I hope I get a dollar for this one.
 
In chapter 18 of the TEI Guidelines, a number of elements used for marking
problematic areas of a transcribed text (eg. supplied, gap, delSpan) all
contain an inherited attribute, "hand."  The purpose of hand, as far
as I can gather, is to refer to an authority responsible for the change
being encoded - an editor or a scribe.  The hand attribute is an idref, so
there should somewhere else in the text be an id for each possible hand
value.
 
The TEI provides a place for this in handList, an element which contains a
list of hand elements.  These too take a "hand" attribute, which is "a
unique identifier ... to refer to the scribe or handwriting style."  From
this I gather that "hand" is here the id that I need for "hand" in the
actual transcription elements.
 
BUT, the "hand" attr in <hand> has a CDATA value, not ID.  It is required,
but the parser won't (and sgmls doesn't) connect it with the "hand" attributes
that are idrefs.
 
I got around this by duplicating the info, such that my header looks something
like:
 
<handList>
  <hand id=Q hand=Q>
  <hand id=Mb hand=Mb>
</handList>
 
This parsed, but it's kind of silly.
 
**********************************************************************
- Gregory Murphy
Text Systems Manager
CETH, the Center for Electronic Texts in the Humanities
e-mail: [log in to unmask]
web:    HTTP://WWW.PRINCETON.EDU/~GJMURPHY
phone:  (609) 258-2460
**********************************************************************