Splitting on more than one letter

Hi,

   I'm having difficulty determining a proper method for handling an
extremely large document. Basically, the document consists of <topic>
elements, and within that <topic> elements there is one <title>
element.

   The goal is to convert this document to HTML and since there are so
many <topic> elements (4795), they all cannot fit on one page - which
is fine in that I can split it up on the first letter of each <title>.

   Unfortunatly, however, this method *again* leaves too many elements
on a particular page (within a <topic> element there may be many other
sub-elements which generate text). So I'm trying to determine the best
way to deal with it.

Possibilities considered so far:

      1) Show only the <titles> of each <topic> on a specific page
         that begin with each letter (A, B, C, etc) and then each
         <topic> generates its own page.

         This would certainly work but there would be many files
         generated.

      2) Split the document into individual pages based on the first
         letter AND POSSIBLY A SECOND LETTER, which is the topic of
         this message.

Using the Muenchian method, its fairly easy to split up the document
into pages of elements whose <title> begins with a certain
letter. But, lets say that there are 2201 elements that begin with A
(this includes the sub-elements, not just <topic>'s). How could one
automatically determine the borders so that there are no more than,
say, 500 elements on a page?

   For example, instead of having "A.html" (which comprises the topics
whose <title> begins with "a"), there may need to be "Aa-Ae.html",
"Af-Ak.html", "Al-Ar.html", and "As-Ay.html" (there are no elements
that have a title beginning with "az").

   Is this type of thing do-able in XSL or should I investigate
handing nodes off to Java or something similar?

   Sincerely and thank you for any help,
   Elizabeth

 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list