This essay starts from the initial presupposition that the landscape is more than the proverbial backcloth to human action found so often in archaeological writing. It is not just the natural feature of the environment putting important constraints on human behaviour, as a popular ‘naturalist’ tradition in archaeology would like to have it, nor is it simply a cultural construction of human conceptions about the world, as its ‘culturalist’ counterpart is advocating (cf. Descola and Pálsson 1996, 2-3). Rather the landscape is doubly-faced, consisting of a foreground actuality of every-day existence and a background potentiality of an imagined timeless ideal (Hirsch 1995). The phenomenal and the imagined landscape are alternately perceived and inextricably intertwined, in that people continuously seek to realize in ordinary life the ideals of an imaginary existence. The ways in which they imagine the world may thus heavily bear on the appearance and organization of the landscape they live in. This is true for all societies in past and present. But while in the modern west the ideals have up to a point a secular character, in all other societies, including those of Roman Gaul, the imagined landscape takes the form of a cosmology.

Landscape is a theme with a long standing in archaeology. But whereas archaeologists in the 1960s analyzed the carrying capacity of the environment in order to measure human adaptation, since the 1990s landscapes are increasingly being ‘read’ for understanding how people related to the land. The shift in perspective is clearly a substantial one which does not merely demonstrate a development of archaeological thought: as related developments in human geography and anthropology show, it is rooted in a wider current in the Western world of rethinking fundamental notions such as ‘nature’, ‘culture’ and ‘society’. The concept of ‘land-scape’ or perhaps more generally that of ‘space’ also fits in this series, as was argued by among others Michel Foucault, who claimed that ‘the present epoch will perhaps be above all the epoch of space’ (Foucault 1986, 22).

Car j'ai découvert une grande vérité. A savoir que les hommes habitent, et que le sens des choses change pour eux selon le sens de la maison. Et que le chemin, le champ d'orge et la courbe de la colline sont différents pour l'homme selon qu'ils composent ou non un domaine. (A. de Saint-Exupéry, 1948: Citadelle, 28).

The Chinese curse – ‘May you live in interesting times’ – has unquestionably fallen upon those involved in landscape studies, as many papers previously published in this journal attest. Derks' contribution adds an attractive case study, drawn from the Roman world rather than from the more frequently visited prehistoric periods, to this ferment. Derks reviews recent interpretive arguments, before turning to the landscapes of northern Gaul: landscapes in the plural, for in his study area Derks teases out the existence of two zones which develop in distinctly different ways. His major achievement is to divert explanations for this development from the more usual channels of thinking, such as pondering economic and ecological variables, or categorizing areas as ‘Romanized’ or ‘non-Romanized’. His addition of social values and cosmological principles as ‘filters’ helping to determine the trajectories followed by these two zones is bold and, whether or not one agrees with the details of his arguments, his basic contention – that such elements are just as vital as imperial policies and environmental conditions in determining the creation of provincial landscapes – cannot be challenged. I predict that within five years a wide variety of such locally sensitive, landscape-based case studies will be available from across the Roman world, leading to fresh analyses of provincial variability within the empire and to cross-regional comparisons no longer strictly determined by administrative boundaries.

While feeling in tune with most of Lemaire's theses, I disagree on some points which are not altogether secondary.

The project presented by Ton Derks seems interesting to me, not only because he offers new data on the northern provinces and their acculturation, but particularly because in studies on the Roman world this sort of analysis is seldom done systematically. To me, his article provokes the following three remarks.

The notion of the landscape has become a central issue in anthropological debates (Hirsch 1995). By comparing different conceptualizations of the relationship between people and land we can get a better understanding of our own categories of nature and landscape. But the notions of nature and landscape are not a universal categories. Each society has its own categories to conceptualize the land. Notions of landscape are always embedded in cosmological notions. Burial places, river-crossings, or mountains are not neutral places. They have meanings and values. A landscape derives its meanings and values from a wider whole encompassing the earth, the sky, the underworld, and the sea. It is populated by people, animals, imaginary beings such as dwarfs, spirits, spirits of the dead. It is also embedded in temporal categories. It has a history. Each place derives meaning from its history. Therefore a landscape is never objectively given, it is a social construction.

Archaeology, anthropology, human geography: three disciplines born out of a nineteenth-century imperative among Europeans to apply a coherent model of understanding (Wissen-schaft) to varied forms of social life within a differentiated physical world; three disciplines stretched between the epistemology and methods of the natural sciences (Naturwissenschaften) which promised certainty, and the hermeneutic reflexivity and critical doubt of the Humanities (Geisteswissenschaften) which promised self-knowledge. Each of these disciplines is today in crisis, and for the same reason. Europe as the place of authoritative knowledge, of civilization, has been decentred upon a post-colonial globe; the white, bourgeois European male has been dethroned as the sovereign subject of a universal and progressive history. Thus, the enlightened intellectual project represented by archaeology, anthropology and human geography, whose findings were unconsciously designed to secure the essentially ideological claims of liberal Europeans, are obliged to renegotiate their most fundamental assumptions and concepts (Gregory, 1993). The linguistic turn in the social sciences and humanities which has so ruthlessly exposed the context-bound nature of their scientific claims — what Ton Lemaire refers to as a critical awareness of their inescapable cultural and historical mediation — forces a recognition that their central conceptual terms, such as ‘culture’, ‘nature’, ‘society’, and ‘landscape’, are far from being neutral scientific objects, open to disinterested examination through the objective and authoritative eye of scholarship. They are intellectual constructions which need to be understood in their emergence and evolution across quite specific histories. Ton Lemaire seeks to sketch something of the history of landscape as such a socially and historically mediated idea: as a mode of representing relations between land and human life, which has played a decisive role in the development of archaeology as a formal discipline. On the foundation of this history he develops a critique of the social and environmental characteristics and consequences of modernity, and seeks to relocate archaeological study within a reformed project of sensitive contemporary ‘dwelling’ on earth.

Lemaire's reflections on the concept of landscape struck a chord with me, and I found myself nodding in agreement with much of what he has to say. I did, however, have some difficulty in coming to terms with the idea that the landscape is wholly an invention of bourgeois sensibility. Moreover, in his emphasis on the landscape as an object of contemplation, ‘totalized by a human gaze’, I feel that he neglects the activity of depiction – whether this be painting, drawing, writing or musical composition – in favour of the end products, finished works. I should like to focus my comment on these two points.

This paper is of fundamental interest to those working on the archaeology of the Roman provinces because it moves us a considerable way forward in relating two previously disparate branches to study to one another. Until now it has been almost impossible to find serious studies which connect settlement and landscape with religion. By drawing on the impressive recent tradition of Dutch landscape archaeology within the Roman period and systematically using the extensive epigraphic evidence for religion Derks provides not only new insights into the nature of Roman provincial society but also a soundly based study which stands well beside the work of prehistorians like Bradley. Derks paper is thus of great interest and should form a model for the study of other regions. I would be particularly interested to see whether the distribution of dedications to Mars and Hercules was similarly structured in other areas, especially in Mediterranean areas.

I wish to thank Augustin Berque, Denis Cosgrove, and Tim Ingold for their critical and stimulating comments on my article. I shall discuss the main arguments in their reactions in the order I received their paper respectively. To begin with, I agree with Cosgrove that the relation between landscape and modernity is ‘more complex and ambiguous’ than in my summary and I agree further with him, quoting Rose (1993), that the viewing subject in its distanced and proprietorial relationship with the environment is mostly a male subject. At least since Merchant's Death of nature (1980) we are aware of the gendered nature of the modern scientific approach of nature. But although it is true that there are ‘several histories of nature’ and that many details may be put forward to complete my view, I nevertheless contend that the main lines in my short history of ‘the invention of the landscape’ are basically right.

I would like to thank the discussants of my essay for their pertinent questions, sincere criticisms and useful suggestions. Though I shall try to remove some of their hesitations to accept my interpretation of the observed patterns and correlations, it is already clear in advance that I shall not succeed in convincing each of them on all points on which we disagree. For while Oosten is defending the proposition that ‘[t]he phenomenal landscape can only be described in terms of an imaginary landscape’ (my italics) and thus seems to deny the existence of two modes of perceiving alternately actualized, Scheid, on the contrary, qualifies my line of inquiry as ‘justified’. According to him, the Ancients had ‘apart from a day-to-day perception of the environment, (…) also a “mythical” representation of space and landscape’. Moreover, while Scheid believes that the present author ‘convincingly demonstrates how in his research area the choice of Mars or Hercules as sovereign deities was linked to the economic model imposed by the landscape’, Oosten judged me to be trapped by ecological determinism. Such contrasting views, which are partly related to different traditions of research, cannot be expected to be solved here, though I shall not avoid to discuss them. I only hope that with respect to other points of the dispute my efforts will have a better chance to persuade the participants of this dialogue. I will begin by answering the questions and criticisms which directly refer to the understanding of my paper and leave my reply to some of the suggestions made for future investigations to the latter part of this response.

This paper results from a seminar and discussion series at Lampeter which was triggered by concern over the increasingly common view presented in archaeological literature which equates a caricature of‘post-processual’ archaeology with an equally naive understanding of relativism. In particular, it was noted that the recent book edited by Kohl and Fawcett (1995) contained many statements and editorial insertions which took this approach. Through a series of presented short papers and discussion we decided to explore why we were unhappy with these characterisations of both relativism and post-processual archaeology. These seminars were then summarised as a consensus document and associated comments (this present paper), which have themselves been through several stages of a joint editing process and further discussion (see also Hamilakis 1996; Holtorf 1996 and Pluciennik 1996 for related comments). We should make it clear that we are not unaware of far more sophisticated approaches to relativism within archaeology (e.g. Fotiadis 1994; Pinsky and Wylie 1989;Wylie 1992, 1993, 1994), but that we are here concerned with addressing a specific, and we believe inaccurate, viewpoint which seems to be gaining increasing currency.

Archaeologists are prone to retrospection. In 1974 Gordon Willey edited a book in which a series of writers looked back over their most famous projects. Some of their contributions were both revealing and frank but in other cases, I suspect, the authors were trying to write themselves into history. It can be dangerous to retrace our steps over the same intellectual ground, but we may still want to attempt this so that we can come to terms with unfinished business. We study some problems which will not let us go. This paper considers a typical example.

Originally my involvement with the stimulating Lampeter Archaeology Workshop paper (hereafter LAW for short) would have been restricted to just another anonymous referee report. However the editors changed their plan and asked me to write a short commentary. I agreed to this change of plan so what follows is basically an expanded version of my initial remarks. (Parenthetically, I belief this format is far better than the old fashioned style referee system were referees can hide behind the protective shield of anonymity.)

The transformation between the two landscapes of Cranborne Chase, the earlier characterized by ritual and ceremonial monuments the later by an agricultural landscape, encapsulates the most significant transformation in the later prehistory of Britain. Indeed, the emergence of intensive agricultural practices which employ a wider range of crops, achieve increasing levels of crop purity, enclose and manage the land according to new patterns of territoriality, and establish long-lived nucleated settlements, is a more general feature of the European Bronze Age. It was upon this foundation that the agricultural practices of the Iron Age were embedded.

It is one o clock in the morning of December 20. Buffalo is cold, five degrees below zero Fahrenheit with a wind chill of minus twenty five degrees. The Niagara River is frozen solid except at the Falls. Sounds travel far on these cold nights. I hear a loud crash as if the ice is beginning to crack. I fantasize about the spring and the ice breaking up into thousands of small icebergs that float down the river. Icebergs bring to mind GIS software. Like icebergs, GIS software is frequently attractive and even starkly beautiful. Yet GIS systems are sometimes foreign and can be very inhospitable. As in the case of those frozen fantasies, GIS software may look better from a distance than close up. (Zubrow 1990b, 184)

The ideas which scientists use to present the known and to advance into the unknown are only rarely in agreement with the strict injunctions of logic or pure mathematics and the attempt to make them conform would rob science of the elasticity without which progress cannot be achieved. We see: facts alone are not strong enough for making us accept, or reject, scientific theories, the range they leave to thought is too wide; logic and methodology eliminate too much, they are too narrow. In between these two extremes lies the ever-changing domain of human ideas and wishes. (Feyerabend 1975, 303)

I very much welcome both the medium and the message of the Lampeter seminar article. Fear of the ‘unscientific’ unknown has long prevented an adequate discussion of relativism in archaeology. Anxiety about the relativist abyss in which ‘all pasts are equal’ has thwarted efforts to evaluate alternatives to ‘objective science’. The dialogical style of the article opens up the debate on relativism without imposing rigid and narrow perspectives. The message itself is greatly enhanced by the helpful distinctions between ontological, epistemic and moral relativism.

Milco Wansleeben provides a useful review of some of the current problems of GIS, and I find much agreement with his comments. As he states most effectively, GIS provides the ability to execute repetitive routes rapidly. When practised well, it will not save time but allow the more thorough interrogation of the data within the same time.