelli in the present issue.


For a healthy exception, cf. Wilson


3 Dobres 2000; Cuomo 2007.


Cool 2006.

5 Morris 1992.


Morley 2007; Mattingly-Salmon 2001.

7 Horden-Purcell 2000.

tempus fugit, «facta» manent. editorial statement 17 «Facta» recognizes that Roman material culture can contribute to our under- standing of the working of regions, in so far as the successful production and distri- bution of a range of artefacts never came about by itself, but formed part of trajec- tories of regional development. Regions1 should not be equated with fixed socio-cultural or political units, however, as this would obscure the potential of Ro- man material culture to contribute to the study of ethnicity. Considering the extent of the Empire, a more dialectic approach towards style and design of artefacts and their apparent combinations in appearance in function of variable contexts of usage and display, could lead to a more fruitful way to approach the very heterogeneous and evolving composition of the population of the empire. As with social interac- tion, no easy answers may be forthcoming, but material culture structures (and is structured) by both in the same, actively constituted and negotiated way, inviting ma- terial culture specialists to start looking beyond their traditional classificatory crite- ria and tools in order to put people on the map.2 Ethnicity is one way of doing that, demography another. Survey archaeologists, for instance, have entered the debate of demographic trends and evolutions, based on their observations, which include for an important part material culture.3 The field of material culture studies, Roman or otherwise, is undisciplined and multi-strategic, and «Facta» does not regard the zones between typical academic dis- ciplines as grey and shady, but as areas of potential intellectual development. In order to succeed, material culture studies must be cross-disciplinary,4 and the journal aspires to become a source of knowledge and inspiration for third parties, such as other Roman archaeologists, fellow material culture enthusiasts, historians (and not necessarily of the ancient ilk), as well as archaeometrists, anthropologists, ethnogra- phers, geographers, sociologists, art historians and designers. In compliance with the multi-strategic approach to Roman material culture, we have no preference for one or the other standpoint in the wider theoretical archaeological debate. This is very much in tandem with the many ways Roman material culture studies are being prac- ticed around the globe – and we consider our openness to more than one approach to have an advantageous potential for growth. In conclusion, Roman material culture «cannot be explained by reduction to [its] component parts», as «the whole is greater than the sum of its parts» whilst its «phe- nomena … are unique in particular but similar in general».5 Its wider interpretation should therefore raise an interest in complexity science. As material culture studies itself, complexity science is a multi-disciplinary undertaking. The study of human society and its material component in the past as an open system of interacting agents, with any given change in action causing many other changes, is an apt ap- plication of archaeology in complexity science, and «Facta» should like to open a door for Roman material culture and its many enthusiasts. John Bintliff has added the theory of punctuated equilibrium to this picture, by which natural scientists con-

1 See the paper by G. Reger in this issue of the journal.

2 Jones 1997, with a useful discussion on Romanisation and ethnicity.

3 Bintliff-Sbonias 1999; Bintliff-Sbonias 2000.

4 The successful Journal of Material Culture, at, is a prime example, along with e.g. Tilley et al. 2006 and Meskell 2005.


jeroen poblome · daniele malfitana · john lund

sider «the past [as] the result of the uncertain interplay between chance occurrences and the adaptive pressures which lead to lasting and ever more complex ecological and biotic structures’.1 «Facta» is well aware that the distance between these theo- ries and the field of Roman material culture studies is great, but as engaged mate- rial culture specialists we predict excellent matches between the complexity of the material culture we study, the complexity of its contemporaneous culture and soci- ety and our archaeologies of complexity.2


We want first of all to thank all our colleagues who have accepted with enthusiasm to become members of the Editorial Board of the journal. We are particularly grateful to Professor Fab- rizio Serra for having accepted to publish the journal in the distinguished publishing house, of which he is director. Dr Rita Gianfelice has competently followed the editing process of the volume, while Giovanni Fragalà was been responsible for the layout of the illustrations. As for the three general editors of the volume, Jeroen Poblome’s research is carried out within the framework of icrates (fwo-g.0152.04), the Belgian Programme on Interuniversity Poles of Attraction (P6/22), the 2007/02 Concerted Action of the Flemish Government and fwo.g.0421.06, while the research of Daniele Malfitana is carried out as part of the pro- gramme ibam – cnr (pc.p002.001: Approcci multidisciplinari integrati per l’analisi dei manufatti:

dalla produzione, alla circolazione e all’uso), which he is directing; John Lund wants to thank the National Museum of Denmark – particularly the Collection of Classical and Near Eastern An- tiquities – for continued support.


Alcock 2002 = S. E. Alcock, Archaeologies of the Greek past. Landscape, monuments and memories, Cambridge, 2002. Alcock 2006 = S. E. Alcock, Small things in the Roman world, in Malfitana-Poblome- Lund 2006, pp. 581-585. Allison 1999 = P. M. Allison, Labels for ladles: interpreting the material culture of Roman house- hold, in The Archaeology of Household activities, ed. P. M. Allison, London-New York 1999, pp.


Allison 2006 = P. M. Allison, The insula of the Menander at Pompeii, 3. The finds, a contextual study, Oxford, 2006. Appadurai 1986 = The social life of things. Commodities in cultural perspective, ed. A. Appadurai, Cambridge, 1986. Bentley 2003 = R. A. Bentley, An introduction to complex systems, in Complex systems and archaeology. Empirical and theoretical applications, eds. R. A. Bentley, H. D. G. Maschner, Salt Lake City, 2003, pp. 9-23. Bentley-Maschner 2003 = R. A. Bentley, H. D. G. Maschner, Considering complexity theory in archaeology, in Complex systems and archaeology. Empirical and theoretical applications, eds. R. A. Bentley, H. D. G. Maschner, Salt Lake City, 2003, pp. 1-8. Bintliff 2003 = J. Bintliff, Searching for structure in the past – or was it “one damn thing after another”? in Complex systems and archaeology. Empirical and theoretical applications, eds. R. A. Bentley, H. D. G. Maschner, Salt Lake City, 2003, pp. 79-83. Bintliff 2004 = J. Bintliff, Time, structure and agency: the annals, emergent complexity and archaeology, in A companion to archaeology, ed. J. Bintliff, Oxford, 2004, pp. 174-194.

1 Citation from Bintliff 2004, p. 186; see also Bintliff 2003.

2 Chapman 2003.

tempus fugit, «facta» manent. editorial statement


Bintliff-Sbonias 1999 = Reconstructing past population trends in Mediterranean Europe (3000 bc-

ad 1800), (The Archaeology of Mediterranean Landscapes 1), eds. J. Bintliff, K. Sbonias, Oxford,


Bintliff-Sbonias 2000 = J. Bintliff, K. Sbonias, Demographic and ceramic analysis in regional

survey, in Extracting meaning from ploughsoil assemblages, (The Archaeology of Mediterranean Landscapes 5), eds. J. Bintliff, K. Sbonias, Oxford, 2000, pp. 244-258. Boivin 2004 = N. Boivin, Mind over matter? Collapsing the mind-matter dichotomy in material

culture studies, in Rethinking materiality. The engagement of mind with the material world, eds.

E. DeMarrais, C. Gosden, C. Renfrew, Cambridge, 2004, pp. 63-71.

Caple 2006 = C. Caple, Objects: reluctant witnesses to the past, London-New York, 2006. Carver 2005 = M. Carver, Key ideas in excavation, in Archaeology. The Key Concepts, eds. C. Renfrew, P. Bahn, London-New York, 2005, pp. 106-110. Chapman 2003 = R. Chapman, Archaeologies of complexity, London, 2003. Clarke 2003 = J. R. Clarke, Art in the lives of ordinary Romans. Visual representation and non- elite viewers in Italy, 100 B.C.-A.D. 135, Berkeley, 2003. Cool 2006 = H. E. M. Cool, Eating and drinking in Roman Britain, Cambridge, 2006. Cumberpatch-Blinkhorn 1997 = Not so much a pot, more a way of life. Current approaches to artefact analysis in archaeology (Oxbow Monograph 83), eds. C. G. Cumberpatch, P. W. Blinkhorn, Oxford, 1997. Cuomo 2007 = S. Cuomo, Technology and culture in Greek and Roman antiquity, Cambridge,


Dobres 2000 = M.-A. Dobres, Technology and social agency. Outlining a practice framework for archaeology, Oxford, 2000.

Ettlinger et al. 1990 = E. Ettlinger et al. Conspectus Formarum Terrae Sigillatae Italico Modo Confectae (Materialien zur römisch-germanischen Keramik 10), Bonn, 1990. Gardner 2006 = A. Gardner, The future of Trac, in Trac 2005. Proceedings of the Fifteenth Annual Theoretical Roman Archaeology Conference, eds. B. Croxford, H. Goodchild, J. Lucas,

N. Ray, Oxford, 2006, pp. 128-137.

Hingley 2005 = R. Hingley, Globalizing Roman culture. Unity, diversity and empire, London,


Hingley-Willis 2007 = Roman finds. Context and theory, eds. R. Hingley, S. Willis, Oxford,


Horden-Purcell 2000 = The corrupting sea. A study of Mediterranean history, eds. P. Horden,

N. Purcell, Oxford, 2000.

Huskinson 2000 = Experiencing Rome. Culture, identity and power in the Roman empire, ed. J. Huskinson, London, 2000. Johnson 2004 = M. Johnson, Archaeology and social theory, in A companion to Archaeology, ed.

J. Bintliff, Oxford, 2004, pp. 92-109.

Jones 1997 = S. Jones, The archaeology of ethnicity. Constructing identities in the past and present,

London, 1997. Laurence 2006 = R. Laurence, 21st century Trac: is the Roman battery flat? in trac 2005.

Proceedings of the Fifteenth Annual Theoretical Roman Archaeology Conference, eds. B. Croxford,

H. Goodchild, J. Lucas, N. Ray, Oxford, 2006, pp. 116-127.

Lund-Malfitana-Poblome 2006 = General introduction, in Malfitana-Poblome-Lund 2006, pp. 19-24. Malfitana-Poblome-Lund 2006 = Old pottery in a new century. Innovating perspectives on Ro-

man pottery studies. Atti del convegno internazionale di studi, Catania 22-24 aprile 2004, Monografie dell’Istituto per i Beni Archeologici e Monumentali, c.n.r., 1, a cura di D. Malfitana, J. Poblome,

J. Lund, Catania, 2006.

Mattingly-Salmon 2001 = Economies beyond agriculture in the classical world (Leicester-Notting- ham Studies in Ancient Society 9), eds. D. J. Mattingly, J. Salmon London, 2001.


jeroen poblome · daniele malfitana · john lund

Meskell 2005 = Archaeologies of materiality, ed. L. Meskell, Oxford, 2005. Meskell-Preucel 2004 = A companion to social archaeology, eds. L. Meskell, R. W. Preucel, Oxford, 2004. Morley 2007 = N. Morley, Trade in classical antiquity, Cambridge, 2007.

Morris 1992 = I. Morris, Death-ritual and social structure in classical antiquity, Cambridge,


Nørskov 2002 = V. Nørskov, Greek Vases in New Contexts: The Collecting and Trading of Greek Vases. An Aspect of the Modern Reception of Antiquity, Aarhus, 2002. Papaconstantinou 2006 = Deconstructing context. A critical approach to archaeological practice, ed. D. Papaconstantinou, Oxford, 2006. Peña 2007 = T. Peña, Roman Pottery in the Archaeological Record, Cambridge, 2007. Poblome-Malfitana-Lund 2006 = J. Poblome, D. Malfitana, J. Lund, A concluding dilemma. Sisyphos versus Daidalos, in Malfitana-Poblome-Lund 2006, pp. 557-579. Sørensen 2004 = M. L. S. Sørensen, The archaeology of gender, in A companion to archaeology, ed. J. Bintliff, Oxford, 2004, pp. 75-91. Tilley et al. 2006 = Handbook of material culture, eds. C. Tilley, W. Keane, S. Küchler, M. Row- lands, P. Spyer, London, 2006. Wilson 2002 = A. Wilson, Machines, Power and the Ancient Roman Economy, «The Journal of Roman Studies» 92, 2002, pp. 1-32. Woolf 2001 = G. Woolf, Inventing empire in ancient Rome, in Empires. Perspectives from archaeology and history, eds. S. E. Alcock, T. N. D’Altroy, K. D. Morrison, C. M. Sinopoli, Cambridge, 2001, pp. 311-322. Woolf 2004 = G. Woolf, The present state and future scope of Roman archaeology. A comment, «American Journal of Archaeology», 108, 2004, pp. 417-428.



Michel Feugère

L a création de la revue «Facta» coïncide avec la restructuration en équipes de l’umr 5140 du cnrs «Archéologie des Sociétés Méditerranéennes», basée à Lattes près

de Montpellier (F).1 Dans le cadre de cette nouvelle organisation, une trentaine de col- lègues ont souhaité se retrouver au sein d’une entité regroupant les chercheurs inté- ressés par la mise en œuvre des matériaux et les multiples dimensions, techniques et culturelles, de cette grande aventure humaine. Ainsi est née l’équipe «tpc» («Tech- niques - Productions - Consommations»), dont l’intitulé embrasse large: difficile en effet de se limiter à une période ou à un territoire, tant il est vrai que des frontières seraient, sur une telle thématique, impossibles à fixer ou à respecter. Tel quel, notre thème d’étude possède cependant une vraie légitimité, que cette contribution me donne l’occasion d’analyser en détail. La matière première de l’archéologue peut être divisée en deux grands ensembles, l’immobilier et le mobilier. Dans la première catégorie se trouvent des sites, dont une partie est fournie par la nature et une autre modifiée par l’homme: ces «structures», au sens où l’entend le fouilleur, ne constituent évidemment que ce qui reste de tels aménagements après que le temps ait fait son œuvre. Bâtiments, clôtures, dépotoirs, tombeaux … etc, mis en place au cours des activités humaines, sont réduits le plus souvent à leur partie inférieure, éventuellement enterrée. Dans les niveaux de terre rapportée peuvent être conservés des éléments mobiliers, de taille et de nature varia- bles, allant de la graine carbonisée à la sculpture de marbre. L’objet de notre étude est l’artefact, l’objet façonné par la main de l’homme pour répondre à un besoin précis. Bien sûr, étudier le mobilier sans le site n’a pas plus de sens que la proposition inverse, chaque culture humaine ne se développant que dans un contexte culturel et technologique qui lui est propre: réunir des chercheurs sur un aspect de la réalité ancienne ne signifie nullement que la problématique puisse se dé- velopper de manière isolée! Mais la production manufacturée fait appel à des connais- sances si variées que la spécialisation apparaît comme un mal nécessaire. Pour conser- ver un contact étroit entre le spécialiste de la céramique préromaine et celui de la pierre taillée, entre celui de la terre crue et celui des petits objets en os, une structure commune est indispensable. Le ciment de cette équipe au programme si varié, pour ne pas dire hétéroclite, ré- side finalement dans les relations de l’Homme avec la Matière: éternel dialogue, dont le beau livre d’A. Leroi-Gourhan avait analysé les prémices préhistoriques, mais qui se prolonge de nos jours, et plus que jamais depuis que la modernité fabrique non seu-

1 umr: «unité mixte de recherche» regroupant le Centre National de la Recherche Scientifique (cnrs), l’Université Paul Valéry de Montpellier, le Service régional de l’Archéologie Languedoc-Roussillon et enfin l’Institut National de Recherches Archéologiques (inrap).


michel feugère

lement de nouveaux objets, mais aussi de nouveaux matériaux. Aujourd’hui comme hier, l’objet fabriqué, l’artefact, est riche d’informations qui nous parlent des hommes de son temps. Explorons ici quelques-unes des pistes de recherche qui s’offrent à nous.


Tout objet produit de la main de l’homme a bénéficié d’une technique: connaître cette technique, c’est d’abord s’intéresser à la connaissance du cerveau qui a produit l’objet. Au fil des années, puis des siècles, les savoirs se multiplient; à un certain moment, qui est probablement arrivé assez vite, aucun homme ne peut contenir tous les savoirs de l’humanité. La diversification des techniques est donc un processus constitutif de la so- ciabilité, un moteur puissant dans les relations entre les hommes et entre les cultures. Dans ses grandes lignes, l’histoire des techniques présente une évolution connue qui certes ne suit pas une courbe régulière, mais au contraire ondulante, avec des ar- rêts, des reprises et de spectaculaires accélérations. La technologie utilisée pour pro- duire un artefact, mais aussi celle qui a été ignorée ou contournée, inscrivent dans la matière un état connaissance, un instantané culturel. L’artisanat, grand absent des nombreuses «histoires des techniques» publiées depuis un siècle, se situe à l’interface entre techniques et technologies. Travailler à écrire cette histoire suppose avant toute chose de disposer de deux sé- ries d’outils: l’un, largement exploré par les spécialistes, est la caractérisation des ma- tériaux: pour repérer d’éventuelles évolutions, qu’elles soient liées à de nouvelles sources d’approvisionnement ou à de nouvelles techniques d’acquisition de la matière première, il faut disposer d’analyses, et notamment de celles qui traquent les compo- sants structurels du matériau. Ainsi, par exemple, les rapports isotopiques du plomb permettent de dessiner, en fonction de l’origine des minerais, de grands groupes géo- graphiques d’objets contenant du plomb: outre ce métal, le cuivre ou encore l’argent peuvent être concernés. La recherche sur les isotopes du plomb permet donc de clas- ser toute une série d’artefacts métalliques selon l’origine du minerai utilisé. Les pratiques métallurgiques, mais aussi les conditions de conservation des objets métalliques, pour rester dans le même domaine, peuvent transformer, voire oblitérer certaines des données en question. Alors que la teneur isotopique du plomb n’est pas affectée par ces processus (et c’est une des raisons de la fiabilité de ces analyses), d’au- tres métaux sont volatilisés au cours des fontes et refontes. D’autres encore peuvent migrer de façon préférentielle dans les couches de corrosion, et donc être éliminés au nettoyage: c’est le cas de l’étain, en particulier, dont la teneur dans la patine d’un bronze est supérieure à celle de l’alliage. Pour toutes ces raisons, il n’est pas forcément simple de comparer un minerai et le métal d’un objet fini, une argile naturelle et la pâte d’une céramique antique. En Gaule du Sud, notamment, on observe une contemporanéité presque parfaite entre les dates d’apparition de la métallurgie du fer, de l’argent et du plomb: tout sem- ble se passer entre la fin du viiie et le viie s. av. n. ère (Fig. 1). Argent et plomb sont as- sociés dans les minerais de plomb argentifère présents dans les gisements de la bor- dure Sud du Massif central: il est donc logique que les métallurgistes aient produit les deux métaux au même moment. Mais le lien avec la métallurgie du fer est plus ténu; c’est peut-être le simple développement de pratiques métallurgiques qui a entraîné,

techniques, productions, consommations: le sens des objets


a b

Fig. 1. Métallurgie précoce du plomb et de l’argent en Languedoc:

a, bouton en plomb de type Grand Bassin I de l’oppidum de Puech-Madame, à Poussan (F, 34); b, fibule hallstattienne en argent de Font-Mingaud à Florensac (F, 34). Ech. diverses.

parmi les artisans eux-mêmes, l’expérimentation et l’apprentissage de nouvelles tech- niques. Après la Méditerranée occidentale et la péninsule ibérique, la Gaule a pu ainsi découvrir un métal étroitement lié, à l’instar du monde grec, aux phénomènes de concentration du pouvoir et à la hiérarchisation des sociétés indigènes. Aux interrogations physico-chimiques, l’historien ajoute celle de la chronologie, couplée à des systèmes typologiques et culturels. Dans le domaine de la métallurgie comme dans d’autres, plus personne ne se satisfait aujourd’hui d’un modèle diffu- sionniste simple, qui permettrait de suivre les transferts de technologie à partir d’un Orient en perpétuelle avance sur l’Occident … Pour le moment, les chronologies dis- ponibles sur les plus anciennes mines de cuivre d’Europe (Rudna Glava en Serbie, Ai- Bunar en Bulgarie, Cabrières en France) dessinent plutôt un tableau dans lequel émer- gent, à des dates voisines mais sans qu’il y ait nécessairement de contact direct entre eux, plusieurs sites de production de cuivre à la fin du Néolithique. L’invention d’une nouvelle technique, quelle qu’elle soit, connaît souvent une pé- riode de latence avant d’être pleinement intégrée aux processus de production arti- sanale. Il en va ainsi, par exemple, du décor niellé, qui fait appel à un sulfure (mono-, bi- ou tri-métallique) utilisé pour sa couleur sombre, contrastant avec celle du support en argent ou en alliage de cuivre.1 Inventée au ier millénaire av. n. ère, peut-être sur les rives de la Mer Noire,2 cette technique ne connaît son véritable essor qu’au ier s. de n. ère, juste avant d’être abandonnée au profit de décors émaillés.3

1 Oddy et al. 1983.

2 Giumlia-Mair 2000; Giumlia-Mair-Rubinich 2002.

3 Deschler-Erb 2000.


michel feugère


Quand il maîtrise la caractérisation des artefacts, qu’il s’agisse du matériau ou de sa technologie, l’archéologue peut définir des séries d’objets pouvant correspondre à une production, c’est-à-dire issues d’un atelier ou de plusieurs unités productives liées par des rapports de proximité géographique, culturelle ou chronologique. La constitution de séries logiques peut s’inspirer des arborescences utilisées depuis le xviiie s. par les botanistes. Mais il faut se garder de tomber dans un système artificiel, une classification hiérarchique qui ne correspondrait pas à une véritable généalogie des productions. Plusieurs méthodes peuvent être utilisées. Historiquement, la recherche sur le mo- bilier est souvent partie des collections de musées, où les spécialistes étaient assurés de trouver ce qu’ils croyaient être des séries homogènes d’objets de même type. Mais ces collections, constituées en-dehors de préoccupations de contexte, et bien souvent de provenance, n’avaient qu’une apparence de cohérence. En rapprochant des arte- facts issus de milieux culturels différents, mais de forme voisine, ces rassemblements ont permis de mettre en évidence des communautés culturelles essentiellement fon- dées sur le partage de modèles stylistiques. L’analyse précise des diverses productions constituant de tels groupes devait faire appel à d’autres sources. C’est, d’abord, la multiplication des analyses physico-chimiques qui, en aidant au classement d’objets de même type en fonction de groupes de composition, a révélé la complexité insoupçonnée de séries d’aspect visuel proche, voire identique. La céra- mologie a largement développé les modèles géologiques et physiques permettant de reclasser de cette manière des lots chronologiquement homogènes: ainsi, par exem- ple, la céramique sigillée de Haltern, analysée dans les années 80: on a mieux compris, à partir de là, le rôle respectif des ateliers italiques et de leurs succursales lyonnaises dans l’approvisionnement en vaisselle de table des garnisons du limes rhénan.1 Dans le domaine des bronzes, les analyses se sont avérées plus délicates à interpréter, mais ont néanmoins donné des résultats utiles, notamment dans la connaissance des al- liages préférés pour chaque catégorie d’objets;2 les analyses de monnaies, de leur côté, permettent désormais de faire coïncider des variantes typologiques, dont l’impor- tance ne pouvait être appréciée, avec des variations d’approvisionnement en métal. Ainsi, au moins une variante des célèbres dupondii de Nîmes au crocodile fait appel au «cuivre gris», comportant des traces d’argent et d’antimoine, des gisements voisins du district de Cabrières.3 Par la suite, la fouille d’ateliers a entièrement renouvelé l’approche que les archéo- logues pouvaient avoir des étapes concrètes de la fabrication. Dans un atelier de po- tiers, on retrouve les bassins de stockage et de préparation de l’argile, les fours bien sûr, avec leurs dépotoirs de rebuts, mais aussi les hangars de séchage et, désormais, la trace des tours installés à proximité.4 La succession technique peut donc être recons- tituée dans le détail, même si les temps de travail, et bien sûr toutes les phases qui peu- vent se passer hors de l’atelier, ou celles qui ne laissent pas de traces archéologiques, demeure difficiles à apprécier.


Schnurbein 1982.

2 Condamin-Boucher 1966-1973.


Besombes-Barrandon 2001.

4 Desbat 2004.

techniques, productions, consommations: le sens des objets 25 Dans le domaine des bronzes, on a d’abord été surpris du caractère souvent mo- deste des installations retrouvées.1 L’atelier proprement dit, où l’artisan peut succes- sivement préparer ses modèles et ses moules, fondre l’alliage et le couler, puis dé- mouler, ébarber, polir les objets finis, peut n’être constitué que d’une seule pièce, ouverte sur une cour. Le stockage de l’argile, des éléments végétaux nécessaires au moules, du charbon de bois pour les foyers…, peut être renouvelé périodiquement et l’artisan n’a donc pas besoin de vastes locaux. Quantités d’artefacts, dont la plus grande partie des petits objets utilisés couramment dans les villes romaines, ont pu ainsi être fabriqués dans de tels ateliers: on en a retrouvé, généralement à la périphé- rie, dans presque toutes les villes romaines. Plus intéressante est l’apparition, dans au moins deux villes éduennes, Alésia et Au- tun, d’ateliers fonctionnant sous la forme d’un regroupement de petites unités cer- tainement complémentaires.2 On voit apparaître, dans ce type d’atelier-archipel, une certaine forme de collectivisation des intérêts qui n’est pas sans rappeler ce qui se passe dans les grands ateliers de fabrication de vaisselle sigillée, comme La Graufe- senque. Si les étapes de production restent contrôlées par chaque artisan dans son unité personnelle, il est clair qu’un échelon supplémentaire intervient pour la répar- tition des commandes et l’uniformisation des produits, sans doute aussi pour l’ache- minement des matières premières et l’expédition. Le stimulus est donné, dans le cas d’Alésia, au moins, par le commanditaire qui n’est autre que l’armée romaine, pour ses équipements de la cavalerie.3 L’atelier artisanal s’inscrit donc dans un réseau qu’il faut pouvoir apprécier dans son contexte: approvisionnement local, liés aux seuls besoins du voisinage; ou au contraire production de masse, destinée aux marchés qui ne sont accessibles que grâce au grand commerce international: Rome ou les zones frontières, pour l’essen- tiel. Au sein de ces différents réseaux, la typologie et le style des productions viennent généralement étayer la recherche de nouveaux marchés, et reflètent donc fidèlement les évolutions des sociétés contemporaines. Il faut dire un mot ici des outils d’analyse morphologique qui, grâce au croisement des approches évoquées ci-dessus, ont pu se développer ces dernières années. Le plus performant est sans doute le logiciel Syslat ® , développé à Lattes par M. Py et son équipe à partir des années 80.4 Ce système de documentation intégrée, accessible à des utilisateurs connectés en réseau, comprend désormais plusieurs dictionnaires de référence qui sont de véritables outils de recherche: ils permettent en effet, non seu- lement de confronter tout nouveau document à l’ensemble des formes connues à ce jour, mais aussi de créer au fur et à mesure les nouveaux types individualisés par les contributeurs. Le dictionnaire de céramiques, Dicocer ® , a déjà fait l’objet de deux li- vraisons papier.5 Il sera bientôt accompagné d’un dictionnaire des monnaies, Dico- mon ® , dont deux volets sont prévus (monnaies gauloises, puis romaines), et un dic- tionnaire des petits objets qui comprendra, lui aussi, plusieurs volumes. En favorisant une meilleure caractérisation des types, de même que l’exploitation statistique ou car- tographique des séries, de tels outils soutiennent à la fois la gestion documentaire et la recherche.


Par ex. Martin 1978.

2 Mangin 1981; Chardron-Picault-Pernot 1999.


Rabeisen 1993.


Py et al. 1997.


michel feugère

26 michel feugère Fig. 2. Tombeau C de Kastamonu, «Evkayası» (Turquie): vue générale de la façade

Fig. 2. Tombeau C de Kastamonu, «Evkayası» (Turquie): vue générale de la façade montrant les enjarrots (blocs provisoirement laissés en place devant le monument) sur le chantier abandonné au ive s. av. n. ère, et partiellement transformé à une date postérieure.

Enfin, le chantier de construction peut être considéré, quant à lui, comme un ate- lier dont le produit fini est le bâtiment, qu’il soit construit ou creusé, dans le cadre des structures rupestres. La recherche sur ce processus particulier peut réserver quelques surprises: ainsi, avec J.-Cl. Bessac, nous avons pu repérer et étudier en Paphlagonie, avec le tombeau C de Kastamonu, «Evkayası», un tombeau dont le creusement, sans doute interrompu par le décès prématuré du commanditaire, a été pratiquement abandonné vers la fin du ive s. av. n. ère, le monument étant ensuite achevé a minima pour fournir au moins une chambre funéraire1 (Fig. 2). L’occasion offerte par un tel monument est rare: c’est celle d’observer, figé au milieu de sa réalisation, une struc- ture dans un état qui n’est pas destiné à être vu, et que l’on ne peut donc que restituer dans les autres cas.2


La troisième manière d’aborder les techniques et les productions artisanales consiste à se placer du point de vue du marché, en examinant des lots de mobiliers homogènes déposés dans le sol à une date connue. Au sein de tels ensembles, les séries étudiées représentent une sélection de ce qui parvient dans la région à cette époque: la pre- mière réflexion à mener est d’ordre taphonomique, afin d’apprécier les critères qui

1 Leonhard 1915; von Gall 1966.

2 Bessac, à paraître.

techniques, productions, consommations: le sens des objets 27 ont pu influencer le tri des artefacts avant leur enfouissement. Ces ensembles ont pu être sélectionnés manuellement, certains éléments pouvant être prélevés en vue d’une réutilisation; la fragmentation peut révéler les traitements subis par le mobilier en cours d’utilisation, ou peu après; enfin, en milieu funéraire ou religieux, les arte- facts sont généralement sélectionnés en fonction d‘aspects symboliques, qui ne seront expliqués que dans un cadre d’analyse élargi. L’étude d’ensembles clos, même quand la date d’enfouissement est connue, ne peut donc pas être abordée avant une réflexion sur le mode de constitution de ces en- sembles. Un dépotoir domestique, par exemple, ne nous renseigne pas directement sur le faciès des céramiques en usage dans la région à une période donnée, comme on le lit trop souvent sous la plume d’archéologues. A la sélection liée à l’environne- ment (géographique, économique et culturel) s’ajoute le tri, beaucoup plus drastique, des gestes humains. Mais cette succession de filtres, loin de diminuer l’intérêt de tels ensembles, les désigne au contraire comme des révélateurs culturels particulièrement chargés de sens. L’étude d’ensembles clos sur les «sites de consommation», c’est-à-dire le point final de la destination des produits commercialisés, débouche donc sur des considérations variées. On peut par exemple, en comparant des ensembles de mobilier chronologi- quement successifs sur un même site, suivre les proportions relatives de différentes provenances pour un même marché, par exemple une forme de vase ou, mieux, un produit alimentaire. Dans ce domaine, les amphores restent à ce jour les conteneurs les mieux étudiés et les plus révélateurs des fluctuations du commerce des denrées commercialisées de cette manière. Car c’est évidemment, au premier chef, l’histoire du commerce international que peuvent alimenter les études de ce type, trop rares encore en Gaule. Depuis quelques années, la généralisation des bases de données de fouilles favorise l’établissement de séries de référence pour diverses catégories de mobilier, au premier rang desquelles la céramique désormais étudiée de manière quantitative grâce à aux systèmes de documentation automatisés mis au point sur de grands chantiers de fouilles. Le meilleur exemple de l’utilisation de ces systèmes en France est, à nouveau, fourni par les fouilles de Lattes, où la quantification des données, effectuée de manière systématique, concerne des volumes considérables1 (Fig. 3). La normalisation des sys- tèmes de comptage, et la généralisation d’outils comme Syslat ® , favorisent l’appari- tion de données solides et d’intérêt général, quand les tendances esquissées sur un gi- sement sont confirmées par les autres données de la même région. Ainsi, les données issues de la céramologie quittent le domaine de l’information ponctuelle pour ap- porter à l’historien des informations désormais fiables et d’intérêt général. L’objet fabriqué, l’artefact, apparaît donc en définitive comme le réceptacle d’in- formations issues de sources multiples: certaines reflètent, comme malgré lui, l’état des connaissances des artisans de son temps; d’autres traces sont imprimées par l’uti- lisateur. Ces deux couches, technique et anthropologique, forment comme des strates de sens qui se cumulent dans l’objet en contexte, celui dont les choix humains ont conditionné le dépôt dans le sol. Ces dimensions sont encore démultipliées dans le cadre de séries stratigraphiées à partir desquelles on peut observer non plus seule-

1 En dernier lieu Sanchez-Adroher 2004.


michel feugère

80000 70000 60000 50000 40000 30000 20000 10000 0 –525 –500 –475 –450 –425 –400
–525 –500 –475 –450 –425 –400 –375 –350 –325 –300 –275 –250 –225 –200 –175 –150 –125–100 –75 –50
a 15 12 9 6 3 0 –575 –550 –525 –500 –475 –450 –425 –400


5 4 3 2 1 0 –300 –275 –250 –225 –200 –175 –150 –125 –100


Fig. 3. Exemples de graphiques quantitatifs de données céramologiques pour le site de Lattes (F, Hérault): a, répartition chronologique du nombre de fragments de céramique décomptés dans les us antérieures à notre ère; b, pourcentages de fragments de céramique monochrome parmi les fragments de vaisselle, et comparaison avec l’arrière-pays du Languedoc oriental; c-d, pourcentages de céramique grise de la côte catalane au sein de la vaisselle (c, en nombre de fragments; d, en nombre de bords) (Py-Adroher-Sanchez 2001).

ment la situation à un moment donné, mais l’évolution des données sur une échelle chronologique. L’étude du mobilier archéologique, au sens large, ne peut se com- prendre qu’au regard de l’éclairage indispensable qu’elle apporte à l’approche d’un site ou d’une culture donnée.


Besombes-Barrandon 2001 = P.-A. Besombes, J.-N. Barrandon, Les dupondii de Nîmes: da- tation, diffusion et nature du métal utilisé, «Revue Numismatique», 2001, pp. 305-328. Bessac à paraître = J.-C. Bessac, Le travail de la pierre à Pétra, éd. erc. Chardron-Picault-Pernot 1999 = Un quartier antique d’artisanat métallurgique à Autun, P. Chardron-Picault, M. Pernot dir. (Doc. Arch. Fr.), Paris, 1999. Condamin-Boucher 1966-1973 = J. Condamin, S. Boucher, Recherches techniques sur des bronzes de Gaule romaine, «Gallia», 24, 1966, pp. 189-215; «Gallia», 25, 1967, pp. 153-168; «Gallia», 26, 1968, pp. 245-278; «Gallia», 31, 1973, pp. 157-183. Desbat 2004 = A. Desbat, Les tours de potiers antiques, in Le tournage, des origines à l’an Mil. Actes du colloque de Niederbronn, octobre 2003, M. Feugère, J.-C. Gérold dir. (Monogr. Instrumen- tum, 27), Montagnac, 2004, pp. 137-154.

techniques, productions, consommations: le sens des objets


Deschler-Erb 2000 = E. Deschler-Erb, Niellierung auf Buntmetall: ein Phänomen der frühen Kaiserzeit, «Kölner Jahrbuch», 33, 2000, pp. 383-396. Gall 1966 = H. von Gall, Die Paphlagonischen Felsgräber. Eine Studie zur kleinasiatischen Kunst- geschichte, (Istanbuler Mitteilungen, Beiheht 1), Tübingen, 1966. Giumlia-Mair 2000 = A. Giumlia-Mair, Solfuri metallici su oro, argento e leghe a base di Rame, in Atti della 6ª giornata “Le Scienze della Terra e l’Archeometria”, Este (Padova), Museo Nazionale Atestino, 26-27 febbraio 1999, Padova, 2000, pp. 135-141. Giumlia-Mair-Rubinich 2002 = Argento: il metallo della luna, in Le Arti di Efesto. Capolavori in metallo dalla Magna Grecia, a cura di A. Giumlia-Mair, M. Rubinich, Trieste, 2002, pp. 122-131. Leonhard 1915 = R. Leonhard, Paphlagonia. Reisen und Forschungen im nördlichen Kleinasien, Berlin, 1915. Mangin 1981 = M. Mangin, Un quartier de commerçants et d’artisans d’Alésia. Contribution à l’histoire de l’habitat urbain en Gaule, (Publications de l’Univ. de Dijon, lx), Paris, 1981. Martin 1978 = M. Martin, Römische Bronzegiesser in Augst BL, in Archäologie der Schweiz, 1, 1978, pp. 112-132. Oddy et al. 1983 = W. A. Oddy, M. Bimson, S. La Niece, The Composition of Niello Decoration on Gold, Silver and Bronze in the Antique and Mediaeval Periods, «Studies in Conservation», 28 (1), 1983, pp. 29-35. Py 1993 = Dictionnaire des céramiques antiques (viie s. av. n. è.-viie s. de n. è.) en Méditerranée nord- occidentale (Provence, Languedoc, Ampurdan), M. Py dir. (Lattara 6), Lattes 1993. Py et al. 1997 = M. Py, avec A. Adr oher, S. Barberan, R. Buxó, F. Conche, M. Feugère, A. Gardeisen, C. Jandot, J. Lopez, V. Mathieu, M. Monteil, H. Pomarèdes, St. Raux, V. Rinalducci, F. Souq, Syslat 3.1. Système d’information Archéologique. Manuel de ré- férence (Lattara 10), Lattes 1997. Py-Adroher-Sanchez 2001 = M. Py, A. M. Adroher Auroux, C. Sanchez, Dicocer (2). Corpus des céramiques de l’Age du Fer de Lattes (fouilles 1963-1999) (Lattara, 14), Lattes, 2001. Rabeisen 1993 = E. Rabeisen, Fourniture aux armées? Caractères et débouchés de la production d’équipements de cavalerie à Alésia au ier siècle ap. J.-C., in Militaires romains en Gaule civile, Y. Le Bohec dir. (Coll. cergr, ns 11), Lyon, 1993, pp. 51-71. Sanchez-Adroher 2004 = C. Sanchez, A. M. Adroher Auroux, La céramique du quartier 30-35. Evolutions, implications historiques et économiques, in Le quartier 30-35 de la ville de Lattara

(fin iiie-ier s. av. n. ère). Regards sur la vie urbaine à la fin de la protohistoire, M. Py dir. (Lattara, 17), Lattes, 2004, pp. 319-344. Schnurbein 1982 = S. von Schnurbein, Die unverzierte Terra Sigillata aus Haltern, Münster,



L’oggetto realizzato dall’uomo, il manufatto, offre pluristratificati livelli di informazioni utili per lo storico: le tecniche alle quali si ricorre per individuare i livelli di conoscenza dell’arti- giano, e più in particolare, la tecnologia utilizzata in quel preciso momento. Si acquisiscono, in questa maniera, informazioni su un singolo individuo o su gruppi sociali che consistono nell’integrare ogni singolo manufatto all’interno di un contesto funzionale, e dunque, in re- lazioni a specifiche ragioni d’utilizzo. Questa lettura ‘di superficie’ può variare, ovviamente, in funzione delle epoche e delle regioni ma anche, e soprattutto, delle culture. Ed è proprio all’interno di questi due approcci complementari che guardano alla creazione dell’oggetto e al suo utilizzo, che lo studio del manufatto archeologico trova la sua piena giustificazione.

L’objet fabriqué par l’homme, l’artefact, enregistre plusieurs strates d’informations utiles à l’historien: les techniques auquel on a recouru pour le mettre en forme témoignent directe-


michel feugère

ment des connaissances de l’artisan, et plus largement de la technologie de son temps. Mais son acquisition par un individu ou un groupe social, qui consiste à retirer l’objet du marché pour l’intégrer dans un contexte fonctionnel, s’effectue elle aussi en fonction de données spécifiques. Cette couche ‘superficielle’ de sens, qui peut varier en fonction des époques et des régions, donc des cultures, est tout aussi éclairante sur les sociétés humaines. C’est entre ces deux approches complémentaires, celle qui relève de la création de l’objet et celle qui découle de son usage, que l’étude des mobiliers archéologiques trouve sa pleine justification.


M aterial culture is constituted by artifactuality, that is, by agent-made proper- ties. But this holds for all aspects of culture, even its non-material aspects (e.g.,

music, oral literature, etc.). What is then constitutive for all agent-made objects – the artifacts? This question will now be discussed in depth. The archaeological record indicates that the making of artifacts has been a signifi- cant human activity for roughly 2.5 million years.1 By making artifacts we construct our social world.2 Settlements, buildings, clothes, art, body adornments, vehicles, food, music, and ritual performances are some examples without which human so- cial and cultural life, as it now is constituted, would have been impossible. As Peter Simons and Charles Dement have pointed out: “More ingenuity and creative energy is investigated in design, production, and application of artifacts than in any other field of human endeavor. We are surrounded by millions of artifacts, we have com- merce with them every day, many of us more than with other human beings. The lev- el of civilisation is literally measurable by the kind of artifacts of which a culture is capable, from the first palæolithic hand axe to the space shuttle and the supercom-


This description holds for the whole of the human cultural do main as such. To be a Homo sapiens is partly but significantly to be able to make, relate to, and use artifacts. Indeed, the human social reality with its cultural objects is inconceivable unless we consider its artifactuality.4 One could have expected that this significant fact of the matter would have at- tracted the theoretical interest among philosophers. For whatever reason, this is not the case.5 There are just a few philosophers that have discussed the nature of arti- factuality as a subject matter in its own right.6 In addition, various issues related to some received view of artifactuality have been discussed. For example, tool func- tions have been widely debated.7 And aestheticians have discussed whether there is an artifact condition for something to be a work of art.8 But for reasons I shall discuss below I do not find the currently available philosophical theories of artifacts entirely satisfying. Three crucial themes regarding the ontology of artifacts will be addressed in this article. The first theme concerns the nature of the stuff that composes artifacts. For example, Stephen Davies9 has argued that artifacts basically are material (i.e., solid,


Mithen 1998, pp. 16-25.


Dickie 1997a; Searle 1996; Simon 1996.


Simons-Dement 1996, p. 255.

4 Mithen 1998, pp. 24-25; Searle 1996; Simon 1996.


Simons-Dement 1996, p. 255.

6 See, e.g., Dipert 1986, Dipert 1993, Dipert 1995; Hilpinen 1992, Hilpinen 1993, Hilpinen 1995, Hilpinen 2004; Houkes-Vermaas 2003, Houkes-Vermaas 2004; Simons-Dement 1996; Thomasson 2003.

Dipert 1995; Houkes-Vermaas 2003, Houkes-Vermaas 2004; Preston 1998; Preston 2003; Searle




johan modée

three-dimensional) objects. I shall reject that view. The second theme concerns the production of artifact-making properties. Most theorists have argued that an agent’s internal ‘intention’ and subjective evaluation is conditional for the property of arti- factuality to exist.1 Following Beth Preston,2 I shall call this approach the intention- alist theory of artifacts. In contrast to this essentially internalist theory – which quan- tifies over possible inner, probably mental episodes of agents – I shall argue for an externalist-functionalist conception of artifactuality.3 Therefore, a third theme con- cerns the function of artifacts.4 I shall argue that if an object is an artifact there is at least one assembly component that determines the artifact’s proper function, given what this property actually may succeed in doing when considered in relation to a de- terminable or possible functional category.5 Arguing this, I emphasize the functional character of artifacts within the social context in which they are constituted. Given these three themes, hence, I shall advocate three theses:

(T1) Artifactuality is not necessarily confined to solid or concrete three-dimensional objects. (T2) Artifactuality is not necessarily dependent on a maker’s intentions and evaluations. (T3) The concept of artifact can be defined in externalist terms, focusing on an object’s agent- caused assembly components and proper function, as constituted in a social context.

The article has the following structure. Firstly I shall take a brief look at the lexical definitions of the concept and then state what I take to be the basic denotation for the concept ‘artifact’. Secondly, I shall argue that a theory of artifacts that takes artifacts to be solid, material objects is insufficient if we want to have a theory of artifacts cov- ering all kinds of agent-made things. Thirdly, I shall discuss and criticize Risto Hilpinen’s intentionalist theory of artifacts. Fourthly, I shall suggest an externalist def- inition of the artifact concept, in terms of agency and proper functions.

Word and Concept

To begin the analysis of the artifact concept, it is useful to take a brief look at the lex- ical definition of the word “artifact”.6 The Oxford English Dictionary (oed) gives for in- stance the following definitions of “artifact” (or “artefact”, in British English). First, we have the non-technical sense: “Anything made by human art and workmanship; an artificial product” (oed). As we can see, what is emphasized is the act of making. This is, according to the oed, the basic meaning of the word. Second, used within nat- ural science the word has a similar sense: “In technical and medical use, a product or effect that is not present in the natural state (of an organism, etc.) but occurs during or as a result of investigation or is brought about by some extraneous agency.” (oed) Here, again, the emphasis is on properties caused by agency. An artifact is not a natural object. Some of its properties are caused “by some extraneous agency,” as the oed puts it. Thus, considering the conceptual issue, it is reasonable to assume that a theoretical conception of ‘artifact’ should denote all made objects. Artifacts are not natural ob-

1 Bloom 1996; Dipert 1993; Hilpinen 2004; Houkes-Vermaas 2004; Simons-Dement 1996; Thomas-

son 2003.

2 Preston 2003.

3 Cf. Preston 2003.

4 See, e.g., Houkes-Vermaas 2003; Houkes-Vermaas 2004; Preston 1998; Preston 2003.

5 Cf. Millikan 1984; Millikan 1993; Simons-Dement 1996; Wright 1973.

outline of a new theory of artifacts 33 jects; they are artificial, caused by some sort of agent activity. The important thing here, as I see it, is a generic matter. It concerns how an object actually received its con- stitutive property.1 Most definitions of ‘artifact’ suggest also that agent activity causes artifacts to have their constitutive property. However, to avoid being “unwarrantedly anthropocentric,” as Simons and Dement writes, by ‘agent’ we could denote any kind of decision-making system, such as persons, organizations, animals, and – perhaps – even computers.2

The Narrow and Wide View of Artifacts

What kind of stuff constitutes artifacts? The narrow view of artifacts says that arti- facts are solid, concrete objects and not, for example, events. By contrast, the wide view says that an artifact is anything made by an agent, regardless of whether it is stuff. However, Stephen Davies has pointed out that the common-sense conception of artifacts usually involves some idea about stuffs.3 As we can see above, however, the lexical definitions in the oed are silent about this matter. But following common sense, artifacts are usually conceived in terms of being three-dimensional pieces of solid matter. This is also the case among some theorists. Indeed, within archaeology a conception of artifacts defined in terms of “material culture” is widely accepted.4 A similar narrow view can also be found among some philosophers. Artifacts are here seen as (more or less) solid objects. For instance, as Stephen Davies has argued:

Implicit in the primary sense [of “artifact”] is the idea that an artifact is manufactured via the direct manipulation of a material item, the existence of which predates the creation of the ar- tifact. This idea suggests in turn that the artifact must share at least some material properties with its progenitor, and hence, that artifacts are items of substance (and so to be distinguished from events, relations, attitudes, and so forth).5

On this view an artifact is a solid, three-dimensional, material product of an action by which a maker modifies the properties of some piece of concrete matter. However, it is strikingly clear that the domain of artificial products is much wider than what Davies’ intuitive account allows for. Consider, for example, musical works. Such works are not solid, three-dimension- al objects. They are rather some sort of events, at least when performed. But why not take them to be artifacts? They are, indeed, caused “by human art or workmanship,” i.e., in the way the oed defines “artifact.” In contrast to Davies, I think ordinary lan- guage could be improved here if we want to use ‘artifact’ as a technical concept for anything agent-made. What kind of ontology do we need for this approach? If physicalism and global supervenience, as metaphysical doctrines, are true of the actual world, the ontology of such artifacts is explainable as basically physical, with all non-material properties supervenient on that structure.6 For example, consider Homer’s original Iliad and Odyssey. Originally they were completely oral works, stored in the memories of the


Iseminger 1973.


Simons-Dement 1996, p. 258.


Davies 1991, p. 133.

4 Iseminger 1973, p. 10; Renfrew-Scarre 1998.


johan modée

bards and instantiated in a more or less improvised form when performed.1 Such oral works exist physically just as performed sound waves and as cognitive bits of infor- mation stored in brains. Nonetheless, despite this apparent immaterial character, they are unquestionably artificial works that exist physically in the actual world due to agent activity. Hence, it is natural to regard them as artifacts. On this wide, theoreti- cal view of artifacts we reject the common sense view that artifacts are material, sol- id, concrete three-dimensional objects. Instead, therefore, we can advocate the fol- lowing thesis: (T1) Artifactuality is not necessarily confined to solid or concrete three-dimensional objects. To accept (T1) is to subscribe to a wide view of what artifacts are. This is not an orig- inalviewwithinphilosophy.Forexample,asRistoHilpinenwrites:“Artifactsinthewide senseformanontologicallyheterogeneouscollection:someof themhaveinstances(lit- eraryworksandmusicalcompositions),othersaresingularobjects(e.g.,paintings),and there are also abstract artifacts, for example, fictional characters, which have authors but are neither concrete particulars nor have such particulars as instances”.2 This “wide sense” applies neatly within the field of aesthetics. For example, it con- forms to George Dickie’s view of artworks as artifacts.3 But it has a general theoret- ical value as well. In particular, a wide view of artifacts is useful for analyzing the products of diverse agent actions. For example, Andy Clark has called language “the ultimate artifact” and Hilpinen regards “belief systems” and certain words as arti- facts.4 In examples like these the stuff issue is of no particular interest, but only what is constitutive for artifactuality as such.

The Ontogenesis of Artifacts

I shall now turn to the second theme regarding the ontology of artifacts. The issue is

how to explain the basic ontology of artifacts qua agent-made or produced objects. A common view of artifacts within philosophy is that the alleged existence and at- tribution of mental events or dispositions called intentions play a more or less decisive

role if anything is to be an artifact. This view can be called the intentionalist theory of artifacts, because it holds that intentions are conditional for artifactuality. Significant theorists within the philosophy of artifacts – e.g., Randall Dipert and Risto Hilpinen

– have advocated this theory. On this approach the alleged existence of a state of af-

fairs that can be conceptualized as ‘intention’ serves as the nexus in the explanation of the relation between agent and artifact. Risto Hilpinen has presented his intentionalist theory on artifacts in a number of papers.5 A recurrent argument in all those papers is that if anything is an artifact it must be a product of the intentional action of at least one agent. This means that un- less an agent in some way or another mentally intended some artifact-making prop- erty P, this being recognized by the agent, the object would not have had P as a prop- erty. A more or less similar view is widely shared among most theorists.6


Ong 1982.

2 Hilpinen 2004.

3 Dickies 1997b.


Clark 1997, pp. 193-218; Hilpinen 1993, pp. 173-175; Hilpinen 1995.


Hilpinen 1992; Hilpinen 1993; Hilpinen 1995; Hilpinen 2004.

outline of a new theory of artifacts


Hilpinen s Theory

In this section I shall only focus on what I take to be the core of Hilpinen’s theory of artifacts and not provide a detailed overview of everything he says in addition to that core. Hilpinen has presented this theory in a number of papers. In all these papers, he has been quite consistent regarding his basic theory. Stating an intentionalist theory of artifacts, Hilpinen claims that if anything is a proper artifact, it must be a product of the intentional action of at least one agent. In his 1993 paper Hilpinen writes this condition as follows (note that Hilpinen uses “au- thor” and “authored” when talking about agency in relation to artifacts): “An object o is an artifact authored by A only if some properties of o depend on the content of A’s intentions”.1 In Hilpinen’s other papers, he states a similar condition. For instance, in his 1992 paper, he proposes “to apply the expression ‘artifact’ only to the intended products of actions”.2 And in Hilpinen’s latest contribution in Stanford Encyclopedia of Philosophy, we read that “[t]he existence and some of the properties of an artifact de- pend on an author’s intention to make an object of certain kind”.3 Hilpinen calls this analysis of the relation between artifact and author the Dependency Condition. How should we then assess this condition? In what way are intentions conditional for the ontology of artifacts? Hilpinen contends that the evaluation of the correlation or degree of fit between an intention and the resulting piece of artifactuality is cru- cial. In his latest contribution, he states this view as follows: “The study of artifacts (qua artifacts) is intrinsically evaluative, since viewing an object as an artifact means viewing it in light of intentions and purposes”.4 On this view, only if the agent’s evaluation is that the act of making actually has been successful, is it to be seen as a genuine or proper artifact. Otherwise, the object is only “scrap,” Hilpinen says.5 In his most recent version of this view,6 we find two conditions for anything to be an arti- fact in this evaluative sense. First, he states what he calls a Success Condition: “An object is an artifact made by an author only if it is satisfies some sortal description included in the author’s productive intention”.7 A crucial term for this condition is that the productive intention of the maker in- cludes a sortal description. What is a sortal description? In an earlier paper, Hilpinen gives this explanation: “A sortal description determines the identity of an object, and the criteria by which it can be distinguished from other objects (or things)”.8 The sortal descriptions Hilpinen talks about are descriptive labels such as (to use his own examples) “house,” “shirt,” “black pudding,” and “bourbon”.9 Particular qualities, such as “red,” “soft,” and the like, are not sortal descriptions.10 Thus, if I understand Hilpinen correctly, sortal descriptions cannot be too wide. ‘Tool’, for instance, is too wide since it applies to many different kinds of objects. So ‘tool’ does not determine the identity of a particular object. For example, if I want a hammer I cannot just ask for a tool – it is not precise enough; it does not determine the identity of the object I am asking for. Similarly, on my understanding of Hilpinen’s theory, if I



1993, p. 159.



1992, p. 60.

3 Hilpinen 2004.




5 Hilpinen 1992, pp. 61-63; Hilpinen 1993, pp. 161-163.


Hilpinen 2004.

7 Hilpinen 2004.

8 Hilpinen 1992, p. 61.



1992, p. 61.

10 Hilpinen 1992, p. 61.


johan modée

have made a hammer, I must have had the sortal description of a hammer in my pro- ductive intention if the Success Condition is to hold; again, this is because the wide cat-

egory ‘tool’ does not determine the identity of this sort of object. It is simply too wide, including not only hammers but also space shuttles and computers. So if the Success Condition is going to make any sense, the sortal descriptions must be very precise. But the Success Condition says not only that an object is an artifact if there was a plan of to make object o that included the sortal description of o, this being the content of

a productive intention, before object o was made, it says also that o actually must sat-

isfy the original plan of the sortal description, this plan being a part of the intention of the maker. Otherwise, the result is not an artifact. Secondly, therefore, Hilpinen also states what he calls an Acceptance Condition for anything to be an artifact: “An object is an artifact made by an author only if the author accepts it as satisfying some sortal description included in his productive intention.” If we take this condition as it stands, it says that only if the agent accepts object o as satisfying the plan of making o, the plan being the agent’s sortal description, object o is an artifact. Consequently, if the agent does not accept object o as satisfying the plan, o is not a genuine artifact – even if an agent makes its constitutive properties. In sum, Hilpinen analyzes the concept ‘artifact’ in terms of an intentionalist theo- ry, by which he claims that the properties constitutive for artifactuality depend on in- tentional actions of agents. This is called a Dependency Condition. Further, when ana- lyzing what constitutes genuine artifacts, he also claims that the relation between agent and artifact must be analyzed as intrinsically evaluative. Hilpinen analyzes this evaluative aspect in terms of a degree of fit between intention and product, and in terms of the agent’s evaluation of the product. This view is expressed by the Success and Acceptance conditions.

Against the Intentionalist Theory

I shall now argue that Hilpinen’s Dependency, Success, and Acceptance conditions are im-

plausible if they are taken as necessary conditions for every instance of artifactuality, including the proper or genuine artifacts Hilpinen wants to characterize. Doing this,

I have two objections against Hilpinen’s view. In the first objection, I argue that artists

and craftsmen can work on routine or habit and be hardly conscious of what they are doing.1 In such cases it makes no sense talking about intentions including sortal de- scriptions each time an object is made, allegedly resulting from conscious conceptions of sortal descriptions in each agent’s mind. But the result must be seen as an artifact anyway. If this is correct, we can reject the claim that ‘intentions’ including sortal de- scriptions are conditional for the making of artifacts. Granted, Hilpinen seems to rec- ognize that there are artifacts that are unintended. For instance, he writes:

[A]rtifact sortals can be essentially or nonessentially (accidentally) artifactual. For example, ‘motor controller’ and ‘paper clip’ are essentially artifactual terms, but a path through a for- est can be intentionally made (an artificial path) or it can be an unintended product of peo- ple’s habit of following the same route when they walk through the forest.2

1 Cf. Dipert 1993, pp. 49-50.

2 Hilpinen 2004, pp. 2-3.

outline of a new theory of artifacts


Apparently Hilpinen regards unintended forest paths as nonessentially (accidentally)

artifactual, belonging to a wide category of all sorts of artifact sortals. Moreover, giv- en his two conditions for the evaluative view of artifacts, he must at least deny that some forest paths are essentially artifactual; that is, they are not genuine artifacts since they are unintended. But, in his 1992 paper, Hilpinen proposes “to apply the expres- sion ‘artifact’ only to the intended products of actions”.1 His Dependency, Success, and Acceptance conditions point in a similar direction. As Hilpinen presents these condi- tions, it appears that he holds the view that some artifacts are more, so to speak, gen- uinely artifactual than others. And, as Hilpinen suggests, the strict concept of artifact should only be applied to “genuine” artifacts – that is, intended artifact sortals. It is hard to see how this view can be reconciled with Hilpinen’s view that some ar- tifact sortals might be unintended. Anyway, I shall argue that the distinction between essential and nonessential artifactual properties cannot be maintained even in cases that Hilpinen probably would take as instances of genuine artifacts (e.g., ‘motor con- troller’, ‘paper clip’). Given Hilpinen’s theoretical claim regarding ‘intention’ as con- ditional for essentially artifactual sortals (i.e., genuine artifacts), I have then the fol- lowing objection against his Dependency and Success conditions:

(a) Agents can produce genuine artifacts without having any intention including a

sortal description to make any particular sort of artifact. By (a), it is thus denied that we need to assume that agents necessarily must have had intentions if the productive results of their agency are genuine artifacts. In other

words, I argue that we can analyze what artifacts are without relying on the intention- alist theory. The second objection is stated against Hilpinen’s Acceptance Condition. Here I want to criticize the evaluative-individualist aspect of his intentionalist theory:

(b) An object can have its constitutive properties (genuinely) artifactually even if its

maker does not accept it as a successful production. By (b), it is denied that the agent’s positive evaluation is necessarily constitutive for artifacthood. If these two objections are sound, they establish my second thesis:

(T2) Artifactuality is not necessarily dependent on a maker’s intentions and evalu- ations. The important thing to note with (T2) is that I deny both (a) that intentions includ- ing sortal descriptions necessarily determine the ontological properties of artifacts (genuine or otherwise), and (b) that the maker’s evaluations of his or her work deter- mine whether or not a given object is an artifact. I shall now discuss these objections.

Against the Dependency and Success Conditions

How should we understand Hilpinen’s view of intentions and sortal descriptions? If I say that I have an intention, I suppose this must mean that I self-attribute myself some sort of conscious mental state which includes planning. But Hilpinen’s notion of ‘intention’ is not particularly clear. We therefore have to consider other literature. For example, John L. Austin explains ‘intention’ as follows:

1 Hilpinen 1992, p. 60.


johan modée

As I go through life, doing, as we suppose, one thing after another, I in general always have an idea – some idea, my idea, or picture, or notion, or conception –of what I’m up to, what I’m engaged in, what I’m about, or in general ‘what I’m doing.’ … I must be supposed to have as it were a plan, an operation – order or something of the kind on which I’m acting, which I am seeking to put into effect, carry out in action: only of course nothing necessarily or, usually, even faintly, so full blooded as a plan proper. When we draw attention to this aspect of action, we use the words connected with intention.1

Although this quotation does not make it entirely clear what intentions are, Austin’s tentative account indicates nonetheless the most important aspect of the notion. To have the intention to do something means having an idea or plan of what one would like to do. According to traditional philosophy, this idea or plan is the content of the propositional attitude or disposition that describes what I think I am going to do.2 I assume that this (or something similar) is what Hilpinen roughly must hold about the use of ‘intention’ in his Dependency and Success conditions. For, to be sure, the sor- tal descriptions Hilpinen talks about appear to be equivalent to some sort of detailed preconceived plan, which the agent, apparently, is consciously aware of. In other words: an object is a proper artifact only if some of its properties depend on the con- tent of the intentions of the maker and if the maker can affirm that this content ex- isted before the object was made. And this content consists in a sortal description, which determines the identity of the object. But is this internalist-intentionalist view really needed for analyzing what is onto- logically constitutive for artifacts? I do not think so. Consider for example doodles, the drawings we occasionally make, for instance while talking on the telephone but which even artists make.3 Such drawings are, I suppose, neither planned in any way, nor con- ditioned by any idea. Nonetheless I suppose that it is natural to regard them as arti- facts. But if an agent is making doodles while talking on the telephone, it is not the case that the agent necessarily intended the doodling, having the sortal description of ‘drawing’ in mind; it was just made while the agent was speaking, having other ideas in mind. It is thus far from obvious that the doodle’s being a drawing has properties dependent on some content of the maker’s intentions. On the contrary, it is plausible that it typically lacks that sort of dependency. If this is correct, there is not always a productive intention with the sortal description of a drawing in mind when we make doodles. Nonetheless, as an agent-made object, I think the doodle is best categorized as an artifact. Here are some reasons. Doodles, qua drawings, might be indistinguish- able from other drawings. That is, they can be representations of flowers, animals, and buildings. Further, doodles usually have a recognizable distinctive style of a particular agent – e.g., Picasso’s style.4 Consequently, doodles can even be of artistic quality, since their properties are dependent on the agent’s drawing skills. And most importantly, doodles are not to be seen as natural outcomes of natural processes; they have typi- cally a genuine artificial character. Thus, doodles fit the oed definition of ‘artifact’:

“Anything made by human art and workmanship; an artificial product.”

1 Austin 1966, p. 437.

2 Cf. Churchland 1981; Dennett 1987; Dennett 1995; Stich 1983; Quine 1960.

outline of a new theory of artifacts 39 Hilpinen must of course insist that doodles cannot be seen as genuine artifacts. But then he needs to explain why all doodles are not mere scrap or accidental artifacts, but rather something that belongs to the category of human art and workmanship. Now it can be argued that this feature of doodles is just a case of a general feature of artistic activities, or even of making in general. Consider Randall Dipert’s illustra- tive example:

Craftsmen, even artists, often work at great speed and with remarkably little conscious thought (apparently) at the time of creation. The average speech act requires a dizzying num- ber of conceptual choices and control of individual muscles, the vast majority of which are surely unconscious. Yet the products of such individuals, the sentences and the furniture they have made or the (even hack) musical compositions they have produced, are still widely con- sidered to be artifacts.1

This is a description applicable to many cases where artifacts actually are made. There are many cases of serial but manual production where the craftsman works on rou- tine, i.e., roughly in the same way that we make doodles. For instance, suppose there is a craftsman making a small figurine on routine based on fifty years of experience. Now, given the applicability of the Dependency and the Success conditions, the crafts- man needs to evoke a productive intention that includes a sortal description each time he makes a figurine that is similar to the thousands he has already made. Otherwise the product made is not a genuine artifact. Suppose then that one figurine is made without this actually happening; this appears to be possible given Dipert’s example. That is, on a certain occasion the craftsman has no sortal description in mind before he makes the figurine. Following Hilpinen’s theory the resulting figurine is not a gen- uine artifact even if it looks the same as the “genuine” ones he has already made. I think that this strange consequence of his theory is less appealing for a general theo- ry of artifacts. There are two ways of taking this problem seriously in the theory of artifacts. One way, which strikes me as being less fruitful, is to try to save the intentionalist claim by an inflation of possible mental notions, which could be applicable in every possible situation of making. Discussing his own example, quoted above, Dipert tries to do this by claiming that there might be mental states that are “half-intentions”.2 What this means is not particularly clear. Another way of taking the problem seriously, which I think is more fruitful, is to reject the assumption that we must use the inten- tionalist theory when analyzing what artifacts really are (genuine or otherwise). This way is illustrated by my initial example: the doodles. Arguably, choosing to take doo- dles as genuine artifacts implies that we can reject Hilpinen’s Dependency and Success conditions.

Against the Acceptance Condition

Hilpinen’s Acceptance Condition says that an object is a genuine or proper artifact made by an author only if the agent causing it accepts it as fitting some sortal description in- cluded in the intention to make it. This is an expression of Hilpinen’s view that the analysis of artifacts is necessarily evaluative, especially from the maker’s point of view.

1 Dipert 1993, pp. 49-50.

2 Dipert 1993, pp. 54-56.


johan modée

Again, I think that the condition is implausible. My objection, (b), is that an object can have its constitutive properties (genuinely) artifactually even if its maker does not ac- cept it as a successful production. Here is my argument. Recall first that a sortal description on Hilpinen’s account “determines the identity of the object, and the criteria by which it can be distin- guished from other objects”.1 That is, the sortal description cannot be too vague or wide: as Hilpinen says, it determines the identity of an object as a particular kind, not as a class of particular kinds. Consider now the following thought experiment about a possible prehistoric case, illustrating objection (b). Say that a Homo habilis named Bob, working in the prehistoric Oldowan pebble tool industries (c. 2 to 1.5 million years ago),2 has plans to make a certain artifact o that fits the sortal description D1. But be- cause Bob is using drugs (perhaps narcotic mushroom) and is hardly aware of his ac- tions, he is making an artifact that fits the sortal description D2 instead of that of D1. That is, instead of making a chopper c he makes a scraper s. When Bob eventually re- alizes this at a later time, he finds that his plan to make c did not succeed. If the plan had been to make s, however, the result must be seen as highly successful. But, for Bob, the product cannot be accepted as fitting the original plan at all. It failed in all identity-determining respects. Does this imply that s is not a genuine artifact? If we follow Hilpinen’s Acceptance Condition, the answer is apparently yes. This is because c, if evaluated in terms of sor- tal descriptions, has a different identity than s. Thus, on this restrictive account, s is not an artifact made by an author, only “scrap,” or, at best, accidentally artifactual. In contrast to Hilpinen, however, I think it clearly can be argued that s is a genuine arti- fact. At least, s will appear to be a genuine artifact because it has a proper function and its constitutive properties due to agent activity. What is the problem with Hilpinen’s view? We note how much emphasis Hilpinen puts on the role of the individual. The maker’s evaluations do not only determine a degree of fit, but also, in effect, whether or not an object actually is an artifact or not. But why must the individual maker’s evaluation determine whether or not the produced object is an artifact? If we consider the observers’ evaluations, the observers – especially if they subscribe to an intentionalist theory like Hilpinen’s – would of course interpret Bob’s product as the result of an intention to make s and not c. They would thus intuitively take s as an artifact resulting from the productive intention including D2 to make s, even if this was not Bob’s plan or idea at all. For, as Dipert says, our attributions of intentions might be “fanciful”.3 My point is thus that the observers interpret s in that way because s has certain artifactual properties. But for the object fitting sortal description s, which they might infer had been part of Bob’s original plan, it is not only irrelevant what mental content Bob – at t1 – actually might have (thought he) had but also irrelevant if Bob accepts the object made as fitting the sortal description he now – at t2 – conceives as a part of his orig- inal productive intention.4 This is because the object has its artifactual properties anyway, fitting sortal descriptions in the social, external sense, regardless of what Bob actually had in mind at t1.


Dipert 1993, p. 61.

2 See Mithen 1998, p. 17.


Hilpinen 1986, p. 402; Hilpinen 1993, p. 164.

4 Cf. Dipert 1986, p. 402.

outline of a new theory of artifacts 41 I think that there is still a problem even if we were to adopt a less restricted view of sortal descriptions (I am indebted to e-mail conversation with Hilpinen for the fol-

lowing example). Say that the sortal description just individuates Bob’s product as a tool while not completely determining its identity (i.e., either as chopper or scraper). If that is the case, Bob’s plan did not fail in all respects – he actually produced a tool. So the sortal description he originally had individuated the object as a tool. The prob- lem here is of course that Bob instead of making a tool could have produced an ob- ject individuated by a different sortal description than the one he originally had. For example, say that the result of his work appears to be a representation of a fish, not

a tool. It is useless as chopper or scraper but an excellent Homo habilis work of art.

Then Bob’s original sortal description did not individuate the object as a certain kind but he might be happy with the result anyway. That is, Bob might realize that his orig- inal idea failed in all its originally planned respects while accepting that the result was successful in some other object-individuating way. Consequently, not even a less re- strictive view of the Acceptance Condition will hold in all possible cases. Finally we must note that Hilpinen holds that an artifact might be “adopted and used for other ends”.1 This means, according to Hilpinen, that “the evaluation of an artifact need not be based on (and limited to) its author’s intention”.2 This view seems to presuppose that the adopted object already is an artifact, since Hilpinen says that unmodified natural objects (i.e., non-artifacts) can be adopted in the same way. So, as far as I can see, the possibility of adoption does not save the Acceptance Condition from the criticism stated above. If this criticism is correct, we can also reject Hilpinen’s Ac- ceptance Condition. A maker’s evaluation in terms of a degree of fit between plan and product is not necessarily constitutive for genuine artifactuality.

An Alternative Approach

I have just argued for the two negative theses. Thus, given (T1), artifactuality is not a property confined to solid or concrete three-dimensional objects. And if we accept (T2), artifactuality is not necessarily dependent on a maker’s intentions and evaluations. But now we must ask: What are artifacts? Addressing that question in what follows,

I shall argue for this thesis: (T3) The concept of artifact can be defined in externalist terms, focusing on an object’s agent-caused assembly components and proper func- tion, as constituted in a social context. The first thing emphasized by (T3) is that artifacts have agents that cause their con- stitutive properties. Necessarily, it can be argued, if anything is an artifact, an agent has manufactured its artifactual properties. Hence I agree with Hilpinen3 that artifacts in the wide sense are agent-made ob- jects, constituted by certain properties. In contrast to Hilpinen, however, I deny the individualist-intentionalist and evaluative theory by which he tries to explain this re- lation for genuine artifacts. Consequently I need an alternative account for determining all kinds of artifact sortals from non-artifacts. Therefore, a second thing emphasized by (T3) is that the


Hilpinen 1993, p. 164.

2 Hilpinen 1993, p. 164.


E. g., Hilpinen 1993; Hilpinen 2004.


johan modée

proper functions of artifacts are constituted in social contexts, where we consider the function of each object as determining its individuation (e.g., as tool) as well as its identity (e.g., as hammer).

Agent Causation

Following what Gary Iseminger1 has argued, if an agent makes an artifact, we can say that the agent is responsible for the mere existence of its artifactual properties. As Iseminger points out, this sort of responsibility should of course not be seen as hav- ing any moral aspects.2 “Responsible” is here only to be taken in the generic, causal sense. Thus, we can say that agent a is responsible for b if and only if a caused the con- stitutive properties of b. For example, we can say that Pablo Picasso is responsible for Les Demoiselles d’Avignon if and only if he caused the artifactual properties constitutive of that painting. This talk about responsibility points to a causal relation between agents and certain objects. Hilpinen would of course insist that intentions make an agent responsible for having caused artifactual properties. But as we have seen, apart from the fact that it is unclear what intentions are, there are possible cases of artifactuality where the in- tentionalist theory does not apply – for instance in the case of doodles and unintend- ed forest paths. What holds universally, however, is that if there is an artifact then there is a causal link between an agent’s activity and the object’s artifactual properties. The link is the ontogenesis by which an artifact receives its constitutive properties. As Iseminger ar- gues: “Being an artifact is not a property that a thing presents either to sense perception or to ‘aesthetic perception;’ it is rather a matter of its history, how it came to have cer- tain properties”.3 On this view, the historically relevant ontogenesis is thus the fact that an agent caused the object (a material object, a performance, or a story or tune conceived in one’s brain) to have some of its properties artifactually. This is to subscribe to what Iseminger calls an adverbial theory, regarding what is constitutive of artifactuality. The focus is on what is constitutive of having properties artifactually. This is not necessarily a question of whether or not we actually can perceive that an object is an artifact (that is, it is not an epistemological issue of how we discover certain properties). But it is necessarily a matter of causation and origin. That is, the focus is on a procedural, historical matter rather than on epistemic considerations. What properties are crucial for the focus of this analysis? It is clear that not just a sin- gle artificial property always makes an object into an artifact as a whole. If I carve my name on a tree I do not make the whole tree into an artifact. The artifact is just the carving, not the tree as a whole. The same holds, of course, for the universe as a whole. Just because I have made some artifactual properties on earth, I have not made the whole earth or the entire universe into an artifact. So in order to have a reason- able view of what kind of objects are artifacts due to a causal history, let us therefore suppose – following Simons and Dement – that an artifact qua object consists of its assembly components.4 For an object to have artifactuality as a property (or to have




2 Iseminger 1973, p. 6.


Iseminger 1973, p. 9.

4 Simons-Dement 1996, pp. 260-264.

outline of a new theory of artifacts


properties artifactually) is to have its assembly components qua a certain object caused by agent activity. For example, a particular token of a car has a finite list of components, which is constitutive for this kind of object. The agent, making a car, caused just these com- ponents for the car to come into existence as artifact. These components do not in- clude other arbitrary components, such as the entire universe or Mount Everest. So the car qua car is not constitutively the assembly components for ‘car’ plus Mount Everest or the entire universe, because there is a possible world where the entire uni- verse or Mount Everest might cease to exist, apart from the particular token of a car. Thus it is perfectly possible to talk about a finite list of assembly components of a par- ticular object. These components are constitutive for a particular object qua that kind of object. Artifacts have these components caused by agents.1

Proper Function and Functional Categories

Now it is obvious that mere agent causation of certain components is not sufficient for explaining artifactuality as property. In a trivial sense of ‘production’, we can say that agents cause biological products, such as sweat and other outpourings, or foot- prints in snow while walking. But such products are not (necessarily) artifacts. Hilpinen’s way of dealing with this problem is his intentionalist theory: the agent has a sortal description in mind and evaluates the degree of fit between this content and the result of action. By contrast, I argue that we can reject the intentionalist theory and its individual- ist-evaluative account of making. So instead of this theory – but given agency causa- tion, which is the procedural history of any given artifact – I want to argue that pos- sible proper functions,2 given external functional categories, determine the identity of artifact kinds. This view suggests itself quite naturally. Artifacts appear always in some context that we – at least in principle – could call a social context (this holds even for Robinson Crusoe). As Davies writes:

The members of all cultures always have engaged in storytelling, drawing, carving and whit- tling, song, dance, and acting or mime. Frequently these activities are tied to social functions, such as the production of tools, the enactment of ritual, the preservation of a historical record. Their pervasiveness suggests that they are integral, not incidental, to the social ends they serve.3

Davies points here to the fact that artifacts always appear in social contexts where they have some sort of function – as tool, ritual, or storytelling – given certain social ends. Thus we can say that for an artifact to serve a certain social end, it needs to have a proper function by means of its intrinsic, identity-determining properties. This view could be compared with Hilpinen’s analysis. Explaining his view of sor- tal descriptions, Hilpinen suggests that “artifact kinds can … be defined by mass terms, e.g., ‘bourbon’ or ‘black pudding’”.4 I agree. “Bourbon” and “black puddings” are terms that apply to certain artifacts that have a certain composition of assembly components as well as certain functions, given a social context. But in contrast to

1 For further discussion, see Iseminger 1973; Simons-Dement 1996.

2 Millikan 1984; Millikan 1993.


Davies 2000, p. 199.

4 Hilpinen 1992, p. 61.


johan modée

Hilpinen’s intentionalist theory, and following Larry Wright’s influential article on functions,1 I want to argue that artifacts as kinds are determined by certain functions

and not by intentions preceding their production. Wright’s basic analysis of function is quite simple. Say that an object X has a func- tion that determines it to be Z. On Wright’s analysis, “[t]he function of X is Z means


X is there because it does Z.


Z is a consequence (or result) of X’s being there”.2

This definition of functions does not rely on an intentionalist theory.3 Instead, it analyzes function as an ontological property, intrinsic to each object. I shall now try to explain why I find Wright’s analysis to be of great value in the theory of artifacts. For my analysis, it is interesting to note that Wright points out that his analysis is com- patible with any kind of agent causation:

The differentiating feature is merely the sort of reason appropriate in either case: specifically, whether a conscious agent was involved or no. But in functional-explanatory context … the difference is minimal. When we explain the presence or existence of X by appeal to conse- quence Z, the overriding consideration is that Z must be or create conditions conducive to the survival or maintenance of X.3

Thus, rejecting the use of ‘intentions’ and individual evaluations in the theory of ar- tifacts, I suggest that we should look for the possible proper function of the things we classify as artifacts. As Wright says, when we analyze the function of a given object, “the overriding consideration is that [the function] must be or create conditions con- ducive to the survival or maintenance” of the object.4 The difference between artifacts and non-artifacts is of course partly, as Iseminger emphasizes,5 causal-historic. Artifacts are results of agent action. An artifact has its as- sembly properties caused by agents. But, in addition, it can be argued that the dif fer- ence also consists in the functions of artifacts. Artifact sortals are determinable in terms of various functions, given the possible functional categories. Let me briefly flesh out this view. Following Wright’s basic analysis, we have an agent-made object X because it does a certain Z. For example, we have a novel because it presents a fic- tional story. That fictional story is a consequence (or result) of that particular novel’s being there. And in this case, ‘being there’ is explainable in terms of certain agent cau- sation, i.e., writing. Another example: we have a doodle because it does present a draw- ing on a surface, usually a piece of paper. That drawing is a consequence (or result) of that particular doodle’s being there, which is explainable in terms of certain agent causation, i.e., drawing while talking on the telephone. Now it is obvious that this theory of function is silent on matters of purpose, in- tention, appreciation, dispositions, etc. But it does allow for an analysis in terms of evaluation. As Simons and Dement say, “[t]he level of civilisation is literally measur- able by the kind of artifacts of which a culture is capable”.7 Thus, in a certain sense, I agree with Hilpinen that a theory of the artificial should assess an evaluative aspect of artifacts.8 However, this evaluation does not, I argue,


Hilpinen 1973.

2 Wright 1973, p. 161.


For a contrasting view, see Searle 1996, pp. 16-18.

4 Wright 1973, p. 164.


Wright 1973, p. 164.

6 Wright 1973.

outline of a new theory of artifacts 45 consist in assessing the degree of fit between the maker’s alleged intentions and the resulting product, but in considering what proper function a particular agent-made object actually might establish due to its assembly components. Here is my argument for this view. Hilpinen points out1 that it is typical in all cases of artifactuality that artifacts can be classified in terms of sortal descriptions or mass terms, which all have a possible function given some functional category, e.g., concep- tualized as ‘car’ or ‘work of art’. That is, they appear in an agent-related social context (whether or not the agent is a human being). In my view, it is this external aspect of artifactuality that we should evaluate, classifying unsuccessful artifacts from success- ful ones, and artifacts from non-artifacts, by assessing a possible proper function. This assessment is not dependent on any human being’s point of view. A car has the proper function as a car independently of its users, as much as a tree ontologically is a tree independently of what any human being might think about this. But, qua an artifact, a car can be evaluated with respect to its proper function in a given context of use. Simons and Dement also point to function as an ontological aspect of artifactual- ity. On their analysis, apart from the fact that artifacts are composed of “assembly components”, they have also “functional components”.2 This is a plausible view. For example, an oil painting can be composed of paint, canvas, and frame. These parts of the painting are its assembly components. In addition, its functional component is the “purpose, office, or function in the working of the artifact”.3 Hence the oil painting, objectively, may or may not succeed as a work of art. At least, its functional compo- nents yield this possibility. Following Ruth Millikan, I want to describe these functional components in terms of “proper functions” of artifacts. This sort of function is what a thing can cause im- personally or objectively, what the thing actually may succeed in doing.4 Now Mil- likan provides several distinctions of proper functions. But I do not need such a fine- grained analysis for the present argument. The interesting question for my analysis is just what an artifact may actually succeed in doing by establishing a possible area of use, because the production and reproduction of certain assembly components estab- lishes or conforms to certain possible functional categories. For example, an object can, due to its constitutive properties, succeed in doing what it takes to be a possible work of art, tool, or representation. Functional categories are thus analytical categories, framing kinds of artifacts in terms of what proper functions they may have. For example, the conceptual category ‘car’ contains only such artifacts that have the assembly components that are constitutive of the artifactual kind we conceptualize as ‘car’ due to its function in a given context. It is in terms of these functional cate- gories we can also assess malfunctioning or less successful artifacts, which nonethe- less are artifacts with respect to a range of their possible proper functions. Given Wright’s theory we can evaluate artifact X in terms of whether or not it succeeds in achieving a social-functional category Z, so that Z “must be or create conditions con- ducive to the survival or maintenance of X”.5 For example, a malfunctioning cd is still an artifact, even if it is unplayable by my cd player, because, as a cd, it is an artifact


E. g., Hilpinen 1992, Hilpinen 1993.

2 Simons-Dement 1996, pp. 263-264.


Simons-Dement 1996, p. 264.

4 Millikan 1984, pp. 28-31; Millikan 1993.


johan modée

that should have had the proper function of transmitting a recording. If the cd is mal- functioning, I can evaluate it in such terms. This holds for artifacts, since they can be evaluated in terms of how successful they are with respect to a certain function of an area of use. For example, we can consid- er an agent-made object that is presented as a candidate for being a work of art and evaluate its possible status in the artworld. In other words, we could check if the ob- ject’s function in that particular context provides the conditions for the survival or maintenance of the object as a work of art.1 If it does not succeed in that respect, we could nonetheless consider its property-determined function; for instance, as a can- didate to be a work of art; as an unsuccessful work of art; as craft, etc. In any case, we can say that the object’s artifact kind is determined by its (ideal) proper function, given its assembly components and possible functional categories as constituted in a social context – i.e., the artworld. This example has a general applicability. For any ar- tifact, we can say that it exemplifies a possible functional category. Given contextual evaluation, we can distinguish artifacts from nonartifacts, even if both are products caused by agent activity. For example, we can distinguish ‘scrap’ from ‘artifact’. The former concept denotes simply mere by-products of agent action; the latter has some sort of proper function.

Defining artifact

We have now arrived at a position that enables us to suggest a definition of the arti- fact concept. This definition will be at odds with the intentionalist theory of genuine artifacts, since it includes all kinds of artifact sortals. Given that the theory just out- lined is sound, we can define ‘artifact’ in the following way. An object is an artifact if and only if the following two conditions apply: (1) Agent activity causes the assembly components of the object. (2) The object has a proper function given a functional cat- egory within a social context. Or, in other words, an artifact is an object that has its assembly properties caused by an agent and has a possible area of use in some social context. This explains what it is to have properties artifactually and how certain objects have identities as artifact sortals. Now, to prevent some possible misunderstandings of this definition, I want to em- phasize the following remarks. First, in (1), “causes” should be understood in terms of physical causality with respect to structurally constitutive assembly components.2 That is, an agent cannot make an artifact of a natural object just by pointing to it while saying that it is some sort of artifact with a proper function. As George Dickie says, artifactuality cannot be merely conferred upon an object; it involves some sort of ma- nipulating action.3 And, as I have indicated above, the identity of a particular artifact consists in its assembly components that determine proper function qua a particular object, e.g., qua building, bottle, sword, etc. Second, regarding (2), it must be emphasized that possible functional categories need not be established before particular artifacts can have proper functions. Instead, artifacts might instantiate such categories, exemplifying a range of possible functions,


Dickie 1997b.

2 Cf. Davies 1991, 133.

3 Dickie 1997b, pp. 44-46.

outline of a new theory of artifacts 47 which then can be evaluated in terms of their contextual functionality. This leaves room for the invention of novel artifacts.1 Third, note also that the definition is independent of epistemic aspects: it defines what artifacts ontologically are. Thus, without access to the appropriate context, we do not always know the exact function of the object. We know only, perhaps, that it might have had some possible functions. This is, for instance, the case with many ar- chaeological findings. But even if we have no clue whatsoever as to whether a given object is an artifact or not, it can nonetheless be argued that my definition frames the

constitu tive prop erties of artifacts qua arti facts as an ontological category. To ex em- plify, a particular washing machine is an artifact since it is an agent-made product with

a certain proper function, fitting the possible functional category of washing ma-

chines in a social context. This property is independent of whether or not it was in-

tended and whether or not I am in the epistemic situation to grasp the object as a washing machine. Ontologically, at any given time t, it is nonetheless possible for the object to be evaluated in terms of its proper function, given the functional category

it exemplifies in a possible social context. Finally, the definition is consistent with the oed definition of the word “artifact”:

“Anything made by human art and workmanship; an artificial product.” For by (1) I

emphasize the role of agency and by (2) I emphasize what oed explains in terms of “art and workmanship” and “artificial product.” Artifacts have proper functions given some possible functional categories. Not any kind of product is an artifact. For example, if I just “breathe in and out half as fast as usual for ten times” (this example

is due to an anonymous critic), the definition does not entail that I thereby have made

an artifact. For, if seen in isolation, this kind of property cannot be pigeonholed into any functional artifact category. Suppose, however, that the breathing inflates a balloon, or that it is part of a musical work like György Ligeti’s Aventures.2 Then it is included among an object’s artifactual assembly components with a proper function, given the functional categories of toys and music. In other words, we must consider what assembly components must be at hand when objects are qualified qua particular objects with certain proper functions. And I have argued that only such objects are, in Hilpinen’s words, “genuine artifacts”.3


Austin 1966 = J. L. Austin, Three Ways of Spilling Ink, «The Philosophical Review», 75, 1966, pp. 427-440. Bloom 1996 = P. Bloom, Intention, History, and Artifact Concepts, «Cognition», 60, 1996, pp. 1-29. Churchland 1981 = P. M. Churchland, Eliminative Materialism and the Propositional Atti- tudes, «Journal of Philosophy», 78, 1981, pp. 67-90. Clark 1997 = A. Clark, Being There: Putting Brain, Body, and World Together Again, Cambridge, ma, 1997. Davies 1991a = S. Davies, Definitions of Art, Ithaca-London, 1991. Davies 1991b = S. Davies, Non-Western Art and Art’s Definition, in Theories of Art Today, ed. N. Carroll, Madison, 1991, pp. 199-216.


Cf. Preston 2003; Wright 1973.

2 (1962).


Hilpinen 1992, pp. 61-63; Hilpinen 1993, pp. 160-163.


johan modée

Dickie 1997a = G. Dickie, Art: Function or Procedure. Nature or Culture?, «The Journal of Aes- thetics and Art Criticism», 55, 1997, pp. 19-28. Dickie 1997b = G. Dickie, The Art Circle: A Theory of Art, Evanston, 1997. Dipert 1986 = R. Dipert, Art, Artifacts, and Regarded Intentions, «American Philosophical Quarterly», 23, 1986, pp. 401-408. Dipert 1993 = R. Dipert, Artifacts, Art Works, and Agency, Philadelphia, 1993. Dipert 1995 = R. Dipert, Some Issues in the Theory of Artifacts: Defining Artifact and Related Notions, «Monist», 78, 1995, pp. 119-135. Gombrich 1960 = E. H. Gombrich, Art and Illusion: A Study in the Psychology of Pictorial Representation, New York, 1960. Hilpinen 1992 = R. Hilpinen, Artifacts and Works of Art, «Theoria», 58, 1992, pp. 58-82. Hilpinen 1993 = R. Hilpinen, Authors and Artifacts, «Proceedings of the Aristotelian Socie- ty», 93, 1993, pp. 155-178. Hilpinen 1995 = R. Hilpinen, Belief Systems as Artifacts, «Monist», 78, 1995, pp. 136-155. Hilpinen 2004 = R. Hilpinen, Artifact, in The Stanford Encyclopedia of Philosophy, ed. E. Zalta (Fall 2004 Edition). url = < 2004/entries/artifact> (January 10, 2007). Houkes-Vermaas 2003 = W. Houkes, P. Vermaas, Ascribing Functions to Technical Artifacts:

A Challenge to Etiological Accounts of Functions, «British Journal for the Philosophy of Sci- ence», 54, 2003, pp. 261-289. Houkes-Vermaas 2004 = W. Houkes, P. Vermaas, Actions Versus Functions: A Plea for an Alternative Metaphysics of Artifacts, «MonIst», 87, 2004, pp. 52-71. Iseminger 1973 = G. Iseminger, The Work of Art as Artifact, «British Journal of Aesthetics», 13, 1973, pp. 3-15.

Jackson 1988 = F. Jackson, From Metaphysics to Ethics: A Defence of Conceptual Analysis, Oxford,


Millikan 1984 = R. G. Millikan, Language, Thought, and Other Biological Categories: New Foundations for Realism, Cambridge, ma, and London, 1984. Millikan 1993 = R. G. Millikan, White Queen Psychology and Other Essays for Alice. Cam- bridge, ma, 1993. Mithen 1998 = S. Mithen, The Prehistory of Mind: A Search for the Origins of Art, Religion and Science, London, 1998. Artefact, n. and a, in Oxford English Dictionary, 2nd edition, oed Online, Oxford 1989, url = (January 10, 2007). Ong 1982 = W. J. Ong, Orality and Literacy: The Technologizing of the Word, London, 1982. Petroski 1992 = H. Petroski, The Evolution of Artifacts, «American Scientist», 80, 1992, pp.


Preston 1998 = B. Preston, Why is a Wing Like a Spoon? A Pluralist Theory of Function, «Jour- nal of Philosophy», 69, 1998, pp. 215-254. Preston 2003 = B. Preston, Of Marigold Beer. A Reply to Vermaas and Houkes, «British Journal

for the Philosophy of Science», 54, 2003, pp. 601-612. Quine 1960 = W. V. O. Quine, Word and Object, Cambridge, ma, 1960. Cognition and Material Culture: The Archaeology of Symbolic Storage, eds. C. Renfrew, C. Scarre, Cambridge, 1998. Searle 1996 = J. Searle, The Construction of Social Reality, London, 1996. Simon 1996 = H. Simon, The Sciences of the Artificial, 3rd edition, Cambridge, ma, and London,


Simons-Dement = P. Simons, C. Dement, Aspects of the Mereology of Artifacts, in Formal On- tology, eds. R. Poli, P. Simons, Dordrecht-Boston-London, 1996, pp. 255-276. Stich 1983 = S. P. Stich, From Folk Psychology to Cognitive Science: The Case Against Belief, Cambridge, MA, and London, 1983.

outline of a new theory of artifacts


Thomasson 1999 = A. Thomasson, Fiction and Metaphysics, Cambridge, 1999. Thomasson 2003 = A. Thomasson, Realism and Human Kinds, «Philosophy and Phenome- nological Research», 57, 2003, pp. 580-609. Wright 1973 = L. Wright, Functions, «The Philosophical Review», 82, 1973, pp. 137-168.


Questo contributo mira ad illustrare una nuova teoria nell’interpretazione del manufatto. Sco- po è quello di rigettare teorie precedenti che affermano che il manufatto è solido, è insomma un oggetto materiale. Inoltre, è anche rigettata la teoria “internalista”. Invece, l’obiettivo di questo contributo è quello di riflettere sulla effettiva realizzazione di un oggetto che è possi- bile anche se un artefice di un particolare oggetto non ha alcuna determinata e specifica in- tenzione. Una nuova proposta interpretativa viene, dunque, presentata. Essa suggerisce che il concetto di manufatto può essere definito in termini “esternalisti”, focalizzando l’attenzione sui motivi che stanno alla base della realizzazione di uno specifico manufatto e dunque, sulle sue funzioni, all’interno del contesto sociale di utilizzo.

This article argues for a new theory of artifacts. Thereby, other theories are rejected. First, the article rejects a narrow theory of artifacts, which claims that artifacts are solid, material ob- jects. Second, the internalist theory of artifacts, which holds that intentions are constitutive for artifacts, is rejected. Instead, this article argues that artifactuality of objects is possible even if a maker of a particular artifact do not have any determined intentional content. Third, a new theory of artifacts is outlined. This theory suggests that the concept of artifact can be de- fined in externalist terms, focusing on an object’s agent-caused assembly components and proper function, as constituted in a social context.


Enrico Giannichedda

Lo scavo

Q uando sono stato invitato a contribuire al primo numero di «Facta» avevo aderi- to con convinzione immaginando di scrivere alcune pagine sui residui, così da ri-

prendere alcune riflessioni personali già svolte in altra sede, ma tuttora inedite. Ri- flessioni relative non tanto ai modi di discriminare reperti in fase e residui, ma, più in generale, alla vita dei manufatti e in gran parte conseguenti alla lettura degli Atti di un convegno tenutosi a Roma il 16 marzo 1996. In tale sede, sia le relazioni sia gli in- terventi alla tavola rotonda conclusiva avevano adeguatamente affrontato un tema mai in precedenza approfondito ridefinendo il «problema residualità» con particolare attenzione per gli aspetti cronologici e di quantificazione dei reperti. Con il passare del tempo, però, ogni qualvolta iniziavo a lavorare sul tema che mi ero autoimposto, provavo un’insoddisfazione di fondo conseguenza di considerazioni più generali – non si può discutere dei residui senza discutere dei metodi della ricerca stratigrafica – o di domande a cui non ho sicura risposta. Prima fra tutte, i metodi della ricerca ar- cheologica sono adeguati per fare sì che ogni scavo diventi oggetto di interpretazio- ne storica? Oppure, quegli stessi mezzi sono di parziale ostacolo a ‘fare storia’ e, ma- gari, finiscono per servire ad altro?1 Con queste perplessità in testa, nell’anno accademico 2006-2007, ho insegnato alla Scuola di specializzazione dell’Università Cattolica di Milano, corso di Metodologia e tecnica dello scavo cercando, credo onestamente, di porre di fronte agli specializzandi, quasi tutti con ampia e diversificata esperienza di scavo, il dilemma se i metodi dello scavo archeologico sono adeguati, se sono migliorabili e come. Nonostante il corso iniziasse rilevando che la bibliografia di riferimento contempla testi scritti oltre ven- ticinque anni fa (in Italia i manuali celeberrimi di Harris, Barker e Carandini) e che a tali lavori quasi nulla aggiungono le opere successive che in genere compendiano quanto sopra, con un qualche personale stupore ho registrato generale soddisfazio- ne per i metodi adottati nelle più svariate situazioni.2 Per gran parte degli specializ- zandi i problemi «dello scavo» sono conseguenza dell’assenza di soldi, di tempo, tal- volta di preparazione individuale. Un problema, quest’ultimo, che ognuno dichiara

1 Di seguito, le indicazioni bibliografiche sono ridotte al minimo indispensabile anche con lo scopo di

sottolineare come da tempo si disponga di lavori che adeguatamente inquadrano i problemi qui posti. Sul ciclo di vita dei manufatti rinvio a quanto in Giannichedda 2006 mentre gli atti del convegno romano sa-

ranno ora citati complessivamente come I materiali residui 1998. Dedicato ai residui e al momento inedito è Giannichedda in stampa.

2 Il riferimento è a Barker 1981, Carandini 1981, Harris 1983, Piuzzi 1990, Santoro 1997, Tron-

chetti 2003. Un approccio diverso e francamente geoarcheologico è caratteristico di Leonardi 1982 e dei contributi in Leonardi 1992. A tali lavori e alla bibliografia ivi citata si rinvia anche per le considerazioni, più avanti soltanto accennate, sui modi di formazione.


enrico giannichedda

di volere risolvere facendo maggiore esperienza sul campo. La situazione è, però, dav-

vero questa o, forse, il problema dell’adeguatezza dei metodi alle situazioni reali (e quindi alla persistente e oggettivamente irrisolvibile carenza di soldi e tempo) per- marrà nonostante gli sforzi di ciascuno? Non è evidentemente questa la sede per una estesa discussione di un tema su cui, credo necessiti lavorare molto, ma può essere utile richiamare le questioni che reto- ricamente ho posto agli studenti e le direzioni verso cui si dovrebbe procedere per tro- vare quelle soluzioni che facciano dello scavo una ancora migliore occasione di rico- struzione storica. Sia comunque chiaro che il paradigma di riferimento di tutta la mia attività è tradizionalmente stratigrafico, addirittura harrisiano nello studio degli aspetti geometrici della stratificazione (la messa in fase) ed il problema non è negare gli strumenti di cui si dispone, ma cercare di migliorarli. Ecco perciò le domande che retoricamente pongo prima di tutto a me stesso. È nor- male che una disciplina che si dice scientifica adotti manuali, ma anche procedure, schede, pratiche di cantiere, vecchie di trent’anni? È normale che ad occuparsi (e scri-

vere) dei metodi di scavo siano solo pochi archeologi e che negli altri prevalga una sor-

ta di «unanimismo stratigrafico» apparentemente acritico? E che dire di quella gran

parte del mondo in cui non si adottano le procedure standardizzatesi in Italia negli an-

ni Settanta? Più in generale, i paradigmi e le procedure possono almeno in parte cam-

biare? Cosa significa qualità: molti riferimenti storici, molti strati, molti cocci, molti gigabyte? Accanimento stratigrafico e accanimento archeometrico sono una possibi-

le soluzione o costituiscono piuttosto una parte del problema? Farne a meno è possi-

bile? Mercato del lavoro e corporativismi vecchi e nuovi influenzano la qualità e la cre- scita della ricerca? Il gran numero di scavi inediti accumulatisi nei decenni o pubblicati solo a brandelli (spesso con riferimento solo alle strutture e ai materiali «significati-

vi») è in qualche misura conseguenza delle stesse procedure di scavo?

E qui, sempre procedendo per cenni che non esauriscono affatto l’argomento, ma

forse ottengono di evidenziarlo, ecco le direzioni in cui credo dovrebbe volgersi chi ha interesse a migliorare il proprio lavoro sul campo: ritornare a porsi il problema dei modi di formazione delle Unità stratigrafiche; riconoscere che si tratta di unità ope-

razionali; che non vale l’affermazione «un’azione uguale un deposito»; che molte uni-

si formano in tempi lunghi per il concomitante concorrere di più eventi. Nella pra-

tica significa che le schede sono strumenti burocratici del tutto inadeguati alla registrazione dei dati e, ancora più importante, che non tutte le Unità stratigrafiche sono degne di pari attenzione e, quindi, è sullo scavo che l’archeologo costruisce l’in- terpretazione. Ed è nel medesimo luogo che costruisce quella che potrà diventare l’edizione dei risultati in tempi e con modalità ragionevoli.1 Un’interpretazione che, con Hodder, è sì ‘in punta di cazzuola’,2 ma è, soprattutto, basata su semplici, e deci- sive, osservazione geoarcheologiche,3 sulla conoscenza di situazioni viventi in cui si

1 Per la definizione di unità operazionale cfr. De Guio 1988. Per le difficoltà a passare dallo scavo al-

l’edizione si vedano i contributi in Francovich-Manacorda 1990.

2 Hodder 1997. In questa sede, per ragioni di brevità e per focalizzare l’attenzione sulla concreta possi-

bilità di migliorare le procedure di scavo e quel che ne consegue si è evitato ogni ulteriore riferimento alla cosiddetta reflexive archaeology e a questioni di natura prettamente teorica.

3 Ancora fondamentale è Mannoni 1970 che dimostra da quanto tempo si dispone di strumenti con-

cettuali utili allo studio delle stratificazioni e di come sia possibile compiere ciò anche senza un iperspecia-

lismo raramente applicabile nel concreto.

lo scavo, i residui, l affidabilità stratigrafica 53 è osservato il formarsi di specifiche stratigrafie (l’etnoarcheologia),1 talvolta su veri e propri esperimenti,2 sulla conoscenza della casistica archeologica da cui ricavare un approfondito ‘dizionario’ di situazioni stratigrafiche ricorrenti in cui raccogliere le va- riabili che contraddistinguono specifiche situazioni.3 Per guardare più approfonditamente in avanti è quindi necessario guardare indie- tro. La ‘rivoluzione informatica’, e quindi i diffusissimi Gis o ancor meno altri stru- menti elettronici, non è, difatti, rivoluzione che scende per davvero nello scavo inter- pretando come si sono formati i depositi, ma, semmai, è operazione che pone in relazione, in una successiva fase del lavoro, un numero sempre più esteso di variabili raccolte in modo tradizionale, o quasi. Quel che necessita è unire ad una diffusa com- petenza stratigrafica (quella che, in primissima istanza, fa ritenere soddisfacente il sa- pere distinguere il colore, i componenti, la consistenza di un qualcosa e potere, quin- di, gridare ad una nuova Us) una diversa formazione degli studenti e degli operatori (naturalistica per quel tanto che serve ed etnoarcheologica per quanto possibile), una certa elasticità delle procedure di cantiere, un modo di passare dallo scavo all’inter- pretazione che non può realizzarsi nella totale separazione fra chi scava e chi scrive.

Auspicabile è, quindi, l’avvio o, se si vuole, la ripresa di un dibattito che, in realtà, at- tualmente sembra carente e che solo sporadicamente, ha guardato più alla teoria del-

la stratificazione che non alla costruzione di metodi adeguati. Adeguati, fra l’altro, a

fare sì che gli scavi non restino inediti, che non diventi prassi diffusa la pubblicazione

di notizie definite preliminari ma in realtà definitive essendo frutto dello studio non

integrale della documentazione e dei materiali.

I residui

A questo punto tornare a discutere di quei residui da cui volevo inizialmente partire

può apparire difficile se non aiutasse una frase di Carlo Pavolini, curatore insieme a Fe- derico Guidobaldi e Philippe Pergola del menzionato convegno romano. Egli, difatti, dichiarava che il convegno dedicato ai residui doveva servire da ‘capostipite’ ad una se- rie di convegni dal titolo Dallo scavo all’interpretazione così da obbligare a discutere non solo di materiali, ma del «primato del contesto», legando quindi i reperti alla terra in cui li si è rinvenuti e alla storia della stessa.4 Ragionare di residui ha, del resto, senso solo nell’ambito dell’archeologia stratigrafica e, nonostante le relazioni al convegno romano fossero incentrate sull’analisi di contesti urbani quasi sempre di età classica, il tema non può certamente trascurarsi nello studio di altri periodi e situazioni. Certamente esistono diverse definizioni di residuo, ma, intuitivamente, residuo è ciò che non è pertinente al contesto dal punto di vista cronologico, ciò che appartie- ne al passato, ciò che non dovrebbe trovarsi dove lo troviamo, ciò che Jean Paul Mo-

1 Vidale 2004, sintetico ma efficace e con bibliografia aggiornata.

2 Significativo esperimento volto alla comprensione dei fondi di capanna preistorici, ma più in genera-

le al riempimento di fosse e alla formazione di strati d’uso è Tinè 1989 soprattutto per la coraggiosa pub-

blicazione del dibattito fra esperti che rivela la consapevolezza di quanto sia difficile, ma necessario, inter- pretare come si è formato in passato ciò che si scava. Una dettagliata analisi e un positivo commento a tale lavoro si ha in Giannichedda 2002

3 Il riferimento è, ovviamente, alla per certi versi pionieristica Biografia di un foro di palo in Barker 1981 ma si veda anche quanto citato supra a nota 13 e il testo corrispondente.


enrico giannichedda

rel dice provocare sorpresa.1 Spesso il riconoscimento dei materiali residui è un’ope- razione empirica e approssimativa non migliorata da affermazioni del tipo: residuo è l’oggetto prodotto in un’epoca precedente al suo ingresso nella stratificazione o, de-

finizione migliore, residuo è l’oggetto scartato prima dell’inizio della formazione del-

lo strato in cui lo si trova. Ovviamente si potrebbe sostenere che ogni coccio è un re-

siduo della propria tazza, ma, in realtà, per capire se un oggetto è significativamente un residuo, qualsiasi sia la definizione che si dà al termine, non è mai sufficiente guar- dare al manufatto in sé, ma necessitano dati di contesto e valutare quindi le relazioni che, quell’oggetto, ha con il mondo circostante (un singolo reperto estratto ‘a forza’ dalla cassetta dei reperti di per sé non è perciò giudicabile). Un reperto è residuo o in fase rispetto a qualcosa d’altro. Nella valutazione di residuo necessita quindi considerare tre diversi caratteri dei materiali e dei contesti legando, fra loro, osservazioni aventi natura diversa. Primo ca- rattere significativo sono le modalità di giacitura del reperto; sembra al suo posto o è

in giacitura secondaria? Si può presumere che l’oggetto una volta scartato sia stato

seppellito e quindi riesumato a distanza di tempo per poi essere riseppellito? Per ca- pire ciò necessita studiare sia i caratteri della stratificazione sia quelli del reperto (usu- re, frammentazione, arrotondamenti) e dell’intera associazione di materiali rinvenu-

ti. Secondo carattere, un avvenuto decorso di tempo differente da quello di altri

materiali associati; il reperto crea stupore, non è nel suo periodo, è cronologicamen-

te fuori posto? Indizio di ciò sono, oltre a quanto menzionato per valutare le giacitu-

re, i caratteri cronotipologici del tipo.2 Terzo carattere, la funzione del reperto. Per come è o per come lo si trova il reperto può dirsi essere stato inutile? Per rispondere necessitano competenze non limitate a discriminare classi e tipi, ma in grado di valo- rizzare i caratteri dell’oggetto in relazione a materia prima, tecnica di realizzazione,

modalità di utilizzo, ma anche modi di frammentazione, tracce d’uso e riuso. Queste tre valutazioni congiunte rinviano, non a caso, ai tre paradigmi “principe” della ricerca archeologica (stratigrafia, tipologia, tecnologia) e quindi ad una guida fra

le più sicure per il prosieguo del lavoro. Nella tabella 1 si schematizzano i caratteri che

distinguono svariate situazioni concrete, ma è ovvio che lo schema è per molti versi semplificante e, oltre a non tenere conto della possibilità di oggetti sepolti e dissep-

pelliti più volte, non considera quale considerazione degli oggetti avessero gli antichi (ad esempio, erano consapevoli che si trattava di manufatti prodotti in altre epoche,

vi attribuivano un valore particolare o solo d’uso).

1 I materiali residui 1998, p. 282.

2 Si tralascia qui di affrontare il tema della vita dei vasi e, cioè, di quanto un frammento ceramico de-

ve essere più vecchio dello strato in cui lo si trova per provocare “sorpresa cronologica” ed essere quindi considerato residuo. O, con altre parole, qual è il lasso cronologico che deve darsi fra residui e materiali in fase? Per ragionare con qualche rigore occorrerebbe sapere quanto dura un vaso (conoscerne cioè la vita), ma su questo si sa davvero poco nonostante il problema sia rilevante e, non risolvendolo, è impossibile ra- gionare di quale fosse la composizione delle associazioni d’uso (trovare fra i reperti in fase cento piatti non significa che in quella casa essi fossero tutti in uso contemporaneamente). Senza dilungarsi è possibile di- re che lo studio della vita normale dei recipienti (che non ha nulla di casuale, ma è determinata da mate- riali e condizioni d’uso) dovrà in futuro divenire uno dei settori di studio da sviluppare maggiormente con, da un lato, la disamina di contesti archeologici particolari (ben datati, inalterati, di breve durata) e dall’al- tro le comparazioni etnografiche (per questi argomenti si rinvia a Mannoni, Giannichedda 1996 e Va- rien-Mills 1997).

lo scavo, i residui, l affidabilità stratigrafica



Il reperto

Il reperto

Il reperto

ha giacitura

pone un

ha una


problema funzione

di decorso

nella Us

di tempo

di giacitura


Coccio romano




Reperto in fase;


in strato romano Anello romano in tomba medievale




giacitura primaria Cimelio


Concio romano






reimpiegato Coccio romano in strato medievale




Residuo archeologico


Anello romano in strato medievale




Cimelio smarrito o residuo


Fr. olla funeraria




Residuo stratigrafico;


in livelli di sepolcreto Oggetto in posto




giacitura secondaria Reperto in situ


Monumento in elevato









Non determinabile

Tabella 1. In maniera intuitiva possono essere interpretate situazioni per certi versi simili: un coccio romano in uno strato medievale è un residuo, un concio romano in unità stratigrafica muraria me- dievale è un riuso o reimpiego. In realtà, la tabella evidenzia il concorso di tre fattori di natura strati- grafica, tipologica e produttiva-funzionale (il ciclo di vita dell’oggetto) nel definire la natura del re- perto. Come discusso nel testo si noti che i termini ‘giacitura alterata’ e ‘avere una funzione’ vanno chiariti caso per caso, mentre ‘decorso di tempo’ è ciò che determina sorpresa per un reperto che non è nello strato in cui sarebbe prevedibile doversi trovare.

Diversamente da quanto alcuni possono credere, in realtà i residui creano meno pro- blemi di quel che si pensa: raramente dovrebbero creare problemi di datazione (a da- tare è sempre il reperto più recente), più spesso problemi di definizione del contesto (inteso come associazione di manufatti coevi). I problemi sono comunque tutti di post scavo benché è sullo scavo che si pongono le premesse per risolverli. Con questo non si vuole peraltro negare né il rischio di datare, in particolari situazioni, un singolo strato con i residui più appariscenti anziché con i meno conosciuti materiali pertinenti alla fase (rendendo, ad esempio, romano ciò che in realtà sarebbe altomedievale), né l’esistenza di tipi prodotti e usati per tempi lunghi senza apprezzabili variazioni (i residui non risultano pertanto facilmente distinguibili e non possono assegnarsi a periodi precisi). In realtà, come ben mostrano molte relazioni in I materiali residui 1998, se ricono- sciuti come tali, i residui possono essere considerati al pari di altri elementi costitutivi gli strati: «da parassiti dei dati archeologici e delle quantificazioni, capaci di sminuir- ne o addirittura di metterne in crisi la validità, ad elementi innocui o perfino utili».1

1 Saguì-Rovelli 1998, p. 175.


enrico giannichedda

Essi possono difatti informare di fasi altrimenti ignote (‘attività fantasma’) a seguito della completa asportazione o distruzione della stratificazione di interi periodi. In ta-


casi i residui informano di una parte della storia del sito comprovando, ad esempio,


svolgersi di attività produttive, sepolcrali, insediative altrimenti non attestate. In fu-

turo il ricorso alla valorizzazione dei materiali residui presumibilmente diverrà sem-

pre più importante con il progredire della distruzione di molti siti archeologici e quin-

di con il formarsi di strati moderni ad alto contenuto di reperti ‘rimaneggiati’. I residui

forniscono anche elementi di novità nel campo degli studi tipologici (ad esempio nuo-

vi tipi o varianti mai riconosciute) o, più frequentemente, elementi di integrazione al

catalogo di quanto già noto per uno specifico sito (manufatti altrimenti non attestati

fra i materiali in fase). Infine, i residui possono contribuire al riconoscimento dei mo-

di di formazione dei depositi in cui vengono rinvenuti. Ad esempio, la presenza di re-

sidui di età romana nella preparazione di un terrapieno altomedievale può contribui-


a informare delle modalità con cui questo fu costruito e addirittura a indicare l’area


prelievo della terra. Su questo argomento, però, si tornerà estesamente più avanti

per quanto ha a che vedere con lo scavo e l’affidabilità stratigrafica e, prima, è utile fa-

re tesoro di quanto approfonditamente discusso nel convegno romano. In I materiali residui 1998 sono stati illustrati, in maniera peraltro diseguale, due ap- procci relativi al trasformare un apparente problema in fonte di informazione. Un pri- mo approccio, certamente privilegiato in quella sede, può ben dirsi ‘aritmetico’ o ‘sta- tistico’ e fondamentale al proposito resta l’articolo di Nicola Terrenato e Giovanni Ricci (1998). Dopo un utile accenno alla storia degli studi, in tale lavoro si affronta

l’analisi dei tassi di residualità nelle stratificazioni urbane con metodi che, in realtà, sono applicabili anche a contesti rurali e non dipendono dall’ubicazione del sito. Una prima fase del lavoro è volta ad evidenziare i limiti insiti nel semplice calcolo della per- centuale di residui per classi di materiali e quelli conseguenti al cercare di valutare la composizione cronologica dei residui ricorrendo alla stima della data mediana di ogni tipo.1 Un originale “approccio alternativo”, sempre nell’ambito aritmetico-statistico,

è quindi proposto proprio per ovviare al difetto derivante dall’imprecisa datazione di

molti frammenti che appiattisce le differenza fra tipi di breve durata e ben datati (ad

esempio, Hayes 58B, 350-375 d.C.), tipi forse di breve durata ma ancora non ben data-


(come potrebbe essere il caso della forma Hayes 81, B 360-440), tipi di lunga durata


che per lo stato delle conoscenze si devono ancora ipotizzare tali (olle grezze spes-

so datate con intervalli di un paio di secoli e più). Il metodo, detto della somma delle medie ponderate individuali2 è aritmeticamente di semplice applicazione, ma come

1 È difatti evidente che se in uno strato medievale ci sono residui preistorici, romani e del secolo prece-

dente a quello della deposizione e in un altro strato solo questi ultimi, ciò avrà qualche motivazione, ma il quantificarli non risolve il problema che evidenzia. Ad esempio, non aiuta a capire se nel primo caso gli stra-

ti preistorici e romani siano stati intaccati nel medioevo o se già allora erano residui in strati solo di poco più antichi.

2 Il metodo che si tenterà di riassumere aritmeticamente è più semplice di quanto appaia a parole e con-

sta dei seguenti passaggi (ovviamente lavorando per singole Us di cui si sono classificati, contati e datati tut- ti i reperti): 1. Dividere il totale dei frammenti di ogni tipo per l’intervallo di datazione del medesimo. Esem- pio: 14 fr. scodelle datate fra 30 e 130 d.C. = 14: (130-30) = 0,14 fr. di olla/anno. 2. Ripetere l’operazione precedente per ogni tipo. 3. Sommare per ogni anno il numero di frammenti ottenuto per ogni tipo. 4. Pre-

disporre un grafico con in ordinata il numero di frammenti e in ascissa gli anni (o i decenni se al punto 1 an- ziché dividere per 100 si è opportunamente deciso di lavorare per decenni e si è quindi diviso per 10). In tal

lo scavo, i residui, l affidabilità stratigrafica 57 evidenziano Terrenato e Ricci non elimina del tutto i problemi, primo fra tutti il doversi necessariamente accettare che ogni tipo era uniformemente diffuso per tutto un periodo (in molti casi è più realistico pensare a distribuzioni gaussiane e, nella realtà i livelli di produzione e di approvvigionamento sito per sito oltre che ignoti potevano essere mutevoli).1 Con questo procedimento si ottiene da ultimo di potere confrontare la residualità in Us diverse, ma questa è evidentemente poca cosa, e gli stessi Terrenato e Ricci2 riconoscono che talvolta si dimostra solo ciò che era co- munque “lapalissiano”. Un secondo modo di affrontare la questione residui può dirsi geo archeologico o, meglio, dei modi di formazione, così da non renderlo, con le parole, terreno per soli specialisti. Esso muove dal chiedersi il perché in taluni strati esistano oggetti che crea- no stupore cronologico, tipologico, funzionale e ne studia l’origine cercando, strato per strato, di capire chi (o che cosa), e in che modo, ha portato le componenti di uno strato a depositarsi. Di particolare importanza nell’occuparsi di residui è perciò la con- siderazione (tutt’altro che banale) che le Us, proprio perché non sono eventi puntua- li, contengono elementi conseguenti a storie diverse e accomunati in certi casi solo dal definitivo seppellimento. Al proposito, i manuali di scavo, da Barker 1981 a Ca- randini 1981, discutono brevemente la questione soffermandosi a considerare anche strati in cui tutti i reperti sono rimaneggiati e oltre a ciò procedono a una casistica delle situazioni ricorrenti: strati orizzontali, mucchi, riempimenti. Ragionando del fatto che reperti in fase e residui giungono all’archeologo insieme perché insieme seppelliti, si può introdurre qualche osservazione sull’ambiente della superficie inteso come la situazione che comprende tutti i piani d’uso coevi in un da- to momento (superficie del pavimento, ma anche dell’arativo, del cumulo di rifiuti dietro casa, del fondo di una buca aperta, della canaletta creata dalle acque piovane eccetera). In queste situazioni si ritrovano insieme, benché possano avere ciascuno un proprio specifico posto, oggetti in uso, oggetti temporaneamente accantonati con la previsione di poterli recuperare, oggetti di cui non si conserva memoria perché ab- bandonati lì da più tempo, oggetti della cui presenza non si ha neppure conoscenza (portati ad esempio in superficie dall’erosione), oggetti scartati definitivamente. E con essi la matrice, intendendo così, genericamente, la ‘terra’ o qualsiasi altro materiale costituente lo strato. Un insieme quindi complesso anche perché con il termine oggetti si devono comprendere sia manufatti, e loro parti, sia ecofatti (ossa, carboni eccete- ra) e perché quasi mai la situazione è del tutto stabile. A influenzare l’ambiente della

modo si otterrà una curva che rispecchia tutto il materiale presente nell’Us e dove “ciascun frammento ha lo stesso peso [statistico] nel determinare la curva, ma la ricchezza dell’informazione cronologica viene man- tenuta” (Terrenato-Ricci 1998, p. 93). Al fine di confrontare fra loro contesti differenti, e indipendente- mente dal numero dei reperti rinvenuti in ognuno, Terrenato e Ricci procedono infine a compiere le seguenti operazioni: 5. Trasformare il numero di frammenti calcolati al punto 3 in percentuali del totale. 6. Dare ad ogni decennio un valore uguale al proprio più quello dei decenni precedenti. In tal modo si ottiene una cur- va, detta cumulativa (o di saturazione), ad andamento crescente dallo 0 al 100%. In pratica si è cambiato l’abi- to con cui si presentano i dati senza introdurre alterazioni e ulteriori passaggi sono possibili senza però ot- tenere nuovi dati, ma solo migliorandone la leggibilità (Terrenato-Ricci 1998, p. 101). Si noti, però, che, in I materiali residui 1998 è significativa l’osservazione di Archer Martin secondo cui non esiste più di una mez- za dozzina di archeologi in grado di utilizzare i metodi statistici e la metà di essi non è interessata a farlo. 1 Il problema è ovviamente ben noto agli autori e ritenuto da approfondire (cfr. ancora l’intervento di


enrico giannichedda

superficie possono concorrere agenti atmosferici (dilavamento, degrado, spostamen- to di oggetti e matrice), calpestio umano e animale (compressione, altri piccoli spostamenti di oggetti e matrice), ripetuta dislocazione di intere parti (ad esempio a seguito di arature), deposizione di nuovi materiali, attività sotterranee di animali e piante (tane, radici), pedogenesi. La situazione ‘in superficie’ è quindi una situazione che va pensata come in co- stante trasformazione (mutano le relazioni spaziali, le associazioni di materiali pos- sono venire smembrate e costituirsene di nuove, magari determinate dalla forza gra- vitazionale e non da scelte antropiche; intere classi di materiali scompaiono, ad esempio, perché degradate dagli agenti atmosferici). I residui di periodi precedenti ri- tornati per qualche motivo in superficie, solitamente subiscono una parte di questo processo, che è insieme culturale e naturale, e successivamente, insieme ai materiali che diremo in fase, vengono seppelliti con quanto ne consegue.1 L’esito di tutto que- sto nelle schede di Unità stratigrafica spesso si riduce a una crocetta nelle caselle Na-

turale o Artificiale, in descrizioni buone per tutte le occasioni («strato di terra marro- ne, mediamente compatta contenente…»), in un inadeguato riconoscimento di come il singolo strato, con tutto ciò che contiene, è giunto laddove lo si trova. Proprio con la consapevolezza della necessità di meglio discriminare i modi di formazione degli strati, è possibile considerare anche i residui come utili elementi di informazione. E non solo, come giustamente indicano Terrenato e Ricci, «all’interno

di ciascun sito, la quantità e la distribuzione cronologica dei residui possono infor-

marci sulla provenienza della matrice che ha formato ogni strato»,2 ma valorizzando

le osservazioni basate su criteri non tipologici di valutazione della residualità: in

primo luogo la differenza di frammentazione riscontrabile in classi diverse di mate- riali (i residui possono essere più frammentati a seguito di seppellimento-scavo- riseppellimento, ma è anche vero che, ad esempio, le ceramiche grezze preistoriche possono rompersi meno di ceramiche con pareti sottili più recenti e meno resistenti), ma anche alcuni caratteri riscontrabili sulle superfici dei reperti (ad esempio, angolo- sità delle fratture, distacco dei rivestimenti, colorazione differenziale con alcuni di questi caratteri utilizzabili anche nello studio di manufatti non ceramici e, addirittu- ra, di resti organici non lavorati come ben evidenziato in Dobney et al. 1997. Nel caso di contesti con molti materiali un qualche indizio relativo alle vicende de- posizionali è dato dalla presenza di pezzi dello stesso manufatto in più strati (diverso è difatti il caso di ‘attacchi’ frequenti in strati fra loro contigui o in strati topografica- mente e/o cronologicamente lontani).

Ritornare allo scavo

Un’inadeguata o, peggio, erronea, determinazione delle associazioni di manufatti d’uso coevo può avere gravi ricadute sugli studi di caratterizzazione del sito e, ad una scala più generale, sugli studi a carattere economico, tecnologico, sociale (dalla stima dei livelli di produzione, alle logiche commerciali…). Da questa considerazione ne di- scende che i residui non possono essere trascurati e, se anche la quantificazione della

1 Sul tema delle alterazioni conseguenti al seppellimento e postdeposizionali cfr. Donato et al. 1986, Ar- noldus-Maetzke 1988 e una discussione meno analitica in Giannichedda 2006.

lo scavo, i residui, l affidabilità stratigrafica 59 residualità potrà, forse, divenire «un utile sottoprodotto» delle operazioni di classifi- cazione,1 l’attenzione va focalizzata su quanto è possibile ottenere ragionando, insie-

me, di reperti (in fase e non) e di strati. In corso di scavo, ragionare del perché si rin- vengono residui può contribuire a rendere meno inadeguate le descrizioni dei modi


formazione degli strati, a cogliere come minimo la molteplicità di apporti in mol-


Us, a ragionare dell’affidabilità stratigrafica delle stesse Us che, è noto, nel corso

del lavoro di post scavo saranno considerate come cassetti da cui estrarre i materiali da studiare e i campioni destinati ad analisi e studi specialistici. Cassetti tutti uguali,

nei grandi siti migliaia e migliaia, e che, se affrontati senza guida, impediscono, o co- me minimo ritardano, qualsiasi studio complessivo del sito. Senza valutare i modi di formazione degli strati, l’archeologo rischia, da un lato di essere vittima del proprio egualitarismo stratigrafico (l’obbligo di considerare tutti gli strati come parimenti si-

gnificativi) e dall’altro di basare l’interpretazione su poche evidenze più chiare (spes-

so i resti di costruzioni a scapito, ad esempio, degli indizi di attività agricole e mani-

fatturiere). Le considerazioni di cui sopra vanno nella direzione di una «archeologia della ma- trice» (così definita da Enrico Zanini in I materiali residui 1998, 293) di cui gli stessi re- sidui sono una componente certamente antropica per produzione, ma di cui restano

da definire le cause che ne determinano la presenza in quella data giacitura. E questo

nella consapevolezza che la matrice, qui intesa come la “terra” in cui giacciono i re- perti, è sempre lei benché cambi ogni volta che venga rimossa, arata, vangata, pres- sata e quindi, in realtà, non sia mai eguale a se stessa. Erroneo quindi sostenere che la

matrice possa essere residuale perché, nel caso, tutto sarebbe residuale e da ricondursi ad antichissimi eventi geologici. Non ha quindi senso sostenere che alta percentuale di residui = alta percentuale di matrice residuale, ma può essere invece vero il caso in cui, in uno strato con materiali residuali, si rinvengono anche porzioni di matrice che con- servano i caratteri dello strato d’origine (è il caso, banale, di parti di murature crolla-

te in cui può riconoscersi la tecnica muraria antica, ma anche di zolle di suoli ancora

riconoscibili all’interno di depositi successivi). Reperti e strati, strati e reperti (residui compresi). Solo in tal modo è possibile af- frontare il tema dei residui non semplificando le situazioni in cui è possibile ritrovar- li. Può, ad esempio, capitare di leggere che i riempimenti di buche sono ipotizzabili contenere numerosi residui (evidentemente ritenuti provenienti dall’azione di scavo delle stesse), ma anche l’esatto contrario. Entrambe le osservazioni potrebbero rite- nersi vere solo per quanto sono generiche. In realtà per ragionare della probabilità che

in una unità stratigrafica positiva vi siano residui non bisogna pensare che si tratta di

un riempimento di unità negativa (perché ciò induce all’errore), ma di un qualcosa che si è stratificato dopo un trasporto (brevissimo o lungo non importa).

Ad esempio, nell’identica situazione di una fossa che ha interessato livelli antropi-

ci preesistenti possono aversi riempimenti con differente residualità. Se si tratta della

fondazione di un muro la terra estrattane e gettata a lato del muro appena costruito è probabile contenga residui, invece, nel riempimento di una fossa fatta per cavare ar- gilla e poi usata come immondezzaio, i residui saranno quasi certamente pochi o as- senti essendo finiti, insieme all’argilla, nelle vicinanze della bottega del vasaio, forse

1 Terrenato-Ricci 1998, p. 102.


enrico giannichedda

nella zona destinata a stoccaggio e depurazione della terra, forse addirittura come chamotte nei vasi. Probabilmente i residui sono presenti nei riempimenti di fosse quasi soltanto quan- do nelle stesse si getta la terra che ne è stata estratta, ma si deve convenire che, fatte salve le fosse sepolcrali e quelle di fondazione (per muri, ma anche per piantare alberi), non è questo un caso ricorrente. La situazione ‘normale’ è la fossa scavata per essere mantenuta aperta o per collocarvi qualcosa d’altro rispetto a ciò che c’era in precedenza. Nel pozzo (per acqua o a perdere), nella cava, nella canaletta, nell’im-

mondezzaio, i residui non dovrebbero esserci e se ci sono occorre chiedersi come ci sono finiti (e ancora prima se davvero vanno considerati come residui). Altre osser- vazioni generalizzanti (e quindi difettose) potrebbero forse tentarsi, ma basti notare che gli strati a crescita continua (livelli di occupazione, discariche, gli stessi crolli) non dovrebbero contenere residui, mentre gli stessi sono possibili negli strati interessati da azioni di spostamento di matrice e/o alterazione del volume e della stessa super- ficie. Un caso quindi frequente ed esemplificato da arativi, depositi colluviali, suoli in- teressati da pascolo intenso unito a dilavamento. I pavimenti ovviamente al loro in- terno possono contenere residui (dipende con che materiali sono stati costruiti), ma quel che conta è non pensare che essi sigillino nulla più di quanto è immediatamen-

te e topograficamente sottostante (e perciò nello strato sopra il pavimento si posso-

no trovare residui provenienti da strati più antichi posti pochi metri al margine dello stesso o derivanti dallo scavo della fossa di fondazione dei muri della medesima casa). Richiamare, come si é sinteticamente fatto, casi ricorrenti ed eccezioni è, eviden- temente, un modo per sottolineare un problema e, in parte, la sua soluzione. Miglio- rare le procedure di scavo significa, a mio avviso, acquisire più mezzi (intellettuali) per capire come si sono formati gli strati e, quindi, cosa è possibile ricavarne. In tal sen-

so, i residui, insieme agli altri reperti e alla “matrice”, possono dare indicazioni sui mo-

di di formazione e, fatto non secondario, tali indicazioni sono a disposizione anche di

chi non è abituato a ragionare tenendo conto di quei pochi fattori ‘geoarcheologici’ che sovrintendono, insieme o contro l’uomo, al formarsi degli strati (caratteri del ma- teriale, pendenza, forza di gravità, azione delle acque meteoriche e superficiali, pe- dogenesi).

Fra l’altro, i residui possono così concorrere a una tipologia delle ‘cose da scavare’ che sulla falsariga della Biografia di un foro di palo – o del seminario sulle testimonian-

ze di combustione o dello studio tafonomico delle sepolture o dei concetti di tenden-

za e fatto come utilizzati da Leroi-Gourhan – costituirebbe un enorme progresso del-

la disciplina non più schiacciata dall’approccio geometrico harrisiano allo studio della

stratificazione, ma libera di ricercare i caratteri ricorrenti a seguito di determinati comportamenti ritenuti storicamente rilevanti.1 All’archeologo, sul campo, non più

1 Cfr. Leroi-Gourhan 1973 per i focolari preistorici e Duday 1990 e Duday 2006 per lo studio tafono- mico delle sepolture. Per il ricorrere nell’esperienza umana di tendenze e fatti fondamentale è Leroi-Gour- han 1993 ripreso in Giannichedda 2006. Significativi esempi di studio dei modi di formazione di eviden- ze significative (ad opera di archeologi) sono Antico Gallina 1996 relativo ai dei cosiddetti drenaggi con anfore, Neri 2004 e Neri 2006 (con ulteriori considerazioni in Giannichedda 2007). Posizioni contrastanti sull’utilità di un approccio esclusivamente harrisiano in Giannichedda 2004 e in Medri 2004 che fanno seguito a Harris 2003 e affrontano da punti diversi il tema delle stratigrafie murarie. Più in generale è fondamentale Carver 2003 che discute della necessità di ‘valutare prima di indagare’ e quindi prima di

lo scavo, i residui, l affidabilità stratigrafica 61

il compito di smontare la stratificazione riconoscendone i limiti, ma quello di capire

nel volume degli strati che cosa ne ha determinato la formazione: ad esempio, attivi-

ripetute di forgia, scarico di rifiuti seguito da compressione per calpestio, utilizzo


un vano come stalla, crollo per cause naturali o mancata manutenzione. Valutare

non solo dal punto di vista quantitativo i residui presenti fra i reperti significa andare

proprio nella direzione di meglio affrontare il tema dell’affidabilità e, quindi, capire cosa si può ricavare, ad esempio, da campioni di ossa, carboni, semi eccetera. Inutile soffermarsi a chiarire perché ciò sia importante: chi dedicherebbe tempo e denaro o farebbe affidamento su un campione di ossa animali da uno strato medievale con re- sidui preistorici indistinguibili pari, ad esempio, al 30 o 50%? In generale, ragionare della formazione degli strati e delle associazioni di manufatti

in essi contenuti significa costruire uno strumento per discriminare, nella stratifi -

cazione archeologica quali pagine sfogliare velocemente, quali trascurare, quali leg- gere con la massima cura. Un compito, questo, per archeologi esperti, presenti sullo

scavo, competenti dei materiali, interessati a capire e a ricostruire, nel presente, una storia di sito che, altrimenti, spesso resta celata in una documentazione cartacea op- primente e a posteriori inestricabile. Chiunque abbia provato a ‘rileggere’ vecchi sca-

vi attraverso la documentazione lasciata da altri sa quanto sia forte la sensazione di

una, ormai incolmabile, perdita di informazioni avvenuta proprio nel momento in cui

erano raccolti, senza interpretarli, i dati. Affrontare con tutti i mezzi possibili il tema

di come si forma la stratificazione è forse il solo modo per ragionare dell’affidabilità

stratigrafica in chiave storica: non ponendosi il problema se i limiti di Us sono netti,

ma quello di scegliere, fra tutte le unità stratigrafiche teoricamente degne di eguale attenzione, le sole meritevoli di essere studiate approfonditamente. Un modo che molti, e da sempre, adottano senza dichiararlo, ma finendo così con celare i criteri scelti e, quindi, impedendo ogni critica e ogni possibile studio alternativo. Chiedersi perché ciò che si sta rimuovendo nello scavo è lì dove lo si trova è fondamentale e lo studio dei residui (quantificazione compresa) può aiutare proprio in questo e nel va-

lorizzare le associazioni di manufatti coevi, le uniche davvero utili a studiare l’equi- paggiamento materiale degli antichi. Lo studio dei modi di formazione non è certo facile, è qualcosa per la quale spesso mancano gli strumenti concettuali (quel poco di geoarcheologia e di etnoarcheologia

a cui si è già fatto cenno), ma è anche un modo per ribadire che la qualità dell’inda-

gine archeologica si ha solo con la presenza sui cantieri di archeologi motivati a capi-

re ciò che smontano e, quindi, sempre più esperti in quelle che devono essere le com-

petenze imprescindibili per lo scavo. Fra queste competenze dovendo rientrare anche

quella di raccogliere i dati interpretandoli per quanto necessita a organizzarli in mo-

di suscettibili di essere pubblicati. Uno dei problemi maggiori di ogni studio che ab-

bia per obiettivo il territorio è difatti il gran numero di scavi inediti, un vero scanda- lo che periodicamente viene menzionato, ma che può trovare soluzione solo abbreviando la catena operativa che attualmente dallo scavo conduce con grandi len-

emettere una sentenza che decide della sorte del sito. Per far ciò, necessita riconoscere preliminarmente i caratteri specifici dei depositi in singoli siti e il loro potenziale informativo con il rischio, anche con di sba- gliare, ma con la consapevolezza che «noi possiamo predire il carattere dei depositi, noi possiamo ricono- scerne il valore, noi possiamo persuadere la comunità a studiare i depositi interessanti e preservare gli altri» (Carver 2003, p. 113).


enrico giannichedda

tezze all’interpretazione e all’edizione.1 Interpretare sullo scavo le associazioni di re- perti e matrice (i modi di formazione) va, a mio avviso, nella direzione di semplifica- re la catena e accrescerne l’efficienza purché l’obiettivo sia chiaro e condiviso dal- l’inizio. Per questo, però, servono archeologi disponibili a procedere ‘senza rete’ e cioè senza le schede compilate meccanicamente a garanzia del lavoro (o con schede nuove e ‘intelligenti’), forse, anche archeologi maggiormente dubbiosi della propria preparazione, ma motivati, con la loro fatica, a fare storia e non soltanto a consenti- re trasformazioni irreversibili del territorio in cambio della possibilità di riempire gli archivi di schede e i magazzini di manufatti.


Antico Gallina 1996 = M. V. Antico Gallina, Valutazioni tecniche sulla cosiddetta funzione drenante dei depositi di anfore, in Acque interne: uso e gestione di una risorsa (Itinera. I percorsi del- l’uomo dall’antichità a oggi, 1), Milano, 1996, pp. 67-112. Arnoldus Huyzenveld-Maetzke 1988 = A. Arnoldus Huyzenveld, G. Maetzke, L’in- fluenza dei processi naturali nella formazione delle stratificazioni archeologiche: l’esempio di uno scavo al foro romano, «Archeologia Medievale», xv, 1988, pp. 125-175. Barker 1981 = P. Barker, Tecniche dello scavo archeologico, Milano, 1981. Carandini 1981 = A. Carandini, Storie dalla terra. Manuale di scavo archeologico, Bari (con aggiunte e modifiche nelle edizioni successive: Torino, 1991 e 1996). Carver 2003 = M. Carver, Archeological Value and Evaluation, Mantova, 2003. De Guio 1988 = A. De Guio, Unità archeostratigrafiche come unità operazionali: verso le archeo- logie possibili degli anni ’90, in Archeologia stratigrafica dell’Italia settentrionale, 1, Atti del Con- vegno di Brescia 1 marzo 1986, 1988, pp. 9-22. Dobney-Kenward-Roskams 1997 = K. Dobney, H. Kenward, H. Roskams, All mixed up but somewhere to go? Confronting residuality in bioarchaeology, in Method and Theory in Histori- cal Archaeology, eds. G. De Boe, F. Verhaeghe (Papers of the Medieval Europe, vol. 10), Brugge, 1997, pp. 81-88. Donato-Hensel-Tabaczynsi 1986 = Teoria e pratica della ricerca archeologica I. Premesse meto- dologiche, a cura di G. Donato, H. Hensel, S. Tabaczynsi, Torino, 1986. Duday et al. 1990 = H. Duday et al., L’Anthropologie “de terrain” reconnaissance et interpretation des gestes funeraires, «Bulletin et Memoires de la Societe d’Anthropologie de Paris», 2, 1990, pp. 29-50. Duday 2006 = H. Duday, Lezioni di archeotanatologia. Archeologia funeraria e antropologia sul campo, Roma, 2006. Francovich – Manacorda 1990 = Lo scavo archeologico, dalla diagnosi all’edizione, a cura di R. Francovich, D. Manacorda, Firenze, 1990. Giannichedda 2002 = E. Giannichedda, Buche, discorsi, esperimenti e dizionari, in Omaggio a Santo Tinè. Miscellanea di studi di Archeologia preistorica e protostorica, Pubblicazioni del, n. 202, Genova, 2002, pp. 43-55. Giannichedda 2004 = E. Giannichedda, L’incorreggibile ‘Harris’ e altre questioni, in Archeolo- gia dell’Architettura, IX, 2004, pp. 33-43. Giannichedda 2006 = E. Giannichedda, Uomini e cose. Appunti di archeologia, Bari, 2006. Giannichedda 2007 = E. Giannichedda, Da Teofilo a Biringuccio: parole e diagrammi per in- terpretare la realtà, in Dal fuoco all’aria. Tecniche, significati e prassi nell’uso delle campane dal

1 In questa sede volutamente si tralascia qualsiasi considerazione ‘politica’ sui modi in cui viene orga- nizzata la ricerca sul terreno, ma anche sulla formazione delle competenze necessarie e sui sistemi di pote- re in vigore.

lo scavo, i residui, l affidabilità stratigrafica


medioevo all’età moderna, a cura di F. Redi e G. Petrella, Agnone 6-9 dicembre 2004, 2007, pp.


Giannichedda (in stampa) = E. Giannichedda, Il problema della residualità della ceramica nel- lo studio dei contesti archeologici, in Atti del Corso Valorizzazione dei manufatti ceramici prove- nienti da scavo archeologico, Università Cattolica del Sacro Cuore, Milano, 2003 (in stampa). Harris 1983 = E. C. Harris, Principi di stratigrafia archeologica, Roma, 1983. Harris 2003 = E. C. Harris, The stratigraphy of standing structures, «Archeologia dell’Archi- tettura», viii, 2003, pp. 9-14. Hodder 1997 = I. Hodder, Always momentary, fluid, flexible’: towards a reflexive excavation me- thodology, «Antiquity», 71, 1997, pp. 691-700. I materiali residui 1998 = I materiali residui nello scavo archeologico, a cura di F. Guidobaldi, C. Ta- volini, P. Pergola, Testi preliminari e Atti della tavola rotonda, Roma 16 marzo 1996, Col- lection École Française de Rome, 149, Roma, 1998. Leonardi 1982 = G. Leonardi, Lo scavo archeologico: appunti e immagini per un approccio alla stratificazione, Corso di propedeutica archeologica, Corezzola, 1982, pp. 97-142. Leonardi 1992 = Processi formativi della stratificazione archeologica, a cura di G. Leonardi, Atti del seminario internazionale Formation processes and excavation methods in Archaelogy: per- spectives, Padova 15-27 luglio 1991, Saltuarie dal Laboratorio del Piovego 3, Padova, 1992. Leroi-Gourhan 1973 = Seminaire de 1973 sur les structures d’habitat. Témoins de Combustion, ed. A. Leroi-Gourhan, College de France, Chaire de Prehistoire (dattiloscritto, pp. 1-44). Leroi-Gourhan 1993 = A. Leroi-Gourhan, Evoluzione e tecniche, vol. 1, L’uomo e la materia, Milano, 1993. Mannoni 1970 = T. Mannoni, Sui metodi dello scavo archeologico nella Liguria montana. Appli- cazioni di geopedologia e geomorfologia, «Bollettino Ligustico», 179, 1970, pp. 51-64. Mannoni-Giannichedda 1996 = T. Mannoni, E. Giannichedda, Archeologia della produ- zione, Torino, 1996. Medri 2004 = M. Medri, Harris 2003: super SuDoku o qualcosa di utile?, «Archeologia dell’ar- chitettura», ix, 2004. Neri 2004 = E. Neri, Tra fonti scritte ed evidenze archeologiche: un modello per interpretare i resti materiali della produzione di campane, «Archeologia Medievale», xxxi, 2004, pp. 53-98. Neri 2006 = E. Neri, De campanis fundendis. La produzione di campane nel medioevo tra fonti scritte ed evidenze archeologiche, Milano, 2006. Piuzzi 1990 = F. Piuzzi, La ricerca stratigrafica in archeologia. Introduzione ai metodi di scavo e documentazione, Udine, 1990. Saguì-Rovelli 1998 = L. Saguì, A. Rovelli, Residualità, non residualità, continuità di circola- zione. Alcuni esempi dalla Crypta Balbi, in I materiali residui 1998, pp. 173-194. Santoro 1997 = Archeologia come metodo. Le fasi della ricerca, a cura di S. Santoro, Parma, 1997. Terrenato-Ricci = N. Terrenato, G. Ricci, I residui nella stratificazione urbana. Metodi di quantificazione e implicazioni per l’interpretazione delle sequenze: un caso di studio dalle pendici settentrionali del Palatino, in I materiali residui 1998, pp. 89-104. Tinè 1989 = Interpretazione funzionale dei “fondi di capanna” di età preistorica, ed. S. Tinè, Mila- no, 1989. Tronchetti 2003 = C. Tronchetti, Metodo e strategie dello scavo archeologico, Roma, 2003. Varien-Mills 1997 = M. D. Varien, B. J. Mills, Accumulations Research: Problems and Pro- spects for Estimating Site Occupation Span, «Journal of Archaeological Method and Theory», 4, 2, 1997, pp. 141-191. Vidale 2004 = M. Vidale, Che cos’è l’etnoarcheologia, Roma, 2004.


enrico giannichedda


Nel presente contributo si affronta la questione dell’adeguatezza delle procedure di scavo alla reale situazione della ricerca archeologica sul terreno. Da oltre un trentennio le procedure e le pratiche di cantiere si sono standardizzate, ma un gran numero di scavi restano inediti o pubblicati solo in parte. I metodi hanno a che fare con questo? L’attuale ‘unanimismo strati- grafico’ apparentemente acritico si deve ritenere positivo? Più in generale, i paradigmi e le pro- cedure possono almeno in parte cambiare? Cosa significa qualità: molti riferimenti storici, molti strati, molti cocci, molti gigabyte? Questi ed altri problemi sono in parte considerati discutendo di residui e reperti in fase. Riconoscere nei primi una giacitura alterata, un avve- nuto decorso di tempo, l’assenza di una funzione, oltre che possibile è difatti relativamente facile e inevitabilmente rinvia allo studio dei modi di formazione degli strati. Uno studio, quest’ultimo, che si ritiene fondamentale perché è l’unico in grado di evidenziare la diversa affidabilità stratigrafica (e storica) delle associazioni di reperti e matrice consentendo, così, la raccolta non meccanica dei dati, una prima interpretazione sul cantiere di ogni distinta porzione stratigrafica, un più agevole lavoro di edizione dei risultati.

The present contribution questions the suitability of current excavation procedures. For thir- ty years or more, the field procedures and practices of archaeological excavations have been standardized. Yet, a large number of excavations remain unpublished or are merely published in part. Do the methods used have anything to do with this? Is it possible to maintain a posi- tive view of the present a-critical ‘stratigraphical consensus’? Or should we envisage an at least partial change in the paradigms and procedures normally applied? Does ‘quality’ mean noth- ing more than many historical references, many layers, a multitude of potsherds, and many gigabytes? The article considers these and other problems in connection with a discussion of the residual evidence and finds in phases. To recognize in the first an altered layer, a passing of time, the absence of a function is – in fact – relatively easy and inevitably refers one to the study of the formation of the layers. This is fundamental because it is the only possible way to elucidate the varying stratigraphical and historical trustworthiness of the find associations, which may lead to a non-mechanical gathering of data: a first interpretation in the field of all distinct stratigraphical parts, leading to a smoother way of publishing the results.


Gary Reger

« L

ocal», «regional», «interregional» (or long-distance) – these three terms serve of- ten to distinguish levels of circulation of goods, and so to analyze economic ac-

tivity. A Koan amphora found in Petra has obviously traveled a long way from home; a coin of Kleonai found in the excavations of the sanctuary Nemea from the years when Kleonai controlled the games is just as obviously local.1 «Regional» must lie somewhere in the middle – although what exactly constitutes a «region», whether dif- ferent criteria produce different regions, and whether the boundaries, however labile, of a given region change drastically enough over time to create a new region, or de- stroy an old one, are very difficult questions, impossible to address in the brief com- pass of this paper.2 «Local» itself is a slippery term, for what constitutes «local» pro- duction, distribution, and use? To be «local» must a product be produced, circulate, and be used within (say) the boundaries of a political unit like a polis? Does the bronze coinage called «Milesian» or «epichoric» in some inscriptions cease to be «local» be- cause examples have turned up at Priene, on the island of Kalymnos, and way off in the western Peloponnesos at Lenchaina?3 Moreover, these levels of activity need not be mutually exclusive. A «local» prod- uct – say a bronze coin of no value outside the territory of the issuing authority, or a cheap locally-produced vessel – may nevertheless show up far from home, transport- ed and lost by a traveler and recovered centuries later by an archaeologist. This is obvious. But «local» and «regional» may intertwine in more interesting ways. For instance, the Delian accounts from the temple of Apollo happen to mention the purchase by the temple, in 279 bce, of roof tiles from the neighboring island Syros. Perhaps the tiles were made specially for Apollo, but even so there must have been a local demand for tiles on Syros as well, and it seems safe to presume that Apollo’s overseers sought out tiles from Syros just because the Syrians were already produc- ing, or showed they had the capacity to produce, tiles of the type the Delians need- ed. There is in this case an interconnection between the purely local circulation of Syrian tiles on Syros to meet Syrian demand, and the regional draw of Syrian tiles to Delos for a Delian demand. Was the Delian market for Syrian tiles steady? Did

1 Schneider 1996, p. 132, for the amphora; Knapp-Mac Isaac 2005, p. 161 with pp. 58-60 for the coin.

2 These matters will come in for much more detailed discussion in work in progress, including Reger

forthcoming; a paper called Formation of Taste and Fashion. Perfumes and Imitations in the Hellenistic and Ear- ly Imperial World for a workshop on «Imitations – Regional Traditions – Economic Impacts» to be held in University of Manitoba in late 2007; and, more discursively, in a book tentatively titled Region and Network in the Hellenistic Economy.


gary reger

Delians, for their houses and warehouses and public buildings, prefer Syrian tiles, so that there was a constant stream of Syrian goods moving to the sacred island? Or was the demand occasional, punctuated, perhaps even dependent on personal relations and interconnections of merchants or workers?1 It is impossible to say, but the exam- ple illustrates one way in which the local and the regional may link up, temporarily, seasonally, or occasionally. Perhaps the best illustration of the phenomenon I am trying to describe would be the civic coinage of a particular polis, since it was struck (legally) only in one place, probably in the case of most poleis indeed in a single workshop, but in some cases can be seen to have entered into a broader regional circulation.2 Here is an example. Around 200 bce, or perhaps slightly later, the Naxians in the Kyklades began to mint the so-called kraterophoric coinage, distinguished by a head of Dionysos on the ob- verse and a krater on the reverse.3 It is no surprise, of course, that the hoard most pro- ductive of examples was found on Naxos itself, and the island has also produced at least one stray find.4 But this coinage entered into circulation off the island as well. An inventory from Delos of 153/2 bce registers these coins,5 and they are preserved as well in two hoards found on Euboia and one in Thessaly near Larissa.6 The distri- bution bespeaks a local product (the Naxian kraterophores) which entered into a ‘re- gional’ circulation. Many other examples of «local» products in the form of coins en- tering into a larger «regional» circulation could be cited. Why did this local product enjoy a certain acceptability outside the polis where it was made? Obviously, at least in part, the answer must be economic. The presence of Naxian kraterophoric coins on Naxos and Euboia surely corresponds to a route of trade running from the central Kyklades to Euboia, as Louis Robert noticed many years ago.7 But there may be more to it than that, as a quick glance at another local- regional coinage may suggest. In the later third century bce, the silver coinage of Rhodos struck to a standard of about 3.4 gr/dr circulated widely in Karia in south- western Asia Minor, apparently serving as the medium of exchange for many Karian cities, which did not strike their own coinage in this period. Around 225 bce, the Rho- dians reduced the weight of the drachmas they struck to about 2.8-2.5 gr/dr, although they did not alter the weight standard of the other denominations of this coinage. When, however, around 190 bce the Rhodians abandoned this coinage for the so- called «plinthophoric» coinage, struck on a standard of about 3.0 gr/dr, several Kari- an cities not under Rhodian authority began to issue their own coins, on the reduced drachma standard but with Rhodian types.8 In other words – if this reconstruction is right – the Rhodians’ decision to change their type and weight standard disrupted a regional currency system based on the older Rhodian coinage, and local authorities rejected the new system, taking up the task of striking coins imitative of the former types and weights. In this case, a conservatism about what “counted” as acceptable

1 ig xi 2 161A73-75. The role of personal relations in economic life deserves much more attention than

it has received; in addition to the Kratinos who transported the tiles, two other Syrian contractors are at- tested on Delos: ig xi 2 199A73, 76, 107 (274 bce) and 161 A52, 57, and 165.37.


I leave aside here the problem of the ‘traveling mint’.

3 Nicolet-Pierre 1999.


igch 255 with Nicolet-Pierre 1999, pp. 103-108, 113.

5 id 1432, Bb i.51-52, also 1449 Ba I.57-58.


igch 226, 210 + 215; ch viii (1994), no. 517, all with Nicolet-Pierre 1999, pp. 114-116.

identifying regions in a greco-roman mediterranean context 67 coinage may have come into play beyond any purely economic considerations, since the Rhodians were now operating with a different coinage that could not simply be interchanged with the Karian emissions – even though the Rhodians controlled much of Karia directly after 188 bce.

Regions and consumption, regions and production

The archaeological record of the Greco-Roman world has yielded many examples of regionalism in material culture: styles of pottery, housing, and so on, too many to name. It may be useful to look briefly at this phenomenon from two opposite ends of economic activity: regionalism in consumption (attested through distribution of goods) and regionalism in production (a concept that has been introduced recently into the literature). A striking example of regionalism in the distribution of trade goods has been iden- tified by John Lund for the cases of Eastern Sigillata A and Cypriote Sigillata.1 East- ern Sigillata A began to be produced around 150 bce, almost certainly – as we know now – in kilns near Rhosos, close to Antiocheia.2 The place of production of Cypri- ote Sigillata remains somewhat uncertain but is most likely to be set in the south- western corner of the island, perhaps near Nea Paphos. Eastern Sigillata A was wide- ly distributed, at least in small quantities, over much of the eastern Mediterranean world, but it was concentrated especially on Cyprus, along the Levantine coast, on the south coast of Asia Minor north and north-east of Cyprus, in southwestern coastal Asia Minor, on Krete, and on the African coast from Egypt to Cyrenaica.3 Cypriote Sigillata enjoyed a somewhat less extensive distribution, though in many re- spects presents a similar pattern.4 When quantities are taken into account, however, notable differences arise: first, while considerable numbers of Eastern Sigillata A ap- pear at sites far distant from the production center, particularly in western Asia Mi- nor, Cypriote Sigillata is found in large quantities only on Cyprus itself and to a less- er extent on the Levantine coast and Asia Minor directly north of Cyprus. Second, in the zones in which Eastern Sigillata A and Cypriote Sigillata constitute more than half of the fine wares dated to the first century bc, a striking regional pattern emerges, in which Eastern Sigillata A dominates in eastern Cyprus, the Levantine coast, and the crook of Asia Minor, while Cypriote Sigillata dominates in western Cyprus and the southern Asia Minor coast west of approximately Tarsus.5 Lund has now added another example to this phenomenon: the so-called «Aradip- pou goblet», identified from finds at Panayia Ematousa in southeastern Cyprus. This vessel, a drinking cup with two upturned handles and an everted shape, seems to have been a peculiarly Cypriote product, perhaps descended from an earlier skyphos shape. But it is the distribution of this type that is most striking: it has been found (so far anyway) only at a handful of sites in southeastern and far eastern Cyprus.6 Its dis- tribution – to which I return below – thus mirrors, on Cyprus, the dominance of East-

1 Lund 1999 and Lund 2005a.

2 In 1999 Lund could only say «North Western Syria and Eastern Cilicia» (Lund 1999, p. 2), but he has

now offered a strong argument for the Rhosos kilns in Lund 2005a.

4 Lund 1999, p. 20, Fig. 11.

3 Lund 2005a, p. 242, Fig. 10.5.

5 Lund 1999, p. 19, Fig. 9 = Lund 2005b, p. 76, Fig. 10.


gary reger

ern Sigillata A on the eastern parts of the island. That we now have two well-attest- ed examples of a regional break on Cyprus between the southeastern area of Kition and Salamis and the western fringe of Nea Paphos supports the view that a genuine regional difference existed on the island. On the production side, Jeroen Poblome and Raymond Brulet have recently sug- gested the need to consider «regional production» in identifying provenance for ce- ramics. They develop this suggestion on the basis of reflections on patterns of pro- duction in Gaul from the second half of the first century bce and the appearance in Asia Minor of the names of potters on Eastern Sigillata B (esb) from sites in western Asia Minor and Corinth. «Different potters,» they suggest, «may have been producing esb at different places within the same region of southwestern Asia Minor. Their products are contemporary, typologically and technologically similar, and are basi- cally eastern interpretations of Italian designs.»1 It is interesting to apply this concept to another, more restricted example. One of the chief products of the Koan wine industry was the amphora with the double-bar- reled handle. Recent work in Kos town has shown that this type began as early as the fifth century bce, and it continued to be one of the most common types of amphora produced on Kos through the Hellenistic period. However, Kos was not the only site that evidently produced such amphoras. Maria Berg Briese has identified double-bar- reled handle amphoras produced in ten different fabrics from excavations at Halikar- nassos. On the basis of these identifications and earlier work attempting to distinguish fabrics of amphoras from the Koan and Knidian region, Briese suggests with some caution that these fabrics may be sorted into three groups: one group of fabrics prob- ably actually from Kos itself; a second set of fabrics perhaps from Knidos, and a third set perhaps produced in Halikarnassos.2 At first blush, this might look like a case of imitation. Production of the double- barreled type began on Kos itself, presumably as a container for Koan wine; the later Knidian and Halikarnassian examples, then, would be subsequent copies. But this idea carries less appeal when one considers the reputation of the wines. Strabo re- marks that Knidos, along with a number of other places in western Asia Minor, «pro- duces extremely good wine, whether for enjoyment or medical purposes» (14.1.15). He also praises Koan wine in the same passage as among the ‘best’ wines, aristos, but evidence from Delos in the early second century indicates that Knidian wine there was in fact considerably more expensive than Koan.3 This casts some doubt on the idea that imitation was operating in the decision to follow Koan lead in the style of amphoras produced in Knidos and Halikarnassos, for why would one place a better wine in a container associated with a cheaper? Here is where regional production may help. For Athenaios, in his famous dissertation of the qualities of wines, classes together as flavored with sea-water wines from Myndos, Halikarnassos, Kos, Rhodos (less so), and implicitly, Knidos (i 32d-e). Perhaps, then, we should read the fashion for double-barreled handle amphoras in Kos, Knidos, and Halikarnassos in the light of production of wines that, while differentiated by consumers (or at least con-

1 Poblome-Brulet 2005, pp. 33-35, quotation at p. 34.

2 Briese 2005, pp. 189-191 for the ten fabrics, all Hellenistic except no. 10; p. 191 for the groupings:

Fabrics no. 1-3, Koan; no. 5-6, Knidian; no. 8-9 and possibly no. 10, Halikarnassian.

identifying regions in a greco-roman mediterranean context 69

sumers with refined taste) on the basis of flavor, effect, and price, nevertheless shared

a similar formulation: that is, constituted a regional type of sea-water flavored wine. If this suggestion is correct, we can see here a match of product and container in a context of regional production that distinguished this southwestern Asia Minor product from other wines in a market that extended well beyond the region where they were made.1

What makes a region?2

In the examples we have seen so far, regions have emerged from distribution of ma- terial goods – coins or pots. In some cases these regions are so small, like the appar- ently highly restricted distribution of the «Aradippou goblet», that we may hesitate to see the territory as a «region» and may opt instead to call the distribution «local». In others, like the Naxian kraterophoric coins, the distribution seems to mark out a genuine region. But all this is very vague. Assuming that the regions identified are re- al, what actually makes them a region? Is it merely the distribution of a type of pot- tery or coin? Or is there something more going on? Can we unravel the strands that weave a region? In classic regional geography, the basic structural feature of a region was its phys- ical geography. A region was regarded as a unit composed of some physically defin- able territory: a basin surrounded by mountains, like the Great Divide Basin of cen- tral Wyoming in the United States, or drainage of a major river, like the Hudson River valley in New York State. Some of the regions we have seen might be defined in this way. The distribution of the Aradippou cup, for example, is restricted to the south- eastern corner of Cyprus and the Famagusta peninsula, cut off from the rest of the island by the Kyrenia mountains to the north and the west-central massif of the Troo- dos mountains.3 So, perhaps, the geography of the island «created» this regional unit,

a cup to contain the Aradippou goblet. But there is another striking feature to the distribution of the Aradippou goblet:

the sites at which it has been found are all on, or very close to, the sea. It is well known that shipment costs in antiquity rose dramatically when products had to be moved by land instead of water, so perhaps rather than a land-based region defined by moun- tain and plain, the physical feature regulating the regional distribution was the trade route. Trade routes themselves were, at least sometimes, set by physical conditions, as in the famous Rhodos-Alexandria route open all year round thanks to cooperative winds and safe seas.4 The distribution of the Naxian kraterophores might follow the same pattern of a region created by well-traveled lanes of trade. Two issues arise with these suggestions. In the first place, the «region» may simply be an artifact of research: perhaps Aradippou goblets will turn up in the future well inland on Cyprus, demolishing the idea that sea trade determined the boundaries of the region; or they may start appearing in numbers outside Cyprus, and show a dis-

1 See Cato, De agr. cult. 112 on imitating Koan wine – a passage I will return to in the paper Formation of Taste and Fashion mentioned at p. 53 n. 2.

2 I will discuss the questions raised here with much more extensive documentation in Reger forth- coming.


gary reger

tribution more like Eastern Sigillata A (more on this below). In the second place, the danger of circular reasoning lurks – are we defining a region based on the distribu- tion of pottery, then using that region to predict…the distribution of pottery? This question gains acuteness when we start looking at more than one type of ob- ject. Let’s assume, for the sake of argument, that the distribution patterns identified for Aradippou goblets and Eastern Sigillata A in fact represent the actual patterns we would find if we could recover every object ever made. Clearly, though both ceram- ic types share southeastern Cyprus as a zone of distribution, Eastern Sigillata A pen- etrated much farther; it occurs from western Greece and Cyrenaica to Palestine and Syria. What governs this difference? It can’t simply be geography or topography or the patterns of trade, for the Eastern Sigillata A found in southeastern Cyprus re- produces the distribution of the Aradippou goblets, and presumably because the same forces are distributing both products, yet for some reason the Aradippou goblet failed to follow Eastern Sigillata A beyond southeastern and eastern Cyprus. And, in the case of Eastern Sigillata A, what part of its distribution should be regarded as «re- gional» and what part as «long-distance» or «interregional?» These are vexing ques- tions with perhaps no obvious or easy answers. Another mechanism for creating and distinguishing regions, now again being rec- ognized by scholars of contemporary regional economics, is polity. The Hellenistic and Roman worlds of the eastern Mediterranean remained always worlds of the po- lis, or asty and chora, distinct from neighbors. The Hellenistic period saw, of course, the elaboration of means of breaking down polis barriers, like sympoliteia, isopoliteia, and synoikismos (not that these institutions did not exist beforehand) and the estab- lishment of federal states, like Aitolia or the Achaian League. But these institutions did not necessarily spell the end of local hostility to neighbors,1 even when they may have facilitated a regional identity. It is difficult, though, to see any political forces shaping the regions in which circulated the objects we have examined – especially giv- en that the supra-regional political authorities in different parts of the distribution area of a given good may be not just different, but even hostile. How about not polity but ethnicity? This is a topic that has attracted enormous at- tention in recent years, both on a diffuse, global level – identifying, for instance, the components of a Greek ethnic identity against non-Greeks during the second and third centuries ad2 – and on a much more restricted level, as in exploring (as I have recently done) the possible ingredients of a Karian identity in southwestern Asia Minor.3 In the case of Cyprus one bumps up immediately against the same objection:

to the extent there was a shared Cypriote identity, it cannot explain either the distri- bution of Eastern Sigillata A or the Aradippou goblet; nor can appeal be made to some putative «Phoenician» identity under girding or separating certain Cypriote poleis from others.4 Ethnicity, however, may suggest another approach: culture. Not in the strictly an- thropological sense, but in the broader, quotidian sense of practices, behaviors, fash-

1 See the agreement between Xanthos and Myra in Bousquet-Gauthier 1994, with its many rules and

regulations hedging about the right of citizens of one polis to register as citizens of the other, despite an isopoliteia agreement. Briefly on these institutions, Reger 2005, pp. 345-346.

2 Romeo 2002; for a recent critique of the application of the concept of ethnicity to the Greek world,

identifying regions in a greco-roman mediterranean context 71 ion, and taste. This is not the place to enter into a long discussion of the sociology of taste and fashion, a pursuit that has been carried on with some considerable sophis- tication.1 Suffice it to say that there are plenty of indications that taste and fashion served in the Greco-Roman world, as in ours, to steer people toward particular ob- jects, products, and practices, and to drive them to adopt (or adapt) them, even some- times in the face of potent ethnic or religious sanctions. For instance, the Hellenizing Jews who adopted the gymnasion and Greek athletic practices in Jerusalem in the run- up to the Maccabean revolt, ashamed of their circumcised penises, wore imitation foreskins, àÎÚÔ‚˘ÛÙ›·È, «in accord with the customs of the Greeks,» ηÙa Ùa ÓfiÌÈÌ· ÙáÓ âıÓáÓ.2 This practice outraged many of their fellow Jews but the Hellenizers persisted in the desire to act like Greeks and be accepted by the people who held the reigns of power. This phenomenon shows itself with especial clarity in the stresses in Jewish society from the Maccabean revolt through the Jewish War of 66-71 ce, and probably later, too.3 In milder form it appears in the adoption by Egyptians of Greek names in the Ptolemaic period or the spread in Gaul in the second century of a taste for Italian wine. That is to say, that it may be useful, along side the other mechanisms we see as defining a region, or perhaps better the regional distribution of an object, to consid- er also the birth, spread, and adoption of a fashion. A new fashion may arise in many ways and may succeed (or fail) for many reasons; «early adopters» may or may not be socially respected, the fashion may or may not offend religious beliefs, it may be too expensive, or it may be overwhelmed by another, more appealing trend. A new tech- nology may bring previous luxury goods into the realm of poorer people, as the in- vention of blown glass seems to have made small glass objects affordable to a wide public and spelled the decline of the ceramic imitations used before. But whatever the precise history of any given fashion, there are certain ineluctable conditions for fash- ion to emerge: there must be change; there must be social differentiation (so that it matters what you use); and there must absolutely be mechanisms whereby people come into contact. Broadly speaking, the first condition is technological, or a conse- quence of invention or experiment; the second is social; and the third, when operat- ing beyond a purely local sphere, demands mobility.

Region outside the economy

In the broadest sense, then, «region» is a cultural phenomenon, not just, or even nec- essarily primarily, an economic one. Taste and fashion must have operated to spread or retard popularity of a good or practice, sometimes within constraints placed by ge- ography, topography, or ethnicity, but sometimes not. This analysis may also suggest that regionalism may appear in non-economic cultural expressions. A possible exam- ple is the Second Sophistic, as discussed by Ilaria Romeo in a stimulating recent arti- cle. Romeo argues that two definitions of «Greekness» competed among the sophists to pick out persons suitable as carriers of Greek culture. One was educational, that «Greekness» derived from acquisition of Greek literary culture, a view she attaches

1 I will discuss this matter in detail the studies referred at p. 53 n. 2; for now, as an example of such a study, I will point only to Gronow 1997, a stimulating essay but with no direction connection to Greco- Roman antiquity. 2 I Macc. 1.15. 3 See Goodman 1988 for some trenchant discussion.


gary reger

particularly to the sophist Favorinus. The other was ethnic, that «Greekness» derived from descent from Greeks; Romeo offers the sophist Polemon as the most ardent sup- porter of this view. This argument is interesting from the point of view of ethnicity and regionalism, and I will address that question elsewhere. For our purposes here, another aspect of Romeo’s discussion, in which she follows V. A. Sirago, is more ger- mane. Virtually all the great sophists of the first two generations of the movement came from western Asia Minor, indeed essentially the Roman province of Asia, and the locus where the movement was born and first flourished was Smyrna itself. Lat- er it spread, and Athens in particular became the most important center. But at its ori- gin, the Second Sophistic looks very much like a regionally-based cultural movement – that is to say, another example of regionalism in culture quite different from the eco- nomic regionalism we have been exploring, but a form of regionalism nonetheless.1

Some final thoughts

Obviously, the suggestion that the Second Sophistic should be seen as a regional phe- nomenon at its start requires testing and much deeper consideration, and should per- haps, at least initially, be thought of as a heuristic device or a metaphor. But I think it at least hints that regionalism may be a robust producer of tracts distinctive in many cultural features, both material, like artifacts, and immaterial, like philosophical views. It becomes very intriguing to ask whether there may have been links between not only these different types of regional «product», but also between a broader eth- nic or religious or cultural identity and regional distributions and production of goods. The shapers of regional identity, then, can be sought in geography, patterns of trade, polity, ethnicity, culture; and the mechanisms whereby new products, new fashions, moved within a region, and between regions, discovered in the ways that people came together: in the case of local entities, institutions like fairs and markets, for regions in festivals, for the whole eastern Mediterranean, pan-Hellenic celebra- tions, wars, long-distance trade. None of this is meant to be exhaustive, but rather a glimmer of ways in which some of the richness of the Greek eastern Mediterranean world can be recovered through the study of objects, ideas, and their relations to space and time.2


Ashton-Reger 2006 = R. Ashton, G. Reger, The Pseudo-Rhodian Drachms of Mylasa Revis- ited, in Agoranomia. Studies in Money and Exchange Presented to John H. Kroll, Peter G. Van Alfen ed., New York, 2006, pp. 125-150. Bousquet-Gauthier 1994 = J. Bousquet, Ph. Gauthier, Inscriptions du Létôon de Xanthos, «Revue des études grecques», 107, 1994, pp. 319-361.

Romeo 2002, pp. 33-36, following Sirago 1989, pp. 40-43; see also Gleason 1995 and for a critique Ruby


I am very grateful to John Lund not only for the invitation to contribute to «Facta» and his patience in

waiting for receipt of the text, but even more so for many years of friendship and profitable discussion. I have benefited time and again from his expertise and sharp intelligence. I appreciate his reading through this paper and commenting on it; any errors are charged of course to my account alone. I am also appre- ciative of the award by my home institution Trinity College of a Charles A. Dana Research Professorship, which helped support work on this paper.



identifying regions in a greco-roman mediterranean context


Berg Briese 2005 = M. Berg Briese, Halikarnassian Wine-production? The Evidence from Two Households, in Trade Relations in the Eastern Mediterranean from the Late Hellenistic Period to Late Antiquity: The Ceramic Evidence, eds. M. Berg Briese, L. E. Vaag, Odense, 2005, pp. 184-201. Gleason 1995 = M. W. Gleason, Making Men. Sophists and Self-Presentation in Ancient Rome, Princeton, 1995. Goodman 1988 = M. Goodman, The Ruling Class of Judaea. Origins of the Jewish Revolt against Rome, a.d. 66-70, Cambridge, 1988. Gronow 1997 = J. Gronow, The Sociology of Taste, London-New York, 1997. Jefremow 1995 = N. Jefremow, Die Amporenstempel des hellenistischen Knidos, Munich, 1995. Knapp-Mac Isaac 2005 = R. Knapp, J. D. Mac Isaac, Excavations at Nemea iii. The Coins, Berkeley, 2005. Lund 1999 = J. Lund, Trade Patterns in the Levant from ca. 100 bc to ad 200. As Reflected by the Distribution of Ceramic Fine Wares in Cyprus, «Münstersche Beiträge zur antiken Handelsge- schichte», 18.1, 1999, pp. 1-20. Lund 2005a = J. Lund, An Economy of Consumption: The Eastern Sigillata A Industry in the Late Hellenistic Period, in Making, Moving and Managing. The New World of Ancient Economies, 323- 31 bc, eds. Z. H. Archibald, J. K. Davies, V. Gabrielsen, Oxford, 2005, pp. 233-252. Lund 2005b = J. Lund, Introducing the Aradippou Goblet: A Case Study of Regional Diversity in Late Hellenistic and early Roman Cyprus, in Trade Relations in the Eastern Mediterranean from the Late Hellenistic Period to Late Antiquity: The Ceramic Evidence, eds. M. Berg Briese, L. E. Vaag, Odense, 2005, pp. 68-82. Marcellesi 2004 = M.-Ch. Marcellesi, Milet des Hécatomnides à la domination romaine. Prac- tiques monétaires et histoire de la cité du ive au iie siècle av. J.-C., Mainz am Rhein, 2004. Nicolet-Pierre 1999 = H. Nicolet-Pierre, Les cratérophores de Naxos (Cyclades): emissions monétaires d’argent à l’époque hellénistique, «Revue numismatique», 154, 1999, pp. 95-119. Poblome-Brulet 2005 = J. Poblome, R. Brulet, Production Mechanisms of Sigillata Manu- factories. When East Meets West, in Trade Relations in the Eastern Mediterranean from the Late Hellenistic Period to Late Antiquity: The Ceramic Evidence, eds. M. Berg Briese, L. E. Vaag, Odense, 2005, pp. 27-36. Reger 2005 = G. Reger, The Economy, in A Companion to the Hellenistic World, ed. A. Erskine, Oxford2, 2005, pp. 331-353. Reger 2007 = G. Reger, Karia: A Case Study, in Regionalism in Hellenistic and Roman Asia Minor, eds. H. Elton, G. Reger, Bordeaux, 2007, pp. 89-96. Reger forthcoming = G. Reger, Inter-Regional Economies in the Aegean Basin, in Demand Creation and Economic Flows, eds. Z. Archibald, J. Davies, V. Gabrielsen, forthcoming. Robert 1951 = L. Robert, Études de numismatique grecque, Paris, 1951. Romeo 2002 = I. Romeo, The Panhellenion and Ethnic Identity in Hadrianic Greece, «Classical Philology», 97, 2002, pp. 21-40. Ruby 2006 = P. Ruby, Peuples, fictions? Éthnicité, identité ethnique et sociétés anciennes, «Revue des études anciennes», 108, 2006, pp. 25-60. Schneider 1996 = G. Schneider, Chemical Grouping of Roman Terra Sigillata Finds from Turkey, Jordan and Syria, in Archaeometry 94. Proceedings of the 29 th International Symposium on Archaeometry, eds. S¸. Demirci, A. M. Özer, D. G. Summers, Ankara, 1996, pp. 189-196. Sirago 1989 = V. A. Sirago, La seconda sofistica come espressione culturale della classe dirigente del ii secolo d.C., in Aufstieg und Niedergang des römischen Welt, vol. 33.1, Berlin-New York, 1989, pp. 36-78. Zimmermann 1992 = M. Zimmermann, Die lykischen Häfen und die Handelswege im östlichen Mittelmeer: Bemerkungen zu PMich I 10, «Zeitschrift für Papyrologie und Epigraphik», 92, 1992, pp. 201-217.


gary reger


Il contributo riflette su alcune definizioni di natura squisitamente economica utilizzati spesso in archeologia. Gli archeologi utilizzano talora la definizione di distribuzione regionale della cultura materiale o di particolari manufatti individuando spesso alcuni modelli all’interno dei quali inquadrare il tema trattato. Gli economisti hanno spesso discusso sul concetto di regio- nalismo come base di partenza per analizzare i movimenti di beni e di capitali, le relazioni commerciali e lo sviluppo di istituzioni economiche. Ma il concetto di «regione» rimane an- cora oggi non facile da definire. In che modo noi siamo in grado di distinguere la distribuzio- ne di uno specifico manufatto al di fuori di una regione ben definita? Quali sono le relazioni tra le diverse regioni, e tra entità locali e sovra-regionali? Le regioni mantengono caratteristi- che fisse o esse mutano i loro confini nel tempo? Questi quesiti inducono alla riflessione aiu- tando ad individuare alcune possibili percorsi interpretativi.

The «region» has been a staple of analysis in archaeology and economics for a very long time. Archaeologists write about regional distributions of material culture or particular artifacts, and, as we will see below, about regional production schemes. Economists (not just econom- ic historians focused on Greco-Roman antiquity) have found regionalism a useful basis for analysis of movement of goods and capital, trade relations, and the development of econom- ic institutions. But «region» as a concept remains not always easy to define. How do we know when the distribution of an artifact marks out a region? What are the relations between dif- ferent regions, and between local and supra-regional entities? Are regions fixed or do they change shape and boundaries over time? These fundamental questions repay reflection, even if, in the end, no firm and final answers can be offered.

LA FUSIONE, AFRODITE E L’EMPORION Lucio Fiorini · Mario Torelli

L o studio sistematico dei dati relativi alle campagne di scavo realizzate negli anni settanta a Gravisca ha permesso di precisare le cronologie di età arcaica del

santuario, ricostruendo una sequenza di quattro fasi costruttive comprese nel lasso di tempo di un secolo, tra il 580

e il 480 a.C., quando si con-

centra il maggior numero di testimonianze relative alla lavorazione del metallo, de- stinate invece quasi a scom- parire con il venir meno del carattere emporico del san- tuario1 (Fig. 1). La necessità di un approdo protetto nell’ambito di una navigazione a piccolo cabo- taggio e la possibilità di poter attingere a risorse di acqua facilmente reperibili, sta al- l’origine della prima frequen- tazione dell’area sacra tra la

della prima frequen- tazione dell’area sacra tra la Fig. 1. Pianta dell’area sacra di Gravisca con

Fig. 1. Pianta dell’area sacra di Gravisca con indicazione dei vecchi e dei nuovi scavi.

fine del vii e gli inizi del vi se- colo a.C., concentrata lungo tutta la fascia sud-orientale del futuro santuario, dove era situato un pozzo protetto esternamente da un recinto

di pali e graticci (Fig. 2).2

Alla fondazione dell’empo- rion da parte di Greci di pro-

1 Sul santuario di Gravisca si vedano: Hanoune 1970; Torelli 1971; Torelli et alii 1971; Tronchetti

1973; Boitani 1974; Gianfrotta 1975; Torelli 1977 (a); Torelli 1977 (b); Boitani 1978; Comella 1978; Slaska 1978; Torelli 1978 (a); Torelli 1978 (b); Torelli 1979; Tronchetti 1979; Solin 1981; Shuey 1981; Torelli 1981; Pandolfini Angeletti 1982; Frau 1982; Slaska 1982; Torelli 1982; Boitani-Slaska 1984; Moretti 1984; Boitani 1985 (a); Boitani 1985 (b); Ehrardt 1985; Frau 1985; Slaska 1985; Torelli 1985;

Boitani 1986; Torelli 1986; Torelli 1988; Boitani 1990; Torelli 1990; Pianu 1991; Gravisca 9; Visonà 1993; Boitani 1994; Torelli 1997; Gravisca 4; Boitani-Torelli 1999; Gravisca 10; Gravisca 15; Gravisca 12; Fio- rini 2001; Fortunelli 2001; Fiorini 2002; Torelli 2003; Gravisca 5; Gravisca 11; Gravisca 16; Torelli 2004; Gravisca 1.1; Fiorini 2005; Fiorini 2006; Fortunelli 2006; Torelli 2006; Gravisca 1.2.

2 Cfr. Torelli 2003, p. 28. Si tratta del primo dei sette pozzi apprestati ad est del sacello di Afrodite

durante il periodo arcaico, quasi tutti associabili a contesti metallurgici; a Gravisca sono stati rinvenuti in tutto dodici pozzi.