1The notion of transition appears indispensable when changes to a system require categories not used to describe the initial state of that system. Even if that initial state could be identified with precision, by determining the system’s limits in relation to its ‘environment’ and specifying its components and their interactions, doubtful points often persist in the margins of these definitions. For example, what changes should be described as ‘system state changes’ versus ‘system changes’? Similarly, when considering factors that might trigger transitions in the system, which should be thought of as exogenous perturbations versus amplification of internal fluctuations? Since we are tasked with formalising observed transformations in settlement systems, we will review some definitions of ‘transition’ and sketch out various applications of this concept in several disciplines.

The concept of transition: tacking between natural sciences and social sciences

2In our consideration of transformations of societies over the course of long-term history we have chosen to focus on settlement systems. Such systems are interesting both from the perspective of their durability in time, and their ability to reflect in space a variety of social models. Settlement is defined as the spatial distribution of a population, materially signaled by more or less long-lasting constructions utilised at least for night-time shelter, which are subsumed within the general concept of habitat. This concept is also employed in ecology but takes on broader connotations in the social sciences.

3Provision of habitat is in fact one of the five or six major universal functions identified by geographers for settlement systems in all world societies, whatever their political structure, mode of production, or culture. These major functions also include appropriation (cadastres and/or land use rights, identification by place-names, symbolic representations, and so forth); exploitation or utilisation of local resources (soil, air, water, etc.); circulation (road networks and other transportation infrastructure); and administration (for management and control of a territory). Habitat itself includes all forms of construction for residential use or for other kinds of activity such as storage or production. These five functions common to all spatial systems have been identified in nearly the same terms by Philippe and Geneviève Pinchemel,1, and Roger Brunet.2 One can identify these functions even in territories of nomadic populations; they are components of territorial organisation in all civilisations. For a given society at a given moment, these five functions are obviously not independent of each other: they are organised in a ‘spatial system’ that constitutes a kind of geographic representation of the territory, localised and mappable. The habitat and the networks of circulation constitute the visible web of settlement systems, whose general organisation they reflect. To analyse their transformation thus enables an analysis of those aspects of social change that societies durably imprint on their landscapes.

Formalising change

4A conception of the universe as in perpetual change has long been common to the Asian world3 but has also become prominent in the West in both the human and natural sciences. This seems to be a result of epistemological convergence around the paradigm of complexity. This paradigm of continual evolution makes it more awkward to think in terms of fixed categories and stable objects. These, though, are the habits of a mechanistic occidental science that has enabled considerable progress in our understanding of natural phenomena. This progress however has come at the cost of drastic paring down of system components and even more drastic pruning in our conceptions of temporality and duration, since these are the categories necessary for the analysis as fixed of objects that are, in principle, in motion.

5According to a constructivist interpretation of science, this new way of thinking about change offered by the sciences of complexity is advantageous because it facilitates transfer of concepts from one domain to another, at least with regard to cognitive representations facilitating comprehension. For example, it has been possible to borrow mathematical vocabulary for changes of state in physics and apply them to the social sphere, as in the case of René Thom’s theory of catastrophes or in that of the attractors4 of non-linear dynamic models – not forgetting the more inclusive metaphor of chaos theory. 5Even the physics of phase transitions6 introduces some uncertainty of duration into the state change associated with it, while the bifurcation7 of the system would re-establish, as a capacity of free-will (or of fate) in destiny, a historicity of the evolution of open systems which ‘puts an end to certitudes’8 regarding the exact predictability of isolated mechanical systems.

6These convergences help to circumvent the prejudice that tends to deny—as too reductive—the possibility of modelling complex social systems using the tools of mathematics and computation. The following chapters will demonstrate that all is a question of granularity, of level of resolution, in our focus. We will take a broad view that permits identification of ‘regimes’ over the course of the history of settlement systems, representing relatively stable configurations, interspersed by moments of upheaval in which the rhythm of change accelerates – moments which we call ‘transitions’ (cf. chapter 2).

Originality of social change

7The formalisation of changes in social systems is very largely guided by direct and detailed observation of multiple dimensions of social change, operating and interacting with fields of analysis generally conceptualised separately by each of the social sciences.9 In this process, the place of mathematics or computer simulations is generally auxiliary, but it is essential for measuring significant differences, discriminating among competing hypotheses and examining whether the explanations proposed are necessary and sufficient. If the human and social sciences have much to gain from these exchanges, as is proved by the following chapters of this volume, they must nevertheless affirm the specific features of the concepts by which they represent the transformations intervening in the social field.

8Indeed, one specific feature – both as a driving force and an expression of social change – requires acknowledgement as distinct from changes studied in the natural sciences. Innovations, understood here in a very broad sense and constituting socially accepted inventions, intervene in all the spheres of human activity, whether technical, economic, political, cultural…10 The creation of new elements marks out the historical time of human societies, according to a temporality which is not that immeasurably long of astronomical configurations nor the desperately short of chemical reactions. By reference to the time-scale of our planet, their tempo is also much shorter than that of plate tectonics or biological evolution. What is more, the interval between major innovations has shortened over the course of historical time: it has taken some hundreds of thousands of years to produce reflective consciousness, individual and collective, several tens of thousands of years, no doubt, to construct languages, but only a dozen millennia to organize sedentary habitats, then to invent writing and cities, and less than three centuries to propagate and generalize the ‘industrial revolution’ and the urban way of life. Since the first industrial revolution, the cycles bearing clusters of economic and technological innovation, as well as of ‘creative destruction’,11 often imagined as being of constant duration by those who cite Kondratieff,12 have in fact grown shorter. Very recently, the digital revolution has spread throughout the world in a matter of just decades. The acceleration of change by the emergence of artefacts, but also via innovative social and cultural practices, is so strong that some have proposed adjusting these shorter and shorter periods by log-periodic laws.13

9The identification of a transition thus presumes a qualitative judgement, often supported by quantitative measures, about a change. This involves evaluating its nature and duration, and its modalities of social function, perhaps modelled according to its institutional, material or abstract components, and in any case its spatio-temporal character. Settlement systems are the objects chosen here as revealing transformations occurring in the social systems that produce them.

An example of socio-spatial change interpreted in the framework of self-organisation theories

10One difference often highlighted between social change and biological evolution lies in the intentionality, the objective, of human actions producing a change. This factor certainly adds to the difficulty of foreseeing the future of the systems studied. It has nevertheless been possible to transpose the notion of self-organisation14 fairly directly in order to explain the structuring of settlement systems.15 This structuring, according to an architecture which is identifiable because it is persistent, can be interpreted as the more or less involuntary product of multiple interactions involving a great number of agents and forces. Each of these acts according to its own objectives and strategies, but it is not necessary to know each of those strategies in detail to reconstruct the general architecture of the system. The mechanisms of the change observed in settlement systems resemble those described by the theories of self-organisation in physics, which some have described as ‘order by fluctuations’ .16 In analysing the transformations of the system of French cities between 1954 and 1975, Denise Pumain et Thérèse Saint-Julien17 have demonstrated the link between a persistence or slow transformation of the structure of the system of cities on the macro-geographic level (indicated, for example, by the distribution of the sizes of the cities, the differentiation of their activities and of their social compositions…) and the fluctuations, sometimes very great, in the relative situation of each city considered on the micro-geographic level. Individual cities can, in fact, pass rapidly from states of growth to stagnation and demographic decline, while their overall size distribution remains stable.

11Fluctuations are still more apparent if one considers the most ‘microscopic’ level of geographic investigation, the individuals, households or businesses situated in the cities on a time-scale of one year (residential and professional moves, creations and closings of businesses), or at the level of the succession of generations. In this evolution, the authors have identified bifurcations as inflections in the economic specialisations of the cities. One old bifurcation reversed the urban attractiveness of the flourishing manufacturing cities in the nineteenth century, generally situated in the northern region of France, towards the now better-off and better educated cities, mainly located in the south, during the second half of the twentieth century. Another more recent bifurcation involved the emergence, during the ‘Glorious Thirty’ years, of cities such as Paris, most of the large metropolises, and the large cities of the Rhône-Alpes region that were better adapted to information society. This was coupled with the relative regression of cities more removed from electronic civilization (for example, certain centres in the Southwest, Périgueux and Angoulême, and many other small and medium-sized cities that were undercut by the attractiveness of the regional capitals).

12Theories of self-organisation and ‘synergetics’18 propose mathematical models to reconcile the rapid dynamic of the micro-level and the slow one of the macro-level. These models use differential equations to represent the temporal evolution of the state variables that define the macroscopic structure of the system, with the microscopic interactions accommodated by the parameters. The advantage of such models is that they make the formation of system structures from multiple interactions among their components intelligible. A number of these models have been used successfully to represent urban change.19 20These examples have not been incorporated in this volume, but they have given rise to a series of multiagent models using Simpop21. These models (we provide an example in chapter 7) employ innovation as a driving force for increasing hierarchy in settlement systems, and for their spatial spread.

Concepts for identifying and modelling transitions

13In what follows we will use the terms ‘transition’ and ‘regime shift’ as synonyms, as many authors do 2223. The first term lays more stress on the process of change itself, while the second refers to the regimes in place before and after the transition. ‘Regime’ describes a system with a characteristic set of functions and structure, which can evolve along a fairly stable trajectory. A regime shift characterises a system reorganisation in which the structure, functioning and feedback relations are different than in the previous regime. Why and how such a radical change can occur—often abrupt relative to the time scales considered—will be explained using theory drawn from non-linear dynamics and complex systems, applied to the evolution of settlement systems. In what follows, we present, in broad outline form, relevant frameworks from the socio-environmental sciences, then from computer science.

Different conceptualisations of system change in the socio-environmental sciences: transition, regime shift, alternative state change

14To understand why a transition has taken place in a given system, the first step is to conceptualise that system, its dynamic, and to define the change that corresponds to a transition. Such a conceptualisation varies according to the authors. For some, a transition corresponds to an evolution of the system towards an ‘alternative state’, while for others the transition manifests the disappearance of the system and the emergence of a new, different one with a new ‘identity’. The first case presupposes that different ‘alternative states’ (also known as ‘attractors’ or ‘stable states’) exist a priori and that the system may move from one to the other under certain conditions. This is the case for certain ecological systems – for instance, a lake that passes from an oligotrophic state (corresponding to clear water), where the concentration of nutrients depends wholly on the flow issuing from its watershed, to a eutrophic state (turbid water), where the internal recycling processes of the lake also play a role. Passage from the first state to the second, and vice-versa, depends on an assemblage of environmental and human factors.24 In the social field, an example of the first case is provided by the pre-Hispanic American Pueblo societies, which maintained two different, but fairly stable, relations between the size of the population and the degree of violence. Thus figure 1 illustrates a relatively stable attractor for the societies of the centre of the Mesa Verde between the years 600 and 1000 (in red), with low levels of population and violence. A second attractor is identifiable in the north of the Rio Grande between 1200 and 1500 (in blue),25 this time with a low level of violence associated with a significant population. Around each of these attractors, the form of the curves manifests the existence of negative feedback between the two dimensions, each of which, in fact, regulates the other. Positive feedback, by contrast, would have led to a concomitant increase in population and violence. The causes of these forms of change are explored in chapter 6.

Figure 1 Two relatively stable attractors for the relationship between violence and population size among prehispanic Pueblo societies, shown in their phase space.²⁶

15Such a conceptualisation is less relevant, on the other hand, to an evolving system registering irreversible changes. In that case, several a priori alternative states do not exist, and the change is such that it may entail the disappearance of the system. Here a new system, characterised by different functioning regulated by different kinds of feedback loops27, will emerge. To characterise such a case, certain authors signal that the system has ‘changed identity’. 2829

16The literature speaks often of an abrupt and radical change to designate a regime shift.30 3132These two notions are relative, however. With regard to the temporal aspect, a few decades constitute a short term when one is concerned with ecological regime shifts (involving climate, for example), while this same time span is extremely long when studying a change of opinions on the climate. Moreover, according to the scale chosen to represent time, the change will appear either abrupt or progressive. More decisive for identifying a regime shift is the fact that there exist two distinct dynamic states, before and after, each evolving in a durable form around a stable tendency. Figure 2 makes it possible to illustrate different situations corresponding to the shift in a system from one regime to another. The state of the system is represented, in simplified form, on axis Y, as if it were characterised by the value of a single variable (several variables are generally involved). Two states (which each correspond to one value of Y) are represented, and for each of them the values fluctuate over time around a central tendency. Faced with a change in its environment, the system can react in two ways: 1) the variable changes gradually in value as the environment changes (case a); or 2) the variable continues to fluctuate around the same central tendency, a situation illustrating a resilient system. Nevertheless, when one parameter crosses a critical threshold, the system changes abruptly and moves to another state (case b). The system subsequently pursues its evolution by fluctuating around a new central tendency. Whether the system goes through process (a) or (b), there has been a regime shift.

Figure 2 Schematic representation illustrating fluctuations around a stable state and the different ways of passing from one regime to another, either by a gradual change (case a) or an abrupt one (case b).

17The ‘stable states’ (or ‘attractors’) themselves evolve through time, changing position or even disappearing. In this case, once it has left it behind, the system cannot return to the same attractor (if it has moved), and in numerous cases, it will not return either to a variant of that attractor (if it has disappeared). Settlement systems, the subject of this volume, are most often of this type. For example, when new functions, such as craft-production or commerce, emerge in population units composed, up to this point, of farmers, they cause the emergence of new forms of interaction and transform the settlement system, usually leading it to a more hierarchical organisation. The new set of functions that moves into place then has every chance of lasting, and a return to the previous state is extremely rare. In that case, one may consider that the system has undergone a change of identity.

Conceptualisation of change in computer science: the engineering of change

18In computer science, modelling has as its primary objective to structure and specify as precisely as possible information relating to the data and the underlying processes of phenomena of interest. Terms referring to change are numerous: evolution, modification, mutation, transformation – all evoking the idea that an entity or a system passes more or less progressively from one state to another under the influence of a process. In this terminology, however, the temporal dimension is implicit, and the objective of the computer treatment is to render it explicit. In software engineering, the UML norm33 has been developed to describe the dynamic of a system on the basis of diagrams of state transition (figure 3), which represent the life-cycle of the elements (objects) of the system under study.

19The concept of “state” corresponds to a stage in the life of a stable and durable object. The initial state corresponds to the creation of the object and the final state to its disappearance. Activities are associated with the state, and these take place during the stage when the system is in the state under consideration (State 1 or State 2 in figure 3). The transition simply represents the passage from one state to another, which, in general, is set in motion following an event that plays the role of stimulus.

Figure 3 : Diagram of state transition (UML)

20The computer formalisation of a phenomenon of interest requires selecting the most appropriate conceptualisation of time and space. If time is considered to be a linear sequence with a past, present, and future, then change is a chronological succession of states. Time may instead be conceived as a cycle, to represent regularly recurring processes. Moreover, time may be subdivided into moments (discrete vision) or into periods (continuous vision), and the order of the relations between these moments and periods may be taken into account. Similarly, space, according to the case in question, may be conceived as a receptacle of delimited entities (plots, municipalities) or as a continuous field (mapping the temperature, for example). Qualitative and quantitative spatial relations (of topology, orientation and distance) may subsequently be defined between such entities.34 35

21By generalising the approach of Mireille Fargette,36 we can define a system on the basis of three concepts made explicit in UML diagrams: 1) the structure, which describes what the system is composed of; 2) its function(s), which refers to what the system does; and 3) the dynamic which determines what the system becomes.37 Structure and function are in fact indissociable and together define a system. Any structure has a form making possible or requiring a specific set of functions that together compose a system. By the same token, it is a regime of functions matched to a structural form that makes a system. The state of the system is produced by the functions acting on the structure. Every system S is plugged into an environment. The latter either provides a stable context for it or causes it to undergo change for reasons extrinsic to the system (climatic change, for instance). The dynamic (represented in figure 4 using UML conventions) takes account of the change, and it is through the dynamic that the notion of time ‘takes shape’, with the realisation of change of form, of regime, of state, indeed of system type. There is, with a temporal point of reference, a before (‘previous state’ corresponding to the instant t) and an after (‘following state’, in the instant t+1) The observer records and evaluates the change between the two states on the basis of the differentials in form, in regime, and in state. The researcher’s objective is then to conceptualise and implement the mechanisms underlying the functioning and the dynamic of the system. Mathematics and computer science furnish generic frameworks for the formalisation of change which make it possible to implement and put to work models of systems with varied components and temporalities, conceptualised in diverse contexts.

Figure 4 : System and Dynamic (UML norm)

Triggering factors: exogenous versus endogenous factors

22When a transition has been identified and described, the question is to understand what has triggered the system toward such a radical change. A number of factors may be at the origin of a transition, and, classically, one distinguishes exogenous factors, which imply the influence of elements not part of the system (often termed forcing mechanisms), and endogenous factors, which belong to the components and the functioning of the system itself. To render this distinction operational, it is essential to define the boundaries of the system properly. The system is, in fact, situated within an environment and is distinguished from it by interactions a priori more important within the system than with its surroundings. The former interactions refer to endogenous factors, the latter to exogenous ones. These two large categories account for the varied discussions in the literature.

23One speaks of an exogenous factor (or, alternatively, of a forcing of the environment) when a change in the environment in which the system is situated entails, in response, a transformation of the system itself. It may have to do with a sudden disturbance (tsunami, war, invasion, etc.) or with the gradual change of a parameter (soil degradation, increase in temperature) which, at a certain moment, crosses a critical threshold. Further reference to figure 2 will clarify the latter case. The environment of a system can be described by a collection of parameters. As long as these remain stationary, the system fluctuates around a certain state (indicated as State 1 in the figure). When the environment changes and one parameter begins to vary, the system can evolve in two ways. Case (a) illustrates a gradual change, which follows the rhythm of variation of the parameter, while in case (b) the system is maintained in State 1 during an initial phase, then changes in an abrupt manner. The system may also change suddenly and profoundly without any exogenous disturbance. The transition may thus also be the consequence of a multitude of interactions between the components of the system which results in an internal reorganisation of the system, manifested by a transition towards a new attractor.38 Faced with a given empirical situation, it is a question of determining what the factors are which have triggered the transition, and sometimes several different theoretical interpretations are possible. The triggering factors may combine, a gradual change in the system’s environment leading to the crossing of a critical threshold which provokes a sudden change.

24Marten Scheffer39 evokes the case of the Sahel as an illustrative example of a transition combining exogenous and endogenous factors interacting around a critical threshold. Until 6000 years ago, the west of the Sahara was a humid zone with abundant vegetation, while less than 1000 years later that zone is desertified. Researchers have been able to reconstitute, for the period from 9000 BP to the present, on the one hand, the amount of solar heat during the summer (which evolves as a function of the variation in orbit of the earth around the sun) and, on the other hand, the proportion of fine dust in the marine sediments of the North-African coast, which are evidence of the covering, vegetal or desert, on the soil of the continent. The author makes the point that an extrapolation of the tendency observed between 9000 BP and 6000 BP does not constitute a basis for imagining an irreversible change as sudden as the one which has taken place. During that period the amount of solar heat in the summer changed very gradually, and in parallel one observes a slight tendency towards an increase in the proportion of fine dust, testifying to a slight decline in vegetation. During the period that follows, the solar heat continues to evolve according to the same progressive rhythm, but, surprisingly, the proportion of dust shows a sudden surge, indicating the abrupt shift towards a desert state. After this abrupt transition, the two indicators continue to evolve progressively. Researchers have been able to show that feedback between the evolution of vegetal cover and the local climate is at the origin of this change at once sudden and important, feedback which began when the level of solar heat crossed a critical threshold.

25Numerous environmental and social phenomena have given rise to studies supported by the theoretical framework of transitions and regime shifts. A community of researchers40 working on resilience have taken the initiative of uniting their work on the site ‘resalliance’.41 One of the objectives has been to present a group of case studies in a harmonised manner: 1) identification of two alternative regimes; 2) specification of the principal variables at issue, those with a rapid dynamic, which characterise ‘ordinary’ change, and those with a slow dynamic, which characterise the environment and whose variation can function as a trigger of a more ‘radical’ change; 3) presentation of an interpretation with regard to the mechanisms at work. Most of the cases concern environmental problems, but certain studies concern social systems.

Effects of scale in the dynamics of a system

26Regime shifts often affect systems which are spatially heterogenous and for which the processes involved belong to several scales.42 This multi-scale dimension contributes to making identification of a transition more difficult in the empirical sphere but also to enriching its interpretation. Since the causes are generally multiple, numerous variables are in play simultaneously, and each of them can be associated with critical threshold effects. Thus, Ann Kinzig and her colleagues43 stress the fact that, most often, social and ecological variables interact. In such cases, the critical thresholds also interfere and ‘cascade’ effects can occur, whereby the crossing of a critical threshold in one field entails the crossing of critical thresholds in others. Moreover, it is possible that the several processes involved may operate on different scales. Thus, numerous authors point out that one cannot understand the dynamic of a socio-environmental system without taking into account the interactions of the object of interest with the elements operating on scales that are superior and inferior to it.44

27Scheffer and his colleagues45 have been especially concerned with the effects of scale in the response of a system to a perturbation. They therefore bring out non-intuitive relations between a ‘local resilience’ and a ‘systemic resilience’. Considering a complex system made up of numerous components, they discuss the respective roles of the heterogeneity of the components and of their degree of connectivity with regard to the capacity of the system to maintain itself when a perturbation occurs. When a system is composed of a set of homogeneous components, strong connectivity is a factor favouring ‘local resilience’, thanks to the mechanisms of subsidiarity. Conversely, in proximity to a critical threshold, a local perturbation, even a small one, may engender a cascade effect and provoke a systemic transition.46 This will have all the more likelihood of taking place if the connectivity is strong, as, for example, in the banking system crisis of 2008.47 The very characteristics that favour the resilience of the system when faced with local perturbations can thus entail a ‘large-scale collapse’48 of the system as a whole. On the other hand, a system whose components are more heterogenous but less interconnected will evolve in a more gradual manner.

28A model is a good tool for exploring the effects of the imbrication of mechanisms operating at several levels. The case of the collapse of the Mayan civilization offers an example of this. Thus Scott Heckbert49 has developed the MayaSim model, using a multi-agent system in which he has integrated precise data on the evolution of the environment (climate, soil) and formulated hypotheses regarding the behaviour of agents representing units of settlement (demography, chosen use of soil, commercial exchanges). The model is used as a ‘laboratory’ to explore how the socio-ecological system corresponding to the dense and interconnected settlement of the Mayas responded to climate changes. The author stresses that most of the simulations point to a moderately intense development of the settlement system, of lesser amplitude than the archaeological remains would suggest. In these cases, the simulations show a system that maintains itself over an extended period. Only a few simulations point to a strong development of the system, with a settlement becoming very important in size and units of settlement well inter-connected. These simulations are associated to degraded soil. At this stage, the system shows itself very sensitive, and a small climatic fluctuation has large effects: the decline of an important node in the network results in a cascade of declines. One of the interests of this model is to show that the famous peak of Mayan civilization, identified from the observation of archaeological vestiges, was perhaps unlikely to be reached, but that once this level of development was achieved, the system was of such a sensitive nature that the probability it would collapse at the least perturbation, even of feeble intensity, was high.50

Identification and anticipation of a transition in the empirical domain

29This discussion of the multiplicity of elements interacting within a system gives a glimpse of the difficulty, when faced with a change observed in a given empirical domain, of determining whether or not it corresponds to a process of transition. In applied research, it can be a challenge to anticipate the arrival of a transition. Foreseeing constitutes, indeed, an essential issue in an operational context. Now, unpredictability is a characteristic associated with regime shift. To be capable of identifying the critical thresholds which may potentially cause the system to alter would be a considerable advantage.51 52Such anticipation would make it possible to take appropriate actions if the regime shift is considered deleterious (disappearance of animal species, for example). An intensive fishing effort may thus suffice to make a lake evolve from a state of turbid water to one of clear water. Certain empirical studies have shown that, before a regime shift, the variability of particular indicators is enhanced.53 Once the statistical relation is established, one can then use such statistical indicators (increase in the variance of a variable or of the temporal correlation of fluctuations, for example54) as predictors of a coming regime change and intervene to prevent it, even if the processes at the origin of it are not understood.55

30The settlement systems that are the object of this volume belong to the human and social sciences, and rather than anticipate a transition, the issue is to identify it on the basis of observations and empirical knowledge. The point, then, is to perfect a strategy for interpreting a change in terms of transition. If one possesses appropriate indicators, which are well adapted to the phenomenon studied and are available over a sufficiently long duration, the statistical approach permits one, for instance, to identify a transition from observing intervals with significantly different values in a temporal series.56 A bimodal statistical distribution of an indicator may be possible to interpret, moreover, as the trace of the existence of two alternative states around which the system fluctuates. The identification of a common type of change for multiple indicators or for the same indicator in two different places may contribute to establishing its generic character. The identification of pertinent indicators capable of providing a signature of a transition thus constitutes an essential step. The studies of Jean-Pierre Bocquet-Appel57 on the demographic transition of the Neolithic are emblematic in this respect. This transition relates to the change in the rhythm of population growth which took place in the Neolithic. The question is: what are cause and effect in the processes of mastering agriculture, in sedentarisation and in the increase in population? In order to estimate the birth-rate for these distant periods, the author applies an original indicator – the number of skeletons of children from 5 to 15 years old in relation to the total number of skeletons found in the cemeteries. Using this indicator, the author evaluates the temporal lag between the beginning of agriculture and the increase in birth-rate. The results of the analyses carried out in Europe, Africa and America, point to a regular lag of 500 to 700 years between these events, the first of which can then be interpreted as an element triggering the transition. Applying this method to the Amerindian sites of the southwest United States, Timothy Kohler and his colleagues58 date the demographic growth around 500 A.D., while the beginnings of the cultivation of maize in this region is much earlier (>2000 BC). The authors deduce from this that the first cultivations of maize probably did not influence the settlement modes of the hunter-gatherers. On the other hand, the appearance of ceramic receptacles testifying to an intensive use of maize corresponds well to this temporal lag of from 500 to 700 years. Even if it is important to bear in mind the degree of uncertainty associated with these observable data, the construction of such indicators contributes to opening up fertile lines of thought.

31In numerous cases, however, we do not possess statistical indicators, and other methods more qualitative, based on investigations, may then be used to identify the nature of the change underway. Shauna BurnSilver and her colleagues59 employ the theoretical framework of regimes and transitions to choose between two points of view regarding the functioning of communities of ‘natives’ in Alaska. The question relates to the present functioning, which corresponds to a mixed economy, associating a market economy with traditional means of subsistence (‘subsistence-based livelihood’) derived from hunting and fishing and based on sharing between members of the community. Does this mode of functioning correspond to a transitory phase – an interpretation implying that the system is being directed towards a classic market functioning – or does this mixed economy represent a regime in its own right, in which case it is durable? Basing their argument on observations (derived from investigations) showing the persistence of exchanges of products derived from hunting and fishing among members of different Indian communities, and notably on the significant and persistent engagement of the persons having the highest incomes in these exchanges, the authors conclude in favour of the durability of a mixed economy.

Conclusion: the particularity of human systems

32One of the factors that differentiate the human species from other animal society relates to the importance of culture – defined as the social transmission of practices and norms – among humans. Culture registers much of the fluidity and plasticity that characterise humans and their organisations, but as it happens, these continual changes create an enigma. Culture can indeed change in the absence of any crisis, according to a dynamic activated by drift-effects in the course of a process of transmission (transformations of content or of context of utilisation) or by influential innovations. Because of these means of transmission linked to content or context, it is difficult rigorously to distinguish ‘ordinary change’ from the kind of ‘major change’ that would correspond to a regime shift.

33We have made the decision to concentrate on types of change that are radical in terms of the number of relations affected and their degree of transformation, rather than on changes operating according to a continuum of more or less rapid modifications. Comparative studies suggest that, on the whole, the dynamic of cultural transmission concerning the population testifies to a considerable constancy over time with regard to the repertoire of practices, even when generations are exposed to very different environments.60 61Thus analysis of practices in 172 small indigenous societies in the west of North America at the moment of contact with Europeans has shown that the ensemble of norms and practices relating to the exercise of power, to interpersonal bonds, to ceremonies, to forms of habitat or to methods of warfare were very largely preserved in the cultural phylogenies62 (as measured by linguistic filiations). Consequently, we can consider that the changes that intervene in the cultural domain signal profound transformations in the structure of the societies. By contrast, those changes that intervene in the field of techniques, subsistence or still other practices of an economic order, which are dependent more on ecology than on phylogeny, seem less to affect the most durable aspects of the social structures.

34We have, therefore, two abstract dimensions of variability which may be verified in our case studies: the speed of the change (measured, for example, by the number of generations) and the depth of the change, considering the least frequent changes in the phylogenies as the most fundamental. It is because the changes in the forms and practices related to settlement (including such variables as the degree of permanence and the density of constructions, the size of the entities in terms of population, and the presence of communal equipment in the habitats) are considered as profound transformations of social structures that we have chosen to study the settlement system in the case studies presented in part 2 of the volume.

35Most of the examples considered in this work show the interest of developing original orientations, particular to each case, for interpreting change in terms of critical thresholds and transitions. It has been amply demonstrated that unforeseeable chance events (such as the upsetting of long-term climatic tendencies, volcanic eruptions or tsunamis, or even invasions) could cause crises of variable amplitude and lead to social and cultural changes. Within the framework of the TransMonDyn project, the accent has been placed on the reasons for change and the mechanisms of transformation, rather than on the destructive effects of these extreme events.