The isonymy structure of the 48 states of the continental United States of
America was studied using the surname distributions of 18 million telephone
users, distributed in 247 towns. The shortest linear distance between near
est neighbor towns included in the sample was 12.0 km. The largest distance
was 4,577 km. The number of different surnames found in the whole analysis
was 899,585. Lasker's distance was found to be significantly but weakly co
rrelated with the geographic distance, with r = 0.21 +/- 0.01. A dendrogram
of the 48 states was built from the matrix of isonymy distances: it divide
s the US into several clusters, in general correlated with geography. A not
able exception is California and New Jersey, which cluster together. Wiscon
sin is separated from all other states. An important cluster is formed by T
exas, Colorado, New Mexico, Nevada, and Arizona, together with Illinois and
Florida. It was observed that Hispanic surnames are among the most frequen
t in Illinois, as they are in New Jersey and California. No main distinctio
n among the states clearly attributable to surnames of French origin was de
tected; however, New Hampshire, Vermont, and Maine which have a considerabl
e number of these surnames belong to the same northeastern cluster.
From the present analysis, the great mobility of the US population emerges
clearly, and it seems relevant that the practical absence of isolation by d
istance is seen also considering only small towns. It appears that groups o
f different origin are well-mixed over the whole area of the United States.
The values of isonymy indicate that the south-central area of the USA has
the highest level of inbreeding. In fact, the heterogeneity in surname comp
osition is greater in the coastal areas, particularly on the East Coast, th
an anywhere else in the USA. (C) 2001 Wiley-Liss, Inc.