This question raises a number of important issues pertaining
to the TEI's ability to deal with the physical, non-textual
aspects of "text objects".  The use of <note> with a TYPE
attribute for editorial as well as authorial and readers'
comments on the grounds that "they are all notes," as Julia
says, "and deserve to be encoded as such", while certainly
not an indefensible position, does not, it seems to me, take
sufficiently into account the fundamental distinction we ought
at least to be maintaining between what is actually physically
present in the manuscript and what isn't. At the moment there
doesn't seem to be much alternative, however, at least if one
wants to transcribe marginal comments as they occur in the
text.  Further overtaxing the <note> element by using it for a
description of the artifact or "text object" itself seems quite
inappropriate.  The MASTER project and TEI working
group in manuscript description have devised an
<msDescription> element, which can go in the source
description in the header and contains a range of sub-
elements for describing the physical object.  In addition to
elements for <support>, <extent> and the like, there is an
element for <additions> (marginalia, glosses etc.), which
may be transcribed in full.  In doing so, one can distinguish
between comments which have been written into the
manuscript and those of previous or present editors, which,
one hopes, have not.

M. J. Driscoll
Arnamagnaean Institute
Copenhagen