This is not just a book; it is the foundational text for modern quantitative genetics and the birthplace of two of the most important concepts in statistics: regression and correlation. In Natural Inheritance (1889), Francis Galton moves beyond the qualitative descriptions of heredity common in his day to provide a rigorous mathematical framework for how traits are passed across generations.
TL;DR
Galton introduces the concept of "Regression to the Mean," proving that children of exceptional parents tend to be more mediocre than their progenitors. By applying the "Law of Error" (the Normal Distribution) to human heights and eye colors, he establishes the mathematical basis for predicting the characteristics of a population over time.
What Problem Does This Paper Address?
In the late 19th century, the mechanism of inheritance was a "black box." While Charles Darwin (Galton’s cousin) had proposed the theory of Pangenesis, it lacked quantitative predictive power. Scientists observed a "curious regularity" in populations: despite the vast differences between individuals, the overall distribution of traits like height remained remarkably stable from one generation to the next.
Galton sought to solve the puzzle of population stability: If tall parents have tall children, why doesn't the human race eventually split into giants and dwarves? He aimed to quantify the "average share" contributed by each ancestor and define the "nearness of kinship" in the precise language of numbers.
Methods
Galton’s approach was revolutionary for its time, blending mechanical modeling with advanced (for the era) statistics:
- The "Mid-Parent" Concept: To handle the complexity of two parents, Galton created an artificial "Mid-Parent" metric. He "transmuted" female statures into male equivalents by multiplying them by 1.08, allowing him to treat a couple as a single data point.
- The Quincunx (Galton Board): He used a mechanical device—a board with pins where falling shot forms a bell curve—to illustrate how a multitude of small, independent "accidents" (genetic and environmental influences) inevitably result in a Normal Distribution.
- Geometric Modeling: Working with mathematician J. D. Hamilton Dickson, Galton analyzed "Surfaces of Frequency." By plotting the heights of parents against children, he discovered that the resulting data formed concentric ellipses, leading to the mathematical definition of the regression line.
- Particulate Inheritance: He argued that inheritance is "particulate" rather than purely "blended," using the analogy of a building constructed from the salvaged bricks of older structures.
Datasets
Galton was a pioneer in data collection, utilizing several distinct sources:
- Records of Family Faculties (R.F.F.): A unique dataset of 150 families obtained by offering prizes for the most complete genealogical and physical records.
- Special Observations: Stature data for 783 brothers from 295 families.
- Anthropometric Laboratory Data: Measurements of nearly 10,000 individuals taken at the 1884 International Health Exhibition.
- Botanical Experiments: Longitudinal data on the weights and diameters of sweet pea seeds, which provided his first clear evidence of regression.
Key Insights and Ideas
- Filial Regression: This is the "aha moment" of the book. Galton found that the deviation of a son from the population average is, on average, only two-thirds that of his "Mid-Parent." The population stays stable because the extremes are constantly "regressing" toward the mean.
- The Law of Ancestral Heredity: Galton estimated the quantitative contribution of ancestors. He proposed that the two parents contribute roughly 1/2 of the total heritage, the four grandparents 1/4, and so on in a geometric series.
- Alternative vs. Blended Inheritance: He noted that while height seems to "blend" (children are usually intermediate), traits like eye color are "alternative." If one parent has light eyes and the other dark, the child usually has one or the other, not a "muddy" mix. This was a significant precursor to Mendelian genetics.
- Negative Evidence for Acquired Traits: Galton argued strongly against the inheritance of "acquired faculties" (Lamarckism), using the example of a mother becoming a drunkard—her children might be "neurotic," but he attributed this to the prenatal environment (alcohol in the tissues) rather than true genetic inheritance.
Key Results
- The Regression Coefficient: For human stature, Galton calculated the ratio of "Filial Regression" to be 2/3. This means if a Mid-Parent is 3 inches taller than the average, their children will, on average, be only 2 inches taller than the average.
- Predictive Accuracy: Using his regression formulas, Galton demonstrated that he could predict the distribution of eye colors in a group of families with surprising accuracy (e.g., predicting 614 light-eyed children when 629 were observed).
- Universal Kinship Rule: He established that the "Parental kinship is only half as close as the Fraternal." In modern terms, he was quantifying the correlation between relatives.
Limitations and Future Work
Galton was transparent about the "penumbra of uncertainty" in his work. He acknowledged that his data on "Artistic Faculty" and "Disease" (specifically consumption/tuberculosis) were "lax" and subjective compared to his height measurements.
He also noted that his "Mid-Parent" model was a simplification that ignored the potential for one parent to be "prepotent" (genetically dominant) over the other. He concluded by calling for more rigorous, longitudinal data collection by insurance companies and medical professionals to refine these statistical laws, a call that eventually led to the development of the field of eugenics and modern biostatistics.
Summary updated