> <l> ... <corr sic="I lov" id="corr1" next="corr2"></corr> </l>
> <l> <corr id="corr2" prev="corr1">I love</corr> ... </l>
> If there's a less roundabout way to achieve the same, I would be grateful
> to know.
I actually think that's a very good solution. There are of course quite
a few variations (aggregate the two CORR elements with JOIN or LINK
instead of next= and prev=; use SIC and CORR or two SICs instead;
fragment the L elements around a single CORR, but that loses some
of the information), but they mostly amount to the same thing.
> ... the conclusion would be that
> <corr sic="XYZ"></corr> [with no content between the tags]
If you've documented CORR to mean a scribal, as opposed to editorial,
correction, or indicate this using attributes, then the above is
essentially true. However, DEL carries hand=, status=, and type=
attributes that CORR does not have; CORR allows a whole bunch of the
linking and alignment attributes that DEL does not have.
On further thought, though, I don't wonder if aggregating the two
CORR elements via pointing with next= and prev= is inappropriate.
After all, although they certainly are semantically intertwined, they
are more readily conceived of as separate textual content objects.
(At least, I find it easier -- does that mean I've been doing this
OHCO stuff for too long?) Perhaps using corresp= on each to point to
the other would be more appropriate.
 The resp attribute is mis-declared on CORR as CDATA; it should
be IDREF. (It is declared as CDATA in lots of other elements,
too, and I think it always should be IDREF (or perhaps IDREFS?),
but I'm not sure -- I haven't checked them all, just CORR.)