Z39.50 is notorious for being difficult to implement, and Merrilee is right
that SRW is growing in popularity as a simpler implementation of Z39.50,
which happens to be in XML.
You might be interested in providing OAI-PMH access to your texts instead of
or in addition to SRW or Z39.50. If you did so, the library could direct
its users to one or more OAI-PMH harvesters, which would locate the WWP
texts. For example, metadata for most of the encoded text collections at
Michigan are searchable through our local OAI harvester, OAIster (
http://oaister.umdl.umich.edu/ ), which in turn is fed into Yahoo!.
For more information on OAI-PMH, see http://www.openarchives.org/ .
It is quite important for textbases to make at least their metadata if not
their full text accessible in some other way than through the textbases's
own interface. In our age, where almost all information seeking starts --
and frequently ends -- with a web search engine, they'll never find it
unless the data or metadata "trickles up" to a federated search system or,
better yet, also further trickles up to a general-purpose web search engine.