The Osteological Paradox (Wood, Milner, Harpending & Weiss 1992) holds that skeletal statistics cannot straightforwardly reveal past population health

"The Osteological Paradox: Problems of Inferring Prehistoric Health from Skeletal Samples" (James W. Wood, George R. Milner, Henry C. Harpending & Kenneth M. Weiss, Current Anthropology 33:4, 1992, pp. 343–370) is a foundational methods critique in bioarchaeology. Its argument is that a cemetery's skeletons are not a fair sample of the living population they came from, so naive tallies of lesions, stature, or age-at-death cannot be read straight off as measures of how healthy past people were.

Three confounds do the work. Demographic non-stationarity: a growing or shrinking population changes the age structure of its dead independent of health. Selective mortality: a skeletal sample contains only those who died at a given age, not those who survived it, so the observed can systematically mis-state the risks the living faced. Unmeasured, individual-level frailty heterogeneity: individuals differ in hidden susceptibility, so — paradoxically — a skeleton showing more healed lesions may signal a person healthy enough to survive insults that killed frailer peers, making "more pathology" ambiguous between a sicker and a heartier population. The paper frames the problem as one of "demographic non-stationarity, selective mortality, and unmeasured, individual-level heterogeneity in the risks of disease and death."

This is the archaeological-method counterpart to the demographic source-criticism running through the vault's paleodemography thread: skeletal age-at-death estimation carries the same reference-sample bias ("age mimicry") that makes ancient population reconstruction from bones untrustworthy, just as Chandler's urban census is load-bearing yet imperfectly grounded. One of the paper's four authors, Henry C. Harpending, is the same figure who later became a documented race-science figure — the co-author whose earlier, uncontested infrastructure work sits oddly beside his later, distrusted interpretive turn.