A Mean and Random Past: The Implications of Variance for History

Abstract

Historians seeking to identify the Irish have overwhelmingly relied upon nominal record linkage, thus limiting studies to periods and contexts in which corroborating records exist. Surname analysis provides an alternative: a subset of 283 Irish surnames was able to correctly isolate 40 percent of known Irish individuals across thousands of entries, which is sufficient for sampling the Irish in demographic studies. This conclusion was based on an analysis of 278,949 names from the London area in the 1841 census, and was tested and refined against 42,248 historical records pertaining to the poor in London between 1777 and 1820.