Document Type



The method of assessing mating and inbreeding patterns by use of the frequency of isonymous marriages in sets of historical data on vital events rests on several key assumptions. Two of these are monophyletism of surname origin and an absence of immigration. Another, however, is that names can correctly be assessed as to their identity and nonidentity. Various types of problems with historical data, which amount to mutations in names, can lead to difficulties in this regard. Spelling variations in names may or may not indicate differences of name origins. Aggregating names into equivalence classes based on spelling or phonetics may produce inflated isonymy by associating names which are truly of different historical origins; on the other hand, separating such names may deflate isonymy falsely by treating as different names which are actually the same, historically. These problems are not trivial in dealing with the kinds of data on which isonymy studies are usually based. Under some simplifying assumptions it is shown that mutations in surnames produce biases in estimates of the random as well as nonrandom component of inbreeding which may be rather large unless mutations in names are infrequent. Studies of isonymy should be much more sensitive to these problems if their estimates are to be taken as accurate reflections of mating structure.