Yes, I have a book by one Peter Rosza. It took me some time to realise that Peter was in fact a woman, and a well-known mathematician at that, whom we might have called Rose Peter. Tsk! These Magyars...! You'd think they would have come to us for advice. As for the Icelandic convention, I knew that there was something funny about all their terminal "-sons" and "-dotters" (sp?) but don't they have any family name at all? Some of the Slavic names might be troublesome too, because they vary the suffix of what I take to be the family name, according to gender: -ski vs -ska and so on. But maybe I have that mixed up as in the Icelandic names. Could it be that the Icelandic convention derives from the fact that they are dealing with a smallish population? Anyway, It seems to me that the indexing convention I proposed would still be easy to apply by anyone that understands the naming convention of the language and the population in question. Simply write the complete name (or whatever part suits the DB in question) in the lexically normal way according to the favoured convention, then rotate it till the first letter after the last non-alphabetic character is first in the string, and voila! Go well, Jon Andrew Sly wrote:
And don't forget that other national traditions you can have more confusion.
For example:
For Hungarian names, the preferred order is [Family name] [Given name] So that in the main text of PG#19433 the author's name is given as: Balazs Bela, with the understanding that the first name that appears is the one we alphbetize by.
And in Icelandic names, what looks to us as a "last name" is not actually a family name, but a patrynomic. It is incorrect to alphabetize by that, so the given name is used instead.
--Andrew