At the University at Buffalo we bind each manifestation of the title
separately (except in the rare occasions where we didn't identify a
title change until later).  We have classified our periodicals, so
it probably isn't absolutely crucial to match the bound volumes with
the bibliographic description, but it seems sloppy (lazy, even?) not to.
We don't very often go back and rebind if we missed a title change, unless
there is some pressing reason for it (call # changes because of one title
splitting into two, for example).  Have I contradicted myself?  We follow
the bibliographic description as best we can, but don't beat ourselves
over the head if we missed something for a volume or two.  Because we
are classified we can get away with this.

Susan Davis/Head, Periodicals
State University of NY at Buffalo