Korean Family Name Distribution in the Past

Hoang Ahn Tuan Kiet, Seung Ki Baek, Hawoong Jeong, Beom Jun Kim

Abstract: We empirically study the genealogical trees of ten families for about five centuries in Korea. Although each family tree contains only the paternal part, the family names of women married to the family have been recorded, which allows us to estimate roughly the family name distributions for the past five hundred years. Revealed is the fact that the unique Korean family name distribution, characterized by a logarithmic form of the cumulative distribution and an exponentially decaying rank-size plot often called the Zipf plot, has remained unchanged for a long time. We discuss the implications of our results within a recently suggested theoretical model and compare them with observations in other countries in which power-law forms are abundantly found.

Introduction

In the field of statistical physics, the distribution of social phenomena—such as city sizes, individual wealth, or word frequencies—often follows a "power-law" distribution. In such systems, a few items are extremely common while the vast majority are rare, following a specific mathematical curve. However, previous studies have identified Korea as a significant outlier. While countries like Japan possess over 130,000 family names, Korea has only about 250. Furthermore, the "Zipf plot" (ranking names from most to least common) for Korean names exhibits an exponential decay rather than a power-law. This study investigates whether this unique distribution is a modern artifact or a fundamental, long-term characteristic of Korean society.

Methodology

To investigate the historical distribution of Korean family names, researchers analyzed Jokbo—traditional Korean genealogical records. Although these books primarily track paternal lineages, they meticulously record the family names of the women married into the family. By extracting the names and birth years of 221,609 wives from ten digitized family books, the researchers reconstructed a "snapshot" of the broader Korean family name distribution dating back to the 15th century.

To ensure data consistency, the Kendall tau rank correlation coefficient was employed to verify if the relative ranking of names remained stable across different datasets and time periods. The analysis spanned records from the year 1254 to the late 20th century.

Historical Stability of the Distribution

The most striking finding of this research is the remarkable stability of the Korean family name distribution over five centuries. Whether examining data from the 1500s or the 1990s, the mathematical signature of the distribution remains unchanged.

Rank-size plots of Korean family names across different datasets (a) and time periods (b), showing consistent exponential decay and stability over five centuries.

As shown in Figure 1, the rank-size plots consistently follow an exponential decay. This stability suggests that the cultural practice of strictly inheriting family names without "inventing" new ones has deep historical roots, leading to a "winner-take-all" scenario for established names like Kim, Lee, and Park.

Mathematical Characteristics

The study confirms that the Korean name distribution is fundamentally different from the power-law distributions found in Europe or Japan. A key metric is the relationship between the number of family names ($N_f$) and the total population size ($N$).

Logarithmic growth of the number of family names ($N_f$) relative to population size ($N$) across different datasets and time periods.

In power-law systems, $N_f$ typically increases algebraically with $N$. In contrast, the Korean data (Figure 2) demonstrates that $N_f$ increases only logarithmically with population size. This slow growth in the variety of surnames reinforces the conservative nature of the Korean naming system.

Discussion and Conclusion

The research also touched upon the role of regional identity. When regional origins (Bon-gwan) were included (e.g., "Kim from Gimhae"), the distribution shifted away from exponential decay toward a power-law pattern. This suggests that regional identity historically functioned as a "sub-surname" to provide social distinction that a common name like Kim could not.

While the study relies on paternal genealogical trees which may have a bias toward literate and wealthy families, the high correlation between datasets suggests the findings are representative of the general population. This work highlights how long-term cultural practices can create unique statistical signatures that persist for centuries, offering a fascinating counterpoint to more common social distribution patterns.

Summary updated

Abstract

We empirically study the genealogical trees of ten families for about five centuries in Korea. Although each family tree contains only the paternal part, the family names of women married to the family have been recorded, which allows us to estimate roughly the family name distributions for the past five hundred years. Revealed is the fact that the unique Korean family name distribution, characterized by a logarithmic form of the cumulative distribution and an exponentially decaying rank-size plot often called the Zipf plot, has remained unchanged for a long time. We discuss the implications of our results within a recently suggested theoretical model and compare them with observations in other countries in which power-law forms are abundantly found.
BibTeX
@article{kiet2007korean,
  author = {Hoang Ahn Tuan Kiet and Seung Ki Baek and Hawoong Jeong and Beom Jun Kim},
  title = {{Korean Family Name Distribution in the Past}},
  journal = {Journal of the Korean Physical Society},
  year = {2007},
  doi = {10.3938/jkps.51.1812},
}
Figure