As witnessed by a series of mails to TEI-L (see e.g. http://listserv.brown.edu/archives/cgi-bin/wa?A2=ind1103&L=TEI-L&P=R7694\), there is a need for using the fragmentary witness elements (witStart, witEnd, lacunaStart, lacunaEnd) outside of the rdg and lem elements where they are now allowed. The reason is that witnesses don't necessarily start or end at a location where textual variation occurs.
Council agrees this should be referred to wg of ms sig
Changed to Red. EM assigned to chase up or close.
July 2006: Birgit Kellner posted to TEI-L ponting out that
<lacunaStart> and <lacunaEnd> can only appear in the context of <rdg> or <lem> inside <app>.
This assumes that the lacuna is within a section of a witness which constitutes a variant reading and is encoded within an <app>.
However, It may be useful to mark a lacuna in the main body of the text, whether there is a variant or not, especially since it may be impossible to tell whether there is a variant at all, as the text is missing. This applies to either endpoint of a lacuna as well as the lacuna as a whole. (pun inadvertent).
Missing text can be indicated by a <gap> but this causes problems if the lacuna is large, or goes across lines or other structural elements. Birgit also points out that it is problematic "semantically, since <gap> seems to refer to a gap in the text, but not to a gap in one or more witnesses used in its critical constitution"
http://listserv.brown.edu/archives/cgi-bin/wa?A2=ind0607&L=TEI-L&F=&S=&P=637
This got no response, and was re-animated in March 2011 by Gottfried Koenen and Hayim Lapin. They are interested in automated collation, and want to mark <witStart>, <witEnd>, <lacunaStart> and <lacunaEnd> in texts without having to use <rdg> or <lem> They are both dealing with fragmentary texts.
It's possible to say that any intervention indicating the beginning or end of a lacuna or the beginning or end of a very fragmentary witness is by definition part of an apparatus, and therefore should be enclosed in <app> and <rdg >. This only makes sense if variants from different witnesses are being encoded within a single file. Hayim's project is encoding each witness separately and is probably not using the apparatus elements.
One solution is to do what is requested, and allow the four elements <witStart>, <witEnd>, <lacunaStart> and <lacunaEnd> to appear in most places that <lb> or <pb> appear by making them members of model.milestoneLike and moving them from the textcrit module to something more general. They function very much like milestone elements although the implicit structure they represent is somewhat less regular than pages, gatherings, columns or lines. On the other hand model.milestoneLike also includes <anchor>.
Another solution is to instead suggest a different way of encoding, and suggest that <anchor> or some other element be repurposed in order to indicate the same information, perhaps by giving it an @type.
The former solution is more semantically pleasing, but the latter is equally functional.
Council: discussed in depth; these belong in rdgPart; other techniques should be used for outside app. close-wontfix
status change per previous comment