Hypermedia systems like the Web heavily depend on their
ability to link resources. One of the key features of the
Web's URIs is their ability to not only specify a resource,
but to also identify a subresource within that resource, by
using a fragment identifier. Fragment identification enables
user to create better hypermedia. We present a proposal
for fragment identifiers for plain text files, which makes it
possible to identify character or line ranges, or subresources
identified by regular expressions. Using these fragment identifiers,
it is possible to create more specific hyperlinks, by
not only linking to a complete plain text resource, but only
the relevant part of it. Along with this proposal, a prototype
implementation is described which can be used both as
a server-side testbed and as a client-side extension for the
Firefox browser.