• 1 D. Grossman, D. Siddle, 1998; F.R. Kaloko, 1983; M. Silberfein, 1998a; R.K. Udo, 1966.
• 2 United Nations, 2019.
• 3 AGRA, 2019.
• 4 C.S. van der Waal, 1991.
• 5 D.F. Bryceson, 1996, p. 107.
• 6 J.O.C. Onyemelukwe, 1980; P.S. Robinson, 2003; C. Southey, 1982; A. Todes, 2001.
• 7 M. Mashiri et al., 2008; A. Zvoleff et al., 2009.

1Rural settlement change across sub-Saharan Africa has long been studied with a particular emphasis on the historical impact of macro-phenomena such as population growth, migration, economic change/growth, and infrastructural development.1 Models built around such factors have generally been broad-brush in their approach and have tended to view change as primarily attributable to exogenous ‘prime movers’, in which rural populations are depicted as essentially reactive entities otherwise given to long periods of tradition-bound cultural continuity. As such, the approximately 57% of sub-Saharan Africans who live in rural areas2—of whom 70–80% are small-holder farmers3—often remain characterised as somehow undifferentiated and unchanging, whilst rural areas themselves are continually perceived as economically stifled poverty sinks.4 Such perspectives have long influenced development-oriented discussions and policy proposals centred on the perceived deficiencies of dispersed and/or mobile settlement,5 many of which have blamed traditional land tenure and social organisation for stymieing rural African communities’ fully productive integration within the modern transnational economy.6 Such tropes have painted rural communities as inherently risk averse and their continued geographical dispersion as comprising a major stumbling block to the successful roll-out of many development projects, from healthcare to electrification.7

• 8 E.C. Green, R.B. Isely, 1988; O. Oki et al., 2015; G. Porter, 2002.
• 9 Cf. G.A. Myers, 1999.
• 10 I.A. Adalemo, 1967; A. Arecchi, 1984; R.B. Dakayne, 1962; N. Georgulas, 1967; D. Grossman, 1971; M. (...)

2Alternative models have conversely argued that development planners should seek to learn from rural African communities’ own adaptive techniques for living, and to engage with extant settlement patterns rather than remould them wholesale, as seen in some infrastructural proposals and successful community-led service provision projects.8 However, whilst such schemes have advocated for more nuanced understandings of localised settlement processes, they have rarely engaged in concrete analyses towards those same ends. Consequently, conceptions of rural settlement within Africanist development literature have often continued to operate along the simplistic lines set-up by earlier geographical and historical studies, whereby the dynamism, internal complexity and contextual variation of sedentary rural communities’ modes of habitation may be recognised but are rarely explicated in any detail.9 Instead, agrarian African communities have commonly been assumed to adopt a dispersed form of settlement according to ecological impetus and social tradition, with nucleation occurring during cases of civil unrest, defence against disease, colonial interference, and/or post-independence governmental resettlement schemes, and as the inevitable outcome of agricultural mechanisation and economic marketisation in the modern era.10

• 11 C.Z. Ashley et al., 2016; A. Ogundiran, 2016.
• 12 O.J.C. Boles et al., 2019.
• 13 P. Lane, 2009.
• 14 P.J. Lane, 2016.
• 15 O.J.C. Boles, P.J. Lane, 2016.

3One interpretive stumbling block which has proved particularly durable in maintaining this conceptual status quo is the tendency of past researchers to view the sedentary settlement of predominantly agrarian African communities as categorically distinct from the habitually mobile forms of habitation employed by so-called hunter-gatherers and nomad pastoralists.11 Within eastern Africa, pastoralist lifeways in particular have provided rich investigative material for scholars from various disciplines, who have together emphasised the ways in which mobile pastoralism has long been entangled within wider ecosystems. For instance, in contrast to relatively recent narratives of erosion and savannah grassland degradation which place the blame on indigenous stock keeping, more historically attentive approaches have revealed a far more complex picture in which recent environmental pressures are in large part legacies of colonial land appropriation12 or indeed issues of great duration that pastoral communities have long learnt to either live alongside or actively manage.13 In many cases, cattle keeping has promoted ecological diversity—for instance, where soils enriched by cattle dung deposited within temporary encampments have facilitated the formation of grassy glades within expanses of scrubland.14 Research into vegetal change and soil chemistry in eastern Africa has tracked the ecological impact of such sites back to at least the 15th century AD,15 highlighting the critical long-term role that such nomadic settlement has played in the formation of contemporary landscapes.

• 16 E.g. N. Arazi, 1999; J. Fleisher, 2001, 2014; A.F.C. Holl, 1996; N. Khalaf et al., 2019; K.C. MacDo (...)
• 17 M.E. Clark, 2003; H. Haskell et al., 1988; R.J. McIntosh, 2005; R.J. McIntosh, S.K. McIntosh, 2003.
• 18 J. Fleisher, 2014; J. Fleisher, F. Sulas, 2015; F. Sulas et al., 2017; S. Wynne-Jones, 2007.
• 19 A. Antonites, 2014; A. Antonites, C.Z. Ashley, 2016; A. LaViolette, J. Fleisher, 2005, p. 336-339.

4In cases such as these, mobile patterns of quotidian habitation are recognised as being integral not only to the social life of the communities in question, but also to the wider landscape settings which they inhabit. Conversely, the archaeological and historical study of sedentary settlements has tended to regard problems of movement through a much wider lens: that of en masse population migration and/or the broad-scale distribution of settlement sites through time. Such themes are particularly evident in the many regional archaeological surveys that have been carried out across the continent,16 as well as in more in-depth projects investigating a variety of contextually specific themes. The latter include the study of variable resources and urban–rural networks throughout the Inland Niger Delta,17 the processes of social negotiation encapsulated within the medieval entrepôts of the Swahili coast,18 and the complex economies of cattle transhumance, arable agriculture, and long-distance trade connecting the ‘imagined communities’ of the so-called Zimbabwean Tradition.19 However, whilst spatio-temporal variance and the study of mobile practices may be central to all these projects, their resolution remains comparatively coarse, and the complex historical specificities of individual sedentary settlements and the cross-generational lives of their inhabitants remain elusive.

• 20 I. Kopytoff, 1987.
• 21 See also J.M. Hunter, 1971.
• 22 A.E. Nyerges, 1992.
• 23 A.E. Nyerges, 1992, p. 872-873.

5In contrast to this archaeological lacuna, there are several anthropological and historical studies that have formulated detailed approaches to the structural mobility which underlies the otherwise sedentary character of many rural African settlement systems. Perhaps the best-known of these is Kopytoff’s model of the “internal African frontier”,20 which focuses on areas of weak political control where constant small-scale migration and kin-group fissioning create permanent “interstitial” frontiers between communities.21 Such areas are often characterised by low population densities and predominantly inhabited by segmentary societies where the principle of ‘wealth-in-people’ underpins the accrual of social power. The model can be extended to encompass ecological factors, particularly the environmental and labour demands of swidden agriculture, whereby social/ecological inequalities are perpetuated within a shifting settlement system in which labour is often co-opted by gerontocratic elites.22 In such cases, labour shortages engendered by habitual impermanence and a general lack of social cohesion can lead to the adoption of more intensive agricultural strategies to boost short-term production, including the cultivation of short fallows and cash cropping. However, intensification carries its own risks—for instance, the suppression of forest succession and the increased degradation of soils, necessitating eventual abandonment and fresh land clearance.23 The internal frontier thus comprises a landscape in which mobility becomes both a driver of, and response to, patterns of risk management across intertwined social and ecological relationships.

• 24 P.W. van Arsdale, 1992.
• 25 D. Turton, 1988, 1996.
• 26 F.P. Conant, 1965; M.I.J. Davies, 2009, 2013a, 2014; M.I.J. Davies, H.L. Moore, 2016.

6However, despite the concept’s rapid acceptance across a variety of disciplines,24 it should be remembered that the “internal frontier” was conceived within a specific political-cum-ecological context and should not be applied uncritically across the entire continent. Nevertheless, Kopytoff’s model sets a well-developed precedent for the recognition of habitual forms of periodically mobile sedentism as crucial components of specific socio-cultural complexes, rather than merely tangential outcomes of broader-scale economic or political processes. Similar iterations of gradually shifting settlement systems have been studied across sub-Saharan Africa by both archaeologists and anthropologists. For instance, Turton’s analysis of the punctuated migration of the Mursi people within Ethiopia’s Omo Valley shows how gradual, cross-generational movements emerge from long-term histories of flexible responses to varying social/ecological conditions, whilst incorporating far-reaching changes to both the community’s territorial occupation and ethnic identity.25 Similar sequential movements have been noted for other so-called segmentary groups, such as the agriculturalist Pokot communities of the Wei Wei Valley to the immediate north of Marakwet, where both the settled and cultivated lands of multiple lineage groups effectively ‘leap-frog’ each other in distinct bands that sequentially progress up the valley as successive strips of land are cleared, farmed and fallowed/abandoned.26

• 27 S. Canós Donnay, 2016.
• 28 S. Canós Donnay, 2016, p. 467.
• 29 See also F.P. Conant, 1962.
• 30 S. Canós Donnay, 2016, p. 465-466.

7Canós Donnay has similarly traced the gradual shifting of settlement across the Upper Casamance region of southern Senegal,27 though in this case the settlements in question comprised nucleated villages/towns whose periodic displacement appears to have been more wholesale in nature. These settlements were moved every few centuries for a variety of reasons, from soil exhaustion to intra-settlement politicking as well as larger-scale conflict and insecurity, but preserved their names, identities and community institutions throughout their various relocations. Constrained sedentary mobility hence seems to have long been “an intrinsic and constant dimension of the articulation of social dynamics” in the region.28 Canós Donnay here adopts the term ‘shifting sedentism’ to encapsulate the means by which mobile practices can be incorporated within modes of communal habitation which are otherwise sedentary in nature and most often tied to arable agriculture as their principal means of subsistence. However, she also notes that these practices (and their resultant spatio-temporal patterning) can vary greatly amongst communities, even within a relatively constrained geographic region.29 Understanding these variances will entail far more detailed analyses of how both individual and group lifeways articulate with/inform wider mobile practices—relationships which currently remain poorly understood within Upper Casamance, as indeed across much of sub-Saharan Africa.30

• 31 W. Balée, 1998; C.L. Crumley, 1994.
• 32 C.L. Crumley, 2018.
• 33 W. Balée, 2006, p. 77-79; C.L. Crumley, 2018.
• 34 C.L. Crumley, 2019, p. 10-13.
• 35 C.L. Crumley, 2019, p. 10; W. Balée, 2006, p. 78.

8One particular body of analytic and interpretative thought that is of potentially use in addressing this lacuna is that gathered under the conceptual umbrella of ‘historical ecology’. Broadly speaking, historical ecology comprises an interdisciplinary research programme centred on the assertion that to understand any given landscape it is necessary to regard both human and non-human elements as equal in their capacity to affect one another.31 It thus views humans as a component of all ecosystems, and history as encompassing both the social and physical past of our own species alongside that of the earth system in general.32 In this vein, historical-ecologists advocate holistic analyses of how human and non-human relations produce landscapes within particular locales, alongside their cross-context comparison. This multi-scalar approach stems in large part from the varied intellectual traditions from which historical ecology draws, including the cultural geography of Carl Sauer, the Annales School of history epitomised by Ferdinand Braudel, the early ecological and ethical work of Aldo Leopold, and Boasian traditions of early US-based anthropology.33 In its present manifestations, complex and adaptive systems theories have also proved heavily influential on historical ecology,34 whilst the Annalistes’ formulation of landscape (paysage) as a multi-temporal entity continues to foster the humanistic tendencies of many associated researchers.35

• 36 P. Sinclair et al., 2018, p. 21-25.
• 37 C.L. Crumley, 2019, p. 10-11; M.I.J. Davies, 2013b, p. 23; P. Sinclair et al., 2018, p. 25-26.
• 38 W. Balée, 2006, p. 83.
• 39 C.L. Crumley, 1995, 2005.
• 40 W.S. McCulloch, 1945.
• 41 C.L. Crumley, 1995, p. 3.

9The aforementioned mode of systemic thinking specifically focusses on the dynamic functions of non-linear systems, in which non-equilibrium is taken to be the norm.36 It emphasises the emergent properties of complexity, in which a given system’s constituent components operate in a gestalt manner, where not all relations progress smoothly or even necessarily in sync and thus trend towards novelty over time. This is not to say that external stimuli cannot precipitate large-scale reorganisation but that they occur within an already mobile and intrinsically integrative context.37 Disturbance is thus regarded as a constant characteristic of all systems, which in an ecosystem context is neither necessarily ‘good’ nor ‘bad’ but depends on analytic perspective and a long-term view of system ‘health’.38 Such systems are also often perceived as possessing heterarchical organisation.39 Originally a neuroscientific concept,40 heterarchy can be defined as “the relation of elements to one another when they are unranked or when they possess the potential for being ranked in a number of different ways”.41 As a conceptual tool, heterarchy can be applied to the interpretation of both ecological systems and patterns of settlement and political organisation—emphasising the contingent nature of agentive capacity across heterogenous contexts.

• 42 Sensu A. Giddens, 1984.
• 43 W.E. Doolittle, 1984, 2019.
• 44 L.S. Cordell, F. Plog, 1979; M.I.J. Davies, 2014.
• 45 T. Ingold, 1993, 2000.
• 46 W.E. Doolittle, 2019.

10Heterarchical organisation can also help account for emergent change in contexts where agency is limited, contested, or even unintentional, yet serves to both perpetuate and modify broader social-cum-environmental systems. Much historical–ecological investigation is thus devoted to the analysis of how the ‘unintended consequences’ of repetitive, practical behaviours42 engender moments of small-scale innovation which in turn lead to long-term changes within ecological systems that feature humans as a keystone species. For instance, Doolittle’s model of incremental agricultural change43 explores how complex agricultural systems can arise from the cumulative efforts of individual farmers who, though operating within shared environmental and cultural constraints, are not themselves focussed on the creation or maintenance of the overarching system.44 This perspective is somewhat akin to Ingold’s conception of “taskscapes”, in which landscapes come into being through the interaction of diverse practices of dwelling, individually quotidian but accruing cumulative affect over the longue durée.45 That said, Doolittle’s approach is perhaps more explicitly processual in its focus on expedient human behaviours aimed principally at survival, particularly when dealing with hazards of low magnitude but high frequency. Such expediency often engenders principles/practices of relative impermanence, in which temporary interventions come and go through time, sometimes accruing into landscape features of unintentional longevity, and sometimes not.46

• 47 M.I.J. Davies, 2013a.
• 48 C.L. Redman, 2005.
• 49 E.g. C.L. Crumley et al., 2018; M.K. Nelson, D. Shilling, 2018.
• 50 M.I.J. Davies, 2012; M.I.J. Davies et al., 2016; P.J. Lane, 2010; A.L. Logan et al., 2019; D. Stump(...)

11In either case, it is the suitability of particular interventions for sustaining life in the present instance which is of prime import for their practitioners, and it is the variability and flexibility of their modified repetition which sustain community lifeways over multi-generational timescales. Such assertions have been central to many recent discussions of ‘resilience’ within archaeology. Contrary to exhortations of ‘sustainability’, which tend to promote static continuities of practice in response to external environmental conditions, resilience theory is akin to historical ecology in its emphasis on continuity through change, alongside the socially embedded nature of human–environmental relations in which decision making is often rooted at a household level.47 This perspective does not preclude the long-term success of intensive agricultural practices, centrally managed or otherwise. However, it does stress that centralisation is not necessary for such systems to function and that when it is present the overarching system typically incorporates smaller, ad hoc modifications that ultimately extend its longevity—which, when no longer effective, can lead to wholesale reorganisation in favour of alternative lifeways.48 It is in this vein that many scholars have called for agricultural development to draw on ‘traditional ecological knowledge’ over industrialised, capital-input interventions,49 and for archaeologists to formulate long-term narratives of diachronic change that comprise ‘useable pasts’ for the contemporary world, Africa included.50

• 51 AAREA, 2018.
• 52 E. Boserup 1981.
• 53 A.L. Logan et al., 2019, p. 423.
• 54 C. Ferro Vázquez et al., 2017; T.K. Kabora et al., 2020; C. Lang, D. Stump, 2017; D. Stump, 2006; S (...)
• 55 African Farming Research Network, n.d.; M.I.J. Davies et al., 2016.
• 56 M.I.J. Davies, H.L. Moore, 2016; M.I.J. Davies et al., 2014; see also M.I.J. Davies, 2009, 2013a, 2 (...)

12The Archaeology of Agricultural Resilience in Eastern Africa (AAREA) Project has been a particularly notable example of the latter trend, pursuing a comparative investigation of the terraced landscapes of Engaruka (Tanzania) and Konso (Ethiopia).51 The project refutes Boserup’s hypothesis that communities invest labour in intensive agricultural practices only to increase yields,52 instead showing household decision making was/is largely based on creative approaches to risk mitigation.53 This proved true both of the still-cultivated terraces of Konso, where intercropping and locally distinct terracing systems were used to create seasonally variable agricultural niches alongside the emergence of new erosional regimes, and of the archaeological landscape of Engaruka, where crop diversification and the gradual, un-centralised accretion of large-scale infrastructure continued for centuries before its final abandonment.54 Similar research amongst the Marakwet communities of the Kerio Valley at the foot of the Elgeyo Escarpment—alongside comparative work in Tivland (Nigeria) and Bokoni (South Africa)55—has similarly demonstrated how household and community decision making regarding the maintenance, modification or abandonment of intensive agricultural techniques and infrastructure operates within a complex system of trade-offs, feedback and other effects. Flexible responses to both adversity and opportunity lie at the core of these systems, even to the point of more occasional instances of large-scale change/reorganisation.56

• 57 M.I.J. Davies, H.L. Moore, 2016, p. 82.
• 58 M.I.J. Davies, H.L. Moore, 2016; M.I.J. Davies et al., 2014; H.L. Moore, 1986.

13This multiplex of entangled social, cultural, and ecological processes has resulted in a heterogenous and inherently dynamic landscape, which over the longue durée has been successfully inhabited by local Marakwet communities through the deployment of a diverse complement of ‘technologies of life’, founded in principles of movement, innovation, and improvisation.57 Many of these ‘technologies’ are themselves rooted in household-based practice, with multiple households both within and across communities contributing to mutual labour projects, modes of social interaction, and cultural events.58 Landscape-spanning historical–ecological dynamics can thus be seen to be rooted within households, in which change must be regarded as a constant of their constituents’ flexible lifeways, even as the overarching characteristics of the social–environmental landscape have (at least over the past few centuries) retained a recognisable consistency. Likewise, it is this kind of multi-scalar approach—rooted in household contexts but extending across the surrounding landscape—which could also help to elucidate the dynamic settlement histories of these communities and to further our investigation of the mechanisms and variances encapsulated within the broader framework of ‘shifting sedentism’. It is with this agenda in mind that I turn to the specific focus of this paper: the Marakwet settlements of the northern Elgeyo Escarpment in north-western Kenya.

• 59 Kenya Forest Service, 2015; H.L. Moore, 1986, p. 21.
• 60 R.M. Shackleton, 1951, p. 346.
• 61 P. Mason, A.B. Gibson, 1957, p. 7.
• 62 C. French, 2015; P. Mason, A.B. Gibson, 1957.

14Marakwet (the name given to both the northern half of Elgeyo Marakwet County and the principal ethnic group who reside within it) presents a highly varied and distinctive landscape. Defining its eastern flank, the Elgeyo Escarpment essentially bridges two distinct environmental zones—rising from the floor of the Kerio Valley at approximately 1000 masl up toward the undulating range of the Cherangani Hills which peak at 3365 masl.59 This great change in elevation across only a few kilometres of longitudinal distance is the result of post-Miocene faulting, with the Escarpment forming a giant fault-scarp along the western wall of the Gregory Rift Valley.60 Although the southern part of the Escarpment is largely defined by sheer cliffs, its northern extension through Marakwet consists of a deeply dissected slope comprising sharp ridges separated by acute, V-shaped valleys.61 The geological base of most of this northern scarp is composed of metamorphosed sedimentary rocks of the Basement System, predominantly banded hornblende-biotite gneisses, frequently cut by anastomosing pegmatitic veins of quartzo-felspathic material. In contrast, the floor of the Kerio Valley consists mostly of dissected lavas and pyroclastic rocks. Several small hills are formed by rocky inliers of the Basement System, whilst much of the surrounding plain is overlain by thick alluvial and/or colluvial deposits.62

• 63 T. Dietz et al., 1987, p. 16-20.
• 64 Kenya Forest Service, 2015, p. 4.
• 65 Joseph Kimutai Cheptorus, pers. comm. November 2020.
• 66 T. Dietz et al., 1987, p. 16, 19.
• 67 Kenya Forest Service, 2015; S. Lunn-Rockliffe, 2018, p. 73.
• 68 P. Mason, A.B. Gibson, 1957, p. 8.

15Rainfall is likewise highly variable, with the Kerio Valley and Elgeyo Escarpment slope typically receiving around 900 mm a year, though this fluctuates with time and location,63 whilst the higher Cherangani Hills receive approximately 1200 mm in the east and 1500 mm in the west.64 The rains across the region occur in two main tranches—the ‘short rains’ beginning around April and the ‘long rains’ in July/August, though again these timings are far from uniform.65 Temperatures can also be high, though in the wet seasons it is much cooler in the highlands as cloud cover tends to be low-hanging and ubiquitous. The Valley is typically hot year-round, with high annual evaporation rates resulting in an arid to semi-arid climate.66 The highlands are altogether much wetter, supporting Afroalpine forest and forming one of Kenya’s main ‘water towers’.67 Small streams and springs are prevalent throughout these upper hills, whilst they are largely absent from the Escarpment slope and Valley floor. These latter regions are instead traversed by larger rivers, which have their upper catchments within the highland forest. Though mostly perennial, only the largest of these rivers, including the Embobut (see below), actually converge with the main Kerio River which runs northward along the centre of the Valley, their waters instead being diverted for irrigation purposes en route.68

• 69 H.L. Moore, 1986, p. 18.

16Such topographic and climatic variation supports great ecological diversity within a relatively small geographical region, though highly localised and unpredictable distributions in resource availability also mean this can be a very difficult place to inhabit. Local populations have thus adopted a diversity of livelihood practices and settlement strategies in their adaptive engagement with this landscape. For instance, agriculture in the Kerio Valley relies heavily on furrow-fed irrigation and shifting field systems, whilst highland fields tend to be rain-fed and more permanent. Sparser water resources in the Valley have also led to the development of larger, semi-clustered settlements encompassing several clans’ territories, whilst the upper Escarpment is characterised by the north–south imbrication of multiple clan territories encompassing a more-or-less continuous spread of dispersed settlement. Connections amongst these communities are both many and various, but the Marakwet maintain a linguistic separation between the three zones of Valley, Escarpment, and highlands.69 Keu hence denotes the flat plain of the Kerio Valley, whilst lagam refers to the slope of the Elgeyo Escarpment, both lower and upper parts. Mosop refers to the highlands beyond, though this particular term is more situationally variable than the other two, such that whilst people living in keu often speak of the Escarpment crest as mosop, those within the latter locale reserve the term for the yet-higher Cherangani Hills to the west (this paper adopts the latter usage).

• 70 H.L. Moore, 1986, p. 12-13.
• 71 B.E. Kipkorir, 1973, p. 73; G. Lynch, 2011, p. 35.
• 72 G. Lynch, 2011.
• 73 D. Branch, 2011.

17The Marakwet themselves comprise but one ethnic group within the wider ethno-linguistic conglomeration known as the Kalenjin—a set of related Southern Nilotic language group speakers broadly distributed throughout Kenya’s Great Rift Valley and Western Highlands.70 The ethnonym Kalenjin is itself a fairly recent phenomenon, now encompassing those communities more commonly referred to as ‘the Nandi-speaking peoples’ prior to the 1940s, Nandi being one of the largest such groups and the ‘type language’ to which close relatives such as Marakwet are typically compared. ‘Kalenjin’ means “I say to you” in Nandi—a distinctive opener for local radio broadcasts that was subsequently appropriated by politicians and other figures seeking to build a supra-tribal identity that cut across colonial ethnic delineations in the decades prior to Kenyan independence in 1963.71 This appropriation was highly successful, and by 1963 the ‘Kalenjin’ were both a self-identifying and nationally recognised ethno-political force.72 As a tribal grouping writ large they have played an integral role in the political life and governance of Kenya ever since.73 That said, the component ethnic groupings that together form the Kalenjin still retain their own dialects, cultural practices, and more specific political affiliations nested within their larger Kalenjin identity.

• 74 H.L. Moore, 1986, p. 13.
• 75 C. Ehret, 1971; J.E.G. Sutton, 1973.
• 76 J.E.G. Sutton, 1973, 1987, 1993.
• 77 M. Bollig, 1990; M.I.J. Davies, 2008, 2012; D.K. Kay, 2021; S. Lunn-Rockliffe, 2018; S. Lunn-Rockli (...)

18The establishment of the various Kalenjin ethnic groups’ traditional territories extends back over several centuries—the product of a complex weave of migration, settlement, and ethnogenesis. The current consensus is that Southern Nilotic populations migrated into Kenya’s Western Highlands no less than a thousand years ago,74 where over time they developed the cultural-linguistic ‘package’ typical of proto-Kalenjin populations, including patrilineal clans, patrilocal residence, male and female circumcision, initiation into age-sets, and predominantly pastoral economies.75 This package was very much one of family resemblance, incorporating a great deal of variation across, and often within, its constituent communities. A similar situation is observed throughout the last millennium in the archaeological remains of the so-called Sirikwa tradition. Beginning in the early 12th century AD, the Sirikwa phenomenon spread across much of the Western Highlands and Central Rift Valley alongside a growth in iron-working technologies and mixed pastoral-arable economies.76 The cohesion of the tradition then seems to have expired a little over 200 years ago, most likely due to the short-lived but highly impactful expansion of Maa-speaking populations into the Rift Valley and Uasin Gishu plains in the 17th–18th centuries. These incursions are argued to have fragmented existing Sirikwa/proto-Kalenjin communities and initiated a complex series of small-scale migrations across the region, ultimately leading to the formation and present distribution of contemporary Kalenjin populations approximately 200–250 years ago.77

• 78 B.E. Kipkorir, 1973; J.E.G. Sutton, 1973.
• 79 H.L. Moore, 1986, p. 10.
• 80 S. Lunn-Rockliffe, 2018, p. 56.
• 81 H.L. Moore, 1986, p. 10-11.
• 82 H.L. Moore, 1986.
• 83 B.E. Kipkorir, 1973; D.A Welbourn 1984.
• 84 W. Östberg, 2004, 2014.
• 85 R.C. Soper, 1983.

19The specific ethnonym ‘Marakwet’ arose from a colonial era corruption of ‘Markweta’, then employed by just one of the five sections comprising its present-day populace.78 The other four sections were the Almo, Endo, Kiptani, and Cherangani,79 though again it is perhaps wise to be somewhat sceptical of both the specificity and fixity of these terms given their reportage through the filter of the early colonial administration. The Cherangani provide a particularly difficult case in this regard, as not only is it also a geographical term applied to the Cherangani Hills, but there seems to be some ethnic overlap with the Sengwer, who though similar to the Marakwet in many ways speak a distinct dialect and have historically pursued a less agriculturally focussed way of life within the highland forests.80 Nevertheless, references to the Marakwet as a wider group become more common from the 1950s onwards.81 As with Kalenjin, ‘Marakwet’ has since become both a governmentally defined tribal grouping and a highly prevalent means of self-identification amongst those to whom it has been ascribed. Moore’s anthropological study of the Kerio Valley settlement of Sibou and adjoining market centre of Tot (aka Tot-Sibou) in the early 1980s82 has since been fundamental in establishing the Marakwet’s renown within academic circles, as have the broadly contemporaneous anthropological endeavours of Kipkorir and Welbourn,83 the slightly later research of Östberg,84 and even Soper’s early attempts to date their complex irrigation system.85 Such research demonstrates both the widespread valency of the Marakwet ethnonym and the continued centrality of locally grounded identities within community lifeways. Moore’s investigations in particular also established a strong precedent for the study of Marakwet settlement, and to which this paper is itself indebted.

• 86 S. Lunn-Rockliffe, 2019.

20Within the broader setting of eastern Marakwet, this paper takes as its specific case study the settlement of Kacheseker, perched upon the crest of the Elgeyo Escarpment at approximately 2400 masl within the area known locally as Embobut, and overlooking the Kerio Valley to the east. Kacheseker bridges the juncture between the dry upper slopes of lagam and the more verdant rolling plateau to the west, much of which lies within the official gazetted boundaries of the Embobut Forest, in which habitation is now legally forbidden.86 The modern-day settlement of Kacheseker, as with many of its neighbours, currently extends eastward from the tarmacked road bounding the Forest and over the Escarpment crest onto the upper slopes of lagam. Although no longer actively settled, the lowermost portions of its territory border the channel of the Embobut River as it flows down the Escarpment to the immediate north of Tot-Sibou and on to its confluence with the Kerio.

• 87 M.I.J. Davies, H.L. Moore, 2016; M.I.J. Davies et al., 2014.
• 88 H.L. Moore, 1986, p. 25-29.

21Unlike the four clans which together comprise Sibou, the settlement of Kacheseker is coterminous with a single patrilineal clan of the same name, as is typical for Embobut. Likewise, individual-clan settlements within Embobut typically abut one another in a continuous spread extending along the length of the Escarpment, as opposed to the more nucleated multi-clan settlements of the Valley below. This distribution is enabled by the relatively plentiful and widespread availability of rain-fed water sources on the upper Escarpment, whereas in the Valley people must rely on the sparser occurrence of rivers and artificial irrigation channels. Similarly, whilst in settlements such as Tot-Sibou habitation tends to cluster close to the foot of the Escarpment slope and farmland to lie further into the Kerio floodplain,87 in Kacheseker both housing compounds and their attendant fields are spread throughout the clan’s total territory. Typically, these housing compounds are linearly arranged along the ridges of hill spurs, with different patrilineage-based kin groups occupying distinct ridges, divided by small streams and gullies or other natural features used to mark their respective territories.88 Steeply pitched fields extend across the slopes between and below these compounds, the majority of them now planted with some combination of maize, beans and/or potatoes.

• 89 J.W. Meyer, C.L. Crumley, 2011.

22Field research into the settlement history of Kacheseker was undertaken by myself from 2017–2020 as a part of my doctoral studies. Although the overarching project focussed on several different aspects of mobility in habitational practice, material engagement, and environmental processes (cf. Kay 2021), this paper focusses on the overarching trajectory and spatiotemporal development of Kacheseker’s inhabited zone since its foundation approximately 200 years ago. Within this temporal arc, I further focus on how present-day and recent oral historical data can be used to shed light on the settlement’s earlier history. In doing so, I draw particular inspiration from the mixed methodological approaches advocated by many historical-ecologists,89 in which different forms of data are used to compliment, support, and critique each other to enable a more holistic understanding of the subject in question. This paper thus combines data gleaned both from extensive archaeological and contemporary household/landscape surveys with a suit of oral historical testimonies recounted by Kacheseker’s current inhabitants.

• 90 D.K. Kay, 2021, Appendix 2.

23The majority of these interviews took place in June–July 2017 and February 2018 and involved 35 community members, most of them clan members and/or residents of Kacheseker itself. All interviews followed a semi-structured format, and were primarily conducted in English and KiMarakwet through the translation efforts of my friend and local collaborator, Mr Joseph Kimutai Cheptorus.90 These interviews all followed a broadly similar trajectory of talking through the interviewee’s personal life history, in addition to what else they knew of Kacheseker’s past, and their own aspirations for the future. They varied from well over an hour to barely 30 minutes long, depending on the interest of the interviewee and their other time commitments. I attempted to interview a broad spread of local community members, though the final corpus remains somewhat weighted towards conversations with older men. However, this situation does confer some benefits, as given the patrilocal nature of settlement in the region such men have lived within the village longer than anyone else. Nevertheless, it should be remembered that this demographic skew will entail its own biases, particularly regarding gender, which future research could do well to redress.

• 91 D.K Kay, 2021, Appendix 3.

24Fieldwork similarly encompassed two distinct phases of landscape survey. The first entailed a comprehensive survey of all currently inhabited domestic compounds in July 2017. Kacheseker’s territorial boundaries were first established using an A0 size satellite image print-out of the area, whereby a group of community members gathered to debate where the boundaries lay and draw them in. Some of these boundaries, particularly those lying within the gazetted Embobut Forest, were later ground-truthed with the assistance of people who lived (or used to live) in that specific location. With the boundaries established, the inhabited portions of Kacheseker were then divided into sub-areas and every housing compound therein was mapped in-person using the GaiaGPS smartphone app. The type and number of physical structures within each compound was also recorded, alongside further information on when it was first built and the makeup of its resident household. A total of 231 housing compounds were recorded in this way, and the resulting data amalgamated using Microsoft Excel and ESRI ArcGIS to create a comprehensive geo-referenced database of all the occupied housing compounds within Kackeseker, complete with information on approximate dating, demography, and architectural composition.91

• 92 M.I.J. Davies, 2008, 2012; M.I.J. Davies, H.L. Moore, 2016; D.K. Kay, 2021; S. Lunn-Rockliffe, 2018
• 93 D.K. Kay, 2021, p. 359-364.

26Community origin stories across Marakwet often centre on tales of migration, socially rooted decision making, and a deep-felt engagement with the landscape.92 Kacheseker is no different, and though the details are often disputed, the settlement’s foundation is widely attributed to a man called Kipkembich who had fled up from the Kerio Valley after accidentally causing the death of his nephew. After a long, arduous journey he eventually encountered the elders of Kamoi clan (to the immediate north of present-day Kacheseker), who subsequently gifted him land and a wife. Following an act of incest with a relative of his new wife, Kipkembich’s offspring were then divided into the two clans of Kaptul and Kacheseker (the former now lying immediately south of the latter).93 The three clans of Kamoi, Kacheseker, and Kaptul thereafter retained a strong sense of their interlinked history and are still often referred to locally by the shared toponym of ‘Noiso’.

• 94 B.E. Kipkorir, 1973, p. 9-11.
• 95 H.L. Moore, 1986, p. 58.

27Establishing the putative timing of these semi-mythic events will always be a matter of supposition. However, the Marakwet’s strong emphasis on the importance of patrilineal descent in establishing rights to land and resources, coupled with their system of age-sets, provides an additional means for establishing an approximate relative chronology encompassing Kacheseker’s origins and subsequent historical development. The cultural institution of age-sets is ubiquitous throughout Marakwet, as it is for the Kalenjin more broadly.94 Age-sets are essentially a system for ordering the separate initiation of male and female peers into named, generationally sequential groups. Traditionally at least, initiation involves a period of ritual seclusion and learning followed by a more public (and nowadays principally male only) circumcision ceremony, from which the initiates emerge as adults ready to fully enter social life, marry, and start families of their own. Initiations occur every few years, with all those initiated belonging to the same age-set, until such time as the elders agree that the set is closed and a new one should commence. Men retain this age-set name alongside their clan identity all their life, whereas women will typically become affiliated with both their husband’s clan and age-set upon marriage.95 It is therefore on male age-sets that the following discussion centres, given their geographical continuity and pre-eminence in community discourse.

• 96 S. Lunn-Rockliffe, 2018, p. 124; H.L. Moore, 1986, p. 57.

30The length of each age-set varies according to time and place, but on average they last for approximately 15 years, with the whole cycle taking 100–120 years to complete.96 The recollection of specific initiation years provides clear start and end dates for more recent age-sets, whilst intervals become more approximate further back in time. It is also notable that the certainty with which people speak of the age-set sequence rarely extends beyond its current cycle. Thus, whilst many interviewees counted back to the previous iteration of Nyongi, there was far less agreement as to the relative ordering of yet earlier age-sets. Present-day events also remind us that the system itself is liable to change. For instance, the most recent set of Nyongi has lasted far longer than is usual, as its supposed successor—Maina—was associated with terrible droughts in its previous incarnation. As age-sets are believed to repeat their key characteristics through time, so the elders were keen to avoid the recurrence of such hardship and delayed its onset for some time. It was agreed after the 2016 initiations that Nyongi should finally be ended, but as yet the community has still to decide whether the next set of initiates should be named Maina, or whether to skip onwards to Chumo in the hope of avoiding the recurrence of past misfortunes.

• 97 Timothy Kipkeu Kipruto, pers. comm. June 2020.
• 98 S. Lunn-Rockliffe, 2018.
• 99 H.L. Moore, 1986.
• 100 M.I.J. Davies, 2009b.
• 101 S. Lunn-Rockliffe, 2018, p. 125.

31Even with such caveats, it is still possible to roughly correlate the dating of age-sets across the region. When community elders announce a set transition, they do so with respect to the concurrent discussions of elders in other nearby communities, so that the wider cultural institution remains more-or-less in sync with itself through time. This creates a kind of ripple effect, whereby the start of a new age-set typically first occurs in the south of the Kerio Valley, with communities further north progressively following suit.97 This contextual basis to local decision making maintains the integrity of inter-community male genealogical reckoning for the Marakwet themselves and enables the comparison of chronologies from across the immediate region (see Table 1). In this respect, my own assessments of age-set dates correlate almost exactly with those of Lunn-Rockliffe98—unsurprising given the Sengwer and Marakwet clans of Embobut typically initiate their youth together—whilst also aligning closely with Moore.99 The discrepancies with the Pokot dates reported by Davies100 are likewise unexceptional given their slight cultural and geographical remove, especially as the Pokot typically ‘follow’ the Marakwet sequence by a few years’ interval.101 The age-set system thus works well as a relatively fine-grained, albeit somewhat approximate, local chronology extending across the past 150 years.

• 102 After M.I.J. Davies, 2009, Table B.1; S. Lunn-Rockliffe, 2018, Table 4.1.

Table : Comparison of age-set chronologies within Embobut and surrounding area102

• 103 M.I.J. Davies, H.L. Moore, 2016, p. 73; M.I.J. Davies et al., 2014, p. 514.
• 104 S. Lunn-Rockliffe, 2018, p. 125-126.

32This chronology of age-sets becomes markedly hazier with time depth, such that genealogical reckoning at any further remove than the last three–four generations is highly ambiguous. This situation is largely due to the highly political nature of reckoning patrilineal descent, and the pre-eminence of particular male ancestors/patrilines within wider community discourse. The individual lineages recounted by Kacheseker men thus rarely accord with each other to any precise degree of exactitude. Nevertheless, it is still possible to use the age-set system to roughly gauge when Kipkembich’s sons Sakero and Kibarynir may have been initiated, and thus when Kacheseker itself may have been founded. Accordingly, though the identification of Sakero and Kibarynir’s age-sets varied greatly, most Kacheseker interviewees’ genealogies were in closer agreement regarding ancestors back to Chumo age-set. Counting back by approximate generation, I tentatively suggest that Sakero and Kibarnyir were initiated one, maybe two, age-sets before the last cycle of Nyongi (see Table 1). Their father Kipkembich (whether an actual historical figure or the mythologised version of a larger group migration) would thus have arrived in Embobut sometime around 1850, with Kamoi already having settled nearby perhaps a decade or so previously. Whilst imprecise, this dating of Kacheseker’s foundation to the mid-19th century fits the regional context—with the Marakwet settlement of the Kerio Valley occurring c. 250–300 years ago103 and the Sengwer clans of the highland Embobut Forest somewhat later during the mid- to late 19th century.104

• 105 H.L. Moore, 1986, p. 91-99.

36Dating the pace of this shift in settlement is again challenging, though analysing survey data alongside oral historical information provides a good starting point. For instance, Figure 11 displays the age-set of the male head of every contemporary household in Kacheseker, revealing a clear preference for younger Nyongi men to live above the Escarpment crest, but less so for the preceding age-sets of Kimyikeu, Kaplelach, and Kaberur. However, present distributions do not necessarily reflect the past habitation locations of either specific individuals or age-sets as a whole, given the preponderance for individuals to move multiple times throughout their lifetime and/or reoccupy older platforms built by previous generations of their patrilineage (see below).105 Thus, Figure 12 displays the decade in which each currently inhabited compound was built. This sequence begins in the 1960s with compounds predominantly built by men of Kaberur or Kaplelach age-set and continues until the 2010s, by which point all four extant age-sets were actively involved in house building. These distributions again reveal a general trend of progressive movement upslope, though it is notable that the 2010s have also seen a significant revival of settlement on lagam (see further below). Figure 12 further highlights the high rate of pan-generational turnover in housing sites, showing that the bulk of Kacheseker’s contemporary house compounds were built within the last four decades, regardless of their present age-set affiliation.

• 106 H.L. Moore, 1986.
• 107 M.I.J. Davies 2009, 2013a.

40At a fundamental level, the manner in which patrilineal descent and land inheritance typically function amongst Marakwet communities itself encourages the gradual expansion of settled/farmed areas within—and even beyond—a given clan’s territory. As recounted in detail by Moore106 (and amongst the nearby Pokot by Davies),107 the division of land by fathers amongst their sons will necessarily involve a decrease in the size of each generation’s individual landholdings. As such, new land must be cleared and made agriculturally viable if sequentially proliferating households within the patriline are to maintain a sustainable level of self-sufficiency with regards to food production. Moreover, the spatially expansive drive of this general trend is exacerbated in environments where arable land requires fallowing to retain its overall productivity, and in which populations are themselves increasing over time.

• 108 D.K. Kay, 2021, p. 84-85; S. Lunn-Rockliffe, 2018, p. 162-165.

41Both these situations pertained to Kacheseker during the 1950s, when settlement was largely restricted to the dry slopes of lagam lying below the upper crest of the Elgeyo Escarpment. Households then cultivated sorghum, millet, and local maize landraces (known as kipengara and kisim) in rotational plots close to their houses and grazed small numbers of livestock (predominantly sheep and goats) along the edge of the Embobut Forest, which then ran along the line of the Escarpment crest. Whilst this forested landscape to the west was deeply important for both individual and community lifeways,108 from the early 1960s onwards increasing numbers of young men from the age-set of Kaplelach began to move upslope to clear the trees for new arable fields. In part, these clearances simply marked the continuation of the gradual westward drift of settlement that had characterised Kacheseker since its foundation in the late 19th century. However, today’s elders are keen to point out that a second major factor was the local arrival of chebolos maize in the years around Kenyan independence in 1963. Chebolos was a much more productive variety of maize than kisim and kipengara, but unlike the latter it did not grow well on the arid slopes of lagam. However, the rain-fed lands to the west of the Escarpment crest were ideal, especially in combination with rich former-forest soils that did not require extensive fallow periods to remain productive over multiple seasons. The young men of Kaplelach thus applied themselves with great effort to expanding this nascent agricultural landscape upslope of their childhood homes, first as dispersed individual plots and then increasingly as an extension of the spatial land-use pattern evident on lagam, whereby dispersed domestic compounds were surrounded by larger attendant field systems.

• 109 E.g. J.S. Dean, 1996; M.E. Moseley, 1972; A.M. Prentiss et al., 2014; A.H. Sirén, 2007.
• 110 D.K. Kay, 2021, p. 88.
• 111 D.K. Kay, 2021, p. 87.

42In contrast to many models of subsistence-supported demographic expansion,109 members of the Kaplelach generation attest that the primary incentive for expanding chebolos cultivation at the expense of preserving existing forest resources was to increase social capital, rather than to achieve greater food security per se. In particular, increased maize cultivation enabled the fuelling of ever-larger ceremonial feasts, especially during the culturally important marriage ceremony of tum, in which the ability to produce large quantities of food and alcohol greatly elevated the social standing of the hosts.110 Indeed, it is often said that amongst the men of Kaplalach there were none who were not married and that so much maize was being made into beer (busaa) that large-scale ceremonies became frequent affairs across the whole region of Embobut. Accordingly, it was the appeal of cultivating chebolos within an existing mode of significant cultural production that underpinned the south-westward expansion of agricultural land within Kacheseker throughout the 1960s and early 1970s. That said, the effects of concurrently greater subsistence capabilities cannot be discounted. Whilst population growth is extremely hard to quantify in this region,111 the uphill expansion of Kaplelach is locally reported as marking the beginning of a period of sustained demographic expansion that continues to this day.

• 112 S. Lunn-Rockliffe, 2019.

45Kacheseker’s recent settlement history also exemplifies how contemporary patterns of residential mobility continue to integrate long-standing practices of land-use within the development of new trajectories. As mentioned above, both the conflict of Kacheseker and Kaptebego over Lemeiywo and the federal government’s forcible removals from the gazetted Embobut Forest have resulted in a revival of settlement across the upper slopes of lagam. In both cases, many of those who have moved back to lagam have done so due to a lack of other viable alternatives. However, their actions still directly draw on the long-standing principles of patrilineal inheritance and communal land ownership which have underwritten the changing land-use of Kacheseker’s overarching territory throughout its history. Furthermore, the expulsion of permanent settlement from within the Forest boundaries has transformed its formerly inhabited arable landscapes into ones characterised by wide grassy meadows separating patches of thicker tree cover. These grasslands have in turn opened up new possibilities for the pasturage of expanding livestock herds, including a marked increase in the number of cattle relative to sheep and goats. As explored in detail by Lunn-Rockliffe,112 whilst many local livelihoods were unquestionably disrupted, if not outright destroyed, by the 2013 evictions, Kacheseker and the adjoining communities of Noiso nevertheless continue to enfold the new ecological affordances of the Forest within their creative and constantly shifting lifeways.

• 113 M.I.J. Davies, H.L. Moore, 2016; M.I.J. Davies et al., 2014; S.F. Derbyshire et al., 2020; S. Lunn- (...)

49As discussed above, the theoretical and methodological tenets of historical ecology are particularly well suited to countering such models, particularly through approaches that investigate the ‘resilience’ of diverse systems and fields of action. The ‘shifting sedentism’ of Marakwet habitational practices offers an exemplary case study in this regard, as it is their in-built flexibility and capacity for adaptive change that have most keenly contributed to the long-term continuity of the community at large (as is likewise evident across diverse other aspects of Marakwet social and environmental history).113 Historical analysis has shown that mobility has long been key to this resilient flexibility, and a principal means by which diverse social, ecological, and material phenomena have coalesced within complex systems of interaction, in which settlement has always occupied a place of relative flux. Accordingly, mobility of settlement can be seen as a near constant within Kacheseker’s overarching history, even though the precise direction and form of that mobility has never been linearly constrained. Instead, historical changes in settlement location and material expression have arisen through the cumulative habitational practices of Kacheseker’s constituent households over the course of many generations—practices that have always been realised in creative tension with the various affordances of the surroundings in which they are immersed, be they material, biotic, socio-cultural, or more intimately personal.

• 114 D.M. Anderson, 1988; O.J.C. Boles et al., 2019; P. Lane, 2009; P.D. Little et al., 2008.
• 115 E.g. M. Graham, 1994; E.E. Jones, J.W. Wood, 2012.

51In this vein, future inquiries into the diverse history of shifting sedentary settlement systems across the African continent would do well to emulate the interpretive strides already made in the study of mobile practices amongst predominantly pastoralist populations.114 As with these prior studies, it is the detailed investigation of specific contexts, combined with their subsequent comparative analysis, that will enable us to move beyond purely axiomatic considerations of mobility and habitational movement within otherwise sedentary communities. Doing so would also place African and Africanist scholars in an excellent position to contribute to global discussions regarding the spatiotemporal diversity of flexible land-use practices and (semi-)itinerant lifeways—for instance as similarly noted amongst both historical and contemporary ‘mobile farmers’ in North America.115

• 116 C.M. Cameron, S.A. Tomka, 1993.
• 117 M. Lamoureux-St-Hilaire, S. Macrae, 2020a.
• 118 M. Lamoureux-St-Hilaire, S. Macrae, 2020b, p. 4.
• 119 See particularly C.M. Cameron, 2020; P.A. McAnany, M. Lamoureux-St-Hilaire, 2020.

52From a specifically archaeological perspective, questions of site (and even wider landscape) abandonment often continue to be approached along somewhat simplistic lines, despite multiple authors having long pointed out that the processes involved are invariably complex and highly context-specific.116 More recent research has reiterated these points, particularly that collated by Lamoureux-St-Hilaire and Macrae117 in their treatment of settlement abandonment as a process of ‘detachment from place’, in which material objects, social relations, environmental factors, and personal emotions all interact within complex systems of settlement and landscape transformation. By regarding “the conundrum of sedentism” as largely an “historical illusion”,118 they instead focus in on the cross-scalar histories of constant—and often recursive—movement that pattern all human experiences of ‘place-making’, and concomitantly of place ‘unmaking’.119 Whilst the patterns of shifting sedentism that form the focus of this paper are perhaps not directly tied to the phenomenon of wholesale settlement abandonment, they too demonstrate that practices of habitual mobility are integral features of what, from a single point in time, we might otherwise regard as a ‘sedentary’ community. Likewise, such mobility necessarily encompasses highly complex trajectories of abandonment, re-occupation, and changing land-use within its overarching, fluid history, and it is this fluidity which is itself so crucial to the realisation of continuity through change.

• 120 Sensu S. MacEachern, 2016; M.C. Beaudry, T.G. Parno, 2013; J. Leary, 2014.

53It is this agentive tension which has commonly been neglected in the portrayal of rural agrarian communities across sub-Saharan Africa, much as the mobile practices that embody it have often fallen below the scope of archaeological enquiry. By instead foregrounding mobility as a central element in the interweaving of diverse strands of evidence concerning past settlement change,120 this paper hopes to have demonstrated that we can productively work towards recentring these communities’ intrinsic historical agency—the same agency with which they move into their futures.