I agree with most of what Stephen says, but it seems to me the goal should be to replace textual added entries with links to identifiers.  So it would not be possible to just "assign a name heading."  That means you have to choose one or the other as the supposed author, accepting that sometimes you'll be wrong (not much different from the present situation).



John Hostage                              Authorities Librarian

Langdell Hall                                 [log in to unmask]

Harvard Law School Library    +(1)(617) 495-3974 (voice)

Cambridge, MA 02138              +(1)(617) 496-4409 (fax)


picture. What to do? My answer would be that if differentiation is distinct from heading form, it becomes possible to assign the name heading to the new book without necessarily ascribing its authorship to either of the two established authorities. This to me would be truer to the state of knowledge about the new book and its predecessors than would creating a third, undifferentiated personal name authority just to accommodate the as yet unascribed title. Perhaps the point of this little digression is that the real problem in the case of very similar authors is one of ascribing authorship rather than of differentiating individuals. We're better off dealing with such cases with a mechanism like OCLC's controlled/uncontrolled headings rather than with undifferentiated personal name authorities.