I was forwarded this conversation earlier today and decided to chime in.
I'm the developer of the University of Virginia Art Museum Numismatic
Collections ( mentioned previously my Mark
Custer.  As a web developer for the UVA library with an avid interest in the
encoding of objects generally contained in museums, I decided to adapt EAD
(a common library tool), to the art museum's collection of Greek and Roman
coins.  Since the coin website, built on cocoon and solr, can be seen as a
successful demonstration of EAD's versatility in describing objects outside
of the typical library/archive realm, the art museum is interested in
exploring EAD further to describe their other collections.  One of the
pieces that is currently missing is an administrative backend that allows
for easy editing of the data without having detailed knowledge of EAD, so I
plan on developing a web-based XForms extension to edit and build the

The site has gotten some publicity, but mainly within the UVA community,
although there are now two papers about it.  One was published in the
proceedings of IS&T's Archiving conference last summer.  Another paper is
due to be published shortly as part of the proceedings for the Computer
Applications and Quantitative Methods in Archaeology (CAA) conference that
took place in March.  This paper, which I can forward to anyone that is
interested, describes my use of EAD for the coin collection and its broader
potential for describing museum objects, including the integration of Solr
as a powerful tool for faceted search to make a very robust and useful
site.  I also reviewed VRA Core, CDWA, and a database scheme developed by
the American Numismatic Society, and why I chose EAD over the other three.
Another member of CAA at the University of Granada, Francisco Melero,
presented on the use of EAD to document cultural heritage monuments.

My experience is that a metadata format should be chosen by what it allows
you to describe and your requirements for a user interface.  I favor EAD's
ability to describe organization (whether conceptual or literal) of
materials and also EAD's ability to allow for complex contextual
descriptions to be attached to collections or individual items.

Ethan Gruber
University of Virginia Library