Abstract
Contemporary housing debates often present affordability pressures, densification and shared living as unprecedented. Long-range histories of housing tend to answer that framing by asserting continuity, but rarely state how continuity would be recognised or what would count against it. This article proposes and applies an auditable procedure for that task. Eighteen residential cases, from Çatalhöyük to contemporary financialised co-living, were coded for three candidate patterns drawn from architectural history, space syntax and housing-regime theory: a buffer space mediating between the public realm and the dwelling; the coexistence of a speculative rental sector with a secure or regulated one; and non-market allocation of dwelling space. Each case was also graded for the quality of the evidence on which its coding rests. The analysis distinguishes recurrence, the appearance of a feature in unconnected cases, from persistence in the Conzenian sense of inheritance within a place-specific morphological lineage; the evidence assembled here tests only the former. Of the eighteen residential cases, six display a clear buffer space, six do not, and four are ambiguous and two are indeterminate. No clear instance occurs after 1920, and none of the three speculative-rental cases displays an unambiguous one. Coexisting speculative and secure sectors are documented in two of five periods and, more weakly, in a third. Shared domestic facilities occur under both market and non-market allocation, so their presence does not in itself indicate a communal housing logic. These results do not support claims of a structural constant in residential form. They support three narrower propositions: that no speculative-rental case in this sample contains an unambiguous domestic buffer, while other tenures do not guarantee one; that tenure stratification is old but is not equivalent to Kemeny’s dualist rental regime; and that facility-sharing and non-market allocation are separate dimensions that claims of communal revival tend to conflate. The article closes by specifying the evidence that would test these propositions, including computed configurational measures and cases from regions absent from the present sample.