The Uninvestigated Variable: Socially Stratified Lead Exposure as a Candidate Factor in Elite Rigidity, 1500-1900

Why do ruling elites lose the ability to adapt? The question has been posed by Pareto, Toynbee, Olson, Marx, Lenin, North, and Mokyr, among others. Each described the symptoms — ossification, institutional sclerosis, failure to respond — but none identified a material variable that satisfies four criteria simultaneously: generational scaling, class-stratified distribution, built-in time lag, and a self-reinforcing feedback loop. This paper proposes chronic lead exposure through status goods (lead white paint, cosmetics, glazed tableware, adulterated wine) as a candidate variable and presents preliminary evidence consistent with the hypothesis. The study focuses on Western civilization and Japan — societies where lead white was produced and used at industrial scale — because the specific channel of socially stratified domestic exposure is a feature of these civilizations. Before the twentieth century, the primary channels of non-occupational lead exposure were tied to elite consumption: the higher the social stratum, the higher the cumulative dose. Lead damages the prefrontal cortex, producing not a reduction in general intelligence but a specific behavioral shift: perseveration, shortened planning horizons, reduced capacity for compromise. A preliminary analysis of 25 political crises and 100 historical actors (33 parameters each) finds patterns consistent with the hypothesis: rigidity predicts defeat, estimated lead exposure profiles correlate with rigidity, and the character of the resulting crisis varies with the distribution of the toxin across social strata. These findings are based on indirect evidence and retrospective assessment by a single researcher; they do not constitute proof. Skeletal lead analysis — direct measurement of cumulative exposure in archaeological bone — is the primary method by which this hypothesis can be verified or falsified, and the central purpose of this paper is to argue that such analysis should be undertaken.This version adds the full case corpus (Appendix A), a human-readable rendering of the dataset (Appendix B), a codebook, and machine-readable data files.