sml 4.1 refs odd sentence on model boundary

9/19 draft has, in subject section: An SML reference is a link from one 
element in an SML model to another element from the same model.
I (think I) know the intent of that sentence, but with my "skeptical 
reader" hat on it seems problematic.
Example:
valid model == docs A, B, C
valid SML refs in this model are A/foo -> B/
I add a new SML ref from A/bar -> D, where D is a well-known popular 
location like http://www.w3.org/TR/2007/WD-sml-20070926/#Schemas
Reading the existing text literally, one of two things just happened.  I 
don't think either is necessarily part of the intent.
1. I made D part of my model
2. A/bar is not an SML reference, although there is no way syntactically 
to know that given A as an instance document.
I think our intent was that either of the above _could_ be true, but not 
_must_ be true.  In other words, the model boundary is at least in some 
sense independent of the documents contained in the model.  A model can 
validly refer to documents outside the model using SML.  Not all documents 
in a model need to be reachable via SML refs.  The problematic sentence as 
I read it essentially defines the model boundary as the set of all 
documents reachable via SML refs from any document (already known to be) 
in the model.

Best Regards, John

Street address: 2455 South Road, Poughkeepsie, NY USA 12601
Voice: 1+845-435-9470      Fax: 1+845-432-9787