We at the New York Botanical Garden have been going through
          a very arduous process of creating a collections management
          and research database for our 6 million specimen herbarium.
          It is the largest collection of this type in the Western
          Hemisphere, and the most active in the world, in terms of
          loans and visitor use. As you can well imagine, a collection
          this size is a real bear to manage.
 
          As a first step, we are extensively modifying a program
          created for the Harvard University Herbarium, called HUHpc.
          Our modification is being done by the developer who created
          HUHpc, with massive staff involvement. This is a Revelation
          application, and any of you who have worked with databases
          probably have real strong feelings about Revelation, either
          pro or con (mostly the latter it seems.) This process of
          modification has taken about 2 years, and we are just now
          getting NYpc (as it is imaginatively called) on to a test
          data load.
 
          The second step is to specify and develop a client-server
          database program using one of the 4th generation languages
          such as Oracle or Ingres. This effort has taken about two
          years to date, and we are just about ready with full specs.
          We are doing this as a collaborative project with another
          major botanical collection. I'm not sure what the current
          projection is for getting this up and running, but very
          detailed implementation plans are being drawn up, with
          associated budgets. When it is ready, it will run on a UNIX
          platform, with PC clients. Of course, all records created in
          the interim database will be convertible to this new
          database.
 
          In 1992, a consortium of freestanding natural science
          collections (ie non-university museum collections) did a
          massive survey and analysis of computing needs. This
          project, called the MITRE project, after the consulting firm
          who prepared the report, specifies in a pretty
          straightforward fashion the model architecture for
          collections management system. W. Wayt Thomas, an NYBG
          botanist, was the PI for this NSF-funded project, and we are
          following the MITRE recommendations. An NSF reviewer called
          this report "no less useful for being blindingly obvious..."
 
          Getting to this stage has required massive institutional
          commitment of resources, including a full time manager of
          science computing, and extensive participation of herbarium
          and research staff. And the big bucks for hardware
          and...gulp... data entry are yet to come.
 
          Though this collection may be an extreme case, and I have
          been watching the process more or less from the sidelines
          (as a fundraiser), I have been amazed at the complexity of
          this databasing project. It requires incredible
          institutional stamina and persistence to get a long-term
          project of this scope underway.
 
          For perspective sake, we keep in mind that the project to
          automate the NYBG library catalog, with several hundred
          thousand items, took from 1967 to 1994. It just went on
          line, by the way, and you can telnet to it at
          librisc.nybg.org or 192.77.202.200.
 
          Now that I've thoroughly scared anyone contemplating a
          collections databasing effort...
 
          Eric Siegel
          [log in to unmask]