MB> Probably the most straightforward way is to use <index> (P4
MB> 6.8.2, same reference and so far unchanged in P5), which isn't
MB> confined to creating indices in the traditional printed book
MB> sense. You can arrange the attribute value in whatever way suits
MB> your sorting needs. <head><index type="articles" level1="operas
MB> of Puccini, the"/>The operas of Puccini</head>
I think if the reason for desiring article-free sorting is to get a
properly sorted index, this is the right approach. Note, however,
that <index> will change in P5, not sure how yet. (I am quite
confident that transforming a P4 <index> into a P5 <index> will be
automatable via XSLT.)
PT> We had the same problem in coming trying to encode bibliographic
PT> entries. We settled on <title><seg type="nonSort">The</seg> Great
While this is obviously workable, I'm still not 100% comfortable with
it. One could very well argue, I think, that <title> is in some sense
a special case of <name>, and therefore deserves a reg=.
Another possible solution is to not bother encoding a "sort me"
version of the <head> at all, and use sorting software that's smart
enough to ignore "The".