Small errors in estimated household size massively distort historical population estimates.

Evidence

In order to move from counts of hearths on to estimates of a global population, it is obviously necessary to multiply hearths by a factor representing the average family size—which is precisely what we do not know. With dedicated rigor specialists debate the exact size of the required conversion factor. It is quite clear that a quite tiny error in the size of the factor chosen may have an enormous impact on the size of the eventual population estimate. — L. P. Harvey, Muslims in Spain, 1500 to 1614, 88

Categorization

• XREF: Connects to general methodological debates in historical demography, and to Bayesian/error-propagation thinking about confidence intervals in estimates.
• SEED: Could anchor an essay on how historical population figures everyone cites are built on unverifiable multiplier guesses.