Lewontin (1972)

TL;DR

This paper critically examines Lewontin's 1972 analysis of human genetic diversity, revealing calculation errors and clarifying the mathematical and theoretical foundations, thereby providing a detailed reinterpretation of his influential work.

Contribution

It offers a detailed reanalysis of Lewontin's original calculations, exposing errors and clarifying the underlying mathematics and population genetics theory.

Findings

Lewontin made calculation errors in his analysis.

Reproduced calculations show slight shifts in diversity apportionments.

Clarifies the mathematical and theoretical basis of Lewontin's work.

Abstract

Richard C. Lewontin is arguably the most influential evolutionary biologist of the second half of the 20th century. In this chapter, I provide two windows on his influential 1972 article "The Apportionment of Human Diversity": First, I show how the fourteen publications that he cites influenced him and framed his exploration; second, I present close readings of the five sections of the article: "Introduction," "The Genes," "The Samples," "The Measure of Diversity," and "The Results." I hope to illuminate the article's basic anatomy and argumentative arc, and why it became such a historically important document. In particular, I make explicit all of the mathematics (e.g., six Shannon information measures) and the general population genetic theory underlying this mathematics (e.g., the Wahlund effect). Lewontin did not make this explicit in his article. Furthermore, in redoing all of his calculations, I find that Lewontin made calculation errors (including rounding errors or omitting diversity component values) for all the genes he analyzed except one (P), and understated the among races diversity component, according to even just his own calculations. In reproducing the original computation, I find that the values of, respectively, within populations, among populations but within races, and among races diversity apportionments shift slightly (86%, 7%, 7%); here, in this "field guide" to Lewontin (1972), as well as in Winther (2022), I discuss this change in light of the values produced in subsequent replications of Lewontin's calculation with other statistics and data sets.