Hi, The discussion about "coordinate variables" reminds us here in Massachusetts of a recent scandal over questionable expenditures by our state treasurer. He assured our citizens that it was okay for him and his cornies to continue scamming the public so blatantly, because they had always done so. Our theory on "coordinate variables" relies heavily on the universally trusted "if-it-walks-like-a-duck" rationale. We use multi-dimensional "coordinate variables" -- admittedly in technical violation of the NetCDF rules -- because they are so intuitive and useful. It is no trouble to test for a variable whose dimension names also happen to be the names of variables (and vice-versa) -- that's what we have always assumed by the term "coordinate variable". The "ncvar/iscoord" method in our "NetCDF Toolbox for Matlab <http://crusty.er.usgs.gov/~cdenham/> invokes that idea. Rather than adopt a scheme involving extra attributes that may have to be parsed, we would prefer that the official definition of "coordinate variable" be expanded to incorporate multiple "coordinate dimensions". Thanks, Chuck Dr. Charles R. Denham e-mail: email@example.com, firstname.lastname@example.org U.S. Geological Survey WWW: http://crusty.er.usgs.gov/~cdenham/ 384 Woods Hole Road voice: (508) 457-2318 Woods Hole, MA 02543-1598 fax: (508) 457-2310 "I again sought solace in gazing at that Morphean orb across whose surface visions of other, and perhaps better, worlds may pass." Mark O'Donnell, "The Teltale Tube."