Recurrence or Persistence? Coding Residential Form from Neolithic Anatolia to the Financialised City

1. Introduction

Housing is more than the physical setting of everyday life. It is where people seek privacy, care for children and older relatives, keep possessions, share meals, express identity and negotiate the boundary between individual and collective life. Residential space is never entirely private: its size, location, tenure, construction and internal organisation are shaped by forces operating well beyond the household, including land ownership, regulation, technology, social convention, environmental conditions and access to finance. As Habraken argues, the built environment emerges through an unceasing interaction between people and the forms they inhabit (Habraken, 1998).

This tension between intimacy and structure makes housing both universal and historically specific. Every society must provide places in which people can live, but societies have met that need in different ways. The roof-accessed houses of Çatalhöyük, the courtyard dwellings of Mesopotamia, the Roman domus and insula, the medieval tenement, the industrial terrace and the contemporary co-living complex were not simply successive architectural styles. Each combined a particular settlement pattern, household structure, tenure arrangement, construction technology and understanding of the relationship between private life and collective order (Després, 1991).

Contemporary debates often lose sight of this longer history. Housing affordability is commonly presented as a crisis produced by recent shifts in finance; densification as a response to twenty-first-century urbanisation and environmental pressure; and shared living as an innovation of digital platforms and changing lifestyles. These readings capture important features of the present, above all the growing treatment of housing as an asset for financial accumulation (Aalbers, 2016). Yet earlier societies also confronted crowding, speculative letting, unequal access to land, collective infrastructure and the regulation of building quality, albeit within economic and political systems very different from those of the modern city (Mallett, 2004). Recognising common elements does not mean treating a Roman insula as equivalent to an investor-owned apartment block, or a monastic dormitory as an early co-living scheme. It means using historical benchmarks to separate conditions that are genuinely new from tensions that have been reorganised by new institutions, technologies and forms of capital (Shirwani, Kamran, & Malik, 2019).

Housing must therefore be studied not only as architecture but as an environment that is continually occupied, modified and transformed. Brand describes buildings as artefacts that acquire a life after construction, adapting to changing users rather than remaining fixed expressions of an original design (Brand, 1994), and Moudon defines urban morphology as the study of the city as a human habitat formed through the accumulated actions of individuals and groups (Moudon, 1997). From this perspective, dwellings, plots and streets form layered historical systems in which physical form both constrains and enables later occupation.

Long-range histories of housing, including synoptic frameworks of the kind developed here, share a characteristic weakness. Because they range across millennia and continents, they can assemble resemblances almost at will, and a resemblance once noticed is easily promoted into a constant. When the cases are chosen because they display the pattern of interest, the comparison cannot fail to confirm it. This article takes a different route. Rather than asserting continuity, it asks how continuity could be recognised and what would count against it. Eighteen residential cases, extending from Çatalhöyük to contemporary co-living, are coded for three candidate patterns derived from existing scholarship. Absences and ambiguous cases are reported alongside presences, and each case is graded for the quality of its evidence. The cases were not originally selected for this purpose, and the article treats that limitation as part of the analysis rather than as an afterthought.

The cases are arranged in five broad periods: Urban Revolution, Classical Codification, Medieval Communalism, Industrial Standardisation, and Modernist and Financialised Housing. The periodisation is a heuristic for organising evidence, not a developmental sequence; it does not imply that housing systems developed linearly, that all regions passed through the same stages, or that pre-modern and modern institutions were equivalent. Two research questions guide the analysis.

RQ1. Which of three candidate patterns (buffer-space mediation, the coexistence of speculative and secure tenure sectors, and non-market allocation) recur across eighteen coded residential cases, and in which cases and periods are they absent?

RQ2. Where do the analytical frameworks of Hillier and Hanson, Kemeny and Aalbers require qualification when applied to pre-industrial evidence?

The article makes two contributions. Substantively, it shows that the patterns most often cited as evidence of residential continuity are less general than synoptic accounts suggest, and that their absences are informative: none of the speculative-rental cases contains an unambiguous domestic buffer. Methodologically, it offers a transparent coding procedure that allows the framework to be extended, contested and falsified by others, which a narrative synthesis does not. The argument is not that history repeats itself in residential form, nor that it does not. It is that claims of recurrence in housing history should be stated in a form that could be shown to be wrong.

2. Background

2.1. Analytical Frameworks

Three bodies of scholarship frame the analysis. Phenomenological accounts, beginning with Bachelard’s The Poetics of Space (Bachelard, 1994) and Heidegger’s treatment of dwelling as an ontological rather than functional category (Heidegger, 1971), establish that the house organises memory and identity as well as shelter. Cross-cultural architectural scholarship, above all Rapoport’s House Form and Culture (Rapoport, 1969), explains vernacular form as the joint product of climate, materials and social organisation, while Rykwert traces Western theory’s search for an original dwelling (Rykwert, 1972). These works supply much of the descriptive vocabulary used below. They are not tested here, because their central claims do not specify conditions under which they would fail.

Hillier and Hanson’s space syntax provides the comparative indicators for the first candidate pattern (Hillier & Hanson, 1984). Its central measures, depth (the number of spaces that must be crossed to reach a room from the exterior) and integration, allow dwellings from different cultures to be compared without imposing modern functional categories, and its argument that configurational depth tends to track social distance underlies the expectation that thresholds between street and dwelling encode privacy. Hillier and Hanson caution that configuration describes spatial logic rather than lived experience. In this article depth is assessed qualitatively from published plans rather than computed, a limitation returned to in Section 3.5.

Kemeny’s housing-regime theory distinguishes dualist rental systems, in which a public or non-profit rental sector is held apart from the profit-driven market and tends to be residualised, from unitary systems, in which cost-rental and profit rental housing compete within a single market (Kemeny, 2006). The distinction concerns the organisation of rental provision in modern welfare states. It is invoked here for a narrower purpose: to ask whether the coexistence of a speculative sector with a secure or regulated one, a looser and older phenomenon, has pre-modern precedents. The article calls that looser phenomenon tenure stratification and does not treat it as equivalent to Kemeny’s dualism.

Aalbers’s account of financialisation describes the subordination of housing’s use value to its value as an asset for financial accumulation (Aalbers, 2016). Speculative tenement construction in the industrial city, where landlords prioritised rental yield over habitability, is a historical antecedent of this tension but not its equivalent (Engels, 1993; Wohl, 1977); the scale, velocity and abstraction of contemporary housing finance, through securitisation, real-estate investment trusts and algorithmic investment, are without precedent (Zaloom & James, 2023). Recent work extends the thesis to co-living, documenting the institutionalisation of shared rental housing (Ronald, Schijf, & Donovan, 2023) and the role of housing ideology in legitimising its expansion (White & Madden, 2024). For the period evidence, the analysis relies on archaeological and architectural-historical syntheses, including Kostof (1995), Rybczynski (1986) and Grenville (1997) for the earlier periods and Hobsbawm (1962), Curtis (1996) and Glendinning (2021) for the industrial and modern periods.

2.2. Recurrence and Persistence

The historical-geographical tradition of urban morphology, developed from Conzen’s town-plan analysis of Alnwick (Conzen, 1960) and consolidated in what Whitehand calls the Conzenian tradition (Whitehand, 2001), offers a precise concept of persistence. A fringe belt, a zone of extensive, low-intensity land use that forms at the edge of a town during a pause in outward growth, is later overtaken by expansion but rarely erased; Whitehand and Morton show Birmingham’s Edwardian fringe belt surviving structurally, if not functionally, across a century of planning regimes (Whitehand & Morton, 2004). Contemporary research confirms that inherited plots and street systems can limit what planning interventions achieve (Gallagher, Sigler, & Liu, 2025). Persistence in this sense is a claim about inheritance: an earlier form survives in a particular place and conditions what is subsequently built there.

That precision matters for long-range comparison. When a courtyard appears in Bronze Age Ur and in nineteenth-century Guise, no chain of inheritance connects the two. The resemblance may reflect functional convergence, since courtyards moderate climate, secure property and control access; diffusion through architectural theory; or coincidence. Calling such resemblances persistence stretches a concept defined by place-specific inheritance to cover unconnected cases, the kind of conceptual stretching Sartori identified as a characteristic hazard of comparative research. This article therefore uses recurrence for the appearance of a coded feature in temporally or geographically separate cases, and reserves persistence for documented inheritance within a morphological lineage. The eighteen residential cases can test recurrence. None documents inheritance across periods, and no claim of cross-period persistence is made.

A further reason for caution is geographical. Housing financialisation is variegated rather than uniform, producing different landlord strategies and rent-gap dynamics in Toronto (August & Walks, 2018), Tokyo (Aveline-Dubach, 2020) and Rotterdam (Nelson, Warnier, & Verma, 2024), even though all three exhibit the decoupling of exchange value from use value that Aalbers identifies. The same caution applies retrospectively. The pre-modern record available for comparison is dominated by Western European and Mediterranean sources, so a pattern that recurs within that record may reflect which regions have been most intensively excavated and archived rather than a cross-cultural regularity.

3. Methods

3.1. Comparative-Historical Design

The study uses a comparative-historical case design suited to questions about recurring patterns across cases too dispersed in time and space for controlled comparison or statistical aggregation. The unit of analysis is the residential typology in context: a characteristic dwelling form situated within its regulatory, economic and cultural setting, rather than an individual building. The units are intentionally heterogeneous, ranging from excavated settlements and reconstructed types to individual buildings and market segments. The comparison is therefore typological rather than a matched comparison of equivalent buildings. Where an entry aggregates several constituent forms, it is coded only when the cited evidence supports the assigned category across the aggregate; otherwise it is coded mixed, ambiguous or indeterminate. The design is descriptive and proposition-generating. It can establish whether a coded feature is present or absent in the cases examined; it cannot establish why.

3.2. Case Selection and Its Limits

Eighteen residential cases were selected purposively to provide chronological coverage across the five periods, some geographical and cultural range within the limits of available sources, and documentation sufficient for configurational and tenure coding. A descriptive matrix of the eighteen cases is provided as Supplementary Material A. Two features of the selection must be stated plainly. First, the cases were chosen before the coding scheme reported here was specified, and several were chosen because they are canonical illustrations of the patterns now being coded. The selection is therefore not independent of the outcome, and risks the bias Geddes describes, in which cases selected on the dependent variable confirm whatever relationship is sought. Vitruvius’s treatise and the Bullitt Center are retained only as contextual references in the period narratives. Neither is a unit of analysis or included in the descriptive matrix or pattern counts.

Two procedures limit, without eliminating, the consequences of purposive selection. Every case is coded on every pattern, so that absences carry the same weight in the results as presences. The sample also contains internal contrast cases, among them Çatalhöyük, the Roman insulae and the industrial speculative tenement, which were included for chronological coverage rather than as illustrations of any pattern, and against which the patterns can fail. A design that added deliberately chosen contrast cases from outside the present sample would be stronger; Section 5.4 specifies what such cases would need to show.

3.3. Coding Scheme

Each residential case was coded on four variables. Buffer space (Pattern 1) combines two analytically distinct attributes: configurational mediation, meaning that one or more intermediate spaces lie between the public access point and the principal domestic area; and access-control function, meaning that a household or residential institution can regulate movement through that sequence. It is coded present where a published plan or description identifies a courtyard, atrium, cloister or equivalent enclosed space that both mediates access and operates as a domestic threshold; absent where the dwelling is entered directly from the street, roof or common stair without such mediation; ambiguous where an intermediate space exists but its access-control function is contested, collective or unclear; and indeterminate where the sources do not permit judgement. This distinction prevents a courtyard, light well, defensive ward and internal street from being treated as functionally equivalent merely because each increases spatial depth. Tenure sector records the sector to which the case belongs: speculative rental, regulated or reformist provision, ownership (including elite ownership), institutional, communal, state-sponsored, or not documented. Pattern 2, tenure stratification, is assessed at the level of a historically delimited urban or political-economic context. It is coded present only where the evidence establishes the contemporaneous coexistence of a speculative rental sector and a secure, regulated or reformist sector within the same city, urban system or closely comparable housing market. Cases separated substantially in time or institutional geography are reported as suggestive contrasts rather than evidence of coexistence.

Allocation (Pattern 3) records the dominant mechanism by which occupants obtained the dwelling: ownership, market rental, non-market allocation (by kinship, institutional role, employer or state), mixed, or not documented. Shared facilities records whether residents shared domestic facilities, such as sanitation, cooking or dining space, or building-level residential amenities. Collective neighbourhood infrastructure, including drainage networks, is coded separately and is not treated as evidence of shared domestic facilities. This variable was separated from allocation during analysis, after it became clear that narrative accounts of communal housing, including earlier drafts of this framework, had treated the two as one (Section 5.3). Codes were assigned from the published sources cited for each case in Section 4. They are the author’s judgements, are reported in full in Table 1, and are open to contestation case by case.

3.4. Evidence Grading

Each case was graded for the evidence on which its coding rests. Grade A denotes primary evidence reported directly, through excavation, survey of standing fabric, contemporaneous documentation or original empirical research. Grade B denotes a specialist synthesis of primary evidence. Grade C denotes a general synthesis, an idealised or unexecuted plan, or a description by the case’s own promoter. Grades describe the evidence used here, not the quality of the underlying scholarship.

3.5. Limits of the Method

Several limits follow from the design (Lange, 2013). Density estimates for pre-modern periods are order-of-magnitude approximations drawn from secondary sources; where figures vary, the range is reported rather than adjudicated, and density is therefore described but not coded. Configurational depth is judged qualitatively from published plans rather than computed from justified access graphs, so Pattern 1 codes are coarser than a full space-syntax analysis would allow. Coding was carried out by a single coder, and no inter-coder reliability can be reported. The evidence is radically uneven across regions, with Western European and Mediterranean cases dominating the pre-modern record. Finally, applying modern regime and financialisation theory to pre-modern contexts is analogical: it is used to frame questions, not to claim equivalence between, for instance, medieval guild tenure and a portfolio owned by a real-estate investment trust.

4. Results

4.1. Period I, Urban Revolution (c. 7500–2000 BCE)

The Urban Revolution, Childe’s term for the transition from dispersed agricultural settlement to organised urban centres (Childe, 1950), produced the first residential forms showing deliberate planning, material standardisation and evidence of communal regulation. Çatalhöyük (Anatolia, c. 7500–5700 BCE), though it predates the Urban Revolution proper, is included as the earliest case: a settlement of contiguous mudbrick houses entered by ladder through roof openings, with no street network (Hodder, 2006; Mellaart, 1967). Similar house sizes and the absence of monumental differentiation led Hodder to argue for a relatively egalitarian society, although Düring identifies household-level ritual differentiation in wall paintings and sub-floor burials (Düring, 2011). For the present analysis Çatalhöyük is a contrast case. Domestic space was entered from above, directly into the main room, and no courtyard or equivalent threshold intervened between the roofscape that served as shared space and the domestic interior.

The Mesopotamian courtyard house (Ur and Uruk, c. 3000–2000 BCE) is organised around a central court that provides passive climate control, private outdoor space and a threshold between street and interior. Cuneiform property records from Old Babylonian contexts document house sale, rental and inheritance, among the earliest evidence of houses as transactable property rather than purely kin-allocated space. This qualifies any reading of tenure markets as specific to modern states in one limited respect: transactable residential property long predates them. It does not establish a tenure regime in Kemeny’s sense. At Mohenjo-daro (c. 2600–1900 BCE), standardised brick sizes and gridded streets were combined with individually walled courtyard houses connected to a shared covered drainage system. The infrastructure is consistent with centralised regulation, although the institutions responsible remain debated. Because a collective drainage network is neighbourhood infrastructure rather than a domestic room or fixture shared by residents, it is not counted as a shared domestic facility. Two of the three Period I cases therefore display a courtyard threshold; the earliest does not. No speculative rental sector is documented for the period.

4.2. Period II, Classical Codification (c. 800 BCE–500 CE)

The Classical period produced the first surviving systematic theory of residential design in Vitruvius’s De Architectura (c. 30–20 BCE), which treated firmitas, utilitas and venustas as co-equal criteria and explicitly linked architectural form to social and civic function (Vitruvius, 1999). The Classical Greek house was typically organised around a single courtyard entered from the street through one door, with the andron, the men’s dining and reception room, often placed near the entrance (Nevett, 1999). Nevett’s reading of the archaeological evidence qualifies the textual image of a fixed women’s quarter: rather than a permanently segregated gynaikonitis, houses show a courtyard through which access to every room could be controlled, so that separation of household members from visitors was achieved by managing movement through a single threshold. The courtyard functions here as a buffer in the strict sense used in the coding.

The Roman domus at Pompeii (c. 2nd century BCE–79 CE) is organised around an atrium and proceeds along an axial sequence from the vestibulum and atrium through the tablinum to more private rooms, a sequence Wallace-Hadrill relates closely to the daily salutatio of patron–client relations (Wallace-Hadrill, 1994). Alongside the elite domus, Roman cities contained multi-storey rental accommodation, often poorly built and prone to collapse and fire, which housed much of the urban population. Wallace-Hadrill cautions, however, that in Pompeii and Herculaneum rented shops and upper-floor units were frequently attached to or embedded within larger houses, so that elite and rental housing were less spatially segregated than a two-sector model implies. Insula accommodation was typically entered from the street or a common stair without an intervening threshold. Period II thus provides the clearest early instance of tenure stratification, with secure elite ownership coexisting with speculative rental, but the stratification is social and economic rather than a separation of rental sectors in Kemeny’s sense.

4.3. Period III, Medieval Communalism (c. 500–1400 CE)

After the fragmentation of the Western Roman Empire, residential architecture reoriented around feudal-military authority and monastic communalism. Conwy Castle (Wales, 1283–1289) combined concentric defences with noble residential quarters (Taylor, 1986). Its inner ward produces great depth between the outer gate and the royal apartments, but that depth encodes defensive logic at least as much as social hierarchy, a qualification to the predominantly social reading of depth in Hillier and Hanson (Hillier & Hanson, 1984). Whether the ward functions as a domestic threshold or as military space is ambiguous, and it is coded accordingly. The Plan of St Gall (c. 820 CE) sets out an idealised Benedictine monastery, with dormitory, refectory and cloister arranged according to the Rule’s stipulations on silence, privacy and hierarchy of access (Horn & Born, 1979). The plan was never built as drawn and records a normative programme rather than an observed settlement, for which reason it is graded C. It nonetheless provides a textually codified case of non-market allocation, in which residence followed institutional role rather than purchase, rent or inheritance, organised around a cloister that functions as an enclosed threshold.

Urban tenements in towns such as London, Florence and Ghent developed from the twelfth century as narrow, multi-storey, timber-framed buildings on constrained plots, often with a shop at the street front and living space behind or above (Schofield, 1994; Grenville, 1997). Regulation of materials, party walls and fire safety after major conflagrations can be read as an early form of municipal building control. The sources used here document this regulated sector; they do not document a contemporaneous speculative sector in a form that would allow the two to be compared. Gothic structural innovation, with its pointed arches and ribbed vaults, permitted larger openings and taller structures that filtered gradually into elite urban residences (Frankl & Crossley, 2000), but because the sources describe structural technique rather than particular dwellings, this case cannot be coded for buffer space. Period III therefore yields one clear buffer (the St Gall cloister), two ambiguous cases and one indeterminate case. Non-market allocation is documented in institutional and feudal forms, and tenure stratification is not established.

4.4. Period IV, Industrial Standardisation (c. 1750–1945)

Rural-to-urban migration that exceeded the capacity of existing housing produced the period’s defining contrast between speculative, minimally regulated rental construction and reformist model housing. Engels’s contemporaneous account of Manchester remains the foundational primary source for speculative industrial housing, with its minimal light and ventilation, shared sanitation and extreme density (Engels, 1993), and Wohl’s analysis of London corroborates the pattern (Wohl, 1977). Back-to-back houses were often grouped around narrow courts reached by passages from the street, and in Berlin the speculative Mietskaserne was built around a sequence of rear courtyards. These courts are formally interposed between street and dwelling but were shared by many households and did not allow access to be controlled, so the case is coded ambiguous rather than absent.

Saltaire (Yorkshire, 1853–1870), built by the textile manufacturer Titus Salt, represents the reformist alternative: stone terraces with rear yards, integrated sanitation and civic institutions including schools, an institute and places of worship (Reynolds, 1983; Ashworth, 1951). Tenure was employer-provided and employer-controlled, offering improved habitability in exchange for oversight of the workforce. New Lanark (Scotland, from 1800) combined mill employment with schooling, cooperative provisioning and welfare institutions under Robert Owen (Donnachie & Hewitt, 2015). At Guise, André Godin’s Familistère (from 1859) arranged apartments on galleries around glazed central courtyards, under a cooperative form of governance described in Godin’s own account (Godin, 2000). Of the four Period IV cases, only the Familistère displays a clear buffer, and its evidence is the founder’s self-description. Tenure stratification is clearly present, with speculative rental and employer-provided or cooperative housing coexisting. Shared facilities appear on both sides: compelled by poverty in the tenements, planned in the model settlements.

4.5. Period V, Modernist and Financialised Housing (c. 1920–present)

The fifth period comprises modernist functionalism, post-war mass reconstruction and, from the late twentieth century, the financialisation and digital mediation of residential space. Villa Savoye (Poissy, 1928–1931) articulated the programme later realised at scale, with reinforced-concrete pilotis, the open plan, ribbon windows and roof garden enacting Le Corbusier’s dictum of the house as a machine for living in (Le Corbusier, 1986; Curtis, 1996). The Unité d’Habitation (Marseille, 1952) scaled that logic to collective housing, with double-height units, internal ‘streets’ and integrated commercial and civic facilities (Curtis, 1996); Jacobs offered an influential contemporaneous critique of superblock housing in general (Jacobs, 1961). Post-war mass housing, from Levittown to the French grands ensembles and East German Plattenbau, produced standardised and often prefabricated units at unprecedented scale, predominantly owner-occupied in the United States and predominantly state-allocated rental in Eastern Europe (Hayden, 2003; Glendinning, 2021).

The Bullitt Center (Seattle, 2013), engineered for net-zero energy, water self-sufficiency and biophilic design, is retained in the descriptive matrix as an articulation of the sustainability imperative but is excluded from coding because it is not residential. Contemporary co-living and micro-unit housing, proliferating in high-cost cities since the 2010s, shares facility-sharing with monastic and utopian precedents, but allocates space through market rental, frequently under institutional ownership. Ronald, Schijf and Donovan document this institutionalisation in Amsterdam, and White and Madden show how housing ideology legitimises the sector’s expansion. No Period V case displays a clear buffer space. The internal streets of the Unité and the shared amenity floors of co-living are intermediate spaces, but whether they act as domestic thresholds is contested, and both are coded ambiguous. Period V contains contrasting state-allocated and financialised rental forms, but the coded cases are separated by several decades and do not establish coexistence within one delimited housing market. The period is therefore not coded as evidence of tenure stratification.

4.6. Coded Results

Table 1. Coded results for the eighteen residential cases

Period	Case	Grade	Buffer space (P1)	Tenure sector (P2)	Allocation (P3)	Shared facilities
I	Çatalhöyük	A	Absent	Communal (interpretive)	Non-market	—
I	Ur/Uruk courtyard house	A	Present	Ownership	Ownership	No
I	Mohenjo-daro house	B	Present	Not documented	Not documented	No (collective drainage infrastructure)
II	Greek house (oikos)	B	Present	Not documented	Not documented	No
II	Roman domus (Pompeii)	A	Present	Elite ownership	Ownership	No
II	Roman insulae	B	Absent	Speculative rental	Market rental	—
III	Conwy Castle	A	Ambiguous	Elite (feudal)	Non-market	—
III	Plan of St Gall	C	Present	Institutional	Non-market	Yes
III	Medieval urban tenement	B	Ambiguous	Regulated (guild)	Market rental	No
III	Gothic elite residence	C	Indeterminate	Elite ownership	Ownership	—
IV	Speculative industrial tenement	A	Ambiguous	Speculative rental	Market rental	Yes
IV	Saltaire	B	Absent	Reformist (employer)	Non-market	Yes
IV	New Lanark	B	Absent	Reformist (employer)	Non-market	Yes
IV	Familistère, Guise	C	Present	Reformist (cooperative)	Non-market	Yes
V	Villa Savoye	B	Absent	Elite ownership	Ownership	No
V	Unité d’Habitation	B	Ambiguous	State-sponsored	Mixed	Yes
V	Post-war mass housing	B	Absent	State / ownership	Mixed	No
V	Co-living / micro-units	A	Ambiguous	Speculative rental (financialised)	Market rental	Yes

Table 1. Coded results for the eighteen residential cases. Evidence grades: A, primary evidence (excavation, standing fabric, contemporaneous documentation or original empirical research); B, specialist synthesis; C, general synthesis, idealised plan or promoter’s self-description. — indicates that the sources used do not allow the variable to be coded. Codes are the author’s judgements from the sources cited in Section 4.

Table 1. Coded results for the eighteen residential cases

Period	Case	Grade	Buffer space (P1)	Tenure sector (P2)	Allocation (P3)	Shared facilities
I	Çatalhöyük	A	Absent	Communal (interpretive)	Non-market	—
I	Ur/Uruk courtyard house	A	Present	Ownership	Ownership	No
I	Mohenjo-daro house	B	Present	Not documented	Not documented	No (collective drainage infrastructure)
II	Greek house (oikos)	B	Present	Not documented	Not documented	No
II	Roman domus (Pompeii)	A	Present	Elite ownership	Ownership	No
II	Roman insulae	B	Absent	Speculative rental	Market rental	—
III	Conwy Castle	A	Ambiguous	Elite (feudal)	Non-market	—
III	Plan of St Gall	C	Present	Institutional	Non-market	Yes
III	Medieval urban tenement	B	Ambiguous	Regulated (guild)	Market rental	No
III	Gothic elite residence	C	Indeterminate	Elite ownership	Ownership	—
IV	Speculative industrial tenement	A	Ambiguous	Speculative rental	Market rental	Yes
IV	Saltaire	B	Absent	Reformist (employer)	Non-market	Yes
IV	New Lanark	B	Absent	Reformist (employer)	Non-market	Yes
IV	Familistère, Guise	C	Present	Reformist (cooperative)	Non-market	Yes
V	Villa Savoye	B	Absent	Elite ownership	Ownership	No
V	Unité d’Habitation	B	Ambiguous	State-sponsored	Mixed	Yes
V	Post-war mass housing	B	Absent	State / ownership	Mixed	No
V	Co-living / micro-units	A	Ambiguous	Speculative rental (financialised)	Market rental	Yes

Table 1. Coded results for the eighteen residential cases. Evidence grades: A, primary evidence (excavation, standing fabric, contemporaneous documentation or original empirical research); B, specialist synthesis; C, general synthesis, idealised plan or promoter’s self-description. — indicates that the sources used do not allow the variable to be coded. Codes are the author’s judgements from the sources cited in Section 4.

The summary above reports the codes for the eighteen residential cases:

On Pattern 1, six cases display a clear buffer space: the Ur courtyard house, the Mohenjo-daro house, the Greek house, the Roman domus, the St Gall cloister and the Familistère. Six do not (Çatalhöyük, the insulae, Saltaire, New Lanark, Villa Savoye and post-war mass housing), four are ambiguous and two are indeterminate. Clear instances occur in four of the five periods, but in none of the cases dated after 1920. The distribution by tenure sector is more informative than the distribution by period. None of the three speculative-rental cases, the insulae, the industrial tenement and contemporary co-living, displays an unambiguous domestic buffer; however, the latter two contain intermediate communal spaces and are coded ambiguous rather than absent. The six clear instances belong to ownership, institutional or cooperative provision, or cases whose tenure is not documented. Non-speculative tenure does not guarantee a buffer, however: Çatalhöyük, Saltaire, New Lanark and Villa Savoye lack one.

On Pattern 2, tenure stratification is clearly present in Periods II and IV, where the coded cases include both speculative rental and secure or regulated provision. Period V contains a temporally separated contrast between state-sponsored and financialised forms, but it does not meet the stricter coexistence rule and is therefore not coded present. It is not established for Periods I and III: transactable property is documented at Ur and building regulation in the medieval town, but the sources used here do not document a contemporaneous speculative sector.

On Pattern 3, six cases are allocated by non-market means (Çatalhöyük, on interpretive grounds, Conwy, St Gall, Saltaire, New Lanark and the Familistère), two by mixed means, four by ownership, four by market rental, and two are not documented. Non-market allocation thus appears in Periods I, III and IV and, in mixed form, in Period V, but not among the Period II cases. Shared domestic or building-level residential facilities are coded for seven cases: St Gall, the industrial tenement, Saltaire, New Lanark, the Familistère, the Unité and co-living. Four allocate space by non-market means, two by market rental and one through mixed provision. Mohenjo-daro’s drainage network is recorded separately as collective infrastructure and excluded from this count. Facility-sharing therefore occurs on both sides of the market–non-market divide.

The evidence grades temper all three results. Six cases rest on grade A evidence, nine on grade B and three on grade C. Of the six clear instances of Pattern 1, two rest on grade A evidence (Ur and the domus), two on grade B (Mohenjo-daro and the Greek house) and two on grade C (St Gall and the Familistère). A third of the positive evidence for the most frequently claimed continuity in residential form thus comes from an unexecuted plan and a founder’s description of his own project. A sensitivity analysis excluding Grade C evidence reduces the number of clear buffer cases from six to four, but does not alter the key descriptive result: no speculative-rental case contains an unambiguous domestic buffer.

5. Discussion

5.1. Buffer Space Is Contingent, Not Constant

The results do not support the claim, common in synoptic accounts and advanced in earlier drafts of this framework, that a courtyard or buffer logic is a structural constant of residential design. Clear buffers appear in a third of the residential cases, are concentrated in the ancient Mediterranean and Near East, and disappear from the coded record after 1920. Two rival explanations deserve consideration. The climatic explanation favoured by Rapoport fits four of the six clear cases, all in warm, dry or Mediterranean climates, but not Çatalhöyük, which occupied a semi-arid plateau yet had no courtyard, nor the glazed courtyards of northern France. The tenure explanation fits the observation that no speculative-rental case has a clear buffer. Together they motivate a first, deliberately bounded proposition. P1: within this sample, no speculative-rental case contains an unambiguous domestic buffer, whereas non-speculative tenure neither guarantees nor excludes one. This is an association, not a sufficient-condition claim: two speculative cases contain intermediate communal spaces whose domestic access-control function is ambiguous. The proposition rests on only three speculative cases and should therefore be treated as a hypothesis for testing rather than a causal finding.

5.2. Tenure Stratification and Kemeny’s Dualism

The coexistence of secure and speculative housing is old, documented clearly in Roman and industrial cities and more weakly in the present. This does not show that Kemeny’s dualist regime predates the welfare state. Kemeny’s distinction concerns how rental provision is organised, specifically whether non-profit and profit rental compete in one market or are held apart, whereas the pre-modern evidence assembled here concerns stratification between owners and renters, or between reformist and speculative landlords. The two phenomena are related, but the first is a specific modern institutionalisation of the second. P2: tenure stratification recurs in the two historically delimited contexts in this sample where a speculative rental sector can be compared with a contemporaneous secure, regulated or reformist sector, but its institutional form varies, and Kemeny’s dualism is one modern form of it rather than its general model. A test would compare the organisation of rental provision in pre-modern cities with well-documented rental markets, such as Rome or Ostia, against Kemeny’s criteria. The same reasoning answers part of RQ2 for Aalbers: speculative tenements are an antecedent of financialisation, but not evidence that it is old. For Hillier and Hanson, Conwy shows that configurational depth can encode defence as readily as social hierarchy, and Pattern 1 suggests that the presence of a threshold may depend on tenure as much as on social relations.

5.3. Facility-Sharing Is Not Communal Allocation

Narrative accounts that place the monastery, the Familistère and contemporary co-living in a single communal lineage conflate two separable dimensions. On the sharing dimension the three belong together; on the allocation dimension co-living belongs with the speculative tenement, in which sanitation was also shared. P3: facility-sharing and non-market allocation vary independently, and claims of a communal revival in contemporary housing should specify which dimension is meant. This bears on policy debate. Co-living may reduce per-person space and cost, but its historical precedents, so far as these cases show, are the industrial tenement as much as the monastery or the cooperative. Community land trusts and housing cooperatives, which combine sharing with non-market allocation, are absent from the sample and would be the appropriate contemporary comparators for the communal cases. The earlier claim that communal allocation recurs cyclically under comparable economic pressure cannot be evaluated with these data, because the cases do not record the economic conditions under which each arrangement arose.

5.4. Geographical Limits and the Evidence That Would Test the Propositions

The propositions are claims about recurrence within a sample dominated by European and Mediterranean evidence, not claims about universality. Historical-geographical research on colonial planning shows why this matters. Yakubu’s study of Tamale, Ghana, documents how colonial decisions to redevelop indigenous housing clusters, classified as informal by colonial administrators, were inscribed so durably into the city’s fabric that they continue to structure housing politics nearly a century later (Yakubu, 2021). This is stratification of a different kind from those coded here: imposed across a colonial–indigenous boundary, with the informal sector defined into existence by the planning apparatus rather than emerging from market dynamics. Even within a single well-documented welfare state, housing outcomes are reproduced through neighbourhood-level path dependency at a finer grain than national regime typologies register (Nelson, Warnier, & Verma, 2024).

Four additions would turn the propositions into tests. First, P1 needs speculative-rental cases with courtyards. The Berlin Mietskaserne and the Glasgow back-court tenement are the obvious candidates, and should be coded as separate cases on whether their courts functioned as thresholds controlled by households; if they did, P1 fails. Second, the eighteen cases should be re-coded from computed justified access graphs rather than qualitative judgement, with a second coder to establish reliability. Third, the sample should be extended to traditions currently absent, including the Chinese siheyuan, sub-Saharan compound housing and Indigenous American residential forms, chosen without reference to whether they display the patterns. Fourth, P3 requires contemporary non-market cases, such as community land trusts and housing cooperatives, coded on the same variables as co-living.

6. Conclusions

This article set out to ask how continuity in residential form could be recognised and what would count against it. Coding eighteen residential cases from Çatalhöyük to contemporary co-living, with absences reported and evidence graded, yields a more modest picture than synoptic histories of housing usually offer. The buffer space often cited as a constant of residential design appears in six of eighteen residential cases, rests in part on weak evidence, and is unambiguous in none of the speculative-rental cases. The coexistence of secure and speculative housing is old, but it is not Kemeny’s dualism. Shared facilities appear under both market and non-market allocation, so contemporary co-living revives communal sharing without communal allocation.

The broader argument concerns method rather than any single pattern. Long-range comparison is valuable precisely because it can show that conditions presented as unprecedented have precedents, but that value depends on distinguishing recurrence from persistence, reporting the cases in which a pattern is absent, and stating claims in a form that further evidence could overturn. The coding scheme and propositions offered here are intended to be extended and tested by others, above all with cases from the regions and tenure forms that the present sample does not reach.