1983 Volume 91 Issue 3 Pages 309-322
An attempt was made to estimate the migration pattern in recent Japan using surnames as migration markers. As an analogy of genes, a mutated surname is transmitted generation to generation to produce its offsprings, and incidentally these decendants migrate from the original population to other ones. From the geographical distribution of surnames at the present time, the migration pattern of Japan in historical age can be estimated. A small city in the Hokuriku District (Takefu) was selected and it was compared with other 53 cities in Japan with respect to the surname distribution. Data were obtained from telephone books and 409 surnames were detected from the A-series of the telephone book of Takefu city. Numbers of these surnames at other 53 cities were counted and analyzed by using four kinds of statistics.
A positive correlation is found between the total size of a surname and the number of cities where this surname is distributed. A negative correlation is found between the total size of a surname and the coefficient of variation of the surname size for cities. The affinity of Takefu city with other cities in relation to the surname distribution was computed under four indices, including KENDALL'S rank correlation coefficient and PENROSE's shape distance. In general, we can see the negative correlation between the affinity and the geographical distance from Takefu city.
To obtain the valid estimate of the migration pattern from the surname distribution, it is important to assume the monophyletic origin of surnames. If we exclude surnames with extremely large total size from migration markers, this assumption might hold better. The more accurate relationship between the affinity of the surname distribution and the migration pattern remains to be studied.